28688095 Parte 11 Encyclopedia of Astronomy and Astrophysics

Universe E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

UniverseThe sum of everything that exists and of which we can beaware; the entirety of space. There is a semantic difficultyin talking about the universe; on the one hand, we defineit to be ‘everything’, but it may be (a) that our universe isfinite, yet unbounded; (b) that the accessible universe isonly a small part of a much larger entity, most of whichwe cannot observe; or (c) that there exist other universesof which we are not ‘aware’.

See also: astrometry, astronomy, astrophysics.

Copyright © Nature Publishing Group 2001Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998and Institute of Physics Publishing 2001Dirac House, Temple Back, Bristol, BS1 6BE, UK 1

Universe: Simulations of Structure and GalaxyFormation

E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Universe: Simulations of Structureand Galaxy FormationThe current cosmogonic paradigm posits that structuresin the universe (such as galaxies and clusters) originatedfrom tiny density fluctuations generated around the timeof the Big Bang and subsequently amplified by gravity.The precise origin of the primordial density fluctuationsis uncertain. One possibility is that they were generatedwhen the universe underwent a brief period of exponentialexpansion known as inflation, a tiny fraction of a secondafter the Big Bang. In the most successful model, themass density of the universe is dominated by an exoticform of matter called cold dark matter which consists ofelementary particles that do not make any contribution tothe luminosity density of the universe. The predictionsof this model have been set out extensively, using bothanalytic calculations and computer simulations, the latterhaving become increasingly important over the past 20years. The build-up of structure in a cold dark matteruniverse is hierarchical: small-mass objects are the firstto condense out from the expanding universe at earlytimes (high REDSHIFT), whilst more massive objects (suchas rich clusters) form later by repeated mergers of smallerobjects. A remarkable confirmation of this framework wasprovided by the discovery, in 1992, of anisotropies in thetemperature of the COSMIC MICROWAVE BACKGROUND radiation(CMB) by the DMR instrument aboard the COBE satellite.The amplitude of these anisotropies was within a factorof two of the extant theoretical predictions. Spurred onby the deep images of the sky produced by the HubbleSpace Telescope, which show galaxies as they were whenthe universe was only about 10% of its present age, a greatdeal of effort is now being directed towards modeling theformation and evolution of galaxies within the setting ofcosmological structure formation in the dark matter. Thedissipative gas dynamical processes involved in GALAXY

FORMATION make this a challenging task that can only betackled using either idealized models or large amounts ofsupercomputer time.

The physics of galaxy formationStudies of the dynamics of galaxies confirm a fundamentalinference made 25 years ago from studies of thekinematics of stars within galaxies: on scales largerthan galactic nuclei, the dominant physical interactionis gravity. Furthermore, it is now well established thatthe predominant source of gravity is dark matter, thatis matter that does not emit detectable electromagneticradiation. Dark matter is now routinely ‘imaged’ throughthe phenomenon of GRAVITATIONAL LENSING, the relativisticdeflection of the light from distant galaxies as it passesnear an intervening cluster of galaxies. Independently ofthe exact identity of the dark matter, the dominance ofgravity leads directly to a general scheme for structureformation, first outlined by Landau and Lifshitz in the1950s, rigorously developed by Peebles during the 1970s,and calculated in detail using computer simulations in

the 1980s and 1990s. The key concept is the gravitationalinstability experienced by small matter overdensities inthe expanding universe. Matter fluctuations present inthe early universe grow in amplitude approximately as apower-law in time and eventually collapse to form self-gravitating objects.

The process of gravitational instability sets the scenefor galaxy formation, the main physical ingredients ofwhich are set out in figure 1. Although the precisedetails depend on the identity of the dark matter, underquite general conditions, galaxy formation is expected toproceed via a two-stage process originally outlined byWhite and Rees in 1978. First, gravitational instabilityacting on collisionless dark matter, results in the formationof self-gravitating dark matter haloes. Gas, initiallywell mixed with the dark matter, participates in thiscollapse, but it is heated by shocks to the thermal (orvirial) temperature of the dark matter haloes. Second,the hot gas cools radiatively, on a time scale set byatomic physics, due to bremsstrahlung, recombinationand collisionally excited line emission. The rate ofcooling depends upon the density of electrons and atomicnuclei, and so it was most efficient at high redshift, whenthe universe was denser. (In practice, the heating andsubsequent cooling of the gas may occur rapidly andchaotically, particularly in small galaxies.) Just prior togravitational collapse, angular momentum is impartedon the generally aspherical perturbations by gravitationaltorques exerted by neighboring clumps, as proposed byHoyle in 1948. Thus, the initial collapse genericallyresults in the formation of a gaseous disk. Once thedisk has become centrifugally supported, the materialin it begins to fragment into stars by processes that arestill poorly understood. In this simplified picture, thespheroidal components of galaxies form by mergers of diskgalaxies which jumble up the stellar orbits, disrupting theirorganized configuration.

The scheme outlined above provides a naturalexplanation for why there are galaxies of two basic types:disks and spheroids. It also elegantly explains whythere is a limit to the luminosity that galaxies attain atthe present day. The most luminous galaxies form inthe most massive dark matter haloes. These have onlyrecently collapsed because, as is the case in most currentmodels, the amplitude of mass fluctuations is smalleston large scales. At the high temperatures and low gasdensities prevailing in recently formed structures, coolingis very inefficient and the gas has not had time to cooland fragment into stars over the lifetime of the universe.One problem of hierarchical structure formation modelsis that the number of low-mass haloes that form exceedsthe number of faint galaxies seen in the local universe.However, a likely solution is that the feedback of energyinto the cooled gas from early generations of stars willhave acted as a self-regulating mechanism. This feedbackprevents substantial star formation activity in shallowgravitational potential wells, thus causing most of thesmall-mass haloes to harbor extremely faint galaxies.




DAMPING

FORMATION OF DEEP GRAVITATIONAL POTENTIAL WELLSDARK MATTER HALOES

QUANTUM DENSITY FLUCTUATIONS

DARK MATTER EXPANSION OF UNIVERSE

PRIMORDIAL SPECTRUM OF FLUCTUATIONS

VIRIALIZATION

RADIATIVE COOLING

STAR FORMATION

FEEDBACK PROCESSES

supernovae, stellar winds

INFLATION

GR

AV

ITA

TIO

NA

L I

NST

AB

ILIT

YD

ISSI

PA

TIV

EG

AS

DY

NA

MIC

S

Figure 1. The ingredients of the standard model for the formation of galaxies and cosmological structure.

Cosmological structure formationThe processes of gravitational instability and collapse, gascooling and star formation operate under quite generalconditions. A quantitative theory of galaxy formation,however, requires that two key cosmological questionsbe addressed: (i) what is the origin of primordial massfluctuations and (ii) what is the identity of the dark matter?Significant progress towards answering these questionswas made in the early 1980s, as the result of a fruitfulinteraction between particle physics and cosmology.

The first influential idea of the ‘New Cosmology’ wasproposed around 1980 by Alan Guth and extended byAndrei Linde. Searching for a solution to the ‘magneticmonopole problem’, Guth proposed that the universehad undergone a period of exponential expansion, orINFLATION, very soon after the Big Bang, triggered perhapsby the transition of a quantum field from a false to thetrue vacuum. Quantum fluctuations generated duringthis epoch would be swept across the event horizonand thus become established as classical ripples in theenergy density of the universe. When they cross thehorizon back again, their amplitude is independent ofscale. Thus, the inflationary model predicts a scale-invariant power spectrum of primordial fluctuations,P(k) ∝ k, with a Gaussian distribution of amplitudes.(More elaborate versions of the same idea can producemodels whose power spectra have an exponent that differsslightly from unity.) The subsequent evolution of thefluctuations depends on the values of the cosmologicaldensity parameter, 0, the Hubble constant, H0, the

COSMOLOGICAL CONSTANT, 0, and the identity of the darkmatter.

The second key idea from the 1980s concerns theidentity of the dark matter. The abundance of thelight elements (H, D, He, Li) produced during Big Bangnucleosynthesis agrees with astronomical data only ifthe present-day density of baryons is low enough fordeuterium to form in at least the abundance measuredin primitive gas clouds at high redshift. The baryondensity required for this is about one order of magnitudesmaller than the total mass density of the Universe inferredthrough a variety of tests. Thus, a fundamental conclusionis that the dark matter must consist of non-baryonicelementary particles.

Particle candidates for the dark matter are conve-niently classified into ‘hot’ and ‘cold’ varieties, a nomen-clature introduced by J R Bond around 1980. The pro-totype of a hot particle is a stable neutrino with a massof the order of a few eV. (A single species of neutrinowould give = 1 if its mass were approximately 30 eVand H0 = 50 km s−1 Mpc−1.) Examples of cold particlesinclude the least massive stable supersymmetric particle,the neutralino, and the much lighter axion. Cold particlesare often referred to as weakly interacting massive parti-cles or WIMPs. There is a fundamental difference in theway in which galaxies are predicted to form in hot andcold dark matter cosmogonies, arising from the differentdamping mechanisms that affect the development of pri-mordial fluctuations in the two cases. If the universe weredominated by massive neutrinos, then fluctuations below




some critical mass would be wiped out because the neu-trinos move at relativistic speeds in the early universe andrapidly free-stream out of overdense regions. For a singleneutrino of mass 30 eV, this critical scale corresponds to afluctuation wavelength of

λc = 2πkc= 13h2

Mpc (1)

where we have parametrized Hubble’s constant as H0 =100h km s−1 Mpc−1. The power spectrum decaysexponentially for wavenumbers k > kc.

In the case of cold dark matter, the damping of densityfluctuations is much less severe. The free-streamingscale of cold dark matter is many orders of magnitudesmaller than that of hot dark matter, and so this effectis not important. The most relevant damping process isthe Mezaros effect, whereby oscillations in the radiationenergy density stifle the growth of matter fluctuations.This situation persists until the energy density of matterdominates over that of the radiation, which occurs after aredshift of 1 + zeq = 23 900h2. The size of the horizon atthis epoch is imprinted upon the cold dark matter powerspectrum, marking the turnover of the primordial form,P(k) ∝ k, to the damped fluctuation spectrum whichasymptotes to P(k) ∝ k−3 at small scales.

In neutrino dominated models, the first structuresthat form are flat, pancake-like objects of mass comparableto that enclosed within the critical free-streaming scale,corresponding approximately to 1016M. These are objectsof supercluster scale which must somehow subsequentlyfragment, in a ‘top-down’ fashion, in order for galaxies toform. Early computer N -body simulations of this processcarried out by Frenk, White and Davis in 1983 showed thatfor a neutrino dominated universe not to exceed the levelof clustering measured in the galaxy distribution today,galaxies would need to form at redshifts z 1. Yet, itwas already known at that time that quasars can havemuch higher redshifts than this and, today, we know thatthere is a large population of galaxies already establishedat z = 3–5. Thus, although rather appealing at first sight,neutrino dominated universes with Gaussian primordialfluctuations were soon abandoned.

The alternative, a cold dark matter universe, provedto be much more successful, as discussed, for example, in aseries of papers in the 1980s based onN -body simulations,by Davis, Efstathiou, Frenk and White. The definingproperty of the fluctuation spectrum in a cold dark matteruniverse is that small-scale perturbations are preserved.Thus, subgalactic mass haloes are the first to collapse andseparate out from the expansion of the universe. Thesehaloes then grow, either gradually by accreting smallerclumps, or in big jumps by merging with other haloesof comparable size. The timetable for the formationof structure in a universe dominated by cold particlesis hierarchical or ‘bottom-up’—small objects form first,larger objects form later. The cold dark matter fluctuationpower spectrum thus specifies completely the evolution

of the merging hierarchy of dark matter haloes into whichthe baryons must fall in order to make the galaxies.

Currently, the most successful version of the cold darkmatter model has around 30% of the critical density in coldmatter and 70% in the form of a vacuum energy densityor cosmological constant term, so that the universe has aspatially flat geometry. This model is in good agreementwith a range of observational data: the amplitude of theangular power spectrum of temperature fluctuations inthe cosmic microwave background (including the locationof the first ‘Doppler’ peak), the abundance of clusters ofgalaxies ranked by their x-ray luminosity, the clusteringof galaxies on large scales, and the expansion rate ofthe universe as deduced from the brightness of distantsupernovae.

The recent detection of oscillations in neutrino flavourby the Super-Kamiokande experiment, which require theneutrino to have a non-zero mass, has rekindled interestin the possibility that neutrinos might, after all, makesome contribution to the density of the universe. In orderto avoid the problems faced by a universe dominatedby hot dark matter outlined above, most of the mass inthe universe must still consist of cold dark matter, withneutrinos providing only a minority contribution. Such‘mixed’ dark matter models have not proved as successfulat matching the data as the pure cold dark mattermodel. Another possibility for accommodating massiveneutrinos is to replace inflation with a different mechanismfor generating primordial density fluctuations. TomKibble suggested in the 1980s that TOPOLOGICAL DEFECTS IN

COSMOLOGY might arise during phase transitions in scalarfields present in the early universe. Certain classes ofdefects, such as strings and textures, can act as seedsonto which matter gravitates, generating inhomogeneitiesin the mass density of the universe. These have thedefining feature of being non-Gaussian. Unfortunately,modeling the formation of structure in defect models hasturned out to be more complicated than in the Gaussiancase because, amongst other subtleties, the properties ofthe defects themselves evolve with time. Defect modelshave not yet been explored with the same degree of rigoras the cold dark matter model. However, the currentindications are that they have difficulties reproducing thespectrum of temperature anisotropies measured in thecosmic microwave background radiation. Thus, froma cosmological point of view, the oscillation data arebest accommodated if the neutrino mass is very small(much less than 1 eV), so that neutrinos make a negligiblecontribution to the cosmic mass budget.

Computer simulations of galaxy formationThe remarkable developments of the past 15 years—theidea of cosmic inflation, the cold dark matter model, thediscovery of ripples in the microwave background, and theobservations of galaxies at high redshift—have laid downvery solid foundations on which to build an understandingof galaxy formation. The ‘initial conditions’ for theevolution of the dominant dark matter component and its




z=3 z=1 z=0

CDM

SCDM

CDM

OCDM

Figure 2. The formation of structure in N -body simulations of representative cosmological volumes of the universe. The intensity ofthe shading indicates the density of cold dark matter. Each row shows results from different versions of the cold dark matter model.The top row is a flat universe with 0 = 0.3 and a cosmological constant; the middle two rows are 0 = 1 universes, with differentpower spectra; the bottom row is an open universe with 0 = 0.4. The images, from left to right, show the evolution of structure ineach model as a function of redshift. The present day corresponds to z = 0 while z = 3 corresponds to the epoch when the universewas approximately 15% of its current age. (Courtesy of the VIRGO Consortium for Cosmological N -body Simulations.)

subsequent gravitational evolution are well understood.Yet formulating an ab initio theory of galaxy formationand evolution over 10 billion years of cosmic historyremains a tall order. The main stumbling block is ourpoor understanding of the behavior of cosmic gas—most probably a complex, dynamic, multiphase medium,of the physics of star formation, and of the feedbackbetween the two mediated by winds from massive starsand supernovae explosions. The best way to addressthese issues is through extensive computer simulation andmodeling.

The basis for present-day cosmological simulations is

the N -body technique, which has been very successfullyapplied to modeling the evolution of collisionless darkmatter. Using various computationally efficient methods,the computer solves the coupled equations of motionof N particles, interacting only through gravity, in theexpanding universe. Progress over the past two decadeshas been driven mainly by dramatic improvements inthe speed and memory of computers. By way ofillustration, the early simulations of the cold dark mattercosmogony in 1985 employed 32 768 particles. In 1999,the largest simulations performed on massively parallelsupercomputers (using essentially the same algorithms as




Figure 3. A high-resolution simulation of the formation of a single dark matter halo in a cold dark matter universe. The brighter colorsindicate higher densities of dark matter. The sequence shows a series of snapshots of the evolution of the halo, at the redshiftsindicated in the legend. The present-day halo displays a significant amount of substructure within its virial radius. (Courtesy of BenMoore, Joachim Stadel, Tom Quinn and George Lake.) This figure is reproduced as Color Plate 69.

those of the 1980s) can follow the evolution of 109 particles.

Snapshots from simulations of representative, cosmo-logical volumes are displayed in figure 2. This figure il-lustrates the evolution of structure in four versions of the

cold dark matter model, differing only in the values of thecosmological parameters, and. At the present day, thedark matter is arranged in a complex network of voids, fil-aments and super-clusters (dubbed the ‘Cosmic Web’ byBond, Kofman and Pogosyan). It is similar in all the sim-




Figure 4. The star formation history of the universe. The curveshows the theoretical prediction for a cold dark matter universe,from the Durham semi-analytic model of galaxy formation. Thepoints show a selection of recent observational determinationsof the star formation rate density expressed in units of thepresent-day value.

ulations, partly because the amplitude of the initial fluc-tuation spectrum was adjusted so that all models wouldapproximately reproduce the observed local abundance ofhot, x-ray emitting clusters. (In addition, the initial fluctu-ations were set up with the same random phases in all thesimulations, so that structures form at the same locationswithin each volume.) The development of structure pro-ceeds at different rates in the different cosmological mod-els. In those with a low value of (top and bottom rows infigure 2), the formation of structure is essentially frozen athigh redshift, whereas in the models with = 1, growthcontinues to the present. This merely reflects the fact thatlow-density universes expand more rapidly at late times,with the result that the expansion overwhelms the accre-tion of matter onto overdensities which lies at the root oftheir growth.

Simulations of sufficient size to resolve the internalstructure and dynamics of dark matter haloes haverecently become possible. The largest of these employseveral million particles to model the formation of a singlehalo, revealing the existence of a rich substructure oflumps within the virial radius (figure 3). The simulationsshow that there is a remarkable uniformity in the densitystructure of dark matter haloes: over a wide range ofscales, the spherically averaged halo density profile infollows a simple form proposed by Navarro, Frenk andWhite in 1996:

ρ(r) ∝ 1r/rs(1 + [r/rs]2)

(2)

where rs is a scale length related to the density of theuniverse at the time when the halo formed. This functionalform appears to be universal, independent of the choice ofhalo mass, power spectrum or cosmological parameters.

The N -body technique can be augmented withnumerical hydrodynamic methods to model the evolutionof gas subject to cooling and heating processes and coupledgravitationally to the dark matter. Two such methods arecurrently in use: Eulerian methods (including adaptivemesh refinements in some cases) and a Lagrangianscheme known as smoothed particle hydrodynamics. Thetwo techniques have advantages and disadvantages, butso far the latter has proved to be the best suited tostudying galaxy formation, because of the formidabledynamic range in thermodynamic quantities involvedin the problem. At present, most progress has beenachieved in modeling the physics of primitive gas cloudsat high redshift, the so-called LYMAN ALPHA FOREST, detectedobservationally as absorption lines in the spectrum ofdistant quasars. The first simulations capable of followingthe evolution of gas to the present, with enough resolutionto model the brightest galaxies, are now being carriedout by various groups around the world. Currently,only a subset of the relevant gas physics, such as theshock heating of gas within dark matter haloes and itssubsequent radiative cooling, are treated reliably.

A complementary technique for simulating galaxyformation ab initio, known as semi-analytic modeling,was developed in the 1990s by researchers at Durhamand Munich. The main difference with the directsimulation approach is the abandonment of the idealof solving the equations of hydrodynamics directly,in favor of a simple, spherically symmetric model inwhich the gas is assumed to have been fully shock-heated to the equilibrium temperature of each halo, fromwhere its cooling and accretion onto the halo can beaccurately calculated. This simplification speeds up thecalculations enormously and has the added advantage ofbypassing resolution considerations which are one of themain limiting factors of full hydrodynamic simulations.Phenomenological models of star formation, feedbackand metal enrichment by supernovae are included in thesemi-analytic program, through simple scaling relations.The semi-analytic machinery may be grafted onto haloesgrown in a cosmological N -body simulation or ontohaloes whose formation histories have been generatedusing a Monte Carlo approach. The models describe theentire star formation and merger history of the galaxypopulation. The free parameters of the model, which,perhaps surprisingly, are rather few in number, can beset by requiring a good match to a selection of propertiesof the local galaxy population, such as its luminositydistribution. This results in a fully specified model thatprovides an ideal tool for comparing the predictions ofthe cold dark matter theory with observations of the highredshift universe.

Confronting the high-redshift universeThe combination of direct simulations and semi-analyticmodeling has revealed in detail the manner in whichgalaxies are expected to form in the cold dark mattermodel. The picture that emerges is one of gradual




Figure 5. The evolution of clustering in the dark matter and galaxies. The left-hand panel shows an N -body simulation of a flat,low-density, cold dark matter universe with a cosmological constant, in a cube of comoving side 141h−1 Mpc, at z = 3. The right-handpanel shows the same simulation evolved to the present day. The gray scale indicates the density of the dark matter. Dark matterhaloes in the simulation have been ‘populated’ with galaxies using a semi-analytic model of galaxy formation. The color of each spotreflects the color of each model galaxy, which is sensitive to the star formation rate. The size of the spot increases with the absoluteluminosity of the galaxy. The inset shows the development of a cluster of galaxies. (Courtesy of Andrew Benson, Shaun Cole, CSF,CMB and Cedric Lacey.) This figure is reproduced as Color Plates 70 and 71.

evolution punctuated by major merging events that areaccompanied by intense bursts of star formation andwhich trigger the transformation of disks into spheroids.Galaxy formation stutters into action around z ∼ 5. Only atiny fraction of the stars present today would have formedprior to that time. By z ∼ 3, the epoch when galaxiesisolated by the ‘Lyman-break’ technique1 are observed,galaxy formation has started in earnest, even though only10 per cent of the final population of stars has emerged.The midway point is not reached until about a redshiftof 1–1.5, when the universe was approximately half ofits present age. These theoretical predictions are shownin figure 4. Observationally, the star formation rate perunit volume can be inferred from the density of ultravioletradiation, which is a measure of the number of high-mass,short-lived stars. Estimates of the star formation densitybased on data taken by ground-based telescopes and by theHubble Space Telescope are shown as the points in figure 4.The major uncertainty in the interpretation of these datais the obscuring effect of dust, a modest amount of whichhas been allowed for in the models. But unless this effectturns out to be much stronger than anticipated, the theoryand data in figure 4 suggest that we have now tracked

1 The so-called ‘Lyman-break’ galaxies are detected in passbandsabove the redshifted Lyman-limit at 912 Å and undetected inpassbands below this limit; the strength of the Lyman-limit breakis enhanced by absorption due to cold gas in the galaxy and inclouds along the line of sight.

most of the star formation activity, and the associatedproduction of chemical elements, over the entire lifetimeof the universe.

Asecond important prediction of the cold dark mattertheory concerns the clustering properties of galaxies athigh redshift. At the heart of the hierarchical clusteringprocess lies the fact that galaxies tend to form first nearhigh peaks of the density field because these are thefirst to collapse at any given epoch. This predilectionfor high-density regions is known as ‘biased galaxyformation’ (a term introduced by M Davis in 1985),because the distribution of galaxies offers a biased viewof the underlying distribution of mass. An importantconsequence of biased galaxy formation is that brightgalaxies tend to be born in a highly clustered state andremain so for long periods of time. The process of biasedgalaxy formation is illustrated in figure 5. The left-handpanel shows a snapshot of an N -body simulation of acold dark matter universe at z = 3, whilst the right-handpanel shows the same simulation evolved to the present.The semi-analytic model of galaxy formation has beenimplemented in the dark matter haloes identified in thesimulation at each redshift, in order to populate them withgalaxies. Galaxies that are bright enough to be detectedat z = 3 may be seen to form at the locations where thedark matter density (depicted by the gray scale) is highest.Observational confirmation of this clustering predictioncame with the discovery that the population of galaxies at




z ∼ 3 identified by the Lyman-break technique is about asstrongly clustered as bright galaxies are today. The relativeclustering strengths of galaxies and dark matter evolvequite differently. The right-hand panel of figure 5 showsthat the dark matter is much clumpier today than it was atz = 3. On the other hand, the clustering pattern of galaxieshas hardly changed over this long period of time. Galaxiestoday are found in a wide range of environments and havea clustering amplitude similar to that of the dark matter.This was not the case at high redshift when bright galaxieswere much more strongly clustered than the dark matter—in other words, when they were very strongly biased.

The next stepsThe two areas of agreement between theory and datahighlighted here—the cosmic history of star formation andthe clustering of high-redshift galaxies—are particularlynoteworthy because they concern fundamental aspects ofthe theory. The broad agreement between models and datasuggests that the main ingredients of a coherent picture ofgalaxy formation are now in place. These ingredients are:primordial Gaussian density fluctuations; collisionless,non-baryonic dark matter; gravitational instability; andthe growth of galaxies by hierarchical clustering. Thereare justifiably high expectations for the next decade. Thenumber of 10 m class telescopes is proliferating: thefirst of the four European Space Observatory ‘Very LargeTelescopes’ came into full operation in 1999, the sameyear in which the Gemini and Subaru telescopes first sawlight. Other large telescopes are under construction inthe USA and Spain. The middle of the first decade ofthe new century should also see the launch of NASA’sNext Generation Space Telescope, that will search forgalaxies out to a redshift of 10, and ESA’s Planck Surveyor,that will map the microwave background radiation withunprecedented precision. Towards the end of the decade,the ‘Large Millimeter Array’, sponsored by a majorinternational partnership, is scheduled to come intooperation in the Chilean desert. It will search for galaxyformation at high redshift and examine star formationin nearby galaxies in the still relatively unexplored sub-millimeter waveband. Ultimately, the cornerstone uponwhich much of modern cosmology rests is the idea thatthe universe is dominated by non-baryonic dark matter.Experiments under way in the UK, Italy and the USAstanda good chance of detecting it, if it really exists, within thenext few years. This will no doubt count as one of themost exciting discoveries in the history of science. Fortheir part, theorists will not be standing still. Increasedcomputing power, more efficient algorithms and, aboveall, a better understanding of the astrophysics of galaxyformation, are likely to result in a pretty good imitation,by computer, of the processes through which galaxies inour universe formed.

BibliographyA pedagogical discussion (at an advanced level) of thephysics of structure formation in the expanding universe

and of the processes that play a role in galaxy formationmay be found in:

Peacock J A 1999 Cosmological Physics (Cambridge:Cambridge University Press)

Numerical simulations are reviewed in:

Bertschinger E 1998 Ann. Rev. Astron. Astrophys. 36 599–654

Some of the observational data discussed in this article arereviewed in:

Ellis R S 1997 Ann. Rev. Astron. Astrophys. 35 389–443

Carlton M Baugh and Carlos S Frenk


Universe: Thermal History E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Universe: Thermal HistoryHUBBLE’s discovery of the expansion of the universe in 1929revealed our beginning from a much smaller and muchdenser initial state (BIG BANG THEORY). Penzias and Wilson’sdiscovery of the COSMIC MICROWAVE BACKGROUND radiation(CMBR) in 1964 implied further that just after creationthe universe was a hot soup of the fundamental particleswhose dynamics was controlled by the energy density ofthe CMBR.

The microwave radiation that A A PENZIAS and R WWILSON discovered is more precisely BLACK-BODY RADIATION

characterized by a temperature of around 3 K. Becausethis black-body radiation fills space and its photonsoutnumber all other photons and known particles bya billion to one, it sets the average temperature of theuniverse. Of course, many hotter places exist in thecosmos—stars, planets and even the interstellar medium.As the universe expands its temperature decreasesinversely as its linear size. While the temperature todayseems unimportantly small, the fact that the universe stillhas a measurable temperature means that it was incrediblyhot in the beginning.

The thermal history of the universe is thus the storyof what happens when a tremendously hot and denseplasma expands and cools. Understanding the veryearliest moments requires knowledge of the fundamentalparticles and their behavior under extreme conditions(high densities and temperatures). The connectionbetween the inner space of elementary particle physicsand the deep outer space of cosmology which plays suchan important role in cosmology today was born with thediscovery of the CMBR.

The known thermal historyThe hot Big Bang cosmological model (see COSMOLOGY:

STANDARD MODEL) provides an account of the universe fromthe hot soup of quarks, gluons, leptons and photons thatexisted before 10−5 s, when the transition from quarks andgluons to neutrons, protons and related particles occurred,until the present. During much of the time the universewas in a state of near thermal equilibrium; needless tosay, the departures from thermal equilibrium are veryimportant and make the universe an interesting placetoday! As the universe expanded, it cooled, T ∝ 1/R(t),where R(t) is the cosmic scale factor which sets the size ofthe universe. As it cooled, layer upon layer of structureevolved, beginning with the neutrons and protons beingproduced from quarks culminating with the building ofthe largest structures seen today, the great walls of galaxies.

During its earliest moments, the temperature wasthe key to describing the state of the universe, becauseit sets the level of thermal particle energies. The thermal-energy scale kBT ∼ 1 MeV (T /1010 K) determines whenit is energetically favorable for the next layer of structureto form, and which particle species are present in greatnumber. When kBT was much greater than the rest massof a particle, it and its antiparticle were easily produced in

pairs and were present in the thermal soup in numberscomparable to photons. During the radiation era, thethermal-energy scale and age of the universe were relatedby: kBT ∼ 1 MeV

√t (in s).

The soup of particles that existed at 10−5 s consisted ofthe up, down and strange quarks and their antiparticles;electrons, electron neutrinos, muons, muon neutrinos, tauneutrinos and their antiparticles; eight types of gluons andphotons (the eight massless gluons are the carriers of thestrong color force). The thermal-energy scale at this timewas about 200 MeV (T ∼ 2× 1012 K), and the other quarks(charm, bottom and top) and the tau lepton were too heavyto be pair produced and were not present (any presentwould have annihilated). At even earlier times, when itwas hotter, they (and possibly other particles) would havebeen present in great numbers too.

According to quantum chromodynamics (QCD), thetheory of the strong color force, a phase transition froma quark–gluon plasma to hadronic matter occurred at atemperature of around 1012 K. Because of the increasingstrength of the color force with distance, all particles withcolor (quarks and gluons) became confined in colorlessquark triplets (neutrons, protons and other particles,known as baryons) and quark–antiquark pairs (pions,kaons and other particles, known as mesons). Collectively,the mesons and baryons are known as hadrons; thehadrons are the particles that experience the strong nuclearforce.

At the end of the phase transition from quarks andgluons to hadronic matter, almost all the hadrons weretoo heavy to be pair produced. They could, of course,still annihilate, and as the temperature approached 1011 Kannihilations had eliminated almost all of the hadrons.Were it not for the slight excess of baryons over antibaryons(whose origin is still a mystery, but see next section),nucleons (neutrons and protons) would have annihilatedand disappeared too. The one additional nucleon for everybillion or so antinucleons, left a few nucleons for everyseveral billion photons without antinucleon partners toannihilate with. The primordial excess of baryons overantibaryons is responsible for all the ordinary matter thatexists in the universe today.

At an age of about 1 s and temperature of about 1010 K,the primary constituents of the universe were photons,electron–positron pairs, and neutrinos and antineutrinosof all three species. There were also a few nucleons forevery ten billion or so photons, about equally dividedbetween neutrons and protons. Over the next 200 s or so,a sequence of nuclear reactions occurred out of thermalequilibrium and synthesized about 25% of the nucleonsinto 4He. Trace amounts of D (a few parts in 105) and 3He(about one part in 105) escaped being incorporated into 4Heand a tiny amount of 7Li (a few parts in 1010) was produced.The other 75% of the nucleons remained as free protons.This series of events is known as Big Bang NUCLEOSYNTHESIS

(BBN). The rest of the periodic table was produced billionsof years later by nuclear reactions in stars.



Big Bang nucleosynthesis occurred rapidly and at lowdensity (around 10−2 g cm−3), while the rest of the elementswere cooked more slowly and at higher density (around102 g cm−3 or higher) in stars and stellar explosions. Thisexplains the great differences in the nuclear yields ofthe Big Bang and stellar nucleosynthesis. In particular,Coulomb barriers and the lack of stable nuclides of mass5 and 8 prevented BBN from producing elements beyond7Li.

Big Bang nucleosynthesis provides the earliest test ofthe standard cosmology as well as a probe of conditions inthe early universe. The fact that the pattern of abundancesseen in the most primitive samples of the cosmos isconsistent with its predictions is one of the experimentalcornerstones of the standard cosmology. Further, the exactyields of the light elements, most especially deuterium,depend upon the baryon mass density; from recentmeasurements of the primeval deuterium abundance inhigh-redshift clouds of largely unprocessed hydrogen, wecan infer that ordinary matter (i.e. matter comprised ofneutrons and protons) today contributes between about4% and 6% of the critical density. (The average density ofthe universe determines its curvature: a critical-densityuniverse is spatially flat; a subcritical-density universeis open or negatively curved and a supercritical-densityuniverse is closed or positively curved.) Because BBN‘weighs’ all the ordinary matter at a simpler time, itprovides the most accurate determination of the amountof ordinary matter. Today, baryons exist in many forms—bright stars, faint stars including white dwarfs and blackholes, clouds of cold gas and of hot gas, and dust—andare more difficult to inventory. Thus far, only about one-third of the BBN-determined baryon abundance has beendirectly accounted for.

Light-element production also depends upon theambient conditions in the universe, and Big Bangnucleosynthesis can thus also be used as a probe ofthe particle soup that existed then. For example, theexistence of an additional neutrino species beyond the tauneutrino would have led to additional 4He production, incontradiction to the observations. This argument againstthe existence of another neutrino species was put forthin the 1980s by David SCHRAMM and his collaborators andwas confirmed by experiments at particle accelerators inthe 1990s.

Two other important thermodynamical events tookplace during Big Bang nucleosynthesis. At a tempera-ture of around 1010 K neutrinos and antineutrinos (all threespecies) ceased interacting with electron–positron pairsand decoupled from the electromagnetic plasma. There-after, they evolved independently of the rest of the uni-verse, interacting only through gravity. Neutrino decou-pling occurred because the decreasing particle energiesand densities made neutrino interactions with other par-ticles increasingly infrequent. When the temperature wasaround 109 K, essentially all of the electrons and positronsdisappeared as pairs destroyed by annihilations were nolonger replenished by thermal pair creation. The slight

excess of electrons over positrons preserved the few elec-trons per ten billion photons required to balance the chargeof the protons. The electron–positron annihilations raisedthe number of photons in the universe by a factor of 11/4and heated the photons slightly relative to the neutrinos;thereafter Tν = (4/11)1/3T .

The radiation era ended when the universe wasaround 40 000 years old and the temperature was about10 000 K. At this epoch, called matter-radiation equality,the energy density contributed by matter (both baryonsand exotic dark matter, more below) and that by relativisticparticles (photons and three neutrino species) were equal.Thereafter, the matter density would exceed that ofrelativistic particles, growing in proportion to the linearsize of the universe. (The matter density decreases as1/R(t)3 due to the volume dilution effect of the expansion;the energy density of the CMBR decreases as 1/R(t)4, withthe extra factor of 1/R arising because photon energies areredshifted with the expansion.) Today, the energy densityof matter is about a factor of 4000 times larger than that ofthe energy density of photons and neutrinos.

The dawning of the matter era marked the beginningof the formation of large-scale structure in the universe(see also UNIVERSE: SIMULATIONS OF STRUCTURE AND GALAXY

FORMATION). The small inhomogeneities in the distributionof the exotic matter that existed (spatial variations in themass density at the level of about one part in 105) begangrowing through the attractive force of gravity; prior tomatter–radiation equality the universe was expanding toofast for this to occur. Their tight coupling to photonsprevented baryons from participating in this growth.

Shortly after the radiation era ended, at a time ofaround 400 000 years and a temperature of around 3000 K,two related and very significant events involving theradiation took place. The first was the transition fromionized matter to neutral matter (called ‘recombination’,which is paradoxical since neutral matter had notpreviously existed). As the temperature dropped below3000 K neutral matter became thermodynamical favored,and all but a few ions combined with the free electronsto form neutral atoms (a residual ionization fractionof around 10−4 persisted thereafter because ions andelectrons became too rare to find one another to combineto form atoms). When the universe became neutral, itsopacity dropped precipitously (free electrons efficientlyscatter light, neutral atoms do not), and matter andradiation decoupled. Photons streamed freely and havenot scattered since; this important event is referred toas last scattering. Once baryons decoupled, they wererapidly pulled into the cosmic structures being formed bythe gravity of the exotic dark matter.

The black-body character of the radiation, establishedby the hot, dense conditions in the early universe, waspreserved by the expansion of the universe (the deepmathematical reason involves the conformal invarianceof Maxwell’s equations and the conformal nature of theexpansion), albeit with a decreasing temperature, T ∝1/R(t)). Today, this black-body radiation, which at last



scattering resembled the light emitted by the Sun today,has been redshifted to the microwave part of the spectrum.In 1996, the far infrared absolute spectrophotometer(FIRAS) instrument on the Cosmic Background Explorer(COBE) satellite made the most precise measurement of itstemperature, T = 2.7277 ± 0.002 K, and showed that anydeviations from a perfect black-body spectrum are smallerthan 0.005%. Because there is no other viable mechanismfor producing such perfect black-body radiation, thespectrum of the CMBR is one of the experimental pillarsof the hot Big Bang cosmology.

The radiation in the CMBR has not scattered sincethe universe was 400 000 years old, and so it provides asnapshot of the universe at a simpler time, when matterwas still nearly uniformly distributed and stars, galaxiesand clusters of galaxies did not exist. The variationsin the intensity (or temperature) of the CMBR acrossthe sky today map the two-dimensional distribution ofmatter at this time because variations in the mass densityproduce temperature variations of the same size. Thus,the temperature variations of a few parts in 105 measuredby the differential microwave radiometer (DMR) onthe COBE satellite and other balloon and ground-basedexperiments imply the existence of variations in the matterdensity of approximately the same size. This level ofinhomogeneity is just what is needed to produce the large-scale structure seen today—provided that the bulk of thematter is exotic dark matter and not baryons (more below).The variations (or anisotropy) of the CMBR also encode awealth of information about the early universe and howlarge-scale structure formed. Higher precision and higherangular-resolution measurements will be made by futureexperiments including NASA’s MAP satellite (scheduledfor launch in late 2000) and ESA’s PLANCK SURVEYOR satellite(scheduled for launch in 2007).

Three cosmic seas of thermal neutrinos shouldbe with us today. Just as with the CMBR, theexpansion of the universe maintained their thermal(Fermi–Dirac) distributions with a temperature that hasdecreased inversely with the cosmic scale factor sincethey decoupled. Because neutrinos did not share in theenergy release from the electron–positron annihilations,the temperature of the neutrino seas is predicted to besmaller than that of the photons, Tν = 1.947 K. If theycan be detected, these neutrinos will reveal the universeas it was about 1 s after the beginning. However,because low-energy neutrinos interact extremely weaklywith ordinary matter their detection presents one of thegreatest challenges in all of science.

Beyond the standard cosmology: the very earlyuniverseThe earliest history of the universe (before 10−5 s) is stilla mystery, but is under intense study. The motivationis twofold: the hope that events which took place mayexplain some of the most pressing cosmological puzzles.For example, the reason for the small excess of matterover antimatter, the explanation for the regularity seen

in the universe, and the origin of the primeval densityinhomogeneities. The second motivation is the possibilitythat the early universe can be used to probe fundamentalphysics more deeply than particle accelerators and otherEarth-based experiments. At the moment, the discussionof the universe at times earlier than 10−5 s is speculative,both because of uncertainties about the microphysicsneeded to describe these early times and the absenceof cosmological tests like Big Bang nucleosynthesis. Inany case, the physics and the cosmology are of sufficientinterest to merit the discussion of the possibilities.

On fairly firm ground, the discussion of the thermalhistory of the universe can be extended back to around10−11 s when the temperature was about 1015 K. This wassufficiently hot that the thermal-energy scale exceededthe rest masses of all known particles. At this time thethermal soup should have included all the quark andlepton species, gluons, photons and the W± and Z0 bosons,all in roughly equal abundance.

The state of the universe earlier than this is muchless certain. The prevailing belief is that a phasetransition occurred and restored the full symmetry of theSU(2)⊗U(1) gauge theory of the electroweak interactions.(SU(2)⊗U(1) is the mathematical symmetry that underliesthe unified theory of the electromagnetic and weakinteractions.) At low temperatures the symmetry betweenthe weak and the electromagnetic interactions is notmanifest: the electromagnetic interaction has long range,while the weak interaction has very short range becausethe mediators of the weak force, the W± and Z0 bosons, arevery massive. The symmetry is said to be spontaneouslybroken, by the Higgs mechanism. When the symmetry isrestored, all of the force mediators become massless. If theHiggs mechanism is correct, then there is at least one morespin-zero particle species, the Higgs boson, whose restmass is greater than about 100 GeV and probably less than300 GeV. Its discovery would be a striking confirmationof spontaneous symmetry breaking, and thus the Higgsis at the top of the ‘most wanted’ list at all acceleratorlaboratories.

As successful as the electroweak theory is, it providesonly a partial unification of the forces, leaving out thestrong color force and gravity. One possibility is thereare other levels of symmetry breaking and symmetrybreaking phase transitions. The simplest idea, grandunification, unifies the color force with the electroweakforce. Estimates for the temperature at which the grandunification phase transition might take place are even moreuncertain, but are around 1029 K, corresponding to a timeof about 10−39 s.

Another interesting feature of symmetry breaking isthe possibility that the phase transition did not occursmoothly and that ‘defects’ are created. (Such defects areknown to be produced in phase transitions in condensedmatter systems: vortices and magnetic flux tubes.) Theseso-called topological defects are concentrations of energy:point-like magnetic monopoles, one-dimensional cosmicstring and two-dimensional domain walls. The kinds of



defects that can be produced depend upon the symmetrybreaking pattern. Thus far, the cosmology of topologicaldefects has not been promising: monopoles shouldhave been grossly overproduced, domain walls havedisastrous cosmological consequences, and cosmic string,once thought to be a possible seed for the formation ofstructure in the universe, predicts a pattern of CMBRanisotropy that is inconsistent with the measurements.

A much more promising idea arising from the con-sideration of cosmological phase transitions is INFLATION.Inflation refers to an enormous burst of expansion whichmight have taken place very early on (probably earlier thanabout 10−32 s). Because of its potential to explain a numberof the most fundamental and most puzzling features of theuniverse, inflation has been the dominant theoretical ideain cosmology over the past 15 years. It can account for thesmoothness of the universe, the origin of the primeval mat-ter inhomogeneity, the heat of the Big Bang, and the natureof the Big Bang itself. Originally inflation was thoughtto be driven by the latent heat (or false-vacuum energy)associated with a first-order phase transition. Most mod-els of inflation no longer involve a phase transition, butinstead rely upon the potential energy of a fundamentalscalar field.

Besides inflation, the most compelling ideas ofearly-universe cosmology are particle dark matter andBARYOGENESIS. It has been known for more than 50 yearsthat most of the matter that holds galaxies and clusters ofgalaxies together is not in the form of visible stars but is‘dark’ (i.e. does not emit or absorb detectable radiationof any form; see DARK MATTER: ITS NATURE). For at least adecade it has also been known that the total amount ofdark matter exceeds by more than a factor of three theamount of matter in the form of baryons as determinedfrom Big Bang nucleosynthesis. Further, with the level ofinhomogeneity measured by COBE, the observed large-scale structure can only form if there is exotic dark matter.These facts are strong circumstantial evidence for a newform of matter in the universe.

The most promising candidates for the dark matterare elementary particles that were present in copiousnumbers in the thermal soup early on and which failedto annihilate away because of the weakness of theirinteractions. Of all the possibilities considered, the threemost attractive are neutrinos (if they have a small mass)and axions or neutralinos (if they exist). (The axion andneutralino are as yet hypothetical particles predicted toexist by theories that unify the particles and forces ofnature.)

Baryogenesis is a higher-level analog of Big Bangnucleosynthesis: BBN explains how baryons cometogether to make nuclei and baryogenesis hopes to explainthe origin of the excess of quarks over antiquarks that leadsto the existence of ordinary matter. The idea is that particleinteractions that violate matter–antimatter symmetry andthe conservation of baryon number and which occurredout of thermal equilibrium produced a slight excess ofquarks over antiquarks. When the quarks formed into

neutrons and protons, this led to the excess of baryons overantibaryons needed to ensure the existence of the ordinarymatter.

The weak interactions violate matter–antimattersymmetry at a small level and are also predicted to violatebaryon-number conservation through subtle quantumeffects. Moreover, baryon-number non-conservation isa generic prediction of grand unified theories. Whilethe details are not currently understood, nor is thereany experimental evidence for the non-conservation ofbaryon number, nonetheless baryogenesis is a promisingframework for understanding how the crucial excess ofmatter over antimatter arose.

To date, superstring theory has been the mostsuccessful approach to the unification of gravity withthe other forces. Superstring theory makes two genericpredictions relevant for cosmology. First, the existenceof a new symmetry of nature that relates fermions andbosons (supersymmetry), and second, the likely existenceof additional spatial dimensions. Because the knownparticles of nature cannot be paired off as fermion–bosonpartners, supersymmetry requires the doubling of thenumber of fundamental particles. The superpartners, asthey are called, are predicted to have rest energies of theorder of 100 GeV. If correct, this implies a doubling of thenumber of particles in the primordial soup only occursat temperatures greater than around 1015 K. The lightestsuperpartner, usually the neutralino, is stable or very long-lived and has a rest energy of order 100 GeV. As mentionedabove, it is a prime candidate for particle dark matter.

If there are extra spatial dimensions they must be‘small’ enough to have escaped detection or be otherwisehidden from us. Small here refers to their being curledup like the circular dimension of a straw. While manyversions of superstring theory predict that the extradimensions are exceedingly tiny—10−34 cm or smaller—some versions suggest that they might be as large as amillimeter in extent! The cosmological implications ofextra dimensions are not well understood and raise ahost of additional cosmological questions, for example theexplanation for the size discrepancy between the familiarthree spatial dimensions and the extra spatial dimensions.

Even if superstring theory does not prove successfulin unifying gravity with the other forces and in providinga quantum description of gravity, ‘interesting physics’should have occurred at times earlier than 10−44 s andtemperatures greater than 1032 K. This is the Planck era,the epoch when quantum gravitational effects should havebeen extremely important and the classical descriptionof gravity given by general relativity should have beeninapplicable. It could be that the universe achieved alimiting temperature due to the exponentially growingnumber of particle species (also a prediction of stringtheory), or that space-time dissolves into a foam. Even bythe standards of early universe cosmology, speculationsabout the Planck era are extraordinarily speculative.



BibliographyBernstein J 1988 Kinetic Theory in the Expanding Universe

(Cambridge: Cambridge University Press)Kolb E W and Turner M S 1990 The Early Universe (Reading,

MA: Addison-Wesley)Schramm D N and Turner M S 1990 Rev. Mod. Phys. 70 303

Michael S Turner


Universitat Radioastronomisches Institut, Bonn E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Universitat RadioastronomischesInstitut, BonnAn institute of the University of Bonn, Germany.Carries out research at infrared, radio and submillimeterwavelengths. Particular research interests are theinterstellar medium and intergalactic medium, especiallygalactic halos, and dwarf galaxies.

The institute was responsible for the discovery of thex-ray halo around the Milky Way in 1991.

Although it has recently disposed of a 25 m radiotelescope in the Eiffel Mountains, the institute has acooperative agreement with the Institute of Physics,Cologne, to use KOSMA, a 3 m radio telescope onGornergrat in Switzerland.

For further information seehttp://www.astro.uni-bonn.de/.


University Observatory, Ludwig-Maximilians-Universitat


University Observatory,Ludwig-Maximilians-UniversitatThe University Observatory of Ludwig-Maximilians-Universitat was founded in 1816. Astronomers whoworked or graduated at the Munich Observatoryinclude: Fraunhofer, Soldner, Lamont, Seeliger and KarlSchwarzschild. At present four professors and ten staffastronomers work here. Funding comes from the BavarianGovernment, the German Science Foundation, and otherGerman and European research programs.

Facilities include the Wendelstein Observatory in theBavarian Alps with a 0.8 m telescope. The Observatoryis also a partner in the Hobby–Eberly Telescope Project inTexas.

The Observatory’s mission is to teach astrophysicsat the Ludwig-Maximilians-Universitat in Munich, toresearch astrophysics and instrument development for theVery Large Telescope (ESO), the Hobby–Eberly Telescope(Texas) and the Wendelstein Observatory.

Research specialties are stellar atmospheres andwinds; binary stars; structure, formation and chemicalevolution of galaxies; galaxy clusters; dark matter andgravitational lensing; plasma astrophysics.

For further information seehttp://www.usm.uni-muenchen.de.


University of Crete Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

University of Crete ObservatoryBased in Heraklion. Scientific activity is evenly distributedbetween theory and observation. Observational researchhas been undertaken using a small observatory on MountIda (Skinakas), and international facilities such as Areciboand Effelsberg (radio), the ATO-Australia (optical) andRosat (x-ray). The observational activities cover a widerange of objects including comets, supernova remnants,planetary nebulae, nearby galaxies, the MagellanicClouds, dwarf galaxies, radio pulsars and the Earth’sionosphere and magnetosphere.

For further information seehttp://www.physics.uch.gr/english/.


University of Hawaii Institute for Astronomy E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

University of Hawaii Institute forAstronomyThe University of Hawaii Institute for Astronomy wasfounded in 1967 to manage Haleakala Observatory onMaui, and later, Mauna Kea Observatory, and to carryout its own program of fundamental research. Institutestaff also make extensive use of spacecraft for astronomicalresearch and planetary exploration. The Institute’s mainoffices are located in Manoa valley, just off the University’smain campus on Oahu. A research institute within theUniversity of Hawaii (UH), the IfA has a total staff ofapproximately 200, including about 50 faculty staff.

The Institute is one of the world’s leading astronom-ical research centers. Its broad-based research programincludes studies of the Sun, planets and stars, as well asinterstellar matter, galaxies and cosmology. Most IfA as-tronomers use the giant telescopes atop Mauna Kea andHaleakala. They also use space observatories, such as theHubble Space Telescope and the Chandra X-ray Observa-tory, to make observations that cannot be made from theground. In addition to doing research, some astronomersdesign and build new instruments to measure and anal-yse the radiation collected by the telescopes. Other IfAscientists develop theories which explain the observationsmade by their colleagues.

During the last 30 years, the State of Hawaii hasbecome the most sought-after location in the world for theconstruction of large ground-based telescopes. The focalpoints for this construction are Mauna Kea (4200 m) on theisland of Hawaii and Haleakala (3000 m) on Maui. Theremarkable clarity, dryness and stillness of the air abovethese isolated high-altitude sites led to the commissioningby the University of Hawaii of the Mees Solar Observatoryon Haleakala in 1963 and the UH 2.2 m telescope on MaunaKea in 1970.

Haleakala is lower in altitude than Mauna Kea,yet when compared to other astronomical sites outsideHawaii, Haleakala is an excellent location for optical andinfrared observations of objects in Earth orbit, solar obser-vations, laser ranging and laser guide star experiments,and the whole range of night-time astronomical observa-tions. The IfA’s Mees Solar Observatory on Haleakala iscentral to the Institute’s solar astronomy program. It isset in a naturally fine location for studies of the extremelyfaint radiation emitted by the solar corona and chromo-sphere. At sea level, these outer atmospheric layers of theSun are normally obscured by light scattered by small air-borne particles, such as dust or pollen. Mees is a perfectuse of Haleakala, which means ‘The House of the Sun’.

The Lunar Ranging Experiment (LURE Observatory)operated by IfA for NASA, measures the motions of thePacific Basin and interior of Earth by bouncing laser beamsoff reflectors on satellites. It has also been used to bouncelaser beams off reflectors left on the Moon by the Apolloastronauts.

Institute for Astronomy staff will receive a shareof the observing time on the University of Tokyo 2 m

‘Magnum’ telescope and on the US Air Force 3.7 mAdvanced Electro-Optical Systems (AEOS) Telescope,both of which are located at the Haleakala Observatory.Rocketdyne Technical Services is contracted by the US AirForce to carryout satellite ranging, surveillance and otherdevelopmental activities at the Maui Space SurveillanceSite on land leased from UH at the Haleakala Observatory.Hawaiian students will share the use of the 2 m Faulkestelescope with students in the UK when it is installed inHaleakala Observatory in 2001.

The summit of Mauna Kea on the island of Hawaiihouses the world’s largest astronomical observatory, withtelescopes operated by astronomers from eleven countries.The combined light-gathering power of the telescopes onMauna Kea is fifteen times greater than that of the Palomartelescope in California—for many decades the world’slargest—and fifty times greater than that of the HubbleSpace Telescope.

In addition to developing its own research program,the IfA has provided the scientific impetus for thedevelopment of Mauna Kea into the world’s premiersite for ground-based astronomical observatories. Moremajor telescopes are located on Mauna Kea than on anyother single mountain peak, and Mauna Kea is widelyrecognized as offering better observing conditions foroptical, infrared and mm/sub-millimeter measurementsthan any other developed site.

The success of the 2.2 m telescope led to proposalsfor the construction of several other telescopes on MaunaKea by national and international organizations. Bythe end of the 1970s, there were three new 4 m classtelescopes—the NASA INFRARED TELESCOPE FACILITY (IRTF), theUNITED KINGDOM INFRARED TELESCOPE (UKIRT) and the CANADA–

FRANCE–HAWAII TELESCOPE (CFHT). Two of these telescopesare specifically designed to collect infrared radiation;the dryness of the atmosphere above Mauna Kea isparticularly advantageous at these wavelengths.

The characteristics that make Mauna Kea a pre-eminent site for optical and infrared astronomy also makeit an excellent site for measuring short-wavelength radiowaves. In 1987, two submillimeter-wave observatorieswere completed just below the summit. The CaliforniaInstitute of Technology and the US National ScienceFoundation supplied the expertise and funds for theCALTEC SUBMILLIMETER OBSERVATORY (CSO). The JAMES CLERK

MAXWELL TELESCOPE (JCMT) belongs to a consortium of theUnited Kingdom, Canada and the Netherlands. Morerecently, the Smithsonian Institution, in collaboration withastronomers from Taiwan, has built an array consistingof eight 6 m submillimeter antennas designed to worktogether as a single telescope.

The 1990s saw the construction of a new series ofgiant optical/infrared telescopes on Mauna Kea. Thetwin telescopes of the W M KECK OBSERVATORY are thelargest optical/infrared telescopes in the world. Their10 m diameter mirrors each consist of 36 hexagonswhose positions are controlled by computer. Japan andan international consortium led by the United States


University of Hawaii Institute for Astronomy E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

have built the Subaru and Gemini Northern Telescopes,respectively. Both of these telescopes contain thinmeniscus 8 m primary mirrors, and are designed for bothoptical and infrared astronomy.

At a lower altitude, on the southern flank of MaunaKea, is the Hawaii antenna of the Very Long BaselineArray,which is part of a 5000 mile wide system of ten 25 m radiodishes that work together as the world’s largest dedicatedfull-time astronomical instrument.

As part of the agreement between these organizationsand the University of Hawaii, astronomers at UH areentitled to 10–15% of the observing time on each non-UHtelescope on Mauna Kea, in addition to full use of the UH2.2 m and 0.6 m telescopes.

For further information seehttp://www.ifa.hawaii.edu.


University of Tokyo, Institute of Astronomy E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

University of Tokyo, Institute ofAstronomyFounded in 1987, the Institute of Astronomy, Universityof Tokyo, is located at Ohsawa, Mitaka, Japan, 30 km westof central Tokyo. Its objectives are research and educationin observational astronomy. It has a staff of 3 professors;5 associate professors; 8 research associates, 2 technicalsupport staff, 2 administrative officers and several part-time staff (secretary, catering etc). Its major facilitiesinclude Kiso Observatory (105 cm Schmidt telescope);two 60 cm submillimeter telescopes; 2 m IR telescope (incollaboration with the Physics Department). Researchspecialities: galactic and extragalactic radio astronomy;galactic center; extragalactic optical/IR observations;observational cosmology/formation and evolution ofgalaxies; infrared stellar physics.

The Institute is supported by an annual fund from theUniversity of Tokyo (Ministry of Education) and scienceresearch aids from the Ministry of Education.

Its member organization (in joint ventures) is theUniversity of Tokyo, Graduate School of Sciences

For further information seehttp://www.ioa.s.u-tokyo.ac.jp.


Unsold, Albrecht (1905–95) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Unsold, Albrecht (1905–95)German astrophysicist, became professor at the Universityof Kiel, where he studied stellar atmospheres and the waythat spectral lines were formed and shaped—the effects ofabundances, radiation damping, Doppler shifts, electricfields and atomic collisions. He analysed the spectrum ofthe star Tau Scorpii, which he obtained on a visit to theYerkes and McDonald observatories. This was the firstdetailed analysis of a star other than the Sun and yieldedthe physical conditions in the atmosphere of the star.


Uppsala Astronomical Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Uppsala Astronomical ObservatoryThe Uppsala Astronomical Observatory (UAO) is adepartment of Uppsala University (UU) with research,undergraduate, graduate and public education as its mainobjectives.

UU was founded in 1477, and preserved lecture notesshow that astronomy was taught in the 1480s. Amongwell-known professors at UAO are Anders Celsius (1730–44), Anders Angstrom (1842–58), Gunnar Malmquist(1939–59), Erik Holmberg (1959–75) and Bengt Westerlund(1975–87).

Current research focuses mainly on galaxies, stellarastrophysics and solar system minor bodies and is char-acterized by the combination of theory and observations.Nowadays the observational material comes from largeinternational facilities like ESO (EUROPEAN SOUTHERN OBSER-

VATORY) or space observatories like the HST (Hubble SpaceTelescope).

At UAO, about 70 km north of Stockholm, there isa 36 + 33 cm double refractor from 1893 currently usedmainly for public shows and at Kvistaberg, 50 km south ofUppsala, there is a 135 cm Schmidt telescope from 1964.

For further information seehttp://www.astro.uu.se.


Uranus ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS

With an equatorial diameter of 51 118 km (at the 1 baratmospheric pressure level), Uranus is the third largest ofthe planets in the solar system, slightly larger than themore distant NEPTUNE. However, because of its lowerdensity (1.27 g cm–3 compared with 1.64 g cm–3 forNeptune), Uranus ranks fourth among solar systemplanets in mass at 14.54 terrestrial masses. Its character-istic blue–green appearance is due to a layer of clouds ofmethane ice in its upper atmosphere that is not present inthe atmospheres of JUPITER and SATURN. This cloud layerand the overlying methane gas preferentially absorb thered portion of sunlight, resulting in the distinctive color.Prior to the flyby of Uranus by VOYAGER 2 in 1989, littlewas known about the rotational or other physical char-acteristics of Uranus (given in table 1).

Uranus has a ring system and many satellites. Therings were first detected from observations of stellaroccultations as the narrow rings passed in front of distantstars, causing the observed light to blink out temporarilyas viewed from Earth. Nine narrow rings were detectedin this manner. Voyager 2 detected a tenth narrow ring, abroad diffuse ring closer to the planet, and tenuous dustrings scattered among the rest (see URANUS: RINGS).

Prior to 1986, only five moons of Uranus wereknown. Voyager 2 found an additional 10 (plus one iden-tified in 1999 from Voyager images). Earth-based obser-vations since 1986 have resulted in the finding of anoth-er five confirmed or probable moonlets much more dis-tant from the planet. The satellites, known and suspect-ed, are listed in table 3.

Uranus is the first of the non-naked-eye planets to bediscovered. Observations of the closer planets (Mercurythrough Saturn) date back to antiquity, but Uranus wasdiscovered by William Herschel in March of 1781 using

Uranus


his home-made telescope in the back yard of his home inBath, England (see HERSCHEL FAMILY). Technically, at sixthmagnitude, Uranus is marginally visible with the unaid-ed eye. However, it was indistinguishable from back-ground stars until Herschel’s telescopic observations.

The early observationsIn the 200+ yr of observation since its discovery, therehave been monumental changes in our understanding ofthe planets in general and of Uranus in particular. Thefirst major change was a recognition that the solar systemwas much larger than previously supposed. Uranus wasmore than twice the distance of Saturn from the Sun; itsdiscovery gave great impetus to searches for even moredistant planets. Perturbations to the orbit of Uranus ledto independent predictions by mathematicians JohnCouch Adams in England and Urbain Jean Joseph LeVerrier in France. On the basis of Le Verrier’s predictions,Johann Gottfried Galle and his student Heinrich LouisD’Arrest found Neptune on 23 September 1846, the sameday they received Le Verrier’s prediction.

In the meantime, Herschel continued his observa-tions of Uranus with a telescope, again of his own mak-ing and far superior to those available to otherastronomers of his day. OBERON and TITANIA were discov-ered by him in 1787, less than 6 yr after the discovery ofthe planet. For more than 10 yr, no other astronomerswere even able to confirm his findings. Herschel wasconvinced that other satellites must also exist, but in 40yr of searching he found none, although some historiansclaim that he may have been the first to see UMBRIEL (fourtimes between 1790 and 1801). Credit for the discovery ofUmbriel and ARIEL is generally given to William Lassell,who reported their sighting in 1851 to the RoyalAstronomical Society. MIRANDA was discovered byGerard Peter Kuiper in 1948, using the 82 in reflector ofthe McDonald Observatory in Texas. The same instru-ment was used by Kuiper and co-worker D L Harris todiscover Neptune’s Nereid in 1949.

The orbits of the Urananian satellites provided thefirst evidence of the unusual tilt of the equator of Uranusrelative to its orbital plane. Excluding Miranda and theirregular satellite discovered in 1997 and 1999, all otherUranian satellite orbits are within 0.4° of being coplanarwith Uranus’s equator. The actual tilt of the rotation axisof Uranus is 97.92°, although by InternationalAstronomical Union convention, which specifies as thenorth pole that rotation pole which lies north of the eclip-tic plane, the tilt is more properly designated as 82.08°with retrograde rotation. Much speculation has sur-rounded this unusual rotation of Uranus. One populartheory is that, during its early history, Uranus had a morenormal tilt, but that during the latter stages of formationthe planet was struck by an Earth-sized PLANETESIMAL,resulting in the present extreme tilt. The only other plan-ets to exhibit more than a modest tilt are VENUS andPLUTO, which are 1–3 orders of magnitude smaller inmass.

Table 1. Orbital and physical characteristics of Uranus.

Semimajor axis of the orbit 19.191 AUEccentricity 0.046Inclination over the ecliptic 0.77°Sidereal period 83.7474 yrMean orbital velocity 6.83 km s–1

Equatorial diameter 51 118 kmEquatorial diameter relative to Earth 4.007Polar diameter 49 946 kmFlattening 0.023Mass relative to Earth 14.54Mean density 1.27 g cm–3

Surface gravity (at P=1 bar) 8.86 m s–2

Surface gravity relative to Earth 0.906Escape velocity 21.28 km s–1

Sidereal rotation period 17.24 hInclination of equator to orbital plane 97.92¬∞Albedo (visual geometric) 0.57Albedo (bolometric Bond) 0.30Main atmospheric composition H2 (83%), He (15%),

CH4 (2%)


It was the low reflectivity of Uranus at the red end ofthe spectrum which led spectroscopist R Wildt to proposein 1932 the presence of methane (CH4) in the atmospheresof each of the gas giant planets of the solar system. Thelow densities of these giant planets had also led to theconclusion that the primary atmospheric constituent washydrogen (H2), but confirmation of that fact came fromKiess et al. in 1960. Formal detection of helium (He) in theatmosphere of Uranus first came from Voyager 2 in 1986.All other gaseous components of Uranus’s atmosphere(above the cloud tops) constitute a combined abundanceof much less than 1%.

By analogy with Jupiter, each of the other gas giantplanets in the solar system was expected to have anintrinsic magnetic field. However, the strength and ori-entation of those fields and the associated radio emis-sions were not determined until Pioneer and/or Voyagerspacecraft flew close to these planets. It was the monitor-ing of these radio emissions that enabled Voyager scien-tists to determine the rotation periods of Saturn, Uranusand Neptune (see also URANUS AND NEPTUNE: ATMOS-PHERES, IONOSPHERES AND MAGNETOSPHERES).

Space exploration The only space vehicle ever to fly close to Uranus wasVoyager 2. This hardy spacecraft was launched in August1977, flew by Jupiter in July 1979, Saturn in August 1981,Uranus in January 1986 and Neptune in August 1989. Asof late 1999, Voyager 2 continued to send back usefuldata about the particle and magnetic field environmentof the outer solar system. Voyager 2 provided high-resolution images of Uranus’s atmosphere, the ring sys-tem and 16 of the 21 known satellites. In fact, 11 of the 16moons it imaged were discovered by Voyager 2. Voyager2 also determined the detailed composition of the atmos-phere, confirmed the presence of a highly tilted and off-set magnetic field and measured the rotation period ofthe planet’s interior. In fact, more than 90% of all weknow about Uranus came from the measurements madeby Voyager 2 and the subsequent interpretation of thosemeasurements.

No other space missions to Uranus are at presentplanned and funded. Nevertheless, continued improve-ments in Earth-based telescopes and techniques are nowproviding updates to some of the Voyager data, especial-ly in the area of atmospheric studies (as well as in the dis-covery of the new satellites mentioned earlier). The pre-sent emphasis on smaller, faster, less expensive missionsdoes not bode well for missions to Uranus or Neptune,each of which requires long flight times and concomitanthigher space flight operations budgets.

Composition and structure Like its larger counterparts (Jupiter, Saturn andNeptune), Uranus is composed primarily of the light ele-ments hydrogen and helium. Even the somewhat heaviercompounds, confirmed methane (CH4) and suspectedammonia (NH3) and water (H2O), are composed largely

of hydrogen. This dominance of light elements is reflect-ed in the low densities of the gas giant planets and in theabsence of distinct liquid or solid surfaces in the interiorsof these planets. However, in contrast to Jupiter andSaturn, methane, ammonia, water and the heavier corematerials are thought to constitute a larger total massthan the overlying hydrogen and helium. These mea-sured or deduced characteristics are thought to be theconsequence of the formation process by which theseplanets came to be. There are a number of competing the-ories about the nature of that formation process. Sinceeach theory, if sufficiently developed, predicts outcomesthat are generally dependent on the starting conditions,the measured or deduced characteristics can becomeboth an effective discriminator between theories and ameans of determining the starting conditions.

All formation theories have certain characteristics incommon. For example, each assumes a decrease in tem-perature in the presolar nebula as a function of increas-ing solar distance. That temperature gradient in turndetermines which compounds are solid at a given dis-tance from the Sun and which are not. With more mater-ial from which to coalesce, it is likely that the planetarycores which formed from the coalescing solids were larg-er at the distances of the gas giants than for the Earth-likeplanets, enabling the gas giants to capture more massiveatmospheres. Uranus and Neptune may be smaller thanJupiter and Saturn as a consequence of the solar nebulabeing less dense at these greater distances from the Sun.Perhaps the growth of Uranus and Neptune thereforelagged behind that of Jupiter and Saturn. If the Sun wentthrough its T-Tauri stage when atmospheric accretionwas more or less complete for Jupiter and Saturn, butincomplete for Uranus and Neptune, most of the remain-ing source material may have been swept away, and fur-ther growth of the planets would have essentiallystopped. In that scenario, the cores of Uranus andNeptune would have occupied a much larger percentageof their interiors than would have been the case forJupiter and Saturn. Jupiter and Saturn were thereforeexpected to more closely represent the mix of materials inthe original solar nebula (and in the present-day solaratmosphere).

Actual measurements confirm the predicted higherrelative abundance of methane in the atmosphere ofUranus than that of Jupiter, Saturn or the Sun. Thesenumbers imply an enrichment of carbon (relative tohydrogen) in Uranus’s atmosphere that is about 24 timesas high as the C/H ratio in the Sun. Carbon enrichmentwas also noted for Jupiter (Galileo probe measurements)and estimated for Saturn, although to a lesser extent thanfor Uranus (as predicted). As in the other gas giants, pho-tochemistry of methane in the upper atmosphere pro-duces a suite of hydrocarbons, but only acetylene (C2H2)has been detected thus far. Water (H2O) has also beendetected in the upper atmosphere (Feuchtgruber et al.1997); its presence reveals an exogenic source of thewater, presumably micrometeoroid ablation.



Helium abundance relative to hydrogen is also high-er for Uranus (and Neptune) than for Jupiter and Saturn.Strangely, the helium abundances for Uranus andNeptune coincide (within the uncertainties of the mea-surements) with that measured for the Sun. Jupiter’satmosphere is moderately depleted in helium relative tothe Sun; Saturn’s is strongly depleted, having only ascant 20% of the solar He/H abundance. At very highpressures (>3 million bar) in the interiors of Jupiter andSaturn, hydrogen is transformed into a liquid metallicstate. It is possible, especially in the case of Saturn, thathigher-density liquid helium droplets slowly sinkthrough the liquid hydrogen owing to the force of gravi-ty, thus depleting the source of helium to the visibleatmosphere. No such mechanism can operate in the inte-riors of the smaller Uranus and Neptune.

The vertical temperature profile of Uranus below the600 mbar level increases monotonically with depth at anadiabatic lapse rate. Voyager radio science occultationmeasurements detected the presence of a methane clouddeck near the 1 bar pressure level, where the temperatureis near 80 K. This value is colder than the minimum tem-peratures in the atmospheres of Jupiter and Saturn; it isfor that reason that no methane clouds are present intheir atmospheres. Uranus’s tropopause is located near100 mbar, where the temperature is near the minimumvalue of about 50 K.

Extrapolation of measured atmospheric tempera-tures into the interior of Uranus at an adiabatic gradientwould lead to a central temperature between 6000 and 10000 K. Unlike the other three gas giant planets, Uranusappears to have little or no heat escaping from its interi-or to warm the upper regions of the planet. Whether thatis a consequence of its extreme axial tilt or is due to insu-lating layering below the visible cloud deck is not knownat present. Interior modeling yields central densities ofbetween 4.5 and 9 g cm–3, depending on whether or notthere is a separate molten rocky core out to a distance ofabout 5000 km from the center. Such a model might havethe larger density values at its center, whereas a relative-ly well-mixed massive core of molten rocky and icymaterial extending through most of the interior wouldhave the lower central density value. The measured sec-ond-order gravity harmonic (J2=0.003 343), combinedwith a rotation period of 17.24 h and an oblateness (polarflattening) of 0.023, tends to favor the well-mixed core.

DynamicsWith little or no discernible internal heat source, Uranusmight be expected to have little atmospheric turbulence.Figure 1 depicts the planet Uranus as seen by Voyager 2.The left-hand view is true color; the right-hand view is ahighly color and contrast-stretched version of the sameimage. It is apparent that essentially all atmospheric fea-tures are symmetric around the rotation axis whose posi-tive pole is just left of and slightly below the center of thedisk in these images. Zonal wind velocities were measur-able for only a few latitudes on the planet, mainly because

of the paucity of non-axisymmetric cloud features.Furthermore, only the illuminated southern hemispherewas visible. The data are best fitted with a latitudinalwind velocity in m s–1 of 200 (0.4cosb–cos3b), where–90°<b<0° is the latitude. This formula predicts a maxi-mum prograde wind speed of about 240 m s–1 near –60°latitude, zero wind speed near –23° latitude and 120 m s–1

retrograde wind speed near the equator.More recent images from the Hubble Space

Telescope (HST) show greatly increased atmosphericactivity for Uranus than was seen during the Voyager 2flyby. Two color-stretched images, each including ringsand satellites, are displayed in figure 2. The increase inatmospheric activity between 1986 and the 1997 HSTimages may be due in part to seasonal changes. Duringthe Voyager 2 flyby, Uranus was near its southern sum-


Figure 1. Uranus in true color (left) and contrast-enhanced falsecolor (right). The planet’s south (positive rotation) pole is leftand below the center of the disk, at the center of the concentricpatterns. Image from Voyager 2 courtesy of NASA JetPropulsion Laboratory.

Figure 2. Uranus as seen from the HST’s Wide Field PlanetaryCamera in 1997. The image on the right was taken later andshows the motion of satellites and atmospheric spots in theintervening time. The bright portion of the ring is the unre-solved epsilon (ε) ring; the dimmer portion is a composite ofseveral of the rings. Image from HST is courtesy of NASA JetPropulsion Laboratory.


mer solstice (beginning of summer). In 2008 Uranus willreach an equinox, and all latitudes will be illuminated bysunlight. If the changes are due to seasonal effects,atmospheric cloud activity may be expected to increaseuntil 2008 or beyond.

The temperature in the troposphere does not exhibitany north–south asymmetry despite the peculiar insola-tion pattern. The horizontal thermal structure shows lit-tle contrast other than shallow minima at mid-latitudes.The observed latitudinal dependence indicates that thezonal winds decay with height. Dynamical models sug-gest that the mid-latitude cool regions result from adia-batic cooling associated with upwelling in the tropos-phere and that poleward and equatorward flow occurs atupper levels.

Uranus’s magnetosphereThe magnetic field of Uranus was first detected about 8 hbefore Voyager 2’s closest approach to Uranus.Amazingly, the best dipole model of that magnetosphereindicates a magnetic dipole tilt of 58.6° and an offset fromthe center of the planet of 1/3 the radius of the planet inthe direction of the unilluminated north pole. The hightilt and offset may indicate that the magnetic field is gen-erated by electrical currents at fairly shallow depthswithin the liquid icy–rocky core. The dipole moment ismore than a factor of 50 greater than the dipole momentof Earth’s magnetic field, but, because of the larger sizeof Uranus, typical surface magnetic fields are of compa-rable magnitude for the two planets. The magnetic fieldof Uranus, like those of other planets in the solar system,is highly distorted by the solar wind. Its typical sunwardextent is about 18 Uranian radii; the magnetic tail proba-bly extends to thousands of radii downwind.

Ultraviolet auroral emissions were observed by

Voyager 2 near the 44.2° north latitude magnetic pole.The auroral region was 15°–20° in diameter and isbelieved to be excited by electrons with energies of about10 000 eV. A variety of radio emissions originating with-in the magnetosphere were detected. It was variations inthe narrowband emissions near 60 kHz that enabledVoyager to determine that Uranus rotates once every17.24 h. Detected charged particle species within themagnetosphere included mainly electrons and protons,although small amounts of ionized hydrogen gas (H2

+)may have been present. It seems evident that the prima-ry source of these particles is the extended hydrogenatmosphere of Uranus, which has a temperature of about800 K at altitudes of 5000–7000 km above the cloud tops.

Rings Uranus has 10 narrow rings with sufficient optical depthto block out starlight as they passed between a star andthe Voyager spacecraft or groundbased observers. By farthe densest and widest of these rings is the ε ring, which


Table 2. Uranus’s rings’ orbital data.

Ring Semimajor Orbit Inclination Periodname axis (km) eccentricity (°) (h)

6 41 837 0.001 01 0.062 6.19885 42 235 0.001 90 0.054 6.28754 42 571 0.001 06 0.032 6.3628α 44 718 0.000 76 0.015 6.8508β 45 661 0.000 44 0.005 7.0688η 47 176 0.000 00 0.001 7.4239γ 47 627 0.000 11 0.002 7.5307δ 48 300 0.000 00 0.001 7.6911λ 50 024 ~0.0 ~0.0 8.1069ε 51 149 0.007 94 0.000 8.3823

Table 3. The satellites of Uranus.

Name Distance (km) Period (days) Incl (°) Eccentricity Diameter (km) Density (g cm–3)

Cordelia 49 792 0.3350 0.08 0.000 ~26 ?Ophelia 53 764 0.3764 0.10 0.001 ~32 ?Bianca 59 165 0.4346 0.19 0.001 ~44 ?Cressida 61 767 0.4636 0.01 0.000 ~66 ?Desdemona 62 659 0.4737 0.11 0.000 ~58 ?Juliet 64 358 0.4931 0.07 0.001 ~84 ?Portia 66 097 0.5152 0.06 0.000 ~110 ?Rosalind 69 927 0.5585 0.28 0.000 ~54 ?Belinda 75 255 0.6235 0.03 0.000 ~68 ?1986U10 76 500 0.637 ? ? ~40 ?Puck 86 006 0.7618 0.32 0.000 154 ?Miranda 129 847 1.4135 4.22 0.027 472 1.15Ariel 190 929 2.5204 0.31 0.0034 1158 1.56Umbriel 265 979 4.1442 0.36 0.0050 1170 1.52Titania 436 273 8.7059 0.14 0.0022 1578 1.70Oberon 583 421 13.4632 0.10 0.0008 1522 1.64Caliban 7 168 900 579 139.68 0.082 ~44 ?Sycorax 12 213 600 1289 152.67 0.509 ~40 ?1999U1 ? ? ? ? ? ?1999U2 ? ? ? ? ? ?1999U3 ? ? ? ? ? ?


varies in width from 20 to 95 km as its particles circle theplanet. The other rings have typical widths of 1–10 km.The ‘equivalent depth’ of the ε ring (and most of theother variable-width rings) is constant; in other words,the same number of ring particles pass through the nar-rower portions of each ring during each orbit as throughthe wider portions of the same ring. Their narrow ringwidths are the result of gravitational interactions withnearby satellites, which act as ‘shepherds’ to keep thering material from spreading. The orbital characteristicsof the narrow rings are given in table 2.

In addition to the 10 narrow rings, Uranus has abroad, diffuse and optically thin ring component seen insome of the Voyager 2 images. It is interior to the otherrings, extending from a radius of about 37 000 to about 39500 km from the center of the planet. During passagethrough the shadow of Uranus, Voyager 2 also snappedan image of the rings that is unlike any other taken atUranus. In this unique geometry, forward-scattered sun-light brings out the optically thin dust bands that occupymuch of the area between the ten narrow rings men-tioned above. There is a lot of radial structure in thesedust rings (see figure 3), little of it understood at present.

SatellitesPrior to 1999, Uranus had 17 known moons. The discov-ery of the five largest (Miranda, Ariel, Umbriel, Titaniaand Oberon) was discussed in the introductory material

earlier in this article. Ten more (Cordelia, Ophelia,Bianca, Cressida, Desdemona, Juliet, Portia, Rosalind,Belinda and Puck) were discovered by Voyager in 1986.All of the moons discovered by Voyager orbit Uranuscloser to the planet than the five ‘classical’ satellites. In1997 Caliban was discovered by Brett Gladman and hisassociates. The same year Sycorax was discovered byPhil Nicholson and his associates. Caliban and Sycoraxare more than 10 times as far from the planet as any ofthe previously known ‘regular’ satellites. Their orbits areretrograde, highly inclined and more eccentric than istrue for the regular satellites.

In 1999, Erich Karkoschka found a previously undis-covered moon in images shuttered by Voyager 2 in 1986.This moon, which has an orbit almost identical to that ofBelinda, was found in seven different Voyager 2 images.Based on a preliminary orbit, it is estimated that Belindalaps (passes) 1986U10 about once a month, a highlyunusual circumstance.

Three more probable satellites of Uranus were sight-ed in 1999 by Gladman, J J Kavelaars and others. Orbitsfor these satellites are not yet well-determined, but theyare probably all distant, irregular satellites. Their tempo-rary designations are 1999U1, 1999U2 and 1999U3. In2002 the discovery of another satellite was confirmed.Named S/2001 U 1, this brings the total number of con-firmed Uranian moons to 21. S/2001 U 1 and five otherslike it have very irregular, eccentric orbits that do notshare the same orbital plane as the larger moons ofUranus. Ranging in size from 10 to 20 km (about 6 to 12mi), these moons are thought to be remnants of ancientcollisions that occurred at the early stage of planetaryformation.

Bibliography The Voyager 2 Encounter with Uranus 1987 J. Geophys.

Res. 92 14 873–15 375Bergstralh J T, Miner E D and Matthews M S (ed) 1990

Uranus (Tucson, AZ: University of Arizona Press)Feuchtgruber et al. 1997 Nature 389 159–62Miner E D 1998 Uranus: the Planet, Rings and Satellites 2nd

edn (Chichester: Wiley)Ellis D Miner


Figure 3. This long-exposure image from Voyager 2 was takenof the back side of the rings of Uranus during the spacecraftpassage through the rings. The image was taken during rotationof the spacecraft and is therefore motion compensated at the leftedge of the image, but the right side and the background starsare smeared. The outermost ring is the ε ring, but most of theother features seen are not identifiable with the narrow, optical-ly thicker rings normally seen in backscattered light.

Uranus and Neptune: Atmospheres, Ionospheres, andMagnetospheres


Uranus and Neptune: Atmospheres,Ionospheres, and MagnetospheresNearly the entire mass of URANUS and NEPTUNE consistsof gas, much like the other gas giant planets Jupiterand Saturn. Therefore Neptune and Uranus might beregarded as essentially all atmosphere, although it isconventional to limit the definition to the outermost layers.In a tenuous outer layer of the atmosphere we find theionosphere (see PLANETARY IONOSPHERES), where ultravioletlight and other radiations knock electrons loose fromatmospheric molecules so that electrical currents can flowthere. The magnetosphere begins even further out, wherethe atmosphere becomes so thin that collisions betweenits molecules are rare, and so they fly freely like orbitingsatellites. In this region, the largest forces on electronsand IONS (electrically charged atoms or molecules) comefrom electric and magnetic fields, and so their motionsare organized by the planetary magnetic field instead ofplanetary gravity (see PLANETARY MAGNETOSPHERES).

Most of what we know about Uranus’s andNeptune’s atmosphere and everything we know abouttheir ionospheres and magnetospheres were learned fromdata collected during the close passes of VOYAGER 2 nearUranus in January 1986 and near Neptune in August1989. Unfortunately this also means that it will be difficultto learn much more without more dedicated spacecraftencounters. As a result, our mental picture of theseplanets, especially Uranus, is dominated by their stateduring their Voyager flyby, and will become increasinglyunrealistic as time progresses.

AtmosphereVisually, Uranus and Neptune are blue planets, chieflybecause of absorption of red light by atmospheric methane(CH4). Uranus is virtually featureless, while Neptunehas large but short-lived cyclonic storms that appear asspots. By tracking cloud features, strong east–west windsvarying with latitude were found (especially at Neptune),indicating the presence of weather.

As tables 1 and 2 show, Neptune and Uranus arenearly twin planets about four times the size of the Earthand at much greater distance from the Sun. During theformation of the solar system, the material coalescing intothe planets was extremely cold because of its distancefrom the Sun, and so most common gases (such aswater, carbon monoxide and dioxide, ammonia andmethane) were frozen. Because solid material moreeasily accumulates into planets than does gas, Uranusand Neptune incorporated much more of this materialthan did planets like the Earth, that formed in warmerregions. Once these frozen gases were incorporated intoUranus’s and Neptune’s warm interiors, they becamegaseous again, but nevertheless were still bound bygravity. Because they were able to accumulate thesemore common materials, Uranus and Neptune grew muchbigger than the terrestrial planets, and thus were massive

enough to attract substantial amounts of the still-gaseoushydrogen and helium.

Thus by number of atoms, Uranus’s and Neptune’satmospheres are mostly hydrogen with some helium andheavier gases, like the other gas giant planets and the Sun.Compared with the Sun, Jupiter and Saturn, however,Neptune and Uranus are highly enriched in oxygen,carbon and other elements heavier than hydrogen andhelium. As a result, these heavy elements constituteat least 80% of Uranus and Neptune by mass. Inthese planets’ atmospheres, carbon mostly combines withhydrogen to form methane, which is slowly broken downby solar ultraviolet light to form smog, just as on Earth (seePLANETARY ATMOSPHERES, EARTH’S ATMOSPHERE). Oxygen mainlyjoins with hydrogen to form water, which freezes out,leaving a dry upper atmosphere just as at Earth. Bothof these condensible materials fall into the interior, wherethey are recycled back into gaseous water and methane.

Table 1. Useful Uranian constants, in units compared with Earth(⊕) and SI metric units.

Quantity Earth units SI units

Distance from Sun 19.18⊕ 2.870× 1012 mRadius 4.01⊕ 2.556× 107 mMass 14.54⊕ 8.683× 1025 kgOrbital period 84.01⊕ 2.651× 109 sRotation period 0.720⊕ 6.206× 104 s

Table 2. Useful Neptunian constants, in units compared to Earth(⊕) and SI metric units.

Quantity Earth units SI units

Distance from sun 30.06⊕ 4.497 × 1012 mRadius 3.89⊕ 2.476× 107 mMass 17.23⊕ 1.024× 1026 kgOrbital period 164.8⊕ 5.200× 109 sRotation period 0.673⊕ 5.800× 104 s

Uranus One of Uranus’s most remarkable traits is thedirection of its rotation axis. Most planets’ orbits lie inapproximately the same plane, and their rotation axes (andthe Sun’s) are roughly perpendicular to this plane. NotUranus; its rotational axis lies nearly in its orbital plane.It is believed that this great tilt was caused by a near-catastrophic collision at the time when Uranus was stillforming out of infalling bodies, one of which must havebeen comparable in size with Uranus itself.

This ‘tipped-over’ orientation means that, part of thetime, the Sun will be nearly overhead at Uranus’s northor south pole. Because the rotational axis stays fixedin direction while Uranus orbits around the Sun, 1/4of an orbit later the Sun will be over Uranus’s equator.Therefore the differences between the solar heating duringUranus’s summer and winter are much more extreme thanat Earth, in addition to being much colder overall because




of Uranus’s much greater distance from the sun. Despitethese extremes of heat input, the temperature extremesare small, apparently because the atmosphere can equalizetemperature differences by transporting heat.

Another anomalous trait of Uranus is its low rate ofinternal heat release. Heat slowly leaks out of the interiorsof the other gas giant planets, at rates comparable withthe planets’ absorption of sunlight. However, Uranus’internal heat loss is undetectable.

Neptune Despite Neptune’s greater distance, its temper-ature at its apparent surface is about the same as Uranus’s.The reason for this is the heat release from Neptune’s inte-rior, which makes up for its relative shortfall of sunlight.In this aspect Neptune is a more typical gas giant planetthan Uranus. Neptune’s rotational axis is also more typ-ical, in being roughly perpendicular to its orbit plane, soNeptune’s seasonal variation of solar heating is much lessextreme than Uranus’s.

IonosphereUranus’s and Neptune’s ionospheres are currently poorlyunderstood, because of the limitations of the observationsand gaps in theoretical knowledge. However, it appearsthat ionization due to solar ultraviolet and galactic cosmicrays is mostly responsible for maintaining the ionosphere,with enhancements from meteoritic impacts.

MagnetosphereUranus and Neptune have planetary magnetic fieldswhose surface intensity is the same order of magnitudeas the Earth’s field, but whose distribution is completelydifferent. At Earth the field is oriented roughly north–south and its strength is roughly the same everywhere.Neptune’s and Uranus’s fields are much more variable instrength and direction.

Like Earth, Neptune and Uranus have AURORAS,which are glows created by electrons hitting the upperatmosphere where the magnetic field lines converge. AtEarth auroras occur near the north and south geographicpoles, but at Uranus and Neptune the magnetic poles lie atmiddle latitudes so that is where their auroras occur also.

Uranus The magnetosphere of Uranus is tilted once byUranus’s rotation axis and once again by the magneticfield’s tilt with respect to the rotation axis. The chargedparticles in Jupiter’s and Saturn’s magnetospheres mostlyrotate with their planets. At Earth charged particles flowin a kind of back-eddy opposite to the direction of the solarwind. At Uranus they do both at the same time, althoughthis may change as Uranus’s orbit carries it from Sun-over-pole to Sun-over-equator orientation.

Uranus’s magnetosphere contains mostly atomichydrogen ions (H+), which probably come from Uranus’atmosphere.

Neptune The magnetic field of Neptune is also drasticallytilted and offset with respect to the rotation axis. DespiteNeptune’s normal rotation axis orientation, chargedparticle trajectories are even more complicated herethan at Uranus. Many of these trajectories exit themagnetosphere, so the density of ions in Neptune’smagnetosphere is very sparse.

Neptune’s magnetosphere contains mostly atomichydrogen ions (H+) and atomic nitrogen ions (N+).These ions are probably derived from the atmosphere ofTRITON, Neptune’s large, icy moon. Triton’s atmosphereis mostly molecular nitrogen with traces of hydrogen-bearing species such as methane.

Lower atmosphere and interiorBecause Uranus and Neptune are gas giants, thedistinction between atmosphere and interior is arbitrary.By convention the boundary between them is placedat a level where the pressure is 10–100 bar (1 bar =105 N m−2 is about the pressure at the bottom of theEarth’s atmosphere). This definition is based partlyon observability and partly on the diminished effect ofweather at that level.

Also by convention, the planetary radius is definedas the distance from the planetary center out to the levelwhere the pressure is 1 bar at the equator. These areabbreviated as RU = 25 559 km and RN = 24 764 km, forUranus and Neptune respectively.

Most of what is known of Uranus’s and Neptune’sinteriors is derived from the slight departure of theirgravity fields from spherical symmetry, detected byperturbations of Voyager’s flyby trajectory. From theseslight perturbations has been deduced an approximatedensity versus radius profile, which suggests the presenceof rocky cores roughly half the size of Earth, representingabout 0.1–1% of Neptune’s or Uranus’s total volume.

The properties of Uranus’s atmosphere are summa-rized in figure 1, while Neptune’s is summarized infigure 2. The number of molecules per cubic centimeter(number density) of each of the more important species isprofiled versus altitude and pressure. The H2 density iscontrolled by the temperature profile via the hydrostaticequation,

dpdr= −mpg(r)

kT (p)

with p is the pressure, m the mass of an H2 molecule, kBoltzmann’s constant, g(r) the acceleration of gravity asa function of altitude and T (p) the temperature versuspressure profile. The densities of the other species arecontrolled by their interactions with the dominant H2.The temperature profile is controlled by the balanceof heat gain and loss, which is in turn controlled bythe mechanisms by which heat is transported. Thedifferent temperatures seen at different levels result fromdifferent heat sources and transport modes dominating atdifferent densities.

Both planets’ tropospheres (from words meaning theregion of changes; the lower atmosphere, where there




Figure 1. Profiles of temperature and density versus altitude inUranus’s atmosphere. The temperature (T ) is read from the topscale, the densities (labeled by chemical symbol) from thebottom. Important boundary altitudes are marked on theprofiles that determine them.

Figure 2. Profiles of temperature and density versus altitude inNeptune’s atmosphere. The temperature (T ) is read from thetop scale, the densities (labeled by chemical symbol) from thebottom. The temperature profile is an approximationconstructed to match the observations; its sharp bends are notreal. Important boundary altitudes are marked on the profilesthat determine them.

is weather) are very cold compared with Earth. Withinthe troposphere the dominant heat transport mode isconvection, whereby heat is carried along with the gas as itrises and falls. Because the pressure changes with altitude,the gas is cooled by expansion as it rises and heated bycompression as it sinks. Thus in the troposphere thetemperature variation with altitude follows the adiabaticrate of change, that is, the rate at which a column of risingair would cool just as a result of expansion without heatexchange.

At Earth the vertical temperature profile is compli-cated by the condensation of water, but, in the colder Ura-nian troposphere, water condenses at a deeper level. AtNeptune and Uranus the visible clouds are probably con-densed methane. These clouds are located near the levelof the sharp bend in the CH4 profile just below the 1 barlevel in figures 1 and 2. Above that level the ratio of CH4

to H2 density abruptly decreases because of methane con-densation.

At the tropopause (the upper boundary of thetroposphere, where the temperature stops falling withaltitude), the temperature is about 50–52 K (much colderthan liquid nitrogen) at both planets. The altitude of thistransition point is marked on the temperature profilesdisplayed in figures 1 and 2. At this altitude the methanevapor pressure (at a given temperature, the pressure ofvapor in equilibrium with its solid form; usually extremelytemperature dependent) is at a minimum. Any methaneabundance in excess of this pressure is frozen out and theresulting ice falls to lower altitude, which is why the CH4

profiles in figures 1 and 2 show abrupt decreases there.

UranusThe temperatures within Uranus’s interior are probablyhigh, but unknown. Heat constantly leaks out of the othergas giant planet interiors, and from the rate of heat loss atemperature versus radius profile necessary to maintainthat loss can be calculated for those planets. Uranus’sinternal heat loss, by contrast, is too small to have beenmeasured, so its internal temperature is unknown.

It may be that Uranus’s interior is actually cooler thanthe other gas giants’, perhaps because the collision thattipped Uranus over also stirred it enough to let its excessheat out. Another possibility is that the settling of densermaterials has stabilized the interior so that hot gas does notrise and carry out heat, because even if the lower regionsare hotter they would be compositionally denser and sonot buoyant.

NeptuneUnlike Uranus, Neptune does have a measurable rate ofheat loss, which is about 1.6 times as large as the rate ofheat absorption from sunlight. From this heat loss rate, atemperature versus radius profile can be calculated, whichis very near to an adiabat. The temperatures within theinterior are high, reaching a maximum of probably about6000 K at the center.

Upper atmosphereAbove the tropopause in each planet is the stratosphere,where the atmosphere is relatively stable (hence the name)because the temperature rises with altitude. Atmosphericstirring does not completely cease here, but it is greatlyattenuated compared with the troposphere. In thelower stratosphere, the temperature profile is mainlycontrolled by the absorption and reradiation of heatabsorbed from sunlight. The upper atmosphere isheated by a high-altitude energy source of unknown




nature. Because the energy is input at such low densities,no known component of sunlight or bombardment bymagnetospheric charged particles can be responsible.

This heat is not reradiated from where it arrives,because H2 and He are very poor radiators of heatand the more effective radiators, the hydrocarbons, arescarce here. Thus thermal conduction transports the heatdown to the level where it can be radiated into spaceby the hydrocarbons. In order for conduction to carrythe heat this far, the heat-source region’s temperaturemust rise to high temperatures. From there up to theexobase (the atmospheric level above which atomic andmolecular collisions are negligible), the atmosphere isroughly isothermal.

Because of the factor ofm in the hydrostatic equation,atmosphere pressure diminishes with altitude at a rateproportional to m, the atmospheric molecular mass.When an atmosphere is a mixture of gases, atmosphericturbulence and other stirring mechanisms tend to keepthe different gases mixed in equal proportions. Thus allgas densities decrease with altitude at the same rate, anaverage value which, at Uranus and Neptune, is aboutequal to the rate for H2. However, above a certain altitude,where the mixing is not vigorous enough to keep theproportions uniform, each gas density decreases withaltitude at its own individual rate. The point wherethis separation of rates begins is called the homopause,which is marked on figures 1 and 2. Notice the change inslope of the He and CH4 profiles there, above which eachconstituent is decoupled from the others and so follows itsown independent profile.

It is because of the rapid falloff of the CH4 densitycompared with the lighter elements that the hydrocarbonsare confined to lower altitudes, to which heat absorbedat high altitudes must be conducted before it can beradiated into space. Thus the level of the homopause helpsdetermine the maximum temperature of the uppermostregion of the atmosphere.

UranusThe stratosphere of Uranus has been discovered to beunusually calm compared with the other giant planets,so that heavier gases such as helium and methane arenot well-mixed with the hydrogen at higher altitudes. Inother words, the homopause lies at lower altitude andhigher pressure, about 10−4 bar in figure 1. Because ofthe low methane abundance at the higher altitudes wheresolar extreme ultraviolet light can penetrate and drivephotochemistry, there are also lower abundances of thehigher hydrocarbons such as ethane and acetylene there.Therefore there are fewer photochemically producedaerosol particles, or smog (an aerosol is a particle so smallthat it settles out of the air very slowly). As a result,Uranus’s stratosphere is unusually clear for a gas giant,so that Rayleigh scattering (light scattering caused byspontaneous microscopic fluctuations of the air’s density)is the dominant atmospheric opacity at wavelengths awayfrom methane absorptions. At Earth, when opacity from

water aerosols (clouds) is negligible, Rayleigh scatteringis also important, turning the sky blue and sunsets redbecause of its λ−4 wavelength dependence.

Because the hydrocarbons lie so deep in thestratosphere, heat conducted from its source around10−11 bar must travel a great distance. In order forconduction to carry the heat this far, the heat-sourceregion’s temperature must rise to about 850 K.

This high temperature inflates the upper atmospheregreatly, particularly the light atomic hydrogen component(a minor component at lower altitudes), so that it fallsoff slowly with altitude. Therefore even at the greataltitudes (1.64RU ≤ r ≤ 2RU ) of Uranus’s rings, whichare basically icy orbiting pebbles, the gas density is highenough to produce frictional drag. As with artificial Earthsatellites, atmospheric drag causes Uranus’s ring particlesto gradually lose their orbit-maintaining momentumuntil eventually they fall vertically into the troposphere.Because the ratio of gas drag to particle inertia is largest forthe smallest ring particles, Uranus’s inflated atmosphereselectively removes the finest ring particles. The steadyshrinkage of ring-particle orbits, piling up against therepulsive perturbations of larger orbiters with slowerorbital decay, may be part of the reason that Uranus’s ringsare a series of very thin concentric bands, in contrast withthe nearly continuous disk that constitutes Saturn’s rings.

NeptuneThe upper atmosphere of Neptune is much morevigorously mixed than is Uranus’s, as shown in figure 2 byits higher homopause altitude (at around 10−6 bar insteadof 10−4 bar as at Uranus). For this reason methane andother hydrocarbons are found at significant densities muchhigher in the atmosphere at Neptune, as is also evident infigure 2. Neptune’s degree of atmospheric mixing is moretypical of the other giant planets, Jupiter and Saturn, thanis Uranus’s.

Probably because the hydrocarbons reach higherin the atmosphere, the heat flow from Neptune’s highaltitude heat source (not well defined, but probably at nohigher an altitude in pressure units than at Uranus) hasa shorter distance to travel before it can be radiated intospace. Therefore the hot uppermost regions of Neptune’satmosphere, at about 550–650 K, do not reach as high atemperature (about 850 K) as Uranus’s does.

IonosphereIonospheres are formed by ionizing radiation andrepresent a balance between the ion production rateand the recombination loss rate. Because both therecombination loss rate and the atmospheric opacityto ionizing radiation are proportional to gas density,which increases rapidly with depth in the atmosphere,ionospheres typically have sharp lower boundaries. Theupper part of the ionosphere, however, typically decreasesupwards at an exponential rate determined by the localatmospheric temperature. This rate has been measured for




Neptune’s and Uranus’s ionospheres and is about right forH+ ions.

In addition to this smooth variation, thin layers ofhigher electron density were observed in Uranus’s andNeptune’s ionospheres. It has been conjectured thatthese are transitory effects of meteor showers, which areobserved to create ionization in Earth’s atmosphere. Thesethin layers are at the bottoms of the two ionospheres, whichlie at about 1000–2000 km above 1 bar at Uranus and about500–1000 km at Neptune.

MagnetosphereThe magnetosphere is the region outside a planet’sexobase, where the planetary magnetic field is strongenough to deflect the solar wind. On the upwind side,the pressure balance between field and wind creates aroughly parabolic boundary, called the magnetopause,with nose distance about 20RU upstream at Uranus and25RN at Neptune. This boundary envelops its associatedplanet and extends downstream in a wake many tensof planetary radii long. Outside of this boundary is aroughly hyperboloidal bow shock standing in the highlysupersonic solar wind.

The inner boundary of the magnetosphere iseffectively its planet’s exobase, below which collisionswith atmospheric molecules control charged-particlemotion. However, atmosphere–magnetosphere couplingcan extend more deeply than the exobase, as magnetic-field-aligned currents can flow into the ionosphere,across to other field lines and back out into themagnetosphere. It is this coupling that drives muchcollective magnetospheric motion.

The magnetosphere encompasses a wide range ofphenomena, described in their individual sections (seefigures 3 and 4). Because of the long-range couplingpresent in the magnetosphere, its seemingly independentphenomena mutually interact more than is the case in theneutral atmosphere.

PlasmaPLASMA is a gas which contains significant numbers ofions and electrons, so that it conducts electricity at leastalong the magnetic field, and often perpendicular to itas well. The plasma at both Uranus and Neptune wascomposed of ions and electrons with kinetic energiescovering Voyager’s entire range of measurement, from 10to about 107 eV (1 eV = 1.6 × 10−19 J is a particle kineticenergy equivalent to a temperature of about 8700 K).Plasmas in different energy ranges behave differently, sowe discuss them separately.

The high-energy (≥105 eV) plasma at both planets isextremely sparse and forms radiation belts similar to thoseat Earth. It diffuses inwards through the magnetosphere,so that its absorption by satellites and rings createsdecreases in density that were observed by Voyager. Afew faint rings and satellites were discovered in this way.When these energetic particles strike icy satellites and ringparticles, hydrogen and oxygen atoms are ejected.

Figure 3. Diagrams of Uranus’s magnetoshere. The curved linesdepict the magnetic field as it is confined within themagnetopause. The arrow labeled is Uranus’s axis of rotation,and is fixed in space. The arrow labeledD is the magnetic dipoleaxis and rotates with the planet. The two diagrams show themagnetosphere at half-rotation intervals. The axis labeled XGSMpoints toward the Sun. (Figure from Voigt et al 1987J. Geophys. Res. 86 15 337.)

The low-energy (≤104 eV) plasma drifts throughthese magnetospheres on paths determined mostly bythe electric and magnetic fields. Both the rotation ofthe planets and the interaction of the solar wind with themagnetospheres create electric fields that fluctuate in timeas well as space. In the following, we describe what isknown of the behavior of the low-energy plasma.

Uranus. Uranus’s magnetospheric plasma is rathertenuous and consists almost entirely of H+. It appearsto originate from Uranus’s atmosphere, because if it camefrom the icy satellites there would also be O+, which is notobserved. It drifts Sunward as at Earth, but because ofthe pole-on configuration simultaneously co-rotates withUranus. However, the co-rotation does not interfere withthe plasma drift, as it does in the Earth’s plasmasphere,because the rotation occurs about an axis parallel to thedrift.

As the plasma drifts toward (away from) Uranus,it sees an increasing (decreasing) magnetic field strengthwhich compresses (expands) the plasma and heats (cools)it drastically. Thus there is a plasma adiabat in the




Figure 4. Diagram of Neptune’s magnetosphere, showing itscomponents. The curved lines with triangular arrowheadsdepict the solar wind flow deflected around the magnetopause.Those with V-like arrowheads are magnetic field lines, exceptthe one labeled ‘Triton’, which is Triton’s orbit. The line with thefull dots is the trajectory of the Voyager flyby. (From Schulz et al1995 Magnetospheric configuration of Neptune Neptune andTriton ed D P Cruikshank (Tucson, AZ: University of ArizonaPress). Copyright © 1995 by The Arizona Board of Regents.Reprinted by permission of the University of Arizona Press.)

magnetosphere just as there is a gaseous adiabat inUranus’s interior.

The plasma consists of a number of components,which are distinguished by their temperatures. Becauseof the adiabatic drift compression, their densities andtemperatures are best compared by referral to a standarddistance from the Uranian magnetic dipole. Two low-energy populations are known; a warm plasma (0.5 cm−3

and 10 eV at 5RU ) and a hot plasma (0.3 cm−3 and1000 eV at 5RU ). The warm plasma drifts by Uranuswithout much change, but the hot electrons continuallylose energy (relative to their adiabatic temperature) as theyapproach Uranus. The hot ions also abruptly disappearinside approximately 5RU , as though their drift paths wereavoiding Uranus.

Neptune. Neptune’s magnetospheric plasma is verytenuous and consists mostly of H+ and N+ ions. Itssource appears to be the atmosphere of Triton, becauseof the presence of N+ ions. In addition, a measure of theplasma density, corrected for compression by the magneticfield, appears to decrease as Neptune is approached. IfNeptune were the source, the plasma density measurewould increase towards the planet instead.

The decrease of plasma density measure towardsNeptune is due to loss of some of the plasma to collisionswith the atmosphere as the charged particles bounce backand forth along the magnetic field lines. Unlike thecase of the aurora, where high-energy (104 eV or more)

charged particles from far out in the magnetosphere hitthe atmosphere near the magnetic poles, this plasma lossrefers to low-energy (tens of eV) plasma on field lines nearthe planet that is impacting the atmosphere far from themagnetic poles.

The motion of Neptune’s magnetospheric plasmais complex and not completely understood. Neptune’sspin axis is roughly perpendicular to its orbital plane,just as it is at Jupiter and Saturn, where magnetosphericplasma rotates with the planet. However, the largetilt of the magnetic field results in Neptune’s magneticdipole alternating between states pointed at the Sun andperpendicular to the Sun–Neptune line as Neptune rotates.As a result plasma motions in Neptune’s magnetospherefluctuate as Neptune rotates. As at Uranus, plasma motionis part Sunward back-eddy and part co-rotation.

AuroraAuroras are regions where the atmosphere glowsbecause of bombardment by high-energy charged particles(usually electrons) which flow into the atmosphere alongthe magnetic field lines. They typically occur near, butnot quite at, magnetic poles because the field lines thereconnect to the largest regions of the magnetosphere.

Uranus. Uranus’s magnetic poles, and therefore itsauroras, are far from the geographic poles because ofits large magnetic dipole tilt (60) and offset (0.3RU ).Moreover, both auroras are on magnetic field lines thatconnect to the magnetospheric tail, so that the aurora-exciting particles (probably 104 eV electrons) streamingdown the field lines into the atmosphere are still driftingtoward Uranus. Because the plasma drift paths continueon past Uranus, on field lines where there is no aurora, itseems likely that the auroras are exhausting the availablehot electrons before they can drift completely past Uranus.

The aurora is visible primarily at in the extremeultraviolet at wavelengths between 90 and 110 nm, emittedby electronic transitions between low-lying states ofmolecular hydrogen. These bands are excited by the low-energy secondary electrons emitted when high-energycharged particles impact the atmosphere.

Neptune. Neptune’s magnetic poles and auroras are alsofar from the geographic poles because of its large magneticdipole tilt (47) and offset (0.55RN). The light fromNeptune’s aurora, like Uranus’s, is extreme ultravioletH2 molecular band emission between 90 and 110 nm,stimulated by auroral secondary electrons. Some diffuseglow is also stimulated by electrons which are emitted bythe day-side atmosphere in response to solar ultravioletlight and which follow the magnetic field lines around toimpact the dark side.

Radio wavesThere are a number of types of radio waves emittedby Uranus and Neptune with frequencies in the range100–1000 kHz, both narrow and wide bandwidth and




emitted both in bursts and semicontinuously. Uranus’sand Neptune’s radio emissions are more similar to eachother than to those of other planets. They can beobserved from outside the magnetosphere, although, atthe present orientation of Uranus, the strongest Uranianradio emissions are beamed away from the Earth

Several types are emitted from regions not farout of the atmosphere on the auroral field lines. Inmost cases they are generated by electron velocitydistributions that are relaxing to equilibrium by means ofcollective instabilities. One of the ways to cause electronvelocity distributions to depart from equilibrium, and sogenerate these waves, is to selectively remove part ofthe distribution by atmospheric absorption in an aurora,which is why these radio emissions are located there.

Most of the radio emissions are confined to a narrowrange of directions, or beamed, and rotate with theplanet. The rotation periods of both Neptune and Uranuswere measured by timing the recurrent detection of theseemissions in the weeks after each Voyager flyby.

Another radio emission emitted at Uranus, but onlymarginally at Neptune, is the radio noise produced bylightning discharges. The strength of these emissionssuggests that lightning discharges are weaker at Uranusthan at Saturn but considerably stronger than at Neptuneand at Earth. At Neptune, however, whistler plasmawaves (see below) originating from lightning were foundinstead. Therefore lightning does occur at Neptune, butthe discharges must be much weaker than at Uranus.

Plasma wavesPlasma waves are collective disturbances in a plasma thatare coupled to perturbations in the electric and magneticfields and that move with respect to the plasma. They are,like the radio emissions, typically generated by departuresof the plasma from thermal equilibrium. Unlike theradio waves, however, they do not propagate out of theplasma, and so cannot be observed from outside themagnetosphere.

The strongest of them, Whistler-mode noise, is causedby the depletion of magnetospheric electrons travelingalong the magnetic field when they impact the atmosphereand create an aurora. This imbalance between the numbersof electrons with velocities parallel and perpendicularto the magnetic field is unstable, and the whistler-modenoise is generated by the motions of the electrons asthey approach equilibrium. In the process, however,more electrons stream along the field and impact theatmosphere, perpetuating the process. Unsurprisingly,therefore, whistler-mode noise was observed at its greatestintensity at Uranus when Voyager was on field lines thatwere connected to the aurora.

Whistlers are plasma waves propagating in the samemode as whistler-mode noise (and vice versa) but aregenerated by a different process—lightning discharges.Whistlers are very short bursts of plasma wave noisethat are spread out in frequency as they propagate alongthe magnetic field. In a radio receiver they sound like

pure tones descending in frequency; hence the name.Whistlers emitted by lightning discharges were detectedin Neptune’s magnetosphere.

Current researchMostly because of the lack of new data, Uranus andNeptune research has declined greatly in recent years.The HUBBLE SPACE TELESCOPE (or HST, a 2.4 m apertureorbiting telescope covering the wavelengths 115–2500 nm)and large-aperture ground-based telescopic infraredobservations of the ionospheric species H+

3 between 3890and 4090 nm are the main sources of new data.

There are many questions for future research. Howdoes Uranus’s atmosphere smooth out such great seasonalextremes of solar heating? Why is Uranus’s internal heatsource so small compared with those of the other gasgiants? Why is Uranus’s stratosphere so much calmer andless mixed than other giant planet stratospheres? Whatis the high-altitude heat source that heats Neptune’s andUranus’s upper atmospheres to such high temperatures?What is the pattern of motion of Uranus’s and Neptune’smagnetospheric plasma? What is the origin of bursty radioemissions? Why are Jupiter’s, Uranus’s and Neptune’sradio emissions rotation controlled, while Earth’s andSaturn’s are solar wind controlled?

BibliographyBergstralh J T, Miner E D and Matthews M S 1991 Uranus

(Tucson, AZ: University of Arizona Press)Cruikshank D P 1995 Neptune and Triton (Tucson, AZ:

University of Arizona Press)Lunine J I 1993 The atmospheres of Uranus and Neptune

Ann. Rev. Astron. Astrophys. 31 217–63

Floyd Herbert


Uranus: Rings E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Uranus: RingsThe Uranian ring system comprise ten narrow rings,one diffuse ring and some tenuous material between thenarrow rings. With a mixed nomenclature arising fromhistorical tradition, the names of the ten narrow rings (inorder of increasing distance from the planet) are 6, 5, 4, α,β, η, γ , δ, λ, ε; the name of the diffuse ring is 1986U1R.The innermost narrow ring lies about 16 000 km above themicrobar pressure level in the Uranian atmosphere, andthe entire ring system is about 10 000 km in radial extent.The distinctive feature of these rings is their narrowness,most having a radial width less than 12 km (table 1).

Although these rings can now be easily imagedfrom the Earth with modern detectors, they were firstobserved with the technique of stellar occultations. Astellar occultation occurs when starlight is blocked from anobserver by an intervening body—in this case the Uranianrings. Observations are carried out with high-speedphotometric equipment, ideally with several observingsites well spaced in the direction perpendicular to themotion of the ring shadows across the Earth. Therecording of the starlight intensity as a function of timethen provides a ‘line scan’ of the transmission of thering material. With multiple observations of the sameoccultation by a network of telescopes, several pointsaround each ring are recorded, and the ring orbits canbe constructed. By repeatedly observing occultationsover several years, one can learn the ring orbits togreat accuracy. One advantage of the stellar occultationtechnique over ordinary imaging observations with Earth-based telescopes is that occultations achieve a spatialresolution (limited by fresnel diffraction) of just a fewkilometers, rather than the much coarser resolutionachievable with Earth-based telescopic observations.

The discovery of the Uranian rings with airborne andground-based observations of a stellar occultation in 1977ushered in the beginning of modern ring research. Withonly the broad, bright ring system of Saturn as an example(see SATURN: RINGS, no one had suspected the existenceof dark, narrow rings. Any thought of narrow ringswas quickly dismissed because collisions between ringparticles would cause them to spread out into a broaderring over time. In order to explain the existence of narrowrings, theorists postulated the existence of small satellitesnear the rings, termed SHEPHERD MOONS, that would confinethe particles through their gravitational interaction. Thisidea was borne out by VOYAGER’s discovery of Prometheusand Pandora, two satellites that shepherd the narrowF ring of Saturn. Voyager also discovered Cordeliaand Ophelia, the two shepherd satellites of the ε ring.However, pairs of shepherd satellites have not yet beenidentified for the other Uranian rings, so the mechanism(s)for maintaining narrow rings may not be completelyidentified.

In the two decades following their discovery, about20 stellar occultations by the rings were observed,mostly in methane bands where the background light

Table 1. Orbital properties† of Uranus’s rings.

Semimajor Radial InclinationRing axis (km) width (km) Eccentricity (degrees)

6 41 837.2 1.0–2.5 0.001 013 0.0625 42 234.8 1–7 0.001 899 0.0544 42 570.9 1–7 0.001 059 0.0323α 44 718.5 4.5–10.5 0.000 761 0.0152β 45 661.0 5.5–12 0.000 442 0.0051η 47 175.9 55 ∼0.004 ∼0.0011γ 47 626.9 1–8 0.109 ∼0.002δ 48 300.12 2–8 0.004 0.0011λ 50 023.9 1.3–2.5 ∼0.0 ∼0.0ε 51 149.3 20–96 0.007 936 ∼0.0002

† Errors in these quantities are in the least significant digit, andvalues preceded by a tilde have an error comparable to thevalue. These results are from French et al (1991).

from Uranus is minimized. In 1986 the Voyager 2spacecraft flew by the Uranian system and recordedseveral high-resolution images of the rings. In addition,the Voyager Photo-polarimeter Subsystem (PPS) andUltraviolet Spectrometer (UVS) instruments observedstellar occultations by the rings, and occultations of theVoyager radio signal (from the Radio Science Subsystem,RSS) were observed from Earth. Due to the much closervantage point of the spacecraft, the Voyager data haveprovided the highest spatial resolution observations ofthe rings that we have so far. In particular, the radio-occultation data at X-band has a resolution of 50 m, whichcan be compared with the 2 km resolution of ground-basedstellar occultation data at visible wavelengths.

Several ring phenomena were first observed in theUranian ring system. After the extreme narrowness of therings, the most surprising of these is the elliptical and/orinclined orbits exhibited by some of the rings. Dynamicaltheories had predicted that rings would be limited tocircular orbits in the equatorial plane of the planet. Theeccentricities and inclinations are excited by the shepherdsatellites, which can also produce an extremely sharp ringedge, if it is located at an exact orbital resonance with thesatellite. Examples of sharp edges produced by resonanceswith the satellites Cordelia and Ophelia are the inner andouter edges of the ε ring, the outer edge of the δ ring andthe inner edge of the γ ring.

Elliptical ring orbits exhibit PRECESSION, which is a slowchange in the orientation of the ellipse. The precession ofthe ring orbits is caused by the oblate mass distributionwithin Uranus, and the rate of this motion gives usinformation about the interior structure of the planet.Here we can consider the rings as ‘test particles’ movingwithin Uranus’ gravitational field. From their precessionalmotions, two of the gravitational harmonic coefficientshave been precisely established (J2 = 3.3434 × 10−3 andJ4 = −2.89 × 10−5). These allow the computation of theUranian internal density distribution, which provides aconstraint on allowable interior models for Uranus.



Figure 1. Discovery observations of the Uranian rings from theKuiper Airborne Observatory. The intensity of starlight beforeand after Uranus occulted the star is plotted versus distancefrom the center of the planet. Nine dips were observed on eachside, which occurred at the same radius, indicating that Uranusis encircled by a system of narrow rings. (After Elliot J L 1979Stellar occultation studies of the solar system Ann. Rev. Astron.Astrophys. 17 445–75. Reprinted, with permission, from theAnnual Reviews of Astronomy and Astrophysics Volume 17, ©1979,by Annual Reviews www.AnnualReviews.org.)

The largest ring of the system (ε) has a minimumradial width of 20 km (at PERIAPSIS, its closest distance toUranus) and a maximum 96 km (at APOAPSIS, its furthestdistance). This width is a strict linear function of thering radius, which implies that the precessions of theindividual particle orbits are ‘locked’ at the same rate.If the precession rates were not all the same, then theorbits of the particles in the outer parts of the ring wouldprecess more slowly than those for the particles closer tothe planet, and the radial width of the ring would becomeuniform around the ring. The radial widths of rings α andβ also vary linearly as with distance from the planet, butthe variation is only 4.5 to 10.5 km for the α ring and 5.5to 12 km for the β ring. One mechanism proposed formaintaining the locked precession of the particle orbits isthe gravity of the ring particles themselves, but this hasnot been conclusively proven to be the case.

In discussing the structure of the Uranian rings,one property of interest is the equivalent depth, whichis the optical depth (minus the natural logarithm of thering transmission) integrated along the radial direction.Thus the equivalent depth measures the amount of

Figure 2. Voyager observations of the Uranian ring system. Thethree faint rings in the lower right are (in order of increasingradius) 6, 5 and 4; these are followed by α, β, η, γ , δ and ε. The λring is not visible. Ring ε contains the most material and is thebrightest because it reflects the most sunlight. (CourtesyNASA/JPL/Caltech.)

material in the ring in a way that is independent of theangle at which the ring occults a star or other source ofradiation. For most rings the equivalent depth is thesame all around the ring, indicating a constant amountof material at each longitude of the ring orbit. Equivalentdepth has the dimensions of length.

Rings 6, 5 and 4 have a structure similar to eachother, being very narrow with equivalent depths of 0.8,1.8 and 1.4 km. These rings have variable widths, butthe widths do not correlate with orbital radius (as theydo for the α, β and ε rings). The Voyager occultationdata reveal significant internal structure. Rings α and β

are more substantial, with equivalent depths of 6.0 kmand 3.8 km respectively. These rings rarely exhibit thesharp edges characteristic of resonances with satellites(two sharp-edged profiles were observed for the α ringin the Voyager data). Rings γ and δ are concentrated, withequivalent depths of 6.6 km and 4.1 km. Each of these ringshas an excited normal mode, with an amplitude of 5 km forthe γ ring and 3 km for the δ ring. The δ ring may also havean internal density wave. Theη ring is unique, with a sharp



Figure 3. Occultation profile of the ε ring. This graph shows thenormal radio opacity (optical depth) of the ring as a function ofradius. Note the sharp inner and outer edges caused by orbitalresonances with Cordelia and Ophelia and the detailed internalstructure. Below the dashed line, the variations are mostly dueto noise, rather than structure of the ring. (After Gresh D L,Marouf E A, Tyler G L, Rosen P A and Simpson R A 1989Voyager radio occultation by Uranus’ rings. I. Observationalresults Icarus 78 131–68.)

inner feature (similar to rings 4, 5 and 6) and an adjacentbroad, low-optical-depth component about 55 km in radialextent and with a sharp outer edge. Both components ofthe η ring follow circular orbits in the equatorial planeof Uranus. The λ ring is tenuous and clumpy, with anequivalent depth of only 0.1 km. Finally, the ε ring hasvery sharp inner and outer edges and an equivalent depthof 83.8 km. Its occultation profile has a ‘W’ shape, showingthat its material becomes more concentrated at its edges(figure 3).

The rings are very flat—with a vertical thickness(perpendicular to the plane of their orbits) of only a fewtens of meters. The ring particles themselves are dark, withALBEDOS of a few per cent. Comparison of the occultationprofiles of the rings at different wavelengths reveals thatthey contain very few particles smaller than centimetersizes, with particle sizes tens of centimeters having beeninferred in the ε ring. A Voyager image of the rings (inforward scattered sunlight) shows a small population ofmicron-sized dust particles in the ring system.

Compared with the other ring systems, Uranus’ ringshave less material than those of Saturn, but more thanthe ring systems of Jupiter and Neptune (see JUPITER: RINGS

NEPTUNE: RINGS). All the Uranian rings lie inside the ROCHE

LIMIT, which is the smallest radius for which a body canremain intact by its self-gravitational forces alone (withoutbeing pulled apart by the tidal forces from Uranus). Fora satellite (such as Cordelia and Ophelia) to exist withinthis region, it must have sufficiently high density and/ortensile strength to prevent the tidal forces from Uranusbreaking it up. In fact, each of the rings could have formedin this manner. Once formed, a ring is subject to erosionprocesses, such as radiation drag, meteoroid impact and

drag from Uranus’ exosphere. One can then estimate howlong these processes could have been operating in the past,which sets limits on the age of the rings. Estimates alongthese lines suggest ring ages of a few × 106 years—muchless than the age of the solar system (4.5× 109 yr).

BibliographyElliot J and Kerr R 1984 Rings: Discoveries from Galileo to

Voyager (Cambridge, MA: MIT Press)Elliot J L and Nicholson P D 1984 The rings of Uranus

Planetary Rings ed R Greenberg and A Brahic (Tucson,AZ: University of Arizona Press) pp 25–72

Esposito L W, Brahic A, Burns J A and Marouf E A1991 Particle properties and processes in Uranus’rings Uranus ed J T Bergstrahl, E D Miner and M SMatthews (Tucson, AZ: University of Arizona Press)pp 410–65

French R G, Nicholson P D, Porco C C and Marouf E A1991Dynamics and structure of the Uranian rings Uranused J T Bergstralh, E D Miner and M S Matthews(Tucson, AZ: University of Arizona Press) pp 327–409

J L Elliot


Urey, Harold Clayton (1893–1981) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Urey, Harold Clayton (1893–1981)Chemist, born in Walkerton, Indiana, Nobel prizewinnerfor Chemistry in 1934 ‘for his discovery of heavyhydrogen’. It was at Columbia University that he isolatedthe isotope deuterium by distilling liquid hydrogen; inthe Second World War, he directed the effort to separateuranium-235 from uranium-238 for the atomic bomb. Atthe University of Chicago, he worked on the origin of theelements, their abundance in stars, and the origin of theplanets, including the chemical properties of the Earth.He invented a technique that used oxygen isotope-bearingminerals to date rocks and to measure paleohistoric watertemperatures (see MILUTIN MILANKOVITCH). He analysedlunar rocks from the Apollo missions.


Ursa Major E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Ursa Major(the Great Bear; abbrev. UMa, gen. Ursae Majoris; area1280 sq. deg.) a northern constellation which lies betweenDraco and Leo Minor–Leo–Canes Venatici, and culminatesat midnight in mid March. Its origin is uncertain, thoughit was known to the ancient Greeks, who identified itwith two mythological figures—Callisto, a mortal whowas turned into a bear after having fallen victim to Zeus’spassion and whom he placed for safety in the sky, andAdrasteia, a Cretan nymph who raised the infant Zeusand whom in gratitude he placed among the stars. Itsbrightest stars were cataloged by Ptolemy (c. AD 100–175)in the Almagest.

The third-largest and probably the best known ofthe constellations, Ursa Major is easily recognized by theasterism of the Plough or Big Dipper, formed by seven ofits brightest stars, α, β, γ , δ, ε, ζ and η, which since ancienttimes has been used for navigation, as a line drawn from β

throughαUrsa Majoris points toαUrsae Minoris (Polaris),a second-magnitude star that lies within 1 of the northcelestial pole. The brightest stars of Ursa Major are εUrsaeMajoris (Alioth), magnitude 1.8, α Ursae Majoris (Dubhe),a very close binary with orange (K0) and pale yellow(F0) components, magnitudes 2.0 and 4.9 (combinedmagnitude 1.8), separation 0.7′′, period 44.4 years, ηUrsaeMajoris (Alkaid or Benetnasch), magnitude 1.9, ζ UrsaeMajoris (Mizar), a multiple system consisting of twowhite (A2 and A7) components, magnitudes 2.3 and 3.9,separation 14.4′′, both of which have a fainter companion,which forms a wide optical double with 80 Ursae Majoris(Alcor), magnitude 4.0, separation 11.8′, β Ursae Majoris(Merak), magnitude 2.3, and γ Ursae Majoris, magnitude2.4. There are 13 other stars of magnitude 4.0 or brighter.The five central stars of the Plough (β, γ , δ, ε and ζ UrsaeMajoris) are part of the Ursa Major Moving Cluster, whichis the closest star cluster to the Sun, its center being about75 light-years distant.

Another interesting multiple star system is ξ UrsaeMajoris (Alula Australis), which consists of two yellow(G0) components, magnitudes 4.3 and 4.8 (combinedmagnitude 3.8), separation 0.9–3.1′′, period 59.8 years,the former of which has an unseen companion whichrevolves around it in 669 days, and the latter two unseencompanions, one of which revolves around it in 3.98 days.The main components were the first binary system to havean orbit computed, by M Savary in 1828. Other interestingstars include Lalande 21185 (magnitude 7.5), which, at adistance of 8.3 light-years, is the fourth closest star to theSun, and Groombridge 1830 (magnitude 6.4), which hasthe third largest proper motion of any star.

Other interesting objects include M81 (NGC 3031), aseventh-magnitude spiral galaxy, its near neighbor M82(NGC 3034), an eighth-magnitude irregular, starburstgalaxy which is thought to have experienced disruptionfollowing a near-collision with the much more massiveM81 some 40 million years ago, and M101 (NGC 5457), aneighth-magnitude face-on spiral galaxy.

See also: Groombridge 1830, Mizar and Alcor, Plough.


Ursa Minor E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Ursa Minor(the Little Bear; abbrev. UMi, gen. Ursae Minoris; area256 sq. deg.) a northern circumpolar constellation whichis surrounded on three sides by Draco and includes thenorth celestial pole. Its origin is uncertain, though it wasknown to the ancient Greeks, who identified it with Ida, aCretan nymph who helped her sister, Adrasteia, raise theinfant Zeus and whom in gratitude Zeus placed amongthe stars alongside Adrasteia (who is represented by theneighboring constellation of Ursa Major). Its brighteststars were cataloged by Ptolemy (c. AD 100–175) in theAlmagest.

A small, rather inconspicuous constellation, UrsaMinor’s only claim to fame is that its brightest star, αUrsaeMinoris (Polaris or Alrucaba), lies within 1 of the northcelestial pole. α Urase Minoris is actually a triple systemwith pale yellow (F7 and F3) components, the primary ofwhich is a Cepheid variable (range 1.86–2.13 decreasing,period 3.97 days) and the secondary a star of magnitude8.2, separation 18.4′′, and a third, unseen component whichrevolves around the primary in 30.5 years. There aretwo other stars brighter than magnitude 4.0: β UrsaeMajoris (Kochab), magnitude 2.1, and γ Ursae Majoris(Pherkad), magnitude 3.0, which together are known asthe ‘Guardians of the Pole’. The seven brightest stars ofthe constellation (β, γ , η, ζ , ε, δ and α) form an asterismsimilar to the Plough and are sometimes called the LittleDipper.

There are no bright star clusters, nebulae or galaxiesin the constellation, the brightest being NGC 6217, aneleventh-magnitude spiral galaxy.

See also: Polaris.


US Astronomy E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

US AstronomyAs the 19th century ended, astronomy underwent a periodof rapid growth in the United States, a growth that wasfueled by both the expansion of the university systemand private philanthropy and which also paralleled thegrowth in astrophysical research. For the first half of the20th century, the US government took little interest in thefunding of astronomical research, concentrating on thoseaspects of astronomy that were of greatest commercialand military interest: navigation and the determinationof time.

Following World War II, the involvement of the USgovernment in funding all branches of science grew andthe National Science Foundation (NSF) established andbuilt the national observatories. With the birth of spaceastronomy in the 1960s, the National Aeronautics andSpace Administration (NASA) became a major force inastronomical research. Despite an increasing governmentpresence, there remained a steady source of privateand institutional funds contributing to the operations ofastronomical facilities and to the support of research.Together with this growth in funding came steady growthin the number of practicing research astronomers in theUS, a number which grew more than six-fold in the threedecades from 1960 to 1990.

Funding sourcesThe dominant portion of US government funding forastronomy is provided through two agencies, NASA andthe NSF. Other agencies supporting astronomical researchinclude the Department of Defense, the Departmentof Energy and the National Oceanic and AtmosphericAdministration.

NASA currently provides about 80% of the totalfederal funding, with most of the balance provided bythe NSF. NASA funds astronomy through the Office ofSpace Science (OSS), which had a budget of 2.1 billiondollars in fiscal year (FY) 1999. This office funds theplanning, construction, launch and operation of space-based astronomical observing facilities. It also providesfunding for the archiving and analysis of data from spacemissions. Furthermore, the OSS operates some high-altitude observing facilities, which fly on balloons orairplanes.

The NSF division of astronomical sciences (AST)operates within the Mathematical and Physical Sciencedirectorate of the agency. Its budget for FY 1999was 119 million dollars. The NSF operates severalnational observing facilities, directly funds astronomicalresearchers and provides some funding for educationalpurposes. The NSF also provides funding for the design,development and construction of national ground-basedobservational facilities. In FY 1999, 8 million dollarswas allocated for the design and development phasesof the ATACAMA LARGE MILLIMETER Array (ALMA, formerlyknown as the MMA or millimeter array), an internationalcollaboration.

There are numerous sources of non-federal fundingfor astronomy. These include private organizations,private universities, state universities, state and localgovernments and private foundations. The amount ofnon-federal funding for astronomical research is difficultto determine owing to the broad array of sources.Universities typically provide a significant level of supportthrough salaries for faculty members. Private universitiessuch as the California Institute of Technology or HarvardUniversity as well as public universities such as theUniversity of California or the University of Texas alsouse institutional funds to support observatory operations.Among private organizations, the Carnegie Institution ofWashington and the Lowell Endowment provide ongoingoperating funds for specific observatories while othergroups such as the Keck Foundation, the Sloan Foundationand the Ford Foundation have provided substantialfunding for the construction of observatories or for specificresearch projects.

FacilitiesTo effectively study the diverse array of objects in the uni-verse, observations must be made using all wavelengthsof electromagnetic radiation. Some wavelengths of lightare not able to penetrate the Earth’s atmosphere. Somemethods of observation require extensive instrumentation.Some direct physical data are gleaned from analysis of col-lected cosmic dust particles or meteoritic material. Forthese reasons, observations must take place both from thesurface of the Earth and from space.

In the United States, observatories are funded by theFederal Government and through private sources. TheNSF owns the facilities of and contracts for the operationsof the National Observatories. These are as follows:

• the National Optical Astronomy Observatories, withobservatories on KITT PEAK, Arizona, CERRO TOLOLO,Chile, and Sunspot, New Mexico;

• The NATIONAL RADIO ASTRONOMY OBSERVATORY, withobservatories at Green Bank, West Virginia, Socorro,New Mexico, and Kitt Peak, Arizona;

• the National Astronomy and Ionosphere Center atARECIBO, Puerto Rico.

In addition:

• The NSF provides operational support to a numberof university radio astronomy facilities and the NSFOffice of Polar Programs, which operates severalastronomical facilities at the South Pole,

• the NSF is a member of an international opticaltelescope partnership named GEMINI, which hasbuilt two identical telescopes, one in the northernhemisphere and one in the southern hemisphere (theUnited States funds approximately 48% of the overallbudget; see GEMINI OBSERVATORY).

The non-federally funded observatories are mainlysupported through universities, consortia of universities



or private foundations. Many observatories operatewith mixed sources of funding. For example, someobservatories are constructed using university funds orwith grants from private foundations and then relyon grants from the federal agencies to provide fundsfor development and construction of instruments andsometimes for operations.

In the last 10 yr, NASA has launched three ofthe four so-called ‘Great Observatories’. These includethe HUBBLE SPACE TELESCOPE, the CHANDRA X-RAY OBSERVATORY

and the COMPTON GAMMA-RAY OBSERVATORY. The final GreatObservatory, the SPACE INFRARED TELESCOPE FACILITY, isscheduled for launch in December 2001. These satelliteshave radically changed our view of the universe andprovided unprecedented discoveries unattainable fromthe ground.

In addition to the Great Observatories series, NASAhas a range of other missions from small to large. Thereare several major mission programs including Discovery,Explorer, Mars Exploration and New Millennium. Eachmission has specific goals. Discovery missions exploreobjects within our solar system such as planets, moonsand other small bodies. The Explorer program providesfrequent flight opportunities at three separate fundinglevels for missions that support one of the four majorthemes of the Office of Space Science: university (lessthan $7.5 million), small (less than $71 million) andmedium (less than $140 million). These themes are theAstronomical Search for Origins, Planetary Systems, TheSun–Earth Connection and the Structure and Evolution ofthe Universe.

NASAmaintains 15 research centers across the UnitedStates, each of which fulfills various portions of NASA’soverall mission. The centers most closely associatedwith astronomy missions include the GODDARD SPACE

FLIGHT CENTER in Greenbelt, Maryland, the JET PROPULSION

LABORATORY in Pasadena, California, and MARSHALL SPACE

FLIGHT CENTER in Huntsville, Alabama.NASA has led the way in archiving astronomical

data. It has created numerous facilities such as the HubbleData Archive, the Astronomical Data Center, the PlanetaryNode system and the Astrophysics Data System, whichprovide researchers access to archival data.

There has been a recent trend toward major projects,whether on the ground or in space, being undertakenthrough international collaborations because of the largecosts and complexity of the projects. This trend is likely tocontinue into the near future.

Demographics and employment patternsAfter a decade of rapid growth in the 1980s, the number ofastronomers in the United States has grown more slowlybut steadily in the 1990s. The median age of astronomicalresearchers has remained approximately the same, but theage distribution has widened somewhat with the numberof astronomers in the age range from 35 to 50 decreasingfrom about 50% at the end of the 1980s to 40% at the endof the 1990s.

Toward the end of the 1990s there has been a levelingoff and even a decrease in the number of students pursuingstudies in physics and astronomy at the undergraduateand graduate levels in American universities. Almostall graduate students in astronomy or astrophysics haveprepared themselves for graduate study by obtaining anundergraduate degree in either physics or astronomy.Among astronomy PhD recipients from US institutionsin 1997, 19% were female and 27% were not US citizens(source: American Institute of Physics, 1997 GraduateStudent Report).

About six out of every ten astronomers in the UnitedStates are employed in academic institutions, colleges oruniversities, and slightly over half of these are in tenuredor tenure-track positions. Most of the others are workingin federal laboratories or research centers or for industrialcontractors supporting these centers.

Research publications

The bulk of the research published in the United Statesappears in one of the four scholarly journals publishedby the astronomical societies. The AMERICAN ASTRONOMICAL

SOCIETY (AAS) publishes The Astrophysical Journal, TheAstrophysical Journal Supplement Series and The AstronomicalJournal. The ASTRONOMICAL SOCIETY OF THE PACIFIC (ASP)publishes The Publications of the Astronomical Society of thePacific. In addition, there is a section of The Physical Review(published by the American Physical Society) devotedto astrophysical research, and both Icarus (AcademicPress) and The Journal of Geophysical Research (AmericanGeophysical Union) publish research in planetary andspace physics.

Annual Reviews, Inc., publishes a review volume,The Annual Review of Astronomy and Astrophysics. TheAstronomical Society of the Pacific and the AmericanInstitute of Physics are the principal publishers ofconference proceedings.

As technology has developed in recent years, therehas been an increasing use of on-line preprint postingservices for rapid dissemination of research results. Themost popular of these is part of the arXiv e-print archiveservice hosted at the Los Alamos National Laboratory(http://xxx.lanl.gov/archive/astro-ph).

Amateur astronomers in the US

Astronomy appeals to nearly everyone. Since the sky isaccessible to all, astronomy is blessed with a large andactive amateur community. The US has a number ofamateur organizations that serve many diverse purposes.These range from small local clubs that meet regularlyto exchange advice on telescopes and host ‘star parties’to large national organizations that coordinate researchactivities easily performed by amateurs. Some of the largernational organizations that include amateurs are listed intable 1.



Table 1. Larger national organizations.

Organization Purpose Webpage

AMERICAN ASSOCIATION OF VARIABLE

STAR OBSERVERS (AAVSO) Variable star research www.aavso.orgAstronomical Society of the Pacific Education, outreach www.aspsky.orgAstronomical League Observation www.astroleague.orgINTERNATIONAL DARK SKY ASSOCIATION Light pollution control www.ida.org

SummaryAstronomy in the United States is ending a period thathas lasted a few years more than a century and has seenthe US emerge as a world leader in astronomical research.At the beginning of the 20th century, the US was leadingobservational astronomy into an era of increasinglypowerful and sophisticated telescopes. The 40 in refractorhad just been commissioned and the two powerfultelescopes on MOUNT WILSON were in the near future. Whenthese were put to work, our view of the universe changedcompletely. Edwin HUBBLE explored the extragalacticdistance scale and we began to understand the expansionof the universe. These discoveries, in turn, spurred theconstruction of newer, more powerful telescopes such asthe 200 in Hale telescope. Following the discovery of radiowaves from cosmic sources, powerful instruments such asthe Very Large Array were commissioned (see VERY LARGE

ARRAY NATIONAL RADIO ASTRONOMY OBSERVATORY). The UnitedStates led the way in space beginning in the 1960s andopened new and previously unexplored regions of theelectromagnetic spectrum for astronomical observations.A series of increasingly ambitious and complex spacecraftwere built and operated, leading finally to NASA’s GreatObservatory Series, which reaches completion in the earlyyears of the 21st century.

At the beginning of the 21st century, Americanastronomers have access to an unprecedented suiteof instruments and techniques with which to observethe heavens. New technology has made possible theconstruction of telescopes of a size that was inconceivablea generation ago, and the Keck Telescopes, at 10 m each,have led the way in showing what can be achieved withsuch powerful instruments. The successors to the NASAinstruments of the 1990s are being planned even while thecurrent generation is at its peak of performance.

Kevin B Marvel and Robert W Milkey


US Naval Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

US Naval ObservatoryThe US Naval Observatory is the oldest astronomicalobservatory in the United States, and the oldestcontinuously operating scientific institution in the USgovernment. Founded in 1830 as a Depot of Chartsand Instruments for rating chronometers and maintainingnavigational instruments, by 1844 it had become thefirst national observatory of the United States, analogousto the ROYAL OBSERVATORY IN GREENWICH, England. Theobservatory’s headquarters are located in Washington,DC, including the historic 0.66 m refractor used to discoverthe two moons of Mars in 1877. The observatory’s largesttelescope, a 1.55 m astrometric reflector, is located at itsstation in Flagstaff, Arizona. The observatory’s missionhas always been to aid in the improvement of navigationas well as to conduct basic research in astronomy. Today,it provides the national time service for the United States,determines the precise positions and motions of celestialbodies, measures Earth rotation parameters includingpolar motion, and produces a variety of almanacs for useby astronomers, navigators and the general public.

For further information seehttp://www.usno.navy.mil/


Ussher, James (1581–1656) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Ussher, James (1581–1656)The Archbishop of Armagh and Primate of All Ireland,Ussher was a churchman and a scholar. He correlatedMiddle Eastern and Mediterranean histories with Jewishgenealogies of the Old Testament, and the resultingchronology was incorporated into the Authorized Versionof the Bible of 1701. The chronology established the year ofcreation as 4004 BC. Lightfoot followed Ussher, improvingon Ussher’s accuracy by declaring that the Earth had beencreated at 6:00 a.m. on 26 October 4004 BC. The dates ofother biblical events followed, for example, thatAdam andEve were driven from Paradise on Monday 10 November4004 BC, and that the Ark landed on Mount Ararat on 5 May1491 BC, ‘on a Wednesday’. Ussher was a great scholar,if over-literal in his analysis of the Bible. He collectedthe earliest available manuscripts into his library, whichformed the nucleus of the great library of Trinity College,Dublin.


Valles Marineris E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Valles MarinerisA complex system of canyons on Mars, centered at 11.6 S,70.7 W, stretching for a total of 4128 km in the east–westdirection just south of the equator, and reaching depthsof over 6 km. It is named after the Mariner probes whichreturned the first close-up images of the planet’s surface,and is also referred to as Mariner Valley. Individualsections of the systems are termed chasmata. At its westernend, where the system abuts the faulted area at the eastof the Tharsis Bulge known as Noctis Labyrinthus, lieTithonium Chasma and, to its south, Ius Chasma. To theireast, the system expands into the three parallel canyonsnamed, from north to south, Ophir Chasma, CandorChasma and Melas Chasma. To the northwest lies theunconnected Hebes Chasma. The main system continueseastward with Coprates Chasma, and then opens out, viaCapri Chasma to the north and Eos Chasma to the south,into the so-called chaotic terrain of the region known asMargaritifer Sinus.

Valles Marineris was created largely by faulting, butother forces have been at work. The deep branchingvalleys running into the southern edge of Ius Chasmasuggest erosion by groundwater, while at the eastern endare teardrop-shaped islands suggestive of flowing water.Close up views of Coprates and the other central chasmataclearly reveal the presence of layered sediments, whichcould only have been deposited under water.

See also: Mars: surface.


Valongo Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Valongo ObservatoryThe Valongo Observatory, part of the Federal University ofRio de Janeiro (UFRJ), sited in downtown Rio de Janeiro,Brazil, was founded in 1881. Its main activity is scientificresearch in the field of astronomy, carried out by its staffof twelve astronomers. The professors/researchers keepscientific collaboration programs with other institutions atboth national and international levels. The main researchareas are: fundamental astronomy, stellar astrophysics, in-terstellar medium, extragalactic astronomy and laboratoryastrophysics. Funding comes chiefly from the Braziliangovernment agencies. The Valongo Observatory offers anundergraduate course in astronomy, attended by approx-imately 90 students. Its astronomical instruments includean original turn of the century Thomas Cook & Sons 30 cmrefractor telescope, and a Zeiss reflector of 15 cm, for un-dergraduate teaching. As extracurricular activities, thereis a visitation program offered to the general public, alsoserving secondary schools, which features telescopic ob-servations and conferences given twice a month.

For further information seehttp://www.ufrj.br/ov.


Van Allen Belts E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Van Allen BeltsTwo toroidal regions surrounding the Earth which containtrapped charged particles, discovered by James Van Allen.

See: Magnetosphere of Earth: radiation belts.


Van Allen, James (1914–) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Van Allen, James (1914–)Magnetospheric physicist, born in Mount Pleasant, Iowa,discovered the Van Allen belts that surround Earth. Afterwar-time service, Van Allen used V-2 rockets on high-altitude experiments, specifying a replacement rocket,the Aerobee, when supplies of the V-2 ran out. Duringthe International Geophysical Year (1957–8), the firstAmerican satellite, Explorer 1, carried a micrometeoritedetector and a cosmic ray experiment designed by VanAllen. Data from Explorer 1 and Explorer 3 (launched 26March 1958) revealed the existence of a doughnut-shapedregion of charged particle radiation trapped by Earth’smagnetic field. Later in 1958, Pioneer 3 led to the discoveryof a second radiation belt.


van de Kamp, Peter (1901–95) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

van de Kamp, Peter (1901–95)Astronomer, born in Kampen, Netherlands, becamedirector of the Sproul Observatory and professor atSwarthmore College, PA. He worked on the astrometricmeasurement of proper motions of stars across the skyand discovered oscillations in their paths. He interpretedthe deflections as due to the revolution of planetarysystems around the stars. The oscillations have notbeen confirmed, and many astronomers think that theywere instrumental effects caused by the telescopes andmeasurement techniques.


van Maanen, Adriaan (1884–1947) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

van Maanen, Adriaan (1884–1947)Dutch astronomer, became a member of the Mount Wilsonstaff, and studied the rotation of spiral nebulae as a meansto establish their distances. His detection of rotationalmotions was illusory and misleadingly suggested that theywere relatively nearby. Through its proper motion andparallax, he discovered the white dwarf van Maanen’s star.


Van Maanen’s Star E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Van Maanen’s StarThis star in the constellation of Pisces, although faint, isof much importance as it is the nearest readily observablewhite dwarf—a class of star of considerable significancein studies of stellar evolution. The white dwarfs Sirius Band Procyon B are both closer, but their proximity to theirmuch brighter companions makes them very difficult tostudy as individual stars.

Van Maanen’s Star, also known by the catalogdesignation Wolf 28, is situated about 2 south of δPiscium. It was discovered in 1917 by the Dutch-Americanastronomer Adriaan Van Maanen (1884–1946), from acomparison of plates exposed in 1914 and 1917, to have thelarge proper motion of 2.978′′ per annum. Its spectrum isclassified DG. With a parallax of 0.227′′, it is only 14.4 light-years distant, yet its apparent magnitude is only 12.37,indicating that it has a very low intrinsic luminosity (itsabsolute magnitude is 14.1). It has been estimated that itsdiameter may be no more than 12 500 km, comparable tothat of the Earth.

See also: white dwarfs.


Van Vleck Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Van Vleck ObservatoryThe Van Vleck Observatory on the campus of WesleyanUniversity in Middletown, Connecticut, was built in theyears 1914–16 upon the disassociation of the departmentsof mathematics and astronomy. It was named in memoryof John Monroe Van Vleck, Professor of Natural Science atWesleyan throughout much of the nineteenth century.

The primary instrument of the observatory is a visualrefracting telescope of 0.5 m aperture and a focal lengthof 8.41 m. The crown and flint lenses were made byC AR Lundin of the Alvan Clark Co. and were installed ona Warner and Swasey mounting in 1992. The two lensesare separated by about 10 cm, allowing cleaning of allsurfaces without disturbing them. Their alignment hasthus not been altered, and this and other features make therefractor among the best for astronomical research. Since1992, the telescope has been mainly used to determinetrigonometric parallaxes and proper motions of manyhundreds of faint nearby stars. In the last few decadesthe Van Vleck astrometric program has been one of theleaders in the use of parallaxes to respond to problems instellar astrophysics.

In 1971, a 0.6 cm reflecting telescope was obtainedfrom the estate of Richard Perkin, founder of the Perkin-Elmer Corp. Like the refractor, the on-campus site ofthis instrument has encouraged a monitoring programusing photoelectric photometry and later CCD imagingcapability, of properties of T Tauri variables and otheryoung stars. The department of astronomy offers theBA and MA degrees, and research programs on bothtelescopes feature student participation at every level.

For further information seehttp://www.astro.wesleyan.edu/astro.html.


Variable Stars ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS

Copyright © Nature Publishing Group 2001Brunel Road, Houndmills Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998and Institute of Physics Publishing 2001Dirac House, Temple Back, Bristol, BS1 6BE, UK 1

Variable StarsStars vary in many ways and for many reasons. Broadcategories of variation include stars that pulsate,fluoresce, become veiled, are disturbed by companions,erupt, explode and even change their spots. Variationsmay be episodic, irregular, semi-regular or periodic. Aseparate wide class comprises the extrinsic variables suchas eclipsing and tidally distorted binaries andgravitationally lensed stars, whose variations result fromchanging aspect, but here we survey the intrinsicvariables.

The first recognized variable stars were novae andsupernovae (‘new’ stars), with some very old recordsbeing astrophysically useful, chiefly because they fix theaccurate date of a stellar explosion. Although the ancientaccounts are obscure and fragmentary and do notrecognize modern distinctions (such as nova versussupernova), the type of event may still be discernible. Forexample, a supernova can be visible for 6 months or evenlonger, whereas ordinary novae are prominent for amatter of weeks, so a nearby nova can mimic a distantsupernova in brightness but is typically a much brieferevent. Observations of exploding stars go back to severalcenturies BC with records of some objects from two ormore places, and would go back further if not for libraryburnings. The records were kept by imperial astrologersin China (later in Japan and Korea) as the ‘new’ starswere considered important omens. Today’s telescopicobservations provide good statistics of novae, but usefulstatistics of the infrequent supernovae in our Galaxy relyon the old records. Events observed in other galaxieshave demonstrated enormous differences in scalebetween supernovae and novae. Theory has shown thosedifferences to be far greater than optical observationsindicate, as supernovae release the vast bulk of theirenergy in invisible forms.

Except for novae, stars were regarded as constantuntil the mid to late 1600s, when the first non-eruptivevariables were recognized. Most notable was thepulsating red giant o Ceti (Mira, the Wonderful) with its≈330d large-amplitude variations. An appearance anddisappearance of Mira had been noticed by D Fabriciusin 1596, but was taken to be a nova. The early historiesof intrinsic and extrinsic variables are inextricably linked,as ideas about causes emerged only slowly over severalcenturies. Systematic recording replaced casual notesover the 1700s, with the improved records leading tofurther discoveries. Thus θ Serpentis was conclusivelyshown to be variable by E Pigott, confirming GMontanari’s suspicions of a century earlier, and JGoodricke found periodic variation in δ Cephei, theprototype of Cepheid variables, and in the eclipsingbinary β Lyrae. The improved records also led to somesuccess in identifying variation mechanisms. Speculationof the late 1700s cycled among eclipses, moving and

transient spots on rotating stars, and rotating distortedstars. All of these phenomena commonly occur, but earlyassignments to particular stars were usually wrong.Pulsation was advocated only much later. Algol was(correctly) said to eclipse by Pigott and Goodricke, butboth later favored other ideas. Curiously, the eclipsehypothesis of Algol was cast into disrepute by discoveryof variable stars that clearly were not eclipsing binaries,apparently from a wish to have all variables follow asingle theory. Spots on rotating stars were very popularin the 1700s and 1800s, even for Mira, and have nowreturned for RS Canum Venaticorum, W Ursae Majoris,FK Comae and BY Draconis type binaries. As late as c.1930, pulsating star velocity curves were fitted in termsof orbital parameters, even when there was little or nodoubt about their pulsational origin. A common findingwas that the orbit of a star’s unseen (and non-existent)companion was actually inside the observed star. Theorbit parameters would be printed anyway after acomment or two on the unphysical situation and the needto fit something. Although the reality of pulsation forLong Period Variables (LPVs) and Cepheids was notseriously disputed, there was much uncertainty in sortingtrue pulsators from a miscellany of competingphenomena. Most authors simply avoided explicitmention of pulsation while discussing effects thatscarcely could be due to anything else, with anoccasional remark that velocity variations might notreally be due to line of sight motion of any kind. Thepopularity of pulsation was undeniable in the 1920s and1930s, however, when up to 10% of AstrophysicalJournal papers concerned Cepheids and related stars—remarkably high for one subject. Nearly all contributionswere observational at that time, although majortheoretical advances had already been made by A Ritterin the 1870s and 1880s and by A S Eddington in the1920s. Full acceptance of spectroscopy’s implications forthe pulsation hypothesis took nearly a half century (c.1890–1935), yet many kinds of pulsating stars wereeventually recognized as a direct consequence of theradial velocity and temperature changes discoveredspectroscopically. Not only the existence of radialmotion, but its phasing compared with that of lightcurves, is a crucial discriminant among models.

Realization that large numbers of variable stars existwas slow to develop until the breakthrough ofphotography. Only ≈10 non-exploding variables wereknown in the early 1800s, but the permanent records inphotographic plates quickly extended the list into thehundreds between 1850 and 1900. Systematicphotographic surveys in the 1900s at Harvard, Bamberg,Leiden and other observatories raised catalog listings intothe tens of thousands. Recent automated gravitationallens surveys with electronic detectors have producedlight curves of thousands of variables as a by-product ofthe surveys’ primary objective, thereby substantially



increasing the discovery rate. The General Catalog ofVariable Stars (‘GCVS’, Kholopov 1984) is a four-volume listing of coordinates, types and otherinformation on 28 211 variables of all kinds, nearly 70%being intrinsic variables. A companion catalog of NewSuspected Variables (‘NSV’, Kholopov 1982) contains14 811 stars of uncertain status.

A reasonably complete list of variable star types withminimal descriptions would exceed available space, butthe GCVS devotes five pages to descriptions. Itrecognizes 88 types of intrinsic variables, including 34pulsating, 7 rotating, 10 x-ray, 22 eruptive and 15cataclysmic types (cataclysmic and other eruptivevariables are kept distinct). A standard namingconvention allows immediate recognition of a star asvariable, although some bright variables have only theirBayer names. To begin the sequence, a single capitalletter is assigned alphabetically in order of discoveryfrom R to Z, then double letters RR, RS, …, RZ, thendouble letters AA, AB, … to AZ, then BB, BC, … to BZ,and ultimately to ZZ, followed by the constellation name(e.g. T CrB; ZZ Cet). Letter J is not used. Designations ofthe form V335 Cygni, V336 Cygni, etc. follow the 334names of the letter system.

There is great interest in variable stars amongamateur astronomers, whose observations areindispensible for professionals. Not only visualbrightness estimates but phototube and charge-coupleddevice (CCD) measures are now made available in largequantities through bulletins, journals and especially theInternet. Variable star organizations include theAmerican Association of Variable Star Observers(AAVSO), the Association Française des Observateursd’Etoiles Variables (AFOEV), the British AstronomicalAssociation Variable Star Section (BAAVSS),Bundesdeutsche Arbeitsgemeinschaft für VeränderlicheSterne (BAV), the Center for Backyard Astrophysics(CBA), the Hungarian Astronomical Association,International Amateur-Professional PhotoelectricPhotometry (IAPPP), the Variable Star Group of theAstronomical Society of Southern Australia, VariableStar Network (VSNET), and the Variable Star Section ofthe Royal Astronomical Society of New Zealand.Information on all these organizations can be found viathe Internet. Their imprints on professional astronomydate at least as far back as the 1920s—for example inJoy’s use of an AAVSO Mira light curve in the 1926Astrophysical Journal. The role of such organizations israpidly expanding today. The central professionalorganization is Commission 27 of the InternationalAstronomical Union.

LPVs and Cepheid variables were not onlydiscovered early but still dominate our catalogs. LPVsare pulsating red giant stars of great luminosity and arethereby rather easy to discover. The name Mira type isreserved for the LPVs of larger light amplitudes,

although there probably is continuity in the overall class,including Galactic distribution, space velocity andcomposition as well as light variation. There may also becontinuity with ‘non-variable’ red giants, which probablyhave some variation. Spectra of LPVs and especiallyMiras show prominent emission lines that are clearly dueto pulsation, as non-variable red giants lack emissionlines but otherwise have similar spectra. Morespecifically, pulsation amplitude correlates with emissionline strength. Miras vary by up to 9m in blue light (afactor of ≈4000), but the enormous optical variations arecaused by emission being ultra temperature-sensitive,with the optical band lying far to the short-wavelengthside of the spectral energy peak. Bolometric amplitudesare similar to those of Cepheids. A problem forspectroscopy, especially prior to modern efficientspectrographs and large telescope apertures, was thefaintness of Miras near minimum light. Observationsmust cover full cycles if much is to be learned aboutpulsation, and it was only in 1926 that A Joy producedthe first complete velocity curve of Mira. The classicalCepheids are highly luminous F and G type pulsatinggiants and supergiants that vary periodically up toroughly a magnitude. Periods range from a few days toover 100 days. Rather early it was realized that theimportant relation between period and mean density

applies to Cepheids and several other prominentcategories of pulsating stars. The relation exists because astar’s dynamical timescale depends on its size and mass,with some dependence on internal structure (thus adifferent constant for each category). Stars with givenHR diagram coordinates have the same size, but notnecessarily the same mass and structure, so pulsationperiods provide a link to structure and thereby tounderstanding evolution. The W Virginis stars, membersof a very old Galactic population, pulsate with their ownPρ1/2 relation and are understood to be in a laterevolutionary stage than classical Cepheids, having lostmuch of their original envelopes. Thus pulsation is animportant evolutionary diagnostic that can even helptrace Galactic origins.

In addition to the essentially periodic and roughlyperiodic classes of pulsating stars, there are giants andsupergiants of spectral classes F to M that pulsateirregularly or semi-regularly. Curiously they lie mainlybetween the more regular Cepheids and LPVs in the HRdiagram. The less regular and typically redder of theseare known as RV Tauri stars, while the somewhat moreregular ones are simply called yellow semi-regularvariables. Periods, in so far as they can be quantified atall, range from tens to thousands of days, and amplitudescan be up to about four magnitudes. These stars may beinteresting in terms of chaos theory, as some show



alternating cycles of larger and smaller amplitude thathint of the period doubling seen in the approach to chaos.

Brightness measurementWe have astrophysically useful records of the brightnessvariations of the supernovae of 1572 and 1604, asobserved by T Brahe and others and by J Keplerrespectively, despite lack of light detection equipmentand even of telescopes 400 years ago. Their usefulnessillustrates two points about astronomical brightnessmeasurement. First, although precision requirements arestringent for small variations, they can be relaxed forlarge variations. Second, in no case can we allow thescale to be wrong, unless mere detection of variability isthe only objective. One might expect eye estimates to bequantitatively useless, but Brahe’s and Kepler’s lightcurves demonstrate otherwise. The situation has beensaved by the ‘secret’ advantage of the photometricastronomer—the remarkable long-term constancy ofnormal stars. Brahe and Kepler recorded theirsupernovas’ magnitudes relative to known stars that, asfar as we know, are essentially unchanged in brightness.Modern photoelectric observations of the reference starshave accordingly placed the supernova estimates on anobjective scale. Of course the old estimates aresubjective, but the largest overall errors are limited by themagnitude spacings of reference stars. An estimationmethod involving magnitude steps, introduced by FArgelander about 1840, further diminished maximumerrors.

The existence of natural constant brightnessstandards largely accounts for the accuracy ofastronomical photometry, which is typically better thanthat of laboratory photometry despite the disadvantage ofobserving through Earth’s irregular atmosphere.Astronomers also have an advantage in the temperaturesof reference sources, as very hot laboratory standardswould vaporize their surroundings, while hot stars haveno surroundings to vaporize. Difficulties due to daylightand weather as well as complete blockage of spectralregions are best dealt with by observing from space,although many less severe problems have beenminimized through a variety of clever techniques.

The history of star brightness measurement canroughly be divided into an early era of qualitative notes(c. 1600 to c. 1800), a middle era of quantitative butinaccurate estimation by eye or photography (beginningc. 1800 and still somewhat active), and a recent era ofaccurate quantitative measurement with variouselectronic detectors (beginning c. 1910). Even the earlyqualitative work was a major improvement on theunplanned occasional notices of preceding times. It madepossible the discovery of periodic variation and, coupledwith long baselines in time, even accurate periods. ThusGoodricke estimated Algol’s period in 1784 with anuncertainty of less than a minute out of 2.87 days.

Entering the middle era, note that the eye can reliablyjudge brightness equality between two stars, a point wellappreciated and exploited by F W Herschel, J Herschel,Argelander, J Hartmann and K Schwarzschild, but eyeestimates of inequality are notoriously subjective. Thebetter visual work accordingly employed a variety ofnow obsolete ways to reduce the apparent brightness ofone star by a known factor so as to equal that of another.Techniques, sometimes applied also to photographicobserving, included partial aperture blocking with twintelescopes, extra-focal imaging, tapered neutral filters(wedges), and crossed polarizers. Photographicmagnitudes were extracted from diameters of focusedimages or blackening of unfocused images. The moderndefinition of astronomical magnitudes, relatingmagnitude (m) differences to light (l) ratios, m1 − m2 =−2.5 log (l1/l2), was suggested in 1850 by N Pogson.Simple as it may seem, Pogson’s contribution wascrucially important, as the quantitative meaning ofmagnitudes had previously been vague. The newaccuracy and increased time lines allowed periodchanges to be measured, with S Newcomb’s text of 1884,Astronomy, already mentioning period changes for Algol,Mira and β Lyrae. The above-mentioned accuracy trickswork only within small fields, so measurement of actualmagnitudes, as opposed to ‘local’ magnitude differences,received a large boost with the arrival of photoelectricdetectors. Invention of the photomultiplier tube in the1940s especially helped to surmount atmosphericattenuation problems via all-sky photometry, and alsogreatly improved accuracy of magnitude changes forvariable stars. Development of modern standardmagnitude systems began when photomultipliers cameinto widespread use. Examples include the U, B, V, R, I,J, K, L wide-band systems and the uvby medium-bandsystems. A standard star has accurately known constantmagnitude in one or more standard systems (e.g. γ Ophhas V = 3m.72, B = 3m.76), while a comparison star has(presumably) constant but not necessarily knownmagnitude. CCD detectors similar to those in videocameras are now replacing photomultipliers because oftheir ability to measure many sources simultaneously andthus provide for accurate differential photometry. Newpractical problems have arisen out of the enormousquantities of data coming from CCD photometry.Advanced computer technology is being applied tostorage and distribution problems, in some cases withdata collection directly from the detecting equipment.Automatic telescope operation is becoming ever morewidespread in many spectral regions, not only indiscovery modes (as in extra-Galactic supernovasearches) and quick follow-up on γ and x-ray transients,but also in routine observation of known variables. Theacronym APT, for Automatic Photometric Telescope,implies remotely operated, programmable instrumentsthat are primarily used for variable star research.



Astrophysical importance of variable starsThe scientific value of variable stars can be seen in twoissues—why they vary (theory) and what their variationsallow us to measure (observation). The theoretical pointconnects physical makeup with behavior, while theobservational one concerns properties such as size,distance and internal structure that are measurable orestimable because of variation.

Most astronomical distance estimates are based onthe standard candle method, a comparison of apparentand intrinsic brightness. The extra-Galactic distance scaleis an important application, and distances to multiplestars and clusters within our Galaxy also are found fromembedded standard candles. The method applies to anystellar group that contains a standard candle and whosediameter is small compared with its distance. Goodstandard candles are consistent (small luminosity rangeamong individuals), highly luminous (for observation atgreat distance) and reliably recognizable. Variable starsnaturally satisfy the recognition requirement, while sometypes satisfy the other requirements reasonably well andin some cases quite well. Cepheids, being giants andsupergiants, can be observed in distant galaxies.Individual Cepheids differ in luminosity by factors up toabout 100 and so might seem not to satisfy theconsistency requirement, but H Leavitt discovered (c.1905) that Cepheid luminosities are closely related totheir directly measurable periods of brightness variation.After calibration, the period–luminosity (P–L) relationallows Cepheids to be used as standard candles, althoughessentially correct calibration required another half-century and is still being refined. However, even faultyearly calibrations established distances to galaxies thatwere correct to an order of magnitude and thereby helpedelucidate the nature of the ‘spiral nebulae’ and the scaleof the Universe. Early applications were E Hertzsprung’sdistance estimate for the Small Magellanic Cloud and HShapley’s work on the extra-Galactic distance scale,following E Hubble’s discovery of Cepheids in severalgalaxies. The distinction between Galactic and extra-Galactic objects was well established by 1925, five yearsafter the famous Shapley–Curtis debates, when Hubblefound 11 Cepheids in galaxy NGC 6822 and derived adistance. The Hubble Space Telescope and large ground-based telescopes have recently produced importantincreases in the Cepheid distance limit. Several kinds ofexploding stars have very high luminosity, but Type Iasupernovae are best for recognition and consistency, inaddition to being especially luminous. However, even thefar less luminous and far less consistent ordinary novaehave been used as distance indicators.

The large-amplitude pulsations of Cepheids andrelated variables can be exploited to measure radii ratherdirectly. Luminosities then follow from combination ofthe radii with effective temperature (Teff) estimates, so asto provide standard candles. Observed variation is due to

separable changes in surface brightness and size.‘Wesselink radii’ (after A Wesselink) require three kindsof accurate observations—a radial velocity curve, a lightcurve and a color index curve (briefly ‘color curve’—formed by differencing light curves (in magnitudes) fortwo effective wavelengths). Color curves of Cepheidsresemble their light curves at first glance, having smalleramplitudes and subtly different shapes. Points with givencolor occur at paired times that are close together near acolor maximum or minimum and well separated atintermediate color. The basic idea is that color index is areliable indicator of Teff and therefore of surfacebrightness. However, a quantitative relation to predictsurface brightness from color is not needed—just thereasonable assumption that when the star returns to agiven color it also returns to the ‘original’ surfacebrightness. Any change in observable light, l, betweenthose two times is due to changed surface area, so theratio of radii at the two times is

The difference of radii follows from an integrationbetween the corresponding points on the radial velocitycurve,

Now having both ratio and difference, we find theindividual radii at the two times. Repeating the procedurefor many time pairs, we find the run of R with time. Thefundamental assumption that surface brightness is aunique function of color for a given star may not be quitetrue, but is close enough for useful applications. Withactual radii in kilometers and with model stellaratmosphere predictions of surface emission per unit area,a final step computes luminosities. The method is one ofthe few direct means to find accurate radii of giant starsand to calibrate P–L relations.

Another natural situation to exploit is the existenceof variable stars in binaries and multiple star systems,where the variable star and binary star characteristicsshould be compatible if our evolutionary understanding iscorrect. For example, the age of a variable should agreewith that of its companion(s), and its absolute dimensionsand distance may be derivable from observations of thebinary or multiple system. Examples include δ Scuti (seebelow) type members in Y Camelopardalis and ABCassiopeia, a Cepheid in V350 Sagittarii, and a β Cepheistar in 16 Lacertae. It is important to discover more suchsystems.

There are few probes of stellar interiors so it isimportant to have a wide variety of checks on global and



surface properties, and the measures provided bypulsation and other forms of variation are welcome. Forexample, Cepheid masses can be inferred from bothevolutionary and pulsational models, so the comparisoneither provides a check or sounds an alarm. Light curveand velocity curve details also can be predicted frompulsation models. The help comes where needed, aspulsational instabilities occur primarily in highly evolvedstars where structure is complicated and theories areuncertain. Not only stellar structure theories but alsoequations of state, opacities, and energy generation rates(‘laboratory physics’ in principle but not in practice) arethereby subject to scrutiny.

On a wide front, galactic evolution is largelyunderstood in terms of population types. For example theRR Lyrae type pulsating stars found in globular clustersand spheroidally distributed Galactic populations are notjust short-period Cepheids, but have quite differentGalactic distribution, motions, history and structure.Insight into RR Lyraes and Cepheids can apply to theirnon-variable neighbors in the HR diagram. On theobservational side of galaxies, certain variable stars canbe recognized at large distances and thus serve aspopulation indicators.

Pulsation studies have now broadened into therapidly developing field of asteroseismology, in whichnew observing methods and theoretical work on stellaroscillations extract structural information from multiple-frequency small amplitude variations in light andvelocity. Applications to the Sun (helioseismology) areparticularly successful because the Sun is spatiallyresolved and because it has a rich spectrum ofoscillations. The directly tapped energy of solaroscillations is that of convective motions in the outerenvelope that generate sound waves. Certain other kindsof stars, such as Ap, δ Scuti (see below) and especiallysome white dwarf stars, show intricate oscillations thatare clues to their evolutionary states. World-wideobserving networks such as the Whole Earth Telescope(WET) allow the long, nearly continuous, coverageneeded to separate closely spaced frequencies.

Explosion mechanismsNova theories of the early to mid 1900s involved rapidcontraction to the white dwarf state or dynamicalresonances combined with nuclear reactions. Thosetheories now seem inapplicable to known kinds of stellarexplosions. A modern schematic model that covers manyparticular categories includes a donor star to provide asupply of gas and an accreting star on which eruptionsoccur. The flow may be a stream from overspilling acritical lobe or it may be a wind, and the eruptions maybe powered by thermonuclear or gravitational energy.Ordinarily the donor is non-degenerate while the accretoris some kind of compact degenerate object, and mostcommonly a white dwarf star. Neutron star accretors are

much rarer, and main sequence donors greatly outnumberevolved donors, but a wide variety of combinationsoccur. An especially abundant class is that of thecataclysmic variables (CVs) where the compact object isa white dwarf, although the white dwarf often cannot bedetected directly. In most CVs the companion is a low-mass main sequence star, the orbit period is well under aday, and the entire binary is similar in size to the Sun,and smaller in many cases. CVs are typically old objectsas shown by their distribution in the Galaxy, which is in athicker disk than ordinary Population I stars.

A major clue into mechanisms and evolution was theinferred presence, via radial velocities and light curves,of white dwarfs in post novae, recurrent novae and nova-like variables in the 1950s to 1960s. Althoughthermonuclear models had been proposed earlier, ideasfor explosions soon focused on accretion-driven surfacehydrogen-burning runaways on white dwarfs. CVcomponent masses are difficult to estimate, but the whitedwarfs are typically much more massive than their non-degenerate companions. Relatively massive donorsshould be subject to unstable mass transfer that wouldradically change the configuration (flow from high to lowmass or between comparable mass stars favorsinstability). Evolved donors (giants and subgiants) havethe same problem—for structural reasons they tend toexpand, overflow their lobes and transfer gas unstably ona large scale. So the lack of evolved and of relativelymassive donor stars is commonly explained by their self-destructive tendencies. However, a CV needs some levelof slow mass transfer to fuel its eruptions. A suitablelevel can be maintained by orbit shrinkage due tomagnetic wind braking or gravitational radiation.

Classical novae brighten by 9m or more over a fewdays and then decline irregularly over weeks for fastnovae or months for slow novae. Symbiotic novae (seebelow) are much slower (duration ≈ decades) than theslowest classical nova and are much wider binaries thanthe ordinary CVs. The mass donor is typically a giant.For classical novae, more or less standard spectraldevelopments occur, with emission and absorption lineepisodes and ejecta velocities of the order of 103 km s−1,but with a large velocity range among examples.Interpretation is difficult, as the state of the gas is hard tocompute and the dynamical situation is intricate, withmultiple velocity systems seen at most times. Classicalnovae have had only one known outburst, but thatcircumstance is attributed to the short history ofobservational astronomy. Thus there are also nova-likevariables that are probably the same as classical novaebut have not exploded in (astronomical) historical times.Observations of post-novae find the underlying binariesnot markedly changed by their outbursts.

Recurrent novae explode like classical novae but theeruptions are frequent, typically decades apart. Currentthinking favors accretors that are close to the white dwarf



mass limit, such that only small accretion build-up isneeded to start a thermonuclear runaway. Chemicalabundances in the ejecta do suggest advanced nuclearprocessing, as expected for remnants of stellar cores nearthe edge of collapse.

A computation of simple energetics shows thatgravitational energy alone can power a large outburstfrom a white dwarf for a plausible sudden accretionepisode. Objects identified with that mechanism are thedwarf novae—CVs that have cyclic outbursts with highrepetition rates. Brightenings of up to about fivemagnitudes are separated by typical intervals of weeks tomonths and complicated shorter-term behavior includesfast flickering. Ideas to account for the episodes includesudden release of matter from a circumstellar disk (diskinstability) and variable supply from the donor star(source instability). Examples are SS Cygni, UGeminorum, Z Camelopardalis and SU Ursae Majoris,each a prototype in a finer classification. Thermonuclearenergy may play some role in dwarf novae. Ideas aboutevolutionary relationships among the several kinds ofCVs are under continual revision.

Impressive as they are, CV explosions are smallfirecrackers compared with supernovae. In contrast toCVs, where the basic configuration remains after anexplosion, a supernova event involves the entire star.Supernova mechanisms include thermonuclearincineration of a massive white dwarf star (in a binary)and collapse of an old, dense stellar core to nucleardegeneracy. Other types that are probably much lesscommon have been proposed. The energy release inburning a white dwarf star can be estimated bymultiplying a white dwarf mass (≈ 3 × 1033 g) by c2 andby an efficiency factor (≈10−3 for helium burning andbeyond) to arrive at an energy between 1051 and 1052 erg,about a million times that of a typical nova. Theminimum energy of a core collapse supernova is evenmuch larger and follows from energy conservation information of a neutron star, whose gravitational bindingenergy, Eg, is of order −1053 erg. Obviously a positiveenergy that at least matches Eg must appear in theradiative and material ejecta. The visible radiationamounts to only ≈1048 to 1049 erg, and thus is a tinyfraction of the energy budget. Theoretical simulationsindicate that nearly all of the energy is carried byneutrinos. The detection of about two dozen neutrinosfrom supernova 1987a in the Large Magellanic Cloudwas a great triumph of supernova theory, as the numberobserved was the number predicted for a core collapse,within the uncertainties.

PulsationPulsating stars undergo true pulsations (oscillations insize or shape). The pulsations are of the envelope, as thecore is static and not directly involved. One couldimagine the high-temperature sensitivity of nuclear

reactions leading to core pulsations (the E mechanism),but theory predicts that core pulsations will be quicklydamped, except for supermassive stars. There are nocandidates for core pulsation among the recognizedvariable stars, except perhaps the bizarre object η Carinae(see below) which may be a supermassive pulsating star.Pulsations grow when there is net conversion of thermalto mechanical energy, so the star is a heat engine. Thereare zones that drive pulsation (net [thermal ⇒mechanical]) and others that damp pulsation (net[mechanical ⇒ thermal]). Elementary thermodynamicsrequires net heat injection in the compression stage ifheat is to be converted to work so that the engine runs.However, direct injection of heat around maximumcompression (as in an ordinary engine) is not the onlyway. Eddington realized in the 1920s that favorablecircumstances could occur if a star is relatively heat-tightwhen compressed. Rather than thermal energy beinginserted, its escape is prevented. This is the famousEddington valve mechanism—a generic means to achievenet driving that covers all specific ways to implementheat-tightness upon compression. Two specific ways arethe κ and γ mechanisms. The κ mechanism traps thermalenergy by making material in driving zones more opaqueupon compression (κ is the usual symbol for opacity).Kramers’ Law, κ = κ0ρ/T3.5, provides roughly correctopacity for given density (ρ) and temperature (T). With ρand T entered from accurate stellar structure models, itpredicts that stellar material becomes less opaque uponcompression, and indeed most stars do not pulsate.However, actual opacity depends in a complicated wayon thermodynamic variables and, although it decreaseswith compression in most parts of most stars, it increaseswith compression in the driving zones of Cepheids andsome other pulsating stars. The γ mechanism operates byincreasing the surface density of absorbing particles uponcompression. Imagine a thin pulsating shell of gas at itsmaximum and minimum radii. The number of containedparticles is the same whether the shell is large or smallbut the number per unit area is greatest when the shell issmall. Therefore the shell more effectively blocksradiation, or is more heat-tight, when compressed. The γmechanism always contributes to driving for obviousgeometrical reasons, while the κ mechanism can produceeither driving or damping.

Driving will be weak if the main driving zones aretoo close to the surface (high) or too deep inside the star(low), and net damping will arrest pulsation. In the highcase, the problem is insufficient density in the drivingzones so that too little matter contributes to driving. Inthe low case, the driving zones lie in the nearly adiabaticinterior where, although there is plenty of mass, eachgram makes only a small contribution. Because driving isassociated with ionization zones of the abundantelements H and He, net driving zones will be high inrelatively hot (blue) stars and low in relatively cool (red)



stars. Thus we have a roughly vertical instability strip inthe HR diagram where Cepheids are found. However, itsred and blue borders are not well defined and many non-pulsating stars lie within the strip. The instability stripextends down to and below the main sequence andincludes the little-evolved or mildly evolved δ Scuti andrelated variables with their typical periods of hours. Itthen continues to the region of the pulsating white dwarfs(ZZ Ceti stars).

Several major distinctions characterize stellarpulsations. To begin, there are radial pulsations thatpreserve a star’s figure (shape) but change its volume,and non-radial pulsations that preserve volume but varythe figure. An example of radial pulsation could be theexpansion and contraction of a balloon under cyclicallychanging external pressure. A small-amplitude exampleof the non-radial case is provided by the tidal distortionsof the Earth and its oceans. A large-amplitude examplewould be the oscillations of a disturbed water globule. Amore formal description of the contrasting behavior is interms of a vector field. In radial pulsation, thedisplacement vectors of local matter elements pass backand forth through zero length but maintain fixedalignments along local radii. Non-radial pulsation is morecomplicated, with the vectors cyclically changing bothlength and direction. Of course, real pulsations need notbe purely radial or non-radial but can involve bothvolume and figure changes. However, many realisticsituations approximate these idealized types.

Another distinction involves fundamental andovertone pulsation. The situation is conceptually thesame as for a flexible string with one fixed and one freeend, where the free end is the analog of a pulsating star’ssurface. The string can have a fundamental oscillation,with a node only at the fixed end, and also overtoneoscillations with 1, 2, 3,…, n additional nodes. At a giveninstant, adjacent inter-node regions have oppositemotions (up versus down), and of course there is nomotion at the nodes. Similarly a pulsating star can havefundamental and overtone pulsations, with extra nodesfor the overtones. For a uniform string the nodes will beequally spaced, but a star has inwardly increasingdensity, with the consequence that node spacings are noteven approximately equal and must be computed.Fundamental and overtone pulsations can coexist and,accordingly, some unusual stars show beat phenomena.

Still another distinction is among pulsation modes,i.e. possible ways to pulsate. The large amplitude radialpulsations of Cepheid and RR Lyr type variables are inpressure modes (p-modes), so called because theyinvolve large local pressure variations. Both radial andnon-radial p-mode pulsations can occur. Gravity modes(g-modes) involve global oscillations about a hydrostaticlevel surface and are necessarily non-radial. They havemuch smaller pressure variations than do p-modes. Of

course, energy can feed from one mode to another andbetween fundamental and overtones.

No one kind of pulsation analysis serves allpurposes. As in other areas, success in computingdetailed behavior does not guarantee understanding. Thegeneral phenomenon is highly nonlinear and one mightexpect maximum insight to follow from all-encompassing computations, but much insight has beengained from linear analyses in which only infinitesimalpulsations are studied. N Baker realized that outcome byrestricting attention to a single mass shell—the One Zonemodel. However, pulsating stars are complicatedstructures with many (thermally and mechanically)coupled zones, so quantitative predictions of periodsneed complete models. To view the problem the oppositeway, a full nonlinear computation should settle into acorrect period, but will not identify all possible periodsfound from linear theory. Although linear analysis leadsto insight on several fronts, it deals only withinfinitesimal pulsations and therefore cannot producecomplete models for comparison with observations.Possible pulsation modes and overtones are identifiedthrough linear analysis, with detailed behavior at finiteamplitude examined via nonlinear analysis, includingpossible interactions of fundamental and overtones.Similarly, growth and decay rates of small-amplitudepulsations can be investigated via linear analysis, butrates for large pulsations and final saturation amplitudesare matters for nonlinear analysis. Of course, specificlight and velocity curve features can be modeled onlywith full nonlinear computations.

A wealth of observed phenomena in pulsating starsremain only partly understood or have only recently beenunderstood, including light versus velocity phase lagsthat are characteristic of a given class, unusual surfaceabundances, changing velocity amplitudes and multipleperiodicity. Magnetic fields, fast rotation and binarycompanions are clearly present in some cases, but theirroles in pulsation are not usually obvious. Specificdriving mechanisms may not be clear. An active testingground for pulsation theory is provided by the β Cepheistars, a class sometimes called the β Canis Majoris stars,although current usage seems to have settled on the nameβ Cephei. These stars have remarkably short pulsationperiods of order 0.d2, with about half being doublyperiodic and thereby showing beat behavior in light andvelocity. Many RR Lyr and δ Sct stars also are doublyperiodic, but the two periods of a β Cep star can differ byas little as 1%. The prototype β Cephei (not itself doublyperiodic) is strongly magnetic and at least a triple system,with weak B-emission (Be) characteristics. Aphenomenon that must occur but is difficult to model isthat of dynamical tides in eccentric and/or asynchronousbinaries, which can be regarded as forced non-radialpulsations. Like ocean waves, such tides are largelystochastic so that prediction is mainly limited to



statistical behavior. Several supergiants with neutron starcompanions show the expected fluctuations in light andvelocity, including GP Velorum = Vela X1 and V884Scorpii, but quantitative matches with tidal theory remainunrealized.

Other variation mechanismsWhile eruptive and pulsating stars vary due to a varietyof dynamical, thermal and cyclic instabilities that aremore or less understood, or at least largely identified,variations in many categories are caused by instabilitiesthat are not so well understood, by random events and byrandom drifts in conditions. Some of these stars live onthe border between intrinsic and extrinsic variables. Forexample, RS CVn, BY Dra and W UMa binaries can beheavily covered by magnetic spots, similar to sunspots.The agent understood to produce such spots is dynamoaction in a fast-rotating convective envelope, so coolstars that have convective envelopes are likelycandidates—if they spin fast. The RS CVns, BY Dras,and W UMas naturally spin fast because they are tidallylocked close binaries, but FK Com and a small number ofsimilar stars are fast spinning single giants! Their fastrotation currently is attributed to their being recentlymerged ex-binary systems, with the binary orbitalangular momentum now existing as spin angularmomentum. Spot-modulated stars are extrinsic variablesin that they vary on the orbital time scale because ofchanging aspect of their spots. However, the spots grow,decline and move, so they are intrinsic variables onlonger time-scales.

Accretion tends to be highly irregular and causes notonly modest brightness fluctuations in active binaries, butvariations of up to several magnitudes in newly formedstars that accrete from their surroundings (T Tauri stars).Much of the luminosity of T Tauris and related stars isfrom the accretion process. Young objects of the FUOrionis type have brightening episodes of up to fivemagnitudes on a time-scale of decades to centuries.Outflows are associated with accretion but are not wellunderstood at present. Some classical T Tau stars aresurrounded by circumstellar disks that have been imagedin the infrared, with spectroscopic and imaging evidenceof bipolar outflows. Accretion onto white dwarfs not onlyleads to circumstances that eventually produce novaexplosions but also directly converts gravitational energyto more tangible forms, as seen above for dwarf novae.Even an occasional non-degenerate binary such as V361Lyrae has a hot spot where high-speed gas impacts one ofthe stars, having flowed in a well defined stream from theother. Variable flows produce variable spot luminositiesand thus another kind of intrinsic–extrinsic variable. Inpost novae and dwarf novae, the hot spot is on a disksurrounding the stream’s target star. Most exploding starsshow variability between outbursts in their ‘quiescent’light curves. A common seat of variation is the hot spot

as it flickers by 0.m1 to 0.m2 on a time scale of minutes,due to irregular inflow. Usually the hot spot itself, ratherthan either star or the overall disk, is the brightest lightsource in the binary. X-ray binaries whose accretingobjects are neutron stars have accretion power at least 30times their thermonuclear power (for H burning; morethan 100 times for He burning), so the promptly releasedenergy is about the same whether the material burns ornot. However, large thermonuclear bursts can occur onneutron stars if substantial amounts of fuel accumulate,providing the sources known as x-ray bursters, wherehelium is the fuel. As a group, x-ray binaries are variablein all spectral regions, from radio to gamma, althoughmost individuals are detected only in restricted ranges.There are two essentially disjoint classes on greatlydifferent scales, low-mass x-ray binaries (LMXBs) andhigh-mass x-ray binaries (HMXBs). At first inspection,LMXBs and HMXBs have little in common except thatthey contain neutron stars, and even a sketch of ideasabout their origins would exceed available space.Configurations are such that HMXBs accrete mainly viawinds from their blue supergiant companions and manypulse in x-rays, while LMXBs accrete mainly via lobeoverflow and very few pulse. LMXBs bear a remarkablesimilarity to CVs, including absolute dimensions, withthe notable difference being that a neutron star replacesthe CV’s white dwarf. It has been noted that someLMXBs may form from CVs in rare cases of accretionbeyond the white dwarf mass limit.

Still another kind of intrinsic–extrinsic hybrid is thesymbiotic star, where ultraviolet radiation from theenvirons of a small, hot accretor interacts with the wind,extended atmosphere or chromosphere of a red giantcompanion. Fluctuations in UV radiation and in windflow lead to intrinsically variable fluorescence modulatedby orbital aspect effects, sometimes including eclipses.Orbit periods are long compared with most variablebinaries, being typically hundreds of days or more.Because of observational difficulties, the definingcharacteristics of the class are necessarily superficial—the accreting objects are not usually directly observedand may be as diverse as neutron stars, white dwarfs andmain sequence stars. Extreme examples of symbioticsinclude symbiotic novae that have outbursts by brightnessfactors of order 100 and are believed to contain whitedwarf accretors. Unlike CVs, a symbiotic nova can be inoutburst for decades. The mechanism usually assumed isa thermonuclear surface flash on the accretor followed byslow cooling. These objects are often called slow novaebut differ radically from the CVs with the same name, so‘symbiotic nova’ should consistently be applied.Examples of symbiotic novae include PU Vulpeculae,RR Telescopii and V1329 Cygni.

Variable polarization can result from scattering bycircumstellar gas, but really spectacular variablepolarization is seen in CVs that contain accreting white



dwarfs with extremely strong magnetic fields. These arethe polars, or AM Her binaries, where cyclotronradiation in an oblique rotator produces strong circularpolarization over a wide range of wavelength, includingoptical light. Polars also are moderately strong x-raysources and natural laboratories for studyinghydrodynamic flows under combined gravitational andmagnetic fields, with variable light and polarization asdiagnostics. Intermediate polars have magnetic fieldsthat are intense but somewhat below those of polars.Examples of intermediate polars are EX Hydrae and GKPersei.

Fluctuations in stellar winds can generate small-scaleemission line and continuum variation, particularly wherewinds are strong. Most Wolf–Rayet (W–R) starsaccordingly vary by a few hundredths of a magnitude. AW–R star is the highly evolved residue of a very massivestar that has lost much of its envelope so as to expose hotinner regions. Often there is an evolved close companionwith its own strong wind, so that wind–wind interactionscause further variation.

Intermittent veiling is a very unusual variationmechanism and a major diagnostic of circumstellarconditions and interior structure in certain (usually giant)stars. In veiled stars we find intrinsic and extrinsicvariation combined, as clouds come and go whilechanging aspect controls their influence on light curves.Hydrodynamic flow in a mass transferring binary canlead to concentrations of partially opaque material beingprojected onto the face of one or both stars, so that lightcurves can help map circumstellar gas. An example isAX Monocerotis. Particularly spectacular examples ofveiling are the R Coronae Borealis stars—extremelycarbon rich and hydrogen deficient (apparently single)supergiants that are veiled by carbon particles at irregulartimes, with brightness drops by factors of up to about1000. These pulsating stars eject clouds of gas whosecarbon condenses to soot when sufficiently far from thehot photosphere, with clouds that happen to lie on theline of sight producing the veiling. Pulsationalcharacteristics, together with the strange abundances,provide strong constraints on possible evolutionaryhistories of R CrB stars. Given that only a few dozen areknown, R CrBs must either represent a very brief stage ofnormal evolution or products of an unusual formationprocess such as a merger.

Large evolutionary changes over the brief history ofobservational astrophysics are extremely unusual.Perhaps the most spectacular example is FG Sagittae,which has evolved from the small, hot exciting star of aplanetary nebula into a pulsating R CrB type red giant inonly a century, with decade to decade developments.Whether FG Sge is typical as a progenitor of R CrBs isan open issue at present.

Flares are brief local eruptions, prominent in bothcontinuum light and lines, from the chromospheres of

magnetically active stars. The UV Ceti or flare stars arelow-mass main sequence stars with unusually high levelsof chromospheric activity. Flares are much hotter thanred dwarf photospheres, so the brightening in magnitudesis highly wavelength dependent, increasing strongly intothe ultraviolet. As with spot-modulated stars, rotation in adeep convective envelope generates strong dynamoaction, with UV Cet stars being especially fast rotators.Flare activity is at least statistically a sign of youth, sincered dwarf rotation decreases with age.

Stars described as variable are traditionally those thatvary in brightness, especially over broad spectral regions.However, several kinds of spectrum variables vary mostnotably in spectral details, with behavior that can be asinteresting as in the more obvious variables. Forexample, strong variable magnetic fields are involved inspectral variations of Ap and Am stars (Ap = peculiarstars of spectral type A; Am = metallic line stars ofspectral type A). Stars that lose matter via powerful andunsteady winds, such as those of spectral type Of, showvariable emission lines with great Doppler broadening.Stars of type Be (B emission stars) were among the firstto draw special attention as spectroscopically interestingin the 1800s, and continue to stimulate hypotheses,observations, and controversy. The emission lines of Bestars are usually ascribed to circumstellar equatorial ringsassociated with centrifugal ejection of matter, butcircumstances leading to that situation are not widelyagreed upon. Even the basic point of whether binarynature is essential or irrelevant to Be behavior remains atissue.

Unique objectsMuch of the fascination of variable star astrophysics isprovided by unique objects that may result fromanomalous formation, short effective lifetime or selectioneffects. A few examples may give some of the flavor.

Most recognized causes of variation, includingpulsation, gas dynamic interactions and rotationphenomena, have been proposed for the spectacular andmysterious η Carinae. Especially notable are its greatluminosity and enormous swings in apparent magnitudeover centuries, including an interval in the 1840s when itapproached Sirius in brightness. It is usually regarded asa supermassive star with mass above (and perhaps farabove) 60 solar masses, and therefore a candidate forcore pulsation via unstable thermonuclear burning.Speculation that it is a binary is supported by recent highresolution imaging of a bipolar nebula that must havebeen ejected in η Car’s great outburst of c. 1840 and bydiscovery of 5.5 year spectroscopic and photometricperiodicities. Understanding of η Car is rapidlydeveloping but still rudimentary.

The recurrent nova T Corona Borealis has the basicred star–blue star binary morphology of novae, but with ared giant in place of the usual main sequence red star and



an orbit period (228d) typical of symbiotic novae.However, it can scarcely be called a symbiotic nova, asthe outbursts last not for decades but only for weeks. TCrB has no known siblings, but does have a rich andextensive literature.

And then there are the pulsating WHITE DWARF stars,the DUSTY CIRCUMSTELLAR DISKS, the QUASIPERIODICOSCILLATIONS IN X-RAY BINARY STARS, the incredibleSS433 and HERCULES X-1, …, but wait—those are otherarticles. No brief synopsis could do them justice. So turnto those articles—your instrumentation is a comfortablechair and a cup of hot chocolate. Enjoy.

BibliographyClark D H and Stephenson F R 1977 The Historical Supernovae

(Oxford: Pergamon)Gallagher J S and Starrfield S 1978 Annu. Rev. Astron.

Astrophys. 16 171Hoskin M 1979 J. Hist. Astron. 10 23Lewin W H G, van Paradijs J and van den Heuvel E P J (ed)

1995 X-ray Binaries (Cambridge: Cambridge UniversityPress)

Kholopov P N 1982 Catalog of New Suspected Variable Stars(Moscow: Nauka)

Kholopov P N 1984 General Catalog of Variable Stars(Moscow: Nauka)

Kippenhahn R and Weigert A 1990 Stellar Structure andEvolution (Berlin: Springer)

Payne-Gaposchkin C 1957 The Galactic Novae (Amsterdam:North Holland)

Warner B 1995 Cataclysmic Variable Stars (Cambridge:Cambridge University Press)

R E Wilson

Vatican Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Vatican ObservatoryThe Vatican Observatory is one of the oldest astronomicalinstitutes in the world. It began with the reformationof the calendar in 1582. At the Roman College, FatherAngelo Secchi first classified stars according to theirspectra. With these rich traditions Leo XIII, in 1891,formally founded the Vatican Observatory on a hillsidebehind the dome of St Peter’s Basilica. In 1935 PiusXI provided a new location for the Observatory at thePapal Summer Residence at Castel Gandolfo. In 1981 theObservatory founded a second research center in Tucson,Arizona. In 1993 the Observatory, in collaboration withSTEWARD OBSERVATORY, completed the construction of theVatican Advanced Technology Telescope (VATT) whichhas pioneered the new technology of creating large,lightweight, stable mirrors in a rotating furnace. Researchincludes cosmological models, spectral classification ofpeculiar stars, photometric studies of metallicity, mass-exchanging binary stars, material in star-forming darkclouds, dust envelopes about young stars, planetarydynamics. The Observatory is supported with anannual budget from the Holy See and by contributionsto the Vatican Observatory Foundation, a tax-exemptcorporation in the State of Arizona.

For further information seehttp://clavius.as.arizona.edu/vo/.


Vega E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

VegaThe star α Lyrae, the only bright star in the constellation.Its name, formerly Wega, derives from the Arabic Al Waki,‘the Swooping (or Falling) Eagle’, hence the form Alvakaused on some seventeenth century celestial globes. Inancient Greek and Latin writings it sometimes sharedthe constellation name Lyra, and consequently appearsas ‘the Harp Star’ in some later texts. It is the dominantcomponent of the Summer Triangle asterism, with Altairand Deneb. It was the Pole Star about 12 000 years ago,and will be again 14 000 years hence.

Vega is the fifth brightest star in the heavens, with anapparent magnitude of 0.03. It is the brightest star withsufficiently high northern declination to appear overheadin north temperate latitudes, and dominates their skieson clear summer nights. It is a striking object whenviewed with binoculars or a low-power telescope, its blue-white color (sometimes described as ‘sapphire’) belying itsrelatively cool spectral type (A0Vvar). Fairly close at 25.3light-years distance (parallax 0.129′′), Vega is in fact thethird closest of the bright stars. Its absolute magnitudeis 0.6. In 1983 it was discovered by the IRAS surveyto be surrounded by a disk of gas and dust, which mayeventually form a planetary system.


VEGA Space Mission E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

VEGA Space MissionVEGA (mission) is a combined spacecraft mission to VENUS

and COMET HALLEY. It was launched in the USSR at theend of 1984. The mission consisted of two identicalspacecraft VEGA1 and VEGA2. VEGAis an acronym builtfrom the words ‘Venus’ and ‘Halley’ (‘Galley’ in Russianspelling). The basic design of the spacecraft was the sameas has been used many times to deliver Soviet landersand orbiters to Venus. In June 1985, these spacecraftsuccessfully delivered the first balloons into the Venusatmosphere and also landers. After this the VEGA 1and VEGA 2 cometary probes were directed to the cometHalley. They encountered it on 6 and 9 March 1986. Sucha trajectory was possible only because of the favorablemutual positions of the comet and Venus at this time. Thegravitational field of Venus turned the spacecraft to thecomet and only a short switch-on of the spacecraft enginewas necessary.

The most important scientific achievements of theVEGA mission were as follows: (a) the first direct tracingof Venusian atmospheric properties (winds, turbulence,illuminance, cloud density) on a long horizontal path; (b)the first look at a cometary nucleus, not as a star-like object,but as a spatially resolved body; (c) the first identificationof a set of components of the cometary atmosphere,including organic matter; (d) the first study of the cometaryenvironment by direct (contact) methods. The VEGAmission was an important step in the development ofinternational cooperation in space studies. Scientistsand engineers of many countries were involved togetherwith Russians in the creation of sophisticated scientificinstruments and supporting systems. Such internationalcooperation in a space project was new not only for Russia.Professor R Sagdeev (leader then of the Institute of SpaceResearch in Moscow) was the head of the whole project,and these innovations were in many points results of hisenlightened activity.

VenusBalloonsThe launch of balloons with meteorological instrumentsis traditional in studies of the terrestrial atmosphere. Itis natural to apply this for atmospheres of other planets.The concept of a Venusian balloon was first proposed byJ Blamont in 1967, but only after 18 yr of technical studies(in the USSR and France) was it realized in the VEGAmission.

The Venus module of the VEGAspacecraft entered theatmosphere of the planet and separated into two parts at aheight of 64 km. One of them was a descent probe (lander)and the other was a canister containing balloon, ballast,parachute and inflating systems. Beginning at 64 km thecontainer went down with the parachute to about 50 km(where the atmospheric pressure is near 1 bar); there theballoon was inflated, and parachute and ballast separated.After this balloon went up to about 53.6 km correspondingto a pressure of 535 mbar and a temperature of 305 K.

Both balloons were inserted on the nightside of Venus,one (VEGA 1) ∼7 north of the equator and the other∼7 south of the equator. They drifted westward withthe predominant zonal wind closely parallel to the lines oflatitude. Each balloon was tracked over a distance of morethan 11 000 km, encountering dawn and then penetratingfar into the daylit atmosphere (figure 1(D)). Transmissionwas stopped after 46 h of flight owing to limited batteryresources.

There were two types of measurements: in situ bymeans of on-board instruments transmitted by telemetryand ground-based determination of balloon motion byvery-long-baseline interferometry (VLBI). The balloon’stransmitter sent directly to the Earth signals at awavelength of 18 cm used ordinarily by VLBI astronomersfor observations of distant galactic OH sources. Becauseof Earth’s rotation, continuous reception of telemetry datarequired the use of many antennas widely distributed inlongitude.

It was known earlier that so-called superrotation isthe dominating dynamical property of Venus’ atmosphere:its rotation period (about 4 days near cloud top) is muchshorter than for the solid surface. This means that a fastwestward wind is a permanent feature on Venus (see VENUS:

ATMOSPHERE). VEGA balloons provided its mean velocity(67–69 m s−1) and its fine fluctuation pattern also.

Strong turbulence was found with amplitude to2 m s−1 with a time scale of 30–100 s (figure 1(C)). Theinfluence of surface topography on atmospheric motionswas found; it was not expected for such high altitudes.The measurements covered the twilight zone which wasfound to be 7 wide in the clouds of Venus. The horizontalprofile of the clouds’ optical properties (backscatteringcoefficient) was measured for the first time. It was foundthat at the height of flight it is smooth in general; even thestrongest variations were no more than a factor of 3 fromthe average level (figure 1(E)). Such an experiment has notyet been repeated; however, balloons are present in manyproposals for future studies of the atmospheres of Venusand Mars.

Descent probesStudies of Venus by means of descent probes were startedby the VENERA 4 mission in 1967. Another 10 SovietVENERA MISSIONS followed during 1969–1983; eight of themincluded descent probes with instrumental packages thatprovided measurements of the chemical composition ofthe atmosphere and soil, images of the landscape aroundlanders, vertical profiles of temperature and pressure, thecloud particles’ properties and solar illuminance withinthe deep atmosphere of Venus. VEGA1 and 2’s descendingprobes landed on the nightside of the planet, unlike mostof the previous descenders, but the general scheme wastraditional: entry to the atmosphere, fast aerodynamicdeceleration of capsule with descenders between 100 and64 km, opening of capsule, about 10 min of descent to46 km, separation of parachute, landing after about 51 minand 20 min of working data transfer from the surface.



Figure 1. Mission history for VEGA 1 balloon showing (A) pressure, (B) temperature, (C) atmospheric vertical velocity, (D) ambientlight level and (E) cloud backscatter coefficient as functions of time after insertion on 11 June 1985. Venus longitudes are also shown onthe abscissa. The approximate position of the terminator is shown as a broken line in the plot (D). Gaps in the data sets indicate timeswhen no in situ measurements were transmitted by the balloon (Sagdeev et al 1986b).

Improved measurements of atmospheric temperaturewere made in the full range between ∼63 km and thesurface. The temperature rises monotonically from upto down but an inversion was found near 61 km. Thelapse rate is not constant; its variations confirmed theexistence of two convective zones, one below 15 km,

another between 45 and 58 km (unlike our planet whereonly one convective zone exists, between ∼10 km and thesurface).

Cloud particles were studied in different ways witha set of sensors that measured size distribution, numberdensity and chemical composition. Sulfuric acid (H2SO4)



as one of the constituents of the Venusian cloud particleswas detected by in situ measurements. This compounddominates the upper and probably the middle clouds(down to 54 km) but below this level cloud particles mostlyconsist of a phosphorus compound (probably H3PO4).

A UV lamp and spectrometer were used to study thestrong UV spectral band created by small traces of SO2

which was found by earlier probes by another method(gas chromatography). Analysis of the new measurementsconfirmed the presence of SO2 but gave different quantitiesand covered larger height ranges.

Measurements of soil composition were madeafter landing by two methods. One of them (x-ray fluorescent spectrometry on VEGA 2) showed thatelemental composition is like that of very ancient lunarrocks (see also VENUS: SURFACE). Another (based on K, U andTh abundances) in both places showed rocks like toleitbasalts.

The VEGAdescent probes were the last in a long set ofSoviet VENERA missions but most of the studies resultedin new questions that require new experiments. However,it is impossible to say at the moment where and when newsteps will be made for prolongation of the study of Venus.

Comet HalleyCOMETARY NUCLEI are probably samples of pristine solarsystem material and their study may be important for anunderstanding of the origin of the solar system. However,they are practically not available for observations fromthe Earth; only products of their evaporation (coma andtail) are observable. Spacecraft give a unique possibility ofapproach to the cometary nucleus and investigation of it asa spatially resolved object. Comet Halley was selected forthe first try. At every periapsis passage (once per 76 yr) itis available for astronomical observations but the relativeposition of the comet and Earth is not same in differentapparitions. For example, the passage of 1986 was notfavorable, because the comet periapsis point and Earthwere nearly in opposite directions from the Sun. So itwas reasonable to use spacecraft during this apparition.Comet Halley was especially attractive for such a missionbecause it is a periodic comet with a well-known orbit, itis the brightest among the periodical comets and it is theonly periodic comet having physical properties like thoseof a young comet.

In reality not one but five spacecraft were sent to thecomet Halley: Soviet VEGA 1, VEGA 2, European GIOTTO

and Japan SUISEI and SAKIGAKE. The VEGA cometary probeswere the largest and had more scientific instrumentsthan the European and Japanese craft put together. Themass of every VEGA spacecraft (figure 2) was about4500 kg, including about 2500 kg for the cometary module.There were two sorts of experiments on VEGA cometarymodules: (a) camera and other optical instruments forremote sensing studies of the nucleus and inner comaand (b) instruments for contact measurements in cometarygas, dust and plasma. Imaging of the comet from thespacecraft required a steerable platform which could be

Figure 2. The Halley probe of the VEGA spacecraft , showingthe locations of some subsystems and scientific instruments:1, solar panels; 2, pencil-beam antenna; 3, omnidirectionalantenna; 4, steerable platform; 5, the pointing sensor; 6, camerasystem TVS; 7, infrared spectrometer IKS;8, ultraviolet–visible–near-infrared spectrometer a; 9, dustparticle sensor SP-1; 10, particle composition analyser PUMA;11, plasma composition analyser PLASMAG; 12, dust protectionshield; 13, astro-orientation unit. The Venus probe is not shown.Its location before separation is on the top of the Halley probe.

automatically pointed with high accuracy to the cometarynucleus. A TV camera, IR spectrometer and UV–Visible–near-IR spectrometer were placed on this platform. Themost difficult task was to locate the nucleus, with its lowalbedo, against the background of the bright coma withgas and dust jets, and this problem was solved successfully.The fly-by speed was extremely high (∼78 km s−1), becausethe direction of orbital motion of the comet is oppositeto that of the Earth, and thus of the spacecraft. Thishigh speed resulted in a strong dust hazard and requiredprotection of parts of the spacecraft.

VEGA 1 passed through the comet Halley coma on 6March 1986 at a nearest distance to the nucleus of 8890 km,VEGA2 did it 3 days later at a distance of 8030 km. Most ofthe data were sent to Earth in real time , because loss of thespacecraft due to the dust hazard was possible. However,this did not happen; both of the spacecraft were alive afterthe fly-by, although the solar-battery current dropped by50%. Figure 3 shows how comet Halley looked from thespacecraft VEGA2 on 6 March 1986. Results of the analysisof images, spectra, and in situ data obtained are brieflyoutlined below.

It was concluded that the nucleus is a single solidbody. Its shape is irregular (like a potato); the size isapproximately 8 km × 8 km × 16 km. It has a very low



Figure 3. How comet Halley looked from spacecraft VEGA 2flying through its coma. (a) The distance to the nucleusd = 29 000 km, the Sun is on the left, the nucleus is near the rightedge of the bright dust fan directed towards the Sun; the framesize is 228 km × 308 km. (b) d = 8000 km, the Sun is above theplane of the image (phase angle ∼30), the frame size is62 km × 84 km; the upper part of the visible spot is the nucleus,the lower is a dust jet. (c) The Sun is to the right, the nucleus is tothe left; the distance and size are the same as in (a). (Moroz 1989).

reflectivity; the geometrical albedo is ∼0.04. So it is oneof the darkest bodies of the solar system. Dust is emittedby its surface from the sunlit side. The dust flow consistsmainly of narrow jets. The thermal IR radiation of thenucleus was measured by an infrared device. Its valuecorresponds to a temperature of about 380 K near thesubsolar point. This was unexpected because a lot ofwater ice with a temperature about 200 K should existin the upper layers of the nucleus to provide the H2Oflux observed by the same instrument and also in many

other experiments. This discrepancy may be explained bya patchy structure of the surface (icy and rocky patches)or by a porous mantle which may be hot on the upperboundary and cold on the lower. The mass of nucleus wasestimated as 3× 1014 t and density about 0.6 t m−3.

Indirect estimates of the chemical composition ofthe nucleus are possible on the basis of measurements ofthe chemical composition of gas and dust going out fromthe nucleus surface to the coma. H2O was identified as themost abundant constituent with a production rate of about1030 molecules s−1. Organic substance outflow (neverobserved earlier) was identified using IR spectra obtainedby VEGA 1. Its production rate may be comparablewith that of H2O. CO2, CO and HCHO were observedwith same instrument as parent species with mixingratios of a few per cent. It is important that mainlysecondary components arising from the decay of suchparent molecules were observed earlier from the Earth.Detailed observations of secondary species with the VEGAspacecraft revealed their production rates, geometry ofejection, lifetimes of species and their parent and speciesvelocities.

Neutral atoms and molecules in the coma areconverted to ions by photoionization. The SOLAR WIND

when it encounters them creates a shock wave that wasmet by the spacecraft at a distance of ∼106 km fromthe nucleus. Nearer, at a distance ∼105 km, anothercharacteristic boundary (not predicted by theory) was met;later it was called the ‘cometopause’. Cometary ionsdominate the plasma inside this. The plasma composition,its energy distribution, plasma waves and magnetic fieldswere measured.

Grains of different masses (from∼10−19 to∼10−11 kg)were counted by a few VEGA instruments. The dust massproduction rate was estimated. One of the instrumentsprovided measurements of the chemical composition ofindividual particles and showed its large diversity. Someof them contain simultaneously C, H, O, N as waspredicted for interstellar particles; others are icy, silicateor metallic bearing. Very small particles (<10−14 g) aredense (1 and 2.5 g cm−3 for organic and silicate particles,respectively), while optically active particles (>10−10 g) areof low density (0.3 g cm−3) and of a complicated structure.

VEGA’s flight to comet Halley was an important partof coordinated international efforts giving a very strongimpulse to studies of comets and other small bodies of thesolar system.

BibliographyMoroz V I 1989 Halley’s comet (part II): space studies

Highlights Astron. 8 17–31Sagdeev R Z, Blamont J, Galeev A A, Moroz V I, Shapiro

V D, Schevchenko V I and Szego K 1986a Vegaspacecraft encounter with comet Halley Nature 321259–62

Sagdeev R Z et al 1986b Overview of VEGA balloon in situmeteorological measurements Science 321 1411–4

V Moroz


Vela E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Vela(the Sails; abbrev. Vel, gen. Velorum; area 500 sq. deg.)a southern constellation which lies between Antlia–Pyxisand Carina, and culminates at midnight in mid February.It was introduced by the French astronomer Nicolas L deLacaille (1713–62), who charted the southern sky in 1751–2, from stars that formed part of the ancient constellationof Argo Navis (the Ship), which had been included byPtolemy (c. AD 100–175) in the Almagest.

The brightest stars in Vela are γ Velorum (Regor), amultiple system consisting of the brightest Wolf–Rayet starin the sky (bluish-white, WC8 + O9, range 1.81–1.87) anda bluish-white (B2) component, magnitude 4.3, separation41′′, the former of which has an unseen companion thatrevolves around it in 78.5 days, and a fourth, bluish-white(B6) component, magnitude 7.7, separation 63′′, δVelorum,a very close binary with a white (A1) primary, magnitude2.0, and a fainter secondary, magnitude 5.6, separation0.74′′, λ Velorum (Suhail), magnitude 2.2, κ Velorum(Markeb), magnitude 2.5, and µ Velorum, another binary,with yellow (G5 and G2) components, magnitudes 2.9 and5.9, separation 2.0′′. There are nine other stars brighterthan magnitude 4.0. δ and κ Velorum together with ι andε Carinae make up as asterism called the False Cross, asit is sometimes confused with the constellation Crux (theSouthern Cross).

The Milky Way passes through Vela and theconstellation contains more than 40 open clusters andmany planetary nebulae, including IC 2391, an opencluster of more than 20 stars scattered across a 1 field,the brightest of which is o Velorum, magnitude 3.6, andNGC 3132 (the Eight-Burst Nebula), a ninth-magnitudeplanetary nebula with a tenth-magnitude star at its center.Other interesting objects include NGC 3201, a seventh-magnitude globular cluster, and the vast Gum Nebula, anexpanding emission nebula some 36 across that extendsinto the neighboring constellations ofAntlia, Pyxis, Puppisand Carina, and is thought to be the result of a supernovaexplosion that occurred more than 1 million years ago,inside which is the more recent Vela supernova remnantand at its center the Vela pulsar.

See also: False Cross, Gum Nebula, Vela pulsar, Velasupernova remnant.


Vela Pulsar E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Vela PulsarThe pulsar PSR 0833–45, discovered in 1968. With a periodof only 9.3 ms, it has one of the fastest pulse-rates known,implying that it is one of the youngest pulsars. This isborne out by an observed deceleration in its pulse-rateof 10.7 ns per day, which sets an upper limit to the timeelapsed since it was formed in a supernova explosionof about 11 000 years. In 1977 its pulsations were alsorecorded in visible light, making it one of the first opticalpulsars to be confirmed.

The Vela pulsar is located in the southern Milky Way,2.8 south of the galactic equator and approximately at thecentre of a triangle formed by the stars ζ Puppis, γ Velorumand λ Velorum. It is surrounded by extensive nebulositymore than 5 across, also a product of the supernova, and isassociated with a moderately strong radio source, Vela X;its distance has been estimated at between 1300 and 1600light-years. The Vela pulsar and supernova remnant liewithin a much older and larger supernova remnant, theGum Nebula.


Vela Supernova Remnant E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Vela Supernova RemnantA supernova remnant in the constellation Vela, positionRA 08 h 34 m, dec. −45 45′. It extends to nearly 5 indiameter, and consists of material expelled by a supernovaan estimated 11 000 years ago. The core of the supernovaremains as the Vela Pulsar.

See: Vela Pulsar.


Velikovsky, Immanuel (1895–1979) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Velikovsky, Immanuel (1895–1979)Physician, psychoanalyst, cosmologist and writer, bornin Vitebsk, Russia. His book Worlds in Collision claimedfrom analysis of mythology and scientific fragments thata breakaway piece of the planet Jupiter had collided withthe Earth, causing various disasters recorded in the Bible,and ending as the planet Venus. None of his theories stoodup to analysis and further exploration.


Venera (Missions) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Venera (Missions)Veneras 1 to 16 are the Russian (Soviet) spacecraft missionsto VENUS launched in the period 1961–1983 (table 1). Venuswas also studied by the VEGA MISSION.

The first three Venera (hereafter V) missions, V1–V3,were developed by the space center headed by SergeyKorolyov, father of the Russian space industry (1906–1966). (This space center is now a division of the EnergiaAssociation.) Since 1967, spacecraft for missions to planetshave been developed by the Lavochkin Association. V1–V8 are the first generation of the Venera spacecraft. Thespacecraft mass varied from 900 to 1200 kg. The three-stage Soyuz missile, with the first two stages similar tothose used for the first manned flights to space, launcheda spacecraft and the fourth stage into Earth orbit. Then thefourth stage accelerated the spacecraft and moved it intothe interplanetary orbit to Venus. Due to the insufficientreliability of the fourth stage and some subsystems in theearly 1960s, some spacecraft were lost or their programswere not completed in full. However, the first generationspacecraft made the first soft entries into the atmosphereand landed on the surface of Venus.

V9 to 16 (and also Vega 1 and 2) represent the secondgeneration of missions to Venus. Their mass was 5000 kg,and they were launched by the Proton rocket. Thosemissions were successful, reliable and, along with ground-based observations and the USA Mariner 2, 5 and 10,Pioneer Venus and Magellan spacecraft, they provided thebasis of our current knowledge of Venus.

Scientific objectivesThe objectives of these missions were to study (i) thesurface relief and properties of rocks, (ii) the ther-mal structure and chemical composition of the loweratmosphere (h < 65 km), (iii) the atmosphere andionosphere above the cloud layer (h > 65 km), (iv) themagnetic field and plasma environment near Venus, and(v) the solar wind, interplanetary magnetic field, cosmicrays and micrometeorites during the cruise phase.

Scientific instrumentationV2. Magnetometer, plasma detectors, cosmic ray and

micrometeorite sensors, telecamera, infrared (5–40 µm)and ultraviolet (1700–3500 Å) spectrometers, radiometerin the decimeter range. Only some data during the cruisephase were obtained.

V3. Flyby: magnetometer, plasma detectors andcosmic ray sensors, photometer at H 1216 Å and O 1304 Å.Lander: atmospheric temperature, pressure and densitysensors, chemical gas sensors, photometer, γ -ray counter.Only some data during the cruise phase were obtained.

V4. The same as V3 with an altimeter instead ofthe lander photometer. Results: the first entry and directmeasurements in the atmosphere down to h = 23 km.Pressure at 55–33 km, density at 55–28 km, temperaturesat 55–23 km, atmospheric composition (CO2 > 90%, N2 <

7%). Hydrogen corona, upper limit to atomic oxygen

density, plasma bow shock at 19 000 km from the planet,low density of the nightside plasma, magnetic moment(< 3× 10−4 of the Earth’s value).

V5, 6. Measurements of pressure, density andtemperature (p, ρ, T , respectively) at 55–17 km, chemicalgas sensors.

V7. First soft landing on Venus. Measurements of Tfrom 55 km to the surface.

V8. p, ρ, T from 55 km to the surface, radar altimeterfor determination of altitude, photometer to study thelight conditions and the cloud structure (the lower cloudboundary was observed at 33 km), measurements of windspeed using the Doppler shift of the radiotransmitterfrequency, γ -spectroscopy of the surface rocks.

V9, 10. Lander (1560 kg): γ -spectrometer, γ -densitometer, p, T sensors, camera, accelerometer (p, Tat 90–65 km), mass spectrometer, nephelometer (anglesof 4, 15, 45 and 175), photometer for five broadintervals in the range 0.44–1.16 µm, narrow bandphotometer for 0.8, 0.82 and 0.87 µm (continuum,H2O and CO2 bands, respectively), onboard transmitter(Doppler measurements of wind speed). Orbiter: imagingsystem (0.3–0.4 µm), spectrometer (0.24–0.75 µm), IRspectrometer (1.7–2.8 µm), IR radiometer (8–28 µm),multiband photopolarimeter (0.35–0.8µm), photometer atH 1216 Å with absorption cells, magnetometer, plasmaelectrostatic spectrometer, plasma sensors, cosmic rays,dual-frequency onboard transmitter (radio occultationsand bistatic radar).

V11, 12. Lander: p, T sensors, accelerometer, massspectrometer, gas chromatograph, optical spectropho-tometer (0.45–1.2 µm), nephelometer, x-ray fluorescencespectrometer, low-frequency electromagnetic detector, on-board transmitter (Doppler measurements of wind speed).Flyby: ultraviolet spectrometer (304–1657 Å), plasma sen-sors, cosmic rays, γ -ray bursts.

V13, 14. Color camera, rock sampling, penetrometer,electric conductivity of rocks, x-ray fluorescence spectrom-eters for (a) rocks and (b) aerosols,p, T sensors, accelerom-eter, mass spectrometer, gas chromatograph, spectropho-tometer (0.32–1.2 µm), electromagnetic and microseismicdetectors, electric conductivity of aerosol, onboard trans-mitter (wind speed measurements), γ -ray bursts (cruise).

V15, 16. Synthetic aperture radar, radio altime-ter, Fourier transform spectrometer (5–40 µm), dual-frequency onboard transmitter (radio occultations).

Main scientific resultsRelief. Detailed mapping of the Venus relief was madewith the synthetic aperture radars and radio altimeters(V15, 16). This study covered the vast area north of about30 that is about a quarter of the total surface of Venus.This mapping had a spatial resolution of 1–2 km, whichis sufficient to distinguish various landforms and terrainsand determine the geological processes responsible fortheir formation. This resolution is better than that of 20 kmin a similar study from the Pioneer Venus orbiter in 1979



Table 1. Venera 1 to 16 missions.

# Launch Encounter Comment

1 12/02/1961 — First interplanetary spacecraft; flyby 100 000 km2 12/11/1965 27/02/1966 Flyby 24 000 km3 16/11/1965 01/03/1966 First hard landing4 12/06/1967 18/10/1967 First soft entry (to 23 km) and flyby5 05/01/1969 16/05/1969 Soft entry (to 17 km)6 10/01/1969 17/05/1969 Soft entry (to 17 km)7 17/08/1970 15/12/1970 First soft landing8 27/03/1972 22/07/1972 Soft landing9 08/06/1975 22/10/1975 Soft landing and first orbiter hmin = 1560 km

10 14/06/1975 25/10/1975 Soft landing and orbiter hmin = 1620 km11 09/09/1978 21/12/1978 Soft landing and flyby at 40 000 km12 14/09/1978 25/12/1978 Soft landing and flyby at 40 000 km13 30/10/1981 01/03/1982 Soft landing and flyby14 04/11/1981 05/03/1982 Soft landing and flyby15 02/06/1983 10/10/1983 Orbiter hmin = 1000 km16 07/06/1983 14/10/1983 Orbiter hmin = 1000 km

but not so good as the 0.12 km achieved by the Magellanorbiter seven years later (1990–1994).

The Venus surface (see VENUS: SURFACE) is dominated byplains formed by vast floods of the non-viscous volcaniclavas, probably basalts. About 15% of the mapped areais covered by so-called tessera, which is a specific terrainformed by heavy tectonic deformation of some unknownprecursor terrain. The morphology of numerous ring-likevolcanic plain features, so-called coronae, is evidence oftheir formation due to tectonics and volcanism relatedto the uplifting plumes in the planet’s interior. Severallarge fracture zones sharing many properties with thecontinental rift zones on Earth were found. Many impactcraters of 8 to 150 km diameter show the age of the mappedarea to be 0.5–1 billion years. The main conclusion is thatthe geology on Venus, contrary to that on Earth, is notdriven by plate tectonics.

Panoramas of the landing sites of V9, 10 (figure 1), 13and 14 show the small-scale structure of the surface. Offour sites, only two are similar (V10 and 13). V9 showsa slope of the mountain ridge covered by irregular andsharp-edged stones having sizes up to 0.5–0.7 m. V10 and13 landed on the rolling plain with outcrops of crystallizedmagmatic rocks changed by deep chemical weathering.Steeply sloped uplands with layered structured rocks areseen sloping away. V14 found a smooth region of thestone plain (lowland) composed of relatively fresh fine-grained rocks having a horizontal layered structure whichresembles the accumulation of volcanic tuff.

Composition of rocks. The abundances of Mg, Al, Si,K, Ca, Ti, Mn, S, Cl and Na given as their oxides weremeasured by the x-ray fluorescence spectrometers at theV13, 14 and Vega 2 landing sites. Abundances of thenatural radionuclides (K, U, Th) were measured by thegamma-ray spectrometers from V8, 9, 10, Vega 1 and 2.Both elemental and radionuclide abundances are typicalof various types of erupted basalts, with the exception ofV8 which is more similar to granites.

Mechanical properties of rocks. The density of rocks wasmeasured by the gamma-ray densitometer from V10 tobe 2.8± 0.1 g cm−3, which is typical of crystalline basalts.Determinations of the bearing strength and firmness ofrocks were made by V13 and 14 using three methods: (i)a penetrometer which impacted the ground, (ii) samplingthe rocks by drilling for the x-ray fluorescence analyses,and (iii) accelerometer observations during the shock oflanding. Heavy clays and compacted dust-like sand arethe terrestrial analogs for V13, and volcanic tuffs andfissured rocks are those for V14.

Temperature, pressure and dynamics in the loweratmosphere (0–60 km) were measured by all landing andentry probes. T and p vary from T = 735 K andp = 90 bar near the surface to 260 K and 0.2 bar at60 km. The temperature lapse rates (gradients) may becompared with the adiabatic lapse rate for the CO2 (97%)+ N2 (3%) mixture which varies from −11.2 K km−1 at60 km to−8.0 K km−1 near the surface. The adiabatic lapserate corresponds to the static stability of the atmosphere.Significant deviations from static stability were observedat 50–35 km and 22–7 km. Wind speed (measured usingthe Doppler shift of the radiotransmitter frequency) variesfrom 70 m s−1 at 50 km to 0.5 m s−1 near the surface. Seealso VENUS: ATMOSPHERE.

Abundances of noble gases and isotope ratios weremeasured by the V11–14 mass spectrometers. The totalargon abundance is 100 parts per million (ppm). Theratio of primary isotopes, 36Ar/38Ar, is that on Earth, whilethe radiogenic isotope 40Ar is depleted by a factor of 270relative to the Earth’s value. The abundance of Ne is8 ppm, and 20Ne/22Ne = 12.15 ± 0.1, which is betweenthe ratios found on the Earth’s (10.1± 0.4) and in the solarwind (13.7±0.3). The abundances of Kr and Xe are 35 partsper billion (ppb) and less than 20 ppb, respectively. Theisotope ratios for C and O are equal to the Earth’s ratios.All these values are important for the origin and evolutionof the atmosphere.



Figure 1. Panoramas of the surface of Venus observed from V9 and V10. Each panorama is 180 × 40.

The chemical composition of the atmosphere was studiedby the V12–14 gas chromatographs: N2 ≈ 3% andAr = 100 ppm (both were also measured by the massspectrometers), SO2 = 130 ± 35 ppm, CO varies from 30ppm at 36 km to 17 ppm at 12 km. Detections of someother gases are disputable.

Strong CO2 and H2O bands observed at variousheights with the V11–14 spectrophotometers resulted inan estimate of the full water abundance of 1 g cm−2 abovethe surface (first done at V11). The absorption at 0.37 µmobserved with V14 below 58 km is due to SO2, which variesfrom 50 ppm at 50 km to 10 ppm at 57 km.

Observations with the V15 Fourier spectrometerresulted in H2O abundances of 5–15 ppm at 60 ± 3 km.SO2 abundances at 70 km were equal to 20–40 ppb at thelow and middle latitudes, 1–10 ppb in the polar collar, and100–1000 ppb near the north pole.

Structure of clouds and aerosols. Illumination. Analysisof the solar radiation scattered in the atmosphere andmeasured by the photometers and spectrophotometersonboard the V8–14 landing probes established three mainlayers within the cloud deck structure and the lower cloudboundary at 48 km (near 35 km at V8 and Vega 1 and 2).Optical depths of the layers were derived at each landingsites. The V14 near-UV (0.32–0.39 µm) photometerestablished two absorption layers, above 60 km and below58 km. 90% of the solar near-UV energy is absorbed above60 km and supports strong winds in the upper cloud level.A few per cent of the solar flux reaches the surface.

A nephelometer is an instrument that illuminates avolume of atmospheric medium by a narrow beam andmeasures the scattered radiation at fixed angles. Four-angle nephelometers were used onboard V9 and 10 andone-angle (backscattering) nephelometers on V11, 13 and14. They also established the three-layer structure of themain cloud deck and determined aerosol properties inthose layers.

The chemical composition of clouds was measured bythe V12 and 14 x-ray fluorescence spectrometers, whichdetermined the presence of S, Cl and Fe, with the Fe/Cl

ratio corresponding to FeCl3. Later, on Vega 1 and 2, thismethod was used to measure the abundances of S, Cl and P.Gas chromatographic and mass spectrometric analyses ofaerosol pyrolysis products (Vega 1 and 2) revealed sulfuricacid and sulfur aerosols with the mass ratio 1:0.1. FeCl3

is probably present as very small (0.3 µm) particles andconstitutes 1% of the total aerosol mass. These particlesmay serve as condensation centers for sulfuric acid andexplain the near-UV absorption. The Vega data favor P asphosphoric acid H3PO4 at locations where the lower cloudboundary extends to 35 km. The observed excess of Cl mayindicate AlCl3.

Lightning. Strong and frequent electromagneticpulses at 10–100 kHz observed at V11 to 14 favorlightning on Venus. The Pioneer Venus electromagneticobservations also support lightning. However, of threesets of observations in the visible (V9, 10, and PioneerVenus orbiters, Vega 1 and 2 balloons), there was only onecase for suspecting a thunderstorm. Overall, the problemis uncertain.

Properties of clouds and the atmosphere near the uppercloud boundary (68 km) were studied by the near-UVcamera, visible spectrometer, photopolarimeter, near-IRspectrometer and IR radiometer onboard the V9 and 10orbiters, and Fourier spectrometer (FS) on V15.

FS covered the range 5–40 µm with a spectralresolving powerλ/δλ ≈ 200. 1500 spectra obtained mostlyin the northern hemisphere showed sulfuric acid as amain cloud species at all latitudes. The cloud deck ishomogeneous with a smooth decrease in particle densityat latitudes< 50, and is horizontally variable with a steepvertical decrease at the polar and subpolar regions.

Limb observations of haze above the clouds by the V9and 10 visible spectrometers resulted in vertical profilesof the haze particle size, their number density and eddydiffusion at 70–100 km.

Pressure, temperature and dynamics in the middle atmo-sphere were studied by three methods: (i) accelerometerson the entry and landing probes, (ii) dual-frequency radio



occultations from the V9, 10, 15 and 16 orbiters, and (iii)FS spectra of the CO2 band at 15 µm.

50 and 90 p–T profiles in the range 40–90 km wereobserved with radio occultations from V9, 10 and 15,16, respectively, to study the thermal balance, day–night variations and latitudinal behavior with a specialemphasis on the polar and collar regions. Temperatureprofiles deduced from the FS spectra are different at lowand high latitudes and depend also on local time. Windvelocities (≈ 100 m s−1 at 70 km) estimated from thetemperature profiles vary with local time, showing thehalf-day period typical of the atmospheric tides.

Airglow. Observations with the high-sensitive visiblespectrometers (V9 and 10 orbiters) solved the three-centuries-old problem of the Venus ashen light (nonzerobrightness of the night side) and revealed spectra of thenightglow. Laboratory simulations of these spectra helpedto identify four O2 band systems. Their intensities, verticalprofiles and variations with local time were obtained.These emissions originate near h = 100 km in the reactionO + O + CO2 → O∗2 + CO2 with subsequent quenching andexcitation trasnfer.

Dayglow emissions of H, O, He and O+ were observedwith the V9 and 10 airglow photometers and V11 and12 spectrometers. The densities of these species and thetemperature of the upper atmosphere were obtained fromthe observations.

Profiles of electron density in the ionosphere weremeasured using dual-frequency radio occultations fromthe V9, 10, 15 and 16 orbiters. About 100 profiles wereused to study variations of the density maximum andits height, the scale height above the maximum, and theionopause height with solar zenith angle, solar activity andother geophysical parameters. The nightside ionosphere isespecially variable and often shows a two-peak structure.

The magnetosphere and interaction with the solar windwere studied by two plasma spectrometers and amagnetometer onboard the V9 and 10 orbiters. The shapeand location of the bow shock indicated that the solarwind is diverted almost completely by a magnetic barrier.Charge exchange between the solar wind protons andatmospheric hot oxygen atoms and their photoionizationadd mass to the solar wind. This mass loading wasobserved in the boundary layer of the solar wind flowaround the planet and plays a crucial role in the formationof the magnetosphere. The planetary ion outflow onthe nightside forms the distinct plasma tail. Its polarityis controlled by the local direction of the interplanetarymagnetic field, thus confirming the induced nature of theVenus magnetosphere. Both plasma flow from the daysideand electron precipitations on the nightside support thenightside ionosphere (see also VENUS: INTERACTION WITH SOLAR

WIND).

BibliographyBarsukov V L, BasilevskyAT, Volkov V P and Zharkov V N

(ed) 1992 Venus Geology, Geochemistry, and Geophysics(Tucson, AZ: University of Arizona Press)

Hunten D M, Colin L, Donahue T M and Moroz V I (ed)1983 Venus (Tucson, AZ: University of Arizona Press)

Keldysh M V 1977 Venus exploration with the Venera 9and Venera 10 spacecraft Icarus 30 605–25

Krasnopolsky V A 1986 Photochemistry of the Atmospheres ofMars and Venus (Heidelberg: Springer)

Moroz V I 1981 The atmosphere of Venus Space Sci. Rev. 293–127

Vladimir A Krasnopolsky


Venus E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

VenusVenus is the second planet from the Sun. It orbits theSun at 0.72 times the Earth’s distance, or 108.2 millionkm. Because of its location relative to the Earth, Venusis always seen close to the Sun, at less than 48 angulardistance from it. The planet appears either as a morning‘star’ or as an evening ‘star’. Actually, as in the case ofMercury, for a long time it was thought that there weretwo different bodies (the ancient names for Venus wereEosphorus and Hesperus), until the Greek astronomersrealized that there was only one body. The distance ofVenus to the Earth varies from 41 million km to 257 millionkm. As seen from the Earth, Venus’s apparent diametervaries from 10 to 64.5 arcsec. However, we see the fulldisk illuminated by the Sun only at Superior conjunction.The planet has phases, like the Moon (figure 1). They werefirst seen by Galileo in 1610. Venus is the brightest objectin the sky, after the Sun and the Moon, but the planet’sbrightness is very variable. Venus is the brightest betweenmaximum elongation and inferior conjunction. At inferiorconjunction, which happens about every 19 months, whenVenus’s size is the largest, the disk is completely dark atvisible wavelengths. The brightness of Venus is due notonly to its close proximity to the Sun (Venus receives twicemore solar light as the Earth) and to the Earth but also tothe high reflecting power of the clouds that reflect 75% ofthe solar light.

The rotation of Venus is nearly synchronous. Indeed,the period of rotation of Venus around the Sun is 224.7days, while the planet’s rotational period is 243 days (Earthdays), which is extremely slow, and in the retrogradedirection (i.e. opposite to the direction of the movementof the rotation of the planet around the Sun). Becauseof the combined effects of the slow retrograde rotation ofVenus and of its orbital motion around the sun, the solarday (time between two sunrises) on Venus correspondsto 117 Earth days. The eccentricity of the orbit and theinclination of the planetary rotation axis are both verysmall. Therefore no major seasonal changes are expected.In fact, Venus is the planet with the most circular orbit.It is also the most Earth-like of the terrestrial planets interms of size and bulk properties. The planetary radiusat the equator is 6051.5 km, which corresponds to 95%of the Earth’s radius (6378 km). Furthermore, Venus’smean density (5.24 g cm−3) and surface gravity (8.6 m s−2)are comparable with those of the Earth (5.97 g cm−3 and9.78 m s−2, respectively). However, now that we knowVenus better, we find many more differences from theEarth than was believed only 50 yr ago. In particular,Venus has a very high surface temperature of 740 K (hotenough to melt lead) and a very dense atmosphere (93times that on Earth). The main atmospheric constituentis carbon dioxide, instead of nitrogen for the Earth. Theclouds are made of sulfuric acid and not pure water ason Earth. Besides, the surface of Venus is essentially ofvolcanic origin and does not present evidence for lateralplate tectonics in recent history. Furthermore, Venus has

no intrinsic magnetic field. In addition, Venus does notpossess any satellites.

Earth-based investigationsUntil the first successful space mission to Venus in 1962,all that was known about Venus came from Earth-basedinvestigations. In 1761, M V Lomonosov, who observed atransit of Venus accross the Sun in an attempt to measurethe distance from the Earth to the Sun, was the first todetect a halo around Venus that he interpreted as being dueto the presence of an atmosphere. Spots or markings onthe planet had been seen since about 1666. Around 1930,ultraviolet measurements detected the Y-shaped featuresof the upper clouds. Studies of these markings establishedin 1966 that the atmosphere was rotating in about 5 daysat the cloud tops. These features appear much better inthe ultraviolet than in the visible. The first identificationof an atmospheric constituent was that of CO2 in 1932by spectroscopy at 8000 Å. Later on, H2O was detectedfrom the ground, but also from balloons and aircraft, invisible spectra of the planet and in very variable amounts.Infrared spectroscopy from Earth around 1967 led to thedetections of CO, HCl and HF in small amounts above theclouds. Ultraviolet measurements allowed the detectionof SO2 in 1979. The cloud composition remained unknownfor a long time. It was believed that the clouds were madeof water, as on Earth. It is only around 1974, through theanalysis of the polarization of Venus’s radiation, that avalue for the refractive index of the particles was obtained,n = 1.45, and it was then correctly inferred that the cloudparticles were in fact droplets of concentrated sulfuric acid(H2SO4). This was confirmed later on by space missions,which provided much more information about the clouds.However, it is interesting to note that an explanation forthe ultraviolet markings in the upper clouds is still lacking,although it is suspected that they are related to sulfurchemistry.

Measurements of the radio emission of the planetat 3.15 cm in the late 1950s seemed to indicate that thetemperature at the surface of Venus was much higher thanthat on Earth. As it was believed at that time that Earthand Venus were very similar planets, this came as a bigsurprise, and a confirmation was needed. This was oneof the goals of the first American spacecraft that went toVenus in 1962.

Concerning the surface itself, it is only by radarmeasurements that are not stopped by clouds thatinformation on the surface could be gained. Informationon the rotation of the planet as a solid body came onlywith the first radar studies of Venus with the Goldstoneradio antenna around 1964–9 that allowed the directionand rate of rotation of the solid planet to be obtained. Theextremely slow rotation thus inferred, and the fact thatVenus was rotating in the retrograde direction were othersignificant surprises. It was also found at that time that therotation axis is nearly perpendicular to the ecliptic. Radarobservations from Goldstone, California, and Arecibo,Puerto Rico, at 12.6 cm (S-band), around Venus’s inferior



Figure 1. A schematic view of Venus as observed from Earth atvarious phases. Venus is largest at inferior conjunction, but thenit is very dark. At superior conjunction, Venus is smallest butcompletely illuminated by the Sun. Venus is brightest aroundeast and west elongations. (From Colin L 1983 Venus ed D MHunten, L Colin, T M Donahue and V I Moroz (Tucson, AZ:University of Arizona Press) p 21.)

conjunction, provided the first radar images of portions ofthe surface at 5–20 km resolution.

The studies of the atmosphere made nowadaysby Earth-based telescopes cover all wavelength rangesfrom the ultraviolet to the radio range. Earth-orbitingultraviolet satellites as well as sounding rockets are alsoused regularly to probe the upper atmosphere of theplanet.

As the planet is covered by thick cloud layers, the partof the atmosphere below the clouds is difficult to probe byEarth-based observations. For many years, information onthe deep atmosphere came essentially from space probesthat went into the atmosphere or from occultations ofspacecraft by the atmosphere of the planet during a spacemission. However, in 1984, it was discovered that thenight-side of Venus emits near-infrared radiation comingfrom deep atmospheric layers below the main clouds. Thisproperty has since been used to probe these deep layersusing Earth-based telescopes. It led to measurements ofsulfur compounds (SO2, OCS), CO, H2O and halides (HCl,HF), below the clouds.

Radio measurements that are made today concernmostly the millimeter range, and they pertain to the upperatmosphere of the planet. They provide information onminor species such as CO and H2O and on the temperatureprofile in the upper atmosphere. They allow the variationswith the time of the day to be studied. For instance, largerconcentrations of CO have been found on the night-sidethan on the day-side. Also, Doppler shifts of the COlines have been used to measure wind speeds. Long-termvariations in CO have also been observed by this method.

Space missionsSpace exploration of Venus started in 1962. Since then,there have been many spacecraft explorations with fly-bys and orbiting spacecraft. The most favorable periodsfor launch occur at times of inferior conjunction, whenVenus is closest to the Earth, which happens about every19 months. Venus is the planet that has been the mostexplored by space missions.

The first fly-by was that of the American Mariner 2spacecraft, in 1962, which carried a microwave radiometerto check whether the surface temperature was high. Then,another American spacecraft, Mariner 5, that went toVenus in 1967, confirmed that the surface pressure wasvery high, about 90 atm. Mariner 10 on its way to Mercuryexplored Venus in 1974 (see MARINER MISSIONS). The Sovietsbegan exploration of Venus in 1967 with Venera 4 (firstsuccessful launch) which carried an atmospheric probe.The first successful landing on Venus was achieved byVenera 7 in 1970 (another Soviet spacecraft), and the firstphotographs of the surface were obtained by the Venera9 and 10 Soviet spacecraft in 1975 (see VENERA MISSIONS).Venera 8, 9 and 10 lander capsules carried gamma-ray spectrometers, which allowed the first geochemicalmeasurements to be made. The amounts of radioactiveelements U (uranium), Th (thorium) and K (potassium)were determined.

Then, in 1978, came the American Pioneer Venus1 and 2 missions, with an orbiter on Pioneer Venus 1and entry probes on Pioneer Venus 2. Pioneer Venusorbiter penetrated the upper atmosphere and made massspectrometer measurements. Four probes reached thesurface. The orbiter part of the mission lasted fromDecember 1978 to October 1992. Topographic maps wereobtained by the radar altimeter on Pioneer Venus 1 in 1978with a resolution of about 50 km in horizontal extent andabout 200 m in altitude (see PIONEER VENUS MISSION). TheVenera 11 and 12 spacecraft, which carried descent landers,were also sent to Venus by the Soviet Union in 1978.Altogether, in December 1978, ten separate spacecraftvisited Venus.

The Soviet exploration continued with other Veneramissions (up to Venera 16) and with the Vega 1 and 2spacecraft that were initially designed for an explorationof Venus but were sent to comet Halley as well as toVenus (see VEGASPACE MISSION). The Venera 13 and 14 probes,which landed on Venus in 1982, and the Vega 2 probe,which dropped a landing module in June 1985, carried aninstrument that was able to measure the x-ray fluorescenceof soil samples introduced into a vacuum chamber. Venera13 and 14 obtained some more images of the surface ofVenus, and the Venera 15 and 16 probes provided radarmapping at a resolution of 1–2 km of the northern 25%of the surface of Venus close to the north pole, starting in1983.

Another major step in the exploration of Venus wasachieved by the American Magellan Radar Mappingspacecraft that was launched from the space shuttleAtlantis in 1989. Between 1990 and 1992, it mapped 98% ofthe surface at resolutions between 120 and 300 m. It alsoprovided altimetric, radiometric and gravity data. Thefinal orbit occurred on October 1994. The radar maps arespectacular. TheAmerican spacecraft, Galileo, designed toexplore the planet Jupiter, accomplished a fly-by of Venusin February 1990. It has allowed us, in particular, to studythe deep clouds of Venus from near-infrared imaging ofthe night-side.



The atmosphereComposition and temperatureOur current knowledge of the composition of theatmosphere of Venus comes from a number of in situanalyses with infrared and ultraviolet spectrometers,mass spectrometers and gas chromatographs on spaceprobes that went into the atmosphere of the planet. Itis also the result of spectroscopy obtained with Earth-based telescopes, airplanes, rockets and satellites. Venushas a very dense atmosphere, with CO2 as the majorconstituent (mixing ratio: 0.965), instead of N2 as onEarth. Minor components are nitrogen (N2), water vapor(H2O), sulfur dioxide (SO2), sulfur monoxide (SO), carbonmonoxide (CO), carbonyl sulfide (COS), sulfuric acid(H2SO4), oxygen (O2), hydrogen chloride (HCl), hydrogenfluoride (HF), argon (Ar), neon (Ne) and krypton (Kr) (seeVENUS: ATMOSPHERE). The atmospheric surface pressure is95 bar, or 95 times that at the surface of the Earth. Thetemperature at the surface is very high: 740 K.

One of the striking characteristics of Venus is itsdryness. At the high surface temperature of Venus,water could not currently exist in the liquid state at thesurface. However, even in the atmosphere, only verysmall amounts of H2O vapor are found. The deuterium-to-hydrogen ratio is very high (about 150 times that onEarth). In contrast, isotopic ratios of C and O are the sameon Venus and Earth.

Concerning the nonradiogenic noble gases 20Ne, 36Arand Kr, 20Ne is about 20 times more abundant on Venusthan on Earth, 36Ar about 70 times more abundant and Kronly 3 times more abundant. Radiogenic argon (40Ar) isabout 4 times less abundant on Venus than on Earth.

The high average surface temperature is somewhatsurprising. Indeed, in spite of the close proximity of Venusto the Sun, because of the high reflecting power of theclouds, only about 25% of the solar flux penetrates intothe atmosphere. However, the high surface temperaturecan be explained by a very efficient greenhouse effect asthe atmosphere of Venus contains gases that are stronginfrared absorbers (CO2, SO2, H2O, OCS) and cloudsthat trap the thermal radiation emitted by the deepatmosphere. This causes an increase in the surfacetemperature of nearly 500 K (compared with the expectedtemperature without an atmosphere).

The measurements of temperature in the atmospherecome from infrared radiometry and radio occultations.Spacecraft radio-occultation measurements indicate thatthe temperature decreases from 740 K at the surfaceto about 240 K at the cloud tops. Temperaturesbelow 35–50 km approximately follow an adiabat. Thisindicates that the temperature structure below theclouds is controlled essentially by atmospheric dynamics(convection). Diurnal and latitudinal variations are verysmall below the clouds.

Above the clouds, in the region called the mesosphere,temperature measurements obtained from Pioneer Venusaround 15 µm have provided maps covering the 60–105 km vertical range. The temperature (averaged over

time) varies from 170 to 330 K. These maps indicate that,at these altitudes, the polar regions are up to 20 K warmerthan the equator, which is very surprising.

At still higher altitudes, in the thermosphere, thetemperature in the day-side is around 300 K, and thereforemuch colder than on Earth (where it is 1000–2000 K).Heating in this part of the atmosphere is caused mainlyby the absorption of ultraviolet photons from the Sun.The difference in temperature with respect to the Earthis due to the greater abundance of CO2 which is veryefficient at radiating heat to space. The temperature variesspectacularly from day to night. At night, it can be as coldas 100 K. The long duration of the days on Venus and thecooling effect of CO2 on the night-side probably combineto produce this strong contrast in temperature.

The homopause is the region of the atmosphere abovewhich the gases are no longer uniformly mixed (gases ofdifferent mass have different scale heights). On Venus,this occurs at around 130–145 km altitude. What is foundis a surprisingly low concentration of the dissociationproducts of CO2—O and CO—above the homopause.These products must be quickly removed and transportedto lower altitudes where they recombine (maybe througha catalytic cycle that involves chlorine, which is about 1000times more abundant on Venus than on Earth).

Still higher up, Venus has an ionosphere, whichwas first detected by the radio-occultation experiment onMariner 5 in 1967 and has been studied in more detail byan ion-mass spectrometer that went to the ionosphere andby remote-sensing observations with other instruments onthe Pioneer Venus orbiter. The day-side ionosphere peaksnear 140 km and stops at about 500 km. It is producedprimarily by solar extreme ultraviolet radiation. The mostabundant positive ions are O+

2 , O+ and CO+2 . O+

2 (and notCO+

2 ) is the dominant ion up to about 190 km becauseof the rapid reaction rate of CO2 + O. Higher up, O+

dominates. Quite surprisingly, the night-side ionospherecan sometimes be rather significant. This happens duringperiods of solar maximum when the day-side ionosphereis so dense and extensive that the solar wind cannotpenetrate the planet’s environment. A large flux of O+

ions then goes to the night-side of the planet and createsa significant night-side ionosphere (the long duration ofthe nights prevents the ions from recombining). At solarminimum, the day-side ionosphere is much less extensive,and the flow of ions to the night-side not as strong (seeVENUS: INTERACTION WITH SOLAR WIND).

CloudsAlthough the upper part of the cloud layers can beobserved and monitored from the Earth, the verticalstructure of the clouds can be studied only during spacemissions. Probes and the Vega 1 balloons have gonethrough the cloud layers, and the Galileo spacecraft hasimaged the clouds on the night-side of the planet around40–50 km altitude during its fly-by of Venus.

Clouds cover Venus globally in generally three layersbetween 48 and 68 km. At some times and at some



Figure 2. A possible model for the Venusian sulfur cycle (fromLewis J S and Prinn R G 1984 Planets and their Atmospheres:Origin and Evolution (New York: Academic).

locations, the lower two layers may combine. However,the variability in structure is much less important than inthe Earth’s clouds. Hazes are present above and below theclouds. The total cloud optical depth has been estimatedto be between 25 and 35. Their total mass is relativelysmall; the Venus clouds are more like the hazes on Earth.The droplets are about 2 µm in size, but the particle sizedistributions differ from layer to layer.

The clouds are made mostly of concentrated sulfuricacid, with about 75% by weight sulfuric acid and 25%water vapor. They resemble terrestrial acid rain (causedby pollution). This composition is probably the result ofphotochemical production of SO3 from SO2. H2O and SO3

then react to make H2SO4 (figure 2). H2SO4 is thermallydecomposed into SO3 and H2O in the lower atmosphere.Chemical reactions with the surface should be removingsulfur quickly from the atmosphere, maybe through theformation of anhydrite (CaSO4). Some sulfur must goback into the atmosphere from the decomposition of sulfurcompounds present in the crust, such as pyrite (FeS2). Toexplain the presence of permanent cloud layers of sulfuricacid on Venus, a source of sulfur (continuous or episodic)is needed, and it is postulated that surface volcanism playsan essential role in the release of sulfur gases from the crust(see also CLOUDS IN PLANETARY ATMOSPHERES).

Atmospheric circulationThe atmospheric circulation of Venus is very differentfrom that on Earth (see EARTH’S ATMOSPHERE). This is notsurprising given the different rotation rates, the differentatmospheric composition and structure, and the lack ofseasons. Furthermore, on Venus, solar light is mainlydeposited within and above the upper cloud layer which

is always present, whereas on Earth it is mostly absorbedat the surface.

What we know about the global circulation in theatmosphere of Venus comes from various sources ofinformation. Direct sources are cloud-feature tracking andthe radio tracking of descending probes (Doppler trackingof Veneras 8, 9, 10 and 12 and interferometric tracking ofthe Pioneer Venus probes) and of balloons that went intothe atmosphere (Vega mission). Doppler-shifted emissionlines from atmospheric gases (such as CO2 and CO) alsoprovide direct measurements of wind speeds. Spatialvariations in gaseous abundances are also used as tracersof the global circulation. In a more indirect way, theoreticalanalyses of planet-wide variations in temperature at agiven pressure, obtained by radio occultations and IRradiometry, have been used. However, winds can bedetermined in this manner only if the atmosphere is incyclostrophic balance, a condition which, on Venus, occursin the non-equatorial regions below 60 km altitude.

As we have seen, Venus rotates extremely slowly,in 243 days, whereas the atmosphere at the cloud levels(around 60 km altitude) rotates in 4–5 days, whichcorresponds to wind speeds of 100 m s−1, in the sameretrograde direction as the planet. This zonal (alonga line of constant latitude) superrotation of the entireatmosphere of Venus from about 10 km up to about 100 km,which is so different from what is found on Earth, is themain mode of circulation of the atmosphere.

The zonal winds are not constant with altitude,however. They decrease regularly from a speed of100 m s−1 at the cloud tops to values close to 1 m s−1 nearthe surface. However, there exist some layers of relativelyconstant wind speeds that alternate with regions of largeincrease in velocity. Differences in zonal wind speeds withtime of the day also exist.

Above the clouds, zonal winds generally decreasewith altitude up to about 100 km, where they reachvery small values. At still higher altitudes, a verydifferent circulation regime exists. The global circulationis there dominated by a strong flow from the day-sidehemisphere to the night-side hemisphere (subsolar toantisolar circulation). Peak velocities reach approximately250 m s−1 across the terminator. This circulation, which isalso different from that existing on Earth, is a consequenceof the high temperature contrasts between the day-sideand the night-side above 150 km. With only a strongsubsolar-to-antisolar flow, one would expect upwellingcentered on the subsolar point and subsidence centeredon the antisolar point, with a return flow around 70–90 km altitude. However, subsidence does not occurat midnight but, instead, at around 2–3 a.m. local time.This asymmetry may be due to the superposition of thesubsolar-to-antisolar flow and of some zonal retrogradewinds (these two flows add across the evening terminator,but go in opposite directions at the morning terminator).

There is, in addition, a slow meridional (in theequator-to-pole direction) circulation that may consist ofHadley cells, as on Earth. Cloud-tracked winds from



Figure 3. A possible model for the general circulation of theVenusian atmosphere (from Schubert G 1983 Venus ed D MHunten, L Colin, T M Donahue and V I Moroz (Tucson, AZ:University of Arizona Press) p 730.)

Pioneer Venus and Mariner 10 have shown equator-to-polevelocities of about 5 m s−1 in each hemisphere. However,the interferometric tracking of the Pioneer Venus probeshas revealed a complex altitude variation, with changesin magnitude and in direction. Interpretations of theseobservations and theoretical considerations suggest thatthe Pioneer Venus probes went through several verticallylayered large meridional cells that each extend from theequator to high latitudes. There may be three differentcells: a Hadley cell near the surface, a weak indirect cell inthe stable layer below the clouds (between 30 and 50 km)and a Hadley cell at the cloud level (that may be ‘thermallyindirect’) (figure 3).

There are many indications of wave motions ondifferent scales in the atmosphere of Venus. In particular,the relative stability of the large-scale dark markings inthe ultraviolet photographs may be due to planetary-scalewaves drifting slowly with respect to the atmosphericsuperrotation. The thermal tides that follow the motionof the Sun in Venus’s sky (the Sun rises in the west andsets in the east) may also play an important role. Thereis also evidence for small-scale turbulence. One of themain problems in Venus’s global circulation is to explainthe existence and maintenance of the superrotation. Theeddies that have been observed in the lower atmospheremay, in combination with the mean meridional circulationand the planetary-scale waves, help in providing theupward transport of angular momentum required toexplain Venus superrotation.

Origin and evolutionAlthough giant planets have retained most of theirprimitive atmosphere, this is not the case for the terrestrialplanets Venus, the Earth and Mars that have lost theirprimitive atmosphere, probably through massive escape,and have acquired a secondary atmosphere. A large partis made of gases outgassed from the interior which isheated during planetary formation by contraction and bydecay of natural radioactive elements. During periods oflarge-scale volcanism, CO2, SO2, H2O and other volatilesare injected into the atmospheres. An extra componentcomes from cometary and meteoritic impacts (for theEarth, biological processes also play a role). Gases can beremoved by atmospheric escape (solar wind interaction),interaction with the surface or major impacts. See alsoPLANETARY ATMOSPHERES.

As Venus and Earth were formed in approximatelythe same region of the solar nebula, they may have hadthe same global composition. There are some indicationthat this may be the case. Indeed, the total estimatedoxidized carbon inventories on Venus and Earth arecomparable, although on Earth the carbon is mainlypresent in carbonate rocks at the surface of the planetand in the oceans while, on Venus, it is mainly presentin the atmosphere, where it can accumulate because of theabsence of liquid water. Nitrogen inventories on Earth andVenus are also comparable.

The underabundance of radiogenic argon 40Ar(produced by the decay of 40K—potassium) could be due todifferent outgassing rates and/or duration on Venus andon Earth. If it is due essentially to a shorter duration ofoutgassing on Venus than on Earth, then the measurementwould indicate that most of Venus’s outgassing happenedduring the first billion years.

The high values of the nonradiogenic noble gases20Ne, 36Ar and Kr on Venus compared with Earth havenot yet found a completely satisfactory explanation. Thepostulated early episode of massive supersonic escape ofthe early hydrogen-rich atmospheres by enhanced extremeultraviolet radiation from the young Sun would probablyhave carried along the noble gases on both Venus and theEarth, in spite of their high molecular weight. The onlymajor difference in the evolution of the two planets maycome from the giant impact with a planetesimal that musthave occurred on Earth (but not on Venus) to explain theMoon’s formation. This would have produced anothermajor erosion of the Earth’s atmosphere, including thenoble gases, and, according to some scientists, it couldbe the main reason for smaller abundances of noble gaseson Earth than on Venus (except for Xe, which may be aparticular case because of its higher molecular weight).The current endowment in these noble gases for theEarth would come essentially from subsequent periods ofoutgassing from the interior, with, maybe, some additionalcontribution of less destructive and more recent cometaryor meteoritic impacts that could enrich the atmosphere innoble gases. More measurements of noble gas abundances



on Venus, as well as measurements in comets, are neededto enable the further development of these hypotheses.

Venus’s atmosphere currently contains about 100 000times less water than is found in the oceans andatmosphere of the Earth. Venus could originally have hadthe same amount of water as the Earth. A significant lossof water could have occurred on Venus, but not on Earth,because of a runaway greenhouse effect due to the muchhigher temperatures of Venus that would have vaporizedthe water. The water molecules dissociate in the upperatmosphere, and hydrogen then escapes. After a periodof massive escape, it has been computed that fractionationwould start (the heavier deuterium D does not escape aseasily as the lighter H) after the amount of water has beenreduced to the equivalent of a layer of water a few metersthick covering the surface of Venus. The very high value ofthe D/H ratio measured on Venus could therefore indicatethat much larger amounts of water were present in the past,with, maybe, the possibility that OCEANS once existed on theplanet. However, such calculations are not simple, as it ishard to track the amount of water present at a given time.Indeed, during the more recent past, significant ejectionsof water could have been produced by a major volcanicepisode or by a massive outgassing event. Furthermore,cometary impacts continually replenish the atmospherewith relatively weakly deuterated water.

Surface and interior of VenusLittle was known about the surface until the 1970s.After the first observations by Earth-based Goldstone andArecibo radar mapping in the 1960s, global-scale radarmappings of large parts of the surface were carried out byPioneer Venus in 1978 and Venera 15 and 16 in 1983. Thiswas then followed by the US Magellan mission in 1990–4that has provided radar images, altimetric and radiometricdata at a remarkable spatial resolution of 120–300 m for98% of the surface of Venus.

Main characteristics of the surfaceThe radar altimetric data indicate that more than 90% ofthe surface has an elevation between −1 and +2.5 km,compared with Venus’s reference mean radius of 6051.8km. The surface is dominated by plains. One candistinguish the lowland plains (or planitiae), such as theAtalanta Planitia, that have elevations between 1 and 2 kmbelow the mean Venusian radius and that cover about27% of the surface, and the upland or rolling plains thatconstitute 65% of the surface and have elevations between0 and 2 km above the mean surface radius. The rest (about8% of the surface) is made of highlands with 2–12 kmelevation.

The lowland plains are generally very smooth (darkat radar wavelengths), whereas the upland or rollingplains are a little more rough and possess many small-scale landforms such as scarps, ridges, troughs, hills,channels, . . . . The plains are interrupted by the large-scalelandforms at higher elevations called highlands (see VENUS:

SURFACE).

The highlands are dominated by two continent-sizefeatures: APHRODITE TERRA and ISHTAR TERRA (figure 4). Thelargest one, Aphrodite Terra (about 10 000 km across),is larger than Africa in surface area. It lies mostly inthe southern hemisphere, close to the equator, at 1–5 km elevation. Ishtar Terra, about 5600 km across(about the size of Australia), and located in the northernhemisphere, possesses the highest mountain belts on theplanet, including Maxwell Montes which is up to about12 km high. Following V L Hansen and colleagues,the highlands can be subdivided into three groups: thevolcanic rises, the crustal plateaux and Ishtar Terra. Thevolcanic rises are more than 1000 km across, they rise 1–2.5 km above the surrounding plains and they have gentleslopes. They include the Atla, Beta, Bell, Dione, Imdrand Themis Regiones. The most prominent volcanoeson Venus and numerous coronae (which are circular toelliptical features with tectonic annuli unique to Venus)are found there. The volcanic rises are mostly located inequatorial regions of Venus. The crustal plateaux suchas Ovda and Thetis in Aphrodite Terra, or the Alpha,Phoebe or Tellus Regiones are 1000–3000 km across; theyhave steep slopes and reach elevations of 2–4 km abovethe surrounding plains. They contain regions of veryrough, highly deformed terrain called tesserae, and somevolcanic features, as well as many other structures suchas ridges, fractures and grabens. Ishtar Terra has uniquecharacteristics; it is an irregular platform with a complextopographic profile that includes a high interior plateau,Lakshmi Planum, with mountain belts in its periphery, andsurrounding high tesserae. In addition, its tesserae aredifferent from the ones found elsewhere. They have morecoherent structures.

Thousands of volcanoes have been identified onVenus, from kilometer-size vents to broad shieldshundreds of km across. Volcanoes and other volcanicfeatures are found everywhere, but they are less abundantin the tesserae regions and more abundant in the Beta–Atla–Themis region which covers about 20% of the planet.So far, 167 volcanoes more than 100 km across have beenfound. Examples are Maat Mons in Atla Regio (figure 5)and Theia Mons in Beta Regio, which have diametersin the range 100–200 km and rise to 1–9 km above themean planetary radius. These volcanoes are surroundedby radar-bright (rough) ejecta with circular or ellipticalshapes that sometimes form kinds of festoons. About 300intermediate-size volcanoes have been found (20–100 kmin diameter), including several ‘pancake domes’ thatare steep-sided, circular, flat-top features. The smallestvolcanoes are mostly found in groups called shield fields.

There are about 1000 impact CRATERS on Venus, whichis not much compared with what is found on Mars orthe Moon. The largest crater, Mead, has a diameter of270 km, and the smallest ones have diameters of about1.5 km. The absence of small craters and the fact thatcraters on Venus seem to come in groups are probably dueto the presence of a dense atmosphere around the planetthat has broken the large meteroids and completely burnt



Figure 4. Topography of Venus in a Mercator projection (from Hunten D M, Colin L, Donahue T M and Moroz V I (ed) 1983 Venus(Tucson, AZ: University of Arizona Press) p 1059).

smaller ones. Higher crater densities are usually foundin the plains. Most craters are pristine. The low numberof craters, their distribution and their shapes suggest thatthe surface of Venus cannot be older than 500–800 millionyears on average.

Information on the surface composition comes fromin situ data and from radar data. The geochemical datafrom the Venera 9, 10, 13 and Vega 1 and 2 landersimply a predominantly basaltic composition, with one siteindicating possibly more alkalinic rocks. Venera 8 found amore granitic composition for the soil. Radar backscatteris a function of both roughness and the bulk properties ofthe target material. The radar properties of the lowlandplains indicate surface dielectric constants of 3–8, whichcorresponds to low values of the dielectric constants ofbasalts on Earth. Much higher dielectric constants havebeen found for some regions at high elevation. Theycan be explained by highly conducting materials such asiron-bearing minerals (pyrite (FeS2) or iron oxides such asmagnetite (Fe3O4)) but identification of a specific phasewhich would exist only in a narrow band of elevation isproblematic.

Volcanism and tectonismVolcanic deposits and volcanic centers are very importantcharacteristics of the surface of Venus. The processes ofvolcanism have been largely preserved on this planet,which is not the case for Mars and the Earth (see VOLCANISM

IN THE SOLAR SYSTEM).

The plains are apparently completely covered byvolcanic deposits. Concerning the volcanic centers, onecan distinguish two main categories. The first category ischaracterized by very significant flows that generally comefrom a vent region during repetitive eruptions. It includeslarge and intermediate-size volcanic edifices, flow fieldsand calderas. To explain the formation of centers ofrepetitive volcanic eruption it is necessary to assume thepresence of stable shallow magma reservoirs. The globaleruption rates corresponding to the population of largevolcanoes have been estimated by L S Crumpler andcolleagues to be about 1.7 × 10−2 km3 yr−1 of magma.

The second type concerns regions of the order of 100–200 km in diameter characterized by the presence of manysmall edifices and by relatively limited associated volumesof lava flows. It includes the groups of small-size edificesthat are called shield fields. Some calculations indicate thatthese shield fields form where magma replenishment ratesare about 10−4 km3 yr−1 or less, but there is no estimate ofthe global rate for this category.

Concerning Venus’s resurfacing history, it looks as ifa substantial decrease in area resurfacing rates from about4 km2 yr−1 during plain emplacement to about 0.5 km2yr−1

over the past 290 million years has occurred. Thisreduction has probably been accompanied by a changein volcanic style from flood-type volcanism in the plainsto more localized development of volcanoes, coronae andrifts.

Concerning the types of lava, some indication comesfrom the existence and morphology of some of the



Figure 5. Magellan image showing a perspective view of Maat Mons, with a vertical exaggeration factor of 10. Maat Mons is a 5 km (or8 km above the mean planetary radius) high volcano located in Atla Regio, at the eastern end of Aphrodite Terra (image P-40175 fromthe NASA/JPL Magellan Radar Mapping Mission). This figure is reproduced as Color Plate 45.

LOWERBOUNDARY LAYER

CORE

spinel

plains

plainstessera

Lakshmi

montes

coronae & chasmata

perovskitevolcanic rises

crustalplateaus

ISHTAR TERRA

decayed plumes

crust

mantle residuum

upper boundary layer

upper mantle lithosphere

plumes / diapirs

LEGEND

Figure 6. A possible global tectonic model for Venus (from Hansen V L et al 1997 Venus II ed S W Bougher, D M Hunten and R J Phillips(Tucson, AZ: University of Arizona Press) p 835.) This figure is reproduced as Color Plate 46.

channels observed on Venus. The longest of the variousclasses of channels observed on Venus are called canali.They have remarkably constant width along very longpaths, exceeding 500 km and up to 6800 km (Baltis Vallis).

They are 1–3 km wide, and their depth seems to beless than 50 m. Highly fluid lavas, that have eruptedduring sustained, high discharges, seem best to explainmany of the channel features, particularly the canali.



Possible candidates for this fluid are native sulfur or,preferably, alkali–carbonatite lavas (very rare on Earth)which would have resulted from the melting of the crustaltered by interaction with the CO2–SO2–halogen-richatmosphere. In contrast, the pancake domes and festoonflows correspond to high-viscosity lavas.

Hundreds of coronae are distributed across thesurface, most with diameters of less than 300 km, buta few exceed 1000 km in diameter. They are not seenon any other terrestrial planetary body. Many of themare associated with rift zones, forming linear chainsthousands of kilometers long. In other places, they occurindividually or in clusters in the plains. All coronae havesome association with volcanism. Coronae may representcollapsed domes over large magma chambers.

There are numerous other tectonic surface featureson Venus that have been formed by compression or byextension: ridges, valleys, mountain belts, . . . . In tesserae,one finds ridges and grooves that intersect and formvery chaotic patterns. They probably reflect a complexdeformation history. One also finds rift zones on dome-like features that are linear depressions that probablyformed where Venus’s lithosphere has ruptured owingto horizontal extension. The most important one is BetaRegio, which presents some analogy with the terrestrialEast African Rift.

Geological evolution and the interior of VenusStudies of impact craters on Venus indicate that about80% of the history of Venus is not accessible to us. Thesurface we see is very young. The oldest terrain onVenus is tesserae. They are probably due to intensivetectonic deformation. Other apparently younger tectonicfeatures must have formed during successive episodesof compressional and tensional deformations: fractures,broad ridges, wrinkle ridges. A period during whichseveral stages of extensive volcanism occurred, buryingareas of tesserae and forming the plains we see now,probably occurred afterwards.

Similarities in mean density and moment of inertiafactor between Earth and Venus could indicate similar bulkcompositions and internal structures. Models suggest thatthe interior of Venus is indeed compositionally similarto the Earth, except for small differences in iron, sulfurand oxygen content. Venus has certainly differentiated.It must have a crust, a mantle and a core. Although therelatively large value of the mean density of Venus stronglysuggests that the planet has an iron core, Venus lacks anintrinsic magnetic field that would be strong evidence thatat least part of its core exists in the liquid state, as is the casefor Earth. Most models predict the radii of the core andmantle to be of the order of 3200 and 2800 km, respectively.Venus probably has a lithosphere, as does the Earth, butgravitational data indicate that it most likely does not havethe low-viscosity astenosphere that exists on Earth belowthe lithosphere.

Most of the sites exhibit compositions similar tobasalt, a volcanic rock that is very common on Earth

and that constitutes most of the Earth’s seafloor. Thecrust, therefore, probably formed from melting of anupper mantle with an Earth-like composition. Theoreticalmodels constrained by the observed depths of impactcraters and the spacings of ridges and rifts imply crustalthicknesses in many parts of Venus of 10–20 km, which isless than the average thickness of the Earth’s continentalcrust (about 40 km) but greater than the thickness of theoceanic crust (about 6 km). Gravitational data indicatethat the mean thickness of Venus’s crust is 20–50 km.Alternatively, arguments based on mineralogy and hightopographic elevations suggest crustal thicknesses in someareas of 100 km or more. Seismic probes are needed toobtain more information on the thickness of the crust.

The depth of the lithosphere is even more uncertain.Some think that Venus’s lithosphere may be thicker thanEarth’s, owing to its lower water content in mantle rocks,thus making its surface geology somewhat less active.Others speak of a lithosphere only about 100 km thick,similar to or thinner than Earth’s. They argue that thehigh temperature of the surface of Venus prevents a thicklithosphere from forming because the rocks are near theirmelting points.

Based on the identification and study of the differenttypes of volcanic and tectonic features on Venus, and ongravitational data, it appears that there cannot be on Venusa global plate-tectonic activity as on Earth. There is insteadevidence for active mantle convection with a recycling thatoccurs mainly in the vertical direction through regions ofupwelling and downwelling. According to some models,volcanic centers must be associated with regions of mantleupwelling. The long-lived plumes (which are narrowregions of hot mantle upwelling) that allow sustainederuptions may be at the origin of the large volcanoes.Short-lived plumes may explain coronae. They maybe due to local upwelling currents in Venus’s mantle(figure 6).

One must also take into account the fact that thelithosphere may have been thin early in the historyof Venus and has thickened with time. This wouldresult in differences in the surface response above mantleupwellings. For instance, as has been suggested by somescientists, crustal plateaux could represent the interactionof ancient deep thermal plumes (or hotspots) with athinner lithosphere. In contrast, volcanic rises may haveformed only in the relatively recent past, after Venus’slithosphere had thickened enough to support them.

The gravitational data and the presence of ridge beltsthat indicate compressive stress suggest that the plainsmay correspond to regions of mantle downwelling on abroad scale. Regions of diffuse mantle upwelling wouldresult instead in extension of the lithosphere and formationof coronae and chasmata (the chasmata are arrangementsof troughs and scarps).

There is certainly a very complex time evolution. Theregions of mantle upwellings and mantle downwellingsmay change location with time, which could result forinstance in formation of coronae and chasmata in plain



regions. In contrast, some early-formed crustal plateauxmay have collapsed owing to changes in temperature orto tectonics and could constitute some of the plains we seetoday.

BibliographyBougher S W, Hunten D M and Phillips R J (eds) 1997 Venus

II (Tucson, AZ: University of Arizona Press)Hunten D M, Colin L, Donahue T M and Moroz V I (eds)

1983 Venus (Tucson, AZ: University of Arizona Press)

C de Bergh


Venus: Atmosphere ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS


Venus: AtmosphereVENUS is the nearest planet to the Earth, both in terms ofdistance and in overall size and mass (table 1). However,Venus has lost most of its atmospheric and surface water,probably because it is significantly closer to the Sun. As aresult of this, much of the carbon dioxide which on the Earthhas been processed by the oceans to produce carbonates isstill free in the atmosphere on Venus. This in turn meansthat Venus’ atmosphere is very massive by terrestrialstandards, with a surface pressure of almost 100 bar and asearingly high surface temperature in excess of 730 K.

The rotational axis of Venus is nearly perpendicular tothe ecliptic, and the orbit is nearly circular, so seasonalchanges in the climate are probably very small. TheVenusian year is 224.7 Earth days, longer than the time forVenus to rotate on its axis, which is 243 Earth days. Thesolar day, defined as the time for the Sun to go from noonto noon as seen from the surface of Venus, is about 117Earth days. This very slow rotation of the solid body ofVenus is retrograde, i.e. backwards compared with theother planets, a curious state of affairs which is difficult toexplain in terms of currently accepted models of theevolution of the SOLAR SYSTEM. The resulting absence of astrong Coriolis force at the surface is likely to be a majorfactor in determining the structure and dynamics of theatmosphere. Persistent high winds of the order of 100 ms−1 are observed near the cloud tops, 50 or 60 km above thesurface, where the density of the air is similar to that nearthe ground on the Earth. It is not known reliably how theseare produced.

The surface of Venus is obscured at visiblewavelengths by planet-wide cloud cover, the upper layers

at least consisting of sulfate aerosol similar to the muchthinner layers of volcanic origin found in the terrestrialstratosphere (see VENUS: SURFACE). On Venus, the cloudshave a complex layered structure and, although nevercompletely absent, their coverage is very variable. Theyplay an even larger role in the energy balance of the planetthan clouds do on Earth, through their contribution to theatmospheric ‘GREENHOUSE EFFECT’.

Localized dynamical or ‘weather’ activity on Venus,as far as is known at present, is dominated by four majorphenomena: the cloud-top zonal super-rotation, theultraviolet markings and associated planetary waves,cumulus dynamics in the deeper layers and the double-vortex structures at the poles. These and other types ofactivity, such as atmospheric tides, are no doubt linked toeach other and to the general circulation, but all remainpoorly understood.

Energy balanceAt a distance of 108.2 million km, Venus is closer to theSun than the Earth by a factor of about 2J, and so hasabout twice the incidence of solar energy. It is also muchhotter at the surface, nearly 2J; times more than theterrestrial mean of about 300 K. These facts are notsimple to reconcile, however, because the ubiquitous andhighly reflective cloud cover on Venus reflects 76% ofthe incoming solar flux and this results in a smaller netsolar constant for Venus than for Earth. The high surfacetemperature must, therefore, be due to ‘greenhouse’warming produced by the thick, cloudy atmosphere,possibly augmented by a contribution from the internalheat of the planet.

Table 1. Physical constants for the terrestrial planets.Venus Earth Mars

Orbital and rotational dataMean distance from Sun (km) 1.082 × 108 1.496 × 108 2.279 × 108

Eccentricity 0.0068 0.0167 0.0934Obliquity (deg) 177 23.45 23.98Sidereal period (days) 224.701 365.256 686.980Rotational period (h) 5832.24 23.9345 24.6229Solar day (days) 117 1 1.0287Solar constant (kW m−2) 2.62 1.38 0.594Net heat input (kW m−2) 0.367 0.842 0.499

Solid body dataMass (kg) 4.870 × 1024 5.976 × 1024 6.421 × 1023

Radius (km) 6051.5 6378–6357 3398Surface gravity (m s−2) 8.60 9.78 3.72

Atmospheric dataComposition See table 2Mean molecular weight 43.44 28.98 (dry) 43.49Mean surface temperature (K) 730 288 220Mean surface pressure (N m−2) 92 1 0.007Mass (kg) 4.77 × 1020 5.30 × 1018 ~1016



Figure 1. The computed transmission of Venus’ atmosphere from the surface to space, showing the spectral gaps to be plugged bycloud opacity in order to maintain the observed ‘greenhouse’ effect (V Meadows and D Crisp).

It is not simple to prove that the observedatmospheric conditions can in fact generate such a large‘greenhouse’ effect. The problem is that the massiveamounts of carbon dioxide are very effective at blockingthe emission of thermal infrared radiation, but only atthose wavelengths where the gas has absorption bands,which are far from covering the entire spectrum.Moderate amounts of water vapor are also required, andeven then considerable spectral gaps or ‘windows’remain (figure 1). These could be blocked by the clouds,since liquid or solid absorbers present some opacity atevery wavelength, the details depending on compositionand particle size. The difficulty for early theorists wasthat using clouds to ‘close’ the greenhouse also tended toblock the incoming sunlight, so that the calculatedequilibrium temperature of the surface remained wellbelow that observed.

This problem began to be resolved when it wasrealized that the clouds are made of sulfuric aciddroplets, at least in the higher, most easily measuredlayers. These have the property of being highly absorbingat thermal infrared wavelengths, while being nearlyconservative scatterers in the visible and near-infrared.Thus, the clouds tend to diffuse downwards those of theincoming solar photons that they do not reflect to space,while blocking thermal emission from the loweratmosphere and surface. This explains the result,surprising at the time, that the VENERA landers in the1970s were able to photograph the surface in naturallight. It also means that radiative transfer models,involving weak as well as strong bands of CO2 and H2O,plus those of the minor constituents CO, HCl and SO2,can account for the high surface temperatures by carefulincorporation of the scattering and absorbing propertiesof the clouds.

The total solar energy diffusing through the cloudcover on Venus corresponds to about 17 W cm−2 ofsurface insolation on the average, about 12% of the totalabsorbed by the planet and the atmosphere. The highopacity of the gaseous atmosphere and cloud at longerwavelengths requires the surface to reach temperatureshigh enough to melt zinc before the upwelling flex isintense enough, and at shorter wavelengths, so thatequilibrium is attained. An airless body with the sameALBEDO and at the same distance from the Sun as Venuswould reach equilibrium for a mean surface temperature

of only about 230 K. This 500 K greenhouseenhancement of the surface temperature compares withonly about 30 K on Earth and 10 K on Mars.

CompositionThe primordial atmosphere of Venus which originallyformed with the solid body, like those of the otherTERRESTRIAL PLANETS, was likely to have been lost in thedistant past as the young Sun went through phases ofhigh activity. The present atmosphere would have beenproduced much later by outgassing from the crust, aprocess which we are probably observing today as activevolcanism, and by the influx of cometary and meteoriticmaterial, which is also still going on. The relativecontributions of these distinct sources can, to someextent, be deduced from the data which are graduallybeing accrued on the composition, and in particular theisotopic ratios, in the contemporary terrestrial planetatmospheres and in comets and meteorites.

COMETS are a rich source of volatile compounds suchas carbon dioxide, water vapor, methane and ammonia. Ifthe last of these was the source of the nitrogen nowpresent, and we allow for processes such as theproduction of argon by the decay of radioactivepotassium in the crust, the contemporary atmospherecould all be of external origin. On the other hand, thehigh abundance of sulfur in Venus’ clouds is stronglysuggestive of extensive volcanic activity and volcanoesare also prolific sources of carbon dioxide, nitrogen andthe other gases required to explain present-day Venus.

Apart from carbon dioxide and water vapor, Venus’atmosphere consists primarily of inert gases, particularlynitrogen and argon (table 2). The amount of waterpresent as gas and bound up with sulfuric acid and othercompounds in the clouds is between 10 and 100 000times less than exists in the oceans and atmosphere of theEarth (see EARTH’S ATMOSPHERE). Thus Venus is overallvery dry compared with the Earth while, at the sametime, deuterium is about 100 times more abundant onVenus than Earth. This suggests that Venus may havehad much more water initially, but that most of it hasbeen lost. Loss takes place by dissociation of the water inthe upper atmosphere by solar ultraviolet radiation andthe subsequent escape of the hydrogen. Both deuteriumand normal hydrogen escape from the atmosphere whilethere is free water on the surface, but the heavier isotope



escapes less efficiently, leading to the observedfractionation.

The loss rate of the water depends strongly on itsabundance in the relatively cool middle atmosphere aswell as the intensity of the solar ultraviolet flux. Modelsof the process suggest that Venus could have lost anocean of present-day terrestrial proportions in only a fewhundred million years. The oxygen produced at the sametime is too massive to escape at any significant rate,according to Jeans’ formula, and must remain on theplanet, presumably most of it bound chemically withinthe crust, mainly as carbonate rocks. As on the Earth, thisprocess would remove atmospheric carbon dioxideefficiently as long as liquid water was available. Once thefree water was all used up, the mixing ratio of watervapor in the upper atmosphere would fall sharply and theloss rates of both forms of hydrogen, and the take-up ofoxygen into minerals, have also declined to the presentrelatively low levels.

In the present-day atmosphere of Venus, chemicalreactions coupled with the transport and radiativeprocesses regulate the abundances of the most importantminor constituents. The most important are the cyclesinvolving water vapor, sulfuric acid and their products,which maintain the cloud layers, and which probably alsoinvolve reactions between the atmosphere and thesurface. However, there are currently very few hard dataon the abundances of reactive species near the surface,and still less on the composition and mineralogy of thesurface itself. It seems likely, however, that the surface isa net sink for sulfur compounds, the combination ofsulfur dioxide with calcite being an example of likelyimportance in this regard. Similar considerations mustapply to interactions between halogen compounds andthe surface, since relatively large (compared with Earth)amounts of hydrogen chloride and hydrogen fluoride arepresent above and below the clouds. Halogen chemistrymay also have an important role in the formation of theclouds themselves.

Another important chemical cycle is that which givesrise to the observed distribution of carbon monoxide. COis very abundant (mixing ratios of the order of a few partsper thousand by volume) in the upper atmosphere ofVenus, as would be expected from the action of solarultraviolet radiation on carbon dioxide. It is stronglydepleted in the cloud layers (<1 ppm), again not toosurprisingly, since it is involved in reactions with SO2and the other species which make up the sulfur cycle.Below the clouds, and near the surface, however, thecarbon monoxide values recovers to around 30 ppm, andshows a marked equator-to-pole gradient. It seems likelythat CO is transported rapidly down from thethermosphere in the polar vortices (see below) to thetroposphere where it is gradually removed by reactions inthe hot lower atmosphere and at the surface.

Table 2. Compositions of the terrestrial planet atmospheres.Venus Earth Mars

Carbon dioxide 0.96 0.0003 0.95Nitrogen 0.035 0.770 0.027Argon 0.00007 0.0093 0.016Water vapor 0.0001(?) 0.01 0.0003Oxygen 0.0013 0.21 0Sulfur dioxide 0.00015 0.2 ppbCarbon monoxide 0.00004 0.12 ppm 0.0007Neon 5 ppm 18 ppm 2.5 ppmHydrogen chloride 0.5 ppm 3 ppb <0.1 ppmHydrogen fluoride 5 ppm 1 ppb

Values are given as fractional abundances except where parts permillion (ppm) or parts per billion (ppb) are stated.

Thermal structureThe solar radiation which penetrates the clouds warmsthe lower atmosphere, which is prevented by the opacityof the overlying layers from cooling by radiation tospace. It therefore forms a deep convective region, thetroposphere (figure 2). This links the high surfacetemperature of around 730 K, produced by the‘greenhouse’ effect as described above, with the level atwhich the temperature is close to the effective bolometrictemperature of Venus (about 230 K), where strongradiative cooling to space can occur. The adiabatic lapserate, which applies when the vertical gradient iscontrolled by convection, is −g/cp (g being theacceleration due to gravity and cp the specific heat atconstant pressure) or about 10 K km−1 for Venus, so thatthe troposphere is around 50 km in vertical extent, muchdeeper than on Earth or Mars.

Above the troposphere the temperature tends to beconstant with height, because the atmosphere here isoptically thin and, to a first approximation, each layertends to find the same equilibrium temperature. Thetemperature is determined by the balance between theabsorption of upwelling infrared from the surface andtroposphere and cooling to space, with no significantabsorption of direct solar energy taking place.

Figure 2. Mean model temperature profiles for the atmospheresof the terrestrial planets.



Figure 3. Measured details of the temperature profiles onVenus and Earth, on a common pressure scale.

On the Earth, this situation is altered by the ozonelayer, which is responsible for substantial heating whichdivides the middle atmosphere into the stratosphere,where temperature rises with height, and the mesosphere,where it falls. There is no corresponding effect on Venus(figure 3), except for small amounts of absorption ofsolar and thermal energy in the near-infrared bands ofwater vapor and carbon dioxide. The two profiles arequite similar near 1 bar and near 10−4 mbar, away fromthe ozone heating region on the Earth which peaks near1 mbar.

The tendency for temperature to increase at the topof the range covered by figure 3 marks the base of thethermosphere, which on Venus begins at a height ofaround 120 km above the surface. Here the atmosphere isvery rarefied, and short-wavelength solar radiation andenergetic particles drive rapid photochemical reactionsand high temperatures by day. When the heating isabsent, efficient radiative cooling by CO2 results in therapid decline of temperature, so there is a sharp gradientacross the terminator, from over 300 K on the illuminatedhemisphere down to 100 K or less on the night side.

Data on the planet-wide variability of temperaturebelow the clouds are sparse and are still mainly limited to

the small number of direct measurements made by thevarious entry probes. Remote sensing of the loweratmosphere has recently been shown to be possible, inparticular at near-infrared wavelengths, as well as in themicrowave, but uncertainties in the distribution ofopacity sources in both spectral regions make thesedifficult to interpret in terms of temperature variations.Such data as exist show that the temperature gradients inthe lower atmosphere are close to adiabatic in thevertical, and close to zero in the horizontal, as would beexpected theoretically from the high opacity and highdensity. (However, there must be temperature structureassociated with the dynamical activity seen in the deepatmosphere cloud structure—cf figure 5—although thishas yet to be measured or even estimated.)

Above the clouds, where the density is lower, moretemperature variability is expected, and this has beenobserved as gradients of the order of several tens ofKelvins on constant-pressure surfaces planet wide. Figure4 shows the time-averaged (72-day mean) globaltemperature field from Pioneer Orbiter infraredsoundings, in which several features clearly related to thegeneral circulation stand out. Note the ‘polar warming’,in which the equator-to-pole temperature gradient leadsto higher temperatures over pole than equator, in spite ofthe fact that the trend in radiative heating is in theopposite direction. The ‘polar collar’ feature stands out inthe meridional cross-section; this is an intense ribbon ofcold air surrounding the pole at about 65º latitude. Theseare both features of dynamical origin, related to the zonalsuper-rotation and the polar vortex respectively. Themeridional average temperature field is dominated by thediurnal variation or solar tide, the temperature cyclewhich is induced by the apparent motion of the Sunoverhead. This contains a whole spectrum of Fouriercomponents, because the forcing is non-sinusoidal; theactual atmospheric response depends on the mean windand the interference between the various components.The solar tide on Venus and classical atmospheric tidaltheory as originally developed for the Earth (see TIDES)can be reconciled, provided that a realistic representationof the zonal wind is incorporated.

CloudsVenus is completely enshrouded by clouds in a complexlayered structure over 30 km deep; their properties aresummarized in table 3. As already noted, the clouds are akey part of a highly interactive climate system (seeCLOUDS IN PLANETARY ATMOSPHERES). ConcentratedH2SO4 droplets have properties which contribute verysignificantly to the atmospheric ‘greenhouse’, byscattering conservatively at short (solar) wavelengthradiation while strongly absorbing at long (planetary)wavelengths. Changes in the optical properties or depthof the cloud layers, for example if the clouds dissipated



Figure 4. Time-averaged temperature fields in the middle atmosphere of Venus: (a) the zonal mean field and (b) the variationsaround a latitude belt from 0º to 30º N, both plotted against pressure and approximate height. The horizontal stepped line representsthe retrieved mean cloud top height.

Table 3. Properties of clouds and dust in the terrestrial planet atmospheres.Venus Earth Mars

Fractional coverage 1.00 0.40 0.05 (cloud); 0–1.0 (dust)Typical average depth 25–40 5–7 0.01–1.00; 0.2–6 (dust)Composition H2SO4·H2O H2O H2O, CO2, magnetite etc (dust)Number density, liquid (cm−3) 50–300 100–1000 0Number density, solid (cm−3) 10–50 0.1–50 30–1000 (near surface)Typical mass loading (g m−3) 0.01–0.1 0.1–10 0.0002–0.1Main production process Chemistry Condensation Condensation, windblown (dust)Equivalent depth (mm) 0.1–0.2 0.03–0.05 1–100Effective radius (µm) 0.2–30 10 0.4–2.5 (dust)Main forms Stratiform Stratiform, cumulus Stratiform, mixed (dust)Temporal variability Slight High HighDominant heat exchange process Radiation Latent heat Radiation

The equivalent length is the estimated thickness of the cloud material if it were deposited on the surface. The effective radius isthe radius of the spherical particles having most nearly the same scattering properties as the cloud at visible wavelengths.

or changed their composition, owing to a reduction in thesupply of SO2 and other source gases from volcanoes, orto an instability in the dynamical regime, would cause thelower atmosphere and surface gradually to cool down (or,conceivably, to become even hotter). Changes intemperature would be likely to further modify theatmospheric circulation and the formation of clouds,

producing feedback which seems more likely to bepositive (accelerating the change) than negative. Atpresent the radiative, dynamical and chemical processesappear to be in balance, but the stability of the currentlyobserved state may be precarious.

Most of our detailed knowledge of the cloudproperties comes from optical measurements: polarimetry



as a function of phase angle from the Earth,nephelometry and particle size measurements from entryprobes, and visible, ultraviolet and infrared remotesensing from orbiter and fly-by spacecraft. It has recentlybeen discovered that near-infrared spectroscopicmeasurements in atmospheric ‘windows’, that iswavelength regions where the main atmospheric gasesare weakly absorbing, penetrate the clouds, in somewindows all the way to the surface. This type ofobservation reveals the cloud morphology all over theplanet and shows that the integrated vertical opacity isvery variable, by factors of 20 or more. Thick andrelatively thin clouds form patterns suggestive of large-scale cumulus dynamics, presumably with the cloudmaterial actively condensing and dissipating in rising andfalling air associated with weather systems (figure 5),although the details are lacking because of a shortage ofhigh-resolution data in space and time.

At the PIONEER VENUS Large Probe entry site, themain cloud deck extends from about 47 to about 67 kmabove the surface, declining gradually at the upperboundary with a scale height of a few km, and with a thinhaze layer some 13 km deep below. Within this verticalstructure, detailed, and presumably variable, layeringoccurs and particles of different sizes congregate atdifferent height levels. The particles range in diameterfrom less than 1 to over 30 µm and tend to a trimodal sizedistribution, with the commonest diameters fallingtowards the ends of the overall range and in the 2–3 µmregion. It is these intermediate size or ‘mode 2’ dropletswhich are visible from outside Venus and for whichspectroscopic, polarimetric and other evidence yields acomposition of 75% H2SO4 and 25% H2O. Thecomposition of the smaller drops, which form an aerosolhaze extending throughout the cloud layer, is unknown.Most of the mass of the clouds is in the biggest drops, forwhich there is some inconclusive evidence of a non-spherical shape, implying a solid composition, perhapscrystalline sulfur. The formation of the cloud droplets canbe explained by a model in which H2O and SO2 (thelatter possibly of volcanic origin) combinephotochemically near the cloud-top level. It is moredifficult to explain the size distribution, particularly theexistence of more than one mode. Compositionalcontrasts and dynamical effects may be at work but onceagain the observations which would elucidate these arelacking.

There is some evidence in infrared maps from thePioneer Venus orbiter of cold, high cloud, presumablycondensed CO2, near the tropopause above the dawnterminator, where the atmosphere is coldest as it comesto the end of the long (>50 h at this altitude) Venus night.

Figure 5. 2.7 µm image of Venus from Galileo–NIMS.

DynamicsAs a result of the slow rotation of the planet, its near-circular orbit and small obliquity, the underlyingcirculation of Venus’ atmosphere is quite simple. TheSun is always above the equator, to within a couple ofdegrees, so the air warmed at low latitudes rises andmoves towards the poles, where it cools and descendsbefore returning equatorwards at lower altitudes.

On top of this simple picture are superimposedvarious complications. One that has been observed formany decades, but remains difficult to explain, is theglobal ‘super-rotation’, which manifests itself in cloudstructure which moves rapidly around the planet in adirection parallel to the equator. The cloud markings,which appear with high contrast through an ultravioletfilter, have their origin at heights 50 or 60 km above thesurface (where the pressure is of the order of 100 mbar)and travel around the equator in 4–5 days, correspondingto speeds near 100 m s−1. This is more than 50 timesfaster than the rotation rate of the surface below.Measurements of the winds below the clouds, andcalculations (from temperature data) of the winds abovethe cloud tops, show that the zonal wind speed declines athigher and lower levels, reaching values near zero atabout 100 km and near the surface respectively.

Direct measurements of the winds 1 m or so abovethe surface by the Russian landers Venera 9 and 10 found



velocities of ≤1 m s−1. Tracking of the Pioneer andVenera landers during their descent showed that there is asteady decrease with height from the 100 m s−1 or soobserved in the ultraviolet markings near the cloud tops.Earth-based observers had earlier shown, by themeasurement of Doppler-shifted emission lines fromatmospheric gases, that the cloud-tracked winds do, infact, apply to mass motions, rather than the phase speedof waves as had also been suggested.

Attempts have been made to explain these high zonalwind speeds on Venus by several mechanisms, all ofwhich fall into one of three main categories, i.e. (i) thegravitational interaction of the Sun with the atmospherictides, (ii) the overhead motion of the Sun in the sky (the‘moving flame’ mechanism) and (iii) the upwardtransport of momentum from the surface. Currentlyprevailing opinion favors a version of mechanism (iii), inwhich momentum from the solid planet is transported bywaves whose interaction with the main flow is complexand in which the mean meridional circulation plays animportant role. Parameterizations have been found whichare able to produce large zonal velocities in dynamicalmodels of the Venusian atmosphere, although of coursethis is not the same as saying that we understand theforcing or dissipation mechanisms responsible for thetransfer of momentum from the surface to the cloud tops.

The cloud motions which trace the zonal winds alsoreveal the pattern of the meridional circulation on Venus.As expected on the simple theoretical grounds outlinedabove, Hadley cells exist in each hemisphere. These areglobal-scale circulation cells characterized by risingmotion all around one constant latitude belt anddescending motion at another. Each cell extends to higherlatitudes than on Earth, in part a consequence of theslower zonal rotation speeds. Near the poles on Venus, acomplex instability develops, resulting in dramatic long-lived wave structures. The polar collar takes the form ofa band of very cold air, some 10 km deep and 1000 km inradius, centered on the pole. Inside the collar,temperatures are some 40 K cooler than outside thefeature. Poleward of the inner edge of the collar lies thepolar dipole, a wavenumber 2 feature consisting of twowell-defined warm regions circulating around the pole.Both the dipole and the collar have so far resistedattempts to model them as normal modes of theatmosphere.

The cloud-tracked winds obtained from PioneerVenus and Mariner 10 both show equator-to-polevelocities of around 5 m s−1 in each hemisphere. Trackingof the Pioneer Venus probes shows winds of thismagnitude at about 50–60 km altitude, with a verycomplicated vertical structure (figure 6). One possibleinterpretation of the alternations in the direction, as wellas the magnitude, of the meridional wind is that thesecould mark the passage of the probe through the differentcomponents of a stack of Hadley cells, each extending

from the equator to high latitudes. The layered eddysources and sinks which could drive the zonal super-rotation may be related to the cell interfaces.

Motions in the deeper atmosphere were observed bynear-infrared imaging carried out by the Galileo probe in1990. The features observed on the night side of theplanet at wavelengths from 1 to 3.5 µm originate in themain cloud deck, illuminated from below by thermalemission from the hot lower atmosphere. The typicalvelocities inferred near the equator were about half asfast as those from UV markings, which is consistent withthe vertical profiles of wind and cloud opacity measuredby the Pioneer and Venera probes, since the cloud layerproviding most of the opacity in the case of the near-IRmarkings is 10–15 km deeper than for the UV markings.Galileo winds feature a zonal jet of more than 100 m s−1

at middle latitudes and equator-to-pole drifts of a fewm s−1.

ConclusionVenus presents an arresting picture. What we appear tosee is an Earth-like planet whose atmosphere failed to

Figure 6. Profiles of the zonal (east-to-west) and meridional(north-to-south) wind on Venus as measured by tracking thedescent of the Pioneer Venus probes.



evolve in the same way, primarily because of the earlyloss of water, and which now supports a thick, primitiveatmosphere in rapid, turbulent motion. Thick, sulfurousclouds, probably originating in volcanic activity on thesurface and possibly requiring present-day volcanism fortheir maintenance, blanket the planet. The resulting‘greenhouse’ effect drives the surface temperature toremarkably high levels.

These conditions make exploration difficult sinceremote sensing is inhibited (but not prevented altogether)by the clouds, while landings are arduous and short lived.Advanced techniques are under development that willallow sample return from Venus’ surface and clouds andsubmarine-like ‘aerobots’ to cruise in the hot, thicklower-atmospheric regions. New remote-sensingtechniques for the lower atmosphere will be applied fromorbit.

The focus of this new generation of Venusatmospheric studies will be an understanding of thedynamical regimes present, the photochemistry of theclouds, the scale of volcanism on Venus and the questionof whether or not our sister planet once had oceans, likethe Earth. Above all, however, will be the need tounderstand the dynamical regimes present. Knowledge ofeven the principle of the origin and maintenance of thezonal super-rotation is lacking; the same is true of thespectacular giant vortices in the high-latitude regions, inparticular the phenomena known as the polar collar andthe polar dipole. The deep atmosphere shows hugeweather systems whose very existence was unsuspecteduntil about a decade ago and for which there exists notheoretical basis at all. How can such a simply rotating,nearby, Earth-like planet be in such an incomprehensiblestate?

Breaking news update (30 April 2002)Scientists have captured the first x-ray view of Venususing NASA’s CHANDRA X-RAY OBSERVATORY. Theobservations provide new information about theatmosphere of Venus. Venus in x-rays looks similar toVenus in visible light, but there are importantdifferences. The optically visible Venus is due to thereflection of sunlight and, for the relative positions ofVenus, Earth and Sun during these observations, shows auniform half-crescent that is brightest toward the middle.The x-ray Venus is slightly less than a half-crescent andbrighter on the limbs. The differences are due to theprocesses by which Venus shines in visible and x-raylight. The x-rays from Venus are produced byfluorescence, rather than reflection. Solar x-rays bombardthe atmosphere of Venus, knock electrons out of theinner parts of the atoms, and excite the atoms to a higherenergy level. The atoms almost immediately return totheir lower energy state with the emission of afluorescent x-ray. A similar process involving ultravioletlight produces the visible light from fluorescent lamps.

For Venus, most of the fluorescent x-rays come fromoxygen and carbon atoms between 120 and 140 km (74to 87 mi) above the planet’s surface. In contrast, theoptical light is reflected from clouds at a height of 50 to70 km (31 to 43 mi). As a result, Venus’ Sun-lithemisphere appears surrounded by an almost-transparentluminous shell in x-rays that appears brightest over thelimb.

BibliographyTwo major surveys of Venus have been published in recentyears. The reader is referred to the atmospheric chapters inthese for a more detailed overview of current knowledge andproblems and for a complete set of references to original work.

Bougher S W, Hunten D M and Phillips R J (eds) 1997 Venus 2(Tucson, AZ: University of Arizona Press)

Hunten D M, Donahue T M and Moroz V (eds) 1983 Venus(Tucson, AZ: University of Arizona Press)

F W Taylor

Venus: Interaction with Solar Wind ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS

The solar wind interaction with VENUS provides thearchetypal interaction of a flowing magnetized PLASMA

with a PLANETARY IONOSPHERE. Mars interacts with thesolar wind in much the same way as does Venus, whilethe rotating plasma in the Saturnian magnetosphere isbelieved to interact similarly with its moon, Titan (seeSATURN: MAGNETOSPHERE INTERACTION WITH TITAN). Theinteraction of the Jovian plasma with its moon Io isintermediate between the cometary and Venus interac-tions (see IO: PLASMA TORUS). From a more academic per-spective, this interaction illustrates how a magnetizedplasma interacts with an unmagnetized plasma, and howmass pickup (of planetary ions, in this case) affects aflowing plasma. Our understanding of the interactionhas been built up over a series of flight missions: Mariner2 and 5, Venera 2, 4, 6, 8–14 and Pioneer Venus. The lat-ter mission remained in orbit for 14 yr investigating thesolar wind interaction with Venus over a complete solarcycle.

Although Venus is famous for its sulfuric acidclouds and sulfur dioxide haze, its main atmosphericconstituent is carbon dioxide of mass 44. Because Venusis similar to the Earth in mass and size (6051 km radius),we might expect a priori its atmospheric behavior to besimilar. Paradoxically, while the lower atmosphere ismuch warmer than the terrestrial atmosphere owing tothe greenhouse effect, the upper atmosphere of Venus,above 100 km, is much cooler than that of Earth becauseof the absence of heating by a magnetosphere at Venus(see VENUS: ATMOSPHERE). As a consequence the scaleheight of the neutral Venus atmosphere is quite small. Inthe cold (cryosphere) nightside it is only a few km. Ingeneral, the atmospheric density falls off according to theequation of hydrostatic equilibrium that balances theupward pressure gradient force with the downwardforce of gravity on that parcel of gas:

where n, m, g, k, T and h are respectively the number ofmolecules per unit volume, their mass, the force of grav-ity, Boltzmann’s constant, their temperature and the alti-tude of the point of interest.

For an isothermal atmosphere the density decreasesas

where Hn = kT/mg. The isothermal description fails atthe highest altitudes (above about 200 km) for some con-stituents, such as atomic oxygen, where nonthermalprocesses, notably dissociative recombination of O2+,increase the average temperature. The important non-thermal processes are both photochemical (e.g. involving

neutralization of ionospheric ions and electrons pro-duced by solar EUV radiation) and ‘mechanical’ (involv-ing collisions of energetic particles from above with theambient gas). Thus atomic oxygen at Venus has a coldcomponent, with a scale height of a few tens of km, anda hot component that extends to high altitudes with ascale height of hundreds of km. This latter componentplays a special role in the solar wind interaction,described below.

As on Earth the Venus upper atmosphere is partial-ly ionized by solar ultraviolet radiation and energeticparticles that enter from the surrounding space. The rateof ionization decreases rapidly with decreasing altitudeat low altitudes where the ionizing radiation and parti-cles are absorbed. It also decreases with increasing alti-tude at high altitudes where the ionizing radiation is constant but the ionizable neutral particles decrease withaltitude. Thus there is a maximum ion production rate atsome altitude hm. The rate of ionization, Q, varies withaltitude for a simple one-component isothermal ionos-phere as

where Qm is the peak production rate and y = (h–hm)/Hnwhere Hn is the neutral scale height and hm is the heightof peak production. This function is sketched for a scaleheight of 25 km in figure 1, representative of oxygen inthe warmer dayside upper atmosphere of Venus. If theionosphere is in photochemical equilibrium in which therecombination of electrons and ions takes place withoutvertical transport, then the rate of recombination is pro-portional to the product of the electron and ion numberdensities which are equal in our simple ionosphere.Hence the electron number density is

where χ is the solar zenith angle and we have referencedthe density to the maximum at the subsolar point, Nm0,and the normalized altitude, z, to its altitude, hm0, andthe scale height, Hn. This function is shown in the bottompanel of figure 1.

At Venus the high-altitude ionospheric electron tem-perature is about 5000 K. The peak of the dayside ionos-phere is at an altitude of about 140 km and has a peaknumber density of just under 106 cm–3. At 400 km thenumber density is about 20 000 cm–3 under solar maxi-mum conditions near the subsolar point. At this altitudethe collision rate is quite low and the electrical conduc-tivity quite high. This is important for the solar windinteraction because the solar wind is a magnetized plas-ma. When it flows against the Venus ionosphere it doesnot generally penetrate below about 400 km unless thesolar wind conditions are disturbed. The ionosphericpressure at 400 km under solar maximum conditions issufficient to stand off the typical dynamic pressure of the

Venus: Interaction with Solar Wind



solar wind. This is not expected to be the usual case atsolar minimum when the ionospheric pressure is less.

The SOLAR WIND is a supersonic flow of electrons andions with a number density near the orbit of Venus ofabout 15 particles per cm3, a velocity of about 440 km s–1,an ion temperature of about 105 K, an electron tempera-ture of about 2–105 K and a magnetic field strength ofabout 10 nT. The velocity of a wave that could compressthe solar wind flow and deflect it around Venus is about100 km s–1 or about 1/4 of the speed of the solar wind.Thus, the deflection has to occur via the formation of astanding shock wave whose strength is measured by theratio of the compressional wave velocity to the incomingflow velocity. This interaction is sketched in figure 2. TheVenus bow shock has a shape similar to Earth’s bowshock shape but, because the obstacle producing theshock is in this case the planetary ionosphere (and not an

internal magnetic field), the size of the bow shock ismuch smaller (see MAGNETOSPHERE OF EARTH: BOW SHOCK).At the nose of the Venus shock the Mach number orshock strength is close to 4 under typical conditions.

In the ideal case when none of the solar wind isabsorbed by Venus and no ions are added to the flow bythe atmosphere of Venus, the shock front, or bow shockas it is called, stands off from the obstacle at a distancesufficient to allow all the material that is compressed anddeflected by the shock to flow around the obstacle. Whenthe shock is strong, the density of the solar wind flow canincrease a factor of 4 across the shock. At solar maximumwhen the pressure of the ionosphere is high, the shockfront is about 2000 km above the surface of Venus at thesubsolar point. If the shock weakens, the density jumpbecomes less and the shock moves away from the planet.In the limit as the Mach number approaches unity, theshock moves away to infinity. At solar minimum the typ-ical ionosphere does not have sufficient pressure to pre-vent some direct interaction of the solar wind with theneutral atmosphere and the shock moves closer to Venusas some solar wind is absorbed by Venus.

The shocked plasma immediately downstream fromthe shock front is simply heated and compressed solarwind whose properties can be calculated by a set of mag-netohydrodynamic equations called the Rankine–Hugoniot equations. However, the flow velocity perpen-dicular to the magnetic field direction drops drasticallyas the subsolar point on the obstacle is approached. Themagnetic field of the solar wind plasma is compressed asthe stagnation point is approached, and the fieldbecomes draped around the obstacle. The thermal veloc-ity of the particles is then able to evacuate the field lines


0

100

200

300

400

0

100

200

300

400

0

100

200

300

400

-6 -5 -4 -3 -2 -1 0

-3 -2 -1 0 1 2 3

0 1 2 3 4 5 6

Log Radiation Flux [I/Io]

Log Ne [cm ]-3

Log Production Function [cm sec ]-1-3

Alti

tude

[km

]

Figure 1. The formation of a Chapman-layer’ ionosphere. Top:UV radiation from Sun drops in intensity as it is absorbed inphotoionization process. Middle: Rate of electron productionwith altitude. As altitude decreases, the rate first increasesowing to the increasing atmospheric density and then decreaseswhen the UV radiation is exhausted. Bottom: The electron den-sity consistent with the production if there is no vertical or hor-izontal transport.

SolarWind

Bow Shock

Magnetosheath

Tail

IonospherePhotoion

SolarRadiation

Tail Rays

Atmospheric Neutrals

Ion Pickup

Ionopause

Figure 2. The solar wind interaction with the Venus ionosphereand neutral atmosphere. The solar UV ionizes the atmosphereand the ionosphere deflects the solar wind. Some neutral atomsare ionized in the solar wind and are picked up by the electricfield.


so that the inner edge of the shocked plasma, or mag-netosheath, becomes a magnetic barrier in which thetransverse pressure is exerted almost totally by the mag-netic field and not by the thermal plasma pressure. Asimilar situation occurs in what is called the ‘depletionlayer’ just outside Earth’s magnetosphere. As sketched infigure 3 the magnetic field wraps around Venus, main-taining a barrier between the flowing solar wind and theionosphere and contributing to the formation of a mag-netic tail in the antisolar direction. The plasma in themagnetosheath accelerates with distance behind Venusand the shock expands and weakens. Eventually farbehind Venus the solar wind that interacted with Venusis indistinguishable from solar wind that did not interact,as long as Venus did not add anything to the solar windflow.

Even though the bow shock marks the upper extentof the pressure wave that Venus launches to deflect thesolar wind, there still are phenomena that can be seen inadvance of the bow shock. These upstream phenomenaconsist of charged particles and waves. The particles areelectrons and ions that either were reflected at the shock

or leak from the hot population behind the shock andmove upstream against the solar wind flow. The geome-try of this region is sketched in figure 4. The waves caneither be generated by the backstreaming particles (andgenerally be convected toward the shock) or be generat-ed at the shock and move upstream. In many respects thewave and particle phenomena in front of the Venus bowshock are the same as those generated in front of the bowshocks of magnetized planets including Earth. Thenature of the obstacle is not important here.

Returning to the obstacle for a moment, we notethat the magnetic barrier has the effect of maintaining acap on the ionosphere. This upper boundary is calledthe ionopause. Ionospheric plasma is not detectedabove the barrier because the atmospheric ions pro-duced there are immediately removed by the interplan-etary electric field (equal to –VswB), either swept awayfrom Venus or deposited into the deeper atmosphere.Two such ions are sketched in figure 2. Nevertheless,most ions are produced within the ionosphere andremain in the ionosphere proper until they recombine.At comets the opposite is true. The recombination rateat high altitudes in a planetary ionosphere is small sothat in steady state ions must flow to low altitudes torecombine. In the subsolar region the typical motion of


MagneticField Lines

Magnetic Barrier

Streamlinesof Solar WindPlasma Flow

Bow Shock

Ionosphere

Figure 3. The solar wind flow and interplanetary magnetic fieldas it interacts with Venus, leading to the formation of a mag-netic barrier.

DownstreamForeshock

Ion Foreshock

Electron Foreshock

TangentField Line

Solar Wind

Upstream Foreshock

Figure 4. The geometry of the electron and ion foreshock atVenus. These regions mark the field lines that intersect the shockand along which electrons and ions can reach a spacecraft at thevelocities to which they have been accelerated.


the ionosphere is downward. There are few ions creat-ed at any altitude in the Venus shadow so that thenightside ionospheric pressure is much less than on thedayside. Away from the subsolar region the pressuregradient induces a supersonic expansion of the upperionosphere into the nightside whenever the ionosphereis present at altitudes of 400 km and above. This flowproduces a significant nightside ionosphere eventhough little ionosphere is produced there. At solarminimum or when the solar wind dynamic pressure ishigh the ionopause is too low to allow horizontal trans-port to the nightside ionosphere, which becomes veryweak as a result.

ComplicationsThe above description is a correct first-order model of theVenus interaction but there are details that need to beadded to complete the picture. We look first at the varia-tions in the ionosphere caused by varying solar windpressure. We then examine the effects of planetary ionmass added to the solar wind flow and finally examinehow the interaction varies with the solar cycle.

The effect of varying pressureThe boundary between the magnetic barrier and theionosphere, the ionopause, is generally a thin currentlayer, at least when the solar wind pressure is low andthe ionopause altitude is high (about 400 km). This thick-ness appears to be about one ionospheric thermal iongyroradius. As mentioned above, the electrical conduc-tivity is so high under these conditions that there is nodiffusion of the draped solar wind magnetic field acrossthe ionopause current layer. The left-hand panel of figure5 shows such a situation illustrated with vertical profilesof the magnetic field and plasma density seen by thePioneer Venus orbiter. Nevertheless, magnetic field canapparently get into the ionosphere in the form of flux

bundles or ropes. These flux ropes cause the large spikesin field strength that are seen low in the ionosphere in theleft-hand panel of figure 5. Observations suggest thatthey are untwisted at high altitudes but become twistedand even kinked as they sink. Figure 6 illustrates the for-


Figure 5. Vertical profiles of the electron density, Ne (symbols),and magnetic field strength, B, (full curve), on days of low (left),moderate (middle) and (high) solar wind pressure.Measurements by the Pioneer Venus orbiter on three early orbitswhen the spacecraft was entering the dayside ionosphere andperiapsis, the closest approach to the planet, were at low alti-tudes.

MagneticBarrier

MagneticBarrier

MagneticBarrier

Ionosphere

Ionosphere

Ionosphere

J

J

J

J

Figure 6. Formation of a flux rope in the Venus ionosphere dueto the sinking of an initially buoyant flux tube from the magnet-ic barrier. In the middle panel the flux tube has become heavierowing to mass addition through photoionization of the neutralionosphere and sinks. The magnetic force due to the curvaturein the plane perpendicular to the page accelerates this motion.


mation of one of these ropes from an originally buoyantflux tube of the magnetic barrier. The twist in the ropeshelps to maintain the coherence of the rope and to bal-ance the outward pressure of the magnetic field. Thetwist is believed to arise as the rope sinks through a ver-tical shear in the horizontal velocity of the ionosphericplasma. Often the magnetic field of the ropes is almost inforce balance by itself so pressure gradient forces in theplasma are not important. Such ropes are called forcefree. A sketch of the magnetic structure of such a fluxrope is shown in figure 7.

Because the ionopause forms where pressure bal-ance with the solar wind pressure occurs, its heightvaries with solar wind conditions. As illustrated by therightmost two panels of figure 5, when the solar windpressure increases to about 20 nPa and the altitude of theionopause current layer falls to 250 km or less, the cur-rent layer thickens and magnetic field diffuses into theionosphere because of ion–neutral and electron–neutralcollisions. The downward ionospheric plasma velocitynear the subsolar region carries the magnetized plasmato low altitude where it recombines and deposits themagnetic flux. Thus the collisional Venus ionosphere canbecome magnetized from the ionopause down to thepeak of the ionosphere. The magnetic flux then collision-ally diffuses through the current layer at the bottom ofthe ionosphere, creating a magnetic field between theionosphere and the planet’s core. The diffusion of mag-netic flux below the dayside ionopause is balanced by theupward diffusion of magnetic field on the nightside.Since field lines in the ionosphere generally have theirends in the solar wind, these field lines on the nightsidehave ends that have moved far down the magnetotail.Hence the field lines are mainly vertical or radial on thenightside and close beneath the ionosphere. Theseregions of radial magnetic field on the nightside preventthe plasma from entering these regions so that theyappear as ‘holes’ in the plasma measurements. The mag-netic pressure in the holes is enhanced by the amountthat the plasma pressure is diminished so that the plas-ma and field pressures are in equilibrium.

If the dayside ionosphere becomes fully magnetized,then the flow of plasma from the dayside to the nightsideis cut off. Under those circumstances the nighttime plas-ma density drops to very low values and the phenome-

non called a ‘disappearing’ ionosphere results. This samesituation prevails at solar minimum when the daysideionospheric pressure is lower than the typical solar windpressure. The magnetic structure of the ionosphere underthese low and high solar wind dynamic pressure condi-tions (relative to ionospheric plasma pressure) is shownin figure 8.

Mass addition to the solar wind flowThe interaction of a comet with the solar wind is wellunderstood (see SOLAR WIND: INTERACTION WITH COMETS).Gas sublimes from the comet nucleus and expandsrapidly into the relative vacuum of space. This gas isionized and swept up by the solar wind. The mass addi-tion to the solar wind in the form of cometary ions slowsit down by the conservation of momentum. The ends ofthe interplanetary field lines threading the cometaryatmosphere are rooted in the undisturbed solar windand thus carried past the comet at the solar wind speed.


Interior Structure of Flux Rope

Figure 7. The magnetic structure of a flux rope. In the interiorthe magnetic field is strong and parallel to the axis of the rope.In the outer parts of the rope the magnetic field is weak andwraps around the axis.

High Dynamic Pressure

Low Dynamic Pressure

Ionosphere

Figure 8. The magnetic configuration of the Venus ionosphere attimes of low (top) and high (bottom) solar wind dynamic pres-sure.


A long tail is formed downstream of the nucleus as aresult.

The rate of addition of mass by Venus to the solarwind is much less than that of a comet because the plan-et’s gravitational pull prevents the atmospheric gas fromexpanding rapidly and keeps it close to the planet.Nevertheless, as noted above, Venus does have a neutraloxygen exosphere that extends to at least 4000 km. Ionsproduced above the ionopause, by photoionization or col-lisions with the solar wind particles themselves, interactdirectly with the solar wind. The atomic oxygen exos-phere is produced by the ionosphere, in particular by thedissociative recombination of O2+ ions at lower altitude,in which ions recombine with an electron to form twosuprathermal oxygen atoms. These atoms have enoughenergy to allow them to reach altitudes above theionopause where, if ionized, they will be carried away inthe magnetosheath or solar wind flow. This additional ionmass can slow down the flow near the magnetic barrierwhere the ion production is highest owing to the greaterneutral densities there. A larger region of stagnationcould increase the size of the barrier and push the shockaway from the obstacle. It could also add additional mag-netized plasma to the ionosphere and enhance the mag-netic tail. As illustrated in figure 9 the magnetic field linesapproaching closest to Venus are slowed down in theirmotion at closest approach while their ends in the undis-turbed solar wind are carried away at the solar windvelocity. This produces a long comet-like ‘induced’ tail.

The ions on these field lines are eventually acceleratedup to the solar wind velocity. However, the composition ofthese ions always tags them as Venus ions. The Venus mag-netic tail was regularly detected at the apogee of the PIO-NEER VENUS spacecraft (12 Venus radii) but the ions havebeen detected as far away as the orbit of the Earth.

The draped magnetic field will apply a stress to theplasma surrounding Venus but under steady-state condi-

tions this stress should be rather uniform. When themagnetic field changes direction the uniformity of thisstress changes and ridges and enhancements at theionospheric plasma–solar wind boundary can form.When spacecraft pass through such a structure, the den-sity enhancement may appear to be a ‘cloud’ of ionos-pheric plasma near the ionopause. These clouds stretchout from the nighttime ionosphere of Venus and are cor-related with interplanetary magnetic field rotations.Other filamentary structures of ionospheric plasmaextend out of the nightside upper ionosphere when solarwind conditions are steady. These ‘tail rays’ may repre-sent a low-altitude component of planetary ions pickedup by the solar wind from the denser cold upper atmos-phere.

Afinal phenomenon that has been seen at Venus thatappears to be associated with the solar wind interactionis the Venus aurora. On Earth AURORAS are often associat-ed with parallel electric fields that accelerate electronsdown into the atmosphere causing atoms to be excitedand to emit photons. Weak, patchy auroras exist on thenightside of Venus also, but no explicit cause of theseauroras has been determined.


Bow Shock Magnetotail

Bow Shock

1 2 3 4 5

6

7

Flow

B (In Tail)

B (Outside Tail)

Figure 9. The formation of the Venus magnetotail. Flux tubesthat pass closest to Venus are slowed the most, while their endsare carried along in the solar wind.

2.4 RV2.1 R V

Solar minSolar max

2.6

2.5

2.4

2.3

2.2

2.1

250

200

150

100

50

075 77 79 81 83 85 87 89

Sunspot N

umbers

Bow

Sho

ck P

ositi

on (

R ) V

PVO Bow Shock

Venera 9/10 Bow Shock Position

Year

Sunspot Numbers

*

*

Figure 10. The motion of the bow shock in the course of the solarcycle. Top: An ecliptic plane view. Bottom: Position above theterminator and sunspot number through a solar cycle.


Solar cycle variationThe above discussion enables us to now address howVenus responds to the solar cycle changes on the Sun.The Earth’s solar wind interaction is sensitive to theSOLAR CYCLE mainly because there are more coronal massejections launched from the Sun near solar maximumthan solar minimum. The Venus interaction is also sensi-tive to these episodic, mainly magnetic, structures (albeitin very different ways than is Earth’s magnetosphericinteraction), but it is more generally sensitive to thechanging ultraviolet flux during the solar cycle. At solarminimum the ionosphere is not as well developed, andso the same solar wind pressure results in an ionopausedeeper in the atmosphere where the ionosphere is lesshighly electrically conducting. This routinely magnetizesthe dayside ionosphere and drastically reduces the flowto the nightside ionosphere. The obstacle becomes a littlesmaller but more importantly some of the solar wind isprobably absorbed. Thus the solar wind does not have tobe deflected so far above the planet as at solar maximum.This effect is so marked that the shock approaches towithin 1000 km of the planet of solar minimum assketched in figure 10.

At solar maximum the dayside ionosphere is denserand better able to stand off the solar wind plasma abovethe ionosphere. In addition the neutral exosphere of hotoxygen is denser and more ions are created within themagnetosheath itself. This extra ionization furtherenhances the deflection of the solar wind.

Web Update (31 July 2002)Scientists have captured the first x–ray view of Venususing NASA’s CHANDRA X-RAY OBSERVATORY. The observa-tions provide new information about the atmosphere ofVenus. Venus in x–rays looks similar to Venus in visiblelight, but there are important differences. The opticallyvisible Venus is due to the reflection of sunlight and, forthe relative positions of Venus, Earth and Sun duringthese observations, shows a uniform half–crescent that isbrightest toward the middle. The x–ray Venus is slightlyless than a half–crescent and brighter on the limbs. Thedifferences are due to the processes by which Venusshines in visible and x–ray light. The x–rays from Venusare produced by fluorescence, rather than reflection.Solar x–rays bombard the atmosphere of Venus, knockelectrons out of the inner parts of the atoms, and excitethe atoms to a higher energy level. The atoms almostimmediately return to their lower energy state with theemission of a fluorescent x–ray. A similar process involv-ing ultraviolet light produces the visible light from fluo-rescent lamps. For Venus, most of the fluorescent x–rayscome from oxygen and carbon atoms between 120 and140 km (74 to 87 mi) above the planet’s surface. In con-trast, the optical light is reflected from clouds at a heightof 50 to 70 km (31 to 43 mi). As a result, Venus’ Sun–lit

hemisphere appears surrounded by an almost–transpar-ent luminous shell in x–rays that appears brightest overthe limb.

BibliographyBougher S W, Hunten D M and Phillips R J (ed) 1997

Venus II (Tucson, AZ: University of Arizona Press)Hunten D M, Colin L, Donahue T M and Moroz V I (ed)

1983 Venus (Tucson, AZ: University of Arizona)Luhmann J G, Tatrallyay M and Pepin R O (ed) 1992

Venus and Mars: Atmospheres, Ionospheres and SolarWind Interaction (Washington, DC: AmericanGeophysical Union)

Russell C T (ed) 1991 Venus Aeronomy (Dordrecht:Kluwer)

C T Russell and J G Luhmann


Venus: Surface E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Venus: SurfaceBecause of its thick and dense atmosphere, VENUS has onlyvery recently revealed the aspect and the characteristicsof its surface. In fact, the only means to penetrate theatmosphere and to see what is underneath is to observethe planet in wavelength regions of the electromagneticspectra where the atmosphere is transparent such asthe radar centimetric wavelengths or some near-infraredwindows around 1 micron. An alternative solution is tosend probes that directly land on the surface. These twomeans have been used to explore the surface of Venus andthey give two complementary kinds of information. Thefirst gives global and large-scale information about thesurface, such as topography, surface roughness, altimetry,etc, while the second gives very local information suchas surface images and/or composition. These twocomplementary techniques have greatly enhanced ourknowledge of the surface of Venus and of its history.

The Venus coordinate systemThe map coordinate system used for measuring longitudeon Venus is different from that used on Earth. On Earth,longitude (an imaginary line stretching from pole to pole)is measured from a starting point (the prime meridian) atGreenwich, England (near London), toward the east andtoward the west with increasing values in degrees untileast meets west at the 180 point (the dateline), which isdiametrically opposed to Greenwich. On Venus, longitudeis measured from 0 to 360 degrees with the prime meridiancentered within a small impact crater named Ariadne,located in Sedna Planitia. There is an arbitrary conventionthat determines the direction of increasing longitude onplanetary bodies other than Earth: longitude shall bemeasured in a direction opposite to that in which the planetrotates. Because Venus rotates in a clockwise directionas viewed looking down on the north pole, longitude onVenus increases in numerical value toward the east fromthe planet’s prime meridian.

Large-scale observationsMost of the large-scale and global observations of theVenusian surface have been obtained by radar. Thereare different kinds of radar data: (a) altimetric data,(b) reflectivity data and (c) emissivity data. The firstgives the altitude of the topography, the second givesinformation about the roughness of the surface and thelast gives information about the brightness of the surface.

The first reflectivity data for the Venusian surfacewere obtained from ground-based radar observatories.The first images were acquired in 1972 from the Goldstonesatellite radar tracking system located in California.Observations with the Goldstone station were made until1988 covering essentially the equatorial regions between15N and 15S latitude and 260E and 30E (through 0)longitude. The spatial resolution, that was of the orderof 5–10 km in the first images, has been improved to near

1 km in the data acquired since 1986. The second ground-based observations from the Venusian surface come fromthe Arecibo observatory in Puerto Rico. On this site, thefirst reflectivity images were taken in 1975 and the lastin 1988. With these data, that covered latitudes up to50N, the spatial resolution was 1.5 km. From space, thefirst almost global observation of the surface of Venus wasobtained by the NASA PIONEER VENUS Orbiter that arrivedat the planet on 4 December 1978. Around 93% of thesurface was covered between 74N and 63S latitude.These observations gave the first large-scale altimetricand topographic coverage with an altimetric precision of150 m at a sampling resolution of about 60 km. Then,from October 1983 to July 1984, the twin Soviet probesVENERA 15 and VENERA 16 surveyed the planet and mappedessentially the north pole area covering approximately25% of the surface from 30N to 88N. The resolutionof the reflectivity images ranges from 1 to 4 km and thetopographic precision was 50 m with a sampling resolutionof about 8 km. The last global data from the surface ofVenus were obtained by the NASA MAGELLAN Orbiter. Theinsertion of the probe into the mapping orbit took placeon 10 August 1990 and the formal mapping of the surfacebegan on 1 September 1990. These operations, dividedin four observation cycles, lasted until 14 October 1994when the probe was lost. The coverage of the surface wasalmost complete. The precision of the altimetric data isbelow 50 m at a sampling resolution of the order of 10 km.The resolution of the reflectivity images is of the order of75 m.

Infrared observations of Venus have been made fromthe ground and during the flyby of Venus by two probes.The first one was GALILEO during its travel to Jupiter on10 February 1990 and the second occured very recentlywith the CASSINI/HUYGENS probe on 24 June 1999 duringits travel to Saturn. These two probes have infraredspectrometers on board (NIMS on board Galileo andVIMS on board Cassini). The surface was detected onthe nightside of Venus by its thermal emission. Theseobservations show that the surface temperature is inequilibrium with the atmosphere at the same altitude (seeVENUS: ATMOSPHERE).

Local observationsAll available on-ground local observations and measure-ments come from the various Soviet Venera/VEGA landers.These probes have made images of the ground surround-ing them and/or composition analysis of the surface. Thefirst of these landers was Venera 8 that landed in 1972 andgave the first natural radioactive element composition ofthe surface. Then three sets of twin probes reached thesurface. The first pair of landers were Venera 9 and 10 thatmade images of the surface and chemical analyses in Octo-ber 1975. They were followed seven years later by Venera13 (1 March 1982) and Venera 14 (5 March 1982) landersthat made color images of the surrounding surface and alsosurface composition analyses. Finally the Vega 1 and Vega2 probes in 1984 were able to analyse the soil composition.



Figure 1. Topographic map of Venus. The highland at the equator is Aphrodite Terra and the highland at the north is Ishtar Terra.

Global topographyAlthough Venus, like most of the TERRESTRIALPLANETS, can bedivided into two broad physiographic regions, highlandsand lowlands, its hypsometric distribution is unimodal,unlike that of the Earth (bimodal) or Mars (trimodal) andover 80% of the Venusian surface lies within 1 km of themean radius of 6051.84 km (figure 1). The mean slopes onVenus are of the order of 1 but average kilometer-scaleslopes greater than 30 are not uncommon.

Highlands• Regio (regiones, pl.) are topographically high regions,

often hosting large shield volcanoes.

• Terra (terrae, pl.) cover vast areas and have variabletopographic relief, as continents do on Earth.

• Planum (plana, pl.). Lakshmi Planum is the onlyplanum recognized on Venus. It is a 3–4 km highplateau, bordered by mountainous ridges.

There are four highland areas on Venus: (1) ISHTAR TERRA inthe north, (2) LADA TERRA in the south, (3) APHRODITE TERRA

in the equatorial zone east of the prime meridian and(4) the last area is defined by BETAREGIO, Phoebe and Themisregiones distributed along roughly the 285 meridian.

Ishtar Terra comprises the high Lakshmi Planumplateau to the west and the extremely elevated region



Figure 2. This Magellan full-resolution image shows MaxwellMontes, and is centered at 65N latitude and 6E longitude.Maxwell is the highest mountain on Venus, rising almost 12 kmabove mean planetary radius. The western slopes (on the left)are very steep, whereas the eastern slopes descend graduallyinto Fortuna Tessera. The broad ridges and valleys making upMaxwell and Fortuna suggest that the topography resulted fromcompression. Most of Maxwell Montes has a very bright radarreturn; such bright returns are common on Venus at highaltitudes. This phenomenon is thought to result from thepresence of a radar reflective mineral such as pyrite.Interestingly, the highest area on Maxwell is less bright than thesurrounding slopes, suggesting that the phenomenon is limitedto a particular elevation range. The pressure, temperature andchemistry of the atmosphere vary with altitude; the materialresponsible for the bright return is probably only stable in aparticular range of atmospheric conditions and therefore aparticular elevation range. The prominent circular feature ineastern Maxwell is Cleopatra. Cleopatra is a double-ring impactbasin about 100 km in diameter and 2.5 km deep. A largeamount of lava originating in Cleopatra flowed through thischannel and filled valleys in Fortuna Tessera. Cleopatra issuperimposed on the structures of Maxwell Montes and appearsto be undeformed, indicating that Cleopatra is relatively young.

surrounding the Maxwell Montes that are, with their 12 kmaltitude, the highest topographic point on Venus (figure 2).

Lowlands• Planitia (planitiae, pl.) are topographic low-lying

regions, generally the most featureless on Venus in termsof tectonic and volcanic structures.

There are five lowlands areas in which lie other smallerhighlands: (1) Atalanta Planitia at the north of AphroditeTerra, (2) Aino Planitia and (3) Helen Planitia at thesouth of Aphrodite Terra, (4) Guinevere Planitia at theequator between Aphrodite Terra and Phoebe Regio and(5) Lavinia Planitia at the south of Guinevere Planitia. Thesmall highlands are Alpha Regio (6E, 25S), Eistla Regio

(39E, 18N), Bell Regio (49E, 33N) and Tellus Regio(82E, 39N).

The landscape of VenusThe landscape of Venus is characterized by some veryparticular and specific geomorphological and tectonicfeatures that illustrate how Venus can be very differentfrom the Earth and that can help to understand itsgeological history.

Volcanic featuresSeveral types of volcanic flows and edifices are recognized.The presence of volcanic features on Venus suggest localmagma sources at depth. The distribution of these featuresmay provide clues about the crustal properties and thermalhistory of Venus. They can be divided into three groupsaccording to their mean size. Brief characteristics are givenfor each below:

(i) Large volcanic forms (<100 km): large shieldvolcanoes are characterized by numerous lava flowsradiating away from a central caldera. Manyindividual flows extend for hundreds of kilometers.Several of the 156 identified shields are locatedin topographically high regions (e.g. Gula and SifMontes at Western Eistla Regio; Sapas, Maat andOzza Montes at Alta Regio) reaching elevations ashigh as 3–5 km above the surrounding area.

(ii) Intermediate-sized volcanic forms (20–100 km):

• Anemones: a type of volcanic edifice characterizedby flows radiating outward, often in bilateralfashion from a central graben or fissure. Theyare relatively rare; only 25 have been identified.Anemones are typically 30–40 by 40–60 km indimension.

• Ticks are volcanic domes, so named because of theirappearance in radar images. They have flat ordepressed circular domes about 25 km in diameterand are flanked by strongly defined radial ridgesand troughs. At one end, the tick edifice mayexhibit signs of a collapsed graben with extensivefault scarps. Occasionally, flows originating fromthe small central calderas appear to be directedoutward along the radial fault-valleys. About 50have been identified.

• Steep-sided domes, commonly referred to aspancake domes: these volcanic domes have well-defined circular outlines and are characterized byradial fractures near the steep perimeter, radial andconcentric fracturing on the flat interior and smallsummit calderas near the center. Over 150 havebeen recognized, very often in groups, sometimesoverlapping one another.

• Calderas and paterae are volcanic depressionsbounded by arcuate fault scarps and are the sourcearea of numerous lava flows. 86 calderas haveidentified.



(iii) Small volcanic forms (>20 km) shield fields coverrelatively large areas (average area 17 700 km2) andare delimited by numerous, but small (typically2–3 km across, but 6–12 km ones are not unusual)volcanic domes or vents. Some shield fieldsdevelop extensive flow fields surrounding the shieldvolcanoes whereas others are located within tectonicstructures such as coronae or linear extensional zones.Overall, 556 shield fields have been recognized onVenus.

Tectonic featuresMany features recognized from Magellan images arereminiscent of TECTONIC structures on Earth. Some of thesefeatures cover regions of thousands of square kilometers,whereas other features are narrow linear structures thatappear to be related to fractures, grabens and perhapsdykes. A classification and nomenclature of Venusianphysiographic features and how they may be related totectonism has been made. Some of these features aredescribed below.

(i) Large tectonic forms (100 to 1 000 km) are:

• Tessera (tesserae, pl.): a terrain network consistingof two or more directions of anastomosing linearridges and troughs. They are tectonically complexterrains, thought to be largely compressional inorigin and may represent the oldest part of thecrust.

• Chasma (chasmate, pl.): a broad trench orlinear zone consisting of a parallel arrangementof troughs or valleys bounded by fault scarps.These zones are interpreted to reflect extensionaltectonics.

• Mons (montes, pl.): large highland provincesare termed montes (e.g. Maxwell Montes, DanuMontes, Akna Montes and Freyja Montes), descrip-tive of their mountain range-like appearance (fig-ure 2).

(ii) Others features likely to be associated with volcanicand/or intrusive activity are:

• Corona (coronae, pl.): (mean diameter = 250 km)are large circular structures whose circumferenceis defined by an elevated ring-like zone consistingof compressional ridges and extensional troughs.The interior part often shows evidence of volcanicactivity and early fracturing. Radial graben-likestructures may extend beyond the outer diameter.Topographically, the interior may be raised ordepressed relative to the surrounding terrain.Corona structures likely represent the surfaceexpression of mantle upwelling. They range from75 to over 2000 km across and occur in groups, inchains or as isolated structures. More than 360 havebeen recognized, of which about half have beenformally assigned names (figure 3).

Figure 3. This region, roughly 100 km on a side, shows agigantic structure known as a corona. Coronae are circular toelliptical features marked by a ring of concentric ridges. Suchfeatures are thought to be the result of hot rising bodies ofmagma that reach the crust. As hot material rises, it weakens theupper layers of the crust, causing the surface to dome upwards.Then as the region cools, the dome begins to subside. As theupper layers rise and fall, they are subjected to stresses thatcrack the surface, creating both circular and radial fractures.Magellan acquired this view of Venus during its first mappingcycle around the planet in 1990 and 1991.

• Arachnoid(s) (mean diameter = 115 km) arecircular to elliptical structures that consist of acentral dome or depression surrounded by anextensive network of radial and concentric linearfeatures. These locally developed structuresmay represent the surface expression of faultingand dyke emplacement associated with magmainjection at shallow depths. About 259 have beenfound on Venus.

• Nova (novae, pl.) (mean diameter = 190 km)are similar to arachnoids, but are dominated byradial structures. Novae are generally centered ona domal uplift and may possibly represent the earlystages of corona formation. Over 50 have beenidentified.

Impact cratersThere are 935 recognized impact CRATERS on Venus. Abouthalf the craters have been formally assigned names, theothers remain unnamed. All have been named afterfamous women in history, but craters with diametersless than 20 km have been given female common names.Venusian craters range in size from 1.4 km in diameter to



Table 1. Uranium, thorium and potassium content in the Venusian surface rocks given by gamma ray spectrometry.

Venera 8 Venera 9 Venera 10 Vega 1 Vega 2

K2O, wt % 4.8 ± 1.4 0.6 ± 0.1 0.4 ± 0.2 0.54 ± 0.26 0.48 ± 0.24U (ppm) 2.2 ± 0.7 0.6 ± 0.2 0.5 ± 0.3 0.64 ± 0.47 0.68 ± 0.38Th (ppm) 6.5 ± 0.2 3.7 ± 0.4 0.7 ± 0.3 1.5 ± 1.2 2.0 ± 1.0

Table 2. Contents of the major elements in the Venusian surfacerocks given by x-ray fluorescence.

Venera 13 Venera 14 Vega 2

SiO2 45.1 ± 3.0 48.7 ± 3.6 45.6 ± 3.2TiO2 1.59 ± 0.45 1.2 ± 0.41 0.2 ± 0.1Al2O3 15.8 ± 3.0 17.9 ± 2.6 16.0 ± 1.8FeO 9.3 ± 2.2 8.8 ± 1.8 7.7 ± 1.1MnO 0.2 ± 0.1 0.16 ± 0.08 0.14 ± 0.12MgO 11.4 ± 6.2 8.1 ± 3.3 11.5 ± 3.7CaO 7.1 ± 0.96 10.3 ± 1.2 7.5 ± 0.7K2O 4.0 ± 0.63 0.2 ±0.07 0.1 ± 0.08S 0.65 ± 0.4 0.35 ± 0.31 1.9 ± 0.6Cl 0.3 0.4 0.3

280 km. Crater Mead is the largest impact crater identifiedon Venus.

In contrast to Mercury, Mars or the Moon, which arecovered with thousands of craters that have accumulatedover the last 4 to 4.5 billion years, Venus is scarred bycuriously few. In fact the spatial distribution of cratersis uniform (random and anticlustered) over the entireplanet, suggesting that Venus experienced complete globalresurfacing in the relatively recent (geologically speaking)past.

In the global resurfacing model, tectonic and volcanicactivity affected the entire surface of Venus whichobliterated the majority of (if not all) previous impactcraters. An observation that lends support to the suddenarrest of these events is the fact that the majority ofcraters, 84%, do not show any signs of modification. Thisresurfacing activity is thought to have ceased between 300to 800 million years ago. The uncertainty of the timing liesin the uncertainty of estimating the impact flux.

Craters on Venus are recognized by their expressionon images and hence classified by their morphology. Thehigh temperature of the Venusian surface (470 C) andits thick atmosphere make Venusian impact morphologyunique among planetary bodies in the solar system. Basedon the development of crater floor structures and degree ofcircularity, a classification of simple craters and five typesof complex crater into a six-fold scheme can be made.

• Structureless craters are simple craters where theinternal floor is flat and featureless. The smallest cratersare generally of this type.

• Central peak craters have a central uplift that rises abovethe crater floor. These craters range in size from 8 to79 km, but are most commonly 16–32 km. Outliningrims are quite circular and often terraced.

• Double-ring craters are defined by an outer rim and acircular arrangement of inner peaks and ridges. Thesecraters are typically greater than 40 km.

• Multiple-ring craters have two or more concentric ridgestructures that rise above the crater floor. The largestcraters on Venus, ranging from 86 to 280 km in diameter,are of this type.

• Irregular craters have non-circular rim outlines andstructural disruptions to otherwise flat crater floors.Almost one-third of the craters on Venus are of this type,most of which are less than 16 km across.

• Multiple crater formation occurs when a falling bodyfragments into pieces. Each fragment creates a separateimpact crater whose rim may overlap with adjacentlyformed craters. Individuals of this type are up to 44 kmin diameter, but most are less than 11 km.

In general, small-diameter craters are flat-floored,have irregular outlines and may be part of a multipleimpact event. Complex internal structures occur in largecraters and tend to develop progressively as a central peak,a double ring, or a multiple ring, with increasing craterdiameter.

Surface properties and compositionA somewhat unexpected finding in the Magellan imageswas the abundant wind streaks. More than 4000 windstreaks have been identified and this evidence of eolianactivity is widespread and is particularly concentratedbetween 17S and 30S latitude and at 5N–52N onsmooth plains. These wind streaks are generally orientedwith the downwind direction towards the equator. Thiseolian activity is also demonstrated by parabolic craterdeposits, made up of fine, centimeters thick debrisdistributed by high-speed zonal E–W winds. Theseobservations are consistent with the radar signature ofthe surface that indicates that weathering, mass wastingand eolian activity operated continuously on depositssimilar to lava flows. This activity confirms the aspectof the surface as it is known by the Venera lander images(figure 4). The reddish color of the images taken by theVenera 13 and Venera 14 landers also supports the volcanicorigin of the surface which is confirmed by the chemicalcomposition.

Our knowledge of the chemical composition of thesurface of Venus comes from two types of experiment onboard the seven Russian landers. These two methods aregamma spectroscopy and x-ray fluorescence spectrometry.The first one gives the contents of the radiogenic elements,



Figure 4. Venera 14 lander images of the surface of Venus at 1S, 310E on 5 March 1982. The lander survived for 60 min. Both imagesshow part of the lander at the bottom. This area is composed of flat basalt-like rocks, but little soil or fine-grained material, as was seenat other Venera lander sites. Near the center of the top image is a lens cover, and the bottom image shows a test arm.

uranium, thorium and potassium (table 1). The secondone gives the contents of the major elements of the surfacerocks (table 2).

Most of these analyses show a trace element and/orbulk chemistry typical of tholeitic basalts. Moreover,all the dominant type of terrain within the Venera/Vegalanding sites is plains so that their measurements canbe considered representative samples of the Venusianplains. Finally morphological observations of long andvast lava flows together with the results of the geochemicalmeasurements show that the plains of the landing sites aswell as the Venusian plains as a whole are predominantlythe result of large basaltic volcanism. Two exceptions comefrom the data of Venera 8 and Venera 13. The gammaspectroscopic analysis at the Venera 8 landing site showsthat the surface material contains relatively high amountsof K, U and Th (table 1). X-ray fluorescence analysis atthe Venera 13 site shows that the bulk chemistry of thesurface rocks is analogous to that of alkaline basalt. Itspotassium content is indeed very similar to that of theVenera 8 site. For these two sites, where non-tholeiticcomposition of the surface material was identified, steep-sided domes have been found which are not present atthe other sites. These particular morphological features,together with the fact that the Venera 8 and possiblyVenera 13 material compositions are the combination ofevolved igneous rocks and more mafic and primitive rocks,suggest that these rocks are the geochemical signature ofthe presence and involvement of material of continentalcrustal origin in their magma genesis. This raises thequestion about the presence or absence of continental-likecrust on Venus, which remains an enigma.

Stratigraphic historyFrom the study of all these features and of theirmorphological and chronological relationships, a tentative

geological history of Venus can be proposed, defining unitsof common ages, their formation and evolution throughtime as described by Basilevsky and Head (1998) (seefigure 5).

The oldest unit, the Fortuna Group, is mainlycomposed of tessera terrains that cover about 8% ofthe Venusian surface. The morphology of this unit isdominated by intersecting systems of ridges and groovesof tectonic origin. This deformation does not extend intothe surrounding plains that clearly embay the tessera.It is not clear if this group was formed in a very shorttime or if it is composed of various subunits of differentages. However, it is clear that all other units overlie orembay this group, making it the oldest recognized unitof Venus like the Precambrian basement of continents onEarth. Although this group may have formed at differentages, its heavy deformation seems to have spanned arelatively short period of time as suggested by the densityof superposed impact craters that is approximately thesame.

The next unit in chronological appearance is calledthe Guinevere Supergroup. It represents an assemblage ofplains-forming units and consists of four groups separatedfrom one another and from the underlying Fortuna Groupby episodes of tectonic deformation.

The first unit consisting of material of denselyfractured plains is called the Sigrun Group and covers3% of the surface of Venus. It appears as uncoveredterrains by the younger plains and is deformed by denselyspaced swarms of faults usually subparallel to each other.The material composing this unit appears to be primarilyplains that were emplaced as floods of lavas.

The second group, called the Lavinia Group, isrepresented by the materials of fractured and ridged plainsand covers about 3% of the surface. The ridges aresometimes clustered forming ridge belts. They appear



generally as elongated islands among the younger plainsbut embay the tessera and the Sigrun Group. The LaviniaGroup material, as the Sigrun Group, was also primarilyemplaced as lava floods (figure 5(a)).

The third group consists of materials of shield plainsand plains with wrinkle ridges. Together this group,called the Rusalka Group, occupies 70–75% of the surfaceof Venus. The wrinkle ridges plains dominate amongthis group and represent 60–65% of the total surface. Acharacteristic of these plains is the presence of wrinkleridges, typically 1 km wide, that often form a network witha dominant trend. Usually subunits can be recognizedin this group because they are deformed by a singleridge network, thus separating them all from the youngerstratigraphic units. The youngest of these subunits formextended flow-like features. This observation and thein situ geochemical measurements made by the Venera9, 10 and Vega 1, 2 landers suggest that these plains formedby extended floods of mafic lavas. The shield plains is thesecond component of the Rusalka group. It is representedby materials of plains formed by clustered and coalescingsloping shields of volcanic origin. These plains occupy10% of the surface of Venus and are usually embayed by thewrinkle ridge plains that appear consequently younger.The volcanic shields, on the basis of their gentle slopes,are made by lavas composed of alkaline basalts or evenmore differentiated material such as andesite as shown bythe Venera 8 analyses.

The last group of the Guinevere Supergroup ismostly made of the materials of lobate and smooth plainsundeformed by wrinkle ridges. This unit covers 10–15%of Venus and is called the Alta Group. It overliesand embays all the already described units. Most ofthe Alta Group materials are associated with rift zonesoccurring in the form of large, gently sloping volcanicedifices. Some materials of this group are associated withcoronae, forming lava flows aprons around them. Theirmorphology as well as their geochemical compositiongiven by Venera 14 suggest that they also are mafic lavas.

The last and youngest group, the Aurelia Group,is represented by the materials of radar-dark parabolasassociated with the youngest impact craters as well withthe eolian patches and streaks.

Absolute age estimatesThe duration of the morphologically distinguishable partof the geological history of Venus is estimated on the basisof impact crater densities. The average age of the surfaceof Venus is estimated to vary between 300 and 500 millionyears, for the lower bound, up to 800 million years for theupper bound. In terms of this average age T , the age ofthe oldest group represented by the tessera is estimatedto be about 1.4T . Then, the average age T of the surfaceof Venus is apparently a good estimate of the age of theRusalkian Group. This means that the total duration of theSigrunian, Lavinian and Rusalkian group is of the orderof 0.2–0.3T that is, approximately a few tens of millionsof years to about one hundred million years. Finally,

the upper boundary of the Atlian time (that is, the lowerboundary of the Aurelian period) was estimated from theproportion of craters with radar-dark parabolas to be about0.1T prior to the present (figure 5(a)).

Geological history of VenusThe history of Venus as recorded on its surface representsonly the last 10–20% of the total history of the planetbecause the morphological signatures of the terrainsbefore those of the Fortuna Group were not preserved.The beginning of this part of the history of Venus ischaracterized by intensive tectonic deformation on aglobal scale which formed the tessera terrains. Earlystages of this deformation were clearly compressionaland later changed into extensional. Termination of thecompression stage is estimated to have occurred at abouttime 1.4T , while the extensional stage lasted for another0.1–0.2T . Numerous internal dynamical processeshave been proposed to account for the tessera-formingdeformation sequence; from chemical instabilities causingmantle overturn, an oscillatory convective behavior ofthe mantle or catastrophic avalanche within the mantledue to the presence of high-pressure phase transitions.These hypotheses raise the question of whether the largecurrently preserved tessera blocks might represent relictsof downwelling or upwelling. Anyway, this intensivetectonism was accompanied by volcanic activity so thatthe emplacement of tessera-forming material and itsdeformation into tessera terrain are the major geologicalevents of Fortunian time (figure 5(a)).

After tessera formation, several stages of extensivevolcanism occurred that buried vast areas of the tesseraand formed the regional plains of Sigrunian, Lavinianand Rusalkian ages. The combined duration of theemplacement of these plains is estimated to be 0.2–0.3T . Plain forming of all these groups is separatedby episodes of tectonic activity that occurred generallysynchronously in different areas of Venus. These episodesare characterized by the dominance of compression, thentension, then again compression, and finally tension.The last globally distributed tectonic episode, extensivewrinkle ridging, happened at about time T and markedthe transition to the present stage of the history of Venus,which is dominated by regional rifting and its localassociated volcanism in the form of large shield volcanoes.The majority of this stage is represented by the Atlianperiod, which appears to have lasted until 0.1T from thepresent. This period is thus the longest in duration amongall the other units, although its resulting tectonic andvolcanic features and deposits represent only 10–20% ofthe surface of Venus. The last period of the history of Venusis the Aurelian period and is characterized by a certainlevel of reworking of the surface by eolian processes. Fromthe point of view of volcanic and tectonic processes, itrepresents simply the continuation of the Atlian periodso that Venus may be today already endogenically activeat a low level (figure 5(b)).



Figure 5. (a) Venus stratigraphic units following Basilevsky andHead (1988) (Tt: tessera terrains; Pdf: densely fractured plains;Pfr: fractured and ridged plains; RB: ridge belts; FB: fracturebelts; Pwr: wrinkle ridge plains; Psh: plains with shield; Ps:smooth plains; Pl: lobate plains; Cdp: dark parabola craters).(b) Summary of the geological history of Venus. The tectonic isdominated by the succession of compression–tension cycles andthe volcanism, intense at the beginning, decreases and changesin style with time (Basilevsky and Head 1998). T , the averageage of the surface of Venus, is of the order of 300–500 millionyears or possibly up to about 800 million years.

In summary, the observable part of the history ofVenus is characterized by two key points that stand incontrast to the comparable period of Earth history whenglobal geodynamical processes are dominated by platetectonics.

(a) Venus shows no signature of plate tectonics. Instead,its global tectonic environment is characterized bytwo successive compression–tension cycles withthe magnitude of the deformation and strain ratedeclining with time.

(b) During the first cycle, plains-forming volcanismoccurred at a rate comparable to terrestrial volcanismbut in a non-plate tectonic style, and then duringthe second cycle Venus was dominated by arift-associated volcanism emplaced at a very lowproduction rate comparable with that of today’sintraplate volcanism.

The surface features, structures and compositionillustrate the distinctive difference between the recenthistory of Venus and the Earth that are in other ways sosimilar.

BibliographyBarsukov V L et al 1986 The geology and geomorphology

of Venus surface as revealed by the radar imagesobtained by Veneras 15 and 16 Proc. Lunar and Planet.Sci. Conf. 16th, Part 2, J. Geophys. Res. 91 (suppl B4)D378–D398

Basilevsky A T and Head J W 1998 The geologic history ofVenus: a stratigraphic view J. Geophys. Res. 103 (E4)8531–44

Bougher, S W, Hunten M and Phillips R J 1997 VenusII-Geology, Geophysics, Atmosphere and Solar WindEnvironment (Tucson, AZ: University of ArizonaPress)

Carlson R W et al 1991 Galileo infrared imagingspectroscopy measurements at Venus Science 2531541–8

Chapman C R (ed) 1992a Magellan at Venus J. Geophys. Res.97 (E8) 13 063–689

Chapman C R (ed) 1992b Magellan at Venus, part 2J. Geophys. Res. 97 (E10) 15921–6380

Surkov Y A et al 1984 New data on the composition,structure and properties of Venus rock obtained byVenera 13 and 14 Proc. Lunar and Planet. Sci. Conf. 14th,J. Geophys. Res. 89 B393–B402

Surkov Y A et al 1986 Venus rock composition at the Vega 2landing site Proc. Lunar and Planet. Sci. Conf. 17th,Part 1, J. Geophys. Res. 91 E215–E218

Olivier Forni


Verbiest, Ferdinand (1623–88) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Verbiest, Ferdinand (1623–88)Flemish cartographer and astronomer, member of theJesuit mission to China in the early seventeenth century,was made President of the Astronomical Board andproduced textbooks on astronomy and maps for theChinese Emperor.


Very Large Array, National Radio Astronomy Observa-tory


Very Large Array, National RadioAstronomy ObservatoryThe Very Large Array (VLA), 80 km west of Socorro, NM,is one of the world’s premier radio-astronomy facilities,offering researchers a unique combination of resolvingpower, sensitivity and observational flexibility. Dedicatedin 1980, the VLA includes 27 25 m diameter dish antennas,arranged in a ‘Y’ pattern, that work together as a singleaperture-synthesis interferometric imaging system.

A facility of the US National Science Foundation, theVLA is open for use by all scientists on a peer-reviewedbasis. Capable of observing at frequencies from 74 MHzto 50 GHz (non-continuous), and with antennas deployedthrough a 16 month cycle of four standard configurations,the VLA provides resolution ranging from 15 arcmin to0.05 arcsec.

The VLA’s versatility has allowed it to serve a widerange of research specialties, including planetary, solar,stellar and galactic astronomy as well as cosmology.It has made important contributions to the study ofboth galactic and extragalactic relativistic jets; the MilkyWay’s central region; galactic structure, dynamics andevolution; supernova remnants; and transient events suchas supernovae and gamma ray bursters.

For further information seehttp://www.nrao.edu.


Very-high-energy Gamma-ray Sources E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Very-high-energy Gamma-ray Sources

Gamma-ray photons at TeV energies have been observedin recent years from a special class of active galacticnuclei, the so-called BLAZARS. The maximum photon energyobserved is now approaching energies beyond 10 TeV,opening up new windows into the high-energy universe.

The basic concept in our present day understandingof ACTIVE GALACTIC NUCLEI is the existence of a central BLACK

HOLE (our Galaxy is now well established to harbor oneof ∼2.5 × 106 solar masses (Genzel et al 1997)) with asurrounding ACCRETION DISK. Perpendicular to the inneraccretion disk is a stream of high-velocity gas, commonlyreferred to as a jet; one expects jets on both sides. Thematerial in this jet is usually moving at very near the speedof light. In addition, this basic picture appears to holdquite well for solar mass black hole systems such as somecompact binary stars in our Galaxy, as well as for the mostpowerful quasars we know, with black hole masses of theorder of 1010 solar masses.

From Hubble Space Telescope observations we havealso learned that these massive black holes always sit inthe dynamical center of their host galaxies. The spheroidaldistribution of older stars within these galaxies has a masswhich appears to be approximately proportional to themass of the central black hole, with a factor of order300. This can be readily understood in an accretion diskscenario by modelling the entire galaxy as a system ofaccreting gas, out of which stars are formed and whichfeeds the budding black hole at its very center. Mergers ofgalaxies then turn these disks of stars into a more sphericaldistribution (Faber et al 1997, Wang and Biermann 1998).

The radio emission from the environment of theseblack holes and their associated jets is usually dominatedby non-thermal emission from the jet, most oftensynchrotron emission from the gyrating motion of highlyenergetic electrons/positrons in a magnetic field. Thisemission process is accompanied by absorption, and sooften synchrotron self-absorption becomes important. Inthe overall radio emission from the various segments of thejet this leads to an approximately flat spectrum, where theflux density (energy per bandwidth per second receivedin a unit telescope area) is approximately constant withfrequency. A survey of the data shows that such flatspectrum radio sources are almost always variable, andshow signs of bulk relativistic motion along the jet.

The power from active galactic nuclei ranges fromunder 1038 erg s−1 to over 1047 erg s−1 (1 erg = 10−7 J).The source sizes, best measured by intercontinental radiointerferometry, can extend from smaller than the solarsystem to many millions of light years. Their emissionin the electromagnetic spectrum has been observed fromaround 10 MHz in the radio to a photon energy beyondorder of 10 TeV (1 TeV = 1012 eV = 1.6 erg = 1.6× 10−7 J).

Basic conceptsWe measure motions close to the speed of light in terms ofthe Lorentz factor γ , where

Lorentz factor γ = 1√1− β2

. (1)

The relative speed is β = v/c, where v is the velocity of theparticle or material such as gas in a jet under consideration,and c is the speed of light.

We call the motion of matter relativistic, when γ ismuch larger than unity. Since the total energy E of amoving particle is given by E = γmc2, where m is therest mass of the particle, the total energy can be verylarge. The largest particle energies directly observed are3 × 1020 eV, which corresponds for protons to a Lorentzfactor of 3 × 1011. The motions of jets in active galacticnuclei can often be characterized by a bulk Lorentz factorof order 10. From the electromagnetic emission fromthese jets, we can infer Lorentz factors of electrons movingaround inside the jet in its frame of reference of order 106

and lower.We observe the electromagnetic emission from highly

relativistic electrons, and possibly protons, from these jetsat all wavelengths, and interpret the radiation at TeVenergies also as emission from a relativistic jet. Theprincipal source of this emission can be synchrotronemission, due to the gyrating motion of highly relativisticelectrons in a magnetic field. It can also be inverseCompton emission, from collisions between relativisticelectrons and photons. Obviously, at high photonenergies approaching or exceeding 109 (1 GeV), it canalso for instance derive from π 0 decay following hadronicinteractions.

Hadronic interactions are interactions between nucleior between other particles where pions (or at extremeenergies nuclei) are created. Since pions have a largerest mass, and in turn decay into photons or muons andneutrinos, hadronic interactions are usually heralded bythe emission of neutrinos.

Observational techniquesAt TeV energies photons are observed on Earth withCerenkov telescopes. These telescopes use the Cerenkovlight emitted from the products of the shower maximumproduced by interactions of the incoming primary par-ticle, to determine its energy and direction. Since theoverwhelming number of all primaries are directional ran-domly distributed charged nuclei, the intrinsic propertiesof photon and hadron initiated air showers (see COSMIC RAYS:

EXTENSIVE AIR SHOWERS) may be also used for a separationbetween photons and hadrons (see Gaisser (1990) and re-views by Weekes (1996), Cawley and Weekes (1996) andWeekes et al (1997)).

Looking into the air shower one sees an image,which is rather different for hadronic showers and purelyelectromagnetic showers. Hadronic showers producean image which is very irregular in its appearance,



while electromagnetic showers, initiated by a high-energyphoton, give an elliptic image, which points towards thesource on the sky. This analysis was originally developedby M Hillas of Leeds, and then evolved into a techniqueby T C Weekes, of Tucson, at the Whipple telescope (in themountains of Arizona, USA).

There are internationally many telescopes using thistechnique, with one of the most active telescope arraysbeing HEGRA, on La Palma in the Canary Islands. There arefurther such telescopes in many countries now, such asin Australia, France, India, Russia, South Africa, Ukraineand USA (Meyer 1997, Ong 1998). Table 1 lists some of themajor telescopes already in operation.

The acceptance of a large initial detectable photonenergy is a strong function of the zenith angle of theobservation, and one needs large zenith angles to reachbeyond 10 TeV.

ObservationsThese observations started in 1992 with the first successfuldetection of Mrk421 by the Whipple observatory (Punchet al 1992).

There are now several blazars, which have beenobserved at such energies, with the most observationsbeing taken of the objects Markarian 421 and Markarian501, originally discovered by the Armenian astronomerB E Markarian. TeV energies means a factor of 1000 higherthan the photon energy (1 GeV = 109 eV) at which a muchlarger class of blazars have been observed with the EGRETinstrument onboard the COMPTON GAMMA RAY OBSERVATORY

satellite (Mattox et al 1997). But all these blazars, whetherseen only at GeV energies or at TeV energies, have commonproperties: all of them have a flat radio spectrum inthe GHz range, are variable at almost all frequenciesobserved, and their radiation is interpreted as emissionfrom a relativistic jet. In the case of the TeV blazars, theiremission appears to be the greatest at these energies, fargreater than at any other wavelength observed (Aharonianet al 1997, Bradbury et al 1997, Funk et al 1998, Hayashidaet al 1998, Krennrich et al 1997, Zweerink et al 1997).

The observations now show that the spectral shapeis variable, and occasionally extends to the edge ofwhat is observable, suggesting that perhaps the emissioncontinues to many 10 TeV. Some of the time the spectrumis flat, which means that the energy output per logarithmicphoton energy interval is constant with energy. Thevariability time scale is as short as can be detected withthe photon counting statistics, at a few hours or less.

Figure 1 shows the spectrum of Mkn501 measuredby HEGRA. The work of Aharonian et al (1999a) showsthat the spectrum continues without an obvious break to24 TeV.

PhysicsThese observations are interesting in their own right: theBig Bang produced a bath of microwave backgroundradiation as a remnant of its early hot phase, whichcan be characterized by a temperature of 2.73 K,

Figure 1. The spectrum of MKn501 as measured by HEGRA(Aharonian et al 1999b).

based on COBE observations (Fixsen et al 1996). Earlygalaxy star formation extended the background radiationfrom dust emission around young stars to the far-infrared. Encounters of TeV photons with this microwaveand far-infrared background produces electron–positronpairs, which translates to an effective absorption.Therefore observing a source at multiples of TeVenergy gives us information about the far-infraredbackground at wavelengths otherwise nearly inaccessible.Interestingly, early models for galaxy formation in theuniverse exceeded the level required by the gamma-rayobservations by a large margin. The gamma-ray dataare still pushing the envelope lower for the possible levelof far-infrared background radiation. Therefore thesegamma-ray observations provide one of the most stringentlimits known for the violent early phases of the universe(Malkan and Stecker 1998, Mannheim 1998, Stanev andFranchescini 1998).

There are two main interpretations proposed for thephysical process of the TeV gamma-ray emission itself, aleptonic and a hadronic process (Mannheim 1996).

The leptonic process proposed is a collision betweena very energetic (and highly relativistic) electron and aphoton from its surroundings either outside the jet, orinside the jet. In such a model electrons are presumedto be accelerated, all the while undergoing losses; thisimplies one stringent limit to their maximum energy,which leads to a maximum energy of any photon withwhich the electrons interact. This encounter, called theinverse Compton process, boosts the energy of the photonto much higher energies. In such a picture the maximumphoton energy is then limited by the maximum energythe electron could have, and by the kinematics of the



Table 1. Main parameters of the Air Cherenkov Telescope.

Sensitivity‡Mirror Threshold at the

Elevation No of area No of Pixel σθ FOV energy thresholdTelescope Site Country Longitude Latitude (m) telescopes (m2) pixels (deg) (deg) (deg) (GeV) (cm−2 s−1)

Whipple Mt Hopkins USA −110.5 31.4 N 2300 1 74 151 0.25 0.1 3.5 250 10−11

CAT Themis France −2.0 42.5 N 1650 1 17.5 548 + 52 0.10 0.1 4.8 250 10−11

HEGRA La Palma Spain −17.8 28.8 N 2200 4 4× 8.4 4× 271 0.25 0.1 500 4× 10−12

HEGRA CT1† La Palma Spain −17.8 28.8 N 2200 1 5.0 127 0.25 0.1 4.6 1500 3× 10−12

TA Dugway USA −113.0 40.3 N 1600 3 3× 6.0 3× 256 0.25 0.1 4.5 500TACTIC Mt Abu India +72.7 24.6 N 1300 4 4× 9.5 4× 81 0.31 2.8 700CANGAROO Woomera Australia +136.8 31.1 S 160 1 11.3 256 0.12 0.18 3.0 1000 4× 10−12

Durham Mk 6 Narrabri Australia +149.8 30.5 S 200 1 (3 dishes) 3× 42.0 109 + 19 + 19 0.25 0.1 3.4 250 5× 10−11

CrAO GT-48 Crimea Ukraine +34 45 N 600 2× 6 dishes 2× 13 6× 37 0.40 0.2 2.7 900 5× 10−12

Nooitgedacht Potchefstroom South Africa +27.2 26.9 S 1440 4 7 4 1.7 1.7 700Patchmari Patchmari India +78.4 22.5 N 1075 25 4 1 3.0SHALON Tien-Shan Russia +75.0 42.0 N 3300 1 10 144 0.4 7.2 1000

† For the old mirrors, i.e. up to November 1997.‡ Defined as the 5σ limit for 50 h observation time.

collision. In standard models this leads to a limitation ofthe maximum photon energy to about 10 TeV.

A hadronic process can easily produce photonsof much higher energy, by starting with energeticprotons, which can clearly produce very energetic photonsthrough proton–photon or proton–proton encounters andsubsequent decay of a pion produced in the interaction.We observe high-energy particles directly from outsidethe Galaxy with energies up to several 1020 eV; since thegyration motion of such particles—if they are protons—is larger than the size of the Galaxy, they must comefrom outside, as argued early by Cocconi (1956). Also,the interaction of these high-energy particles with themicrowave background ought to limit their energy uponarrival at Earth to less than 5 × 1019 eV, if they areprotons; this expected cut-off is called the Greisen–Zatsepin–Kuzmin cut-off. It is not observed. We are stilltrying to identify the sources for particles at these extremeenergies. If these active galactic nuclei could be shown torequire protons at extreme energies, then we would havea very good candidate class. The experiments AGASAin Japan, HIRES in the US, and in the future AUGER inArgentina, are looking for more events at these extremeenergies, to beyond 1021 eV (Biermann 1997).

We also expect TeV photon emission from the interac-tion of the known COSMIC RAY population (Berezinsky et al1990, Wiebel-Sooth and Biermann 1999) in the Galaxy withthe interstellar medium (Hunter et al 1997); since the cos-mic ray spectrum of probable Galactic sources extends to3×1018 eV, very high photon energies are expected from thedecay of pions resulting from nucleus–proton encounters.Since the cosmic ray population is likely to be stronger inthe galactic center region, we expect the as yet unobservedTeV emission to peak at the galactic center.

There is another class of sources for which TeVphotons are expected to be detected at some point in thefuture: the now famous gamma-ray bursts (GRBs; seealso GAMMA-RAY ASTRONOMY). GRBs are characterized by asudden outburst of gamma-ray photons at typical energiesof tens of keV to MeV, occasionally to GeV. The time scaleof the emission is a few seconds, with substructure down

to milliseconds. The time-integrated spectrum can beapproximated by two power laws which break around150 keV, suggesting a non-thermal process just as in activegalactic nuclei jets. These GRBs are now known to beat cosmological distances (Metzger et al 1997), and thusrequire enormous powers to be released; it is still unknownwhat physical picture can explain all these features.

One immediate and clear consequence of the hadronicpicture is the emergence of a cosmological neutrinobackground at high neutrino energies. This neutrinobackground would be much stronger at high energiesthan both the atmospheric neutrinos (which have nowbeen used to present impressive evidence for neutrinooscillations by the Super-Kamiokande experiment), andalso higher than the Galactic neutrino background, fromp–p interactions of cosmic rays in the Galaxy (which inturn can be estimated from the observed GeV gamma-rayspectrum). There is currently an experiment at the SouthPole, AMANDA, which hopes to detect this expectedcosmological neutrino background.

Similarly a gamma-ray background is expected,which leads to another implicit use of those TeV gamma-ray data: what fraction of the diffuse gamma-raybackground is due to active galactic nuclei such as blazars,and what fraction, if any, is left to be derived from differentphysics? Standard BIG BANG THEORY predicts ubiquituousrelics, among them topological defects, which can decayinto particles of the order of 1024 eV in most versions ofthe theory, which in turn decay via a cascade into protons,neutrinos and gamma rays, thus producing a truely diffusegamma-ray background (Sigl et al 1995). If we could findan upper limit to the GeV and TeV contribution fromthe decay of topological defects, it would give a majorconstraint to particle physics near the energy scale of GRAND

UNIFIED THEORIES for particles (GUTs).Once we have thus identified a class of sources re-

sponsible for very-high-energy protons, possibly observ-able in high energy photons following interactions, andalso other high-energy particles, then these accelerators inthe cosmos and their environment provide us with tools—albeit very distant tools—to do very-high-energy physicsat energies far beyond any physicist’s dream.



AcknowledgmentsI wish to thank Torsten Ensslin, Heino Falcke, GalenGisler, Todd Haines, Norbert Magnussen, Karl Mannheim,Sera Markoff, Hinrich Meyer, Rene Ong, Rainer Plaga,Ray Protheroe, Wolfgang Rhode, Gunther Sigl, TodorStanev, Heinz Volk and Yiping Wang for contributions,discussions, perusing the manuscript and/or suggestingcorrections.

BibliographyAharonian F et al 1997 Astron. Astrophys. Lett. 327 L5–L8Aharonian F et al for the HEGRA collaboration 1999a

Astron. Astrophys. 349 11–28Aharonian F et al for the HEGRA team 1999b Astron.

Astrophys. 342 69–86Berezinskii V S et al 1990 Astrophysics of Cosmic Rays

(Amsterdam: North-Holland)Biermann P L 1997 J. Phys. G: Nucl. Phys. 23 1–27Bradbury S M et al 1997 Astron. Astrophys. Lett. 320 L5–L8;Cawley M F and Weekes T C 1996 Exp. Astron. 6 7Cocconi G 1956 Nuovo Cimento 10th Series 3 1433–42Faber S M et al 1997 Astron. J. 114 1771–96Fixsen D J et al 1996 Astrophys. J. 473 576–87Funk B et al 1998 Astrophys. Part. Phys. 9 97–103Gaisser T K 1990 Cosmic Rays and Particle Physics

(Cambridge: Cambridge University Press)Genzel R 1998 Nature 391 17–8Genzel R et al 1997 Mon. Not. R. Astron. Soc. 291 219–34Hayashida N et al 1998 Astrophys. J. 504 L71–L74Hunter S D et al 1997 Astrophys. J. 481 205–40Krennrich F et al 1997 Astrophys. J. 481 758–63Malkan M A and Stecker F W 1998 Astrophys. J. 496 13–6Mannheim K 1996 Rev. Mod. Astron. 9 17–47Mannheim K 1998 Science 279 684Mattox J R et al 1997 Astrophys. J. 481 95–115Metzger M R et al 1997 Nature 387 878–80Meyer H 1997 Proc. TAUP97 ed A Bottino, A Di Credico

and P Monacelli p 392Ong R 1998 Phys. Rep. 305 93Punch M et al 1992 Nature 358 477–8Sigl G et al 1995 Science 270 1977–80Stanev T and Franchescini A 1998 Astrophys. J. Lett. 494

L159–L162Wang Y and Biermann P L1998 Astron. Astrophys. 334 87–95Weekes T C 1996 Space Sci. Rev. 75 1Weekes T C et al 1997 Proc. Extremely High Energy Cosmic

Rays: Astrophysics and Future Observations ed MNagano (ICRR, University of Tokyo) p 160

Wiebel-Sooth B and Biermann P L 1999 Cosmic rays inLandolt-Bornstein, Handbook of Physics vol 6 (Berlin:Springer) pp 37–90

Zweerink J A et al 1997 Astrophys. J. Lett. 490 L141–L144

Peter L Biermann


Vesta E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

VestaVesta, with a mean diameter of 529 km, is the thirdlargest main-belt ASTEROID and was the fourth asteroidto be discovered. It was found in 1807 by the GermanAstronomer HEINRICH OLBERS and named after the goddessof fire and the hearth in the Roman mythology. Vesta isthe sole ‘intact’ asteroid that may have undergone heatingat temperatures able to produce a complete planetary-type DIFFERENTIATION (when the heaviest element sank tothe core of the asteroid while lighter minerals remainednear the surface). Telescopic observations at high angularresolution, either from space or from the ground, cannow provide images of Vesta’s surface that help to betterunderstand its formation and collisional history. Suchimages show that Vesta has undergone a large impactevent, 4.5 billion years ago. This discovery supports theidea that Vesta is the parent body of a group of small Vesta-like asteroids and possibly the source of a particular typeof METEORITES (the basaltic achondrite meteorites) collectedon Earth. Although space agencies have not yet selecteda large main-belt asteroid as primary target for a roboticmission, Vesta is considered an excellent candidate forfuture in situ exploration of the ASTEROID BELT.

Surface propertiesAbout every 18 months, Vesta is at OPPOSITION. Becauseits surface is made of high-ALBEDO material (table 1),the asteroid can be, at that time, visible with a nakedeye (its visible magnitude becomes close to 6th). Earlyphotometric measurements made by Bobrovnikoff in 1929showed that the surface of Vesta reflects irregularly thesolar flux over a rotation period (∼5.34 h). Thesebrightness changes could be more easily explained bystrong albedo variations over its surface than througheffects purely induced by an irregular shape. Theparticularity of its surface properties was confirmed inthe 1970s when McCord et al obtained the first spectraof Vesta at visible wavelengths. They attributed a strongabsorption band centered at 0.93 µm to the electronictransitions on the M2 site of a magnesian pyroxene,transitions characteristic of magmatic minerals. Thenature of its basaltic surface was confirmed soon afterwith the detection of the second absorption band ofpyroxene near 2.0 µm. Figure 1 shows an averagedreflection spectrum of Vesta at visible and near-infraredwavelengths. In addition to the 1.0 µm and 2.0 µmbands attributed to pyroxene, a wide absorption band,also diagnostic of basaltic material, is visible between1.1 µm and 1.4 µm and is produced by the presence offeldspar on Vesta. The strength of these bands variesacross the surface, depending on whether the material hasbeen excavated from underneath the crust by impactorsor is a relic of the original eucritic layer. When comparedwith the spectra of other asteroid classes, the depth ofthe absorption bands displayed in the reflection spectrumof Vesta is among the strongest. Vesta has undergonepowerful but localized impacts in its history that have

0.6

Rel

ativ

e re

flec

tan

ce

0.5 1.0 1.5 2.0 2.5Wavelength ( m)µ

PyroxeneOlivine

Pyroxene

Feldspar

0.8

1.0

Figure 1. Mean reflectance spectrum of Vesta’s surface at visibleand near-infrared wavelengths. The deflections from thecontinuum are attributed to pyroxene and feldspar minerals.Olivine is also present on Vesta but locally only, and itsabsorption band is not visible in this rotationally averagedspectrum. Pyroxene, feldspar and olivine are characteristic ofmagmatic material. Their presence on Vesta is the signature ofthe high temperature it has undergone. Therefore, complete—orpartial—melting occurred on Vesta, which is the only largedifferentiated asteroid that has escaped catastrophic and fullydestructive collisions with other bodies of the asteroids’ mainbelt. Spectrum from Gaffey (1997).

preserved part of its primordial crust. As a result of theseimpacts, the variegated surface of Vesta produces strongfeatures in the asteroid’s lightcurve.

Table 1. Vesta’s ID card.

Mean diameter 529 km a

Density 3.5 g cm−3 a

Rotation period 5.34 h b

Semimajor axis 2.362 AUOrbital period 3.63 yrOrbital eccentricity 0.097Orbital inclination 7.14Pole coordinates RA2000 = 301, DE2000 = 41 a

Asteroid type VAlbedo 0.423 c

Composition Pyroxene, feldspar, olivine

a Thomas et al (1997).b Gaffey (1997).c IRAS catalog (1992).

Link to HED meteoritesThe distribution of minor planets as a function of theirheliocentric distance is such that, the further from theSun, the more primitive are the asteroids. All asteroidshaving surface properties characteristic of magmaticmaterial are concentrated at the inner edge of the belt.Asteroids of types A (olivine + metal), M (metal), R (iron-poor pyroxene + olivine) or V (pyroxene–plagioclase,diogenite) can be found up to 3.2 AU from our star.Vesta’s eucritic crust is nearly intact and is typical of adifferentiated object that has undergone thermal heatingat temperatures above 1000 K. The source of this heating is



still not certain but the actual adopted theories implicateelectrical heating induced during the early T Tauri phaseof the Sun or decay of long-lived radionuclides. BasalticACHONDRITE (or HED, for howardite, eucrite, diogenite)meteorites have similar composition to Vesta: the eucritesdisplay the pyroxene absorptions as well as the weakerband of feldspar while diogenites are pyroxene rich and donot display the feldspar feature. Because of the similarityof their spectra, Vesta is suspected to be the parentbody of the HED meteorites. Among the populationof differentiated asteroid, Vesta, with a semimajor axisof 2.36 AU, does not occupy a remarkable position thatcould provide a significant flux of meteorites to Earth.Delivering pieces of Vesta to Earth would require somevery energetic impacts that could produce fragments withvelocities of the order of ∼1000 m s−1. Only with sucha kinetic energy could the fragments reach the nearestresonance zone (3:1 resonance with Jupiter) and be ejectedtowards inner regions of the solar system, and, for some ofthem, into Earth-crossing orbits. Binzel and Xu discoveredin 1993 a family of small Vesta-like asteroids extendingfrom the secular resonance ν6 (a = 2.18 AU) to the 3:1resonance with Jupiter (2.5 AU). Their visible spectrumdisplays the 0.9 µm absorption band which is also presentin the spectrum of Vesta, eucrites and diogenites. Morerecent observations with the HUBBLE SPACE TELESCOPE andground-based telescopes equipped with ADAPTIVE OPTICS

showed the presence of a large impact zone located nearthe Vesta south pole (figure 2). These observations supportthe big picture in which Vesta could be the source of thefamily of small Vesta-like asteroids and be the parent bodyof the HED meteorites.

Although the pieces of the puzzle fit together nicely,alternative solutions still exist. Indeed, Vesta is theonly differentiated asteroid that has survived catastrophiccollisions, but we know from spectroscopic studies thatasteroids from class A represent the inner mantle ofsome differentiated asteroids now disrupted. Similarly,asteroids from class M could be fragments of the core ofsome fully differentiated asteroids while class R asteroidsare likely to be the product of the disruption of a Vesta-like asteroid, exposing material from the mantle andouter crust. These three asteroid classes originate fromVesta-like asteroids that have been entirely disrupted byimpactors and constitute an additional potential sourceof basaltic achondrite meterorites. Astronomers havefound several Earth-approaching asteroids whose surfacecomposition is similar to HED meteorites. In addition,measurements of the oxygen isotope ratios in HEDs matchclosely those found in stony-iron meteorites and sustainthe possibility that the HEDs might come, in fact, from acompletely disrupted parent body, more than originatingfrom removal of the outer layers of Vesta.

Regardless of whether the HED meteorites comefrom Vesta itself or have been created by disruption ofone or several Vesta-like asteroids, the undoubted pointis that Vesta is unique as the sole ‘intact’ differentiatedasteroid. This point alone provides strong constraints tothe collisional history of the whole asteroid belt.

B

A

Figure 2. Asteroid Vesta imaged from the ground and space.(A) Image obtained at 0.83 µm from Mt Wilson Observatory on14 June 1996 using the adaptive optics system on the 2.5 mtelescope. A 20 phase angle is responsible for the strongincrease in brightness of the right-hand side (eastern) limb ofVesta. The sub-Earth coordinates of Vesta are latitude = −3 andlongitude = 255 in this picture. (B) Image obtained by theHubble Space Telescope and its visible camera WFPC2 on 6 May1996 with a 5 phase angle. The central coordinates on Vestacorrespond to latitude = −9 and longitude = 12. Both imagesare oriented so the spin axis is vertical, north pointing up. Notethe strong albedo marks visible on Vesta that correspond tovarious geological units. These albedo changes are related tospatial variations in the surface composition as well as to theaction of space weathering (aging process). A complete study atnear-IR wavelengths is needed to fully understand the natureand history of these geological units. A simple picture would beto attribute the low albedo regions to old eucritic assemblagesand the high albedo features to fresher material excavated frombelow the crust. The irregular shape of the south pole is believedto be the result of a major, but not fully destructive, impact thatoccurred on Vesta early in its history. These recent observationssupport the theory that HED meteorites and small Vesta-likeasteroids might originate from such impact. Credit for (A):Christophe Dumas and Chris Shelton. Credit for (B): Alex Storrs(image processing) and NASA/STScI.

Surface mappingBecause of its early differentiation and volcanism, Vestacan be considered as the smallest terrestrial ‘planet’ ofthe solar system. Detailed mapping of its surface isimportant to understand how PLANETESIMALS accreted fromthe material that was present in the inner regions of thePROTOPLANETARY NEBULA and what endogenic and exogenicprocesses these planetesimals have undergone. Only afew techniques are currently available to return imageswith the required spatial resolution for mapping. The



most direct and expensive solution consists in obtainingin situ measurements of Vesta’s surface from a spacecraft.At the present time, only three main-belt asteroids havebeen visited by a spacecraft: GALILEO flew over GASPRA

and IDA (and discovered Ida’s satellite named Dactyl)while en route to Jupiter, and the NEAR spacecraft visitedMathilde during its journey to a rendezvous with the near-Earth asteroid EROS. Ground-based optical observationswere, until recently, limited in spatial resolution becauseof the deformations undergone by the wavefront ofthe light while traveling through atmospheric layers ofdifferent optical indices. For this reason astronomersused the Hubble Space Telescope to obtain the firstresolved images of Vesta. However, new ground-basedobservational techniques, such as adaptive optics, cannow measure the wavefront deformation and correctit in real time. Adaptive optics allows the largetelescopes to image at angular resolution equivalent totheir theoretical diffraction limit, providing performancesat least comparable with HST. Other techniques suchas speckle interferometry can provide diffraction-limitedimages but are still strongly limited to the observationof the brightest objects. Recent improvements in thesensitivity of radar detectors will allow radar observationsof asteroids to be extended to the largest bodies in the mainbelt and their topography to be derived from Doppler-delay measurements.

The surface of Vesta is mostly a pyroxene–plagioclaseassemblage with few regions where the excavatedcrust displays the feldspar-poor inner mantle material(diogenite). At a few locations, impacts occurred atenergy sufficient to expose the deeper olivine-rich layerof the mantle. The core of Vesta is presumably madeof metallic-rich material or even pure metal. In orderto understand how Vesta differentiated, we need tomeasure the distribution of the diverse geological unitsover its surface. Depending on how Vesta melted, thedistribution of diogenites would be different. In caseof total melting of the asteroid, the diogenite-like layerwould be confined underneath the eucritic crust whilea partial melting would implicate a crust compositionmade of a mixture of feldspar-rich (plagioclase) andfeldspar-poor (diogenites) units. Prior to the HST andadaptive optics observations, the first successful attemptto map the surface of Vesta from the ground was realizedin 1981 by Gaffey using rotational spectroscopy. Thistechnique consists in measuring the variation in depthand position of the bands with the rotational phase ofthe asteroid to derive the location and nature of the maingeological units. Such measurement is better constrainedin longitude than in latitude but a compositional mapof Vesta could be derived and several geological unitsidentified across Vesta. Thanks to the particular inclinationof Vesta’s rotation axis at the time it was observed, it waspossible to locate several diogenite units near Vesta’s southpole as well as one olivine-rich region near the equator.These units correspond certainly to a partial removal ofthe outer crust, or lava flows that transported diogenitic

material to the surface through cracks in the mantle. Theimages recorded at various visible wavelengths by HSTconfirmed the existence of a large crater near the southpole. This feature is apparently the result of a large, butnot entirely destructive, impact which had occurred onVesta soon after its crust cooled down. Recent ground-based and HST results show an increase in the banddepth of pyroxene at high southern latitude that could beexplained by a local removal of the crust by the impactorto expose fresher material from the mantle. The family ofsmall Vesta-like asteroids and HED meteorites could haveoriginated from the EJECTA produced during this violentimpact. The topology of the asteroid derived from theHST images agrees with this picture and the estimatedmass excavated from Vesta during this impact could easilyaccommodate the formation of Vesta-like asteroids andHED meteorites. Complete mapping of Vesta at similarand even higher spatial resolution in the near-infrared(at a wavelength that provides a better diagnostic ofthe nature of the minerals) is currently under progressusing HST and ground-based telescopes equipped withadaptive optics systems. These observations will bringsome important insights to understanding the processesof planetary differentiation that occurred on Vesta andinvestigating the existing scenarios that link Vesta to theHED meteorites.

BibliographyBinzel R P and Xu S 1993 Chips off of asteroid 4 Vesta:

evidence for the parent body of basaltic achondritemeteorites Science 260 186–91

Binzel R P, Gaffey M J, Thomas P C, Zellner B J, Storrs A Dand Wells E N 1997 Geological mapping of Vesta fromthe 1994 Hubble Space Telescope images Icarus 12895–103

Gaffey M J 1997 Surface lithologic heterogeneity of asteroid4 Vesta Icarus 127 130–57

Gaffey M J, Bell J F and Cruikshank D P 1989 Reflectancespectroscopy and asteroid surface mineralogy Aster-oids II ed R P Binzel, T Gherels and M S Matthews(Tucson, AZ: University of Arizona Press) pp 98–127

Thomas P C, Binzel R P, Gaffey M J, Storrs A, Wells Eand Zellner B H 1997 Impact excavation on asteroid4 Vesta: Hubble Space Telescope results Science 2771492–5

Christophe Dumas


Viking Mission E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Viking MissionThe Viking mission was a spacecraft mission to MARS whosemain purpose was to search for life. Four spacecraftwere involved: two identical orbiter–lander pairs. Eachpair was launched separately in the summer of 1975 andarrived at Mars the following year. At Mars the orbiterssearched for safe and scientifically interesting sites. Thelanders were then targeted to those sites and conducted avariety of life detection experiments. The consensus viewwas that no compelling evidence for life was found (seealso LIFE ON OTHER WORLDS).

The spacecraftThe Viking Orbiters (VO-1 and VO-2) were designed andbuilt by the JET PROPULSION LABORATORY (figure 1). Theorbiters were solar-powered, three-axis stabilized vehicleswith substantial propulsive capability. The launch mass ofthe orbiters was over 2300 kg of which more than 1400 kgwas fuel required for orbit insertion. The orbiter sciencepayload was mounted on a moveable scan platform whichprovided both thermal control and pointing capability.In addition to communicating directly with Earth, theorbiters were capable of relaying data from the landersas well.

The Viking Landers (VL-1 and VL-2) were builtunder contract to Martin Marietta Aerospace Corporation(figure 2). The contract was managed by NASA’sLangley Research Center. The landers housed thescientific instruments and were powered by radioisotopethermoelectric generators. Each lander had a samplingarm with a scoop for returning samples to selectedinstruments for analysis. Meteorology sensors weremounted on a boom, and a seismometer was mountedon the lander underside. The landers had a two-waycommunication link with Earth and a one-way (transmit)link with the orbiters. The mass of the landers was about600 kg.

To avoid possible contamination from Earth organ-isms, the landers were heat sterilized and encapsulated ina bioshield. The encapsulated landers, including the entrysystem, had a total mass of almost 1200 kg. The landerswere then mated to the orbiters and the combined craftplaced in the shroud of a Titan III launch vehicle. The finallaunch mass of the orbiter–lander pair was 3530 kg.

ExperimentsWhile the primary scientific objective was to search forlife, other objectives were to image the surface and moons,determine the composition of the atmosphere and surface,monitor the weather and climate, and detect marsquakes.To meet these objectives, 13 investigations were conducted(table 1): three mapping experiments from orbit, anatmospheric investigation from the lander during entry,eight investigations on the surface and a radio scienceexperiment.

The orbiter mapping experiments were carried outby three instruments: (1) a pair of high resolution

vidicon cameras incorporated into a Visible ImagingSubsystem (VIS), (2) an Infrared Thermal Mapper (IRTM)to determine the thermal properties of the surface andatmosphere and (3) a Mars Atmospheric Water Detector(MAWD). Each of these instruments was mounted on thescan platform and aligned along a common axis to viewthe same area of the planet.

During the time the landers were traversing theatmosphere, an entry science investigation was conducted.This consisted of a retarding potential analyzer to measurethe ion and electron abundances of the upper atmosphere,a mass spectrometer to measure neutral species anda three-axis accelerometer to measure the spacecraftdeceleration history. Pressure and temperature were alsomeasured (see also MARS: ATMOSPHERE).

The lander cameras were designed to characterize thelocal geology, look for macroscopic evidence of life andsupport the other lander experiments. The two camerasalso provided a stereoscopic capability. Furthermore,by monitoring the brightness of the Sun and sky, thecameras could provide information on the opacity of theatmosphere.

The lander biology investigation consisted of threeexperiments. Each of these acquired small samples ofMartian surface material and exposed them to substancesof various kinds and then monitored the results. TheCarbon Assimilation Experiment was designed to detectphotosynthesis of organic compounds by introducingradioactively labeled CO and CO2 gases incubated underMars-like conditions. The Labeled Release Experimentsought to detect metabolic processes by introducing anaqueous solution of labeled simple organic materials.The Gas Exchange Experiment also looked for metabolicprocesses, but by wetting Martian soils with a morecomplex and unlabeled nutrient solution.

While not principally a life detection instrument,the landers also carried a GCMS which could detectorganic compounds in surface material, and measure thecomposition of the atmosphere. They also carried an x-rayfluorescence spectrometer to measure the abundance ofselected elements in Martian surface material. In additionto determining the composition of surface materials, thesedata provided chemical and mineralogical information tohelp interpret the biology experiments.

The remaining lander experiments were not directlyrelated to the biology investigation. The landermeteorology sensors recorded the pressure, temperatureand wind velocity to help characterize Martian weathersystems. The lander seismometers were designed tocharacterize the background noise level and to listen formajor seismic events which would provide informationon the interior structure of the planet. An arrayof magnets of differing strengths was attached to thesampling arm and to the top of the lander. By monitoring(with the camera) the accumulation of particles on thesemagnets, information on the abundance and compositionof magnetic minerals could be obtained. Finally, thephysical properties of the soil could be ascertained from



Figure 1. Viking Orbiter.

Figure 2. Viking Lander.

engineering data such as variations in motor currents fromthe sampler arm during digging operations (see also MARS:

SURFACE).The communication system (along with a tracking

system) on the orbiters and landers was also usedfor science. The orbiter radios mapped the planet’sgravity field, measured its surface relief and determinedlocal atmospheric temperature profiles during occultationevents. Lander radios were used to determine landerlocations, the spin rate of the planet and the orientationof the spin axis.

Mission profileViking 1 and 2 were launched respectively on 20 Augustand 9 September 1975 and were inserted into ellipticalorbits around Mars on 19 June and 7 August 1976. After

several weeks of surface observations, a landing site wasselected and the landing sequence was initiated. Thelander separated from the orbiter and oriented itself for a4 km s−1 entry into the Martian atmosphere at about 300 kmabove the surface. An ablatable aeroshell protected thelander from the heat of entry. At 6 km above the surface,a parachute was deployed, the aeroshell was jettisonedand the three lander legs were extended. At 1.5 km, retro-rockets were fired to slow the lander to several m s−1 andeliminate horizontal drift. VL-1 touched down on 20 July1976 in Chryse Planitia (22N, 48W) at about 4 p.m. localtime; VL-2 landed on 3 September 1976 in Utopia (44N,226W) at about 10 a.m. local time. Both landings weresuccessful.

After landing, there were several months of intenseactivity. The cameras systematically returned images



Table 1. Science investigations and instrumentation.

Investigation Instrumentation

Orbiter investigationsOrbiter imaging Two vidicon camerasWater vapor mapping Near-infrared grating spectrometerThermal mapping Solar and infrared radiometersEntry science Retarding potential analyzer, mass spectrometer,

pressure, temperature and acceleration sensorsLander investigations

Lander imaging Two facsimile camerasBiology Carbon Assimilation, Labeled Release, and

Gas Exchange ExperimentsMolecular analysis Gas chromatograph–mass spectrometer (GCMS)Inorganic analysis X-ray fluorescence spectrometerMeteorology Pressure, temperature, wind velocity sensorsSeismology Three-axis short-period seismometerMagnetic properties Permanent magnets on sampler armPhysical properties Various engineering sensors

Orbiter and lander investigationsRadio science Orbiter and lander communication systems

of the entire panorama around the lander. They alsophotographed the sky, the Sun, small permanent magnets,trenches dug by the sampler arm and rocks that wereintentionally displaced. The meteorology boom wasdeployed and began returning weather data. Seismicexperiments were initiated, although the seismometer atVL-1 failed to uncage (the VL-2 seismometer operatednominally). Numerous soil samples were acquired andsubjected to chemical and biological analysis.

The primary mission was to last 90 days for eachlander. However, the mission was repeatedly extendedas long as the spacecraft remained healthy and returneduseful data. The first hardware problem occurred inSeptember of 1977 when VL-2 was no longer able tocommunicate directly with Earth, but it was not shut downuntil 12 April 1980. Leaking attitude control propellantleft the VO-2 spacecraft unable to align itself with the Sunand it ceased operations on 25 July 1978. VO-1 was alsorunning out of fuel and was commanded off on 7 August1980. The mission finally ended on 13 November 1982when contact was lost with VL-1.

Scientific resultsBiologyThe most significant result was the lack of detection oforganic compounds at either landing site by the GCMSexperiment. The GCMS was extremely sensitive to organiccompounds and could detect them at the parts per billionlevel for large molecules, a level some 100 times less thanthose found in deserts on Earth. Yet the results wereunambiguously negative. The absence of organic matter,regardless of the results of the biology experiments, arguedstrongly against the possibility for life in Martian soil.

While the three biology experiments produced resultswhich in some ways simulated life, in detail they weremore consistent with the presence of soil oxidants. TheGas Exchange Experiment showed a significant increase

in CO2 and O2 after the soil was humidified. The CO2

can be explained by the displacement of adsorbed CO2

molecules by water vapor, but the O2 requires some otherexplanation. The observed 200-fold increase in O2 isbest explained by the decomposition of peroxides in theMartian soil. Such peroxides are theoretically expectedfrom photochemical processes in the Martian atmosphere.Also, a highly reactive oxidant in the Martian soil couldreadily explain the absence of organic material.

The Labeled Release Experiment showed a surge inradioactive CO2 after wetting the soil with its mixture ofnutrients. Taken by itself, this result satisfied the criteriafor a biological interpretation. However, when viewed incontext with the other biology experiments and the GCMS,this too was believed to be a chemical rather than biologicalresult. Peroxides present in the soil would have oxidizedone of the compounds in the nutrient solution (formic acid)and produced CO2 gas. Also, if microorganisms wereinvolved, a second injection of the nutrient solution shouldhave produced a further increase in CO2 levels, yet nonewas observed.

The results of the Carbon Assimilation Experimentswere more complicated, but they too favor a chemicalexplanation. The amount of fixed carbon produced inthese experiments was generally higher than the expectedbackground levels, particularly the first experiment atVL-1. This experiment also showed a temperaturesensitivity to the amount of fixed carbon produced. Itis now believed that an iron-catalyzed reaction was thesource of the signals seen in the Carbon AssimilationExperiment.

MeteorologyLander meteorological measurements showed that the av-erage surface pressure on Mars is about 7 mbar. However,it varies semiannually owing to the condensation and sub-limation of CO2 in the polar regions. The amplitude of the



fluctuation is about 2.3 mbar, or about one-third of theannual mean. Thus, the seasonal polar caps on Mars aremade of dry ice (frozen carbon dioxide).

The pressure data also revealed eastward-travelingweather systems and westward-propagating thermaltides. The temperature at the Viking sites ranged from180 K to 240 K depending on time of day and seasonand was more repeatable during summer than winter.Winds were found to be strongest during winter with gustsup to 30 m s−1. During summer winds were calm bycomparison.

During autumn of the first year, the sky opacity atthe Viking sites increased dramatically on two occasions.These increases were associated with the development ofglobal scale dust storms. No such storms were observedduring the second and third (Mars) years of operations.

SeismologyNo major seismic events were detected. A magnitude 2.8event (on the Richter scale) was detected 110 km from theViking Lander 2 site. The signal was significantly dampedwithin minutes, indicating the likely presence of waterand gas within the crust. From shear wave reflections,the crust was estimated to be about 15 km thick in thisregion. The natural background level of seismic activitywas found to be very low. Winds were a major source ofthe seismic background because of the poor coupling ofthe seismometer to the ground.

Atmospheric compositionThe composition of the Martian atmosphere was measuredby the Viking landers during entry and descent andwhile on the surface. The principal components are CO2

(95.32%), N2 (2.7%), Ar (1.6%) and O2 (0.13%). The meanmolecular weight is 43.34. These gases were found to bewell mixed up to 120 km altitude. Above 120 km they beginto diffusively separate. In the ionosphere O+ was foundto be the dominant species with CO+ about an order ofmagnitude less abundant.

Isotopic ratios of various elements were alsomeasured. The stable ISOTOPES of carbon and oxygen werefound to be similar to Earth’s, but 15N/14N was enrichedby 60%. The enrichment is due to the selective escape ofthe lighter isotope with respect to the heavier one, andit implies that Mars may have had a denser atmospherein its distant past. 40Ar/36Ar and 129Xe/132Xe were alsofound to be enriched with respect to terrestrial values. Inthis case, however, the enrichment is thought to be due toremoval of a substantial fraction of Mars’ early atmosphereby impacts events followed by the accumulation of theradiogenic isotopes 40Ar and 129Xe.

Atmospheric water detectionWater vapor in the Martian atmosphere was found to varyseasonally and spatially. On average, a column of Martianair contains the equivalent of a layer of liquid water about10 µm deep. EARTH’S ATMOSPHERE, by comparison, contains2–4 cm of water. Approximately 100 µm of water vapor

was observed over the north polar cap during summer.Very little water vapor was observed over the south polarcap during its summer. This difference in behavior is dueto the different compositions of the summer residual caps:the north summer cap is made of water ice, while that inthe south is made of CO2 ice.

Atmospheric structureDuring their descent to the surface, the Viking landersrecorded the spacecraft deceleration history. These datawere used to construct a temperature profile from near thesurface to about 200 km. The profiles at the two landingsites were found to be very similar even though they werecollected at different locations and times. Temperaturesdecreased from 230 K near the surface to about 150 K aloft,and they showed considerable structure possibly due tothermal tides and/or gravity waves.

Thermal mappingSurface temperatures were mapped by the IRTM instru-ment. Highest temperatures (300 K) occur in early af-ternoon in southern subtropical latitudes during summer.Lowest temperatures (130 K) occur in polar regions dur-ing winter. The daily variation in surface temperaturewas used to map the thermal inertia of the soil. Ther-mal inertias range from about 50 to 500 SI units and in-dicate fine sandy material and well-consolidated bedrock,respectively. Atmospheric temperatures near 25 km werealso mapped and showed great sensitivity to dust.

Orbiter imagingOver 50 000 images were returned by the Viking orbiters.The images covered the entire planet at a resolution of200 m and large areas at resolutions as high as 7.5 m. Muchof the planet has been photographed in color and stereo,although at lower resolution. Orbiter cameras also imagedthe moons of Mars (PHOBOS AND DEIMOS), and photographedthe shadow of Phobos as it crossed the VL-1 site.

For almost two Mars years, the orbiter camerasmonitored weather patterns, the advance and retreat ofthe polar caps, dust storm activity and the continualredistribution of fine-grained material on the surface.They returned pictures of enormous volcanoes, a vastequatorial canyon system, numerous impact craters,polar layered deposits and fluvial features suggestingcatastrophic flooding and possible climate change.

The northern hemisphere was found to be geologi-cally young and covered with volcanic fields. The giantshield volcanoes in the Tharsis region have relatively fewcraters indicating that volcanism lasted for an extendedperiod of time. In contrast, the southern hemisphere isold and heavily cratered. Parts of its surface may haveformed at the end of the heavy bombardment period 3.8billion years ago.



Soil propertiesThe fine-grained material at the two Viking sites hassimilar elemental abundances. Silicon is the mostabundant element (21%), followed by Fe (13%), Mg (5%),Ca (4%), Al (3%) and S (3%). The surprisingly uniform soilcomposition suggests planetary-scale homogenization ofthe soil by winds. The soil is weakly cohesive and hasa bulk density of about 1.2 g cm−3. Magnets on top ofthe landers attracted soil-derived airborne dust. About2% of the soil contains magnetic material with maghemite(γ -Fe2O3) being the most likely candidate. A plausiblemodel for the soil material is that it is a basaltic weatheringproduct.

BibliographySoffen and Young 1972 Icarus 16 1–16

give a description of the Viking mission as it was originallyenvisioned. For an historical perspective of the Vikingmission see

Ezell and Ezell 1983 On Mars: Exploration of the Red Planet,1958–1978 NASA SP 4212

The Viking Project as flown is described by

Soffen 1977 J. Geophys. Res. 82 3959–70

The first results of the Viking mission were published in

1976 Science 193 (27 August)1976 Science 194 (1 October)1976 Science 194 (17 December)

More comprehensive papers were published in specialissues:

1977 J. Geophys. Res. 82 (September)1978 Icarus 34 (June)1979 J. Geophys. Res. 84 (December)1982 J. Geophys. Res. 87 (November)

The state of the art in our thinking about Mars has beensummarized in the book

Kieffer et al (ed) 1992 Mars (Tucson, AZ: University ofArizona Press)

Robert M Haberle


Vilnius University Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Vilnius University ObservatoryVilnius University Observatory, the astronomical obser-vatory of Vilnius University, Lithuania, was founded in1753. In 1831, when Vilnius University was closed, the ob-servatory was entrusted to the Academy of Sciences of StPetersburg (Russia) and continued its activities until 1881.The Observatory resumed its operation in 1919 when Vil-nius University was reopened. Activities of the observa-tory were again interrupted by Nazi occupation (1941–44)and resumed after World War II. Its main instruments area 60 cm reflector located at Moletai Observatory 80 kmnorth of Vilnius and a 48 cm reflector located at the Maid-anak observing site in Uzbekistan, Central Asia. Presentstaff consists of eleven astronomers and six technicians.Activities are in the field of construction of photometricequipment, investigation of physical properties of stars,interstellar matter and galactic structure.

For further information seehttp://www.ff.vu.lt/astro/inde40.htm.


Virgo E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Virgo(the Virgin; abbrev. Vir, gen. Virginis; area 1294 sq. deg.)an equatorial zodiacal constellation which lies betweenLeo and Libra, and culminates at midnight in mid-April.Its origin dates back to Babylonian times and it has beenassociated with numerous female deities, including Ishtar(Babylonia), Isis (ancient Egypt), Demeter (ancient Greece)and Astraea (ancient Rome). Its brightest stars werecatalogued by Ptolemy (c. AD 100–175) in the Almagest.

Virgo is the second-largest of the 88 constellations butis not particularly prominent, save for its brightest star,αVirginis (Spica or Azimech), magnitude 1.0. Other brightstars include γ Virginis (Porrima), a fine binary with paleyellow (F0) components, both magnitude 3.6 (combinedmagnitude 2.7), separation 2.7′′, period 168.7 years (closest2005), ε Virginis (Vindemiatrix or Almuredin), magnitude2.9, ζ Virginis, magnitude 3.4, and δ Virginis, alsomagnitude 3.4. There are five other stars of magnitude4.0 or brighter.

There are no bright open star clusters or nebulaein Virgo, but the constellation contains a large numberof galaxies, many of which are members of the VirgoCluster, which extends into neighboring Coma Berenices.The brightest members of the Virgo Cluster are the giantelliptical galaxies M49 (NGC 4472), 8th magnitude, andM87 (NGC 4486, VirgoA), 9th magnitude, which is a strongx-ray and radio source and from which emanates a jet ofmaterial thought to have been expelled from a black hole atits center. An interesting non-cluster galaxy is M104 (NGC4594, the Sombrero Galaxy), an 8th-magnitude spiral witha dark lane of obscuring dust presented almost edge-on.

The constellation also contains the first quasar to havebeen detected (in 1963), 3C 273, magnitude 12.9.

See also: Sombrero Galaxy, Spica, Virgo A, Virgo Cluster.


Virgo A E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Virgo AThe brightest radio source in the constellation of Virgo.Virgo A, otherwise known by the catalog number 3C 274,coincides with M87 a giant elliptical galaxy, and activegalaxy, located at a distance of some 50 million light-yearsin the Virgo cluster of galaxies. Although part of the radioemission comes from two elongated lobes, one on eitherside of the center of the galaxy, the lobes, which span anoverall diameter of about 16 000 light-years, are containedwithin the optical galaxy (rather than, as with most radiogalaxies, extending beyond its visible perimeter). Themajority of the radio output comes from a jet that emergesfrom the core of the galaxy and stretches out some 8000light-years into one of the two lobes.

The jet radiates at all wavelengths from x-ray to radio.Like the lobes, it emits synchrotron radiation (radiationthat is generated by electrons moving at very high speedsin magnetic fields). The motion of bright knots in the jetindicate that they are traveling outwards at about half thespeed of light and implies that the electrons themselvesmust be traveling within the jet at speeds at least as high,if not higher, than this.

With a mass of at least 1012 solar masses, M87 is anexceptionally massive galaxy. Observations of the rates atwhich stars and gas clouds revolve within its central core,and of the way that stars appear to be concentrated at itscenter, imply that M87 contains a compact massive object,most probably a black hole, of about 3 billion solar masses.

See also: active galaxy, astrophysical jets, radio galaxy,synchrotron radiation, M87, Virgo cluster.


Virgo Cluster E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Virgo ClusterThe Virgo cluster is the nearest and best-studied richCLUSTER OF GALAXIES, lying at a distance of about 55 millionlight-years in the constellation of VIRGO. Figure 1 is amap of the distribution of nearby galaxies (as determinedby their REDSHIFTS) in a 1500 square degree region ofthe constellation of Virgo. The Virgo cluster is thestrong, somewhat irregular, concentration of galaxies atthe center. In three-dimensional (3D) space, the Virgocluster constitutes the nucleus of the Local Supercluster(LSC) of galaxies, in whose outskirts our MILKY WAY GALAXY

is situated.As early as 1784, CHARLES MESSIER noted an unusual

group of ‘nebulae’ in Virgo. Fifteen out of the 109 famous‘Messier objects’ are, in fact, member galaxies of the Virgocluster. However, only in the 1920s, following EDWIN

HUBBLE’s proof of the extragalactic nature of those ‘nebulae’,was Messier’s group understood as a self-gravitatingsystem of hundreds of galaxies, and the first systematicinvestigations of the Virgo cluster, as it was henceforthcalled, were carried out by HARLOW SHAPLEY and others.

Virgo was the first galaxy cluster to be studieddynamically by Smith and Zwicky in the 1930s. Theirwork showed that the dynamical mass of Virgo, estimatedby using the virial theorem

2T + U = 0

where T and U are the time-averaged kinetic energyand potential energy of the relaxed system, was muchlarger than the mass inferred by integrating the lightof all the galaxies in the cluster and multiplying by amass-to-light ratio (M/L) like the average of stars in thesolar neighborhood. This was the first clear detectionof ‘DARK MATTER,’ or more properly, non-luminous mass.The distribution of galaxies in the direction of Virgo inredshift space is shown in figure 2, and figure 3 displays thehistogram of galaxy velocities inside a circle of 6 radiuscentered on the core of the cluster at 12h28m and +1028′.

The Virgo cluster lies at the center of the LSC,first dynamically studied by G de Vaucouleurs andcollaborators in the 1950s. The Milky Way appears tobe falling into the Virgo cluster, relative to the generalexpansion of the universe, with a velocity of∼250 km s−1.That is to say, we are still moving away from the clustercore with an apparent velocity, cz, of ∼1100 km s−1, butthat velocity is 250 km s−1 less than it would have beenif the cluster had no mass. Curiously, the M/L for thewhole LSC derived from this Virgo infall is similar to thatderived via the virial theorem for the dynamically relaxedcluster core; both M/L values, when applied to the meanluminosity density of the universe derived from redshiftsurveys, imply a mean matter density of the universe thatis only 1/4 to 1/3 of the critical density.

The kinematics of the cluster is quite complex ascan be seen in figures 1 and 4. There are two mainconcentrations of galaxies; the largest and densest is

centered on the well-known galaxy M87 (NGC4486) whichis both a strong radio source (VIRGO A) and a strong x-ray source. This subcluster, at 12h30m.8 + 1223′, is also astrong x-ray source produced by a reservoir of extremelyhot intracluster gas. This INTRACLUSTER MEDIUM is at atemperature of ∼107 K, emits thermal bremsstrahlungradiation and is a major contributor to the total mass ofthe system. The southern large subcluster is centered onM49 (NGC 4472) at 12h29m.8 + 800′, the most luminousgalaxy in the cluster, but is significantly less massive thanthe M87 subcluster. The total mass of the Virgo core regionis ∼1015M, that of the M87 subclump ∼3 × 1014M andthe M49 subclump is ∼1 × 1014M, assuming a distanceto the cluster core of 16 Mpc. The cluster distance of∼16 Mpc has been estimated via a variety of techniquesincluding measurements of Cepheids to six galaxies withthe Hubble Space Telescope, studies of globular clustersand planetary nebulae (PNe) luminosity functions, andsurface brightness fluctuations (SBFs).

At present, approximately 1300 member galaxies ofthe Virgo cluster are known, most of which are very faintDWARF GALAXIES (but this number is bound to grow fastin the near future, as ever fainter members are going tobe detected with improving techniques). Their projectedpositions within the cluster—spread over a large sky areaof roughly 100 square degrees, or 500 full moons (seefigure 4)—reveal a rich substructure. The Virgo clusterindeed represents the most common type of clusters ofgalaxies that are loosely concentrated and ‘irregular’.Several gravitationally bound subsystems of galaxies havebeen identified. Of these, the subcluster centered onMessier 87 is by far the largest and most massive structure(the Virgo cluster proper, so to speak)—and M87 itself,the giant, active galaxy with its famous jet, can truly beregarded as the heart of the Virgo cluster. The smallersubclusters seem to be in a state of merging with the M87subcluster. In this sense, the Virgo cluster (like many, ifnot most, clusters of galaxies) is still in the making.

The Virgo cluster has always been, and still is, one ofthe most important stepping stones for the cosmologicaldistance scale. Much of the current debate on the value ofthe HUBBLE CONSTANT boils down to a debate on the meandistance of the Virgo cluster (see below).

A new and very exciting era of Virgo cluster researchhas recently been opened by the first detection ofintracluster PLANETARY NEBULAE (PNe) and single RED GIANT

STARS, i.e. stellar objects that are not bound to a single galaxybut to the cluster as a whole. This adds a whole newpopulation to the cluster, which in the future will be usedto explore the 3D structure and dynamical history of theVirgo cluster.

Global structureThe primary optical database for the galaxy content ofthe Virgo cluster is the Las Campanas photographicsurvey, carried out in the 1980s by Allan Sandage andcollaborators, and encapsulated in the Virgo ClusterCatalog (VCC). Figure 4 is a map of the ca 1300 galaxies in



Virgo

RA

DEC

Figure 1. The distribution of nearby galaxies (apparent recession velocities less than 3000 km s−1) in the direction of Virgo. Each circlerepresents a galaxy; the size of the circle represents its apparent brightness. Galaxies with apparent velocities less than 500 km s−1 areshown in red, those above 2100 km s−1 in green, and those between 500 and 1300 km s−1 and between 1300 and 2100 km s−1 in blueand magenta, respectively. This figure is reproduced as Color Plate 63.

the sky area covered by the Las Campanas survey thatwere judged to be members of the Virgo cluster. Themembership criteria were based on (1) the morphologicalappearance of the galaxies, e.g. dwarf ellipticals have acharacteristically low surface brightness, and/or (2) themeasured radial velocities, which for members have to besmaller than ca 2700 km s−1 (heliocentric). Velocities are atpresent available only for the brightest 400 members (butthis number, too, will steadily grow in the future).

The magnitude limit of completeness of the LasCampanas survey is around apparent blue magnitudeB = 18 or, if we assume a mean distance of 16 Mpc,absolute magnitude MB = −13.0. However, faintermembers (up to B = 20) were included as well. Forcomparison, the brightest cluster members, M49 and M87,have B 9.0 and B 9.5, respectively. Thus the knowncluster population spans a range of≈10 000 in luminosity.

From the distribution of symbol sizes in figure 4 one canobtain a feeling of the exponentially growing luminosityfunction of galaxies. Of the 1300 known members, 850alone are of the dwarf elliptical (dE) type. The basiccharacteristic of dEs is a relation between luminosity andsurface brightness: fainter dwarfs have also lower surfacebrightness. As the detection limit for extended objects suchas galaxies is set by the surface brightness rather than totalmagnitude, it is clear that many hundreds, if not thousandsmore, of extremely faint and diffuse dE members of theVirgo cluster have yet to be discovered by future deepsurveys. These will be the analogs of the dwarf spheroidalcompanions of our Galaxy, which are as faint as MB = −8(corresponding to B ≈ 23 at Virgo distance).

Among the remaining 450 member galaxies, thereare roughly 80 elliptical (E) and S0 galaxies, 130 spirals,90 irregulars and 90 dwarf galaxies of intermediate type



Figure 2. The distribution of galaxies in redshift space in the direction of the Virgo cluster. We are at the apex of the wedge. The mainbody of the cluster is the picturesquely named ‘Finger of God’ at the center of the wedge. This is not a real feature. Its extension alongthe line of sight in this plot is a measure of the internal velocity dispersion (and thus the mass) of the galaxy cluster so that a galaxywith a given radial velocity is placed in front of or behind its true position. The large and diffuse structure seen to the right (10.5h–11h

and 1100 km s−1) of Virgo is the Leo group.

Figure 3. Histogram of apparent velocities for galaxies within a 6 radius of the center of the Virgo cluster. The breadth of thedistribution from 0 to 300 km s−1 accounts for the ‘Finger of God’ structure in figure 2.

(Irr–dE). The distribution of these morphological typeswithin the cluster varies considerably—in accord withDressler’s GALAXY ‘MORPHOLOGY–DENSITY RELATION’: Es and S0s

are strongly confined to the regions of highest galaxydensity, defining the ‘skeleton’ of the cluster (dwarf Es onlyto a slightly lesser degree), while spirals and irregulars



Figure 4. Map of the Virgo cluster as it appears in the sky. Allcluster members are plotted with luminosity-weighted symbols.The symbol size (area) is proportional to the luminosity of thegalaxy. The most prominent Messier galaxies are indicated. Thebroken lines indicate the boundary of the Las Campanas survey.

(S+Irr) are very weakly clustered, lying preferentially inthe cluster outskirts.

This morphological segregation is also reflected in thevelocity distributions. An analysis of the radial velocitydata so far available gives a velocity dispersion (1σ ) of≈600 km s−1 for early-type galaxies (E, S0, dE), but one of≈750 km s−1 for late types (S+Irr), i.e. late types are moredispersed in space and velocity. Moreover, the velocitydistribution for S + Irr is distinctly non-Gaussian witha low-velocity and a high-velocity wing. These wingsare probably caused by infalling and expanding shells ofgalaxies around the cluster core. Most Virgo spirals andIrrs may have fallen only recently, or are still in the stage offalling, into the cluster from the surroundings, explainingtheir lack of dynamical relaxation (virialization). Thosespirals observed to be H I deficient have apparentlyalready fallen through the cluster core.

The all-cluster mean heliocentric velocity is 〈v〉helio 1050 km s−1 (there is no difference with respect to type).Note that this is not the cosmic (Hubble flow) recessionvelocity of the Virgo cluster, 〈v〉c, which can be derivedby (1) correcting for the solar motion with respect to thecentroid of the Local Group by subtracting ≈100 km s−1

and (2) correcting for the Virgocentric infall (deceleration)of the Local Group by adding ≈250 km s−1. This finally

gives 〈v〉c 1200 km s−1.

Subcluster dynamicsThe primary structural characteristic of the Virgo cluster(cf figure 4) is certainly its high degree of irregularityand subclustering. Several subclumps (gravitationallybound subclusters) of galaxies suggest themselves: a mainnorthern subclump around M87, M86 and M84, a southernclump around M49 and possible subgroups around M60and M100. The global structure of Virgo seems definedby two main axes: one N–S, i.e. M100–M86/M87–M49,and one E–NW, i.e. M60–M87–M86/M84. Remarkably, theformer axis is nearly perfectly aligned with the positionangle of the outer isophotes of M87, while the latter isperfectly aligned with the jet axis of M87. This shows oncemore that M87 is the heart of the Virgo cluster.

The reality of the main N–S double structure cannotbe doubted, because the northern and southern subclumpsare sufficiently well separated. Both have a very similarmean radial velocity, which would suggest that they areat the same distance, i.e. lying in the plane of the sky(however, see below). The southern M49 clump is ratherspiral rich and has a surprisingly small velocity dispersionof σv ≈ 500 km s−1.

The core region with M87, M86 and M84 is muchharder to disentangle. The key observation here isthat the velocity distribution of galaxies in this northernclump (especially for dEs) is strongly skewed towardslow velocities. In the low-velocity tail we find the mostblueshifted galaxies known in the sky (the record holder,VCC846, has v = −730 km s−1). These objects tend to beclustered around M86, which itself has a negative velocity(v = −227 km s−1). On the other hand, the velocitydistribution is peaked around v = 1300 km s−1—nearlycoinciding with the velocity of M87 (v = 1258 km s−1).

A clear asymmetry in the velocity distribution of acluster of galaxies is almost certainly an indication ofongoing subcluster merging. In the Virgo cluster we seemto witness the merging between a subclump around M87and another clump around M86 (or rather, the infall of theM86 subcluster into the more massive M87 subcluster, seebelow). Both giant galaxies must be the centers of hugeswarms of dwarf galaxies. M86 is apparently falling into,or through, the M87 subclump from the back, hence with ahigh relative (negative) velocity, dragging along its dwarfcompanions. (The giant galaxy M84, very close to M86 inthe sky, but with v = 1000 km s−1, has to be a member ofthe M87 subcluster.)

This whole picture is fully confirmed by an analysisof the x-ray structure of the Virgo cluster. Figure 5shows, as a gray-scale image, the distribution of the x-rayintensity in the Virgo cluster as measured by the ROSAT x-ray satellite. The x-rays originate from the hot intraclustergas via thermal bremsstrahlung. The gas feels the samegravitational potential of the cluster as the galaxies. Theglobal appearances of the cluster in the x-rays and in theoptical (figure 5 versus figure 4) are therefore very similar.However, the main subclusters around M87, M86 and M49



Figure 5. X-ray image of the Virgo cluster from theROSAT All-Sky Survey. Various foreground stars, quasars, Abellclusters and Virgo cluster members (NGCs and Messiers) areindicated. The large, bright spot is centered on M87. Comparewith figure 4. Courtesy of Dr H Bohringer and the MPEGarching.

are popping out much more clearly in the x-ray image.M87 has a huge x-ray halo and is obviously the center of thedominating subcluster in Virgo. M86 has a large gas haloas well—larger than expected for an isolated giant, but ofthe right order for a whole subcluster. The infall of thissubcluster from the back is supported by the observationof a ram pressure plume in the x-ray structure of M86.

Subcluster merging has turned out to be a generalfeature of cluster of galaxies, even with so-called ‘regular’ones, such as the COMA CLUSTER. This is not unexpected ina bottom-up (clustering) picture of structure formation, asdescribed by the much favored ‘cold dark matter’ scenario.

Under the (tested) assumption of hydrostatic equilib-rium, the x-ray halo profiles can be used to calculate thegravitational masses of the subclusters that are bindingthe gas. For the dominating M87 subcluster the integratedmass out to 1 Mpc from M87 amounts toM 1.4×1014M,of which 14% is in the gas, and only 4% in the galaxies.This means that the cluster is essentially dark matter, sec-ond comes the intracluster gas and only in third place theluminous matter in galaxies, with a correspondingly highmass-to-light ratio of M/L ≈ 500. This is quite typical forclusters of galaxies.

Distance and depthThe mean distance of the Virgo cluster is still a matter ofdebate. Distances quoted range from 14 to 22 Mpc. Ingeneral, the most reliable extragalactic distance indicatorsare the Cepheids. Cepheids at (and slightly beyond)

the Virgo cluster distance are now within the reach ofthe HUBBLE SPACE TELESCOPE (HST). This achievement wasso long awaited that the first Cepheid-based distancedetermination of a Virgo cluster spiral in 1994 (M100 at17 Mpc) had an enormous impact. The caveat with thiswas, and is, that spirals tend to avoid the cluster core andmay be in the field far off the cluster (cf also above). It hasbeen conjectured that M100, as well as other spirals locatedby Cepheids with HST later on, might lie at the near sideof the cluster.

On the other hand, from work based on the TULLY–

FISHER (TF) RELATION, which allows the distance to anindividual spiral galaxy to be given with an accuracy of≈0.4 mag, there is consistent evidence that Virgo late typesare distributed in a prolate cloud, or filament, stretching—nearly along our line of sight—from the cluster backwardsto the so-called ‘W cloud’ at twice Virgo’s distance.Probably this is part of a very long filament that is runningway back to the ‘Great Wall’ at the distance of the Comacluster. On the near side of Virgo it might even beconnected with the ‘Coma–Sculptor cloud’ that is runningthrough us, i.e. includes the Local Group.

There is a hint, again from TF distances, that thesouthern M49 subcluster is lying significantly in the backof the M87 subcluster. If so, the M49 subcluster mustbe infalling from the back with a velocity of several100 km s−1, as the mean observed velocities of the twosubclusters are very similar. The merging of subclustersalong the large-scale filament in which they are embeddedis plausible.

To determine the distance of the core of the Virgocluster one should avoid late-type galaxies. The safestwould be to use only elliptical and dwarf ellipticalmembers. Unfortunately, the primary distance indicatorshere, RR LYRAE STARS, are much too faint at the distance ofVirgo even for HST. The secondary distance indicatorswhich can be applied to Virgo ellipticals give controversialresults: globular clusters, Dn–σ and novae tend to givelarge distances (D ≈ 20 Mpc), SBFs and PNe lead toa small D ≈ 16 Mpc. Great efforts are spent in theapplication of the SBF method because its claimed distanceuncertainty for an individual galaxy is almost as small aswith Cepheids (≤0.2 mag) and hence would allow us toresolve the cluster depth (the front-to-back depth is about2 Mpc, or 0.2 mag, if the cluster is spherical). IndividualVirgo E distances, with a surprisingly large scatter, haveindeed been reported, but there is still some concernwhether all variations of the stellar content of ellipticals,on which the method critically depends, are sufficientlywell understood. For instance, the SBF distance of M49 ismuch smaller than the (probably more reliable) TF distanceof the whole M49 subcluster, which means that eitherM49 is lying in the foreground, being projected on topof the background ‘M49’ subcluster by chance, or the SBFmethod is wrong.

Some hope for the future is resting with dwarfellipticals, of which there is an almost inexhaustiblereservoir in the cluster. A first, very promising



application of the tip-of-the-red-giant-branch (TRGB)distance indicator to a particular dE based on HSTobservations has given D 16 Mpc. A recent claim thatdEs are distributed in a prolate structure pointing towardsus, based on the shape of the luminosity profile of thesegalaxies, will soon be tested by an extension of the SBFmethod to dEs.

A new tool to determine the mean distance anddepth of the Virgo cluster core may be provided by therecently discovered population of free-floating giant starsand PNe. Such a population has long been suspected inclusters of galaxies, as stars will be ripped from galaxiesby tidal encounters in the cluster core, a process called‘galaxy harassment’. The data available so far suggest thatapproximately 10% of the stellar mass of the Virgo clusteris in intergalactic stars. There are ongoing programswith large telescopes to measure the radial velocities andmetallicities of a large number of intracluster PNe. Thesedata will provide crucial constraints on the dynamical stateand history of the Virgo cluster.

BibliographyBinggeli B 1999 The Virgo cluster—home of M87 The

Radiogalaxy M87 ed H-J Roser and K Meisenheimer(Berlin: Springer) p 9

Binggeli B, Tammann G A and Sandage A 1987 Studies ofthe Virgo cluster. VI. Morphology and kinematics ofthe Virgo cluster Astron. J. 94 251

Bruno Binggeli and John Huchra


Virgo Gravitational-wave Interferometer (VIRGO) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Virgo Gravitational-waveInterferometer (VIRGO)VIRGO is a collaboration between Italian and French re-search teams to construct an interferometric gravitationalwave detector at Cascina, near Pisa, Italy. It relies on atechnique called laser interferometry to measure with im-mense accuracy the minute changes in distance inducedby gravitational waves from astronomical sources.

The VIRGO project plans to set up a laser interferom-eter made of two, 3 km long, orthogonal arms. Light froma laser is split into two beams which travel down the armsand are then reflected back by mirrors. In effect, multiplereflections will increase the optical length of each arm to120 km. Variations in the pattern of interference betweenthe returning beams will enable gravitational waves to bedetected and their sources to be identified.

VIRGO will be sensitive to gravitational waves atfrequencies from 10 to 6000 Hz. It should be able to detectradiation produced by supernovae and by the coalescenceof binary systems in the Milky Way and other galaxies, forinstance from the Virgo cluster.

It will use high-power ultrastable lasers and high-reflectivity mirrors, as well as seismic isolation and controlof position and alignment. Each optical component isisolated by a 10 m high elaborate system of compoundpendulums, called a ‘superattenuator’. The 6 km long,1.2 m diameter evacuated tube through which the lightbeam passes will be one of the largest vacuum vessels inthe world. The signals will be detected, registered andelaborated by computer.

For further information seehttp://www.virgo.infn.it/.


Visual Binary Stars E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Visual Binary StarsAlready since the early development of telescopes, anincreasing number of stars—single to the naked eye—were found to consist of two or more components. Theybecame known as ‘double stars’ (etoiles doubles, dvoinyezvozdy, Doppelsterne) and soon were too numerous tobe explained as random pairings of stars in the sky.With increasing observing-time span, some pairs showedaround 1800 that the components orbited one another.The English term ‘BINARY STARS’ tends to emphasize thisgravitational connection. In distinction from close doublesrecognized by stellar eclipses or by periodic changesof spectral-line shifts (photometric and spectroscopicbinaries), the directly resolved objects are termed ‘visual’although eyepiece observations are no longer the onlymethod of study. Telescopes and techniques of the 20thcentury have improved the angular resolution to below 0.1arcsec, but in order that components be separated, that isbe recognised by their different positions in the sky, visualpairs are subject to at least one of two limitations.

(a) They are not very distant from Earth. Visual objectsthus are predominantly of the more populous typesin the solar neighborhood: stars from the middle andlower main sequence;

(b) The true distance between the components is not toosmall. The orbit periods are therefore generally muchlonger than in ECLIPSING and SPECTROSCOPIC BINARY STARS,and so it is important that measurements dating back100 yr and more are available for many of them.

In addition, detection is impeded when the stars havea large brightness difference. Hence not many lowest-luminosity dwarf components, or companions of brightsupergiants, are known.

Designations of visual double starsThe celestial position for J2000 (the beginning of the Julianyear 2000) is generally used, for instance 07346 + 3153means right ascension 7h 34.6m and declination 31 53′

N (Castor). Cataloguing confusion can occur when thepositions of faint stars are not well enough determined orwhen the transfer to J2000 ignored proper motions.

Also used are discoverers’ codes, e.g. Bu 1077 for αUrsae Majoris (Dubhe) after the finding by S W BURNHAM.These letter codes are unambiguous but owing to theirvariety a bit cumbrous. The master file of visualdouble stars, maintained at the US Naval Observatory,Washington DC, and currently being revised, lists boththese identifiers for nearly 100 000 entries, and in additionthe Durchmusterung number (Bonn, Cordoba, Cape), ifany.

The double-star catalogue by R G AITKEN containsobjects discovered before 1927 in the sky north of −30declination. Since it includes many well-known binaries,the ADS number is still used. Some publications cite theHIP number after the star catalogue from the HIPPARCOS

satellite mission.

Some results of statisticsAdding all kinds of binaries, and with a statistical estimateof the occurrence of undetected, low-mass components,the fraction of stars on the upper and middle mainsequence which are members of binary and multiplesystems comes to about 80%. This membership ratedecreases toward RED DWARF STARS, and for faint dM types—of which only the nearest are well studied—may be aslow as 40%. Thus the most abundantly occurring stellarmasses, around 0.2 of the solar mass, are less frequentlyfound in binaries.

Very wide pairs have only a weak gravitational bond,which may be disrupted by fairly close random encounterswith passing stars. As estimated from the star density inthe Galactic disk, the limit of separations, below which apair is expected to remain bound through its lifetime isat 10 000–20 000 AU (astronomical units), or 0.05–0.1 pc.Existence of orbit periods up to some millions of years isthus to be expected.

Inevitably some pairs of stars appear close togetherby chance alignment, although the components are atdifferent distances and unrelated. These are ‘optical’or ‘perspectivic’ double stars. Statistics show that theprobability of a random alignment to within 5′′ or so isquite low for stars down to about 12th magnitude. Mostof the catalogued double stars thus are genuinely physicalsystems.

Observing techniquesVisual measurement by micrometerThe FILAR MICROMETER contains a set of thin wires, onemovable against another with a precision screw, andmounted on a rotatable frame in the focus of a high-power eyepiece. This simple device supplies the positionof one double-star component relative to the other inpolar coordinates: the position angle (direction in thesky) counted from N over E, S, W from 0 to 360 byorientation of the frame, and the separation between starsin arcsec by the spacing of the wires. Since the eyehas a good time resolution of 0.05 s, it can suppressmuch of the atmospheric turbulence, often reaches thediffraction-limited resolving power provided by thetelescope aperture and may even recognise elongateddouble-star images below resolution. (The often-heardstatement that atmospheric blurring limits ground-basedinstruments to about 1 arcsec of resolution is false; itapplies only to long recording times.) Although a 17th-century invention, the filar micrometer became a high-precision tool only in the hands of F W STRUVE in 1820.Since then most double-star positions have been obtainedthat way, quite close binaries being reached already withthe refractors around 1900 (Aitken and others), at least inthe northern hemisphere. The southern sky lagged behinduntil W H van den Bos began his micrometric productivityin the 1920s, and R A Rossiter followed with the discoveryof many thousands of fainter pairs.

A modification favored by some observers is thedouble-image micrometer which uses a birefringent prism.



The four star images produced from a double star arebrought into a well-measurable configuration by rotatingand shifting the prism.

PhotographyDuring much of the 20th century, photography was thebest direct-imaging method in astronomy, but its resolvingpower was limited by adjacency effects from the chemicalprocessing and by blurring during the fairly long exposuretimes needed. Extensive double-star data were obtainedfor wider pairs, typically over 2 arcsec. The long-termphotographic monitoring of stars has helped in otherways: with PARALLAX results for nearby binaries, their massratios from the orbital shifts of blended images againsttheir centre of mass and sometimes the detection of unseencomponents from periodic shifts in the motions of stars.Such work could now be done with digitizing cameras(CCDs) which have replaced photographic emulsions indirect imaging.

InterferometryOwing to high instrumental demands, the original formof interference was little used. A successful variant is theeyepiece interferometer as was constructed and widelyused by W S Finsen.

In the last 25 yr a combined high resolution in bothangular separation and recording time, now also formodest-size telescopes, has been achieved by ‘speckle’interferometry (see SPECKLE IMAGING OF BINARY STARS). Itresults in superior precision by correlating numerous,very short-exposure images (speckles, tavelures). Mostcurrently used devices consist of microscope optics asobjective and collimator, a CCD camera with imageintensifier as detector and computers for very fastrecording, centering, autocorrelation and processing;colour filters and usually a pair of prisms, to removeatmospheric dispersion in both amount and direction,are added. The full theoretical resolving power atlarge reflectors, around 0.03 arcsec, can be reached. Animportant development toward yet higher resolution isthe application of long-baseline interferometry (long usedin radio astronomy) now at optical wavelengths, asevidenced by results from devices such as the Mark IIIinterferometer.

Space instrumentsThe Hipparcos satellite measured around 1991 numerousdouble stars and added some 3000 new pairs, mostly thoseof larger brightness differences whose detection from theground was impeded.

OrbitsApart from their significance for the understanding ofstellar origin and of evolutionary patterns, binary starssupply direct determinations of the masses of stars asfundamental quantities. This requires knowledge of theirorbits.

Main-sequence stars have a relation between massM and luminosity L, as expected from theory, because

for stars of nearly identical composition the equilibriumconditions governing interior stability make the luminosi-ties and radii dependent only on the mass: L increaseswith the kth power ofM . The exponent k varies along themain sequence, and it also depends on the colours withinwhich the luminosities are defined; for visual magnitudes,k = 3.8 for the middle main sequence, but k decreases toabout 2.6 for both high- and low-mass (B and M) types.

The apparent orbit of a visual binary as seen in the sky(at right angles to the line of sight) is a parallel projection ofthe true orbit ellipse. Seven constants, the so-called orbitalelements, are needed to describe the motion. Three of themare angles specifying the projection: the inclination is theangle between orbit and projection planes, the node is thecelestial direction in which the planes intersect and thelongitude of periastron (the point of closest approach ofthe components, and of fastest orbital motion), countedfrom the node, gives the orientation of the ellipse. Theother elements fix the motion within the orbit: the period(P ) in years, the eccentricity, the time of passage throughperiastron, and the semimajor axis (a) in arcsec. Theposition in the orbit at any time is then given by formulaewhich represent Kepler’s first and second laws (see KEPLER’S

LAWS). There are several computer routines on how tosolve for the seven elements as best fitting a given set ofobservations.

Visual binaries with known orbits (some very precise,others tentative and subject to future revision) nownumber about 1000. The orbit sizes (a) are mostly under1 arcsec; the periods range from a few years to somecenturies (in cases with well-observed periastron passagesup to about 1000 yr), and almost all objects are within150 pc from Earth. Contrary to the planetary system andto close binaries, orbits of medium or high eccentricitiesdominate among visual pairs.

To determine the total mass (M1 +M2), Kepler’s thirdlaw is used in the form a3 = P 2(M1 + M2) only if a isdetermined in astronomical units:

a (length in AU ) = a′′ (angular size in arcsec)

× d (distance in pc).

Inaccuracies of distances d cause often the largestuncertainty in the masses. This can be remedied in otherways.

(a) By spectroscopic observations of the orbital motion.Doppler-shift amplitudes alone yield only theone-dimensional radial motion, leaving the orbitinclination and hence the true orbit size and the massundetermined, but the combination with the visualelements supplies the missing link. The speckletechnique in particular has resolved many short-period pairs discovered by spectroscopists.

(b) When the components can be assumed to bemain-sequence stars, a mass–luminosity relationcan be invoked, along with the known apparentmagnitudes, to connect the masses and the distance.



Cases where this assumption does not hold arerevealed by a disagreement between spectral typeand the computed mass. Many stars spectrally stillclassified as main sequence are found to be alreadyevolved in luminosity to above ‘zero age’ level.

Multiple systemsAbout a third of binary stars contain at least one additionalcomponent. However, the separations are almost alwaysquite different: the ratio of wider versus closer pair is atleast 6:1 and often much more. Long-term stability ofmultiple systems with stellar (not very unequal) massesdemands this ‘hierarchical’ order. So the periods are verydifferent, and there are few visual systems in which bothorbits are reliably known.

More common are close spectroscopic subsystemswithin visual pairs, and there are also cases of definite,periodic deviations from the elliptical motion indicatingthe presence of an unresolved component. These objectsrequire elaborate analysis since two orbits (14 elementsinstead of seven) have to be computed simultaneously.Slow changes in the orbits owing to perturbations betweencomponents are known in two of them: ξ Ursae Majoris(11182 + 3132) and ζ Cancri (08122 + 1739).

Some noteworthy objectsWithin 3 parsec from the Sun, four apparently single dwarfstars and the following three interesting binaries are found.

α Centauri (14396 − 6050), the nearest system,consists of two solar-type stars (masses 1.1 and 0.9) inan orbit of period 79.9 yr, now slowly closing in aftermaximum separation was reached in 1980, and a red-dwarf companion, a full 13 000 AU distant from theprimary pair, but even closer to Earth, hence calledProxima Centauri.

Sirius (06451 − 1643) has a companion with a periodof 50.1 yr, which was predicted from periodic changes inthe proper motion of the bright star. Resolved in 1862, thecompanion was found to be almost 10 magnitudes fainter,and the identical white color indicated an enormousdisparity of the radii of the stars. It was the firstidentification of a white dwarf star. (40Eridani B =04153− 0739 had been observed long before but its 252 yrorbit with a faint red-dwarf companion and its white-dwarf nature were not known until later.)

LDS 838 (01390−1757), a 12th magnitude object witha period of 26.5 yr is one of the lowest-luminosity pairs ofred dwarfs. Each component has only 0.12 solar masses;one of them is the very active flare star UV Ceti.

A few other remarkable binaries and multiples maybe added.

Castor (α Geminorum, 07346 + 3153) is a sixtuple at adistance of 14 pc. The bright pair has been observed foralmost three centuries, and the latest orbit gives it a periodof 467 yr. Both components are close spectroscopic binarieswith low-mass companions (periods of 9 and 3 days), andthe distant 9th magnitude star YY Gem, which belongs to

the system, is a unique case of a red-dwarf eclipsing binary(type dM) with a period of 19.5 h.

Capella (α Aurigae, 05167 + 4600), a long-knownspectroscopic binary with P = 104 days, and at adistance of 13 pc, was resolved and well observed byinterferometers and speckle cameras. It consists of two Gtype giants of nearly equal masses (2.5 solar masses each)and luminosities, evolving synchronously. Also part of thesystem is a later-discovered, distant red-dwarf pair whichmay have a period around 400 yr.

The closest pair resolved by a large interferometer isSpica (α Virginis, 13252 − 1110) at a separation of only0′′.0015, with a period of 4 days.

ε Hydrae (08468 + 0625), at a distance of 36 pc, isof interest because of the diversity of its five knowncomponents: a G type and a fainter F type subgiant ina 15 yr orbit, an F type main-sequence star (period about1100 yr around the primary pair) which is a spectroscopicbinary with a low-mass companion (P = 10 days), anda still more distant, subluminous dK dwarf star whoseperiod may be around 10 000 yr.

Finally, GC 20393/4 (15102 − 1623) is a faint, widepair of very old subdwarf stars, separated by 5 arcmin,and 30 pc distant. They race through the solar vicinity ata record speed of nearly 600 km s−1 relative to the Sun,or practically at escape velocity within the Milky Way,suggesting an origin in the outer galactic halo. It shouldbe noted that, although separated by at least 10 000 AU,they have kept together on the long trip for twice the ageof the solar system.

W D Heintz


VLBI (Very Long Baseline Interferometry) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

VLBI (Very Long BaselineInterferometry)The Michelson stellar INTERFEROMETER was developed byMICHELSON as an optical instrument to measure the angulardiameter of stars. In recent times, the method hasbeen extensively used at radio wavelengths, to achieveextremely high angular resolution, greater than thatachieved by single-aperture telescopes at any wavelength.The principle is as follows: two RADIO TELESCOPES, receivinga signal from a distant radio source, amplify the signalsand send them via transmission lines to a radio receiver.the radio signals arrive at different times at the tworadio telescopes, so the shorter path length is lengthenedby a delay line that equalizes the time at which thesignals are combined by the receiver. If the signals arein phase, they will reinforce one another and the signalamplitude will be twice that of a single telescope (i.e.four times the power). If the signals are 180 out ofphase, there will be no signal, since they destructivelyinterfere. If the telescopes are rotating with the Earth, andthe time delay is held fixed, the alternating constructiveand destructive interference will give a sinusoidal outputwhose amplitude and phase can be measured. (The opticalMichelson stellar interferometer is functionally identical;the radio telescopes are replaced by plane mirrors that sendplane waves into a telescope, where they are combinedat the focus, where interference fringes are detected.) Byextension from the optical case, the sinusoidal radio signalis also called a fringe, even though it is seldom seen directlyin modern interferometers. This is because the receiverhas, in effect, two channels that cross-correlate the twosignals, directly and in quadrature. From the amplitudesof these two outputs, the fringe amplitude and phase arederived.

The great power of the Michelson interferometer liesin its extension to APERTURE SYNTHESIS, in which the signalsfrom an assemblage of many telescopes are combinedpairwise. As in so many areas of modern science,the Fourier transform technique is used to derive animage. Given a radio source with a brightness distributionB(α, δ) on the sky, where α and δ are right ascensionand declination, one constructs an approximate Cartesiancoordinate system x, y. The distribution B(x, y) hasa Fourier dual b(u, v), and it can be shown that theamplitude and phase of the radio fringe from a Michelsoninterferometer is proportional to the Fourier transform ofthe source brightness distribution. The dual coordinates(u, v) are determined by the baseline bλ, measured interms of the observing wavelength λ, projected on the skyperpendicular to the line of sight to the radio source. Thisplane is usually taken withu in the east–west direction, andwith v north–south, and is commonly referred to as the uv-plane. Asufficiently large sample of fringe amplitudes andphases, with interferometers sampling many points on theuv-plane, can give a good approximation to the Fouriertransform b(u, v) of the brightness distribution, and by

numerically calculating the Fourier inversion, the actualbrightness distribution B(x, y) can be constructed.

It follows that an array ofN radio telescopes, suitablydistributed on the Earth, will yield N(N − 1)/2 pairs, thusallowing one to map radio sources with an instrumenthaving an effective diameter, and a corresponding angularresolution, of a telescope of diameterDmax, the length of themaximum interferometer spacing. Such an instrument iscalled an aperture-synthesis array; the density of Fouriertransform samples can be increased by observing as theEarth rotates, thus changing the projected baseline lengthon the uv-plane. In this case, the technique is knownas Earth-rotation aperture synthesis. The VERY LARGE

ARRAY (commonly known as the VLA), operated by theNational Radio Astronomy Observatory on the Plains ofSt Augustine, near Socorro, New Mexico, with 27 antennasin a variable-length, Y-shaped array, is a good example ofan aperture-synthesis instrument.

Techniques of VLBIThe technique of Very Long Baseline Interferometry, orVLBI, is an extension of Michelson interferometry, madepossible by the exquisite control that modern electronicsexerts over the time domain. If the transmission linesfrom the telescopes to the receiver are removed, it is stillpossible to observe the interference fringes by having ahighly stable clock (usually a hydrogen maser frequencystandard) at each telescope. The amplified signals arerecorded on wide-band magnetic tape recorders (typicallyat a data rate of the order of 108 bits per second or more),using the hydrogen standards for time control. The tapesare then shipped to a central processing facility, wherethe received signal is now a string of discrete samplese(ti) at known times ti . The two data streams are thencross-correlated numerically, giving the fringe amplitudeand phase. (For a general treatment of interferometry,including VLBI, see Burke and Graham-Smith (1997).)

Several VLBI arrays are in operation: the VLB Array(VLBA) of the National Radio Astronomy Observatorywith its processor in Socorro NM, the EuropeanVLBI Network (EVN), with a processor in Dwingeloo,Netherlands (see JOINT INSTITUTE FOR VLBI IN EUROPE), andthe Southern Hemisphere Array, with its processor inNarrabri, NSW, Australia. There is also a worldwidenetwork, a collaboration among the above VLBI arrayswith the addition of other national radio telescopes.

The VLBA can be taken as an example of a dedicatedVLBI system. It consists of ten similar radio telescopesspread across the USA and its outlying territories; thearray is bounded by telescopes in St Croix, US VirginIslands, on the southeast, Hancock NH on the northeast,Brewster WA on the northwest, and Mauna Kea HA onthe southwest, with a maximum baseline of 10 000 km.At the present time its shortest operating wavelength is7 mm, giving a maximum angular resolution of about150 microarcseconds. The operating wavelength canbe switched rapidly from one band to another, and theobserving schedule, sent out to the individual telescopes



some time in advance, includes the time and duration ofobservation the source to be observed, and the frequencyband.

There are some practical considerations that arisewhen making VLBI observations. The masers atthe individual telescopes will have slightly differentfrequencies, and therefore the time base on each magnetictape will have systematic errors, showing up as timedifferences and differences in fringe rotation. The VLBAcorrelator, therefore, performs the cross-correlation fora number of different time delays and fringe rotationrates, in order to correct for these differences. In thecase of a ‘hard-wired’ interferometer such as the VLA,timing variations occur, and these are usually correctedby observing a nearby calibration source and derivingthe necessary corrections. For VLBI observations, this issometimes possible, but in many cases it is not feasible todo so. For an array of several telescopes, however, thereare closure relations that allow extra information to bederived. If three telescopes are observing the same source,there is a sum condition on the three fringe phases thatgives useful phase information, and there is a conditionon the fringe amplitudes, when observing with fourtelescopes, that gives useful constraints on the individualamplitudes. When an array of ten telescopes is used, as inthe case of the VLBA, there is enough redundancy in thevarious combinations to constrain the uncertainties in theFourier transform, giving maps of remarkable fidelity.

Science applications of VLBIThere is a surprising variety of scientific problems thatcan be addressed by the methods of VLBI. Quasarsand the nuclei of active galaxies (BL Lac sources, radiogalaxies and Seyfert galaxies) all have structures in themilliarcsecond range or smaller. Pulsars and gamma-rayburst sources are even smaller, requiring baseline lengthsfar longer than anything currently feasible, Molecularmaser sources are also of milliarcsecond size, and theseoccur in association with star-forming regions and cool,evolved stars such as the Mira variables. Three generalclasses of observation can be identified: the mappingof milliarcsecond structures, the measurement of relativesource positions to achieve higher astrometric accuracy,and the use of known sources as references to studygeophysical phenomena such as continental drift andpolar wandering.

Active galactic nucleiThe detailed study of ACTIVE GALACTIC NUCLEI (AGNs)has shown that the jet-like structures seen on largescales persist at the smallest angular scales that havebeen studied. Paradoxically, the (relatively) long radiowavelengths are closely related to phenomena at x-raywavelengths. The radio phenomena observed by VLBIare in large measure relativistic, since the radio wavesare generated by synchrotron radiation from relativisticelectrons circulating in magnetic fields. A large fractionof the sources involve massive black holes, surrounded

by accretion disks (the hot gas in these accretion disks,observed at ultraviolet and x-ray wavelengths, is in theprocess of spiralling down into the black hole at thecenter). The exact process is still unclear, but seems toinvolve magnetic fields in the accretion disks, twistedby differential rotation in the disks, ultimately releasingenormous amounts of energy by accelerating electrons torelativistic energies and confining the particles to a narrowjet. Shortly after VLBI was developed, it became clearthat rapid changes were developing in the structure ofthe jets, and it is now well-established that the jets exhibitapparent motions that significantly exceed the speed oflight, sometimes by an order of magnitude. Althoughthe structure of fundamental physics might be questioned,such a dramatic revision as superluminal velocity does notseem to be needed. The effect is almost certainly causedby the aberration effects observed when a relativistic beamis directed at the observer. The effect requires a beam ofparticles having bulk motion close to the speed of light,directed within a few degrees of the observer.

A relativistic effect that must be considered inconjunction with these compact, relativistic sources ofSYNCHROTRON RADIATION is the self-Comptonization of theradiation. As the sources increase in energy density, therecan be so many radio photons generated that they areCompton-scattered by the electrons and become x-rayphotons, which can be observed by X-RAY TELESCOPES. Thishas the effect of limiting the possible surface brightnessat radio wavelengths; expressed in terms of ‘brightnesstemperature’ (the temperature a source would need tohave at the observing band to give the observed specificintensity, not a physical temperature necessarily). As aconsequence, it is not possible for a synchrotron sourceto exhibit a brightness temperature in excess of 1012 K.Unfortunately, it is not possible to verify this predictionby interferometer baselines on Earth. The flux of aRADIO SOURCE is proportional to the product of its averagesurface brightness and its angular size, and the wavelengthdependences cancel out. At any wavelength, therefore, aninterferometer with a baseline length of an Earth diameter,observing even the strongest radio source, can only set alimit of 1012 K on the surface brightness of the compactstructures.

This limitation to our knowledge can be overcome byusing interferometer baselines that exceed the diameter ofthe Earth. By placing a VLBI telescope in space, there is inprinciple no limit on the angular resolution that might beachieved. It has turned out that the technical difficultiescan be overcome; a successful demonstration was carriedout, using the TDRSS (a data relay satellite system), thatshowed the practical feasibility of such measurements.This was followed up by HALCA, a dedicated VLBI satellitelaunched by the Japanese Space Institute ISAS in February1997. The results are clear: a large fraction of quasarsand AGNs have radio nuclei that exceed the self-Comptonlimit.



Maser sourcesThe extragalactic maser sources have been another majorobject of VLBI research. The lambda-doublet lines ofthe hydroxyl radical at 1612, 1665, 1667 and 1720 MHzall exhibit high surface brightness (1010 K or more)that can only be a consequence of gas clouds beingpumped by radiation or by collisions to invert the levelpopulations, resulting in a one-pass maser that can amplifythe background radiation. The same phenomenon occursin the rotational lines of water, silicon oxide and methanolthat are accessible in the radio spectrum. Star-formingregions frequently have many such maser sources, as do awide variety of evolved stars. In such cases, the difficultyin determining the fringe phase can be overcome by self-referencing: a particular compact line feature can serve asa reference within the maser complex, and the positionsof other line sources, often separated in Doppler shiftby several km s−1, can be measured with respect to thereference maser.

MegamasersExtragalactic maser sources have also been discovered.These are referred to as ‘megamasers’ since the necessaryenergies are so high. One of the best studied of these, themegamaser in the active galaxy NGC 4258, appears to beemitting 300 solar luminosities in the 22 GHz water line.The line structure turned out to be extraordinary: therewas a complex of strong central lines, having very littlevelocity relative to that of the galaxy, but in addition therewere satellite maser lines with relative Doppler velocities±900 km s−1 to either side of the central complex. Adetailed analysis of the VLBI-derived map showed thatthe velocities could be accurately explained by a simplemodel. A ring of gas surrounds the (presumed) black holeat the center, and we see the ring nearly edge-on. Thecentral line complex changes linearly in Doppler velocityacross the source, and is amplifying the radiation from thecentral source; this allows the acceleration of the moleculargas to be measured. The high-velocity lines come frommasers at the terminus of the edge-on ring, and these allowthe Keplerian velocity to be determined. The result is firmevidence of a central black hole having a mass of 70 millionsolar masses (see Miyoshi et al 1975, Herrnstein et al 1999).

BibliographyBurke and Graham-Smith 1997 An Introduction to Radio

Astronomy (Cambridge: Cambridge UniversityPress)

Herrnstein et al 1999 A geometric distance to the galaxyNGC 4258 from orbital motions in a nuclear gas diskNature 400 539

Miyoshi et al 1975 Evidence for a black hole from highrotation velocities in a sub-parsec region of NGC 4258Nature 373 127

Bernard F Burke


Vogel, Hermann Carl (1841–1907) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Vogel, Hermann Carl (1841–1907)German astronomer, became director of the Potsdam As-trophysical Observatory, where he made spectroscopicanalyses of stars, planets, comets and the Sun. He usedphotography to measure Doppler shifts of stars to deter-mine their radial velocities, discovering spectroscopic bi-naries and, with Julius Scheiner, determining stellar diam-eters and masses.


Voids E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

VoidsGalaxies have a strong tendency to cluster. They havebeen drawn together by mutual gravitational attraction.Seventy per cent of galaxies are found in groups or clustersthat are probably gravitationally bound. In turn, thegroups of galaxies lie along filaments that interconnect.The largest clusters of galaxies tend to be found at theintersections of several filaments. Even the 30% of galaxiesthat are not in bound groups lie, almost entirely, within thefilaments.

The filaments occupy only a small fraction of space.Most of space has no observable matter. These emptyregions are the ’voids’.

The discovery of voidsAlthough the general tendency for galaxies to cluster wasknown for a long time, the extreme segregation of galaxiesinto a small fraction of the available volume was notappreciated until recently. With information about thedistribution of objects on the plane of the sky augmentedby only crude information about distances from galaxysizes and magnitudes, there is too much confusion fromprojection overlay to get a good sense of structure. Itis necessary to have the REDSHIFTS of galaxies in order toplace them in distance and, hence, to be able to constructa three-dimensional map of the distribution of galaxies(see DISTRIBUTION OF GALAXIES, CLUSTERS, AND SUPERCLUSTERS).Only in 1978, with galaxy redshifts in hand, did StephenGregory and Laird Thompson comment that there wereregions nearly devoid of galaxies to the foreground ofa supercluster that extends from the Coma cluster tothe cluster Abell 1367. Several other empty regionswere found within the next few years and, in 1981, theinstantly famous Bootes void was reported by RobertKirshner, August Oemler, Paul Schechter and StephenShectman. This void extends over roughly a million cubicmegaparsecs (30 million cubic light-years).

Over the years since, there have been searches tosee whether there is anything to be found in the Bootesvoid. A few small galaxies have turned up. Still, itis an empty place. The Bootes void may not even bethe biggest empty region within the currently accessibleuniverse. It is distinctive in part because it is surroundedby the prominent supercluster domains of Hercules–Corona Borealis, Leo and Ursa Major. There are even largerregions of substantial underdensity. The larger regions arehard to characterize because they extend so far across thesky that they become lost in the obscuration of our MilkyWay Galaxy and they extend so far in distance that redshiftsurveys become incomplete.

Nearby voidsThere is no need to go far to find a void. The Milky WayGalaxy is part of the LOCAL GROUP which in turn is partof the Coma–Sculptor cloud, a filament that lies in theequatorial plane of the Local Supercluster. In the planeof this filament there are galaxies in all directions and to

the south of this plane in supergalactic coordinates there isanther filament. However, to the north of this plane, justbeyond the last galaxies of the Local Group, is the LocalVoid. This empty region is poorly surveyed because itis intersected by the Milky Way. However, there are fewnearby galaxies on either side of the Galactic plane over awide segment of the sky.

There are several voids that are somewhat fartheraway but better studied because they cover a restrictedpart of the sky and stand out with good contrast againstpronounced filaments. The Perseus–Pisces filament isparticularly well defined, at a distance of 60 Mpc (200million light-years). Immediately to the foreground is alarge region devoid of galaxies.

The topology of voidsThe complement of voids are the superclusters and figure 1in SUPERCLUSTERS AND THE LOCAL SUPERCLUSTER illustrates boththe filamentary nature of the distribution of galaxies andthe voids. The filament–void network can be comparedwith a sponge. The material of the sponge can be equatedwith the filaments and the sponge holes with the voids.The sponge material is all interconnected, else the spongewould fall apart, but also the holes are all interconnected,so water can pass through. However, with this analogy,the universe is a very tenuous sponge with frail materialconnections and big holes.

The universe must have begun with only minutedensity variations. Over time, matter has gatheredthrough self-attraction in the high-density regions andevacuated the low-density regions. Hence, even in co-moving coordinates the domain of the voids has beengrowing. In metric coordinates, the voids are much biggernow than in the past. It follows that matter is moving awayfrom voids, toward high density concentrations.

There is still debate about how much residual matterthere is in voids. Galaxy formation may only occurabove a certain density threshold which leaves open thepossibility that there could be a considerable amount oftotal mass tenuously spread through the voids. Stringentlimits are set by failure to detect intergalactic gas overa wide range of temperatures. The only hint that theremight be residual gas comes from studies of absorptionlines in distant quasars. These absorption lines arise fromtenuous intergalactic gas in the foreground of the quasars.However, at low redshifts (the present epoch) there arerelatively few absorbing gas clouds and it remains tobe demonstrated whether that gas is associated with theintersection of filaments with the line of sight or with voids.

BibliographyRood H J 1988 Ann. Rev. Astron. Astrophys. 26 245

R Brent Tully


Volans E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Volans(the Flying Fish; abbrev. Vol, gen. Volantis; area 141 sq.deg.) a southern constellation which lies between Mensaand Carina, and culminates at midnight in mid January.It was first shown on Petrus Plancius’ celestial globe ofc. 1598 as Piscis Volans, though it is usually attributed tothe Dutch navigators Pieter Dirkszoon Keyser (also knownas Petrus Theodorus) and Frederick de Houtman, whocharted that part of the southern sky in 1595–7.

A small, inconspicuous constellation, the brighteststars in Volans are β Volantis, magnitude 3.8, γ Volantis, abinary with yellow (G8) and pale yellow (F2) components,magnitudes 3.8 and 5.7, separation 14.11′′, ζ Volantis,magnitude 3.9, and δ Volantis and α Volantis, bothmagnitude 4.0. Another interesting star is ε Volantis,a double with bluish-white (B6 and B9.5) components,magnitudes 4.4 and 7.5, separation 6.0′′, that have thesame proper motion, the former of which has an unseencompanion that revolves around it in 14.17 days.

There are no bright star clusters, nebulae or galaxiesin the constellation, the brightest being NGC 2442, aneleventh magnitude barred spiral galaxy.


Volcanism in the Solar System ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS


Volcanism in the Solar SystemVolcanism is the main endogenous process (a processoriginating within the planet) which gives PLANETARYSURFACES most of their characteristic features.Volcanism consists of the superficial outflow of liquidmixed with a variable quantity of gas and crystals, whichcomes from the PLANETARY INTERIOR. These liquids areproduced by the melting of the planet’s or satellite’sinterior. They are called ‘lava’ or ‘magma’ and come upto the surface owing to buoyancy forces. Water andhydrocarbon springs on Earth are not part of thesemagmatic liquids.

Volcanism not only modifies the structure and themorphology of the planetary surface. During and justafter the formation of planetary bodies, accretionalheating partially separated volatile from refractoryelements, degassed the mantle and produced primitiveatmosphere. Volcanism causes this degassing of themantle to continue, and it is gradually modifying both theplanetary interior and the atmosphere.

The morphological types of volcanic featuresstrongly depend on the magma viscosity and gas content.Very-low-viscosity magma induces flat lava flows andmay produce lava plains in the case of abundant flooding.An increase in the viscosity produces shield volcanoeswith shallow sloping flanks (inclined at about 10º) forintermediate-viscosity lava or domes with steepperipheral flanks (inclined at about 25º) for high-viscosity lava. A significant gas content fragments themagma during volcanic explosions and producesfragments of rock called ‘tephra’. During low-energyexplosions, the tephra again fall around the outflowpoints, forming scoria or cinder cones. High-energyexplosions cover large areas with ash or pyroclasticdeposits expelled aerially from a vent. Some volcanoesmay produce continually the same magma with the samegas content that produces monogenic volcanicconstruction (built up by one kind of eruption).Elsewhere, the gas content and the magma viscosity maychange during a single eruption or throughout thevolcano’s history, inducing a complex polygenic volcanicconstruction (built up by several kinds of eruption).

Volcanism timing allows the distinction betweenthree kinds of worlds: (1) active worlds, with presentevidence of endogenic activities, (2) recently activeworlds on which active resurfacing processes haveoccurred on a large scale after the end of the heavybombardment, but which now seem dead, and (3) deadworlds, on which the majority of the surface pre-dates theend of heavy bombardment and where the geologicalactivity has occurred only occasionally after the end ofthis bombardment.

Two types of solid planets and satellites exist in thesolar system. There are silicated and rocky bodies in theinner solar system and icy bodies in the outer solar

system. Io and Europa, two Galilean satellites of Jupiter,are silicated bodies, although they belong to the outersolar system. Volcanism exists on almost all of thesebodies.

Volcanism in the inner solar system

Active worldsThe Earth (diameter 12 750 km), the basis forplanetology. Volcanism on Earth has been studiedconsiderably since Pliny the Younger died studying thefamous eruption of Vesuvius in AD 79. Terrestrialvolcanism is the reference for comparative planetaryvolcanism studies, and the definitions of rocks, eruptionsand feature types were proposed on Earth.

The geographical distribution of volcanoes on Earthis one of the characteristics of our planet. About 95% ofactive volcanoes are grouped together and form volcaniclines. The first group is located on the islands ormountains near oceanic trenches around the Pacific andnear Indonesia and the West Indies. This volcanism ismainly andesitic (SiO2 55%), basaltic (SiO2 50%) andrhyolitic (SiO2 60%). The second group lies underoceans, on the top of oceanic ridges. This volcanism isbasaltic (SiO2 50%). The location of these two volcanictypes is strongly correlated with the global dynamics ofthe EARTH’S INTERIOR (plate tectonics). Volcanismmainly occurs near plate boundaries. The other 5% ofvolcanoes are isolated volcanoes, spread over oceanfloors or continents. These volcanoes are basaltic onesand are called ‘hot spots’.

The popular notion of the Earth’s interior is a world-wide layer of magma beneath a solid thin shell.Following this notion, volcanism is simply the arrival ofmagma through a hole or a fracture crossing the crust.This popular notion is totally false. The Earth is dividedinto three levels: the superficial crust 5–40 km thick, themantle from the base of the crust to 2900 km and the corefrom 2900 km to the centre of the Earth. The Earth isentirely solid with the exception of the outer core, whichis made of molten iron. This liquid outer core representsless than 15% of the Earth’s volume, 2900 km below thesurface, which is not directly involved in volcanism. TheEarth’s interior is solid, but hot, with a temperature of upto 5000 ºC. This heat is generated by natural radioactivedecay. The solid state of the Earth despite this hightemperature is due to the very high pressure created bythe weight of overlying rocks.

Temperature differences between the cold superficiallayers and the hot interior produce density differencesand enormous buoyancy, circulatory and convectiveforces. Impelled by these convective forces, the solidmantle flows very slowly like ice in a glacier. Thesemantle motions induce superficial motion of the coldsuperficial layers, called continental drift or platetectonics. Downward-flowing superficial oceanic rocks



introduce oceanic water deep into the mantle. This waterlowers the melting point and produces a small proportionof molten material. Upward-flowing mantle reaches alevel where the drop in pressure allows a smallproportion to melt. If enough is formed, the formedmagma rises away from the denser residue of solidinfusible rocks, owing to buoyancy forces. On Earth theconvective cells are more than 1000 km wide; the upperparts of these cells are the tectonic plates with onlyhorizontal motion. Volcanism occurs at the plates’boundary where upwards or downwards motions existand induce a partial melting of the mantle. Some othercirculatory motions exist inside the mantle, withlocalized upward flows of solid mantle. These localizedupward flows which induce localized drops in thepressure, partial melting and volcanoes, and are far fromthe plates’ boundaries, are called ‘hot spots’.

In 2001, a NASA mission to study Alaska’s uniqueterrain provided scientists with their first detailed look atthe changing topography of one of Earth’s most activevolcanic regions. Researchers at the Alaska SyntheticAperture Radar Facility, Fairbanks, created a high-resolution digital elevation model of Umnak Island,home to the Okmok volcano. This model can be used toproduce new, accurate geologic maps. The most recenttopographic map of the region was made in 1957 fromaerial photographs. Okmok has erupted four times sincethen, dramatically changing the landscape. The Alaskascientists used data gathered in October 2000 by theNASA/JPL Airsar instrument, a side-looking imagingradar system carried aboard a NASA DC-8. It collectedthe Alaska data as part of its PacRim 2000 Mission,which took the instrument to French Polynesia, Americanand Western Samoa, Fiji, New Zealand, Australia, NewGuinea, Indonesia, Malaysia, Cambodia, Philippines,Taiwan, South Korea, Japan, Northern Marianas, Guam,Palau, Hawaii and Alaska. Researchers there combinednumerous Airsar strips of data into a mosaic, fused it toLandsat imagery, checked its accuracy and generated anumber of data products, including the mosaic of UmnakIsland.

Venus (diameter 12 102 km) (figure 1). There is no doubtthat volcanism has been responsible for the generation ofthe majority of the Venusian surface. The Venusiansurface is affected by only few impact craters, indicatinga geologically young age. Venusian plains exhibit longflow-fields, with long lava channels in some places.Many thousands of volcanoes exist on Venus, either onplains or in the highlands (see VENUS: SURFACE). TheMAGELLAN pictures revealed 550 concentrations of smallvolcanoes <20 km in diameter, 270 intermediatevolcanoes between 20 and 100 km in diameter with agreat variety of morphologies, 150 large volcanoes inexcess of 100 km in diameter, 80 caldera-like featuresand more than 100 steep-sided domes sometimes called

‘pancakes’. Some strange volcanic features withoutequivalents on Earth or other planets exist on Venus.There are complex overlaps of bulges, faults, cindercones and lava flows, and they are called ‘coronae’,‘arachnoids’, ‘novae’…. The majority of these volcanicfeatures are consistent with basaltic composition, ahypothesis confirmed by the findings of Soviet landerswhose soil analyses are chemically similar to those ofbasalt (SiO2 50%). The steep-sided domes may representmore silicic magma (SiO2 60%), while lava channels mayrepresent ultramafic lava flows (SiO2 45%). Only a littleevidence was found for extensive pyroclastic deposits.

The volcanic features show a broad globaldistribution, in contrast to the distribution of volcanoesalong plate boundaries on Earth. Some highlands showan excess of volcanic features, while some lowlandsshow only lava floods.

There is no direct evidence in the Magellan data forcurrent volcanic activity. Some atmospheric analyseshave shown sulfur dioxide variations, which may be dueto volcanic activity. However, the dynamics of theatmosphere may also explain these observed variations.Crater counts would indicate that the whole Venusiansurface is younger than 0.8–0.5 Gyr. An equilibriumresurfacing model for explaining the impact craters’density implies a current volcanic lava flux of0.5 km3 yr−1, a value comparable with the terrestrialintraplate volcanism. Thus there is hardly any doubt that,as is the Earth, Venus is a geologically active planetdespite the lack of active eruptive processes identifiedduring the Magellan mission.

Figure 1. A volcano on a rocky body: Sapas Mons on Venus.Sapas Mons is 400 km across and 2 km high. It is a typicalshield volcano. The sides of the volcano are covered withnumerous overlapping lava flows which appear clear in thisimage. This volcano looks like Hawaiain volcanoes or largeMartian volcanoes. No impact craters are visible on this picture,which indicates a very young age for the Venusian volcanism.This three-dimensional perspective view is produced with radardata from Magellan (NASA).



The density of Venus indicates that the Venusianinterior is theoretically more or less similar to the Earth’sinterior. On Venus, mantle convection produced and stillproduces superficial motions and tectonic features, whichare different from plate tectonics. The volcanismdistribution confirms that the global dynamics of Venus’sinterior is quite different from that of the Earth. Thelocalized ‘hot-spot-like’ convection strongly dominatesthe plate-tectonics-like convection. The origin of thisdifference is poorly understood.

Recently active worldsMars (diameter 6794 km). Volcanoes are unquestionablyon Mars (see MARS: SURFACE). Four enormous shieldvolcanoes and many smaller ones rise 15–25 km abovethe surface. These volcanoes are mainly located on a8000 km wide broad topographic bulge which standsabout 10 km high. OLYMPUS MONS, the largest volcano inthe known solar system, rises 26 km above thesurrounding plateau. These volcanoes possess nestedsummit calderas and have numerous lobate lava flows ontheir flanks. Other smaller shield volcanoes exist in otherplaces. Low-relief shields with irregular summit cratersand numerous radial flows on their flanks are called‘paterae’. Some domes (called ‘Tholus’) and cindercones have also been identified across Mars. Volcanicplains with multiple overlapping flow lobes are foundaround the periphery of the shield volcanoes. Thecomplex flows can extend for >1000 km beyond theshields. Large plains covering wide areas in the northernhemisphere (but rare in the southern hemisphere) havecontroversial origin. The preferred origin is a volcanicone, based primarily on morphological similarity to lunarmaria. No or only little evidence was found for extensivepyroclastic deposits.

In 1976 and 1997, respectively, the two VIKINGlanders and MARS PATHFINDER made in situ chemicalanalyses of soil and rocks. These analyses are consistentwith volcanic rocks (mainly basalts, and one andesite-like analysis). Some meteorites (the SNC METEORITES) aresupposed to have a Martian origin. These meteorites aremainly volcanic rocks (basalts and pyroxenites).

Volcanism existed on Mars from the earliest epoch,the lower Noachian, to the latest one, the upperAmazonian. The maximum of visible volcanic materialswas produced during the Hesperian (from 3.5 to −2 Gyr).The younger volcanoes, Olympus and Arsia Mons, areestimated to be near 0.5±0.25 Gyr.

The Martian interior is theoretically more or lesssimilar to the Earth’s interior. On Mars, mantleconvection produced superficial motions and tectonicfeatures which are different from plate tectonics. Thelack of active volcanoes since 500 million years agoindicates that the mantle convection now is either verylow or has completely stopped. In the past, the mantle

convection was active, and looked like hot-spot-likeconvection rather than plate tectonics convection.

Dead worldsThe Moon (diameter 3476 km figure 2). On the MOON,84% of the surface which was called ‘terrae’ by GALILEOin the 17th century consists of old anorthositic crateredcrust. However, about 16% of the surface consists ofyounger flat basaltic plains and flows which fill the oldlow-lying basins and form black patches on the Moon’sface. These plains were called ‘maria’ by Galileo. Thelava flows inside the plains are very long (up to 1000 km)with very low slopes (less than 1º), a factor that indicatesa very low viscosity for the lava, with very low gascontent. Many of these flows present sinuous rilles,which may be interpreted as collapsed lava tubes orerosional features caused by very hot lava flows. Thethickness of lava may reach 10 km under Mare Imbrium.A very few cinder cones, shield volcanoes and volcanicdomes rise up from the lava flows. The majority of mariaare located on the near side.

This volcanism became evident during the waningstage of heavy bombardment which occurred before−3.9 Gyr. The volcanism was important from −3.9 to−3.5 Gyr, then decreased and virtually stopped 3 Gyrago. The larger maria fill very old impact basins, but thesuperficial flows are some hundreds of millions of yearsyounger than the impact itself. The maria are moreabundant on the nearside than on the farside. Apollosamples indicate that lunar lava chemical composition isrelatively similar to that of terrestrial basalts.

Figure 2. A lava plain on a rocky body: Mare Serenitatis (left)on the Moon. White mountains (on the right) are ‘terrae’. Novisible volcanic structures are seen inside this part of MareSerenitatis, which only consists of very flat lava flows.Numerous small impact craters are visible on the mare, becausethe lava flows are 3.5 Gyr old (NASA–Apollo).



On the Moon, volcanoes ceased 3 Gyr ago. Thedynamics of the lunar mantle during these very oldepochs is poorly known. It is only known that thisvolcanism was not accompanied by significant tectonicmotions, contrary to the Earth, Venus and Mars.Volcanism is more abundant on the nearside than on thefarside, because the crust thickness is more significant onthe farside, and the buoyancy forces seldom allowed themagma to reach the surface. There is a geographicalrelation between volcanism and large impact on theMoon, but genetic relationships are not clear because ofthe large chronological interval between large impactsand maria flooding.

Mercury (diameter 4878 km). On MERCURY, maria-likeplains called ‘smooth plains’ are probably volcanic inorigin. Unfortunately, the modest resolution of MARINER10 images makes it difficult to clearly identify volcanicfeatures. These smooth plains are mainly inside andaround CALORIS BASIN, the major impact basin ofMercury. These smooth plains cover large areas ofrelatively ancient ‘intercrater plains’ which are possiblyvolcanic plains flooded during the heavy bombardment.

Crater chronology would indicate that the maximumof smooth plain flooding is geologicallycontemporaneous with the Caloris impact. The end ofvolcanism probably occurred 3.5 Gyr ago.

The dynamics of the Mercurian mantle during thesevery old epochs is poorly known. As on the Moon, it isonly known that these periods of volcanism were notaccompanied by significant tectonic motions, in contrastto the Earth, Venus and Mars. There is a geographicaland a chronological relation between volcanism and largeimpacts on Mercury, suggesting genetic relationshipssuch as a pressure fall in the mantle under the Caloriscavity.

The meteorites’ parent bodies and the asteroid belt.About 8% of the catalogued METEORITES are volcanicrocks, mainly basalt or pyroxenite. There are five mainkinds of these volcanic meteorites and they are calledeucrite, howardite, aubrite, diogenite and urellite. Themeteorites are suposed to be Earth-crossing fragments ofasteroids. Thus, some ASTEROIDS with past volcanicactivity exist in the asteroid belt. Some asteroids have aspectral signature of volcanic rocks. VESTA is supposed tobe the diogenite and eucrite parent body, and Nysa issupposed to be the aubrite parent body.Radiochronological dating of volcanic meteorites indicatesthat the volcanism in the asteroid belt occurred only 10–100 million years after the formation of the solar system.

The relationship between the duration of volcanismand planetary sizeAfter the end of ACCRETION, the heat inside a planet ismainly produced by radioactive decay. This heat

production by massive bodies is supposed to be the samefor all the silicated bodies and identical to the chondriticheat production value (about 10−11 W kg−1). This heatproduction inside the planets’ interior balances the heatlost to space. This balance fixes the interior temperature.If the loss is efficient with regard to the production, thetemperature is low; if the loss is not efficient, thetemperature is high. The heat is produced inside theplanet and is related to its volume; the heat is lost by thesurface. This balance is thus regulated by thevolume/surface ratio, which is proportional to theplanetary radius: a small planetary body with a lowvolume/surface ratio has efficient heat loss and a lowinternal temperature ; this is the case now for the Moon,Mercury and the asteroids. A large planetary body with ahigh volume/surface ratio has inefficient heat lost and ahigh internal temperature ; this is now the case for Venusand Earth. An intermediate planetary body has anintermediate volume/surface ratio and has anintermediate internal temperature; this is the case forMars today.

At the beginning of their history, all the planetarybodies were hot because of the accretional heating and theshort-lived radioactivity. After the end of this primevalheat production, the heat production was mainly due tolong-lived radioactive elements (K, U, Th). Whenconsidering the solution to the spherically symmetric heatequation, the accumulation of heat from long-livedradionuclides produces a maximum in temperature afterseveral hundred million years in large bodies. This hasbeen confirmed by the extensive magmatic phase on theMoon which started 500 million years after its formation.After this maximum in temperature, because of the naturaldecay of radioactivity, there has been a secular cooling forall the planetary bodies. This cooling has induced agradual stoppage of volcanism, with end times related tosize and volume/surface ratio of the planets: this stoppageoccurred very early for the small bodies (the meteorites’parent bodies), 1–1.5 Gyr after the accretion of Mercuryand the Moon, 4 Gyr after the accretion of Mars and hasnot yet occurred for Venus and the Earth.

Volcanism in the outer solar systemIn the outer solar system, where the temperature is 100K, one planet (Pluto) and almost all large satellites have alow density (≤2 g cm−3) and water ice spectralreflectance. This indicates that water ice is a majorcomponent of these bodies, with a proportion larger than50%. Other volatile compounds (N2, NH3…) may bemixed with water ice. These bodies are called icy bodies.Only two satellites of Jupiter, Io and Europa, have adensity >3 g cm−3 and are mainly rocky bodies.

Io, Europa and tidal heatingIo (diameter 3630 km, figure 3) is one of the fourGalilean satellites of Jupiter. Its density of 3.55 g cm−3



indicates a global composition of silicates and iron. Io’ssurface displays a wide variety of colors and spectrawhich are dominated by sulfur compounds. Io’s surfacegeology is dominated by volcanic vents, craters, fissures,flows and other morphological forms attributed tovolcanic processes (see also IO: VOLCANISM ANDGEOPHYSICS). Active plumes of volcanic materialreaching heights of 300 km were seen by Voyager andGALILEO MISSION spacecraft in 1979 and 1997. Theseplumes are thought to be driven up by SO2 or sulfurvapors. Sulfur compounds are certainly present on thesurface, but, because of the low viscosity of solid sulfur,the heights of mountains, 9000 m, indicate a dominantlysilicated composition for high mountains. Temperatures>1500 ºC measured by Galileo above active volcanoesprove the existence of molten silicates. It is not knownwhether low topography lava flows are composed of puresulfur or sulfur rich silicates.

New large volcanic deposits and flows were seen bythe Galileo spacecraft 18 yr after the Voyager fly-bys. Noimpact craters can be seen on Io’s pictures, anobservation which indicates that the volcanism is a veryactive resurfacing process. The global average height ofIo’s resurfacing is about 100 m (106 yr)−1.

Io’s volcanism supplies a very thin atmosphere(10−2–10−7 Pa). Around Jupiter, a plasma torus is a resultof material escaping from Io (see also IO: PLASMATORUS).

Io is a small planetary body, with theoretically notenough radiogenic heating to be active now. However,Voyager revealed that Io is the most active solid world inthe solar system. Another heat source other thanradiogenic heating must be present for Io. The orbitof Io has a measurable forced eccentricity, which is

Figure 3. Global view of Io, showing numerous lava flows, andtwo active plumes: one on the limb (right) and the other on theterminator (left). These two plumes are about 200 km high.Detailed views of these plumes are shown on the right. Note theshadow of the plume located near the terminator (NASA–Galileo).

particularly pronounced because of the phenomenon oforbital resonance with Europa and Ganymede. The resultof this forced eccentricity is that Io’s tidal bulges arecontinually flexed and moved with respect to Io’sinterior. The tidal bulge is about 7 km in height, and itsperiodic flexion is ±100 m. This is a significant source offriction and of tidally generated heat. In the case of Io,heat generated by tidal dissipation (about 1014 W,corresponding to 10−9 W kg−1) is probably 100 times therate of radiogenic heating. This is enough to melt at leastsome parts of Io’s interior.

EUROPA (diameter 3138 km) is a satellite of Jupiter,3.04 g cm−3 in density. This density indicates a silicatedcomposition. It may be briefly described in this sectioneven though its visible surface is completely covered bywater ice. Density calculation and modelling indicate asilicated sphere of radius 1400–1500 km, surrounded bya 100–150 km thick H2O layer. Voyager and Galileoimages revealed geologically active or recent resurfacingprocesses bringing liquid water or ice-crystal mush to thesurface. The 100–150 km thick H2O layer is probably notcompletely frozen, and a liquid water ocean is probablyoverlain by an ice field. The flooding of liquid water onthe surface does not correspond to volcanism sensustricto, but rather to polar-like phenomena. The H2Olayer is probably warmed at the H2O–silicate boundaryby volcanism, similar in origin to Io’s volcanism, butwith a lower intensity.

The icy bodiesThe surfaces of 13 of these icy bodies larger than 450 kmin diameter have been well imaged by Voyager andGalileo spacecraft. On these kinds of bodies, volcanismmay bring to the surface the molten compound of theinterior: liquid water. The morphology created by thiswater volcanism is very different from that of theirsilicated counterparts. Fractures, vents and dykes mayexist, but volcanic morphologies similar to the popularnotion of volcanoes do not exist. On many satellites(Ganymede, Enceladus, Miranda, Triton), the surface ofresurfaced ice is affected by sets of parallel orcrosscutting ridges and grooves, poorly understood inorigin, despite their frequency in the solar system.

Active worlds: Enceladus (Saturn’s satellite) and Triton(Neptune’s satellite). ENCELADUS (diameter 500 km)displays two different terrain units: old heavily crateredterrain units, cut and overlain with smoother units closestin morphology to bright grooved or smooth terrains ofGANYMEDE. There are essentially two types of grooveson Enceladus: Ganymede-like sinuous grooves, whichmay have the same origin (extruded and ridged low-viscosity fluid, probably liquid water or a mush of water-ice crystals), and straight grooves generally interpreted astensional or volcanic fissures. The grooved and smoothterrains of Enceladus have no detectable impact craters,



which implies a very young age, of less than a fewhundred million years. The albedo of Enceladus isuniform and virtually equal to 1. Analysis of the way thesurface scatters light suggests that it is covered by freshice frost. A likely cause of this is a deposit of frozenspray from young volcanic eruptions perhaps continuingon Enceladus up to the present day. Another argument insupport of the present dispersion of spay from Enceladusis that Enceladus orbits in the middle of the E ring, whichconsists of spherical particles about 1 µm in size. Suchsmall particles would not survive more than ten thousandyears in the interplanetary medium, suggesting that activevolcanism from Enceladus is at the origin of the E ring.

TRITON (diameter 2720 km) is the largest satellite ofNeptune. The entire visible surface is crater poor andrelatively young. The older terrains (the hummocky andthe cantaloupe terrains) show volcanic evidence: flow-generated terrains containing a variety of domes andridges which appear to be viscous extrusions which haveerupted along the fault lines. These older terrains areoverlain with smooth plains, which in some places formfull flat-floored depressions 200 km wide. These youngerterrains seem to have been formed as a series of large-volume, viscous flows. In the center of the flat-flooreddepressions is a collection of pits and flows, which maybe the site of the most recent volcanic activity. Thesouthern part of Triton is covered by a polar cap probablymade of frozen nitrogen. In 1989, two active plumeswere seen by Voyager inside the polar cap. Each plumesends a jet of dark material upwards to an altitude of8 km and produces dark deposits. Triton is the only icybody in the solar system where eruption processes havebeen seen without doubt in action.

Recently active worlds: Ganymede (Jupiter’s satellite),Dione, Tethys (Saturn’s satellites), Ariel, Miranda andTitania (Uranus’s satellites). These are worlds wherethere are evidences of both tectonic proceses on a globalscale and vast volcanic outpourings which occurred afterthe end of late heavy bombardment.

Ganymede (diameter 5262 km, figure 4) representsthe best known of these bodies. Ganymede is divided intodark heavily cratered terrains that occur as polygonalregions and poorly cratered bright terrains in the form ofbands that separate the heavily cratered polygons. Thebright terrains are usually strongly grooved and ridgedand provide unequivocal evidence of widespread-to-global volcanism and tectonism. In some place, brightterrains are smooth, notably where they overlay the darkterrains. This suggest that the bright terrains wereemplaced during extensional tectonic events, such as anextruded low-viscosity fluid, probably liquid water or amush of water-ice crystals. The age of the end of thisflooding is strongly dependent on cratering models in theouter solar system, and may be estimated to be 3–3.5 Gyr.

Figure 4. (a) Global view of the south pole of Ganymede. Flatterrains probably consist of large frozen water flows. Ridgedand grooved terrains are deformed ice flows. Numerous impactcraters indicate that this flooding stopped 3.5 Gyr ago (NASA–Voyager). (b) Detailed view of Ganymedean surface. Flat andridged terrains are crosscut by an elongated caldera-like feature.Viscous flow (probably ice–liquid mush) is visible inside thecaldera (NASA–Galileo).

Large-scale resurfacing and flooding withouttectonics is visible on DIONE (diameter 1120 km) and onone area of TETHYS (diameter 1060 km), while evidenceof water volcanism and resurfacing related to tectonismexists on other parts of Tethys, and on TITANIA (diameter1580 km) and ARIEL (diameter 1158 km). On Ariel,canyons called ‘Chasma’ have clearly been flooded bysome kind of volcanic flow processes which somewhereproduced sinuous rilles with scales and morphologycomparable with their lunar maria counterparts.

MIRANDA (diameter 472 km) is one of the mostremarkable and complex worlds in the solar systemdespite its small diameter. Half of Miranda is a uniformdensely cratered plain, while the other half is occupied by



regions of younger grooved terrains, the detailedappearance of which has never been found elsewhere inthe solar system. These regions are called ‘corona’ andexhibit belts of parallel or crosscutting ridges. Theseridges are commonly interpreted as fracture-controlledextrusions of warm ice or mush. The pattern of ridges isnot understood at all. Somewhere, the ridges seemsubmerged by lava-flow-like structures.

Dead worlds: Callisto (Jupiter’s satellite), Iapetus,Mimas, Rhea (Saturn’s satellites), Oberon, Umbriel(Uranus’s satellites). The Voyager and Galileo fly-bysrevealed six satellites as intensely cratered worlds. Ifsome kind of volcanic resurfacing processes had existed,they ceased before the end of late heavy bombardment.Possible volcanic morphologies have now almostcompletely disappeared, with the possible exception ofCallisto and Rhea.

On CALLISTO (diameter 4800 km), some flat plainsare associated with fractures surrounding a large impact(Valhalla) and may be due to some water resurfacing.

On RHEA (diameter 1530 km), one-third of theknown surface is less cratered than the other two-thirds.Possible clues that there was endogenic activity on Rheaare given by features such as linear troughs and ridges,which may be interpreted as subdued volcanic vents.

Origins of volcanism on the icy satellitesIcy bodies are not large bodies (smaller than or equal insize to Mercury), and their volcanism should becompletely extinct. Moreover, the size does not display asignificant effect on internal activity: the small Enceladusis still active, while the large Callisto never hassignificant activity. Two physical reasons may produceabnormally active volcanism.

(1) With the exception of the unimaged Pluto, all the icybodies are satellites of large planets, and permanentor transitional orbital resonances with other satellitesmay produce permanent or transitional orbitaleccentricity and tidally generated heat. That wouldexplain the difference between Ganymede andCallisto: Ganymede is closer to Jupiter than Callisto;tidal heating may have been greater for Ganymedeand sufficient to produce volcanism during the firstbillion years. Tidal heating was never sufficient forCallisto. The heat sources of Enceladus and otherbodies smaller than 1000 km in diameter are moreuncertain and may be searched for in temporaryorbital resonances with other satellites. The natureand shape of convective motions inside icy mantlesare unknown at present.

(2) If a temperature of 1100 ºC is necessary to meltsilicates, a temperature of about 0 ºC is sufficient tomelt pure water ice, and a temperature of −100 ºCinduces the melting of a water–ammoniac ice

mixture. Even low internal heating could increasethe temperature sufficiently to reach the meltingtemperature and to produce water or water–ammoniac volcanism.

The case of Triton is more complex. Volcanism-induced resurfacing which occurred everywhere isprobably due to radiogenic and tidal heating, as forGanymede or other large icy satellites. Such mechanismscould explain the active plumes. However, these activeplumes are only located on the southern polar cap, and anexogenous origin is possible. The polar cap is made oftransparent nitrogen ice covering a darker substrate. Thiswas Tritonian ‘springtime’ when Voyager 2 imaged thesouth pole of Triton in 1989, and the temperature wasincreasing. Nitrogen ice is transparent and allows most ofthe sunlight to heat the underlying darker layer. A rise ofonly 10 ºC of this dark layer above the surfacetemperature of −236 ºC would cause the vaporization ofthe base of the nitrogen ice layer and its eruption as anitrogen gas plume, through the icy superficial crust.

Pierre Thomas

Voyager Mission ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS

If one were to ask a space scientist which of all the spacemissions ever flown he or she would consider the mostscientifically successful, there is a good chance thatNASA’s Voyager Mission to Jupiter, Saturn, Uranus andNeptune would be mentioned more than all others com-bined. Launched from Cape Canaveral, Florida, in 1977(Voyager 2 on 20 August and Voyager 1 on 5 September),these two hardy spacefarers celebrated their 20thanniversary in late 1997. Although their last planetaryencounter was in 1989, they continue to collect usefuldata about the outer solar system and will probably cel-ebrate their 40th anniversary in the year 2017 before theiruseful lives come to an end.

As early as 1966, engineers at NASA’s JET PROPUL-SION LABORATORY (JPL) realized that it was possible witha single spacecraft to visit all of the giant gaseous plan-ets in the solar system (JUPITER, SATURN, URANUS and NEP-TUNE). After rejecting a number of more costly and moreambitious missions proposed by JPL, NASA approved atwo-spacecraft Mariner Jupiter/Saturn Mission to takepartial advantage of this fortuitous planetary align-ment. The initial approval was for a 4 yr mission,including flybys of Jupiter and Saturn only by each ofthe two Voyager spacecraft. The official project startdate was 1 July 1972, and launch was slated for the latesummer of 1977. The mission was renamed Voyagershortly after project start. (The name Voyager had pre-viously been selected by JPL for an unsuccessfully pro-posed mission to Mars. The basic concepts of that earlier Voyager mission were essentially reproduced in

the successful dual-orbiter, dual-lander VIKING MISSION

to Mars.)The two Voyager spacecraft were built with identical

engineering designs and identical scientific payloads.Part of that design was based on results from the smallerPIONEERS 10 and 11. Neither Pioneer spacecraft detecteddust particles in their passages through the Asteroid Belt.As a consequence, the Voyager dust detector was dis-carded in favor of a plasma wave detector. The eleveninvestigations and their Principal Investigators or TeamLeaders are given in table 1. The two Pioneer spacecraftalso experienced very high levels of radiation in thevicinity of Jupiter; this led to a campaign to build theVoyager engineering and scientific hardware to be resis-tant to damage from these higher radiation fields.

Although the Voyager Mission was approved byNASA for the 4 yr flight including Jupiter and Saturnencounters only, NASA gave approval for JPL to designthe mission of Voyager 2 (but not the spacecraft or its scientific payload) to go beyond Saturn to Uranus andNeptune (or to Pluto). As noted earlier, Voyager 1 waslaunched more than two weeks after Voyager 2.However, the flight to Uranus required a slower flyby ofSaturn; Voyager 2’s flight to Jupiter and Saturn wastherefore a few per cent slower than that of Voyager 1. Itwas prior knowledge of that fact that prompted the‘reverse numbering’ of Voyagers 2 and 1. Voyager 1encountered the planet Jupiter on 5 March, 1979. Voyager2 followed on 9 July of that same year. Voyagers 1 and 2flew by Saturn on 12 November 1980 and 25 August1981, respectively.

Following Voyager 1’s successful flyby of Saturn,which included a very close approach of Saturn’s largest

Voyager Mission

Copyright © Nature Publishing Group 2002 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2002 Dirac House, Temple Back, Bristol, BS21 6BE, UK 1

Table 1. Voyager Investigations and their Lead Scientists.

Investigation and acronym Lead Scientist Affiliation

Cosmic Ray Subsystem (CRS) Rochus E Vogt (1972–84) California Institute of TechnologyEdward C Stonea (1984–) California Institute of Technology

Infrared Imaging Spectrograph Subsystem (IRIS) Rudolph A Hanel (1972–87) NASA Goddard Space Flight CenterBarney J Conrath (1987–91) NASA Goddard Space Flight Center

Imaging Science Subsystem (ISS) Bradford A Smith (1972–91) University of ArizonaLow Energy Charged Particles Subsystem (LECP) S M (Tom) Krimigis (1972–) APL, Johns Hopkins UniversityMagnetometer Subsystem (MAG) Norman F Ness (1972–) NASA Goddard Space Flight Center

and University of DelawarePlasma Science Subsystem (PLS) Herbert S Bridge (1972–87) Massachusetts Inst. of Technology

John W Belcher (1987–97) Massachusetts Inst. of TechnologyJohn D Richardson (1997–) Massachusetts Inst. of Technology

Photopolarimeter Subsystem (PPS) Charles F Lillie (1972–78) University of ColoradoCharles W Hord (1978–9) University of ColoradoArthur L Lane (1979–91) NASA Jet Propulsion Laboratory

Planetary Radio Astronomy Subsystem (PRA) James W Warwick (1972–91) Radiophysics, IncorporatedPlasma Wave Science Subsystem (PWS) Frederick L Scarf (1972–88) TRW Systems

Donald A Gurnett (1988–) University of IowaRadio Science Subsystem (RSS) Von R Eshleman (1972–8) Stanford University

G Leonard Tyler (1978–91) Stanford UniversityUltraviolet Spectrometer Subsystem (UVS) A Lyle Broadfoot (1972–91) University of Southern California and

University of Arizonaa Note that Edward C Stone also served as Project Scientist from project start through the present time.


moon, TITAN, NASA gave approval for the VoyagerProject to target Voyager 2’s Saturn encounter for a con-tinuation on to Uranus. The transition between the pri-mary mission and the Voyager Uranus/InterstellarMission (VUIM) occurred on 1 October 1981. Similarapprovals were given for subsequent portions of theextended Voyager Missions. The VoyagerNeptune/Interstellar Mission (VNIM) started 1 April1986; the Voyager Interstellar Mission (VIM) began 1October 1989. Because of Voyager 1’s close approach toTitan, the spacecraft’s trajectory passed well south ofSaturn. The massive gravity of Saturn pulled Voyager 1’spath northward of the ecliptic and hence northward ofthe orbits of Uranus, Neptune and PLUTO. However, themeasurements Voyager 1 makes of interplanetarycharged particles and magnetic fields continue to be ofgreat interest to solar system scientists. Voyager 2encountered Uranus on 24 January 1986, and Neptune on25 August 1989; like Voyager 1, its mission is now direct-ed at interplanetary and interstellar measurements.

The intensive data collection periods surroundingeach of the planetary encounters generally includedabout 3 months prior to closest approach and one monththereafter. The first approximately two months of theencounter period constituted an ‘Observatory’ phase.‘Far Encounter’ spanned the period from the end of theObservatory phase to about 1 day before closestapproach. ‘Near Encounter’ generally lasted about 2days, more or less centered on the closest approach to therelevant planet. A ‘Post Encounter’ of about 1 month con-cluded each planetary encounter period. Each of thesephases included two or three sequence computer ‘loads’,which updated the central computer and sequencer(CCS) contents. The CCS would then control spacecraftactivity, including recording to and playback of the digi-tal tape recorder storage device. A full tape recorderwould hold the equivalent of about 100 full-resolution(i.e. uncompressed) images.

Because Voyager was designed to operate at greatdistances from the Sun where solar panels were ineffec-tive, power was (and is) provided for each of the Voyagerspacecraft by means of radioisotope thermoelectric gen-erators (RTGs). These ‘batteries’ generate electricity bytransforming heat radiated by the natural decay of plu-tonium dioxide. These batteries, which utilize non-fis-sionable plutonium-238, are highly reliable and longlived, a perfect combination for a mission like Voyager.The RTGs carried by each Voyager generated about 500W of power at launch and are decreasing in output byabout 5 W yr–1. This will permit them to continue topower the two spacecraft with usable amounts ofwattage until about the year 2020.

Communication with the spacecraft occurred viadigitally modulated radio signals. Commands transmit-ted from the ground occurred at S-band, near a frequen-cy of 2113 MHz. Data were transmitted to the ground atX-band, near a frequency of 8415 MHz. Early in its mis-sion, the Voyager 2 primary radio receiver failed.

Following a week of no command receipt by Voyager 2,a computer-stored failure algorithm automaticallyswitched to the spacecraft’s backup receiver. It was soondiscovered that the backup receiver was ‘tone deaf’. Inother words, if the S-band signal arriving at the space-craft differed by more than 96 Hz from the receiver’s frequency, it would not be detected. Factors such as thevelocity of the spacecraft and the rotation and revolutionof the Earth would cause changes (due to Doppler shift-ing of the transmitted radio signal) that easily exceeded96 Hz. While adjustments in the transmitted S-band frequency could be made for the factors mentioned, inde-terminate shifts in the central receiver frequency wouldoccur each time power usage (and hence temperature)would change by more than a couple of watts. Powerchanges were therefore minimized; when power usagechanges were necessary, a command moratorium of from24 to 72 h was honored, providing enough time to deter-mine the new receiver frequency. Voyager engineersbecame very proficient at dealing with this shortcomingof Voyager 2, and it never substantially limited the space-craft’s capabilities.

Another Voyager 2 hardware problem occurred dur-ing the Saturn flyby. Voyagers 1 and 2 are three-axis sta-bilized spacecraft. Except for brief periods of time, the 3.7m wide high-gain communications antenna is keptpointed at Earth, and the star tracker is locked on aselected bright star. The cameras and spectrometers werethen pointed at their targets by means of a steerable two-degree-of-freedom scan platform. One of the two axes ofthat scan platform seized about half an hour afterVoyager 2’s closest approach to Saturn. The problem wastraced to the migration of lubricant out of the gear trainduring periods of heavy usage. Motion was restored afew days later. Careful study of the characteristics of thefailure, both from spacecraft engineering data and fromlife tests on similar gear trains at JPL, enabled engineersto devise a usage plan that would avoid another seizure.The scan platform operated flawlessly during the Uranusand Neptune encounters.

There are several examples of engineering ‘tricks’that improved the Voyager science return. Normal limitcycle motion of the spacecraft was controlled within0.05° in each of the three spacecraft axes. Attitudeknowledge was provided by monitoring the Sun Sensorand Star Tracker. Attitude control was accomplished by200 ms bursts of hydrazine fuel through appropriatethruster jets. Using that scheme, the motion of the cam-era field of view across the sky was slower than themotion of an hour hand on a clock. Nevertheless, longexposures and low light levels demanded even slowerrates. The Attitude and Articulation Control System(AACS) aboard Voyager 2 was programmed to shortenthe thruster pulses, slowing the resultant rates acrossthe 0.05° deadband to less than 10% of the normal qui-escent rate, almost eliminating image smear in morethan 90% of the images obtained at Uranus andNeptune.



Another scheme to reduce image smear involvedrotating the cameras at a rate that matched the apparentmotion of a nearby satellite (moon) during flyby of thatsatellite. This scheme involved using onboard gyro-scopes for attitude reference (instead of the Sun Sensorand Star Tracker). The AACS was fooled by doctoreddata in the spacecraft computer into thinking the gyro-scopes were drifting. As the AACS compensated (as itwas designed to do) for this false ‘gyro drift’, it causedthe spacecraft to turn in a way that kept the cameraspointed at the nearby moon without having to use the(somewhat jerky) movement of the scan platform. Thistechnique was first used to obtain unsmeared images ofSaturn’s moon Rhea and was particularly successful inimage sequences of Uranus’s moon MIRANDA, andNeptune’s moon TRITON.

The scientific findings of the Voyager Mission havebeen the most notable of its contributions. The individualscientific articles based on Voyager data now number inthe thousands, and Voyager data continue to be thesource of detailed comparative studies of the giant plan-ets and their rings, satellites and magnetospheres (themagnetic fields surrounding each of the giant planetsand the electrically charged particle population trappedwithin or traversing those magnetic fields).

Atmospheres and interiorsVoyagers 1 and 2 obtained the first detailed composition,temperature and pressure profiles of the atmospheres ofSaturn, Uranus and Neptune and improved our under-standing of those characteristics of the atmosphere ofJupiter, previously visited by Pioneers 10 and 11. TheGALILEO MISSION has dropped an instrumented probe intoJupiter’s atmosphere and completed its orbital missionaround the solar system’s largest planet, improving onmany of the measurements made by Voyagers 1 and 2.Ulysses also completed a high-latitude flyby of Jupiter onits way to study the polar regions of the Sun (see SOLAR

WIND: ULYSSES). The CASSINI/HUYGENS MISSION is designedto make enormous improvements in our understandingof Saturn’s atmosphere and interior as it circles the plan-et for 4 yr starting 1 July 2004. However, for Uranus andNeptune, the Voyager in situ results and the high-resolu-tion and high-phase-angle remote-sensing observationswill probably remain unique for the foreseeable future.

Our understanding of the interiors of the giant plan-ets is based largely on theoretical considerations.However, there are many bits and pieces of information,mostly from Voyager results, that serve to constrain therange of viable models. Careful measurements of thepaths of the Voyager spacecraft helped determine accu-rate measurements of the masses of the giant planets andof their entire systems. Voyager determined the rotationrates of the interiors of Saturn, Uranus and Neptune.(Jupiter’s interior rotation rate was determined byradiowave measurements made from Earth.) The orien-tations of the strange magnetic fields of Uranus and

Neptune (see below) offer additional clues. Excess heatin the form of infrared energy escaping from each of thefour planets is also important in constraining models ofthe interior.

RingsThe Voyager spacecraft revealed the enormous amountof detail in the rings of Saturn, discovered the rings ofJupiter and provided the first detailed images of therings of Uranus and Neptune. Ring data from Voyagerhave raised even more questions on the nature of ringsystems than they have answered. The answers to thosequestions may be important to studies of the asteroidbelt, the Oort cloud of comets, galactic structure andclusters of galaxies within the universe.

SatellitesVoyager imaged Earth’s MOON. Three new satellites ofJupiter (Metis, Adrastea, and Thebe) were discovered inVoyager data, and Voyager images were obtained ofeight of Jupiter’s 16 known moons. Only the outer eightmoons, all thought to be tiny captured asteroids, weremissed. Voyager revealed active volcanoes on IO, an obvi-ously fractured and geologically young icy surface onEUROPA (first evidence of a global subsurface ocean), avariegated array of surface structures on GANYMEDE andan ancient cratered icy surface on CALLISTO. The sizes,masses, densities and reflectivities of the four Galileansatellites were also measured, showing Ganymede to bethe solar system’s largest satellite, Europa to be thebrightest Galilean satellite and a decrease in bulk densi-ty with increasing distance from the planet (presumablydue to water depletion from the closer satellites due toheat from Jupiter).

At Saturn, four new satellites (Pan, Atlas,Prometheus and Pandora) were discovered in Voyagerdata. These data also showed that Janus and Epimetheus,previously thought to be a single satellite, were actuallytwo satellites with remarkably similar orbits. Three satel-lites (Telesto, Calypso and Helene) were discovered fromEarth as a result of the increased interest in Saturn gen-erated by the approaching Voyager. Voyager revealedthat Titan was the solar system’s second largest satellite,and the only one with a substantial atmosphere. Titan iscovered by a high atmospheric haze, which hides the sur-face from visible imaging. However, infrared andradiowave measurements penetrated to the surface ofthis strange moon, which may be the only body in thesolar system (other than Earth) with both liquid andsolid material at its surface. Voyager obtained images ofall 18 known satellites of Saturn. These images revealeda geologically young surface on ENCELADUS, erratic rota-tion for HYPERION and a remarkably sharp boundarybetween dark material on Iapetus’ leading face andbright material on its trailing face. Phoebe, apparently acaptured asteroid, was revealed to have the darkest sur-face among Saturn’s satellites.



Voyager added 10 new satellites (Cordelia, Ophelia,Bianca, Cressida, Desdemona, Juliet, Portia, Rosalind,Belinda and Puck) to the five previously known atUranus. Miranda has unique surface features known ascoronae owing to their strange concentric patterns. Sixnew satellites (Naiad, Thalassa, Despoina, Galatea,Larissa and Proteus) were added to Neptune’s priorrepertoire of two. Triton was discovered to have nitrogenice, extremely cold temperatures (32 K), and geyser-likeplumes spewing dust through cracks in the icy surface.

MagnetospheresLittle was known about the magnetospheres of the giantplanets before the Voyager flybys. Pioneers 10 and 11, asVoyager ‘pathfinders’, had made preliminary measure-ments of Jupiter’s magnetosphere, and Pioneer 11probed Saturn’s magnetosphere. Voyager made signifi-cant improvements in those measurements and providedthe first measurements of the magnetospheres of Uranusand Neptune. The magnetic fields of the latter two plan-ets were found to be highly inclined to the planetaryrotation axes, something previously expected only forcertain stars.

The final frontier for Voyager is a search for the outeredge of the heliosphere (the Sun’s magnetosphere).Theoretical calculations show that the boundary shouldbe at a distance of between 80 and 120 astronomical units(AU) (1 AU is the mean distance between Sun and Earth).Voyager 1 crossed 80 AU in 2001; Voyager 2 reaches thatdistance in 2006. Once Voyager 1 is beyond the helios-phere, humanity will be able for the first time to samplethe magnetic fields and charged particles in the spacebetween the stars.

The two Voyager spacecraft are still sending backdata, 25 years after their launch. Voyager 1 is now themost distant human-made object; in August 2002 it wasabout 85 times as far from the Sun as the Earth andVoyager 2 was about 68 times the Sun–Earth distance.They are both continuing outwards at about 3 AU peryear (slightly more for Voyager 1 and slightly less forVoyager 2). The Voyager team at JPL still receives infor-mation almost daily from the Voyagers which are exam-ining the far reaches of the solar wind.

BibliographyA detailed pre-launch description of the Voyager mis-sion, the spacecraft and the scientific investigations isgiven in Space Science Reviews 21 pages 75–376 (1977).

Voyager scientific results have been chronicled inmultiple dedicated issues of Science, Nature, Icarus andthe Journal of Geophysical Research. They have also result-ed in at least six separate many-hundred-page scientifictexts on planetary science published by The University ofArizona Press.

Ellis D Miner


Vulcan E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

VulcanThe name given to a hypothetical planet believed byUrbain Le Verrier to exist within the orbit of Mercury.At the request of Francois Arago, Le Verrier first studiedthe orbit of Mercury in the early 1840s. He managedto explain the greater part of the discrepancy betweenits calculated and observed positions (the advance of itsperihelion) as being caused by gravitational perturbationsby the other planets. The residual discrepancy eludedhim, and he abandoned the problem as intractable. Hereturned to it in 1859, having in the meantime successfullypredicted the existence of Neptune. Now he invokedan intramercurial planet or asteroid belt as the cause ofMercury’s irregularities. Shortly after, he heard that aphysician and amateur astronomer, Edmond Lescarbault,had observed a small body in transit across the Sun’s disk.Le Verrier examined Lescarbault’s observations closelyand, convinced they were genuine, announce the existenceof a new planet which he named Vulcan. Subsequentsightings were reported but never confirmed. Vulcan isnow known not to exist.

See also: perihelion.


Vulpecula E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Vulpecula(the Fox; abbrev. Vul, gen. Vulpeculae; area 268 sq. deg.)a northern constellation which lies between Cygnus andSagitta–Delphinus, and culminates at midnight in late July.It was introduced as Vulpecula cum Ansere (the Fox andGoose) by the astronomer Johannes Hevelius (1611–87) ofDanzig (Gdansk), who included it in his atlas FirmamentumSobiescianum sive Uranographia of 1687.

A small, inconspicuous constellation, the brighteststar in Vulpecula is α Vulpeculae, magnitude 4.4. TheMilky Way passes through Vulpecula and the constellationcontains a number of open star clusters and planetarynebulae, including Cr 399 (Brocchi’s Cluster, the Coat-hanger), which consists of about a dozen stars ofmagnitudes 6–8, elongated east–west, NGC 6940, anotheropen cluster with about 100 stars fainter than ninthmagnitude, and M27 (NGC 6853), the Dumbbell Nebula,an eighth-magnitude planetary nebula.


W M Keck Observatory ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS


W M Keck ObservatoryThe W M Keck Observatory, located on the island ofHawaii, operates the world’s two largest optical/infraredtelescopes, each with a primary mirror 10 m in diameter,near the 4200 m summit of Mauna Kea. Made possiblethrough grants totaling more than $140 million from theW M Keck Foundation, the observatory is operated bythe California Institute of Technology, the University ofCalifornia and the National Aeronautics and SpaceAdministration (NASA), which joined the partnership inOctober 1996. The Keck I telescope began scienceobservations in May 1993; Keck II in October 1996.

A staff of 80 scientists, engineers, technicians andsupport personnel operate the observatory whoseadministrative facility is located in Waimea, Hawaii, andwhose mission is to provide a world-class researchfacility for astronomers from Caltech, University ofCalifornia, NASA and the University of Hawaii. Over400 astronomers per year are involved with observationsfrom the Keck telescopes, which are carried out fromWaimea via a fiber-optic link to the summit of MaunaKea.

The Keck telescopes employ a unique segmenteddesign for their primary mirrors: 36 1.8 m diameterhexagonal segments are fitted together like a floor-tilemosaic to form each primary, the segments being alignedwith respect to each to a tolerance of one-millionth of aninch under computer control. This expandable technologyis that most likely to be adopted for the giant telescopesbeing planned for the 21st century.

A $55 million project funded by NASA is aimed atjoining the two Kecks, along with four 1.8 m ‘outrigger’telescopes, as an infrared interferometric array to achieveunprecedented angular resolution, by the year 2003.

The first test observation obtained by linking the twoKeck 10 m telescopes was made on 12 March 2001.Light from HD61294, a faint star in the constellationLynx, was captured by both Keck telescopes. Thecollected light waves were combined and processed witha beam combiner and camera. To phase the twotelescopes properly, adaptive optics on both telescopesremoved the distortion caused by the Earth’s atmosphere.In addition, the optical system in the tunnel adjusted thelight path to within a millionth of an inch.

Major discoveries from this young observatoryinclude: the discovery of several planetary systemsaround other stars; the identification of gamma-raybursters as being at cosmological distances; the discoveryof the most distant objects in the universe; themeasurement via supernova observations of the apparentacceleration of the universe.

For further information seehttp://www2.keck.hawaii.edu:3636/.

Wallenquist, Åke Anders Edvard (1904–94) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wallenquist, Åke Anders Edvard(1904–94)Swedish astronomer, worked at the Bosscha Observatoryin Indonesia and became professor in Uppsala at theKvistaberg Observatory. He worked on double stars andopen star clusters.


Ward, Seth (1617–89) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Ward, Seth (1617–89)Born in Aspenden, Hertfordshire, became SavilianProfessor of Astronomy at Oxford, formulated an ‘emptyfocus’ alternative to KEPLER’s law of areas. (A planet movedwith uniform angular velocity around the empty focus ofits ellipse.)


Warner & Swasey Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Warner & Swasey ObservatoryLocated at Washburn University in Topeka, Kansas, homeof the Warner & Swasey 29 cm refractor. Built in the late1800s, the telescope was displayed at the 1912 World’s Fair,then acquired by Washburn College. Crane Observatorywas built on campus to house the telescope. The Warner &Swasey survived a tornado in the 1960s. During telescoperefurbishment, which was completed in 1998, all of theoriginal parts were retained.

For further information seehttp://www.icstars.com/warner swasey/wsmain.htm.


Warner, Worcester Reed (1846–1929) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Warner, Worcester Reed (1846–1929)American mechanical engineer, designed the 36 intelescope for the Lick Observatory and built telescopes forCanada andArgentina. With AMBROSE SWASEY he establisheda machine manufacturing company, and established theWarner & Swasey Observatory.


Water Cycle E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Water CycleThe Earth’s water cycle is the endless circulation of wateron a planetary scale, from ocean to continent throughthe atmosphere and back to the ocean, under the drivinginfluence of solar energy and gravity. This main cycle infact consists of many overlapping loops and branches atall scales, involving more or less durable storage, more orless rapid transfers, more or less intense transformationsbetween water phases. Because of this complexity, it wasonly in the late 16th and 17th centuries that the concepts ofwater cycle and balance arose, with the pioneering workof B Palissy, P Perrault, E Mariotte and E Halley foundingscientific hydrology. Earth is a system approximatelyclosed for water but largely open for energy. The particularproperties of water together with the strong couplingbetween water and energy cycles regulate the Earth’s meantemperature and shape weather and climate patterns andtheir fluctuations (see CLIMATE). The water cycle interactsalso strongly with most biogeochemical cycles, with long-term environmental feedbacks. Human activities havean increasing impact on the water cycle and the coupledfluxes, with some alarming consequences for the quantityand quality of water resources.

Properties of waterWater (H2O) is a very stable molecule (covalent link),which has very specific physicochemical properties,mostly because of its dipolar structure and the relatedhydrogen link between molecules. These propertiesexplain its multiform importance within the Earth’shydrosphere and its four main functions as resource,agent, vector and biotope.

• Its three phases (solid, liquid, vapor) are present withina relatively small range of temperature, correspondingto the temperature range observed on the Earth.

• Its very high latent heats of vaporization (2.454 MJ kg−1

at 20 C) and fusion (0.334 MJ kg−1) play a major role inthe heat transfer by water through the atmosphere.

• Its high specific heat and low thermal conductivity makewater bodies very good thermal regulators and waterfluxes very good heat convectors: oceans play a majorrole in heat storage and transfer.

• Water vapor absorbs most of the infrared and ultravioletradiation but is transparent to visible radiation;atmospheric vapor leads to the ‘greenhouse effect’,while liquid water and solid water have contrastingreflectivities. These radiative properties are central tothe thermal budget of the Earth (see GREENHOUSE EFFECT).

• Because of its high dielectric constant, water is anexcellent solvent, able to alter most minerals: naturalwaters are in fact more or less dilute aqueous solutions,providing nutrients to aquatic life and contributingby their movement to the redistribution of substances;water can also combine with many substances.

• Its density is maximal at 4 C, permitting ice to float overliquid water: freezing and warming of water bodiesoccur from the top, progressive freezing of the wholeocean being avoided.

• Its low viscosity makes it a very mobile liquid, able tooccupy all the space available, even within very smallpores, and its surface tension allows soil water retentionby capillarity against gravity, by which vegetation canpersist between two rainfalls.

• Water molecules present several isotopic combinations,whose relative abundances provide very useful (pa-leo)thermometers and tracers of the water’s origins.

Origin and uniqueness of water on the EarthEarth, the ‘blue planet’, is unique within the solar systemby the presence in great abundance of water in its threephases.

The Jovian planets (Jupiter, Saturn, Uranus, Neptune)are composed of light elements, including hydrogen andhelium, and most of their satellites contain a large fractionof water ice, because of the very cold conditions prevailingso far from the Sun. The Jovian satellite Europa, inaddition, might contain a water ocean below its icy crust.In the Telluric planets (Mercury, Venus, Earth, Mars)constituted of heavy elements, water was partly outgassedby tectonic and volcanic activity and partly brought byimpacts of meteorites of asteroidal or cometary origin. Thefate of water in the Telluric planets has differed largelyaccording to the very different temperature and pressureconditions prevailing on each planet, mostly as a result oftheir distances from the Sun (figure 1).

While Mercury no longer has an atmosphere andprobably still contains only a little ice in cold regions,on Venus the thick carbon dioxide atmosphere is muchtoo hot for water condensation because of a stronggreenhouse effect, and most outgassed water vapor wassubsequently photodissociated (photolysis) (see VENUS).Mars is currently cold and dry because of its thin carbondioxide atmosphere, but there is much geomorphologicalevidence of past water-related erosion, and some limitedwater is probably still contained in permafrost and icecaps (see MARS). Earth is the only planet where thethree water phases coexist: a greenhouse effect dueto the presence of water vapor and carbon dioxide inits atmosphere has maintained temperature conditionspermitting the condensation of very huge amounts ofoutgassed water, and therefore the development of life.Uniformly distributed over the whole planet’s surface, thedepth of all the water on Earth would amount to about3000 m, as compared with rough estimates of about 200 mfor Mars and 0.20 m for Venus.

Main reservoirs and fluxesThe total amount of water within the Earth’s hydrosphere(that is, above and within the upper 2 km of the crust)can be considered as constant, even though there arevery small net water exchanges with both space (loss by



triple point

Earth

1,000

100

10

1

10,000

0.1

0.001

0.0001

0.01

ICE

ICE

ICE

Jupiter

Uranus

Mars

WATER

WATER VAPOR

Mercury ( daylight side)

PR

ES

SU

RE

(at

m)

TEMPERATURE (°C)

-200 -100 0 100 200 300 400 500

Venus

Figure 1. Planetary positions on the phase diagram of water. Adapted from National Research Council (1991). Reproduced bypermission of National Academy of Sciences.

diffusion, gain by icy comets) and the Earth’s mantle (lossby subduction, gain by volcanic activity), where there isa much slower tectonic water cycle. This huge amount ofwater (about 1386 million km3, of which only 2.5% is freshwater) is stored in several main interconnected reservoirsdiffering by the volume and quality (salt, fresh) of thewater they contain. The fluxes they permanently exchangewithin the water cycle control their renewal rates, whichdiffer largely according to their size and the water phasethey store. Current estimates of global storages and meanannual fluxes are still affected by rather large uncertaintiesespecially for groundwater (compare for instance valuesin table 1, table 2 and figure 2).

The largest reservoir is by far the world ocean, whichcovers 71% of the 510 million km2 of Earth’s surface with amean depth of 3700 m. It contains 96.5% of all the water, assalt water, and has a long mean residence time of 2500 yr.

Continents store the remaining 3.5% of water inseveral main reservoirs with a very large range of sizesand renewal rates.

• Ice caps, glaciers and permanent snow cover store 1.74%of total water over 3% of Earth’s surface (essentially inthe Antarctic). This solid phase represents 69% of allfresh water, with a very long residence time in ice caps(104 yr).

• Groundwaters within the upper 2 km of the Earth’scrust contain about 1.7% of total water (45% only asfresh water), with very variable residence time (1–103 yr) according to their depth and the porous mediaproperties. Soil moisture accounts for 0.001% only.

• Surface water (lakes, rivers, swamps) represents only0.01% of all water, most as fresh water. The mean

residence time is short for rivers (about 18 days) andmuch longer for lakes (up to 1000 yr).

• Biomass contains only 0.0001% of total water, but thehigh water content (>70%) of most plants and livingorganisms is renewed within a few hours or days.

The atmosphere stores 0.001% of total water, mostly asvapor. Liquid and solid phases are also present in clouds,which cover more than 60% of Earth’s surface on average.Atmospheric water corresponds to an equivalent heightof only 26 mm but is frequently and rapidly renewed in9 days, ensuring an intense recycling of water.

The global water cycle presents two main intercon-nected loops of different intensity (figure 2).

• Ocean and atmosphere exchange large fluxes: the oceanloses much water by evaporation (434 000 km3 yr−1;1202 mm yr−1), a flux only limited by the energyavailable. Most of this flux returns directly to the oceanby precipitation (398 000 km3 yr−1; 1102 mm yr−1).

• Continent and atmosphere exchange fluxes that aresmaller because less water is available: precipitationon continents (107 000 km3 yr−1; 719 mm yr−1)is in great part compensated by evapotranspiration(71 000 km3 yr−1; 477 mm yr−1)—that is, waterevaporated from the land surface and transpirated byvegetation.

• The water balances of both loops are equilibrated bythe fluxes they exchange: vapor flux in excess from theocean (36 000 km3 yr−1; 100 mm yr−1) is transportedby atmospheric circulation over continents, while anequivalent flux of precipitation excess on the continent(36 000 km3 yr−1; 242 mm yr−1) returns to the ocean bygravity as surface and subsurface runoff.



Table 1. Water reserves on the Earth.

Distribution Water Water Proportionarea volume depth of total water Residence

Reservoir (103 km2) (103 km3) (m) (%) time

World ocean 361 200 1 338 000 3700 96.5 2500 yrGlaciers, permanent snow 16 230 24 064 1463 1.74 Up to 104 yr

Antarctic 13 980 21 600Groundwater (upper 2 km) 134 800 23 400 174 1.7 Up to 103 yr

Fresh water 10 530 78 0.76Soil moisture 16.5 0.2 0.001 Days–months

Ground ice–permafrost 21 000 300 14 0.022Lakes 2 060 176.4 85.7 0.013 Up to 103 yr

Fresh 91Saline 85.4

Swamps 2 680 11.5 4.3 0.0008Rivers 148 800 2.12 0.014 0.0002 18 daysBiomass 510 000 1.12 0.002 0.0001 Hours–daysAtmosphere 510 000 12.9 0.025 0.001 9 days

Total water reserves 510 000 1 386 000 2718 100Total fresh water 148 800 35 029 235 2.53

Adapted from Shiklomanov (1993).

Table 2. Estimates of mean annual water fluxes at global scale.

National ResearchCouncil (1991) Shiklomanov (1998)

Water flux km3 yr−1 mm yr−1 km3 yr−1 mm yr−1

OceanEvaporation 434 000 1202 502 800 1392Precipitation 398 000 1102 458 000 1268Advection to continent 36 000 100 44 800 124

ContinentPrecipitation 107 000 719 119 000 800

Endorheic areas 9 000Exorheic areas 110 000

Evapotranspiration 71 000 477 74 200 499Endorheic areas 9 000Exorheic areas 65 200

Runoff to ocean 36 000 242 44 800 301From rivers 42 700From groundwater 2 100

TotalEvaporation 505 000 990 577 000 1131Precipitation 505 000 990 577 000 1131

Endorheic/exorheic areas: not connected/connected to the ocean.

Globally precipitation and evaporation fluxes areequal and amount to 505 000 km3 yr−1 (990 mm yr−1).

Space and time variationsMean global values of storage and annual fluxes in factmask very large variations at different scales in both timeand space. Variations in evaporation (E), precipitation(P ) and the resulting continental runoff are mainly dueto seasonal and latitudinal variations in solar energymodulated by seasonal thermal contrasts between oceanand continent (figures 3 and 4). An observation networkand remote sensing mainly from satellites help to assessand monitor the variability of water storages and fluxes

and the environmental factors controlling them (see SPACE-

BASED OBSERVATIONS OF THE EARTH).In the intertropical zone, where solar energy is

maximal, the redistribution of water and energy withinthe atmosphere is ensured by the rather stable verticalcirculation within the Hadley cells generated by largevertical contrasts in temperature, in the absence of a strongCoriolis effect. In the equatorial zone, the convergence oftrade wind systems generates the uplift (low pressure) ofwarm air moistened by high evaporation from the ocean:this results in heavy convective precipitation exceedingthe local evaporation, which is reduced by frequent cloudcover (P > 2000 mm yr−1, E < 1600 mm yr−1). Inversely,



Precipitation398

Precipitation107

Wind36

(River, etc...)36

MIXED LAYER 50,000

THERMOCLINE 450,000

ABYSSAL 890,000

LAND

Evaporation &Transpiration

71

Evaporation434BIOMASS

2

SURFACE WATER360

UNDERGROUND WATER15,300

Reservoirs, volumes in 103 km3

Fluxes in 103 km3 yr-1

Total Reservoir Volume = 1.46 x 109 km3

ICE & SNOW43,400

MARINE ATMOSPHERE 11

OCEAN

RunoffOCEAN

1,400,000

TERRESTRIAL ATMOSPHERE 4.5

Figure 2. The water cycle at a global scale. Adapted from National Research Council (1991). Reproduced by permission of NationalAcademy of Sciences.

subsidence of warm and dry air (high pressure) in thesubtropical zones favors evaporation, which exceeds localprecipitation (P < 1200 mm yr−1, E > 1800 mm yr−1).

In the mid-latitude zones (between 30 and 60), theredistribution of water and energy from tropics to polesis rather organized horizontally, through successions ofunstable eddies moving from west to east and generatedby horizontal contrasts in temperature and by the Corioliseffect: cyclonic depressions provide precipitation, whiledry anticyclonic conditions favor evaporation. At highlatitudes, because of minimal solar energy and subsidenceof cold and dry air (high pressure) precipitation andevaporation are very low (only a few mm yr−1 at thepoles). In the extratropical zones, precipitation generallyexceeds evaporation, but, while evaporation decreasesrather regularly toward the poles, precipitation presentsin the mid-latitude zone a maximum (P > 1200 mm yr−1)associated with the depression systems.

This latitudinal climate zonation is most clearlyobserved on the ocean, where evaporation is only limitedby the energy available. Differences in water (P −E) andenergy balances associated with each climatic zone alsogenerate a salinity pattern at the ocean surface (increasingby sea-water evaporation and freezing, decreasing byprecipitation and sea-ice melting). Redistribution ofwater, heat and solutes within the world ocean isensured by large-scale circulations and eddies (see OCEANS).

While rapid surface currents are mainly induced by theprevailing surface winds, slow deep currents are mainlydue to differences in water density (depending on watertemperature and salinity). This thermohaline circulationgenerates the oceanic ‘heat conveyor belt’: in its mainbranch, cold dense water sinks in the Norwegian Sea, flowsas deep water through the Atlantic ocean to the Indian andPacific oceans where it warms and wells up and returns tothe North Atlantic as dilute surface water in about 1000 yr.

On continents, where evapotranspiration is limited bywater rather than energy, the climate pattern exhibits stillmore contrast, with warm deserts belts (P < 300 mm yr−1)in the dry subtropical zones. This pattern is disturbed byother factors depending on the position, size and shape ofeach continent, whose cumulative effects result in a verylarge range of mean continental precipitation (from almost0 to 12 000 mm yr−1).

• Ocean–continent interactions: in the tropical zone, thewestern coasts of continents receive less precipitationthan the eastern coasts, which are supplied by easterlytrade winds carrying much moisture evaporated fromthe tropical ocean—for instance in theAmazonian basin.

• Continentality effect, especially in the middle latitudezones: precipitation on continents decreases from westto east, because of a progressive drying of air mass.Precipitation is more than 600 mm yr−1 in temperate



over land

1000800600400200

mm.yr-1

over seamm.yr-1

200016001200

800400

90° N

60°N

30°N

0°

30°S

60°S

90°S180°W 120°W 60°W 0° 60°E 120°E 180°E

Figure 3. Global distribution of mean annual evaporation. Adapted from Robinson and Henderson-Sellers (1999). Reproduced bypermission of Pearson Education.

300020001000

500250

mm.yr-1

90° N

60°N

30°N

0°

30°S

60°S

90°S180°W 120°W 60°W 0° 60°E 120°E 180°E

Figure 4. Global distribution of mean annual precipitation. Adapted from Robinson and Henderson-Sellers (1999). Reproduced bypermission of Pearson Education.



Western Europe, less than 100 mm yr−1 in continentalKazakhstan.

• Orographic effect: mountain ranges on continents,islands in oceans, act as barriers to dominant humidair mass. The windward side receives much moreorographic precipitation than the leeward side—as isthe case for the Western Cordillera in North America orthe Hawaiian Islands.

These mean annual spatial patterns are modulatedby diurnal and seasonal variations in insolation or longer-term fluctuations.

• At high altitudes as at high latitudes, a great part ofwinter precipitation falls as snow, forming a seasonal oreven permanent snow cover affecting both water andenergy cycles.

• Slight seasonal fluctuations of the Hadley cells aroundthe equator explain the alternation of rain and dryseasons in the tropical zone.

• Seasonal variations in thermal contrasts between oceanand continent also explain the monsoon, providinghuge amounts of summer precipitation to the Indiansubcontinent.

• Multiannual fluctuations can also occur in the ocean–atmosphere system—as for instance the ‘El NinoSouthern Oscillation’ (ENSO) affecting the Pacific buthaving dramatic effects on the water balance of thewhole tropical zone.

In the much longer term (10–104 yr), cyclic variationsin solar activity and Earth’s orbital variations modifythe insolation and the seasonal contrasts between bothhemispheres, and therefore the climate and watercycle patterns. Ice caps and oceanic sedimentsprovide paleoclimatic archives allowing us to date suchfluctuations and characterize their hydrological impact,especially during the past glaciation–deglaciation cycles.

Precipitation generally exceeds evaporation on con-tinents, but both are lower than on oceans at the samelatitude, because less energy (higher albedo) and water(unsaturated soils) are available on continents. Water inexcess within continental basins is collected by aquifer andriver networks and returns in great part to the ocean asriver runoff, with inputs spatially concentrated at the rivermouths. The relative contribution to total river runoff tooceans is 16% for the Amazon alone, 27% for the five majorriver systems (draining 10% of continental areas) and 50%for the 50 major rivers, while 40% of river runoff is con-centrated in the equatorial region (flowing mainly into theAtlantic). Endorheic basins, that are not connected to theocean, represent 22% of continental areas (mainly in dryregions): runoff disappears by infiltration or accumulatesin closed lakes and seas, where water evaporates—as inLake Chad or the Aral Sea.

River runoff varies during the year with usually astrong seasonality of high and low flows, and successions

of relatively short floods separated by longer recessionperiods—with the extreme case of ephemeral streams.Runoff variations are controlled by the time and spacevariabilities of precipitation and evaporation, which arevery different. Precipitations are usually episodic andintermittent, more or less concentrated during rainyseasons and irregularly distributed in space, while theenergy available for evapotranspiration is much moreregularly and continuously distributed but more affectedby diurnal and seasonal cycles. Evaporation intensity,which is limited by the insolation, varies within amuch smaller range (0–10 mm day−1) than precipitationintensity, whose maximal recorded local values are greaterthan 40 mm in 1 min, 400 mm in 1 h, 1900 mm in 1day, 26 000 mm in 1 yr. Precipitation is often bufferedseasonally in snow cover and glacier at high latitudes oraltitudes, while evapotranspiration depletes soil moistureand is often reduced by summer water stress. Theinterannual variability is much larger for precipitationthan for evapotranspiration. There is a large varietyof river runoff regimes as a result of combinations ofprecipitation and evaporation regimes on their basins.

Water cycle on the basin scaleAs the oceanic and atmospheric branches of the watercycle are presented elsewhere (see OCEANS and EARTH’S

ATMOSPHERE), only its continental branch is detailed here,especially on the basin scale. A basin (or catchment)is defined as the total drainage area of a river at somespecified outlet. Catchments, which can be defined at anyscale and in any environment, are functional units veryuseful for both the study and the management of waterresources.

The water balance of a catchment expresses theprinciple of water conservation: the incident precipitationis redistributed between runoff, evapotranspiration andwater storage within the catchment, through a competitionbetween three main sources of energy and drivingforces: (1) gravity, which controls most flows of liquidand solid water and depends mainly on topographyand morphometry; (2) the evaporative demand of theatmosphere, which controls vapor fluxes and dependson energy input (radiative, advective) and climaticconditions; (3) capillarity, which controls soil waterretention and depends on the pore size distribution. Atany point and any time, the dominant form of energy andthus the dominant component in the water balance dependon soil moisture conditions.

There is a large variety of surface and subsurfaceprocesses and pathways governing the hydrologicalresponse of a catchment to water and energy inputs fromthe atmosphere. They result in a strong vertical andlateral redistribution and recycling of precipitated waterwithin a catchment. The main fluxes (evapotranspiration,groundwater recharge, streamflow) are usually generatedsimultaneously or successively by several processes,whose combinations are very variable in time and space.They are indeed controlled by local characteristics and



variations of atmospheric inputs, by the water storageand transfer properties of the various compartments andtheir geometry. Because of many functional thresholds,catchment response is usually highly non-linear, muchdepending on initial water conditions.

After some interception by vegetation, precipitationreaches the soil surface: snow remains stored on thesurface until it melts, while rain infiltrates into the soil andrecharges its water storage. When soil moisture exceedssome threshold depending on both soil texture andstructure, water percolates further vertically to rechargegroundwater. Biological activity, soil cracks and bedrockfissures, chemical dissolution (karstic cave systems,)often create networks of non-capillary macropores, whichprovide preferential pathways for free water, bypassingthe less permeable bedrock or soil matrix. If thepercolating flux exceeds the hydraulic conductivity ofsome subsurface layer, water in excess forms a perchedgroundwater or flows laterally as hypodermic interflow.Groundwater flows are essentially lateral and controlledby the conductivity of the porous media and the slope ofthe piezometric surface (which is the water table slopefor unconfined aquifers). If the water table is shallow,precipitation can create locally rapid groundwater ridgingby saturating its capillary fringe.

If precipitation intensity exceeds the infiltrability(depending on surface conditions), the soil surface issaturated from above: water in excess accumulates insmall depressions, then flows as diffuse surface runoff.If subsurface lateral flows exceed locally a thresholddepending on soil slope, depth and conductivity, soilsurface is saturated from below: the excess flux exfiltratesas return surface flow. Precipitation on already saturatedareas also tends to flow as surface runoff. Whateverits generating process (infiltration–excess, exfiltration,saturation–excess), surface runoff tends to concentrate insmall rills to reach the stream channel but can also infiltratedownslope, depending on local surface topography,roughness and infiltrability.

Surface and subsurface processes contribute tostreamflow, provided that the generated fluxes areconnected to a stream channel. While the slow drainageof groundwater and soil water supplies stream baseflowbetween successive rainy events, floods are generatedby rapid surface and subsurface flows and by directprecipitation on the river. Streamflow is routed downwardas a turbulent flux controlled by the bed slope androughness. Within the channel network, it increasesat each confluence, decreases at each diffluence andaccumulates in lakes and artificial reservoirs. Along thechannel, streamwater and groundwater exchange fluxeswhose direction and intensity depend on their waterlevels and the bed’s conductivity. Flooding occurs whenstreamflow exceeds some threshold fixed by the bedgeometry.

Both surface and shallow subsurface waters aresubject to evapotranspiration, returning water vapor tothe atmosphere. Evaporation combines a phase change

(liquid or solid water to vapor), which consumes muchenergy (radiative or thermal), and a turbulent vaportransfer within the air, which increases with air saturationdeficit and wind speed. Transpiration of the vegetationis the vaporization within leaf stomata of the sap flowextracted from soil by roots. Evapotranspiration is limitedby the water and energy availability: evaporation offree water (wet vegetation, surface water, groundwater)depends on the available energy only, while capillaryretention makes water in unsaturated soil less and lessavailable when soil dries out. During dry periodstranspiration, which exploits water within the wholeroot zone, lasts much longer than evaporation, whichaffects the soil top layer only. Stomatal regulation limitstranspiration in dry soil conditions, to avoid vegetationwilting.

Any catchment presents a functional, spatial andtemporal organization, because of the non-uniform andnon-random distributions of the hydrological processes,factors and inputs: (1) at each point, process activationor deactivation results from a balance between watersupply from above and local water storage or transfercapacities; (2) spatio-temporal variations of factors leadto some recurrence of conditions favorable or unfavorableto each process in some areas of variable extent andsome periods of variable duration; (3) these variableactive areas and periods are also contributing areas tomain outfluxes only if they are hydraulically connectedto the catchment boundaries. Because of many possibleinteractions and processes, zones of interface betweenatmosphere, vegetation and soil, or between surface waterand groundwater are very sensitive to any change.

Coupled water, energy and matter cyclesWater and energy cycles are intimately coupled in theatmospheric and oceanic general circulations, controllingthe radiation and energy balances at both global and localscales, and carrying energy from equator to poles andwater from ocean to continent.

• Atmospheric water vapor maintains a rather hightemperature at Earth’s surface by a ‘greenhouse effect’,while the net solar radiation on surfaces is reduced byan ‘albedo effect’ depending on water phase and contentand generating local thermal contrasts: reflection is veryhigh for low clouds and ice or snow cover, low forwater bodies and higher for dry bare soils than for moistvegetated soils.

• At evaporation sites, cooling occurs by vaporizationconsuming much energy (solar radiation, sensible heat),which is stored as latent heat in the water vapor;both are transported within the atmosphere by verticalconvection and lateral advection to the precipitationsites where warming occurs by condensation releasingthe latent heat as sensible heat in the atmosphere.

• Huge amounts of energy are stored as sensible heat inocean water and transported from warm to cold regionsby ocean currents (for instance the great ‘heat conveyor



belt’): the ocean acts as a thermal regulator, temperingalso seasonal climatic contrasts in many continentalregions.

Water and matter cycles are also strongly coupled.On continents, water is a powerful agent of transforma-tion and vector of material, depending on climate, topog-raphy, bedrock and soil types and vegetation cover. Wa-ter in porous media (rocks, soils) increases their weath-ering by low-temperature biogeochemical alteration andmechanical fragmentation and reduces their mechanicalresistance. Surface flows of water, snow and ice gener-ate soil and river bed erosion and carry downward theeroded material, while subsurface flows transport soluteswashed out of the soil matrix and favor mass movementson slopes (landslides, solifluxion). At river outlets, theocean receives large amounts of fresh water and its so-lute and particulate load, which modify locally its salinityand induce mixing processes. These point inputs of matterfrom continents contribute to compensate for the chemicalprecipitation and mechanical sedimentation occurring inthe ocean and to maintain constant its mean salinity. Trans-port of material by water is still much more discontinu-ous and irregular than water flow, with many possible cy-cles of erosion–sedimentation, dissolution–precipitation,concentration–dilution. Evaporation and freezing tend toconcentrate water solutions, while precipitation and melt-ing tend to dilute them, even though precipitated watercan reach high concentrations in some elements throughthe washing out of the atmosphere. As a result, waterquality varies all along the different branches of the watercycle. In the long term, water activity modifies the land-scape, with some possible feedback on water pathwaysand flows.

Since life originated in the ocean about 3.4 billionyears ago, water is essential in the biosphere also, whereits cycle is strongly coupled with other biogeochemicalcycles. Water is the main component of biomassand is involved in major metabolic reactions such asphotosynthesis and respiration. In terrestrial ecosystems,the redistribution within plants of nutrients extracted byroots is ensured by the sap flow induced by transpiration,and the development of vegetation modifies the surfacewater and energy properties. Vegetation maps reflectwell at all scales the patterns of climate and waterbalance—as can be clearly seen in Africa, with thelatitudinal succession of equatorial rain forest, tropicalsavanna, semi-arid steppe and subtropical desert. Inaquatic ecosystems, which have some ability for self-purification, fresh, brackish or salt water bodies arebiotopes for many living organisms linked in complextrophic chains, structured according to the verticaland horizontal patterns of water’s physicochemicalparameters. Continental wetlands, estuaries and coastalregions are aquatic interface ecosystems very sensitive toany change in water quality and supply.

Humans and waterWater is an essential part of humanity’s environment: itis one of the most important natural resources for humanlife and activities; it is also a risk factor because of humanvulnerability to water-related hazards such as floodsand droughts, avalanches and landslides, waterbornediseases and toxicities. Humankind has therefore alwaysattempted to regulate the availability of water resourcesand to protect itself against such risks. Human activitieshave also many other major impacts on the water cycleand the coupled fluxes, directly by modifying continentalsurfaces and altering water quality and indirectly throughthe climate changes they could provoke. While the waterresources available tend to decrease, there is a rapidincrease in water demand by several uses competing forlimited resources. The struggle for water is becoming amajor source of regional conflicts, which could be avoidedonly by a strong international collaboration in order toincrease the necessary solidarities between upstream anddownstream regions.

Very fast population growth and economic develop-ment have dramatically increased in recent decades thedemand for high-quality fresh water, which is at presentestimated to be about 3750 km3 yr−1: about 2/3 for ir-rigated agriculture (providing 50% of food production),1/4 for industrial production and 1/10 for domestic use.Because of their easy accessibility, rivers, lakes and su-perficial aquifers are the main sources of renewable freshwater resources. Evaluated by the difference between con-tinental precipitation and evapotranspiration rates, cur-rent mean global estimations of renewable fresh water re-sources vary between 36 000 and 45 000 km3 yr−1, about95% as river runoff and 5% as groundwater. This globalamount is much larger than the present demand, andshould also largely meet the expected increase of about10% each 10 yr. Technological development should helpreducing water demand.

Unfortunately, these resources are very unevenlydistributed throughout the year and present largevariations from year to year, as shown by the intenseSahelian drought during the 1970s. They are alsovery unevenly distributed over continents, with patternsthat do not match the spatial distribution of populationand economic activities. Significant fresh water deficitalready occurs in many regions, affecting their social andeconomic development: 35% of the world’s populationhas very low or catastrophically low water supplies (forinstance in North Africa) and 75% lives in regions wherewater use exceeds 20% of available renewable waterresources, leading to a mining-type exploitation of deepergroundwaters, which rapidly become depleted becauseof their much lower renewal rates. Disparity in waterresources, population growth and economic developmentalready generates a very large inequality in fresh wateravailability per person: more than 500 l day−1 in developedcountries, less than 50 l day−1 in some arid regions. Watersupply is already one of the key issues for humankind.



Human activity also has a rapidly increasingimpact on the natural behavior of most hydrosystems.River and aquifer regimes are often severely modifiedby water intake for irrigation, artificial reservoirsregulating discharge, or massive water exchanges betweenbasins. Large-scale changes in land use (reafforestation–deforestation, urbanization, wetland drainage, fieldmanagement) have deeply affected the precipitationredistribution between evapotranspiration, surface runoffand groundwater recharge. Irrigation with poor-qualitywater has generated soil salinization in many arid regions.The discharge in river courses of untreated domestic andindustrial wastewater as well as groundwater rechargeby polluted irrigation-excess waters have also drasticallydepleted water quality, further reducing the availability ofgood-quality fresh water. In the longer term, the possibleglobal warming due to the anthropogenic increase inatmospheric carbon dioxide is likely to alter precipitationvolume and seasonality, especially in arid regions, andto raise the sea level because of ocean thermal dilatationand ice cap melting: this would submerge many coastallowland regions, which are among the most denselypopulated.

ConclusionsThe water cycle is very complex at all scales, because of thenumerous physical, chemical and biological processes in-volved and the high spatial and temporal heterogeneitiesof the controlling factors. Despite much progress recentlyachieved thanks to large interdisciplinary scientific collab-orations and international effort to develop measurementand observation networks, there is still much to be doneto understand and model all feedbacks and interactionswith coupled energy and matter cycles. Impacts of in-creasing anthropic change and possible climate change onwater cycle and resources are therefore still difficult to an-alyze and predict with some confidence. Nevertheless, thegrowing awareness of the numerous interactions control-ling the water cycle and the scarcity and fragility of freshwater resources should contribute to improve the protec-tion and management of both surface and subsurface wa-ter resources and the control of long-term effects of humanactivities on the Earth’s water cycle. It is crucial for the fu-ture of both humankind and the ‘blue planet’.

BibliographyDingman S L 1994 Physical Hydrology (Englewood Cliffs,

NJ: Prentice-Hall)Gleick P H (ed) 1993 Water in Crisis—a Guide to the World’s

Fresh Water Resources (Oxford: Oxford UniversityPress)

National Research Council 1991 Opportunities in the Hy-drologic Sciences (Washington, DC: NationalAcademyPress)

Robinson P J and Henderson-Sellers A 1999 ContemporaryClimatology 2nd edn (London: Longman)

Shiklomanov I A1993 World fresh water resources Water inCrisis—a Guide to the World’s Fresh Water Resources ed

P H Gleick (Oxford: Oxford University Press) pp 13–24

Shiklomanov I A 1998 World Water Resources: a NewAppraisal and Assessment for the 21st Century (Paris:UNESCO)

Shiklomanov I A (ed) World Water Resources at theBeginning of the 21st Century (IHP Monograph) (Paris:UNESCO) in preparation

Zebidi H (ed) 1998 Water: a Looming Crisis? Proc. Int.Conf. on World Water Resources at the Beginning of the21st Century (Paris, 3–6 June 1998) (IHP-V TechnicalDocuments in Hydrology No 18) (Paris: UNESCO)

Bruno Ambroise and Michel Vauclin


Wave–Particle Duality E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wave–Particle DualityThe concept that electromagnetic radiation and subatomicparticles behave in some respects like waves and in otherslike particles.

Various phenomena, such as interference anddiffraction, clearly demonstrate the wave-like nature oflight (and other forms of electromagnetic radiation).For example, if light of some particular wavelength,originating from a single source, passes through twoadjacent narrow slits, each slit then acts as a source oflight. Where the light waves spreading out from eachslit meet, they interfere with each other, their amplitudes(‘heights’) adding together where they are in phase (e.g.where two wavecrests coincide) and canceling where theyare completely out of phase (e.g. where a crest meetsa trough). On the other hand, the particle behaviorof light is demonstrated by the photoelectric effect, aphenomenon whereby certain substances emit electronswhen illuminated by a beam of light, but do so only ifthe wavelength of the light is shorter than a particularminimum value. This behavior is consistent with lightbeing a stream of particles (called photons) each of whichcarries a discrete quantity of energy that is inverselyproportional to the wavelength of that light. If the energyof a photon exceeds the minimum energy that is neededto expel an electron, an electron will be ejected, but if thephoton energy is less than this minimum, no electrons willbe ejected.

Streams of subatomic particles exhibit wave-likebehavior through phenomena such as interference anddiffraction. For example, if a beam of electrons passesthrough two narrow slits before falling on a phosphorscreen, which registers the arrival of each electron as a spotof light, the distribution of light spots takes the form of aninterference pattern consistent with the pattern that wouldbe produced by waves of a certain particular wavelength.

As Prince Louis Victor de Broglie (1892–1987)proposed in 1924, the wavelength (λ) associated with aparticle of momentum, p, is given by λ = h/p, where h isthe Planck constant (=6.63 × 10−34 J s). Since p = mv

(momentum = the mass of a particle multiplied by itsvelocity), λ = h/mv. The wavelength associated witha particle is known as the de Broglie wavelength. Forexample, the de Broglie wavelength for an electron of mass9.11×10−31 kg traveling at a speed of 3×106 m s−1 (1 percentof the speed of light) is: 6.63 × 10−34/(9.11 × 10−31 × 3 ×106) ≈ 2.4×10−10 m = 0.24 nm. For light of wavelength λ,the energy, E, of the photon is given by E = hc/λ, where cdenotes the speed of light, and the momentum associatedwith a photon is E/c = h/λSee also: electromagnetic radiation, electron, light, pho-ton, quantum theory, subatomic particles, wavelength.


Wavelength E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

WavelengthThe distance between two successive crests of a wavemotion. Usually, in transverse waves (waves withpoints oscillating at right angles to the direction of theiradvance), wavelength is measured from crest to crest. Inlongitudinal waves (waves with points vibrating in thesame direction as their advance), it is measured fromcompression to compression. The term is applied toelectromagnetic radiation which is regarded as a wavemotion. For example, blue light has a wavelength ofabout 440 nanometers and red light about 700 nanometers.X-rays have wavelengths of the order of 10−10 meters andradio waves of the order of meters. Wavelength is usuallydenoted by the Greek letter lambda (λ); it is equal to thespeed (ν) of a wave train in a medium divided by itsfrequency (f ): λ = ν/f .

See also: frequency.


Wegener, Alfred Lothar (1880–1930) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wegener, Alfred Lothar (1880–1930)German climatologist and geophysicist who suggested thephenomena of continental drift and plate tectonics. Hisidea was that a super-continent he called Pangaea hadbroken up, the pieces drifting to their present positions.His evidence was the fit of South America and Africa, andsimilarities in climate, fossil record and geology acrossthe join. His path had been prepared by ALEXANDER VON

HUMBOLDT and Frank Taylor, but his book, published inEnglish in 1924, was scorned. His ideas are now widelyaccepted.


Weight E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

WeightThe force experienced by a body resting on, for example,the surface of a planet. A person standing on the Earth’ssurface experiences weight because the surface on whichhe is standing resists the effect of the force of gravity whichotherwise would accelerate that person towards the centerof the Earth. In other words, there is a reaction up throughhis feet equal and opposite to the gravitational attractionexerted upon him on Earth. The weight of a body dependsupon the gravitational force to which it is subjected. On thesurface of a planet it is equal to the mass of the body timesthe surface gravity. For example, a body that weighed100 kg on the surface of the Earth would have the followingweights on the surfaces of the bodies listed below:

the Moon 16 kgMars 38 kgJupiter 264 kgthe Sun 2790 kga white dwarf 30 000 000 kg

See also: zero gravity.


Weizmann Institute of Science E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Weizmann Institute of ScienceAt the Weizmann Institute of Science in Israel, astrophysicsis practiced by a small theory group in the physics faculty.Main interests are high-energy astrophysics (currentlymostly gamma-ray bursters, compact stars, and cosmicrays) and galaxy dynamics (alternatives to dark matter).

For further information seehttp://www.weizmann.ac.il/physics/physics.html.


Werner, Johann (1468–1522) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Werner, Johann (1468–1522)Astronomer, mathematician, instrument-maker and ge-ographer, born in Nuremberg, Germany, follower of Re-giomontanus. Regiomontanus had suggested that the tim-ing of eclipses and the orbits of comets could be used asclocks to determine longitude. Werner developed a prac-tical version of this idea with the method of lunar dis-tances (i.e. measuring the angle of the Moon from the Sun).He published this concept in In Hoc Opere Haec CotinenturMoua Translatio Primi Libri Geographicae Cl’Ptolomaei (1514)and described an instrument with an angular scale on astaff from which degrees could be read off, to measurethe lunar distances. This founded the movement to deter-mine longitude by astronomy and led to the state-fundingof observatories at Greenwich and Paris, the invention ofthe chronometer by JOHN HARRISON, and the accurate mea-surement of star positions and planetary orbits.


Westerbork Synthesis Radio Telescope E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Westerbork Synthesis RadioTelescopeThe WSRT (Westerbork Synthesis Radio Telescope) inHooghalen, The Netherlands, is an aperture synthesisinterferometer that consists of a linear array of 14 antennasarranged on a 3 km east–west strip. The array works bycombining the signal from all the antennas and simulatesa 3 km aperture telescope.

The antennas are equatorially mounted 25 m disheswith an f/D ratio of 0.35. This type of mounting ensuresa fixed orientation of the receiving systems with respect tothe sky. Ten of the telescopes are fixed, 144 m apart, whiletwo nearby dishes are movable along a 300 m track andtwo others are on a 180 m track at a distance of 1.5 km.In the array, the baselines extend from 36 m to 3 km. Thepointing accuracy of the dishes is 15 to 20 arcseconds, thesurface accuracy is of the order of 1.7 mm.

The WSRT telescope with the first 12 antennas on a1.5 km baseline was inaugurated in 1970. The secondphase with two antennas on the 3 km extension wascompleted in 1980. A major upgrade due to be finalizedin 2000 has added multi-frequency front ends capabilityand new state-of-the-art backends. The array can routinelyoperate at 92, 49, 21, 18, 13, 6 and 3 cm wavelength andchange frequency in less than a minute. An observingcapacity with instantaneous bandwidth up to 160 MHz isbeing added.

The WSRT is used by astronomers from TheNetherlands but also from many other countries for a widerange of scientific research starting from our Galaxy toobjects at the far reaches of the universe. There are twoproposal deadlines each year. The WSRT also participatesin the European (EVN) and Global Very Long BaselineInterferometry networks.

The WSRT is operated by the Netherlands Foundationfor Research inAstronomy (NFRA), which is funded by theNetherlands Organization for Scientific Research (NWO).

For further information seehttp://www.nfra.nl.


Whipple, Fred Lawrence (1906–) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Whipple, Fred Lawrence (1906–)Astronomer, born in Red Oak, IA, became director ofthe Smithsonian Astrophysical Observatory. While stilla graduate student he helped compute the orbit of newlydiscovered Pluto. Using a new method of photographyfrom two separated wide-angle cameras, he triangulatedon meteor tracks and determined their orbits. Hededuced that nearly all are made up of bits from comets.He proposed the ‘dirty snowball’ model for comets,suggesting that comets have icy cores inside layers of dirt.This was confirmed in 1986 when the Giotto spacecraftflew near to, and imaged, Comet Halley. Whipple trackedartificial satellites to determine the shape of the Earth.


Whirlpool Galaxy E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Whirlpool GalaxyA face-on, spiral galaxy, M51, which derives its name fromits bold and clear-cut spiral pattern. Located at a distanceof some 20 million light-years in the constellation of CanesVenatici, M51 (also known as NGC 5194), which has twowell-defined arms that spiral out from its relatively smallcentral bulge, is a classic example of a ‘grand design’ spiral.It is classified as an Sc galaxy in the Hubble classificationscheme. Although, with a diameter of 65 000 light-yearsand a mass of about 5× 1010 solar masses, it is somewhatsmaller than the Milky Way Galaxy, it is several times moreluminous, its spiral arms being laden with bright youngclusters and HII regions.

M51 has a smaller, fainter companion (NGC 5195),which lies at the end of one of the larger galaxy’s spiralarms. Although NGC has traditionally been classified asan irregular galaxy, it contains an elongated bar structureand hints of incipient spiral arms. NGC 5194 appears tobe orbiting around M51 in a period of about 500 millionyears. Tidal interaction between the two, during their lastclose encounter, which took place some 70 million yearsago, probably played a large part in establishing the boldspiral pattern in M51 and stimulating a vigorous bout ofstar formation within it.

M51 was the first galaxy to be recognized as having aspiral shape. This discovery was made in 1845 by WilliamParsons (1800–67), third Earl of Rosse, with the aid of the1.8 m (72 inch) telescope that he had set up at Birr, inIreland, and which, at that time, was the largest telescopein the world.

See also: barred spiral galaxy, Hubble classification,interacting galaxies, irregular galaxy, Messier Catalog,spiral galaxy.


Whiston, William (1667–1752) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Whiston, William (1667–1752)Born in Norton, Leicestershire, became the controversialthird Lucasian Professor of Mathematics at CambridgeUniversity, succeeding ISAAC NEWTON. He used theprinciples of the Principia in a popular book to explainthe Book of Genesis and the Flood, which he envisaged ascaused by the near approach of a comet, an interestingforeshadowing of current theories. He published in1716 the first undergraduate course based on Newton’sprinciples under the cumbersome title Sir Isaac Newton’sMathematic Philosophy More Easily Demonstrated with Dr.Halley’s Account of Comets Illustrated. He wrote onnumerous problems of astronomy and observed an auroraborealis, solar eclipses and sunspots. His downfall in theso-called Whiston’s affair was due to his popularizationof what were regarded as heretical ideas about the Trinity,theology and politics. He wrote letters to archbishopsthat raised eyebrows, caused debates in the Housesof Parliament and caused his dismissal from his post.Plunged into poverty, he lived off the income of a smallfarm near Newmarket and lectured in the coffee housesof London, giving scientific demonstrations for a fee as anentertainment.


White Dwarfs E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

White DwarfsWhite dwarf stars, also known as degenerate dwarfs,represent the endpoint of the evolution of stars withinitial masses ranging from about 0.08 to about 8 solarmasses. This large range encompasses the vast majorityof stars formed in our Galaxy and thus white dwarf starsrepresent the most common endpoint of STELLAR EVOLUTION.It is believed that over 95% of the stars of our Galaxywill eventually end up as white dwarfs. The definingcharacteristic of these objects is the fact that their massis typically of the order of half that of the Sun, while theirsize is more akin to that of a planet. Their compact naturegives rise to large average densities and surface gravities.

The first glimpse of the existence of these objectscame in 1844, with the study carried out by F BESSELL,the great German mathematician and astronomer, of theproper motion of SIRIUS, the brightest star in the nightsky. The irregularities in the apparent motion of Sirius onthe celestial sphere led Bessell to suggest the presence ofan unobserved, solar-mass companion orbiting the brightprimary. The companion, Sirius B, was first observed byA G Clark in 1862 and represents, together with 40 Eri B,the first known examples of white dwarf stars.

The luminosity of the white dwarf in the Siriussystem is ∼10 000 times smaller than that of its bright,main-sequence companion. Its effective temperature wasthus thought to be suitably smaller than that of the Astar. However, the first spectrum of Sirius B, securedby W Adams in 1914, showed that the white dwarf hadan effective temperature quite comparable with that ofits companion. Since the stellar luminosity scales asR2T 4

eff , for the white dwarf to be both fairly hot andquite faint required it to have a very small radius, anda correspondingly very high average density. A similarconclusion was reached in the case of 40 Eri B whosespectrum was also secured at about the same time.However, how could such stars withstand the tendencyto collapse onto themselves under the influence of theirgravitational field?

The answer to that question came in 1925, whenR H Fowler first applied the newly-developed principlesof QUANTUM MECHANICS to stars. He showed that, in whitedwarf stars, the density is high enough for the gas offree electrons to become degenerate. Electrons are saidto be degenerate when a majority of them occupy thelowest possible energy states available to them. Thisoccurs, at fixed temperature, when the electrons arepacked sufficiently close to each other. Because of thePAULI EXCLUSION PRINCIPLE that no more than two electronswith oppositely directed spins may occupy the sameenergy state, the electrons retain kinetic motions evenwhen cooled to zero temperature. The amplitude of thiskinetic activity increases with increasing density, whenelectrons become more degenerate. In a white dwarf,the pressure generated by this kinetic motion, clearly ofquantum mechanical origin, prevents the gravitationalcollapse of the star. The first detailed stellar models

appropriate to white dwarfs were calculated in the 1930sby S Chandrasekhar, who received the 1983 Nobel Prize inPhysics for his achievements.

Observed propertiesSeveral properties of white dwarf stars can be determinedfairly directly from observations. Analyses of their energydistribution, as well as of their optical and ultravioletspectra, fix their effective temperature; they range from∼150 000 K for the hottest stars to the coolest degeneratedwarfs, at Teff near 4000 K. Spectroscopic analyses alsoyield the surface gravity (g = GM/R2, but traditionallygiven in terms of its logarithm to the base 10, log g) sincethe strength and width of spectral features are sensitiveto the density of particles in the atmosphere, which iscontrolled by the surface gravity. The average surfacegravity of white dwarf stars is log g ∼ 8 (cm s−2),compared with log g ∼ 4.4 for the Sun. This value implies,as we discuss below, that the mean STELLAR MASS mustbe of the order of half that of the Sun. The luminosityrange encompassed by white dwarfs exceeds 7 ordersof magnitude and reflects the large range of observedeffective temperatures (L ∝ T 4

eff); the faintest known whitedwarfs have L ∼ 10−4.3L, while the rare intrinsicallybrighter ones, just entering the cooling sequence, reachL ∼ (102–103)L.

Because of the intrinsic faintness of the most commonwhite dwarfs, the observation of these objects tends to berestricted to small distances; the bulk of the sample of∼2200 spectroscopically confirmed white dwarfs availabletoday resides within ∼1 kpc from the Sun. It mustbe noted, however, that work with large, ground-basedtelescopes and with the Hubble Space Telescope nowallows the observation of white dwarfs in GLOBULAR

CLUSTERS located at several kpc from the Sun. Traditionally,however, white dwarfs have been culled from samples ofobjects showing significant PROPER MOTION, and thus locatedrelatively nearby the Sun. A selection criterion basedon color then allows the distinction to be made betweentrue white dwarfs and nearby main-sequence stars,and their degenerate nature is then confirmed throughspectroscopic observations. More recently, large numbersof hot white dwarfs have been detected in colorimetricsearches. Selected fields are typically photographedthrough both U (∼3650 Å) and B (∼4400 Å) filters. Acomparison of both images for a given object allows theselection of blue or very blue objects, whose nature can beconfirmed through spectroscopic means. Surveys of thiskind, carried out both at high Galactic latitude and in theplane of the Galaxy, have yielded substantial numbers ofnew hot white dwarfs, selected only on the basis of theircolors without regard for their proper motion.

The existence of homogeneous samples containingsubstantial numbers of white dwarfs allows analyses of thestatistical properties of these objects. Their distribution inthe Galaxy is consistent with that of an old disk population,with an estimated scale height of 250–300 pc. Their spacedensity is of the order of 0.005 pc−3 forMbol < 15, and their



Figure 1. Sample optical spectra of white dwarf stars: from topto bottom, a DA spectrum, dominated by the regular pattern ofthe Balmer lines of hydrogen, seen here near their maximalstrength, a DO spectrum, dominated by the lines of ionizedhelium, in particular the n = 3 to n = 4 transition at 4686 Å, a DBspectrum, featuring many strong lines of neutral helium, and afeatureless DC spectrum, characteristic of a helium-rich star toocool (Teff < 12 000 K) to show neutral helium lines.

birthrate is of the order of (1.5–2.3)×10−12 pc−3 yr−1 whenallowance is made for the contribution of unseen whitedwarfs in BINARY SYSTEMS.

The optical spectra of white dwarf stars arecharacterized by a rich variety which reflects, to aconsiderable extent, complex and varying patterns ofatmospheric abundances. Sample spectra are shown infigure 1. About three-quarters of white dwarfs have aspectrum dominated by the Balmer series of hydrogen,which originates on the first excited, or n = 2, level ofthat atom. These objects are termed DA stars; they arefound over the whole effective temperature domain ofwhite dwarfs, from the very hottest stars above 100 000 K,where the hydrogen lines are fairly shallow, to the coolestwhite dwarfs near 4000 K, where the lines are sharp andextremely weak. Another important group of objects haveoptical spectra dominated by the lines of helium. Thespectrum reflects the vastly different chemical makeup oftheir atmosphere, which is dominated by helium. At higheffective temperatures (Teff > 45 000 K), the lines observedare those of ionized helium. These stars are called DOstars. Between 30 000 K and 12 000 K, neutral helium isthe dominant ion, and the spectrum is dominated by abevy of transitions from that atom; these are the DB stars.Finally, below 12 000 K, the effective temperature is toocool for a helium-rich atmosphere to show transitions ofneutral helium; the spectrum is then featureless, and istermed DC.

The scheme of spectral classification of white dwarfsis flexible enough to accommodate several other importantclasses of objects, not illustrated in figure 1; hence the DQstars, whose spectrum is characterized by the presenceof carbon, generally in molecular form, in a helium-richatmosphere too cool to show lines of neutral helium.Similarly, the DZ stars show lines of heavy elementsother than carbon, for example calcium, magnesium oriron. Sometimes, the star is hot enough for weak carbonor heavy element features to be present simultaneouslywith the dominant neutral helium lines: the stars arethen classified DBQ or DBZ stars, respectively. At higheffective temperatures, near Teff ∼ 100 000 K, another classof objects, termed PG 1159 stars, includes stars with spectrafeaturing ionized helium as well as highly ionized carbonand nitrogen features.

As hinted above, the chemical composition of theatmosphere of a white dwarf star is intimately related toits spectral appearance. The hydrogen line DA stars haveatmospheres where hydrogen is the dominant element,to the near complete exclusion of any other. For theirpart, the so-called non-DA stars, which encompass objectsof the DO, DB, DC, DQ and DZ spectral types, all havehelium-dominated atmospheres. Within both classes, thispurity is understood today as the result of gravitationalsettling in the intense gravitational field of the whitedwarf. Under its influence, all elements heavier thanthe dominant atmospheric constituent rapidly sink intothe deep atmospheric layers of the stars and remain outof sight. In fact, this settling mechanism is so efficientthat the presence of any element heavier than hydrogenin the atmospheres of DA stars, or helium in those ofnon-DA stars, constitutes a puzzle. For example, lines ofheavy elements (in the DAZ stars) or of helium (in theDAB stars) are occasionally seen in DA stars at effectivetemperatures below 25 000 K. Similarly, traces of heavyelements are observed in the DBZ and cooler DZ stars,while carbon is seen in the cool DQ stars. We discuss belowthe mechanisms which may permit the existence of suchchemical impurities in the photospheres of white dwarfs.

Another point of interest for white dwarfs is theirmass distribution. Only in a handful of cases are direct,dynamic masses obtained for these stars. In general,surface gravities are obtained either from the optical colors(e.g. on the Stromgren uvby system, suitable for DA starswith effective temperatures between 16 000 K and 8000 K)or, preferably, from optical spectra which tend to bemore sensitive to surface gravity. The M/R2 relationprovided by the log g determination can then be coupledto an assumed mass–radius relation, characteristic ofthe mechanical structure of these stars (see below), todetermine M . The mass distributions (based on opticalspectroscopy) available for samples of two spectroscopicsubclasses of white dwarfs are shown in figure 2. Themass distribution of the DA stars is sharply peaked, andthe mean mass is M = 0.590M, with a dispersion σ =0.134M. Tails extend at both ends of the mass spectrum:the low-mass tail of the distribution is thought to be



Figure 2. Mass distribution of DA (thick outline) and DB(shaded area) white dwarfs. The average mass appears thesame, but the DB sample studied seems to lack both thelow-mass and the high-mass end of the distribution seen for DAstars. These wings are probably related to binary evolution.

populated by low-mass white dwarfs with helium coresprobably resulting from binary and common envelopeevolution, whereas the high-mass tail could well be theresult of merging events in binary systems. Nevertheless,the case of Sirius B (M = 1.05M), located in a wide binarysystem, shows that there are genuine high-mass whitedwarfs which have not been produced by close binaryevolution. It is worth noting, as well, that independentmass determinations for roughly three dozen DA stars canbe obtained from the gravitational redshift of their Balmerlines. General relativity predicts that the wavelength ofphotons emitted in a strong gravitational field should beredshifted by an amount λ which depends on the M/Rratio. Provided that the star under scrutiny belongs to anopen cluster or a binary system with a well-determinedradial velocity, and that the spectral line shifts induced bythe high pressures of white dwarf photospheres are wellunderstood, the gravitational redshift can be isolated, andthe mass can be determined.

The mass distribution of the DB stars, while based ona smaller sample of well-studied objects, shows at the veryleast that there are no substantial differences between themean mass of the DA and of the DB stars (the latter beingM = 0.585M, with a dispersion σ = 0.063M). Thereis a tantalizing suggestion, however, that the low-massand high-mass tails present in the DA sample may not bepresent in the DB sample; this absence accounts for thesmaller dispersion of the DB mass determination.

While most of the white dwarfs identified todaycan be considered normal, garden-variety objects, twofamilies of peculiar, and fascinating, white dwarfs havealso emerged over the years. One such family is that of the

magnetic objects, which currently includes nearly 50 stars.The detected magnetic fields range from 105 G all the wayto 109 G and are measured through a variety of techniqueswhich include the detection of circular polarization in theoptical continuum (Be ≥ 107 G), the direct observation ofthe Zeeman pattern in absorption lines (106 G ≤ Be ≤107 G), and Zeeman spectropolarimetry for low fields(Be ≤ 106 G). In the high-field objects, the spectrumis so perturbed by the magnetic field that spectroscopicclassification is often difficult if not impossible. Theinferred field morphologies tend to be rather simpleand generally range from centered or offset dipoles toquadrupoles. Statistical arguments favor an evolutionarylink between the chemically peculiar main-sequence Apand Bp stars and the magnetic white dwarfs. The otherdistinct subgroup of interesting white dwarfs is that of thevariable white dwarfs, or pulsators, examples of whichare known among the hot PG 1159, the DB and the DAstars. The usefulness of these objects for our sounding ofthe internal structure of white dwarfs is discussed brieflybelow, as well as in the entries on ZZ CETI STARS and DB

PULSATING STARS.Perhaps somewhat surprisingly, white dwarfs appear

to be slow rotators, in the sense that they rotate moreslowly than would be expected if the ANGULAR MOMENTUM

present in earlier evolutionary phases had been conserved.Determinations of the rotation velocity of white dwarfsrely on three different techniques: (i) measurement ofthe additional broadening of the core of absorption linescaused by rotation; (ii) measurement of the variation ofthe degree of circular polarization caused by rotation inmagnetic white dwarfs; (iii) measurement of the rotationalsplitting present in the Fourier spectrum of the light curvesof pulsating white dwarfs. While the use of any ofthese techniques is restricted to small subsamples of whitedwarfs, the picture which emerges is one where substantialquantities of angular momentum must have been lost bystars on their way to the white dwarf stage. While nodefinite model exists to explain this loss, current ideasfocus on the fact that most stars will need to shed a lotof mass, perhaps concomitant with angular momentum,before they do become white dwarfs.

Origin and generic propertiesIt is generally believed that the immediate progenitorsof most white dwarfs are nuclei of PLANETARY NEBULAE,themselves the products of intermediate- and low-massmain-sequence evolution. As mentioned above, starsthat begin their lives with masses less than about 8M,that is the vast majority of them, are expected to becomewhite dwarfs. Among those which already have had thetime to become white dwarfs since the formation of theGalaxy, a majority have burned hydrogen and heliumin their interiors. Consequently, most of the mass of atypical white dwarf is contained in a core made of theproducts of helium burning, mostly carbon and oxygen.The exact proportions of C and O are unknown because ofuncertainties in the nuclear rates of helium burning.



The observed narrow mass distribution of isolatedwhite dwarfs discussed above is a remarkable propertyof this category of stars. Apparently, the process of massloss in white dwarf progenitors, which may have a widerange of initial masses, is regulated by mechanisms thatare tuned finely enough to leave remnants with similarmasses consistently. Also, the empirical evidence suggeststhat small amounts of helium and hydrogen are left overafter the mass-loss phases have subsided. Taking intoaccount previous thermonuclear history and the efficiencyof gravitational settling, the expected structure of a typicalwhite dwarf is that of a compositionally stratified objectwith a mass of ∼0.6M consisting of a carbon–oxygencore surrounded by a thin, helium-rich envelope itselfsurrounded by a hydrogen-rich layer. Such an objecthas an average density of ∼106 g cm−3, a millionfoldthat of a normal star such as the Sun. The thicknessesof the hydrogen and helium outer layers are not knowna priori and must depend on the details of pre-white-dwarfevolution. On theoretical grounds, however, it is expectedthat the maximum amount of helium that can survivenuclear burning in the hot planetary nebula phase is only10−2 of the total mass of the star and that the maximumfractional amount of hydrogen is about 10−4. Althoughthese outer layers are very thin, they are extremely opaqueto radiation and regulate the energy outflow from the star.They consequently play an essential role in the evolutionof a white dwarf. The question of the exact masses ofthe hydrogen and helium layers present in white dwarfsconstitutes a topic of intense research interest in the field.

The large opacity of the outer layers of a whitedwarf implies that radiation escaping from the staroriginates from the outermost region—the atmosphere—which contains, typically, less than 10−14 of the total massof the star. Spectroscopic and photometric observationscan only probe these outer regions, which are usuallydominated by hydrogen. Thus, a majority of whitedwarfs are referred to as hydrogen-atmosphere objects(or DA stars as discussed above). It turns out, however,that about 25% of the known white dwarfs do notpossess such a hydrogen layer. These are called helium-atmosphere white dwarfs (or non-DA stars) with, again,the understanding that the underlying carbon–oxygencore must contain essentially all of the mass, even thoughit is not directly observable.

Mechanical structure and coolingAs former nuclei of planetary nebulae, most white dwarfsare born in the form of extremely hot, collapsed objectswhich can only cool off: their nuclear energy sourcesare depleted, and gravitational energy can no longerbe tapped efficiently as degenerate electron pressureprevents additional contraction. Because this pressureis independent of the temperature, a white dwarf iscondemned to evolve at essentially constant radius.The mechanical structure of such a star is thereforespecified by the degenerate electrons. In particular,electron degeneracy is directly responsible for the curious

Figure 3. Evolutionary tracks (solid curves) of five (M = 0.4M,0.6M, 0.8M, 1.0M and 1.2M, from top to bottom)representative models of DA white dwarfs in theHertzsprung–Russell diagram. Each model is a compositionallystratified object made of a pure carbon core surrounded by apure helium envelope containing 10−2 of the total mass of thestar and an outermost pure hydrogen layer containing 10−4 ofthe total mass. The thick solid curves are isochrones, that is, lociof constant evolutionary timescales. The number next to eachisochrone gives the cooling time in units of 109 yr. The smallfilled circles indicate the onset of crystallization at the center ofeach evolving model. The open circles at high (low) luminosityindicate the onset of superficial convection (the convectivecoupling of the surface with the thermal core).

relationship between the mass and the radius of a whitedwarf: the more massive the star, the smaller its sizeis. Likewise, relativistic degeneracy is also responsiblefor the existence of a limiting mass above which a whitedwarf cannot exist. This limiting mass is known asthe Chandrasekhar mass, and is of the order of 1.4M.The effects of this peculiar mass–radius relationship areapparent in the diagram presented in figure 3 which showsthe cooling tracks of five representative DA white dwarfmodels that differ in their total mass.

Degenerate electrons also possess another propertyof high relevance for white dwarfs: they are excellentconductors of heat, and thus they thermalize the internalregions of white dwarfs efficiently (a familiar illustrationof this property is provided by the conduction electronsin ordinary metals). We can thus envision a typicalwhite dwarf as consisting of a nearly isothermal core thatcontains more than 99% of the mass, surrounded by athin, opaque, insulating, nondegenerate outer envelope.In the range of effective temperatures 16 000–8000 K,where the bulk of the known white dwarfs is found, coretemperatures vary from ∼2 × 107 K to ∼5 × 106 K. Thevery large temperature drop between the central regions



and the surface takes place mainly in the stellar envelope.In the cooler models, this temperature gradient leads to theformation of superficial convection zones, similar to thosefound in the Sun. When present, atmospheric convectionplays an essential role in the determination of the emergentflux from a white dwarf and, thus, in the interpretationof its spectrum and colors. The small open circles athigh luminosity in figure 3 show the onset of a superficialconvection zone due to the recombination of hydrogenin evolving models of DA stars. Convection plays alsoa key role in the subsequent evolution of a white dwarfby affecting directly the cooling rate. This occurs whenthe base of the superficial convection zone reaches intothe degenerate interior, thus coupling the surface with thecore and, thereby, increasing the rate of energy transferacross the outer opaque envelope beyond what is possiblethrough radiative transfer alone. The small open circles atlow luminosity in figure 3 show where convection startsto play a significant role in the cooling process.

Largely decoupled from the electrons, the (nondegen-erate) ions provide the thermal energy that slowly leaksthrough the outside, thereby producing the star’s lumi-nosity. Thus, an isolated white dwarf shines at the ex-pense of its thermal reservoir. In this context, the electronsdo not contribute significantly to the energy reservoir be-cause degenerate particles, already occupying their statesof lowest energy, cannot be cooled. As thermal energy isgradually lost from the star in the form of radiation, thekinetic motions of the ions lose amplitude, become corre-lated, and the ionic state evolves from a gas to a fluid to asolid. Ultimately, the reservoir of thermal energy becomesdepleted, and the star disappears from sight in the form ofa cooled-off, crystallized object known as a black dwarf.

It should be noticed that the timescale for this finaldemise is strongly dependent on the total mass of thewhite dwarf. Indeed, because of their larger masses andsmaller radii, more massive stars have larger internaldensities (for comparable temperatures) and, therefore,develop a crystallized core earlier, at higher luminosities.By the same token, they also reach earlier the statewhere the specific heat in the solid regime plunges tovery small values, a phenomenon well explained withinthe framework of the simple Debye theory of solids inquantum statistical mechanics. In effect, matter underthese conditions has lost its ability to store thermal energy,the energy reservoir of the white dwarf has become emptyand the star must then disappear from sight in a relativelyrapid and final phase sometimes referred to as ‘Debyecooling’. Figure 3 illustrates, among other things, howDebye cooling is a strong function of the total mass ofthe white dwarf. It is easily seen that the isochronesare strongly dependent on the total mass; crystallizationand subsequent Debye cooling are responsible for the‘accelerated’ evolution of the more massive models at lowluminosities. To complement this, figure 4 provides moredetails on the evolving structure of a representative whitedwarf model in a phase diagram.

Figure 4. Evolving structure of a representative model of a DAwhite dwarf in a phase diagram. This is one (0.6M) of the fivemodels presented in the previous figure. Each curvecorresponds to the density–temperature distribution from thesurface to the center of the model at an effective temperaturegiven by the number alongside. The solid, dotted, and thicksolid (for the three cooler epochs) portions of each curve indicatethe radiative–conductive, convective and crystallized regions.Electrons become degenerate to the right of the small filled circleon each curve. Likewise, the small open circle on each curveindicates the location where the ions become strongly correlated(fluid phase). The dashed curves define the compositiontransition zones, H–He at lower densities, and He–C at higherdensities.

A final point about the cooling evolution of whitedwarfs should be made. In the early, short-lived phaseof evolution following immediately the planetary nebulaphase, white dwarf interiors are still hot enough thatNEUTRINOS can be formed in great quantities there througha number of processes involving the electromagneticand the weak interactions. The vast majority of theneutrinos escape directly from the central regions wherethey are created to the outer space, thus contributing toan important stellar energy sink. For instance, neutrinoluminosities may become 2 orders of magnitude largerthan photon luminosities in these objects. The evolutionof a very hot, young white dwarf is thus dominatedby neutrino cooling. Neutrino processes largely specifythe cooling timescale and lead as well to a temperaturereversal in the stellar core. Such a reversal is still visiblein the hotter model shown in figure 4. By the time a whitedwarf has cooled down toTeff ∼ 25 000 K, however, the starhas lost its memory of the neutrino cooling phase, and itssubsequent evolution and structure depend exclusively onthe properties of its degenerate electrons and thermal ions.

It should be clear from this discussion that there



exists an intimate relationship between dense-matterphysics and the structure and evolution of white dwarfs.For instance, a detailed knowledge of the opacity andthermodynamics of strongly coupled plasmas is necessaryto compute the cooling rate of a white dwarf. Indeed, thisrate basically depends on how much thermal energy isstored in the interior of a star and how rapidly this energyis transferred from the hot core to the cold interstellarmedium through the thin, opaque outer layers. Hence, areliable description of the constitutive properties of denseplasmas is required to build a theory of evolving whitedwarfs. By the same token, the observed properties ofcooling white dwarfs can be used, in principle, to testtheories of strongly coupled plasma physics.

Spectral evolutionThere is strong observational evidence that spectralevolution takes place among white dwarfs, that is,some of the hydrogen-atmosphere stars become helium-atmosphere objects, and vice versa, during variousevolutionary phases. Indeed, the ratio of DA to non-DA white dwarfs changes as a function of effectivetemperature along the cooling sequence. In particular,the cases for the existence of a so-called DB gap—an interval of effective temperature from 45 000 K toabout 30 000 K in which no helium-atmosphere object hasbeen found—and of a cooler and narrower non-DA gapbetween ∼6000 K and ∼5000 K are well documented.Completely convincing explanations for these phenomena(especially in the case of the cooler gap) have not beenworked out yet, but the very existence of ‘holes’ in thedistribution of helium-atmosphere objects as a functionof effective temperature is a strong empirical proof that,at least, some of the white dwarf stars must change theirsuperficial chemical composition from helium-dominatedto hydrogen dominated and back to helium rich again ascooling proceeds. It is suspected that a complex interplaybetween mechanisms such as hydrogen and heliumseparation (through diffusion) and convective dilutionand mixing is responsible for the fact that a white dwarfmay show two different ‘faces’ along its cooling track. Forexample, in DA stars below 15 000 K, convective mixing isthought to be effective in bringing deep-lying helium to thesurface and drastically changing the surface composition.Unfortunately, at these effective temperatures, the heliumbrought to the surface is spectroscopically invisible, and itspresence must be inferred through rather indirect analyses.The case for spectral evolution is further strengthened byanother observational datum of importance: as pointedout above, there is no significant difference between themass distribution of DB white dwarfs and that of theDA white dwarfs, excluding those objects that have beenformed through binary evolution. This is what one wouldexpect for stars changing only surface compositions. Thisbeing said, while spectral evolution appears unavoidable,it cannot currently explain all the peculiarities of theabundance patterns observed in white dwarfs. Alternativeschemes, for example different channels feeding the

cooling sequence, should be considered in order to accountcompletely for the rich variety of spectral types.

White dwarf stars are not only unusual in thatthe abundance of their main atmospheric constituent(hydrogen or helium) may change in a complex way asa function of time, but they also show an amazing varietyof heavier trace elements in their atmospheres, arguablymaking them the most fascinating of all chemicallypeculiar stars. This is because the cooling phase ofwhite dwarfs, a relatively uneventful phase from anevolutionary point of view as discussed above, is, incontrast, a most active phase for the evolution of thechemical composition of the envelope. Indeed, it is nowwell established that the often puzzling variety of surfaceabundances observed in white dwarf stars can be tracedto the simultaneous operation, in the outer layers ofthese stars, of a variety of physical processes which willalso erase the abundances present in the photosphere atthe onset of cooling. As discussed above, downwardelement diffusion in the intense gravitational field of adegenerate star is perhaps the mechanism which is themost closely identified with white dwarf stars. However,in its presence, the observed abundances of variousatmospheric impurities, while small, are much too largeto be accounted for.

Mechanisms which compete with the downwardsettling must thus be called on to explain the presenceof these impurities in the atmospheres. At high effectivetemperatures, sayTeff > 20 000 K, the dominant competingmechanism is thought to be the selective radiative supportof elements in the atmosphere. This mechanism, whichinvolves the transfer of momentum from the intenseradiation field to ions of heavy elements such as carbon,nitrogen, silicon, iron and helium, is able to counteractthe downward gravitational force exerted on these ionsand allows a small, but measurable, amount of impuritiesto remain in the atmosphere. Unfortunately, detailedcalculations of this radiative support still do not reproducethe observed abundances, and it is currently thought thata small mass loss rate (of the order of ∼10−13M yr−1),when coupled to radiative levitation, might be able toaccount for the observed abundance patterns. For coolerstars, it is believed that ACCRETION from the interstellarmedium plays a role in accounting for the traces of heavyelements occasionally seen. Individual accretion events,probably related to encounters between a white dwarf anda small patch of neutral gas in the interstellar medium,might be able to account for the small fraction of DA andnon-DA stars which display impurities in their spectra.In the DQ stars, it is thought that the convection in thehelium envelope is deep enough to dredge up traces ofcarbon from the deeper carbon-rich layers. Much energyhas been expended, in the last two decades, to untanglethe relative importance of these competing mechanismsand to decipher the complex patterns of photosphericabundances observed in white dwarfs. Nevertheless,much work remains to be done.



Some areas of current interestIn the era of the Hubble Space Telescope and of giant,8–10 m ground-based telescopes, faint white dwarfpopulations are being routinely discovered and studiedin open and globular clusters. In addition, evidencehas been provided for the existence of a very old, faintpopulation of white dwarfs in the Galactic halo on thebasis of an interpretation of some MICROLENSING eventsand, more directly, on the basis of the observed propermotions of some very faint objects in the HUBBLE DEEP

FIELD whose positions in a color–magnitude diagram areconsistent with very cool (Teff ∼ 2000–3000 K) DA whitedwarfs. These exciting developments have led to arenewed interest in white dwarf cooling calculations andmodel atmosphere calculations using upgraded inputphysics and extending into the regime of very low effectivetemperatures. Such efforts are clearly worthwhile sincewhite dwarf physics holds the potential for providingthe best indicators of the ages of, and distances to, thesesystems.

The usefulness of white dwarfs as cosmochronome-ters was firmly established about a decade ago when itwas first demonstrated that the white dwarf populationin the solar neighborhood—a population characteristic ofthe Galactic disk—could be used to estimate the age ofthe disk to much improved accuracy. This method ap-plies because (intrinsically) faint white dwarfs cool downextremely slowly (see the isochrones in figure 3). This im-plies that, if white dwarf formation has been going on moreor less constantly over the distant past, many more faintwhite dwarfs than bright white dwarfs should be presentin a given volume of space. This is indeed what the distri-bution of observed white dwarfs in the Galactic disk gen-erally shows (figure 5). However, a very important andsignificant observational result of the last decade has beenthe realization that there is a significant decrease in the lu-minosity distribution of these stars: there is a real deficitof low-luminosity white dwarfs in the solar neighborhood.The simplest model to account for this observational fact isto assume that the oldest white dwarfs in the Galactic diskare still visible. In other words, the corpses of the very firstgeneration of intermediate-mass stars formed in the diskof the Milky Way have not yet had the time to cool to invis-ibility, beyond the reach of our telescopes. By comparingthe location of the observed low-luminosity decrease inthe white dwarf distribution with cooling calculations, itis possible to infer the age of the white dwarf population inthe Galactic disk (see figure 5 for an example of this). Whilethe method continues to be refined through numerous nu-merical simulations of evolving white dwarfs in the con-text of the Galactic disk, its potential for the white dwarfpopulations in open and globular clusters as well as forthe putative population in the Galactic halo is evident.

Another area of current active interest is related tothe presence of so-called instability strips along the cool-ing sequences of white dwarfs in the HERTZSPRUNG–RUSSELL

DIAGRAM. Helium-atmosphere white dwarfs become un-stable against nonradial gravity-mode pulsations as they

Figure 5. Luminosity function of white dwarfs in the solarneighborhood, that is the number of white dwarfs per unitvolume per unit luminosity interval as a function of luminosityin solar units. The observed values are indicated by error bars.Note, in particular, the deficit of low-luminosity stars shown inthe last bin. The solid curve is a fit to the data points based oncooling calculations which assume an age of 9.3× 109 yr for theGalactic disk.

evolve through an interval of effective temperatures from∼25 000 K to about ∼22 000 K. The instabilities are inti-mately connected to the recombination of helium in theenvelopes of these stars and the concomitant formation ofa superficial helium convection zone. There also exists ananalogous instability strip for the hydrogen-atmospherewhite dwarfs, related, this time, to the recombination ofhydrogen in the outer layers, and, consequently, whichis located at lower effective temperatures, in the range12 500–11 000 K. In both cases, the pulsational instabili-ties manifest themselves as temperature waves at the stel-lar surface that cause multiperiodic luminosity variations.The pulsating DB white dwarfs are referred to as V777 Herstars, while their DA counterparts are known as ZZ Cetivariables. The importance of these two instability stripsstems from the fact that they provide windows throughwhich the internal structure of white dwarfs can be probed.For instance, a detailed comparison of the observed periodstructure of pulsating white dwarfs with those of mod-els provides a unique way of inferring the internal consti-tution of white dwarfs and, in particular, the run of thechemical composition as a function of depth. Althoughthe potential of this technique has been barely tapped, itis likely that asteroseismological studies of white dwarfswill soon become a major contributor to our knowledge ofthe internal structure of these stars. One of the major po-tential impacts of these studies is the determination of thethicknesses of the outer layers of helium and hydrogen (in



the DAstars) in individual objects, which will serve as cali-bration of evolutionary models in more refined attempts atthe cosmochronology of various white dwarf populations.

BibliographyD’Antona F and Mazzitelli I 1990 Ann. Rev. Astron.

Astrophys. 28 139Liebert J 1980 Ann. Rev. Astron. Astrophys. 18 363Rudermann M 1971 Sci. Am. 224 (2) 24Van Horn H M 1979 Phys. Today 32 23Wesemael F, Greenstein J L, Liebert J, Lamontagne R,

Fontaine G, Bergeron P and Glaspey J W 1993 Publ.Astron. Soc. Pac. 105 761

Gilles Fontaine and Francois Wesemael


White Dwarfs in the Galaxy’s Halo ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS

The Galaxy’s large spherical halo (see GALACTIC METAL-POOR HALO and HALO, GALACTIC) may harbor as many asseveral hundred billion WHITE DWARFS, a population aslarge in number as the total number of stars in theGalaxy’s disk (see DISK GALAXIES and GALACTIC THIN DISK).Although this assertion is controversial, several astro-nomical surveys provide strong support for it and theimplications affect fields of astronomical inquiry asdiverse as dark matter and star formation. The reasonthat this population of white dwarfs may be related todark matter is tied to the fact that white dwarfs cool andfade as they age. This means that a white dwarf as old asthe ancient halo is extremely faint. Current large-scaleimaging surveys can only see relatively nearby whitedwarfs, and only a tiny fraction of the halo’s mass existsnear the Sun, while the vast majority of the Galaxy’s halois thousands of light-years away.

The theory of gravity requires that the mass of thehalo must dominate that of the Galaxy’s disk. Since thehalo is largely invisible, it is almost entirely made of darkmatter (see DARK MATTER IN GALAXIES). Indeed, the knownpopulation of normal stars in the halo constitutes a triv-ial fraction of the total mass. However, those stars exhib-it properties distinct from disk stars, including rapidmotion relative to the Sun. As a result a halo star thathappens to be within the solar neighborhood can be dis-tinguished from a disk star by examining the rapiditywith which it passes through the solar neighborhood.This is a statistical criterion, meaning that for any givenstar one can assign a probability that it belongs to thedisk or halo. For a survey of many stars with such prob-abilities one can estimate the actual number of stars inthe disk and halo populations.

The first hint that white dwarfs may be a significantpart of the halo emerged in 1996, when a group conduct-ing a MICROLENSING experiment claimed to detect indi-rectly a large population of objects in the Galaxy’s haloall with masses characteristic of white dwarfs. Instead ofactually seeing the white dwarfs themselves, theyobserved the effect they have on images of stars in theMilky Way’s companion galaxy the Large MagellanicCloud (see WIMPS AND MACHOS). The microlensing results,now backed by several independent, similar experi-ments, have found that at most 20% of the total mass ofthe Galaxy’s halo may be in this supposed population ofwhite dwarfs.

In 2001 a search for moving stars in an archive ofphotographic plates taken between 1950 and 1998revealed about 20 nearby white dwarfs which are mov-ing so fast that they must be members of the Galaxy’shalo. Statistical analysis of the data showed that at least1% of the halo’s mass is due to a population of white

dwarfs. Alternative interpretations of this survey suggestthat a fraction of this population may be part of theGalaxy’s thick disk. In all of the interpretations, howev-er, there still appears to be an unexpectedly large numberof white dwarfs in the halo, constituting at least 0.5% ofthe halo’s mass.

With 0.5–20% of the halo’s mass accounted for bywhite dwarfs, a fraction of the Galaxy’s dark matter hasfinally been identified. Furthermore, the halo whitedwarf population is important to STAR FORMATION theory.Ongoing star formation creates many low-mass stars forevery high-mass star. If the same proportions of high-and low-mass stars formed 13 billion years ago, when thehalo formed, only one to five halo white dwarfs shouldhave been found by the recent survey. This is knownbecause the low-mass halo stars have not had enoughtime to evolve into white dwarfs, and we know the num-ber of low-mass stars in the halo. That number, assuminga universal star formation process, determines the num-ber of high-mass stars that have evolved into whitedwarfs in the halo. The actual numbers observed implythat high-mass stars formed more readily in the halothan they do at present.

Ben R Oppenheimer

White Dwarfs in the Galaxy’s Halo


Whitford, Albert Edward (1905–) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Whitford, Albert Edward (1905–)Astronomer and photometrist, worked as a student withJOEL STEBBINS and succeeded him as professor at Wisconsinand director of the Washburn Observatory. Becamedirector of the Lick Observatory. As a student he madea successful device for measuring very small currentsfrom photoelectric cells, and used it to measure the lightfrom stars and galaxies. This turned into his career as anastronomer.


Whittaker, Edmund Taylor (1873–1956) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Whittaker, Edmund Taylor (1873–1956)Mathematician, born in Southport, Lancashire, becameAstronomer Royal of Ireland, director of DunsinkObservatory and professor of astronomy at the Universityof Dublin. Wrote the influential History of the Theories ofEther and Electricity, from the Age of Descartes to the Close ofthe Nineteenth Century (1910).


Whole Earth Telescope (WET) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Whole Earth Telescope (WET)A world-wide network of cooperating astronomicalobservatories which was established by astronomers fromthe University of Texas in 1986. WET HQ moved to IowaState University in 1997.

Intended to obtain uninterrupted time-series mea-surements of variable stars (white dwarfs and Delta Scutistars) and cataclysmic variables. This is done by resolv-ing the multiperiodic oscillations observed in these objectsinto their individual components. The temporal spectrumallows astronomers to probe the interiors of the target ob-jects using the technique of asteroseismology.

Since the first campaign in March 1988, the WET teamhas coordinated global photometry campaigns so that thetarget objects are visible from the night side of the planet24 hours a day. These campaigns typically take place twicea year.

For further information seehttp://ceti.as.utexas.edu/wetpage.html.


Widefield CCD Imagers E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Widefield CCD ImagersSince the early 1980s, the CHARGE-COUPLED DEVICE, or CCD,has emerged as the primary detector for astronomicalobservations. At the present time, CCDs are virtuallythe only detectors used for observations in the opticalportion of the spectrum (300 nm–1000 nm), and their usehas been extended into the UV (200 nm–300 nm) and x-ray(100 eV–10 keV) regions as well. Modern CCDs are superbdetectors. When thinned and backside illuminated,and with new multilayer anti-reflection coatings, CCDscan offer quantum efficiency (QE) exceeding 70% andapproaching unity over most of the optical waveband.The best CCD amplifiers now have readout noise below2e− rms and some have achieved the elusive 1e− noiselevel. One of the main areas of development has been,and continues to be, in device size and pixel count. Thevery first CCD imagers had 100×100 pixels, while the firstastronomically useful device was the RCA 320× 512 with30µm pixels, followed by the 800×800 15µm pixel devicebuilt by Texas Instruments for the WF/PC-I on the NASAHUBBLE SPACE TELESCOPE.As the 1990s began, the sizes of thesedevices began to grow. 2048 × 2048 pixel devices wereproduced by Ford Aerospace (which eventually becameLoral, Loral/Fairchild, and now Lockheed-Martin) andTektronix (now SITe). These 2K×2K pixel, and similar-sized, devices are now produced with good yields byseveral manufacturers.

The technical advances in ground-based instrumenta-tion have included increases in telescope apertures and im-provements in image quality and SEEING. It is not unusualto obtain 0′′.5 images or better at premier astronomicalsites. Consequently, optics and detector systems must bedesigned with sufficient resolution to sample these sharpimages, requiring that the detector pixels be no larger than∼0′′.25. This requirement for small pixels, coupled withthe need for large fields, drives the high pixel count neededin widefield imagers. For example, to be optimally sam-pled while spanning a 0.5 × 0.5 field requires that thearray measure of order 8000 × 8000 pixels. In the nearfuture, the image quality is likely to improve even more,driving the pixel sizes smaller and the pixel count higherin order to maintain the same field of view.

Monolithic devices are not suitable for such very largefocal planes. Limitations in wafer size and yield set apractical limit to device size. In addition, other factors suchas finite charge transfer efficiency (CTE) and readout time,as well as cost, must be considered when the number ofpixels increases. In order to build the very large detectorsneeded to image wide fields of view at seeing-limitedresolution, one must construct mosaics of smaller devices.

Large-gap CCD mosaic focal planesThe simplest approach to constructing a large mosaic focalplane is to mount large, packaged devices as tightly aspossible on a common base. The resulting mosaic willhave large gaps (1 cm or more) between the packagedCCD imagers that come from the wire bond pads that are

Figure 1. Drawing illustrating the layout of the SDSS CCDmosaic. The CCDs are arranged in six parallel columns eachhaving five CCDs. Each CCD in a ‘row’ is covered by a differentfilter (ugriz). The imaged scene ‘drifts’ along the columns andthe clocking of the charge is synchronized with the imagemotion. As a 2.5 strip of sky is scanned, images in five colorsare obtained from the camera. For this specialized application,the large gaps are not considered to be a serious problem.

required to make electrical connections to the silicon dieand the wire bond pins that are usually located around theperimeter of the typical device package.

Nevertheless, several widefield mosaic CCD imagershave been constructed using this technique, for examplethe 4K×4K pixel Big Throughput Camera constructedby the University of Michigan and Bell Labs, the4K×8K mosaic imager built by the National AstronomicalObservatory of Japan and the 10K×12K pixel camera builtby Princeton University for the SLOAN DIGITAL SKY SURVEY

(SDSS).The SDSS camera (figure 1) consists of a 5 × 6

array of SITe 2K×2K CCDs with 24 µm pixels and issomewhat special in that it is designed to operate in the‘drift scanning’ mode where the sky image is allowed toscan along the device columns while the accumulatingphotoelectron packets (or charge) are clocked to followthe image. This application differs from the conventional‘shift and stare’ technique where the telescope is pointedat some region, an image is taken (typically only a fewminutes integration), the telescope is shifted by an amountlarger than the gaps separating the CCDs, another imageis taken, and so on. The multiple images at differentpositions allow the gaps between the mosaic elements tobe filled while also enabling the removal of cosmic raysthat hit the device during the exposure.



Figure 2. This photomicrograph of the lower left-hand corner ofa three-edge buttable CCD illustrates several key design featuresfor edge buttability. This particular device (a 2K×4K LLCCID-20 with 15 µm pixels) can be operated in framestore mode(if desired) so there are six metal bus lines distributing thethree-phase parallel clocks to the imaging and storage regions.These can be seen as the vertical lines to the far left of thephotograph. The saw cut would be made at the edge of thisimage. The last three or four rows are tapered inward, and theserial register is wrapped around vertically so the outputamplifier does not increase the width of the device (photographcourtesy of Dr Barry Burke, MIT Lincoln Laboratory).

At one time, drift scanning may have offered somepractical advantages such as continuous readout with nodead time, and improved flat fielding in one dimension(the pixel-to-pixel sensitivity non-uniformities in a columnare effectively smoothed out by clocking the scene along acolumn). However, except for specialized projects like theSDSS, these advantages no longer exist. Modern mosaicscan be read out quickly (under 1 min), so the readouttime is only a small fraction of the typical integrationtime. Also, modern CCDs do not have the gross flat-field non-uniformities that plagued earlier devices, sothe column smoothing obtained by drift scanning isnot only unnecessary but viewed as a disadvantage.Furthermore, requiring a telescope and camera to operatein drift scanning mode places strict tolerances on theoptical system field distortion because the image trackscannot deviate from straight lines that must map ontodevice columns or else the image quality will deteriorate.Considering the advantages and limitations of bothmethods, the large majority of mosaic cameras have beendesigned to operate in the shift and stare mode.

Figure 3. Shown are four 2K×4K three-edge-buttable CCDswith 15 µm pixels from four different manufacturers. Thesedevices are the present-day building blocks for large,close-packed mosaics. From left to right are devices made byLoral (now Lockheed-Martin), MIT Lincoln Laboratory, SITe andEEV. Note that the EEV package has considerably less structurealong the wire-bonding edge making it the preferred design forfocal planes with more than two rows of devices.

Close-packed CCD mosaic focal planesFor most applications, a continuous focal plane withminimal gaps between the CCD imagers is preferred.While it is not possible to construct a mosaic withzero gaps, there are techniques in the CCD design andpackaging that allow for the construction of close-packedmosaics where the gaps are of order 0.5 mm or less.

Device design considerations for close-packed mosaicsTo enable CCD imagers to be used in close-packed mosaics,the devices must be specially designed from the outsetto be edge buttable. This involves confining the wirebonding pads to only two or, better yet, one of the fouredges of the device. Consider a three-edge buttable design.The only structures that need to run up the sides ofthe device are the metal clock bus lines that distributethe parallel clock signals to the polysilicon gates thatestablish the potential wells in the imaging region of thedevice. To keep the dead regions along sides of thedevice as small as possible, the output amplifier mustbe designed so it does not not protrude from the sideof the device. This can be accomplished by turning acorner with the serial register and tucking the amplifierinside the edge boundary defined by the outermost clocklines (see the photomicrograph in figure 2). Allowing thesmallest reasonable space for the vertical bus lines andthe boundary to the saw cut, the minimum gap from theimaging area to imaging area on two adjacent arrays is∼200 µm.

Currently, at least three manufacturers (EEV, SITe andMIT Lincoln Laboratory) are producing thinned, high-QE2K×4K devices with pixels of sizes in the 10–15 µm rangethat are three-edge buttable and suitable for building largemosaic focal planes (see figure 3).



Figure 4. This 8K×12K pixel CCD mosaic camera built by theUniversity of Hawaii for the 3.6 m Canada–France–HawaiiTelescope is the largest close-packed mosaic in operation at thetime of this writing (larger mosaic focal planes are underconstruction). The mosaic consists of 12 thinned, high-QE2K×4K three-edge buttable CCDs with 15 µm pixels. Thedevices were made by MIT Lincoln Laboratory. The imagingarea, corresponding to the black rectangular region in the centerof the photograph, measures 185 mm×122 mm. When used atthe prime focus of the CFHT, the mosaic spans a field of view of45′ × 30′ at a scale of 0′′.21/pixel. The gaps separating CCDs areless than 0.5 mm or less than 10′′ on the sky.

Packaging and mosaic constructionOnce an edge-buttable device is designed, it can bemounted close to its neighbors to form a CCD mosaic.The very first attempts at this involved permanentlyattaching (i.e. gluing) the CCD dies to a common substrate.There are several disadvantages to this approach. Amosaic assembled in this manner cannot be taken apartsafely if one of the CCDs is damaged. Furthermore,the selection of the CCDs that will be used in a mosaicinvolves thorough testing at cryogenic temperatures. Thisis difficult to accomplish with unpackaged CCD dies, soone could permanently assemble a mosaic and find after itis completed and tested that some of the mosaic elementsare not suitable for scientific observations. These problemsled to the development of edge-buttable packages that donot compromise the tight packing ability of the buttable

imagers and yet allow a mosaic focal plane to be assembledand dis-assembled. Examples of such packages are shownin figures 3 and 4. In choosing a package material, onemust consider that the device will be operated at ∼170 K,so the coefficient of thermal expansion difference betweenthat material and silicon is important. The standardpackage materials that are a reasonable thermal expansionmatch to silicon are aluminum nitride (AlN), Invar andmolybdenum.

Using the devices shown in figure 3, several close-packed CCD mosaic focal planes containing at least8K×8K pixels have been constructed and are in operation(e.g. UH 8K×8K, NOAO 8K×8K mosaic imagers for theKitt Peak and CTIO 4m telescopes, ESO 8K×8K, CFHT8K×12K—see figure 4).

Devices that are three-edge buttable are suitable forclose-packed mosaics having two rows and an unlimitednumber of columns. If one wants to build a larger mosaicwith more than two rows, one must tolerate a larger gapbetween the rows on the fourth, non-buttable edge. Withthese larger mosaics in mind, some designs have beendeveloped for pseudo four-edge buttable packages thatminimize the packaging structures along this fourth edge.EEV has developed a package (see figure 3) that keeps thedead space along this fourth edge to under 3 mm fromthe edge of the bottom imaging row to the edge of thepackage. Another prototype design with a 3 mm deadspace is shown in figure 5.

The futureWidefield CCD imagers are at present being constructedby many observatories. Several current projects aredesigned to span fields in excess of 1 × 1 using CCDmosaics measuring 18K×18K pixels. If widefield imagingtelescopes with 3 × 3 fields are constructed, they willrequire CCD mosaics having 36K×36K pixels or more.Work is also progressing on the designs for the individualbuilding blocks that make up the mosaic elements. Three-edge buttable devices with 3K×6K 10 µm pixels are nowunder development, and wafer-scale devices that can filla 150 mm wafer are being considered as yields improve.

On-chip image motion compensation—the OTCCDOne exciting CCD variation that has emerged in recentyears is the multi-directional or orthogonal-transfer CCD(OTCCD)—a device that can shift charge in up toeight directions, thus allowing active image motioncompensation or fast guiding ‘on chip’ (see figure 6). Amosaic array of close-packed, independently addressableOTCCDs would allow compensation for image motioncaused by the atmosphere over very wide fields, as well ascorrecting for windshake and other mechanical pointinginstabilities common to the telescope as a whole.

Large, buttable OTCCDs with 2K×4K pixels havebeen successfully fabricated and the first mosaics of suchdevices are at present under construction. Even largerdevices subdivided into small (∼1′ × 1′) independentlycontrollable OTCCD cells are under development.



Figure 5. This drawing shows one design for a package that can be used in close-packed mosaics having more than two rows. Aceramic piece with metallized traces that wrap around one edge is mounted at right-angles to the device. Wire bonds are made to thetraces on the short edge of the ceramic and the traces connect the signals to a surface mount connector that emerges from the backsideof the package. Mounting holes and alignment holes are shown for locating the package in focal plane mosaic.

1

3

24

Channel stop

Figure 6. This diagram illustrates the gate layout and charge clocking directions for a multi-directional OTCCD. The symmetry of thegate structure allows clocking of the charge in eight directions: horizontally and vertically (as shown), or diagonally. Subpixelstepping is also possible. In this image, the triangular structures are polysilicon gates that can be biased to form the potential wells inthe silicon below. This type of structure requires four gates per pixel, whereas the typical CCD has only three gates. In the illustrationat the left, the dark triangular gate is biased low to serve as a barrier phase while the other three phases are modulated to transfer thecharge in the vertical direction. At the right, a different gate serves as the barrier phase, and, by modulating the voltage on threedifferent gates, the charge can be transferred in the horizontal direction.

Bright stars and antibloomingBright stars present a serious problem for widefieldCCD imaging. Even at high GALACTIC LATITUDE, there arenumerous bright (m < 15) and several very bright (m < 10)stars in any 0.5 × 0.5 field. Larger fields only makethe problem worse. Light from these stars can reflectoff optical surfaces, including the surface of the CCDitself, and create out-of-focus halos and other scatteredlight artifacts. Large diffraction spikes can extend wellaway from the bright stars and break up into faint pieces,resembling faint galaxies, especially in very good seeing.Bright stars will also produce saturated, bloomed chargetrails; the result of filling the pixel potential well capacity(typically 150 000–300 000 e−) and having the excess chargespill up and down the columns into the neighboringpixels. For some brighter stars, the bloomed trails canextend the full length of the device, destroying anyinformation about objects in the path of the bloomedtrail. Device manufacturers can address the bloomingproblem by implementing antiblooming drains in theirCCD structures. An antiblooming drain is an implant onthe CCD with a potential level near full well, but slightly

less than the barrier potential of the neighboring verticalphases. When the charge in a pixel exceeds the potentiallevel of the drain, but before the charge can spill over tothe neighboring pixel, it is intercepted by the drain. Suchstructures can be made so that they do not obscure anyof the imaging regions (e.g. the drain can be run downthe center of the channel stop). Implementation of suchstructures will be very valuable in future devices designedfor widefield imaging.

Gerard A Luppino


Widmanstatten, Alois von Beckh- [Alois Beck, Edlervon Widmanstatten] (1753/4–1849)


Widmanstatten, Alois von Beckh-[Alois Beck, Edler vonWidmanstatten] (1753/4–1849)Printer and businessman, born in Graz, Austria, becamehead of the Fabriksproduktenkabinett, a private technol-ogy collection of Emperor Francis I. Discovered the crys-talline structure of iron/nickel meteorites by etching pol-ished slices from an iron meteorite from Zagreb and print-ing from the etched surfaces. The patterns are known asWidmanstatten figures.


Widmanstatten Pattern E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Widmanstatten PatternA characteristic, roughly hexagonal pattern of intersectinglines that appears on the surface of an octahedrite, a type ofiron meteorite, when it is sectioned, polished and etchedwith acid. The Austrian mineralogist Aloys Joseph vonWidmanstatten discovered the pattern in 1804. It is formedby the intergrowth of two nickel–iron alloys under theconditions of slow cooling that pertained in the solidifyingcore of an asteroidal parent body that had undergonedifferentiation, and is found only in meteorites. The twoalloys are kamacite, with a low nickel content, and taenite,which is richer in nickel.


Wien, Wilhelm (1864–1928) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wien, Wilhelm (1864–1928)Born in Gaffken, East Prussia (now Poland), Nobelprizewinner (1911), became professor of physics at Munichand discovered Wien’s law, for the distribution of lightin the spectrum of a black-body. Discovered the protonin an early mass-spectrometer experiment (confirmed byRutherford).


Wilcox Solar Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wilcox Solar ObservatoryThe Wilcox Solar Observatory (WSO) at Stanford Univer-sity measures the Sun’s large-scale synoptic magnetic andvelocity fields with the goal of understanding solar vari-ability and how it affects our terrestrial environment. Withmore than a 22 year solar cycle of spectrograph observa-tions since 1975, WSO staff investigate the solar interior,photosphere, corona, wind and cycle.

The Observatory was first funded by the Office ofNaval Research (ONR), the National Science Foundation(NSF) and the M C Fleischman Foundation. Continuingsupport comes from NSF, NASA and ONR. TheObservatory was rededicated in honor of its first director,J M Wilcox, in 1984, shortly after his death. StanfordUniversity is located south of San Francisco, CA.

For further information seehttp://wso.stanford.edu.


Wilkins, John (1614–72) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wilkins, John (1614–72)Churchman, born in Fawsley, Northamptonshire, becameWarden of Wadham College, Oxford, and Master of TrinityCollege, Cambridge. Founded the Royal Society from adiscussion group of scientists at Wadham. In 1638 Wilkinswrote a book describing the Moon as a habitable planet andpredicting that, one day, space travel to the Moon wouldbe possible.


Wilson, Alexander (1714–86) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wilson, Alexander (1714–86)Born in St Andrews, Scotland, became professor atGlasgow, observed sunspots and showed that theywere depressions in the Sun (following LA HIRE andCASSINI). Published Thoughts on General Gravitation (1770),answering NEWTON’s question ‘What hinders the fixed starsfrom falling upon one another?’ with the speculativeanswer that the entire universe rotates about a center.


Wilson, Olin Chaddock (1909–94) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wilson, Olin Chaddock (1909–94)Astronomer and spectroscopist, became a staff memberat Mount Wilson Observatory. He studied stellarchromospheres and stellar activity cycles, showing byintensive analysis of the H and K lines of ionized calciumthat other stars besides the Sun have cycles of activity.With M K VAINU BAPPU, he found a means of determiningluminosity, and thus distance, of stars from the widthsof the emission in these two lines that comes from thechromosphere (Wilson–Bappu effect). He studied spectra ofnebulae, eclipsing stars, Wolf–Rayet stars and planetarynebulae.


Wilson, Robert Woodrow (1936–) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wilson, Robert Woodrow (1936–)Born in Houston, Texas, Nobel prizewinner for physicsin 1978) with ARNO PENZIAS ‘for their discovery of cosmicmicrowave background radiation’. Interested in radioas a boy, drawn to radioastronomy by working withJOHN BOLTON at CalTech mapping the Milky Way. JoinedBell Laboratories at Crawford Hill where with ArnoPenzias he shared a small allowance given to the Labfor radioastronomy projects. With new millimeter wavereceivers at 100–120 GHz they discovered unexpectedlylarge amounts of carbon monoxide in a molecular cloudbehind the Orion nebula, including isotopic spectral linesso that it was possible to determine isotope ratios as aprobe of nucleogenesis. With a large radio telescope (theHolmdel horn) and a new sensitive, low-noise receiverdiscovered the cosmic microwave background radiation.


WIMPs and MACHOs ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS

WIMP is an acronym for weakly interacting massive par-ticle and MACHO is an acronym for massive (astrophys-ical) compact halo object. WIMPs and MACHOs are twoof the most popular DARK MATTER candidates. They repre-sent two very different but reasonable possibilities ofwhat the dominant component of the universe may be.

It is well established that somewhere between 90%and 99% of the material in the universe is in some as yetundiscovered form. This material is the gravitationalglue that holds together galaxies and clusters of galaxiesand plays an important role in the history and fate of theuniverse. Yet this material has not been directly detected.Since extensive searches have been done, this means thatthis mysterious material must not emit or absorb appre-ciable electromagnetic radiation in any known wave-band. Thus it is called dark matter. The nature of thismaterial is one of the biggest unsolved problems in sci-ence. It is important to identify the dark matter, but sinceit is easy to invent substances that could fill intergalacticspace and yet have escaped detection, there are a verylarge number of dark matter candidates. Thus severalgeneric classes of dark matter candidates have been sug-gested, and each dark matter search experiment concen-trates on one of these classes. The most important darkmatter classes, in terms of detection efforts, are neutri-nos, WIMPs, MACHOs and axions.

For example, NEUTRINOS are weakly interacting par-ticles that were almost certainly created in great abun-dance during the big bang. These fill the Galaxy, movingfreely through the Galaxy and even the Earth, and yet arealmost impossible to detect. This is because they can besensed only through the very small electroweak interac-tion. If each neutrino had a mass of several electronvoltsthey would contribute enough mass to make up the bulkof the dark matter. For various reasons, it is unlikely thatneutrinos of the type discovered in particle acceleratorsand nuclear reactors on Earth make up much of the darkmatter.

However, it is very possible that some as yet undis-covered weakly interacting particle was created duringthe big bang and today remains in large enough abun-dance to make the dark matter. The masses typicallyrequired for these particles are in the range 1 GeV–1 TeV,and these hypothetical dark matter particles are calledWIMPs. There are hundreds of elementary particles thatfall into this class of dark matter particles, includingsupersymmetric particles such as neutralinos, photinos,higgsinos or sneutrinos, and new heavy neutrinos.

This hypothetical WIMP is well studied andattempts to detect these particles have been mountedboth by creating them in accelerators and by sensingthem in underground detectors as they pass through theEarth. However, there is a large class of astronomical

objects that could be the dark matter and still escapedetection. For example, if the Galactic halo were filledwith Jupiter mass objects (10–3Mo.) they would not havebeen detected by emission or absorption of light. Browndwarf stars with masses below 0.08Mo. or the black holeremnants of an early generation of stars would be simi-larly invisible. Thus these objects are examples ofMACHOs. Other examples of this class of dark mattercandidates include primordial black holes created duringthe big bang, neutron stars, white dwarf stars and vari-ous exotic stable configurations of quantum fields, suchas non-topological solitons.

An important difference between WIMPs andMACHOs is that WIMPs are non-baryonic andMACHOS are typically (but not always) formed frombaryonic material. As discussed in the article on big bangnucleosynthesis (see UNIVERSE: THERMAL HISTORY), baryon-ic material probably cannot make up all of the dark mat-ter, although it could make up most of the dark matter inthe halos of spiral galaxies such as the Milky Way. Thereis preliminary, although controversial, evidence for theexistence of large numbers of MACHOs, but becausethey probably cannot make up all the dark matter, thesearch for WIMPs continues unabated.

WIMP thermal relics as dark matterAmong the particle dark matter candidates an importantdistinction is whether the particles were created thermal-ly in the early universe, or whether they were creatednon-thermally in a phase transition. Thermal and non-thermal relics have a different relationship between theirrelic abundance Ω and their properties such as mass andcouplings, so the distinction is especially important fordark matter detection efforts. For example, the WIMPclass of particles can be defined as those particles that arecreated thermally, while dark matter axions come mostlyfrom non-thermal processes. Light neutrinos are alsothermally created relics, but because of their very smallmass have a different history.

In thermal creation one supposes that early on,when the universe was at very high temperature, ther-mal equilibrium obtained, and the number density ofWIMPs (or any other particle species) was roughly equalto the number density of photons (particles of light)1.This is just equipartition of energy among all possibledegrees of freedom. As the universe cooled the numberdensity of WIMPs and photons decreased together.When the temperature finally dropped below the WIMPmass, however, creation of WIMPs became very rarewhile annihilation still proceeded. Thus in equilibrium,the number density of WIMPs dropped exponentially :exp(–mWIMP/T). If equilibrium were maintained until

WIMPs and MACHOs


1 There is a small difference between the number density of photons andneutrinos since photons obey Bose–Einstein statistics and neutrinos obeyFermi–Dirac statistics.


today there would therefore be very few WIMPs left, butat some point the WIMP density dropped low enoughthat the probability of one WIMP finding another to anni-hilate became small. Note, we must assume that an indi-vidual WIMP is stable if it is to become the dark matter.The WIMP number density stopped dropping at thispoint and we are left with a substantial number ofWIMPs today. Detailed evolution of the Boltzmann equa-tion that describes this process can be done for an accu-rate prediction, but roughly the predicted relic densitytoday of WIMPs is inversely proportional to their inter-action strength. The remarkable fact is that, for a relicdensity equal to the known dark matter density, theinteraction strength must be that expected for particleswith electroweak-scale interactions: thus the ‘W’ for‘weakly’ in ‘WIMP’. There are several theoretical prob-lems with the standard model of particle physics that aresolved by new electroweak-scale physics such as SUPER-SYMMETRY. Thus these theoretical problems may be cluesthat the dark matter does indeed consist of WIMPs. Saidanother way, any stable particle that annihilates with anelectroweak-scale cross section is bound to contribute tothe dark matter of the universe. It is interesting that the-ories such as supersymmetry, invented for entirely dif-ferent reasons, typically predict just such a particle.

The fact that thermally created dark matter hasweak-scale interactions also means that it may be withinreach of accelerator experiments such as LEP and LHC atCERN, and the proton collider experiments at Fermilab.Thus many accelerator searches for exotic particles arealso searches for the dark matter of the universe. Also,because of the weak-scale interactions, WIMP–nuclearinteraction rates are within reach of many direct andindirect detection methods, as discussed below.

Supersymmetry and dark matterSupersymmetry is a new hypothetical symmetry ofnature that relates bosons and fermions. If supersymme-try exists in nature then every known particle shouldhave a supersymmetric partner. Bosonic ordinary parti-cles have fermonic superpartners with the same nameexcept with the suffix ‘ino’ added, while fermonic ordi-nary particles have bosonic (scalar) superpartner nameswith the prefix ‘s’ added. Examples of proposed super-symmetric particles include photinos, higgsinos, Z-inos,squarks and selectrons. Some supersymmetric particleshave the same quantum numbers as each other andtherefore can mix together producing particles that arenot exact partners of any standard model particle. Forexample, the photino, Higgsino and Z-ino can mix intoarbitrary combinations called the neutralinos.

In most models, the lightest supersymmetric particle(LSP) is stable, and since supersymmetric particles haveelectroweak-strength interactions, the LSP makes anexcellent dark matter candidate. Typically the neutralino

is the LSP so most investigations of WIMP dark matterhave concentrated on the neutralino. However, there aremany possible supersymmetric models and many freeparameters in the models, so precise predictions ofsupersymmetric relic abundance and supersymmetricparticle detectability are not possible. Typically experi-ments attempt to probe a range of model parameters. Sofar no concrete evidence of any supersymmetric partnerexists. If even one supersymmetric partner is found, thetheory predicts that they all must exist.

Note that the parameters that determine the relicabundances also determine all the particle productionand rare decay cross sections, as well as the rate in vari-ous detectors. Thus once these parameters are specifiedor measured, one can compare the model predictionswith experimental results.

Search for WIMPsAccelerator searchesExtensive unsuccessful searches for the particlesinvolved in supersymmetric models have been per-formed at particle accelerators throughout the world.Thus substantial regions of prime neutralino dark matterparameter space have already been eliminated. This doesnot yet mean that low-energy supersymmetry is unlikelyto exist, since only a small portion of the allowed massrange under 1 TeV has been explored. Since supersym-metry predicts a Higgs boson with mass under about 120GeV, such a discovery would be very important, espe-cially if the Higgs boson showed non-standard proper-ties indicative of supersymmetry. It is correct to think ofthe particle physics search for supersymmetry as a pow-erful search for the dark matter.

Direct detection of WIMPsA satisfying solution to the dark matter problem wouldbe the detection of WIMPs from our Galactic halo as theymove past and through the Earth. This would also allowmeasurement of the local density of dark matter andestablish beyond doubt that the dark matter is non-bary-onic cold dark matter. There are several ways to do this,and currently two methods are being aggressively pur-sued.

The most exciting result would be direct detection ofthe WIMP particles in the laboratory. Since we roughlyknow the speed (~220 km s–1) and the density (ρ~0.3 pro-ton masses cm–3), we can say that for a WIMP of mass oforder 10–100 GeV, roughly 100 000 dark matter particlesper second pass through every square centimeter of theEarth. However, if WIMPs exist, they are very weaklyinteracting particles, so it is quite rare that one of themwill interact at all; most of them pass right through theEarth unimpeded. In addition, if a WIMP does elastic-ally scatter off a nucleus, the deposited energy is usuallyin the keV to 100 keV range, too small to be noticed



except by exquisitely sensitive equipment. These difficul-ties, however, have not stopped many groups through-out the world from developing devices capable of detect-ing WIMPs. The detection rates turn out to be within andjust beyond the reach of current experimental efforts.

The basic idea is to detect the small energy deposit-ed when a WIMP scatters off a nucleus in some well-instrumented piece of material. When a WIMP scattersoff a nucleus in a crystal, the nucleus recoils, causing dis-location in the crystal structure, vibrations of the crystallattice (i.e. phonons or heat) and also ionization. Thesesignals can be detected. For example, some experimentsuse kilogram size germanium and silicon crystals andattempt to detect the ionization and phonon signals.Other groups use sodium iodide crystals and look for thescintillation light caused by the ionized electrons orsearch for crystal dislocations in samples of billion yearold mica. Another possibility is to record the recoil of anionized atom in a gas using drift chamber techniques.

The main difficulties in these experiments comefrom the fact that the WIMP events are rare and thatthere are many backgrounds that deposit similaramounts of energy on much more frequent time-scales.Thus the experiments operate deep underground, whereionizing COSMIC RAYS are less frequent, and typicallyoperate their detectors at extremely cold temperatures tokeep thermal excitations low. Also many types of shield-ing, as well as redundant detection methods, are nowbecoming standard. Even so, these are difficult experi-ments and tiny amounts of radioactivity in the detectoror shielding can swamp the expected signal. With effort,a background rate of under one event per kilogram ofdetector per day can be achieved. The expected signal ishighly dependent on the supersymmetry model, but typ-ically is in the range from 10–5–10 events kg–1 day–1. Theevents can be separated from the background in twoways. In some detectors the background (non-WIMP)interactions can be recognized and simply ignored. In thelarger detectors this is not possible, so they use the factthat the WIMP event rate is predicted to be larger in Junethan in December. This annual modulation in event rateis caused by the Earth’s orbit either being aligned withthe Sun’s motion in the Galaxy (in June) or anti-aligned(in December).

The current generation of detectors have detectionthresholds of around 1 event kg–1 day–1, with hopes thatwithin the next few years signals as small as 102 eventskg–1 day–1 will be detectable. Thus there is a reasonablechance that dark matter neutralinos will be detected bythis type of direct detection within the next few years. It is also clear, however, that there are many values of the supersymmetry parameters that predict detectionrates of below the 10–2 events kg–1 day–1 threshold, andso would not be detectable in the near future by thesemethods.

Indirect detection of WIMPsA great deal of theoretical and experimental effort hasgone into another potential technique for WIMP detec-tion. The idea is that if the halo is made of WIMPs, thenthese WIMPs will have been passing through the Earthand Sun for several billion years. Since WIMPs will occa-sionally elastically scatter off nuclei in the Sun or Earth,they will occasionally lose enough energy, or changetheir direction of motion enough, to become gravitation-ally captured by the Sun or Earth. The orbits of such cap-tured WIMPs will repeatedly intersect the Sun (or Earth)resulting in the eventual settling of the WIMPs into thecore. As the number density increases over time, the self-annihilation rate will increase. Since ordinary neutrinoscan result from WIMP self-annihilation, one predicts astream of neutrinos coming from the core of the Sun orEarth. Neutrinos easily escape the Solar core and detec-tors on Earth capable of detecting neutrinos coming fromSun or Earth have operated for some time. The energy ofsuch neutrinos is roughly 1/2 to 1/3 of the WIMP mass,so these neutrinos are much higher energy than the MeVscale solar neutrinos from nuclear reactions that havealready been detected. The higher energy of these WIMPannihilation neutrinos makes them easier to detect thanordinary solar neutrinos and somewhat compensates fortheir much fewer numbers. It also makes them impossi-ble to confuse with ordinary solar neutrinos. Thus thepresence of a source of high-energy neutrinos emanatingfrom the centers of the Sun and Earth would be taken asevidence for WIMP dark matter.

While the above chain of reasoning may seem long,it appears to be robust, and several experimental groupsare in the process of designing and building detectorscapable of seeing such a neutrino signal. For this signal,it is not the mass of the detector that is relevant, but thesurface area. Neutrinos from the core of the Sun or Earthproduce muons in the atmosphere and rock around thedetectors, and it is primarily these muons that the detec-tors watch for. Muons are also copiously created by cos-mic rays entering the Earth’s atmosphere, so there is asubstantial background of ‘downward’ traveling muons.These detectors, then, are located deep underground,where the rock shields many of the background muons,and they also focus on ‘upward’ traveling muons, thatare much more likely to have been created by neutrinosthat have traveled through Earth and interacted in therock just below the detector. Thus surprisingly, the bestway to see high-energy neutrinos from the Sun is to godeep underground at night (when the Sun is ‘under’ theEarth).

The new generation of detectors are designed tohave very large surface areas. A comparison of direct andindirect detection methods indicates that for a typicalneutralino a kilogram of direct detector germanium hasabout the same sensitivity as 104–106 m2 of indirect



detector. The new generation of detectors should haveareas in this range and should be able to start to proberealistic supersymmetry models, but again viable super-symmetric models exist which predict rates too small forthese detectors to measure, and a definitive test of theWIMP hypothesis does not seem possible in the nearfuture.

MACHOs and microlensingAn exciting development in the dark matter story is thedetection of MACHOs by three separate groups,MACHO, EROS and OGLE. All three groups monitoredmillions of stars in the LARGE MAGELLANIC CLOUD (LMC),in the SMALL MAGELLANIC CLOUD (SMC) or in the GALACTIC

BULGE, for signs of gravitational microlensing, and allthree groups have found it. It has now become clear thatthese objects constitute some new component of theMilky Way, but whether they constitute a substantial partof the dark matter or new stellar components of theGalaxy or LMC is not clear.

MICROLENSING is a powerful new tool for discoveringand characterizing populations of dark objects in ourGalaxy, and the current experiments may have the capa-bility to give a definitive answer to the question ofwhether the dark matter in our Galaxy is baryonic. Thereason is that the microlensing searches are probably sen-sitive to any objects in the range 10–8Mo.<m<103Mo., justthe range in which such objects are theoretically allowedto exist. Objects made purely of H and He with massesless than ~(10–9–10–7Mo.) are expected to evaporate owingto the microwave background in less than present age ofthe universe, while dark matter objects with massesgreater than ~103Mo. would have disrupted known GLOB-ULAR CLUSTERS. So unlike the searches for WIMP darkmatter, which if they detect nothing will remain incon-clusive, if the MACHO searches find nothing, we wouldat least know what the dark matter is not. However, theMACHO searches have found something, as we describebelow.

Microlensing is also described in the article on GRAV-ITATIONAL LENSING. The idea of microlensing rests onEinstein’s observation that, if a massive object lies direct-ly on the line of sight to a much more distant star, thelight from the star will be lensed and form a ring aroundthe lens. The ring is called the ‘Einstein ring’ and it setsthe scale for all the microlensing searches. In the lensplane, the radius of the ring is given by

where Ro. and Mo. are the solar radius and mass, m is theMACHO mass, L is the distance to the star being moni-tored and x is the distance to the MACHO divided by L.The formation of a ring is very unlikely, but even with

imperfect alignment two images result and a large mag-nification can occur.

Since the MACHO, Earth and source star are all inrelative motion, the star appears to brighten, reaches apeak brightness, and then fades back to its usual magni-tude. Thus the signature for a microlensing event is atime-symmetric brightening of a star occurring as aMACHO passes close to the line of sight. When amicrolensing event is detected, one fits the lightcurveand extracts the peak magnification Amax, the time of thepeak, t0, and event duration t^. The primary physicalinformation comes from t^, which depends on theMACHO velocity, the MACHO mass, the source dis-tance, and the lens distance. The source distance can bedetermined since it is visible, but unfortunately, one can-not determine the other three physical parameters from t^.However, statistically, one can use information about thehalo density and velocity distribution, along with thedistribution of measured event durations to gain infor-mation about the MACHO masses. Using a standardmodel of the dark halo, MACHOs of Jupiter mass(10–3Mo.) typically cause events lasting 3 days, whilebrown dwarf mass MACHOs (0.1Mo.) cause events last-ing about a month.

Assuming a halo made entirely of MACHOs, theprobability of any MACHO crossing in front of a star isabout 5×10–7. Thus many millions of stars must be mon-itored in order to see a handful of microlensing events. Inaddition, if one wants to see microlensing from objects inthe dark halo, the monitored stars must be far enoughaway so that there is a lot of halo material between usand the stars. Therefore, the best stars to monitor arethose in the LMC and SMC at distances of 50 kpc and 60kpc respectively, stars in the galactic bulge at 8 kpc andstars in nearby galaxies such as M31 at 750 kpc.

Microlensing experimentsThere are several experimental groups that have under-taken the search for microlensing and have returnedresults. All together about a dozen events have beendetected towards the LMC, a couple towards the SMC,and more than 400 towards the galactic bulge. For detec-tion of dark matter MACHOs, it is primarily the LMCevents that are relevant. All survey collaborations oper-ate a medium-size telescope and monitor millions ofstars nightly. Since the fields are crowded, each CCDframe contains hundreds of thousands of stellar images,the brightness of each which must be determined bycomputerized photometry. These stellar brightnesses arearranged sequentially in a lightcurve, each of which isfinally searched for microlensing-like bumps. All togeth-er many terabytes of data have been analyzed by the sur-vey experiments.

Most of the monitored stars are constant brightnessas one expects, but about one-half of 1% are variable.



These are mostly identified as variable stars of knowntypes. Several signatures of microlensing, including theunique shape of a microlensing lightcurve are used topick out microlensing events from this background ofvariable stars. For example, the MACHO collaborationanalyzed about 9.5 million lightcurves from its 2 yr LMCdata set and found six to eight microlensing events.

Experimental resultsIn order to interpret microlensing events, the efficiencywith which an experiment can detect microlensing isneeded. Bad weather, bad seeing, crowded stars, etccause microlensing events to be missed by the experi-ment. The MACHO collaboration finds an efficiency ofaround 30% for microlensing events that last 50 days. Forevents lasting less than a few days and events lastinglonger than 6 months the efficiency is very low.

Using the efficiency and a model of the dark halo,the number of microlensing events one expects to see ifthe halo consisted entirely of MACHOs is found to beabout a dozen for the MACHO collaboration 2 yr dataset. This quantity is related to the lensing ‘optical depth’,the roughly 5×10–7 probability that at any time a sourcestar is lensed (if the halo is made entirely of MACHOs).

Using the eight observed MACHO collaborationLMC events, the observed optical depth is ~(2.5±1×10–7),roughly half the value if the halo consisted entirely ofMACHOs. A careful likelihood analysis of these eventsgives, for a standard dark halo model, a most likelyMACHO halo fraction of 0.5±0.3 and a most likely massof the MACHOs of around 0.5Mo.. This result depends onthe assumption that the events are due to lenses in theGalactic halo and on the model of the galactic halo used.

Another analysis based on noticing that none of theeight detected events had durations of less than 20 dayscan rule out low-mass MACHOs. This is because t^:m1/2,and no short-duration events have been observedtoward the LMC. This analysis gives the strongest con-straints to date on the baryonic content of the dark halo.The EROS and MACHO collaboration limits show thatobjects in the mass range from 10–7Mo. to 10–3Mo. cannotmake up the entire dark halo. Objects in the range from3.5–10–7Mo. to 4.5–10–5Mo. make up less than 10% of thedark halo. Thus we now know that the dark matter is notmostly objects of Earth mass, or Jupiter mass, or anycombination thereof. The only compact baryonic darkmatter candidates left are objects in the brown dwarf andhigher mass range. This result is independent of anyassumptions about the observed microlensing events,but does depend on the model of the dark halo.

In 2002, an international team of astronomersobserved a dark matter object directly for the first time.Images and spectra of a MACHO microlens were takenby the NASA/ESA HUBBLE SPACE TELESCOPE (HST) and theEuropean Southern Observatory’s Very Large Telescope

(VLT). This result is a strong confirmation that gravita-tional microlensing is the cause of the events detected byteams searching for MACHO dark matter.

In addition, Christopher Kochanek of theHarvard–Smithsonian Centre for Astrophysics inCambridge, Massachusetts and Neal Dalal of theUniversity of California, San Diego have used radio tele-scopes and gravitational lensing to search for cold darkmatter. They have studied seven galaxies, each magni-fied by four nearer ones. Because each lensing galaxy isin a slightly different position, the researchers got fourdifferent images of each of the seven distant galaxies. Thefour images should have been identical. But each is actu-ally slightly different. The difference was enough to havebeen caused by the kind of clumps of dark matter aroundlensing galaxies that mathematical models predict.

Interpretation of results: dark matter or not?The naive interpretation of the microlensing results isthat between 20% and 80% of the dark matter in theMilky Way has been identified. However, the resultthat the mass of the objects is above the brown dwarflimit of 0.1Mo. is surprising. Main sequence stars withmasses above 0.1Mo. would have been seen and there-fore cannot be the dark matter. Several interpretationsare possible.

First, perhaps the MACHOs are white dwarf stars orneutron stars. These are dark remnants of an earlier gen-eration of stars, but it is problematic to have enough ofthese around to be the dark matter and not have detect-ed the other byproducts of such an early stellar popula-tion. Second, perhaps the model of the Galactic halo usedis incorrect, and the masses of the MACHOs are actuallysafely below the brown dwarf limit. However, most rea-sonable halo models investigated do not have this prop-erty. Next, perhaps MACHOs are primordial black holes,or other exotic objects not currently known. This is pos-sible, but quite surprising. Most importantly, perhaps themicrolensing events are not due to halo lenses, and there-fore are not telling us about the dark matter. In a typicalmicrolensing event, the distance of the MACHO is notdetermined, so it is not known where the lens populationis located. The estimate of the amount of MACHO darkmatter relies on an assumed distribution of lens material,and therefore on the model of the Galaxy and LMC.

It has been suggested that the lenses could be faintstars in the LMC itself, or in some small undiscovereddwarf galaxy between the Sun and the LMC. These pos-sibilities are being vigorously pursued, but strong argu-ments have been given against both possibilities. Thusthe outcome is very unclear at the moment.

If one could measure the distance to the MACHO,that would be enough to distinguish between the abovepossibilities. Thus a ‘microlensing parallax’ satellite hasbeen proposed to measure these distances. Other ways to



measure the distance to lenses, and therefore determinewhether or not they are part of the dark halo, includeground-based lensing parallax, binary lens caustic cross-ing, and the finite source star effect. Unfortunately,events where these effects can be measured are rare, soluck, or new larger surveys and better follow-up, is need-ed to resolve this issue of what the MACHOs are.

The next generation of microlensing surveys and fol-low-up efforts are underway or being planned andshould determine some lens distances, as well as gathermore events. Thus, while the question of baryonic darkmatter remains open, the next few years should bring ananswer.

BibliographyDalal N and Kochanek C S (in press) 2002 Direct detec-

tion of CDM substructure Astrophys. J.

An older very nice review of WIMP dark matter is

Primack J R, Seckel D and Sadoulet B 1988 Annu. Rev.Nucl. Part. S. B 38 751–807

and a newer review of supersymmetric dark matter is

Jungman G, Kamionkowski M and Griest K 1996 Phys.Rep. 267 195

A survey of experimental results in particle dark matterdetection can be found in the book

Bottino A, di Credico A and Monacelli P 1997 TAUP 97:Proceedings 5th Workshop on Topics inAstroparticles and Underground Physics, Nucl.Phys. B (Suppl.) 70

Recent reviews of microlensing include

Gould A 1996 Publ. Astron. Soc. Pac. 108 465–576Paczynski B 1996 Annu. Rev. Astron. Astrophys. 34 419–59

The analysis of the MACHO collaboration data can befound in

Alcock C et al 1998 Astrophys. J. Lett. 499 L9Alcock C et al 1997 Astrophys. J. 486 697

Kim Griest


Wind E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

WindNASA satellite, part of NASA’s Global Geospace Scienceprogramme and the International Solar Terrestrial Physicsprogramme. Launched November 1994. For the first ninemonths it followed a double-lunar swingby orbit withapogee of 80–250 Earth radii and perigee 5–10 Earth radii.In this orbit, lunar gravity assists maintained apogee overEarth’s day hemisphere for magnetospheric observations.Later inserted into a ‘halo’ orbit at the sunward Sun–Earthgravitational equilibrium point (L1) to measure the solarwind, magnetic fields and particles, and provide a one-hour warning to other ISTP spacecraft of changes in thesolar wind. Since October 1998, placed in ‘petal’ orbitsthat take it out of the ecliptic plane.


Wing, Vincent (1619–68) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wing, Vincent (1619–68)Born in North Luffenham, Rutland, he supported himselfas a surveyor, almanac compiler (his almanac sold50 000 copies per year), astrologer and prolific writer ofastronomical works.


Winthrop, John (1714–79) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Winthrop, John (1714–79)Astronomer, mathematician, born in Boston, MA, becameprofessor of mathematics and natural philosophy atHarvard and observed sunspots, a transit of Mercury andof Venus, eclipses and the weather. He predicted the returnof Halley’s Comet in 1759. He is credited as the firstprofessional scientist in America. He was an ardent patriotduring the American Revolution.


Wise Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wise ObservatoryWise Observatory, in Mitzpe Ramon, Israel, is ownedand operated by Tel Aviv University, and has a well-equipped 1 m telescope. Since construction in 1971,the large percentage of clear nights at its desert siteand its unique longitude have made the observatoryparticularly useful for long-term monitoring projects (e.g.reverberation mapping of quasars and active galaxies),and as a part of global monitoring networks (e.g. thefirst detection, via gravitational microlensing, of a planetorbiting a binary star system).

For further information seehttp://wise-obs.tau.ac.il.


Wittich, Paul (c. 1546–86) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wittich, Paul (c. 1546–86)Born in Breslau, Silesia (now Wroclaw, Poland), invented(or at least developed) ‘prosthaphaeresis’, a formalismof trigonometry that allowed one to multiply and dividetrigonometric functions by the easier process of addingand subtracting instead (in the manner of logarithms).He wrote a commentary on COPERNICUS’s De Revolutionibus,which foreshadowed the Tychonic system. He workedwith TYCHO BRAHE for four months at Uraniborg.


WIYN Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

WIYN ObservatoryLocated at Kitt Peak in Arizona. The WIYN Observatoryis owned and operated by the WIYN Consortium, whichconsists of the University of Wisconsin, Indiana University,Yale University and the National Optical AstronomyObservatories (NOAO). Most of the capital costs of theobservatory were provided by these universities, whileNOAO, which operates the other telescopes of the KITT

PEAK NATIONAL OBSERVATORY, provides most of the operatingservices.

The 3.5 m WIYN Telescope, which was completed in1994, is the second largest telescope on Kitt Peak. The sizeof the telescope enclosure is kept to a minimum by theshort focal length of the primary mirror, which results ina shorter telescope, while the alt-azimuth mount requiresless space. The moving weight of the telescope is a mere46 tons.

Other innovative design features are active primarymirror supports, primary mirror thermal controls andactive ventilation of the telescope mount. The supportsystem for the primary mirror includes 66 actuators, whichadjust the back face of the mirror to maintain the bestoptical figure. The primary mirror thermal control systemkeeps the temperature of the mirror’s surface to within0.2 C of the ambient air temperature, eliminating localturbulence. These innovations enable the WIYN Telescopeto produce much sharper images than any of the othertelescopes on Kitt Peak.

WIYN is equipped with the latest instruments forastronomical spectroscopy and imaging. A multiple-object spectrograph employing optical fibers allows thesimultaneous observation of the spectra of 100 objects.The imaging cameras employ highly sensitive arrays ofelectronic detectors.

For further information seehttp://www.noao.edu/wiyn/.


Wolf, Charles J E (1827–1918) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wolf, Charles J E (1827–1918)French astronomers Charles Wolf and GEORGES RAYET, usingthe 40 cm Foucault telescope of the Paris Observatory,visually observed the spectra in 1867 of several eighthmagnitude stars in Cygnus before the systematic use ofphotographic plates and found very broad emission lines.The ‘bands’ were originally thought to be hydrocarbonmolecules. The stars became known as Wolf–Rayet (WR)stars.


Wolf, Johann Rudolf (1816–93) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wolf, Johann Rudolf (1816–93)Born in Fallanden (near Zurich), Switzerland, becameprofessor of astronomy at the University of Bern anddirector of the Bern Observatory, then professor ofastronomy in Zurich where he founded an observatory.He devised a system now known as Wolf’s sunspot numbersused to quantify solar activity by counting sunspots andsunspot groups and used it to confirm the sunspot cyclediscovered by HEINRICH SCHWABE and measure its period at11 years. He also co-discovered with EDWARD SABINE itsconnection with geomagnetic activity.


Wolf, Max[imilian] Franz Joseph Cornelius (1863–1932)


Wolf, Max[imilian] Franz JosephCornelius (1863–1932)Astronomer, born in Heidelberg, Germany, foundedand became first director of the Konigstuhl Observatoryat the University of Heidelberg. He took wide-fieldphotographs of the Milky Way and counted stars ofdifferent brightnesses, plotting the results in a Wolf diagramof number versus magnitude to prove the existence ofclouds of obscuring dust. He showed that the spiralnebulae have absorption spectra typical of stars, ratherthan emission spectra from gas. He pioneered the use ofphotography to discover hundreds of asteroids.


Wolf–Rayet Nebula E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wolf–Rayet NebulaNebulosity surrounding a Wolf–Rayet star. Wolf–Rayetstars are of around 10 solar masses and have very highsurface temperatures, up to about 40 000 K. This givesthem powerful stellar winds, up to 2000 km s−1, and anenormous rate of mass loss. Material is usually ejectedin the form of a spherical shell or ring (the term Wolf–Rayet bubble is sometimes used), and the accumulatingenvelope from successive ejection episodes comprisesthe nebula. Examples of Wolf–Rayet nebulae are NGC2359, surrounding the star HD 56925, and NGC 6888,surrounding the star MR 102.

See also: Wolf–Rayet stars.


Wolf–Rayet Stars E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wolf–Rayet StarsWolf–Rayet (W–R) stars are a class of peculiar stars firstidentified in 1867 by C J E WOLF and G RAYET. Unlike thespectra of most stars, which are dominated by narrowabsorption lines, the spectra of W–R stars show broademission lines. The rich emission line spectrum makesthem easy to identify, by spectroscopic observations, evenat large distances.

W–R stars are divided into three broad spectroscopicclasses (WN, WC and WO) based on the emission linespresent in their spectrum. WN stars show emission linespredominantly of helium and nitrogen, although emissiondue to carbon, silicon and hydrogen can readily be seen insome of these objects. In contrast, the spectra of WC starsare dominated by carbon and helium emission lines withhydrogen and nitrogen emission absent (figure 1). WOstars, which are much rarer than either WN or WC stars,are similar to WC stars except that oxygen lines are moreprevalent, and there is a tendency to exhibit lines arisingfrom atomic species of higher ionization.

These spectral classes are further divided intosubclasses on the basis of line ratios, yielding aclassification by ionization. The WN stars which exhibitspectra showing emission from high-ionization species(e.g. He II, N V, O VI)1 are designated WN2. Those showingemission from low-ionization species (e.g. He I, N III) areclassified as WN9, although recently the W–R spectralclassification has been extended to WN11. Similarly,WC stars showing emission from high-ionization species(e.g. He II, C IV, O VI) are designated WC4 while thoseexhibiting the lowest ionization (e.g. He I, C II) aredesignated WC9. In the literature there is also a tendencyto refer to WN stars of classes 2–5 as early type (WNE)and classes 6–9 as late type (WNL). Similarly, WC4–6 starsare designated as WCE, while WC7–9 stars are designatedas WCL. Although there are important exceptions, WNEstars generally show no evidence for H emission while Hemission is present in WNL stars.

The distribution of population I W–R stars, which arediscussed in this article, is similar to that of O stars; they areprimarily located in the spiral arms of our galaxy and nearH II regions. W–R masses range from an uncertain lowerlimit of about 5M to in excess of 60M, while surfacetemperatures range from a lower limit of 25 000 K to greaterthan 100 000 K. Because of their spatial association with Ostars, and their peculiar surface abundances, W–R stars aregenerally believed to be descended from O stars.

Approximately 220 W–R stars are known in ourGalaxy but this number is certainly incomplete. Most arehidden from our view by dust, which absorbs and scatterslight (a process termed interstellar extinction) within ourGalaxy. Estimates of the total number of W–R stars inour Galaxy range from 1000 to 2000. The rarity of W–R

1 He II is a spectroscopic designation used to indicate the ion fora transition between bound levels in singly ionized helium (i.e.He+).

stars is due to the initial mass function, which favors theproduction of low-mass stars, and the short evolutionarylifetime of W–R stars, which is only a few ×105 yr. Theirrarity belies their importance. All stars more massive thanapproximately 25M (for solar metallicity) pass througha W–R phase. Further, over the lifetime of a galaxy, W–Rstars (and their progenitors) have an important influenceon the energetics, dynamics and chemical evolution of theinterstellar medium.

W–R stars are expected to end their life via aspectacular supernovae explosion. In some cases aneutron star is formed, while for the more massive starsa black hole is formed. In the latter case the details of themechanism that actually produces the supernova are stillvery uncertain. As the supernovae ejecta expands, it willinteract with the complex circumstellar environment thatreflects the previous mass loss history of the progenitorstar.

In addition to the population I W–R stars, someplanetary nebula central stars also show W–R emissionfeatures (see PLANETARY NEBULA CENTRAL STAR MASS LOSS/WINDS,W–R NEBULA). Their spectral types are inserted in [] todistinguish them from population I W–R stars. They areof type [WC] and have lower masses (less than 1M) andlower luminosities (generally <3× 104L). The spectra ofplanetary nebula W–R stars are often dominated by strongnarrow nebula emission lines. In some cases they canbe difficult to distinguish spectroscopically from normalpopulation I W–R stars although in some [WC] stars N andH emission is seen. Because of their distinct evolutionaryhistories, they will not be further discussed in this article.

Basic modelThe basic model for W–R stars is that of a hot star whichis suffering extreme mass loss. The mass loss occurs viaa continuous stellar wind which is accelerated from lowvelocities near the surface of the star to velocities thatexceed the surface escape speed. The observed spectrumoriginates over a range of radii with the optical continuumforming close to the stellar core, while the emission linesoriginate from a volume that can extend beyond 10 stellarradii.

The observed mass-loss rates (i.e. the amount ofmaterial lost per year) are extreme, typically in excess of10−5M yr−1. These mass-loss rates are sufficient to affectthe evolution of the star (see STELLAR EVOLUTION) and mustbe incorporated into stellar evolutionary calculations.The (average) maximum velocity of material in W–Rwinds (called the terminal velocity, V∞) ranges from800 km s−1 to in excess of 3000 km s−1, and typicallyexceeds the escape velocity from the surface of the star. Itis generally believed, although it has yet to be rigorouslydemonstrated, that the mass loss is driven by radiationpressure acting through numerous bound–bound atomictransitions of Fe and other atomic species in the extremeUV (λ < 900 Å).

The bulk of the material in the wind is believed tobe cool—that is, it has a temperature substantially lower



Figure 1. UV and optical spectra of the WN5 star HD 50896 and the WC5 star HD 165763. The major emission line features areidentified, although it should be noted that many lines, particularly in the WC star, are blends. Notice the very distinct differencesbetween the WN5 and WC5 spectra. The optical spectral region, which can be observed from the ground, has typically been used toclassify W–R stars. Since the advent of space astronomy, the UV spectral region has provided additional invaluable diagnostics on theproperties of W–R stars.



than the effective temperature of the star. Energy inputinto the wind primarily occurs through photoionization bythe intense UV radiation field emanating from the centralsource. Thus photoionization is the ultimate source ofthe line emission that originates in the wind, althoughindividual emission lines form through a variety ofdifferent processes—recombination, collisional excitationand continuum fluorescence.

Determination of stellar parametersThe spectra of most stars are determined by three basicparameters: the effective temperature (Teff), the effectivesurface gravity (geff) and the chemical abundances. Foran assumed metallicity, large grids of models can beconstructed simply by varying Teff and geff .

The effective temperature is defined by the relation

L = 4πR2∗σT

4eff

where L is the luminosity and σ is the Stefan–Boltzmannconstant. The effective temperature is the surfacetemperature that a star of radius R would have if itradiated as a blackbody (a perfect thermal emitter andabsorber).

The effective surface gravity is defined by

geff = (1− )GM

R2

where M is the star’s mass, G is Newton’s gravitationalconstant, and is a correction for the influence of radiationpressure. In conjunction with the equation of hydrostaticequilibrium

dPdr= −ρgeff

(R

r

)2

where P is the pressure and ρ is the density, this setsthe scale height of the atmosphere, h. For an isothermalatmosphere, h is given by

h = kT

µmHgeff

where µ is the mean particle mass in atomic mass units(amu) and mH is the atomic mass of hydrogen. For mostnormal stars h R, curvature effects can be ignoredand the atmosphere can be treated as a plane-parallel slab.As a consequence of the small scale height, the Sun has adefinite radius at optical wavelengths.

For W–R stars the situation is quite different. First,the atmosphere is extended, and consequently radiationescapes from the star over a range of radii. Further,the radius of the star at an optical depth (τ ) of 2/3depends on the adopted mass-loss rate and is a functionof wavelength (figure 2). The difficulty of uniquelydefining R has led to difficulties in comparing Teff

derived from evolutionary models with that obtained fromspectroscopic analyses. Second, geff does not have a directinfluence on the stellar spectrum, simply because emission

from the stellar wind dominates the spectral appearance ofthe star. Third, the abundances are non-solar and must bedetermined observationally. Indeed it is the abundancesthat determine to which class (WN, WC or WO) a W–R starbelongs (see also STELLAR ATMOSPHERES: EARLY-TYPE STARS).

In addition to the abundances (primarily of H, He, N,C and O) it has been found from numerical experimentsthat the spectra of W–R stars are determined primarily bytwo parameters: Teff and a wind density parameter, Wρ .The latter, which plays a similar role to geff , can be definedby

Wρ = (M/V∞)R−3/2c

where Rc is the radius of the hydrostatic core. Twostars will have very similar spectra if they have similarabundances, and ifTeff andWρ are similar. The dependenceof the spectra on Wρ arises because most of the radiativeprocesses in an extended atmosphere depend on thesquare of the density. As a consequence of the scaling,it is impossible to deduce the distance of a W–R star fromthe Sun using its spectrum2. In principle, M andV∞ shoulddepend on the other stellar parameters—composition, M ,L and R, but as yet our theoretical understanding of massloss from W–R stars is not sufficiently advanced to deducethe relationship.

Because of the low wind densities (108–1014 electronscm−3)3 the simplifying assumption of local thermo-dynamic equilibrium (LTE) cannot be made whenmodeling W–R spectra. When LTE holds, it can beassumed that the ionization state of the gas and thepopulations of the atomic levels can be found viaapplication of the principles of statistical mechanics andthus are (simple) functions of the local temperature anddensity only. For LTE to prevail, collisional processes,which couple the atomic populations with the electrons(and hence to the local electron temperature), need to occurfaster than radiative processes.

In contrast, in W–R atmospheres radiative processestend to dominate over collisional processes, and hence it isnecessary to solve the equations of statistical equilibriumat each depth. For each atomic level of each species weassume that all the processes (radiative and collisional)populating the level are in equilibrium with all processesdepopulating the same level. The major difficulty arisesbecause the rates are a function of the radiation field, whichin turn is a function of the unknown populations. Thus theradiation field and atomic populations must be solved forsimultaneously, and in general an iterative procedure isnecessary to obtain consistency.

2 Stellar evolution introduces a correlation between spectraltype and luminosity which may statistically allow distancesto be derived for stars of a given spectral type. However,for an individual star the derived distance may be grossly inerror, particularly if the spectral type can originate via differentevolutionary sequences. This is exemplified in the difficulty ofdetermining whether some W–R stars belong to population I orare the central stars of planetary nebula.3 For comparison, the density of water on Earth is of order 1022

molecules cm−3.



Figure 2. Illustration of how the ‘radius’ of the star varies with wavelength for the WC5 star HD 165763. The solid curve shows the‘radius’ at which the ‘continuum’ optical depth, τ , is 1/3, which occurs in the wind at all wavelengths. The broken curve is identicalexcept that the electron scattering opacity was not included in the computation of τ . The bulk of the observed flux in this model isemitted between 300 and 2000 Å. Rc is the radius of the hydrostatic core—that is, the approximate radius the star would have in theabsence of a stellar wind. For the illustrated model it was 1.8R.

Initial modeling of W–R spectra concentrated on theH and He spectra only. The second generation includedCNO elements, while the most recent generation of modelsinclude iron and other species. The inclusion of iron (andsimilar species) in non-LTE calculations has been a majorstumbling block for atmospheric calculations for O andW–R stars for many years. The iron-group ions have botha wealth of atomic levels and an enormous number ofbound–bound transitions. The advent of faster computerswith large memories, new numerical techniques and theavailability of atomic data have now made it feasible toinclude iron and other species in non-LTE calculations.

Determination of Teff

The effective temperatures of W–R stars are determinedprimarily using ionization arguments. Consider asequence of models with prescribed mass loss and velocitylaw, but with different effective temperatures. Such asequence of models will exhibit a smooth variation in lineratios for lines from two successive stages of ionization.In early WN modeling it was customary to compare He II5411 with He I 5876 since both lines are easily observed andare relatively blend free. In more recent modeling N linescan also be used to constrain the effective temperature,while in WC stars C and O lines can be used. Theanalyses generally give consistent results. Discrepantresults do occur, and probably result from a poor treatmentof line blanketing (the generic name given to the influenceof thousands of bound–bound atomic transitions on anatmosphere) and/or density inhomogeneities in the stellarwind.

Analyses of the spectra of ring nebula aroundW–R stars offers a method of checking on the energydistributions predicted by atmospheric modeling, since

the nebula are ionized by the star’s radiation field. Forring nebula around WNE stars these analyses have usuallyshown reasonable consistency. However, for some WNLand WC stars the observed nebular spectrum was of lowerexcitation than would be predicted using the stellar UVradiation field derived from the modeling. The recentinclusion of line blanketing in the stellar atmosphericmodels has removed this discrepancy for at least one ringnebula around a WN8 star.

AbundancesThe severe non-LTE conditions in W–R stars initiallymade it difficult to understand their peculiar emissionline spectra. Do W–R stars possess peculiar (i.e. non-solar) abundances? Is the difference between the WNand WC stars due to an abundance difference, or isit an excitation effect? Detailed recombination andspectroscopic analyses have now firmly established thatW–R stars are characterized by non-solar abundances.

In WN stars H, C and O are depleted, while N andHe are enhanced. For WN stars, N (H)/N (He) ratios (bynumber) range from approximately 4 to <0.1 (the solarvalue is 10). The observed abundances are consistentwith the idea that material processed by the CNO nuclearburning bi-cycle has been revealed (or mixed) at thesurface (see ‘Evolution’).

In WC stars, He, C, O and Ne are all enhanced.N (C)/N(He) = 0.1–0.5 while the less certain N (O)/N (He)ratios are typically 0.1. H and N are not expected in WCstars and are not detected. The variation of N (C)/N (He)and N (O)/N (He) with WC subtype is still the subject ofmuch debate.



Determination of mass-loss ratesMass-loss rates can be determined in several ways. First,free–free radiation in the dense stellar wind gives rise to adetectable flux at radio wavelengths. From measurementsof the radio flux, Sν , the mass-loss rate, M , can bedetermined from the simple formula

Sν = 23.2

(Mz

V∞µ

)4/3(γgν

d3

)2/3

where d is the distance to the W–R star in kpc, M is themass-loss rate in M yr−1, V∞ is the terminal velocity ofthe wind in km s−1, z is the mean ionic charge, µ is themean ionic mass (in amu), γ is the number of electronsper ion, ν is the frequency in Hz and g is the free–freeGaunt factor (which is the quantum correction factor to thesemiclassical formula for free–free radiation) at frequencyν. Sν is measured in janskys (1 jansky = 10−23 erg cm−2 s−1

Hz−1). The greatest uncertainties in the derived M arisefrom uncertainties in stellar distance and in the ionizationstate of the gas in the radio-emitting region.

Second, the IR flux can be used in a similar manner,although in this case it cannot be assumed that the windhas reached its terminal speed. Third, optical and UVemission lines can be used. Typically, recombination linesare used since they are less sensitive to the precise detailsof their formation. Mass-loss estimates obtained fromdifferent methods generally agree to within a factor of 2.

It has generally been assumed that the winds of W–Rstars are spherical and homogeneous. Thus at any locationin the wind the density can simply be found from theprinciple of mass conservation, giving

ρ = M

4πr2v(r)

where v(r) is the velocity as a function of distance andis assumed to monotonically increase with r . However,emission line variability studies and analyses of emissionline profiles suggest that the wind is clumped (i.e. non-homogeneous) on small scales. If this is true, mass-lossestimates are too large—possibly by factors of 3 or more.This has very important implications for stellar evolutioncalculations. A change in M by only a factor of 2 (overa star’s life) has a profound influence on the evolutionof massive stars. This is seen in theoretical evolutionarycalculations and can also be indirectly inferred from thedifferent WN/WC ratios in the galaxy and our nearestextragalactic neighbor, the Large Magellanic Cloud (LMC).In the LMC the mass-loss rates are expected to be lowerbecause of the lower metallicity.

The extreme mass-loss properties of W–R stars canbe characterized by the ‘wind performance parameter’, η,defined by

η = MV∞L/c

.

MV∞ is the (scalar) momentum of the wind, while L/c

is the momentum that could be transferred if all the

photons, and hence all the momentum in the radiationfield, were absorbed. For O stars, η is typically less thanunity. For W–R stars values as high as 100 have beenobtained, although by allowing for inhomogeneities andline blanketing it has been possible to reduce the values toless than 10. Values of η in excess of unity do not rule outa radiation-driven wind—they simply indicate that eachphoton has to scatter many times within the wind so thatit delivers η times its momentum to the wind. It has beendifficult to produce radiation-driven winds for W–R starssince current models do not have the necessary number ofbound–bound atomic transitions to perform the requirednumber of scatterings.

BinariesApproximately 50% of W–R stars occur in binaries—anumber comparable with O stars. In the past there hasbeen considerable discussion on the importance of binarityfor the W–R phenomenon. For example, it was oncethought that W–R stars could only originate in a binarysystem. Mass loss from the W–R progenitor would thenoccur by Roche-lobe overflow. The major uncertaintyin evolutionary calculations of binary systems is howmuch material is lost from the system during Roche-lobeoverflow (rather than being accreted by the companion).More recently the binary channel for the production ofW–R stars has virtually been ignored. There is little doubt,however, that the binary channel is important, and it mustbe considered when linking W–R types with evolutionarycalculations (although many researchers would disagreewith this statement).

At least three (broad) distinct classes of W–R binarysystems can be envisioned: W–R + OB star, W–R + W–R,W–R + compact companion (neutron star or black hole).All (confirmed) W–R binaries belong to the first class;W–R + W–R systems are expected to be rare, althoughWR98 (where WR98 denotes the 98th W–R star in thesixth catalogue of W–R stars) may be an example of sucha system. The third class is expected on evolutionarygrounds. Although the existence of such systems hasbeen difficult to verify, several good candidate systems areknown. Cyg X-3 can be considered a possible example.

W–R binary systems are extremely useful. First,and foremost, they allow a direct determination of stellarmasses, independent of evolutionary models. The useof these masses in constraining single-star evolutionarymodels presupposes that W–R stars in binaries haveproperties similar to single W–R stars—a proposition thatcannot be reliably tested because of poor statistics anduncertainties in the properties of W–R stars.

Second, the O star can be used to probe the structureof the W–R stellar wind. Indeed it was this technique,applied to the W–R + O binary V444 Cygni, which gavethe first direct evidence that W–R stars are hot (i.e. surfacetemperature in excess of 60 000 K). More recently it hasbecome evident that polarization studies of binary systemsmay allow determinations of mass-loss rates which areinsensitive to the presence of inhomogeneities within thestellar wind.



X-ray emissionMost W–R stars are thermal x-ray sources. They can beclassified into two categories—single W–R stars and binaryW–R stars. In single W–R stars the observed ratio of x-rayto bolometric luminosity is approximately 10−7, althoughthere is a scatter of at least a factor of 3 about this value.The x-rays are believed to originate within the stellarwind through radiation-driven wind instabilities whichlead to high-velocity shocks (Vshock = 100–600 km s−1)and clumping of the wind. These shocks generate x-rayemitting gas characterized by temperatures between 106

and 107 K. The shocks probably permeate most of the wind;however, because of the high wind densities most of thex-rays are absorbed in the wind.

A few W–R stars are stronger and/or variable x-rayemitters. In these binary systems the x-rays can begenerated from the high-temperature gas (of order 107 K)created via shocks generated in a wind–wind collision.The two best examples are the WN +O system V444 Cygniand the WC + O system HD193793 (WR140). Both exhibita periodic x-ray variability broadly consistent with thatexpected from a wind–wind collision in a binary system.Alternatively the x-rays could be generated via accretionof wind material onto a compact companion.

Dust formationThis topic, by all rights, should not belong in a discussionon W–R stars since dust is destroyed by intense UVradiation and requires low temperatures (<3000 K) for itsformation. Yet dust is seen around some WC stars, andmoreover dust is seen to be created in the vicinity of someWC stars. Two distinct dusty WC classes may exist: binarysystems and single stars.

In many WC9 stars, presumed to be single, conditionswithin the C-rich stellar wind appear to allow dustformation. The dust nucleation routes and how thedust (and the necessary prerequisite molecules) formsdespite the presence of an intense UV radiation field areunclear. If the stars are single, inhomogeneities generatedby radiation instabilities probably play an important rolein allowing dust to form.

The second class are the binary systems, with eccen-tric orbits, with HD193793 (WR140) being an excellent ex-ample. In these systems dust does not normally form;however, binary interaction near periastron (i.e. minimalorbital separation) can facilitate dust formation. Appar-ently the high densities generated in the wind–wind inter-action have the right conditions for dust formation. As inthe single WC stars, the dust formation is not understood.

Recent interferometric observations of WR104 withthe KECK telescope have revealed directly the dustoutflowing from the interaction region of the binary(figure 3).

Related starsIn the upper part of the Hertzsprung–Russell (H–R)diagram, many different classes of massive luminous starsexist (see HIGH-LUMINOSITY STARS): Of stars, blue supergiants

Figure 3. A gray scale image at 2.27 µm showing thedistribution of dust in WR104, as found by interferometricobservations with the KECK telescope. The dust moves awayfrom the system radially—the apparent spiral motion is anillusion. It results from the rotation of the dust formation zoneas a consequence of the orbital motion of the binary system. Thering of dust, as illustrated, has an angular diameter ofapproximately 160 AU (or 0.1 arcsec). (Picture courtesy ofW Danchi, J Monnier and P Tuthill, Berkeley.)

(BSGs), red supergiants (RSGs), LUMINOUS BLUE VARIABLES

(LBVs) and WN/Of stars. One of the goals of massive starevolution is to understand the links between the variousclasses of objects and the distribution of massive starsbetween the different classes. Below we briefly discusssome of the salient features of each class.

Of stars are O supergiants exhibiting emission lines inthe optical. They are O stars that have evolved off the mainsequence. WN/Of stars exhibit spectral characteristicsof both Of and WNL stars. It was this intermediatecharacteristic that suggested an evolutionary link betweenthe Of stars, and bona-fide WN stars.

LBVs, as their name suggests, are luminous blue starswhich show irregular variability on a time scale of hoursto centuries. Some LBVs have exhibited giant outbursts inwhich their visual brightness and bolometric magnitudeincreased by several magnitudes. During such outburstsseveral solar masses of material may be ejected. OtherLBVs, such as AG Car, show moderate outbursts on atimescale of a decade. During these outbursts the effectivetemperature changes but the bolometric luminosity andmass-loss rate are almost constant. The most famous LBVsare P Cygni, which suffered a giant outburst in the 1600s,and η Car which underwent a major outburst in the 1840s.The outburst suffered by η Car ejected a bipolar nebula,referred to as the Homunculus, which has a major axisdiameter of 17′′ (approximately 4000 AU). Images of theHomunculus are amongst the most spectacular obtainedby the Hubble Space Telescope.



A major breakthrough in our understanding ofmassive stars was achieved when one LMC Of/WN star(now classified as WN11), R 127, was observed to undergoan LBV-like outburst, suggesting an evolutionary linkbetween LBVs and Of/WN stars.

LBVs are now regarded as a key phase of massivestar evolution. It is believed that during the LBV phasea massive star ejects most of its hydrogen-rich outerenvelope, allowing it to become a W–R star. A keyobservation that has led to this scenario is the absenceof RSGs with luminosities comparable with the mostluminous O stars. Evolutionary calculations show that,during the LBV phase, extensive mass loss can preventthe star from evolving into an RSG. The mechanism ofmass loss during the LBV phase is not yet understood,although it may be related to the star evolving towards themodified Eddington limit. The classical Eddington limitprovides a lower limit on the mass of a star, of a givenluminosity, based on the assumption that the radiativeforce arising from the electron scattering opacity cannotexceed gravity. The modified Eddington limit takes intoaccount that other radiative processes also contribute tothe opacity, and further that stellar rotation can effectivelylower the surface gravity. The maximum luminosityexhibited by supergiants in the H–R diagram is termedthe Humphreys–Davidson limit.

EvolutionW–R stars are believed to be descended from O stars. Thebasic evolutionary sequence, first proposed by Conti in1976, is

O→ Of→W–R

Since that time observational and theoretical work has ledto refinements in this basic sequence. From the theoreticalwork of Maeder and collaborators, one such sequence forstars with initial masses greater than 50M is

O→ OIf→ BSG→ LBV→WN→WC→ supernova

while stars between 35M and 50M have the alternativesequence

O→ BSG→ YSG→ RSG→ YSG→WN

→WC→ supernova.

Other sequences have also been proposed. Suffice it tosay that the precise evolutionary path that an individualmassive star follows (which depends on the star’s initialmass and composition, and possibly its rotation rate andwhether it has a companion) is still uncertain. No firmlink has been established between the different ionizationclasses within the WN and WC sequences, although therehas been some success in linking spectral types with initialstellar mass. The unknown roles of binary interactions androtation only add to the confusion.

Both WN stars and WC stars are generally believedto be on the helium-burning main sequence, althoughsome of the luminous hydrogen-rich WN stars may

still be core hydrogen burning. If the latter is true, itmeans that the spectroscopic and theoretical definitionsof W–R stars are inconsistent. This creates difficultiesin comparing observed W–R/O number ratios withtheoretical predictions. The W–R/O ratio is an importantobservational constraint since it provides a method (atleast in principle) of determining the minimum stellarmass which will evolve into a W–R star. For a solarmetallicity this is generally believed to be approximately25M, but higher values cannot be ruled out.

In massive stars, H burning occurs via the CNObi-cycle through a sequence of reactions, with the CNOspecies acting as catalysts. In the CN cycle the followingreactions occur:

126C + 1

1H→ 137N + γ

137N→ 13

6C + e+ + ν

136C + 1

1H→ 147N + γ

147N + 1

1H→ 158O + γ

158O→ 15

7N + e+ + ν

157N + 1

1H→ 126C + 4

2He.

The fourth reaction is the slowest, and as a consequencemuch of the original C (and O from the other reactionsin the CNO bi-cycle) is converted to N. The total numberof CNO nuclei remains unchanged. When equilibrium isobtained, the ratio of 14N to 13C nuclei is approximately 50,very different from the solar ratio of 0.27.

In normal stars the nuclear processed materialremains within the stellar core and cannot be observed.However, in O stars and their descendants, extensive massloss peels off the outer hydrogen-rich layers. Nuclearprocessed material, once inside the convective core of thestar, is eventually revealed at the surface. In addition tomass loss, it is now believed that mixing, possibly inducedby stellar rotation, can help reveal nuclear processedmaterial at the stellar surface.

In WC stars the mass loss has been so extensive thatthe products of He burning are revealed at the stellarsurface. The predominant reactions for helium burningare

42He + 4

2He 84Be

84Be + 4

2He→ 126C + γ

126C + 4

2He→ 168O + γ.

The variation in surface abundances as a function ofcurrent mass for a star with an initial mass of 40M isshown in figure 4.

W–R stars in external galaxiesW–R stars are moderately easy to detect in externalgalaxies owing to their strong emission lines. Typicallythey are found by performing a photometric survey in twofilters. The passband of one filter is centered on a strongemission feature (generally the He II–C III/C IV complexat 4640–4690 Å) while the second passband is centered on



Figure 4. A diagram showing the evolution in surface abundances as a function of mass for a star with an initial mass of 40M. TheWN phase occurs when the star’s mass is between (approximately) 15M and 30M, while the WC phase occurs when M < 15M.For most of its 4.8× 106 yr life the star has a mass close to its initial value (e.g. after 4.2× 106 yr, its mass is still ≈36M). The lifetimesof various stages are very dependent on the adopted assumptions (e.g. overshooting and mixing) and the adopted mass-loss rates.Overshooting refers to the phenomenon of convective motions extending into a convectively stable region because the convectivevelocities are non-zero at the interface between the convectively stable and unstable regions. Mixing refers to the process of mixingtwo chemically distinct regions of the star—for example rotation might induce nuclear processed material to be transported to thesurface layers. The calculations were undertaken by the Geneva group (Maeder, Meynet, Schaller and Schaerer) for solar metallicity,with overshooting and ‘normal’ mass-loss rates.

the continuum only. W–R stars will be relatively brightin the emission line passband. Alternatively, they can befound using low-dispersion prism spectroscopic surveys.

Both the Large Magellanic Cloud (LMC) and the SmallMagellanic Cloud (SMC) have been extensively surveyedfor W–R stars. In the LMC 134 W–R stars are known,while in the SMC only nine are known. The differencein the number of W–R stars is believed to be due to acombination of the star formation rates and the lowermetallicity of the SMC (which inhibits W–R production).Interestingly, the ratio of WN to WC stars in the LMC is4.5, substantially larger than the observed ratio of 1 inthe solar neighborhood. This is generally interpreted asa metallicity effect.

W–R stars have also been found in many Local Groupgalaxies, e.g. M31, M33, IC1613 and NGC6822. Thestudy of individual W–R stars in these galaxies is in itsinfancy. To date, efforts have been directed primarily intodeterminations of the WN/WC, O/W–R and RSG/W–Rnumber ratios which allow global issues, such as theeffect of metallicity and the star formation rate on W–Rproduction, to be addressed.

W–R stars have also been found in many galaxiesexhibiting extensive star formation (often called STARBURST

GALAXIES). Indeed some galaxies are termed W–R galaxiesif they exhibit strong W–R features in their integratedspectra. The presence of W–R stars in these galaxiesimmediately provides an age determinant. The starbursthas to be older than approximately 2 million years so that

the most massive O stars that formed in the burst have hadsufficient time to evolve into W–R stars. They also providean upper limit of about 7 million years since after this timeall massive stars that pass through a W–R stage will havedone so. Both age limits are metallicity dependent.

Outstanding problemsThere are many outstanding problems related to W–Rresearch. Several of the most important problems relatedspecifically to W–R stars are discussed below.

(1) What initiates and drives mass loss from W–Rstars? Is radiation pressure, as currently believed,responsible for mass loss from W–R stars? Why aremass-loss rates for W–R stars an order of magnitudehigher than those of their O star progenitors?

(2) What is the role of rotation in W–R stars (and theirprogenitors) in modifying the spectral appearanceof the star? What is the role of rotation in massivestars in enhancing mixing processes and in enhancingmass-loss rates, and hence in modifying stellarevolution?

It is essential for evolutionary calculations,whatever the mass-loss mechanism, that we are ableto derive mass-loss rates from first principles. Ideally

M = M[Minit, t, x, y, z,((t)]

where Minit is the initial stellar mass, t is the currentage of the star and x, y and z are the chemical



abundances (hydrogen mass fraction, helium massfraction and metal mass fraction), and where therotation rate ((t) is itself determined via the initialrotation rate and the subsequent mass loss.

A related question is the role of magnetic fieldsin massive star evolution. While the magnetic fieldsmay be too weak to affect mass-loss rates they couldbe important in determining angular momentumlosses, and hence play an indirect role in mass lossthrough the dependence of ((t) on magnetic fieldstrengths.

(3) What is the detailed structure of W–R winds (shapeand homogeneity)? How does the presence ofinhomogeneities affect the determination of thefundamental stellar parameters? How coupled is themass loss to the details of the wind structure?

(4) What is the role of binaries in massive star evolution?Are there classes of W–R stars (and LBVs) that onlycome from a binary evolutionary scenario? Howmuch is our understanding of single-star evolutionbeing confused by the evolution of OB stars ininteracting binary systems?

(5) Because of uncertainties in the treatment of convec-tion, mass-loss processes and rotation, there are stillmany issues in single-star evolution which must beaddressed. Also of concern is whether W–R stars giverise to black holes (as presumed), and to what class aW–R supernova belongs.

AcknowledgmentsThis contribution is a brief synopsis of our currentunderstanding of W–R stars. This understanding is thecumulative result of 130 years of W–R research by manyastrophysicists whose individual contributions cannot beacknowledged.

BibliographyAbbott D C and Conti P S 1987 Wolf–Rayet stars Ann. Rev.

Astron. Astrophys. 25 113 (an excellent review articlewhich discusses W–R stars in greater depth)

Bappu M K V and Sahade J (ed) 1973 Wolf–Rayet and HighTemperature Stars (IAU Symp. 49) (while some of thearticles are somewhat dated, the discussion betweenthe symposium participants is rewarding and is sadlymissing in most recent symposia)

Maeder A and Conti P S 1994 Massive star populationsin nearby galaxies Ann. Rev. Astron. Astrophys. 32 227(an excellent review article which discusses massivestar populations and evolution, with an extensivediscussion of W–R stars)

Tylenda R 1996 Wolf–Rayet Central Stars of Planetary Nebulae(ASP Conf. Ser. 96) ed C S Jeffery and U Heber (review:several other papers on W–R central stars are alsocontained within the same volume) p 101

van der Hucht K A, Conti P S, Lundstrom I and Stenholm B1981 The sixth catalogue of galactic Wolf–Rayet stars,their past and present Space Sci. Rev. 28 (3)

van der Hucht K A, Koenigsberger G and Eenens P R J(ed) 1999 Wolf–Rayet Phenomena in Massive Stars andStarburst Galaxies (IAU Symp. 193) (AstronomicalSociety of the Pacific) (review and research articleson different aspects of W–R research)

Vreux J M, Detal A, Fraipont-Caro D, Gosset E and Rauw G(ed) 1996 Wolf–Rayet Stars in the Framework of StellarEvolution: Proc. 33rd Liege Int. Astrophysical Colloq.(Universite de Liege) (contains many review articleson different aspects of W–R research)

D John Hillier


Wollaston, William (1766–1828) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wollaston, William (1766–1828)Chemist and physicist, isolated two new metals fromplatinum ore, one of which he named palladium (afterthe newly discovered asteroid, Pallas). He was one of thefirst scientists to observe ultraviolet radiation, and in 1801discovered the dark spectral lines in the solar spectrumwhich were later investigated by FRAUNHOFER. This was thefirst observation of spectral lines. A Wollaston prism is usedto study the polarization of light.


Woolley, Richard van der Riet (1906–86) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Woolley, Richard van der Riet(1906–86)Astronomer, born in Dorset, England, and followeda varied career in England, South Africa, Californiaand Australia. Became director of the CommonwealthSolar Observatory and steered it away from solarobservations and wartime optics work to become theMount Stromlo Observatory and part of the AustralianNational University. He worked on stellar and solaratmospheres. In 1955 he became Astronomer Royal inBritain, directing the Royal Greenwich Observatory inSussex. He worked there on the motions and astrophysicsof stars and clusters of stars, directing a small army ofworkers on the measurements of star positions on thethousands of photographic plates that he caused to betaken. He discovered the orbit of the globular clusterOmega Centauri, which moved radially in the Galaxy.He caused the Isaac Newton Telescope to be erected atHerstmonceux: it was later moved to the clearer sky of LaPalma. He retired from the Royal Greenwich Observatoryand became director of the South African AstronomicalObservatory at the Cape.


Wormhole E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

WormholeA hypothetical shortcut, or ‘tunnel’, that, in principle, maylink the interior of a black hole to another universe orto another location in our universe. During the 1930s,Albert Einstein (1879–1955) and Nathan Rosen (1909–95)showed that the sharply curved spacetime of the interiorof a black hole may open out again into another spacetime(another universe). The hypothetical connection betweenthese two regions of spacetime came to be known as anEinstein–Rosen bridge. An alternative interpretation isthat the bridge, or tunnel, links two different regions inthe spacetime of our own universe. More recently, theterm ‘wormhole’ has been used to describe a spacetimetunnel of this kind. Although it has been speculatedthat wormholes could be used to facilitate virtuallyinstantaneous interstellar travel, in practice it seems likelythat, even if wormholes do exist, they will be too smalland too short-lived (and too physically hazardous) to beutilized in this way.

See also: general theory of relativity, spacetime.


Wren, Sir Christopher (1632–1723) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wren, Sir Christopher (1632–1723)Architect and astronomer, born in East Knoyle, Wiltshire,became professor of astronomy at Gresham College, andSavilian Professor of Astronomy at Oxford. He rebuiltLondon after the fire of 1666, planning the entire cityand rebuilding 51 churches, including St Paul’s Cathedral.Newton acknowledges Wren as a mathematician in thePrincipia. Wren independently proved KEPLER’s third lawand formulated the inverse-square law of gravitationalattraction. He solved Kepler’s problem on cutting asemicircle in a given ratio by a line through a given pointon its diameter (it had arisen as a problem in Kepler’s workon elliptical orbits). Wren was a founder member of theRoyal Society, and its president for two years.


Wright, Thomas (1711–86) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wright, Thomas (1711–86)English astronomer, born near Durham, clock-maker andmathematics teacher, wrote in 1750 a curiously arguedbook in which he suggested that the Milky Way was adisk-like system of stars with the solar system near thecenter. Wright suggested that nebulae were star systemssimilar to the Milky Way, but very far away. Conjecturedthat the gap in the planets between Mars and Jupiter hadbeen cleared by the collision of a comet with a planet whichhad then been ejected from its orbit.


Wrinkle Ridge E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Wrinkle RidgeA long ridge on the surface of a planetary body. Wrinkleridges were first identified on the Moon, where they areoften associated with rilles. As the lunar maria solidified,tensile forces in the outer regions opened up the faultsthat produced rilles, while compressive forces nearer thecenter pushed up the surface to form wrinkle ridges.Some wrinkle ridges may result from small-scale extrusionof lava along fissures; others extend from the mariainto surrounding upland terrain. Lunar wrinkle ridgesare typically several hundred meters high and severalhundred kilometers long. On Venus wrinkle ridges arecommon features on the plains, where they extend for 10 to50 km. The alignment of many of them suggests that theyare associated with the compressive forces that upliftedthe northern upland region of Aphrodite Terra.


Xenocrates of Chalcedon (396–314 BC) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Xenocrates of Chalcedon (396–314 BC)Philospher and mathematician, born in Chalcedon (nowKadikoy, near Istanbul), Bithynia (now Turkey), succeededSpeusippus as head of the Athenian Academy whichPLATO had founded. Believed that matter is composed ofindivisible units (and thus an early atomist). He believed(and perhaps originated the notion) that people have athreefold existence, mind, body and soul, and that they dietwice, once on Earth, then on the Moon, when the mindseparates from the soul and travels to the Sun.


Xenophanes of Colophon (c. 570–c. 480 BC) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Xenophanes of Colophon (c. 570–c.480 BC)Greek philosopher, born in Colophon. He believed Earthto be the fundamental element of the universe, noting thatthat because seashells are sometimes found on mountaintops, the physical arrangement of the Earth changes withtime.


XEUS (X-ray Evolving Universe Spectroscopy Mis-sion)


XEUS (X-ray Evolving UniverseSpectroscopy Mission)Proposed European Space Agency x-ray observatorymission to be located on the International Space Station.Specifications include a 10 m diameter mirror with 25–50 mfocal length, spatial resolution of 1 arcsecond, covering theenergy range 0.1–100 keV.


X-Ray Astronomy E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

X-Ray AstronomyX-ray astronomy is an achievement of the space age asthe Earth’s atmosphere is completely opaque at photonenergies beyond the ultraviolet region.

In 1949 the first X-RAYS from the solar corona weredetected by a Geiger counter on a V-2 rocket. In 1962 thediscovery of the first x-ray source outside the solar systemfollowed—SCORPIUS X-1. With the same rocket experimentan apparently diffuse x-ray background was found. Sincethese early days x-ray astronomy has developed at anenormous pace. Today, we know more than 150 000x-ray sources in the sky and they include almost allastrophysical species—from the nearby comets to the mostdistant quasars at the edge of the universe, from the tinyneutron stars to the clusters and superclusters of galaxiesas the largest physical formations in the cosmos. Somespecies radiate most of their power in x-rays, for instanceblack holes or neutron stars accreting matter from a binarycompanion, as well as supernova remnants and single,cooling neutron stars.

Many of the known objects shine in x-rays becausethey are hot—having temperatures of millions to billionsof kelvins. Other emission mechanisms are synchrotronradiation of extremely energetic electrons spiraling inmagnetic fields or the inverse Compton effect which occurswhen high-energy electrons scatter at low-energy photons,e.g. from stellar light or from the 2.7 K cosmic backgroundradiation. In any case, the emission of x-rays points toextreme physical conditions in the source region. Also,these x-rays carry information which is not available fromobservations in other spectral bands. In this article weshall give a brief summary of the development of x-ray astronomy, describe the evolution of experimentaltechniques and x-ray space missions and highlight some ofthe results which have had a major impact on astrophysics.

The beginnings with sounding rocketsLike GAMMA-RAY ASTRONOMY, the field of x-ray astronomywas pioneered by physicists. Herb Friedman, whodetected the first solar coronal x-rays in 1949, had beenworking previously in ionospheric physics; Giacconi andRossi, who led the group discovering Scorpius X-1 andthe x-ray background in 1962, were nuclear and cosmicray physicists.

The instruments used in these pioneering experi-ments were rocket-borne Geiger counters and spectral dis-crimination was achieved by means of windows and fil-ters. A little later proportional counters became the stan-dard instrument, working in the 1–10 keV band and havinga modest spectral resolution (∼20% at 5 keV). Observationswere first performed by scanning the sky by spinning orprecessing the rocket. Later on more sophisticated attitudecontrol systems allowed pointed observations to be madewith greatly improved sensitivities.

The first discoveries of cosmic x-ray astronomy weretotally unexpected. Sco X-1 showed a luminosity manyorders of magnitude larger than that of the Sun. It

soon became clear that the mechanism by which suchsources are powered was the infall of matter into thedeep gravitational potential well of a NEUTRON STAR (as inSco X-1) or a BLACK HOLE. Most of the few dozen sourcesfound during the rocket era, i.e. until the end of the 1960s,were of this kind. The other major class was SUPERNOVA

REMNANTS. The identification of the CRAB NEBULA as a brightx-ray source by means of a lunar occultation was one ofthe early highlights. Other important discoveries of thisera were the x-ray pulsations of the CRAB PULSAR and thex-ray emission from the active galaxy M87 at the center ofthe Virgo cluster.

After the advent of x-ray satellites rocket-borneexperiments with their short observation time (∼5 min)were generally not competitive any more. However, theycontinued to play a role as a test bed for instruments to beflown on satellites or for special observations, in particularthose ones of short lead time (e.g. on the famous SUPERNOVA

1987A).

Hard x-ray balloon experimentsThe rocket experiments had shown that the spectra ofcompact x-ray sources are quite hard, extending muchbeyond 10 keV. This opened up the possibility of usingballoon-borne instruments as the atmosphere becomestransparent for altitudes above 40 km at high x-rayenergies (>20 keV). Balloon flights offered the advantageof long-duration observations (up to 100 h) and the crystalscintillation counters used in these experiments allowedspectra to be taken up to ∼500 keV. Balloon observationswere pioneered by the MIT group in the 1960s. Importantresults of these early years were the measurement of theCrab spectrum to high energies (∼500 keV), the discoveryof variability in GX1 + 4 and the detection of a 20 min fluxfrom Sco X-1. This field of research culminated in the 1970swith experiments of the MIT–Leiden and the Tubingen–MPE groups which flew very large detectors operating inthe pointing mode. Highlights of these activities werethe observation of a lunar occultation of the Crab, thediscovery of a spectral break in the Cyg X-1 spectrumcorresponding to a temperature of about ∼20× 106 K andthe discovery of cyclotron lines at∼40 keV in the accretingneutron star Her X-1, allowing the first direct measurementof its polar magnetic field (∼5× 1012 G).

X-ray astronomy satellitesMost significant progress in x-ray astronomy came withthe advent of satellite observatories. Their ancestor, thefirst satellite entirely devoted to x-ray astronomy, wasUHURU. Launched in 1970 it was a spinning spacecraft witha simple, but very powerful, instrument package: an arrayof proportional counters of 840 cm2 area working in the 2–20 keV band. It performed the first all-sky survey andlocated 339 objects, mostly X-RAY BINARIES and supernovaremnants, showing a strong clustering near the galacticplane. At fainter flux levels an isotropic distribution ofSeyfert galaxies and clusters of galaxies was found.



The discovery of two pulsating and eclipsing x-ray binary systems, Cen X-3 and Her X-1, proved inan impressive way that these systems contained matter-accreting neutron stars. The discovery of short timevariability of Cyg X-1 by Uhuru led to follow-up rocketexperiments yielding accurate source positions whichenabled the optical counter-part to be identified and themass of the accreting compact object to be determined.This mass turned out to be larger than the limiting massof a neutron star. Thus it had to be a black hole.

Uhuru also marked the beginning of an impressiveseries of satellites with ever-increasing capabilities listedin table 1. We cannot discuss them in detail here, but ratherdescribe the main directions of developments.

Imaging x-ray telescopesImaging of x-rays is possible in different ways. Earlysolar observations used pinhole cameras and Fresnel zoneplates, but the sensitivity of such devices is insufficient forstudies of the rather weak cosmic x-ray sources. Therefore,the standard X-RAY TELESCOPES use Wolter optics whichconsist of paraboloidal–hyperboloidal mirrors reflecting x-rays under grazing incidence (see GRAZING INCIDENCE OPTICS).The first x-ray telescopes of this kind using x-ray film as adetector were used for solar observations from rockets andSkylab. The first x-ray satellite carrying a Wolter telescopewas the Einstein observatory launched in 1979. It providedimages with ∼10 arcsec resolution and represented a realbreakthrough, putting x-ray astronomy on equal footingwith optical astronomy.

X-ray sky surveysAll-sky surveys have traditionally been a foundation ofastrophysical research, and in the era of multiwavelengthastronomy their importance has dramatically increased.Such surveys provide an unbiased view of the sky, theydeliver large homogeneous samples of objects and theyallow rare species to be discovered. Cross correlating thesurveys from different wavelength bands—e.g. optical andx-rays—is a very effective method to select sources of acertain type, e.g. active galactic nuclei and quasars.

The Uhuru and Ariel V surveys have revealed∼350 sources in the standard x-ray band (2–6 keV). Thesubsequent HEAO-1 sky survey was not much moresensitive (840 sources) but widened the energy bandconsiderably (0.1–200 keV). The limitations of all thesecollimated counter surveys in terms of angular resolution(<1 deg2) and sensitivity were overcome by the German-led ROSAT which performed the first all-sky survey in softx-rays by sweeping an imaging Wolter telescope across thewhole sky. Although it took only half a year of the morethan 8 yr of ROSAT’s life this survey discovered 80 000x-ray sources and located them with 25 arcsec resolution.In addition, the survey provided a complete map of thediffuse x-ray emission with 12 arcmin resolution (figure 1).

ROSAT PSPC ALL-SKY SURVEY Soft X-ray Background

red: 0.1-0.4 keV green: 0.5-0.9 keV blue: 0.9-2.0 keV

ROSAT PSPC ALL-SKY SURVEY Sources

Energy range: 0.1 - 2.4 keV

Aitoff Projection Galactic II Coordinate System

Figure 1. The x-ray sky as known at the end of the 20th centuryon the basis of the ROSAT All Sky Survey. Top: distribution ofthe ∼19 000 brightest x-ray sources; Bottom: diffuse emissionfrom hot interstellar matter heated by supernova explosions.This figure is reproduced as Color Plate 61.

Recent highlights of x-ray astronomyA rather detailed account of the development of x-rayastronomy until∼1990 with a description of missions andtheir results can be found in Bradt et al (1992). Here wewant to highlight the progress made in the 1990s by thepowerful x-ray telescopes on ROSAT and ASCA.

In addition, there have been two very successfulrecent missions which must be mentioned here. The ROSSITiming Explorer (RXTE) carrying large-area collimatedcounters covering a wide energy range (2–200 keV) hasdeepened our understanding of compact sources by hightime resolution and spectral studies. One of its mostexciting results was the discovery of 2.75 ms pulsationsoccurring during the bursts of the low-mass x-ray binary(LMXB) 4U 1728-34. The data suggest that the pulsationcame from the rotation of a thermonuclear hotspot on thesurface of the neutron star.

In another LMXB which had been previouslyidentified as an X-RAY BURSTER by BEPPOSAX, RXTE found2.5 ms pulsations in the persistent flux which must bedue to the rotation of the neutron star. This is the first



Table 1. X-ray astronomy satellites 1969–2000. Adapted from Charles and Seward (1995) updated January 2000.

Satellite Country Launch Demise Type

Vela 5 A,B USA May 1969 June 1979 Scanning, small scintillation counter,gamma-ray range

Uhuru USA December 1970 January 1975 Scanning, proportional countersOSO-7 USA September 1971 May 1973 Scanning, proportional countersCopernicus USA–UK August 1972 February 1981 Pointed, x-ray telescope (non-imaging)ANS Netherlands–USA August 1974 July1976 Pointed proportional counters, Bragg

crystal spectrometerAriel-V UK October 1974 March 1980 Scanning, rotating modulation collimators

(RMCs) + proportional countersSAS-3 USA May 1975 April 1980 Scanning, RMCOSO-8 USA June 1975 October 1978 Scanning, proportional and scintillation

counters, Bragg crystal, polarimeterHEAO-1 USA August 1977 January 1979 Scanning + short pointings, proportional

and scintillation counters, RMCEinstein USA November 1978 April 1981 Pointed, imaging x-ray telescope, Bragg crystal,

transmission gratingsHakucho Japan February 1979 April 1985 Scanning, RMCsTenma Japan February 1983 ∼1985 Gas scintillation proportional counter (GSPC),

all-sky monitorEXOSAT ESA May 1983 April 1986 Pointed, imaging x-ray telescope, large

proportional counters, GSPCGinga Japan–UK February 1987 October 1991 Pointed, proportional countersKvant USSR–UK– June 1987 – Pointed, GSPC, coded mask, scintillation counter

Netherlands–GermanyGranat USSR–Russia December 1989 August 1999 Pointed, coded masks, all-sky monitorROSAT Germany–UK–USA June 1990 December 1998 Scanning, pointed imaging x-ray and

EUV telescopesASCA Japan–USA February 1993 – Pointed, imaging x-ray telescopes, imaging GSPCRXTE USA December 1995 Pointed, large proportional counters,

scintillation counters, all-sky monitorsBeppoSAX Italy–Netherlands April 1996 Pointed imaging x-ray telescope, coded mask,

scintillation counterChandra USA June 1999 Pointed imaging x-ray telescope, spectrometersXMM-Newton ESA December 1999 Pointed imaging x-ray telescope, spectrometers

and so far only LMXB showing both persistent pulsationsand bursts. BeppoSAX is an Italian–Dutch satellite withinstruments covering a wide energy band (0.1–200 keV).By discovering the x-ray afterglows of gamma-ray burstsit pointed the way to the solution of an old puzzle, thephysical nature of gamma-ray bursts.

Highlights from ROSAT and ASCAROSAT and ASCA have complementary properties.ROSAT carries a large x-ray telescope with a position-sensitive proportional counter (PSPC) providing moderatespectral resolution (∼40%) in the 0.1–2.4 keV band.With its High Resolution Imager (HRI), a microchannelplate detector ‘black-and-white’ images with 5 arcsecresolution can be taken. The telescopes of ASCA coverthe energy band 0.5–10 keV with CCD detectors andimaging gas scintillation proportional counters havingsuperior energy resolution (∼20%), but worse angularresolution (∼3 arcmin), compared with ROSAT. Bothsatellites have been used by many astrophysicists to studya wide variety of problems. The numbers of scientificpublications resulting from ROSAT andASCArun to about4000 and 1600, respectively, covering almost all fields

of astrophysics. In the following a few highlights arepresented.

ROSAT took the first x-ray picture ever of the MOON

(figure 2(a)). The Sun-lit side of the Moon containsa uniform brightness distribution as in optical light.This is due to solar coronal x-rays undergoing Thomsonscattering in a very thin layer of the lunar surface. ThePSPC spectrum shows a broad spectral bump at 0.6 keVwhich is due to fluorescent resonance scattering by oxygenof the minerals of the lunar surface layers. The effectivereflectivity of the Moon in the ROSAT band is only∼0.01%;this means that the Moon behaves as a black body at x-rayenergies. At the same time it casts a shadow on the ‘diffuse’x-ray background (see later section). The small flux of x-rays apparently coming from the dark side of the Moonis probably produced in the Earth’s upper atmosphereby charge-exchange processes of solar wind ions. Thisis the same mechanism which is responsible for the x-rayemission from cometary comas (see below).

COMETS are cold objects that have been described asdirty snowballs. Therefore the discovery with ROSATof x-rays from COMET HYAKUTAKE on 27 March 1996 wassurprising to many scientists (figure 2(b)). Later, another



(a) (b)

Figure 2. Solar system objects in x-rays: (a) the Moon reflecting solar coronal x-rays and casting a shadow on the x-ray sky; (b) x-rayemission from comet Hyakutake superimposed on an optical amateur photograph.

four comets were found in the ROSAT All Sky Surveyarchive, including comet Levy. In total about a dozencomets have been detected by now in x-rays, mostly withROSAT. Various physical processes have been proposed toexplain the observed extended x-ray emission, includingscattering of solar x-rays by cometary dust, x-raysproduced by collisions of cometary and interplanetarydust particles and bremsstrahlung x-rays from electronsaccelerated at the shock between the solar wind and thecoma. However, all these mechanisms fail to explain theobserved characteristics. The most successful and widelyaccepted model suggests that charge exchange betweenhighly charged ions (such as C5+, C6+, O6+ and O7+) inthe solar wind and neutral particles (such as water) in thecometary coma is the dominant source of the observed x-ray emission.

ROSAT has given many exciting results about starsof all types. The study of a complete sample of starsof solar type has revealed the existence of a sharp lowerbound to the x-ray flux, measured at the surface of the star.Interestingly, this minimum stellar x-ray flux is identicalto the flux observed in the coronal holes of the Sun. Thisresult suggests that the stars of minimal x-ray flux arecompletely surrounded by stellar analogs of solar coronalholes.

With ground-based optical follow-up observationsof unidentified ROSAT All-Sky Survey sources, severalhundred new T TAURI STARS have been identified basedon Hα emission of hydrogen and lithium absorption.Surprisingly, many new T Tauri stars have been found faroutside regions of ongoing star formation. Such ‘off-cloud’T Tauri stars must have been either ejected from theirbirthplaces in the clouds with high velocities or formed insmall cloudlets which have largely dispersed since then.

Recently, deep ROSAT observations have led to thediscovery of young (<106 yr) brown dwarfs showing x-ray emission. The x-ray properties of these brown dwarfs,in particular their x-ray to optical light ratio, are similar

to those of nuclear burning stars of low mass. Thismay suggest that young brown dwarfs have hot coronaeresponsible for the x-ray emission just like nuclear burningstars of low mass.

Early in the ROSAT mission a number of objectswere discovered emitting extremely soft x-rays. They arevery luminous and show temperatures of a few hundredthousand kelvins. It turned out that these sources are aspecies which had been predicted to exist but which hadnot been found before. They are WHITE DWARFS in binarysystems accreting matter from their companions at a ratejust sufficient to sustain steady nuclear burning on thesurface of the white dwarf. Thus, they represent a uniquesituation in which steady nuclear burning is observed atthe surface of a compact star.

Massive stars explode giving rise to a supernovawhen the nuclear fuel in their cores is exhausted. Thecore collapses to a neutron star or a black hole whilethe shell of the star is expelled in a giant explosion. Alarge fraction of the kinetic energy is converted by shocksinto high-temperature x-ray emitting plasmas. Supernova1987A was ROSAT’s first-light target on 16 June 1990, butit turned out to be too faint to be seen at that time. Itsremnant was first discovered in soft x-rays with ROSATin 1992 and has steadily become brighter since then. Alarge increase is expected to occur in the near futurewhen the shock reaches the high-density regions of thered giant wind. In total, some 200 supernova remnantshave been found with ROSAT. Three of them are clusteredin the Vela region (figure 3). One is the VELA SUPERNOVA

REMNANT, which, at a distance of 1500 ly, is one of theclosest supernova remnants. Its diameter is about 200 ly, itsage about 20 000 yr. Protrusions discovered with ROSATat the periphery of the shell are probably produced byfragments of the exploding star; x-ray spectroscopy withASCA has revealed that they show different chemicalcompositions (figure 3(a)). The Puppis A remnant atthe north-western rim of the Vela supernova remnant



ROSAT PSPC ROSAT PSPCVELA SNR PUPPIS APUPPIS A

(a) (b) (c)

Figure 3. The Vela complex of supernova remnants: (a) Velasupernova remnant with the Vela pulsar and Puppis A;(b) Puppis A with its central source, probably the remnantneutron star; (c) A young, ∼680 yr old supernova remnant,superimposed on Vela supernova remnant, showing titanium-44emission. This figure is reproduced as Color Plate 62.

is located at a much larger distance (figure 3(b)). It isyounger and has a much higher temperature than theVela supernova remnant. Surprisingly, a third supernovaremnant has been discovered recently in the Vela complexwith ROSAT. Radioactive titanium-44 has been detectedfrom it with the Compton Gamma-Ray Observatory(figure 3(c)). Titanium-44 is produced in supernovaexplosions and has a mean lifetime of only 90 yr, whichmeans that the remnant is very young (∼680 yr). Actually,this supernova remnant looks like a twin of SN 1006. Likethe latter it must have been very bright at maximum (abouta quarter of the full Moon) and it is unclear why it has notbeen recorded in the Chinese and Japanese annals.

Neutron stars shine in different ways. If they arehighly magnetized and rapidly rotating, they appearas radio PULSARS. These objects emit beamed radiationproduced by high-energy electrons (and positrons) thatare accelerated in their magnetospheres. In about halfa dozen young pulsars, optical and gamma-ray pulseshave also been seen. With ROSAT and ASCA, 34 radiopulsars have been detected through their magnetosphericx-ray emission, including the 89 ms Vela pulsar. Thecharacteristic features of this magnetospheric radiationare power law spectra and sharp pulses. Four of theradio pulsars seen with ROSAT, including the Vela pulsar,exhibit an additional thermal spectrum corresponding toa temperature of the order of 106 K, which is interpretedas the thermal radiation from the surface (photosphere)of the neutron star. A few point sources discovered nearthe centers of young supernova remnants also show verysoft x-ray emission, which must be attributed to neutronstar surface emission as well. A prominent example is thecentral source in Puppis A (figure 3(b)).

Since the ages of radio pulsars and supernovaremnants can be determined quite well, one can usethese observations to test cooling models of neutron stars.The result is that standard cooling models describe theobserved temperatures and luminosities quite well. Rapidcooling as expected in the presence of a Bose–Einsteincondensate of pions or kaons in the neutron star core canbe excluded for the observed objects.

A few x-ray sources have been found in the ROSATdata showing very soft thermal spectra, (temperatures be-low a million kelvins) and very faint optical counterparts(25m–26m). These objects might be either single neutronstars accreting interstellar matter or, more likely, old (105–106 yr) cooling neutron stars whose supernova remnantshave already vanished. The importance of all these obser-vations lies in the fact that the surfaces of these tiny stars(with radii 10 km) have now become visible. Future x-ray spectroscopy should make it possible to measure theirenormous gravity as well as their radii, which depend onthe physical properties of matter at supranuclear densities.

The ROSAT deep survey of the Andromeda galaxyled to the discovery of 550 x-ray sources, comparable innumber with the bright x-ray sources in our own galaxyfound by Uhuru. As in the Milky Way, the brightest ofthem are young supernova remnants or binary systemswith neutron stars or black holes accreting matter from acompanion. These bright source populations have beenstudied with ROSAT in many galaxies. In addition, thehot interstellar medium, which is heated by supernovaexplosions, has been investigated.

A small fraction of all galaxies have an ACTIVE GALACTIC

NUCLEUS (AGN) emitting huge amounts of energy in allspectral bands. The radiation is variable, indicating that itis emitted from a small region, generally not more thana few light-weeks or light-months across. Such AGNsoften display jets, originating in their core. It is generallybelieved that the central engine is a supermassive blackhole swallowing matter at a high rate. QUASARS arethe most extreme representatives of the class of AGNsources. ASCA spectroscopy of the AGN MCG-6-30-15and a few other sources has led to the discovery of ironKα emission lines with broad and asymmetric profiles.This profile is most probably caused by gravitational andrelativistic Doppler shifts of the rapidly rotating matterin the accretion disk near the central black hole. TheseAGNs are very bright x-ray sources, and more than 50%of all 150 000 ROSAT sources belong to this class. Becauseof their enormous brightness, they can be detected at largedistances or redshifts. In the ROSAT Deep Surveys witha total accumulated observation time of 2 weeks, about1000 sources are detected per square degree in the sky.Optical spectroscopy has shown that most of them arequasars and other AGNs at cosmological distances. TheROSAT Deep Surveys have also answered one of the oldestquestions of x-ray astronomy, namely the origin of theextragalactic x-ray background: at least, in the soft x-ray band, around 1 keV, some 80% of the backgroundhas been resolved into discrete sources, mostly quasars.An important current question is what evolved first, thegalaxies or the supermassive black holes.

Galaxies are not distributed randomly in the universebut form clusters or groups of galaxies consisting of adozen to thousands of members which are gravitationallybound. With diameters of millions of light-years, they arethe largest physical objects in the universe. As early as1932, Fritz Zwicky found that CLUSTERS OF GALAXIES must



Figure 4. X-ray image of the coma cluster showing the mergingwith a smaller cluster of galaxies.

contain much more gravitating mass than accounted bythe total visible mass of all member galaxies. Actually, theamount of ‘DARK MATTER’ had to be of the order of 95% ofthe total cluster mass.

One early surprise of x-ray astronomy was thedetection by Uhuru of large quantities of hot plasma inclusters of galaxies shining in x-rays. Later Einstein andROSAT observations showed that the mass of the hotplasma is typically a factor of 4 or 5 larger than that ofthe galaxies and represents some 20% of the cluster mass.In other words, a large fraction of the ‘dark matter’ turnedout to be hot, visible only in x-rays and not in opticallight. ASCA spectroscopy has allowed us to measurethe heavy element (iron) abundance of the hot matter inclusters, which is typically only one-third of the so-calleduniversal abundance. This can be explained by the infallof primordial matter onto the clusters.

Many clusters show double structures, indicating themerging of two clusters; others exhibit complicated innerstructures which must be due to earlier merging processes(figure 4). Thus, the x-ray images reveal how these largeobjects evolve with cosmic time.

In total, several thousand clusters have been foundwith ROSAT. About 1500 of them have been opticallyidentified and have known redshifts, and thus knowndistances. From this information one can derive theevolution of the cluster population with time. Acomparison with simulations shows that the observedtime dependence of the cluster population is significantlysmaller than expected in a universe having the criticaldensity. Actually the matter density inferred from cluster

evolution is only about one-third of this critical densitywhich is necessary to close the universe.

The futureDuring the years to come enormous progress is expectedin x-ray astronomy owing to two new very powerful andcomplementary x-ray telescope missions: CHANDRA andXMM-NEWTON. Both satellites carry x-ray CCD detectors anddispersive spectrometers for high-resolution spectroscopy.The special strength of Chandra, launched in summer1999, is its high angular resolution (∼0.5 arcsec) whichallows us to resolve fine structures such as jets and togo at least a factor of 1000 deeper than the ROSAT deepsurveys did. On the other hand XMM-Newton, launchedin December 1999, provides a very high collecting power(∼4–10 times that of Chandra, depending on energy) atmoderate angular resolution (∼10 arcsec). It will be verypowerful for spectroscopic and time variability studies.The first results of these two powerful missions are verytantalizing.

Plans for the future thereafter are already quiteconcrete. NASA is discussing a fleet of four x-raytelescope satellites called ‘Constellation’ to be launchedin∼2007. They can observe the same object with differentinstruments simultaneously or point to different regionsof the sky. ESA’s XEUS (x-ray mission for spectroscopy inan evolving universe), to be launched after 2010, foresees agiant x-ray telescope with a huge mirror system and focalinstruments sitting on two separate satellites, which meansthat pointing requires orbital maneuvers. The collectingarea of the telescope (∼30 m2) is a thousand times thatof ROSAT and about a hundred times that of XMM orRXTE. It should be able to penetrate into the ‘dark ages’of our universe at redshifts of 5–10, where galaxies andsupermassive black holes have been formed.

BibliographyBradt H V, Ohashi T and Pounds KA1992 X-ray astronomy

missions Ann. Rev. Astron. Astrophys. 30 391–427Charles P and Seward F 1995 Exploring the X-Ray Universe

(Cambridge: Cambridge University Press)

Joachim Trumper


X-ray Binary Stars E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

X-ray Binary StarsX-ray binary stars are luminous (∼1036−38 erg s−1) x-ray sources that consist of a neutron star or black holecompact object accreting material from a close (usually≤1R) orbiting companion star which, in most cases, hasproperties similar to those of normal main sequence stars.Such x-ray binaries are rare (only ∼200 are present inour Galaxy and are distributed in the Galactic Plane, theGalactic Bulge and a dozen or so as GLOBULAR CLUSTER X-

RAY SOURCES, see figure 1), whereas the related (and inmany respects similar) cataclysmic variables (where thecompact object is a white dwarf) are much more common.The extremely high intrinsic luminosity of x-ray binaries(the brightest are at the ‘Eddington Luminosity, a level atwhich radiation pressure limits the rate at which mattercan accrete’ for a 1M star) accounts for their discoveryin the early days of x-ray astronomy (via rocket-basedexperiments in the 1960s) and their domination of the fieldfor most of the next two decades.

X-ray binaries can be divided into two basic classesof objects according to the mass of the mass-losing (ordonor) star. The division is remarkably clean with those≥5M classified as ‘high mass x-ray binaries’ (HMXBs) andthose ≤1M as ‘low mass x-ray binaries’ (LMXBs). Thenature of the mass transfer from the donor also dividesbetween these classes, with the (radiation driven) stellarwind dominating in HMXBs and Roche-lobe overflowbeing the principal mechanism in LMXBs (although thereare circumstances in which a combination of the two canoccur). The x-ray characteristics also depend on the natureof the compact object. High magnetic field neutron starsproduce x-ray pulsators, whereas lower magnetic fieldsmake it much harder to detect the spin of the neutronstar, although ‘x-ray bursts’ can occur. Note that coherentpulsations and bursts are never seen in black hole systems.

X-ray binaries are extremely important objects asthey demonstrate that stellar and binary evolution canproduce compact objects via supernova explosions inwhich the binary survives. They currently providethe only circumstance in which stellar-mass black holemasses can be derived within our Galaxy (see BLACK HOLE

CANDIDATES IN X-RAY BINARIES) as well as a majority of theaccurately known neutron star masses. They are alsoa well-defined and constrained laboratory for studyingaccretion onto highly degenerate matter and the effect ofultra-high magnetic fields, and for testing the effects ofgeneral relativity in the strong gravity limit.

High mass x-ray binariesX-ray pulsarsThe very first x-ray astronomy satellite, UHURU, discoveredX-ray pulsations at 4.8 s and 1.2 s in Cen X-3 and HerX-1, respectively, and also found that those pulsationswere modulated (and eclipsed) on their orbital periods of2.1 d and 1.7 d (see figure 2). These results immediatelyestablished the binary star model for these sources beyondany doubt. In combination with optical spectroscopy of

Figure 1. Distribution of x-ray binaries in the Galaxy. Top:HMXBs, associated with young, massive stars in the spiral arms.Bottom: LMXBs, an older population concentrated in theGalactic Bulge (also including globular clusters as opensymbols).

the donor star (relatively bright, early-type stars for bothsources), and the fact that they are eclipsing, they allowfull orbital solutions and hence accurate masses for bothdonor and compact object (see table 1). Furthermore theduration of the eclipse provides an accurate estimate of thesize of the donor which can be compared with the size ofits Roche lobe.

Spinning up and downThe spinning neutron stars are orbiting their massivecompanions and accreting material from the denseand (presumed) uniform stellar winds. This physicalsituation is essentially identical to Bondi–Hoyle accretionas calculated for a single object moving and accreting froma uniform interstellar medium. Matter is accreted onto theneutron star from within a cylinder of radius racc, definedas the point at which the kinetic energy of matter in thewind is equal to the potential energy due to the neutronstar. If the wind’s velocity is vw, and the orbital velocity ofthe neutron star is v then the fraction accreted is given by(Mns/Msec)

2(v/vw)4[1+ (v/vw)

2]−3/2 and is usually∼10−3 to10−5 for typical HMXBs.

The angular momentum of this material will cause itto form a small accretion disk around the compact object,but its final capture by the neutron star is controlled by theintense magnetic field at the magnetospheric radius wherethe ram pressure of the (presumed spherical) infall isbalanced by magnetic pressure. At this point the accretingangular momentum is transferred to the spinning pulsar,which then spins up at a rate P that is broadly proportional



Figure 2. Uhuru observations of the Doppler delays in the 4.8 s x-ray pulsations from Cen X-3 as a function of the 2.1 d orbital period.The x-ray light curve (below) shows the eclipse and hence geometry of the binary.

Table 1. Orbital parameters and masses for eclipsing HMXBs.

Pulse period Orbital period f (M) Mns MsecSource (s) (d) (M) e i (M) (M)

SMC X-1 0.72 3.89 10.85 <0.000 04 65 1.1 18Her X-1 1.24 1.70 0.85 <0.0001 > 79 1.5 2.3Cen X-3 4.82 2.09 15.34 <0.0016 70 1.21 20.5LMC X-4 13.5 1.41 9.86 0.006 65 1.5 15.8Vela X-1 283 8.96 19.74 0.09 >74 1.9 23.54U1538-52 530 3.73 11.8 <0.06 68 1.1 16.4

Figure 3. Approximately five-year history of the variations in the spin rate of the Cen X-3 pulsar as monitored by BATSE on CGRO.The general trend shows that on average it is spinning up, but there are significant fluctuations about this trend indicating that it canspin down as well as up.

to the observed x-ray luminosity (since this is directlyrelated to the accretion rate).

Figure 3 shows the variation of spin period in Cen X-3as observed by CGRO over almost five years. While thegeneral trend is a spin-up, it is clear that the actual processinvolves periods of (almost equal) spin-up and spin-down.It might be expected that the spin-up rate would decline

with x-ray luminosity if the mass transfer rate were todecline, but how does it manage to reverse sign? Thisrequires that material is being accreted whose torque isreversed with respect to the ‘normal’ expectation, andimplies that the disk rotation itself has reversed. Detailsof this process are unclear, but could be due to non-uniformities in the stellar wind.



Figure 4. Optical spectrum of Sco X-1 which is dominated by continuum from the disk and x-ray driven high-ionization emission lines.

Table 2. Be x-ray binary parameters.

Spectral Spin period Orbital period LX(peak)Source type V (s) (d) e (erg s−1)

A0538-66 B2IIIe 15 0.069 16.65 ∼0.7 1039

4U0115+63 OBe 15.5 3.6 24.3 0.34 8× 1036

V0332+53 Be 15.3 4.4 34.25 0.31 1038

A0535+26 O9.7IIIe 9 104 111 ∼0.3 2×1037

GX304-1 B2Ve 14 272 133 ? 3× 1036

4U1154-61 B1Vpe 9 292 188 ? 1036

X Per O9.5ep 6 835 ∼580 ? 1034

Note also that spin-up will only continue untilthe neutron star is rotating so fast that material atthe magnetospheric radius would have to exceed theKeplerian speed at this radius in order to enter themagnetosphere (and hence would be ejected). Thus amagnetospheric ‘barrier’ then exists to counter any furtheraccretion until either the neutron star has spun down orthe mass transfer rate significantly increases.

Cyclotron linesThe magnetic fields of neutron stars that are x-ray pulsarsare inferred to be∼1012 G. Such fields control the accretionof material onto the neutron star’s surface, leading to well-defined hot spots. x-rays are emitted either from the hotspot or from a shock immediately above the surface (within∼102−3 m), producing either pencil-beam or fan-beamemitting patterns. The very strong field will then leadto quantization of the energy levels within the accretioncolumn, which produces resonant scattering cross sectionsand hence cyclotron emission and absorption features.In fact, cyclotron emission lines have only been seen in

magnetic white dwarfs in the infrared, whereas cyclotronabsorption features (most in the ∼12–40 keV range) havebeen detected in a dozen x-ray pulsars, implying magneticfields of ∼2–3× 1012 G.

Be star transientsMany of the x-ray pulsars have long orbital periods withB-type companions which exhibit emission lines. Theserepresent a sub-class of HMXBs and are associated withthe rapidly rotating Be stars, so-called because of theirstrong and variable emission lines (usually hydrogen)superposed on otherwise normal B star spectra. A numberof transient x-ray sources (most of them x-ray pulsators)have been identified with Be stars, establishing that bothsingle and binary Be systems exist. The Be transientshave rare outbursts (∼ hundreds of days) and long orbitalperiods (many weeks). Some (e.g. A0535+26, GX304-1)have poorly determined orbital periods>100 d, even withthe accurately determined x-ray pulsation.

The Be phenomenon is interpreted as being due tothe presence of an equatorial ring around the B star as



Figure 5. Schematic diagram of an LMXB showing the effects ofdisk shadowing and obscuration by the secondary star. Abovean inclination of about 80, LMXBs are seen only as ADCsources. Below about 75, orbital modulation can still be seen inthe form of regular dips, but these tend to evolve in form fromcycle to cycle.

Figure 6. Continuous monitoring of X1916-053 by EXOSATreveals both the regular x-ray ‘dipping’ due to azimuthalstructure in the accretion disk, and the way in which thisstructure can evolve from orbit to orbit.

a result of its rapid rotation. Significant mass is lost viathis ring and this material can be accreted by the orbitingneutron star, thereby producing x-rays. Almost all of theBe transients with measured orbits show significant orbitaleccentricities (see table 2) and this may be related to thetransient x-ray outbursts. An example here is A0538-66in the LMC, which showed dramatic x-ray and opticaloutbursts on a 16.6 day cycle which was stable for atleast 50 years (based on archival plate material), eventhough these outbursts did not always occur. At theirpeak these outbursts exceeded 1039 erg s−1 and yet thecompact object was shown to be a neutron star based on anobserved 69 ms pulsation. The most likely interpretationis that of a highly eccentric orbit (e ∼ 0.7) in which theneutron star interacts (in a complex way) with the Be ring atperiastron. Longer term (hundreds to thousands of days)variations are present in almost all Be systems, the natureof which is unclear. A recent extended (18 year) period

of inactivity in A0538-66 has been attributed to the effectof the magnetospheric barrier given the very fast spinperiod of the neutron star. This could only be overcomeby another interval of very high mass transfer from thesecondary.

Low mass x-ray binariesThe brightest (and first discovered) extra-solar x-raysource is the x-ray binary Sco X-1, which was opticallyidentified with an unremarkable 13th magnitude blue starin 1967. However, proof of its binary status was extremelyhard to establish and this was not accomplished untilthe mid 1970s. This was due to the optical light beingcompletely dominated by the x-ray heated accretion disk,making the secondary undetectable (see figure 4). UnlikeHMXBs, the principal mass-transfer mechanism in LMXBsis Roche lobe overflow from the cool, low-mass secondarydirectly into the accretion disk. Furthermore, the thicknessof the accretion disk itself can render a direct view ofthe compact object impossible at all phases (and not justwhen it is behind the secondary star), thereby makingevidence for binarity less obvious than in HMXBs. Thisis demonstrated in figure 5 where the effect of shadowingby the disk leaves only a narrow range of viewing angles atwhich an orbital modulation can be seen. X-ray irradiationof the disk has been demonstrated in simultaneous multi-wavelength observations of X-RAY BURSTERS, which show theoptical burst lagging the x-ray as a result of reprocessingof x-rays in the accretion disk and heated face of thecompanion.

A significant subset of LMXBs are globular cluster x-ray sources. Indeed, they are grossly over-representedin globular clusters when compared with the density ofLMXBs in the Galaxy as a whole. Furthermore, all theglobular cluster sources (see table 3) exhibit type I x-raybursts, and recent HST studies of the extremely crowdedUV/optical fields have revealed that several of thesesources are ultracompact binaries with orbital periods<1 hour (including X1820-30 which has the shortest orbitalperiod of any known object). It is likely that stellarinteractions within the cluster core lead to the formationof such exotic objects.

X-ray dippersA number of LMXBs display sudden ‘dips’ in their x-ray light curves which occur at particular orbital phases(around ∼0.8). These are attributed to extended verticalstructure in the edge of the accretion disk at the pointwhere the mass transfer stream impacts the disk. Thismaterial temporarily obscures our direct view of the x-ray source, absorbing the x-rays more effectively at lowerx-ray energies (a signature of absorption which confirmsthe process). Extended observations of these x-raydippers shows that the vertical disk structure evolves fromorbit to orbit and on much longer timescales, sometimesdisappearing altogether, at other times producing dipfeatures throughout the orbital cycle (see figure 6).



Table 3. Properties of LMXBs.

OrbitalSource period (h) X-ray type V MV Notes

X1820-303 0.19 Burster >19 globular cluster (NGC 6624), degenerate companion4U 1850–087 0.34 Burster 21 5.6 globular cluster (NGC 6712), degenerate companionX1626-673 0.7 Burster, pulsar 18.5 7.7s pulsar, degenerate companionX1832-330 0.73 Burster 19.4 globular cluster (NGC 6652), degenerate companionX1916-053 0.83 Burster, dipper 21 5.3 x-ray and optical periods different, degenerate companionJ1808.4-3658 2.0 Burster, pulsar, transient 16.5–20 millisecond (2.5 ms) pulsar, ∼0.1M companionX1323-619 2.9 Burster, dipperX1636-536 3.8 Burster 17 1.3X0748-676 3.8 Burster, dipper, transient 17 1.4 no declineX1254-690 3.9 Burster, dipper 19X1728-169 4.2 17X1755-338 4.4 Dipper 19X1735-444 4.6 Burster 17.5 2.2X2129+470 5.2 ADC 16.5 now ‘off’; triple?X1822-371 5.6 ADC 15.5X1658-298 7.2 Burster, dipper 18.3A1742-289 8.4 Burster, transient eclipsingX1957+115 9.3 18.7X2127+119 17.1 Burster, ADC 15.5 1.0 eclipsing, globular cluster (M15)Aql X-1 19 Burster, transient 21.6 2.9 frequent outbursts, kHz QPOSco X-1 19.2 Prototype LMXB 12–14 0.0 Z sourceX1624-490 21 Dipper2S0921-630 216 ADC 15.3Cyg X-2 235 Burster 14.7 −2.0 Z sourceJ1744-28 283 Burster, pulsar, transient 0.47s pulsar, type II bursts

Figure 7. Comparison of the x-ray light curves of the prototype ADC source X1822-371 (top) with the eclipsing ‘dipper’ X0748-676.The orbital periods are very similar, but the orbital inclinations are only slightly different, with that of X0748-676 being just low enoughto allow a direct view of the neutron star whilst still allowing the companion star to eclipse it (see also figure 5).

One particular source, EXO 0748-676, displays notonly x-ray dips but also an (almost) total eclipse, albeitvery brief (see figure 7). The inclination in this LMXB (a

transient that turned on in 1985, and has remained ‘on’)must be in the very narrow range (73–83) that allowsboth dips and a short total eclipse by the secondary star.



Accretion disk coronaeHowever, a closer examination of figure 7 will show thatthe eclipse in EXO 0748-676 is not in fact completely total.There is a small but significant residual x-ray flux whichis about 3% of the normal, out-of-eclipse level. Thisresidual flux is due to x-rays which are scattered into ourline of sight from material (possibly a disk ‘wind’) aboveand below the disk. Hence for sources at even higherinclinations, we would not expect to see total eclipses,but instead partial eclipses of this extended ‘accretion diskcorona’ (or ADC). Figure 7 also shows the x-ray light curveof X1822-371, the classical ADC source which combines along smooth modulation of the x-ray flux by the accretiondisk rim structure with a partial eclipse by the secondarystar. This is essentially what would be expected if thedipper light curve were to be smoothed out. ADC sourcesand x-ray dippers are important objects for the informationthat they provide on the azimuthal structure of accretiondisks in interacting binaries.

X-ray pulsationsThe much larger accretion discs in LMXBs compared toHMXBs should lead to substantial transfer of angularmomentum onto the compact object and hence very fastspin periods (comparable to millisecond radio PULSARS).However, QUASIPERIODIC OSCILLATIONS IN X-RAY BINARIES wereseen with large area x-ray detectors, an effect attributedto an interaction between the inner accretion disk andthe neutron star magnetosphere. Very recently, the firstmillisecond x-ray pulsar was discovered (SAX J1808.4-3658) with a coherent pulsation at 401 Hz. It also displaysx-ray bursts that indicate a weak (<1010 G) magneticfield, and has a 2 h orbital period with a very low masscompanion. This object is considered to be a direct linkbetween LMXBs and the ‘black widow’ millisecond radiopulsars.

There is a further group, the ‘anomalous x-raypulsars’ (AXPs), the nature of which remains very unclear.Only half a dozen are known, the best studied of which is1E2259+587. They all exhibit coherent x-ray pulsationswith periods in the range 5–12 s, have x-ray luminositiesfar too high to be explained by their observed spin-downrates and yet show no evidence for binary motion in thesepulsations, nor has any been optically identified. It hasbeen suggested that they may be linked to the ‘soft gamma-ray repeaters’ (a sub-class of gamma-ray bursters, seeGAMMA-RAYASTRONOMY) which also display similar pulsationperiods during their bursts, an effect which led to theproposal that they had much larger magnetic fields thanhitherto suspected (the so-called ‘magnetars’).

Long-term or superorbital modulationsFrom the early days of x-ray astronomy there have beensatellites that scan the sky repeatedly and regularly (usu-ally because they are spin-stabilized). This automati-cally provided the ability to monitor x-ray sources formany weeks (sometimes continuously), providing long-term light curves. These yielded the presence in a number

of x-ray binaries, both LMXB and HMXB, of yet anothermodulation, this time quasi-periodic and on a timescaleof tens to hundreds of days. The first discovered was the35 d on–off cycle in HERCULES X-1, and which is attributedto a tilted, precessing disk. A similar ∼30 d modulationhas been seen in LMC X-4, but a substantial change in thistimescale over a year or so is difficult to account for. Sub-sequent long-term modulations (from 70 to 200 d) havebeen found in a variety of sources, including both LMXBsand HMXBs. These are currently attributed to the effectsof a warped accretion disk driven by the central x-ray lu-minosity.

BibliographyBildsten L et al 1997 Astrophys. J. Suppl. 113 367Charles P A and Seward F D 1995 Exploring the X-ray

Universe (Cambridge: Cambridge University Press)ch 7–9

Lewin W H G, Van Paradijs J and McClintock J E (eds) 1995X-Ray Binaries (Cambridge: Cambridge UniversityPress)

Philip A Charles


X-Ray Bursters E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

X-Ray BurstersX-ray bursters are a class of sources which emit burstsof X-RAYS lasting from a few seconds to a few minutes.They are further classified as either type I or type II.Typical type I bursts have a faster rise than decay andthe average energy of the x-rays decreases as the burstsmoothly decays. Typical type II bursts stop abruptlywithout changing the average photon energy. There arenow over 50 sources of type I x-ray bursts known, whileonly two sources are known as type II bursters, and oneof these also has type I bursts. Because of this imbalancein numbers, ‘x-ray burster’ is assumed to mean a type Ix-ray burster, although there are other kinds of sources ofshort bursts or flashes of x-rays, ranging from the Sun toblack hole candidates. Type I bursters are NEUTRON STARS onwhich gas accreted from a binary companion undergoessudden nuclear burning. The sources which have beencalled type II bursters are also neutron stars accreting gasfrom a binary companion. The type II burst is a sudden,large, but brief, increase in x-ray producing ACCRETION.

The neutron stars in x-ray bursters of both type Iand type II have binary companion stars in close orbitsuch that the companion loses matter to the neutronstar’s gravitational attraction. (The companion fills thegravitational equipotential called the ROCHE LOBE.) Thetransferring matter has ANGULAR MOMENTUM around theneutron star, so that it circulates around it and mixeswith gas already there in a flat ring distribution called anACCRETION DISK. The disk material slowly moves inwardsuntil it reaches the neutron star, where it accumulateson the surface. As helium and hydrogen accumulateunder conditions of very high density and increasingtemperature, the collisions of the ions can produceinelastic nuclear reactions which generate heat. The risingtemperature increases the rate of collisions, raising thetemperature further, until a nuclear fusion chain reactionis set off. This is the thermonuclear flash. The heat fromburning the thin layer of helium, and sometimes hydrogen,diffuses out of the region. It heats the surface layersof the neutron star to a temperature at which x-rays arethe dominant radiation emitted. Generally the radiationincreases very quickly to its peak, sometimes in less than1 s. The heat of the flash spreads to the surface and theburst decay corresponds to cooling of the surface.

From the burst x-rays and information about thedistance or with assumptions about the peak brightnessthat the source can have, the radius and mass of theneutron star can be estimated. These values are closeto those predicted for neutron stars and to the valuesmeasured in other ways, so that the conclusion that thestar is compact enough to be a neutron star is firm. Themeasurements may eventually determine neutron starparameters, identify the composition of the matter fallingon the neutron star and confirm ideas about the sequencesof nuclear reactions that turn the accreted hydrogen andhelium into the heavy elements of a neutron star’s crust.The neutron stars have accreted angular momentum as

well as mass. In some cases during the x-ray bursts thestar is not uniformly hot and the flux has been found to bemodulated with a period of 2–3 ms. These modulationsare believed to indicate the spin of the neutron star and thecharacteristics of the modulation also provide informationabout the parameters of the neutron star. Thermonuclearbursts provide one of the few ways to obtain informationabout the densest matter.

Type I x-ray burstsIn 1975 there were three X-RAY ASTRONOMY satellitesexploring the properties of sources found in the firstx-ray survey mission, UHURU. Strong, short bursts of x-rayswere first seen with the Astronomical Satellite of theNetherlands, then with SAS-3 and OSO-8, as sources inthe region of the center of the Galaxy were observed. Itturned out that the phenomenon had been seen as far backas 1969, by the Vela-5 satellites.

The sources from which the bursts appeared were inthe direction of the bulge of the Galaxy, approximately10 kpc away. From burst peak fluxes of 10−8 erg cm−2 s−1,factors of 5–20 times normal or persistent flux of thesources, the peak luminosities are ≈1038 erg s−1. Thisis close to the maximum luminosity possible for a 1Mobject, at which the radiation pressure due to photonscattering balances the force of gravity. (This ‘Eddingtonlimit’ depends on the chemical abundances of the gas.)

The bursts were observed mostly in the photon energyrange 1–20 keV. Generally they have a distribution withenergy at a given time that is well approximated by a blackbody function with a temperature of T ≈ (1–4) × 107 K,corresponding to kT ≈ 1–3 keV. Equating the estimatedluminosity to the radiation rate from a spherical blackbody,

Lx = 4πR2σT 4

the radius R can be estimated to be ≈10 km. This isconsistent with the expected radius of a neutron star. Itis one of the best pieces of evidence that the objects areneutron stars. Figure 1 contains schematics of the x-rayluminosity, temperature and apparent radius derived fora type I x-ray burst. The integral of the excess luminositydue to the burst is its fluence.

When a source is burst active, the bursts recur atintervals, usually of several hours to days, as illustrated infigure 2. The phenomenology points to a process in whichthe fuel builds up until it catastrophically ignites. Theenergy output is consistent with the nuclear burning of theaccumulated matter. If the gravitational energy releasedin accretion is turned into x-ray flux Lx = GMM/R, themass accumulated in time T is

M = Lx

GM/RT .

Scaling to nominal observed values for Lx and T ,

M ≈ 1021 Lx

1037 erg s−1

(R

10 km

/M

1.4M

)T

5 hg.



Rise Time

Decay Time

Eddington Limit Lx

Persistent Lx

kT

R

7 km

15 km 1 keV

2 keV

time (sec)0 5 10

37

log

Lx

38

Figure 1. Type I burst parameters. The upper panel shows aschematic of the x-ray luminosity of a burst as a function of time.The lower panel shows a typical behavior of the fitted blackbody kT (right-hand scale) and the deduced apparent radius of aspherical emitter (left-hand scale).

T

37

38

log

Lx

time (hr)0 5

Figure 2. Persistent and burst luminosities. The average burstluminosity is the fluence in a burst per the recurrence time T .

There are a number of nuclear reaction sequences bywhich accreted material with the abundances of the outerlayer of the companion star can convert into the iron andhigher atomic number elements of the neutron star crust.The amount of energy released per nucleon is 2–8 MeV, sothat the potential for energy release from burning, in themass accumulated, is

E = (15–76)× 1038

× Lx

1037 erg s−1

(R

10 km

/M

1.4M

)T

5 hrerg.

If the peak flux were the Eddington limit 2.0 ×1038M/1.4M erg s−1 (assuming cosmic abundances here),it would take 7–40 s to cool the hot layer, with the heatconvecting and diffusing to the surface in ≤1 s.

Generally, in the thermonuclear flash, the ratio of thepersistent x-ray luminosity to the average burst luminosity(the fluence per burst divided by the recurrence interval)should be the ratio of the gravitational energy releasedper gram falling on the neutron star to the nuclear energyreleased in burning it. The observational quantity, denotedα, is equal to the ratio of the fluxes. The predictedvalues of α are in the range 25–200, depending onthe nuclear reaction track. Some reaction tracks implyparticular correlations between burst characteristics andthe persistent luminosity. Theoretical calculations indicatethat the nature of the reactions is more likely to go throughcycles than to be a steady state, as the accretion ratechanges and with it the ambient temperature of the settlingmaterial. The general agreement is regarded as a strongconfirmation of the thermonuclear flash model of type Ibursts.

Type II x-ray burstsType II x-ray bursts usually (but not invariably) havethe ratio of persistent luminosity to burst luminosity (α)too small to be consistent with what is known aboutthermonuclear reactions, well below 25. The first andrecurrent example of this type is MXB 1730 − 335,appropriately called the Rapid Burster. The sourcebecomes bright approximately every 6 months and fora few weeks goes through several metamorphoses ofbehavior, sometimes exhibiting rapid-fire bursts a fewminutes apart. The Rapid Burster is also a source of type Ibursts, so that the two different types of bursts can becompared for the same source. The second source was thetransient GRO J1744 − 28, which first became known in1995 as the Bursting Pulsar. This transient has reappearedonce, so that it is recurrent, although irregularly. Burststypically came about 30 min apart when it was active.

An increase in the rate at which matter falls on theneutron star can explain these bursts. Several possiblemechanisms have been identified to store up the mattercoming in, at some distance from the neutron star, and torelease it suddenly to complete its journey to the neutronstar.

The matter in accretion disks works its way to smallerradii as viscous interaction with neighboring cells of theaccretion disk removes its angular momentum. As matteris continuously fed into the disk by the companion star,the viscous properties change and a sudden transitioncan occur from a state of low viscosity to high viscosity.The higher torques allow matter to start moving rapidlythrough the disk. The emission that we see as the burst cancome both from the disk itself and from the neutron starwhere the matter falls on it. This instability can operate on



a range of recurrence time-scales, from years to minutes,so that it could give rise to bursts as well as to dwarf andx-ray NOVAE.

A different possible cause of the intermittent flowonto the neutron star is the interaction of the neutronstar’s magnetic field with the matter approaching the star.The magnetic field of the neutron star, if strong enough,interacts with the inner accretion disk. In the case of strongfield PULSARS at some radius the field channels the diskmaterial out of the plane and along the magnetic field linesto the magnetic poles. For the case of weaker fields, thedisk may interact with the field over an extended region.It has been suggested that material could accumulateoutside the magnetosphere until some threshold conditionoccurred that would allow the material to suddenly feedonto the neutron star.

For the Bursting Pulsar, pulsations with a period of0.5 s continued throughout the bursts. Considerationsof how low the accretion rate can go before pulsationscannot occur have allowed the surface magnetic field tobe estimated at 5 × 1011 G. The Rapid Burster is likely tohave a field weaker by several orders of magnitude andpossibly the two sources represent two different instabilitymechanisms.

Thermonuclear flashesHydrogen and helium burningAs the accreted material collects on the neutron star, it willspread out over the surface if the magnetic field is too weakto confine the plasma (below 1011 G). If the companion isstill on the main sequence, the material is, by weight, 70%hydrogen, 30% helium, with no more than a few per cent inheavier elements (C, N, O). However, in at least one case,4U 1820−30, the companion is known to be a helium WHITE

DWARF. When there is a layer of≈1021 g on the surface of theneutron star, the density at the bottom of the accreting layerreaches ≈105 g cm−3. Nuclear reactions begin to converthydrogen into helium and helium to oxygen. Thesereactions release heat. At very low and very high rates theheat can be radiated as fast as the energy is released andthe hydrogen burning is stable, but under many conditionsthe burning is unstable, and the temperature increasesrapidly until the fuel is completely burned. Helium burnsby the triple-alpha reaction 3α→12C and by 12C(α, γ )16O,usually unstably. Detailed calculations have examined thekind of succession of bursts that should occur, for a rangeof accretion rates, central neutron star temperatures andneutron star masses and radii.

Observed and modeled burst propertiesFor the intermediate accretion rates at which instabilitiesoccur, three regimes have been distinguished. Heliumburns unstably but, depending on the accretion rate,hydrogen, if present, may burn stably or unstably. In themiddle of these three regimes the hydrogen burns stably(unless the interior temperature is very hot) and produceshelium which flashes with α ≥ 100, recurrence times≥10 h, and burst rise times≤1 s. At either higher or lower

accretion rates, the hydrogen and helium burn together,with complex reactions, building to iron and heavierelements. These reactions give the burst a longer rise timeand can leave unburned fuel, so that the thermonuclearluminosities, and therefore the values of α, are lower.

Type I bursts from different sources are observed tobe similar, but not identical. Bursts rise times range from0.3 to 10 s and decay times range from 5 to 100 s. Thedecay times (τ ) and the values of α vary with the persistentluminosity. The EXOSAT Observatory caught several sourcesin different states. Comparison of the behavior of theparameters with the predictions of models has led toinvestigations of other degrees of freedom, for example,dependence of the internal temperature on the history ofthe burning and the extent of the burning area on theneutron star.

For many of the x-ray bursters the persistentluminosities vary, sometimes consistently, but not exactlyperiodically. Some are transient persistent sources inwhich accretion may hardly occur in between episodesyears apart. Then they rise to maximum in hours anddecay over several weeks to a year. The sources 4U1608−52 and Aquila X-1 are regularly recurrent transients.As the persistent flux may vary by factors of a thousandand the supply of mass to the neutron star envelope isbelieved to vary accordingly, the integrated effect must beconsidered.

Radius expansion during burstsAs the temperature during bursts rises and falls, theprojected area of the hot emitter can be tracked. Formany burst measurements the size of the emitter hasbeen consistent with it remaining approximately constantthroughout the burst. For a subset of bursts the apparentsize increases near the beginning of the burst to values asmuch as 10 times sizes theoretically estimated for neutronstars. The apparent size then drops back and levels off at anasymptotic value consistent with that of a neutron star (asin figure 1). In these cases the bolometric luminosity staysapproximately constant during the expansion episode.The temperature goes down during the expansion. Thepeak luminosity appears to be the Eddington limit forhelium for most ‘radius-expansion’ bursts.

Theoretically, helium flashes can generate enoughenergy to cause photospheric expansion of the neutronstar. When the surface luminosity reaches the Eddingtonlimit, it drives a wind from the neutron star. There isevidence within bursts from a single source that somereach a plateau of luminosity ≈1.7 times lower than theradius-expansion plateau luminosity of others. This hasbeen interpreted as the Eddington limit for hydrogen-richgas, a factor 1.7 below that of helium. These characteristicscan be identified from the spectra, without knowing thedistance of the source, and are used as the basis forestimates of the distances to some burst sources. In thecases in which a distance is known on other grounds, theyare in approximate agreement.



General relativity corrections and radiationtransfer

The assumptions of black body emission and Eddington-limited flux seem to offer the possibility of measuring theradii and masses of neutron stars. These are interestingbecause they depend on the properties of the matter atdensities of 1015 g cm−3 at the center of the neutron star.At least seven of the type I bursters are in GLOBULAR CLUSTERS

and some of their distances are known to within 10%.However, several important effects must be included ininterpreting the measurements. For neutron stars, thecorrections of general relativity can be important; for anominal 1.4M star, the Schwarzschild radius is 40% ofthe nominal radius of 10 km and the relativistic effectscan be larger for some equations of state. Further, thesurface temperature of the neutron star exceeds 107 K andthe neutron star has a scattering atmosphere. The surfacehas an energy-dependent emissivity. These effects needto be taken into account in understanding and using theobservations to determine the neutron star parameters.

A spherical black body neutron star has an apparentradius larger than the radius at the star, because of thegravitational bending of rays. Taking this into account,with

1 + z∗ = (1− Rg/R∗)−1/2

R∞ = R∗(1 + z∗)

(as long as R∗ ≥ (3√3/2)Rg), where Rg = 2GM/c2 is theSchwarzschild radius, R∗ the real neutron star radius andR∞ the apparent radius we measure, which exceeds thereal radius. The Eddington limit for the luminosity asobserved by a distant observer gives another constrainton M and R∗. Together with possible theoretical mass–radii relations the Eddington limit implies a limit on the‘effective’ temperature (defined by Te = (Lx/4πR2σ)1/4),such that kTe should never exceed 2 keV. The black bodyfits determine a ‘color’ temperature Tc, which at its peakoften exceeds this limit on the effective temperature. Thisis a measure of the error in the assumption of black bodyemission. When the competition between scattering andabsorption, the temperature increase with depth in theatmosphere and incoherent scattering in the atmosphereare all included, the ratio Tc/Te has been found to varybetween 1.4 and 1.7 as the flux approached the Eddington-limited flux.

Detailed data for bursts have been obtained byx-ray experiments on many satellite missions, notablyOSO-8, SAS-3, HEAO-1, EXOSAT, Tenma, Ginga, RXTEand BeppoSAX. Extensive work was done attempting todetermine the properties of the neutron star using theEXOSAT and Tenma data. Measurement and perhapssystematic errors prevented definitive results from beingobtained. Similar projects to examine RXTE observations,which have much smaller statistical error and higher timeresolution, are still in progress.

The neutron starsX-ray bursters are in very old BINARY SYSTEMS, as shown bythe population in globular clusters. At least three may wellhave degenerate dwarf companions (4U 1916 − 053, 4U1820−30 and SAX J1808.4–3658) and all of the companionsare low-mass stars. It has been suggested that in accretingover 108–109 yr the neutron star’s magnetic field hasdecayed. When radio pulsars with 108–1010 G fields werediscovered with rotation periods of a few milliseconds,it was suggested that accretion in the low-mass x-ray binaries spins up the neutron stars to millisecondperiods. Then irradiation could evaporate the companion,leaving a rejuvenated, millisecond, radio pulsar. Recentobservations with RXTE discovered evidence tendingto confirm this scenario. The transient and bursterSAX J1808.4–3658 is now known to be an x-ray pulsar. Ithas a magnetic field estimated to be 5×108 G and a rotationfrequency of 401 Hz (a period of 2.5 ms).

In six other bursters an inhomogeneity in temperatureon the star appears during bursts and appears to rotate.The periods are similar to the period of SAX J1808.4–3658. At the beginnings of bursts a high-amplitude (ashigh as 60%) sinusoidal oscillation starts. The oscillationamplitude decays along with the rise in flux. This behaviorwould be consistent with a growing burning region. Theperiods are 1.7–3 ms. During the decays, 10 s trainsof oscillations occur with typical amplitudes of 10%.Whether the asymmetry is associated with one spot ortwo is uncertain. Together with new and related results onfast quasi-periodic oscillations in the persistent flux, theyprovide new, and potentially precise, tools to determineproperties such as the masses and radii of the neutron stars,their magnetic fields, the radii at which the disks terminateand how the accreting gas traverses from the disk to theneutron star. With RXTE, BeppoSAX, Chandra and XMMit will be possible to address many questions about x-raybursters, the evolution of the systems, the flow of accretinggas, the progress of thermonuclear flashes and the neutronstar solutions for collapsed stars of ≤3M.

BibliographyThe literature on the subject of x-ray bursters includesa number of reviews, which are well represented in thechapters by

Lewin W H G and Joss P 1983 X-ray bursters and the x-raysources of the Galactic bulge Accretion Powered X-RaySources ed W H G Lewin and E P J van den Heuvel(Cambridge: Cambridge University Press) pp 41–115

and by

Lewin W H G, van Paradijs J and Taam R E 1995 X-raybursts X-Ray Binaries ed W H G Lewin, J van Paradijsand E P J van den Heuvel (Cambridge: CambridgeUniversity Press) pp 175–232

Brief summaries of the newest results are in

Bildsten L and Strohmayer T 1999 Phys. Today 52 4



and in articles in

Bucceri R, van Paradijs J and Alpar M A(ed) 1998 The ManyFaces of Neutron Stars (Dordrecht: Kluwer)

and in

Scarsi L, Bradt H, Giommi P and Fiore F (ed) 1998 TheActive X-Ray Sky (Amsterdam: Elsevier)

Jean Swank


X-Ray Telescopes E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

X-Ray TelescopesX-ray telescopes produce images of x-ray-emitting objectswithin the telescope’s FIELD OF VIEW by reflection fromprecisely shaped mirrors. Hans Wolter’s design in theearly 1950s of an x-ray microscope using reflective opticsled Riccardo GIACCONI to suggest an ‘inverted’ set of optics,not subject to the fabrication limitations of the microscope,could be used as a cosmic x-ray telescope. As describedin 1960 by Giacconi and Bruno ROSSI, imaging with x-raytelescopes offers a significant reduction in noise, both fromcosmic ray induced events and the soft x-ray background.This is because most background counts are the resultof charged particles and are uniformly distributed overthe detector area. An imaging system concentrates thesource counts in one or a few pixels. As an example,the EINSTEIN (HEAO-2) Observatory x-ray telescope offered animprovement in sensitivity of about a factor of 1000 overthat of the large-area, collimated proportional counterson High Energy Astrophysical Observatory (HEAO) 1satellite. The improved instrument sensitivity resultingfrom increased signal to noise enables the study of faintersources, extending the number and age of detectableobjects.

Imaging observations enable deeper study of ex-tended cosmic x-ray sources. Imaging of CLUSTERS OF GALAX-

IES reveals details of the intracluster gas temperature pro-file and isolates the x-ray emission from component galax-ies. Imaging of GALAXIES allows study of any central sourceas well as the identification and location of the galac-tic emitters. Study of SUPERNOVA REMNANTS allows the vi-sualization of the x-ray-emitting regions and potentialidentification of a remnant’s compact object. None ofthese types of studies is readily undertaken with non-imaging instruments. Other imaging approaches employ-ing such devices as modulation collimators or scanned,tightly collimated detectors provide much poorer resolu-tion (∼1 arcmin.) than that available with current x-raytelescopes and suffer from image reconstruction artifacts.

Studies of point sources also benefit from x-raytelescope resolution. Using the large-area proportionalcounters on board HEAO-1, positions accurate to ∼0.1square degrees were obtained for weak point sources.On average, there will be one star brighter than 12thmagnitude in the position error box. For comparison, theoptical counterparts of Sco X-1 and Cen X-1 are about 13thmagnitude, and that of Her X-1 about 15th magnitude.Imaging enables more accurate optical identifications.With better than 0.5 arcsec resolution of the Chandra X-rayObservatory (CXO) positions of point sources accurate to∼10−7 square degrees may be obtained—on average lessthan a single unrelated 20th magnitude star will be foundin the error box.

This article will first briefly discuss how x-rays maybe reflected and what optical systems can be used to formimages. Key considerations for performance and systemdesign will be discussed, followed by the various waysof implementing x-ray telescopes. We will review the

more significant past and current x-ray telescopes, andthen discuss future trends in x-ray telescopes.

X-ray reflectanceX-rays specularly reflect from a surface under twoconditions, when striking a surface at grazing incidenceand when constructive interference takes place betweenmany layers of a material whose atomic number varies ina periodic fashion.

Reflection at grazing incidenceAt x-ray wavelengths the real part nr of the complex indexof refraction nr − ik of the reflecting surface is less than 1.Thus x-rays in vacuum are incident upon an interface thatis less optically dense. From Snell’s law the angle of thewave transmitted θt into the less dense material is

sin θt = (n1/n2) sin θi (1)

where θi is the angle of incidence and n1 and n2 arethe indices of refraction for vacuum and the reflectingmaterial, respectively. When n1 > n2 one obtains a realvalue for θt only for incident angles θi < sin−1(n2/n1). Forexample, if n2 = 0.99, then the maximum real value forθi is 81.9: at this incident angle the transmitted angle is90, parallel to the interface. For larger incident angles(shallower grazing angles) no radiation is transmitted.This is seen by allowing θt to be complex, with

cos θt = ±i[(n1/n2)2 sin2 θi − 1]1/2. (2)

Substituting equation (2) into the Fresnel equations,solving for the transmitted electric and magnetic fields,and determining the time-averaged Poynting vectorreveals that no energy flows through the interface—the field intensities normal to the interface decayexponentially into the less optically dense surface.

Reflectance is calculated from the complex index ofrefraction using the Fresnel equations. The real andimaginary parts of the index are computed from the atomic(forward) scattering factors f1 and f2:

nr = 1− δ = 1− ρ

WmA0reλ

2f11

2π

k = ρ

WmA0reλ

2f21

2π(3)

where ρ is the material density, Wm is the molar weightof the reflecting material, A0 is Avogadro’s number, re

is the classical electron radius and λ is the incidentwavelength. The scattering factors are material andwavelength dependent. They are derived in a varietyof ways: they may be computed from measurement ofthe forward scattering amplitude or back calculated frommeasured x-ray reflectance. Well-known sets of constantshave been published by Henke et al, Auerbach and Tirsell,and Windt. As better reflectance is achieved with materialswith high electron density grazing incidence optics aretypically coated with metals to enhance reflectance.



0.1 1 100.01

0.1

1

30 arc-min. graze angle60 arc-min. graze angle90 arc-min. graze angle

Reflectance of Iridium

Incident Energy (keV)

No

rma

lize

d R

efle

cta

nc

e

0.1 1 100.01

0.1

1


Reflectance of Gold


No

rma

lize

d R

efle

cta

nc

e

0.1 1 100.01

0.1

1


Reflectance of Nickel


No

rma

lize

d R

efle

cta

nc

e

Figure 1. X-ray reflectance as a function of incident energy and graze angle.

Common materials used as reflecting coatings are iridium,gold and nickel. A plot of x-ray reflectance as a function ofincident energy and grazing angle is shown in figure 1.Note the local minima in reflectance at the location ofx-ray absorption edges. The product of reflectance andgeometric collecting area (the entrance aperture) is calledthe effective area. Reflectance decreases as a functionof increasing graze angle and starts to decrease rapidlynear the so-called critical angle. The critical angle may beexpresses as cos−1(nr) ≈ (2δ)1/2 for δ 1 (see equation(3)).

Reflection from multilayer coatingsMultilayer coatings contain alternating layers of high-Z(atomic number) and low-Zmaterials. Reflection occurs ateach interface, and the layer optical thicknesses are chosenso as to produce constructive interference between eachlayer. Reflection occurs only for those wavelengths whichsatisfy the Bragg equation,

mλ = d sin θ (4)

where d is the layer spacing and θ is the incident angle.In practice the coating reflects a bandwidth λ that is afunction of various fabrication parameters including layerthickness uniformity, density, and layer materials but istypically only a few ångstroms wide. Typical layer ma-terial pairs include (but are not limited to) nickel/carbon,tungsten/silicon, rhodium/beryllium, rhodium/carbon,molybdenum/carbon and platinum/carbon. A typicalcoating may contain between 40 and 500 layer pairs, withthe larger number of layers yielding higher reflectance. Re-flectance of 10–80% has been achieved, depending on theincident wavelength (higher reflectance at longer wave-lengths) and material choices.

Several soft x-ray–EUV, normal incidence, multilayertelescopes have been launched on board soundingrockets and small satellites, including the recentlylaunched Transition Region and Corona Explorer (TRACE).To overcome the spectral bandwidth limitations ofmultilayer coatings, experimenters have resorted tointegrating several smaller telescopes into a singlepayload, each telescope with a different multilayer. In



the case of TRACE, different multilayer coatings wereapplied to each of the four quadrants of the optics so thata single telescope was sensitive to four separate spectralbandwidths (one of which was visible light).

An alternative to several different telescopes ormultiple different coatings on a single telescope is gradedmultilayers, or ‘supermirrors’. Here, the layer thicknessvaries as a function of position (depth) in the layer stack.The thickest layers are at the top of the stack to reflect thelongest wavelengths with a minimum of absorption. Goodbroadband reflectance has been achieved using a thicknesslaw

di = a(b + i)−c (5)

where di is the thickness of the ith layer (starting from thetop), a and c are positive parameters that are a function ofthe materials used, and b is a constant greater than −1.

Substrate and coating roughness degrades multilayerperformance. ‘Low’ frequency roughness (>1µm periods)causes scatter which degrades image contrast. ‘High’frequency roughness (<1µm periods) reduces reflectance.

Types of telescopesTwo approaches are used to achieve imaging: grazingincidence (GI) and normal incidence (NI). Grazingincidence telescopes utilize large angles of incidence (near90) so that the x-rays graze or glance off the surface andare reflected. Normal incidence x-ray telescopes use moreconventional shaped optics with multilayer coatings.

Grazing incidence telescopesIn 1952 the German physicist Hans Wolter was ableto show that a two-element system containing an evennumber of confocal conic optics will come close tosatisfying the Abbe sine condition; rays that reflect offboth surfaces are focused and form an image. Wolter,attempting to produce systems for x-ray microscopy,produced three designs, referred to as Wolter types I, II andIII. The type I telescope (figure 2) consists of a paraboloid asa primary mirror and a confocal and coaxial hyperboloid asthe secondary mirror. The paraboloid focus is coincidentwith the back hyperboloid focus. X-rays strike theparaboloid at the grazing angle (approximately the fieldangle of the source plus half the angle subtended by thebest-fit cone to the paraboloid), are reflected and strike thehyperbola. The image is formed at the front hyperboloidfocus. Usually, the hyperboloid cone angle is three timesthat of the paraboloid so that on-axis x-rays are incidentupon both mirrors with essentially the same grazing angle.X-rays that strike the forward end of the paraboloid reflectto strike the aft (back) end of the hyperboloid, and viceversa.

The entrance aperture is the projection of the primarymirror in the aperture plane. This results in an annularaperture whose width is approximately the product ofthe optic length and the half-angle of the best fit cone(the cone angle). For the CXO the largest paraboloid hasa surface area of approximately 3.2 m2 but an entranceaperture of only 0.047 m2, a reduction of a factor of

F F

CONFOCALHYPERBOLOID

PARABOLOID

REFLECTINGSURFACES

TYPE I

Figure 2. Schematic representation of the Wolter type I telescope.

Figure 3. Cutaway schematic drawing of the nested four shells(paraboloid and hyperboloids) of the Chandra X-rayObservatory (courtesy of the Raytheon Co).

∼68 times. Collecting area is built up by nesting mirrorpairs within one another (figure 3). Each mirror pair, orshell, is co-aligned and confocal. The telescope entranceaperture consists of a set of concentric annular apertures,all contributing to the same focus. Ideally, the shells aredesigned so that each one has the same focal length andtherefore the same plate scale (the proportionality constantrelating angular distance on the sky to linear distance atthe focal plane). All shells do not necessarily contributethe same to the image as a function of x-ray energy. Asshown above in the section ‘X-ray reflectance’, reflectanceis a function of graze angle and incident energy. The innershells have a shallower graze angle than outer shells. Thusthe inner shells of a nested telescope have a larger spectralbandwidth (reflect higher-energy x-rays) and comparableor slightly higher reflectance than the outer shells, whilealso having a smaller entrance aperture.

The type I design yields a perfect image for on-axis illumination. Off axis, the design suffers from fielddependent coma and spherical aberration. In addition,



FF

CONFOCALHYPERBOLOID

PARABOLOID

REFLECTINGSURFACES

TYPE II

Figure 4. Schematic representation of the Wolter type IItelescope.

the telescope focal surface is curved. Van Speybroeck andChase (1972) approximated the off-axis rms blur diameterfor a flat focal plane detector at the Gaussian focus as

σblur ≈ 4ξ + 1

10(tan θfield)

2

tan αL

Z0+ 8 tan(θfield)(tan α)2 (6)

where θfield is the field angle (the off-axis angle), α is thegraze angle, Z0 is the axial distance from the intersectionof the paraboloid and hyperboloid surfaces (the ‘virtualjoint’) to best on-axis focus, and L is the axial length of theparaboloid. The first term in equation (6) represents thefield-dependent coma. The second term is split betweena GI equivalent of spherical aberration and image defocus(relative to a flat detector) due to a curved focal surface. Forthe Chandra X-ray Observatory at a field angle of 5 arcmin,the rms blur diameter ranges from 2.5 arcsec to 4 arcsec.Alternative Wolter I type designs exist that mitigate theoff-axis aberrations at the expense of degrading on-axisperformance. These designs consist of using two coaxial(but not confocal) hyperboloids or using generalized (non-conic) mirror prescriptions such as a power law. Becauseof the relative mechanical simplicity of the design and theability to increase collecting area by nesting telescopes thetype I design (or its variants) is the most common form ofgrazing incidence designs used in x-ray astronomy, beingemployed on the Einstein Observatory, Roentgen Satellite(ROSAT), CXO, X-ray Multi-mirror Mission (XMM) and manyothers described later.

The type II design also consists of a grazing incidenceparaboloid primary and hyperboloid secondary mirror,but the outer surface of the hyperboloid is used, asshown in figure 4. In the type II the image is formedat the back hyperboloid focus. Type II telescopes enablelonger focal lengths than a type I with comparable grazingangle and entrance aperture, affording an increased platescale. Field-angle-dependent aberrations are greater withtype II designs than with type I designs. Also, nestingof telescopes to increase collecting area is impractical.

Instead, collecting area may be increased by increasing thegrazing angle, thereby increasing the projected area on theaperture plane, while maintaining a usefully long focallength and acceptable plate scale. The increase in grazeangle, however, limits the useful spectral bandwidth of thetelescope so that most practical type II telescopes are usedfor the extreme to far ultraviolet and longer wavelengths(>100 Å).

The type III design employs the outer surface of aparaboloid as the primary element and the inner surfaceof an ellipsoid as the secondary mirror. The paraboloidfocus is coincident with one of the ellipsoid foci, and theimage falls upon the other ellipsoid focus. The type IIIdesign has never been used for x-ray astronomy.

Wolter developed variants of the original designsby extending results obtained by Karl SCHWARZSCHILD in1905 for normal incidence telescopes. The modifiedprescriptions are called Wolter–Schwarzschild (W–S)designs and differ slightly in their second-order figure(shape). These optics satisfy the Abbe sine conditionstrictly and so do not exhibit any comatic aberration. Theimprovement in off-axis performance of W–S designs overconventional designs is graze angle dependent. Thus W–S designs provide negligible improvement at the shallowgraze angles (<1) employed on many x-ray telescopesto achieve good reflectance at shorter wavelengths. W–S-type designs have been used on a number of extremeultraviolet telescope applications which use much largergraze angles (>5). The EXTREME ULTRAVIOLET EXPLORER (EUVE)

contained a W–S type I EUV telescope and a W–S type IIfor the stellar spectrometer. ROSAT, besides carrying thelarge Wolter I x-ray telescope, also carried a W–S type IEUV telescope—the Wide Field Camera—which had a 5field of view, ∼450 cm2 aperture and image half-powerdiameter of ∼1.7 arcmin (on axis).

Contamination of the optical surface will degradeperformance. Particulates both absorb and scatter the x-rays, degrading the point spread function and reducingthe effective collecting area. Hydrocarbons absorb x-rays.Grazing incidence telescopes are extremely sensitive toboth. This is a direct result of the shallow grazing angle,as can be seen in figure 5. A round particle of cross sectionπa2 maps onto an entrance aperture area of 2πa2 because x-rays may strike the particle before or after reflecting off themirror surface. The mirror surface area is 2πRLwhereR isthe average radius of the nearly conical optic. The mirrorentrance aperture has an area of 2πRL sin α. Thus thefraction of mirror surface area Fn covered by the particleis

Fn = πa2

2πRL(7)

but the fraction of the entrance aperture FA ‘covered’ bythe particle is

FA = 2πa2

2πRL sin α= Fn

2sin α

(8)

For a 1 graze angle FA is ∼114 times larger than Fn.Similarly for a hydrocarbon layer t thick, the x-rays



2a2a

ParticulateContamination

Optical Surface

Figure 5. Illustration of the effects of particulate contamination.Incident x-rays from the upper left are absorbed or scattered bythe particle either when directly impinging upon it or whenimpinging upon it after reflection from the mirror.

must traverse a distance 2t/ sin α. In addition, becauseof their construction with little space between telescopeshells and very narrow entrance and exit annuli, it isextremely difficult to clean a dirty grazing incidenceoptic. Contamination control is a critical issue for grazingincidence optics.

Finally, an earlier realization than the Wolter designswas the Baez–Kirkpatrick telescope in which an array ofparallel plates, parabolic in the plane of incidence, focusthe incident x-rays to a line image. Placing a secondset of plates oriented at a right-angle to the first setfocuses the line image to a point image. This design hasmore aberrations than the Wolter designs, but may beapproximated using optically flat plates bent in one axisso as to inexpensively build up large collecting area withmoderate resolution.

Multilayer telescopesConventional optical telescope designs, when used as thesubstrate for multilayer coatings, are suitable for x-raytelescopes. The distinction between normal incidence x-ray and optical telescopes is that optical telescopes arediffraction limited and x-ray telescopes (to date) are not.To achieve diffraction-limited performance, the telescopewavefront error must be less than ∼λx/13, rms, where λx

is the x-ray wavelength. For a two-mirror system beingused at 171 Å, this equates to a surface error figure of∼5 Å,rms, or better than λ/1300 (λ = 633 nm), rms (neglectingother error contributors such as mirror figure degradationdue to coating, support deformation, misalignments, etc).The ability to manufacture such optics is only startingto be feasible at the end of 1999. In addition, becauseof the short wavelength (relative to visible light) andcurrent detector technology, extremely long focal lengths(>10 m) are required to make use of diffraction-limitedperformance.

Mechanical stresses in the multilayer coating canproduce deformations of the optic, degrading imaging.Recent studies have shown that the stresses do not appearto vary as a function of the layer pair thickness, butdo vary as a function of the equivalent d thicknessand the materials. Thin-film technologists have foundthat coating stresses are also functions of a number ofdeposition parameters including substrate temperature

and deposition process. Thin-film stresses may beestimated from Stoney’s equation:

σf = Est2s

6(1− νs)tfC (9)

where Es, νs and ts are, respectively, Young’s modulus,the Poisson ratio and the thickness of the substrate, tfis the thickness of the coating, C is the change inradius of curvature of the substrate after the coatingis applied and σf is the induced stress (a negativevalue of σf is compressive). Typical stresses rangefrom −1200 to +200 MPa. These stresses can producesignificant deformations on thin substrates, requiringeither a modification of film or substrate design, orthe use of post coating stress reduction techniques suchas annealing (annealing, however, may cause diffusionacross the interfaces, degrading reflectance).

Design, specification, and performanceTypical design constraints for an x-ray telescope willinclude focal length, spectral bandwidth, collecting area,resolution or fractional encircled energy, and size andweight. Not always independent of one another, theseconstraints determine whether a normal or grazingincidence telescope is desired, mirror figure requirements,coating design and element size.

Spectral bandwidth can determine whether an NI or aGI telescope is more appropriate. For extreme ultravioletand very soft x-rays with wavelengths greater than∼100Å,NI multilayer telescopes offer lower cost, better imagingand larger collecting area. Spectral regions above 0.1 keVare better served by GI optics which offer much greaterreflection efficiency and better performance.

Diffraction by the annular entrance aperture can besignificant at low energies. The intensity as a function ofradial position x is expressed as:

I (x) = 4[1− (ri/ro)2]2

[J1(xo)

xo−(ri

ro

)2J1(xi)

xi

]2

where

xo = 2πrox

λfand xi = 2πri

x

λf(10)

where ro and ri are the outer and inner radii of theannular aperture, respectively, f is the focal length, J1

is the first-order Bessel function and λ is the incidentwavelength. For NI telescopes diffraction does not limitimaging performance until one produces extraordinarilyprecisely figured optics, as discussed above in the section‘Multilayer telescopes’. Depending on telescope size,aperture diffraction can be more significant for GI optics.For example, the X-Ray Telescope (XRT) on the joint US–Japan Solar-B satellite has an aperture width of about0.6 mm and an outer radius of∼170 mm. At a wavelengthof 60Å, the 68% point of the fractional encircled energy (theradial integral of the normalized point spread function)



is about 1.4 arcsec in diameter. For the large Chandraaperture diffraction is negligible.

Grazing angle, focal length and radius are related bythe relation

α = 14

tan−1 r0

Z0(11)

where α is the graze angle at the intersection of theparaboloid and hyperboloid surfaces, and r0 is the radiusof the virtual joint (Z0 is not the focal length but itis fairly close). The choice of graze angle affects theshort-wavelength limit of the telescope because of theprecipitous loss of reflection efficiency near the criticalangle. Therefore spectral bandwidth coupled with focallength can drive the maximum allowable optic size.From these few parameters one can derive the opticalprescription of the paraboloid and hyperboloid usingthe method of Van Speybroeck and Chase (1972). Ifwe constrain the graze angle of the hyperboloid at thevirtual joint to be 3 times that of the paraboloid, then theprescription of the optics can be determined as

P = Zo tan(4α) tan α (12)

d = P (13)

ande = cos(4α)[1 + tan(4α) tan(3α)] (14)

where the optical surfaces are represented as

r2p = P 2 + 2PZ + 4e2Pd/(e2 − 1) paraboloid (15)

andr2

h = e2(d + Z)2 − Z2 hyperboloid (16)

and Z represents the coordinate along the optical axis andis zero at the system focus and e is the eccentricity. Theeffective focal length of the system, fL, is expressed as

fL = 2e2d

e2 − 1. (17)

(The ratio of fL/Zo is a function of α and is approximatelyequal to 1 + 0.0025α2

deg.)Typically, performance of GI systems is expressed in

terms of the encircled energy (usually as a fraction or as aper cent). This is the integral of the point spread functionwithin a given angular diameter. One can think of theencircled energy as the fraction of incident flux that fallson a detector pixel or some other suitable area (e.g. a slitfor some spectroscopy applications). It is a measure ofhow compact or spread out the light from a point sourceis imaged. The encircled energy may be normalized orscaled two different ways. In the first, normalization iswith respect to the flux that leaves the last focusing optic(reflection losses are ignored). In the second approachthe normalized encircled energy is scaled by the entranceaperture area and has units of area.

Image quality is affected by mirror surface imperfec-tions (figure error and microroughness) which scatter the

incident flux and broaden the point spread function. Forconceptual purposes mirror errors may be loosely groupedinto three categories. Low spatial error frequencies (longerror periods) produce small-angle scatter which has onlya small effect on the image ‘core’. Mid-spatial-frequencyerrors produce intermediate-angle scattering and can havea significant effect on the image core, limiting resolution—the ability to discern two closely placed sources. Highspatial error frequencies (typically microroughness) pro-duce large-angle scatter (this might be on the order of 10–100 arcsec and greater) which degrades image ‘contrast’—the ability to find a dim source in the presence of a brightsource. The definition of what error frequencies corre-spond to the various bandwidths is dependent on the sys-tem requirements and grazing angle. On Chandra, errorsthat scatter x-rays by 0.5 arcsec would be considered mid-frequency. On XMM, mid-frequency errors might be con-sidered as those that scatter by 5 arcsec.

A zeroth-order estimate of encircled energy can bemade using the total integrated scatter, or TIS. This termrepresents the fraction of incident energy that is scatteredby a surface (or equivalent surface) with a given rmssurface figure error. The fractional encircled energy (orEE) is approximately 1 − TIS. The EE is both incidentwavelength and included angle dependent, so the choiceof rms figure error must take this into account. This is doneby using a bandlimited rms amplitude and making theassumption that this amplitude includes all surface errorfrequencies that will scatter the light outside the desiredregion of interest. The encircled energy is expressed(approximately) as

EE ≈ exp[−(2kσ sin α)2] (18)

where k is equal to 2π/λ, σ is the bandwidth-limitedeffective rms surface error (in the plane of incidence)and α is the average graze angle. Using encircledenergy goals we can estimate an acceptable value forσ , or, alternatively, σ can be used to estimate EE. (Ofcourse, we want to leave some additional margin as werecognize that pointing stability–jitter, alignment, etc willall degrade performance.) To determine the bandwidth forwhich σ applies we use the grazing incidence equivalentof the grating equation (making use of the small-angleapproximation)—

λf = θs sin α (19)

where λ is the incident wavelength, f is the spatialfrequency of the surface error and θs is the angle throughwhich the radiation is scattered. All errors of higherspatial frequency will scatter through larger angles and falloutside the region of interest. In doing this analysis we aremostly concerned with errors in the axial direction. Thisis because the deviation of specular rays out of the planeof incidence by azimuthal errors is reduced by a factor ofsin α. Similarly, the scattering distribution is elongatedin the plane of incidence by a factor of 1/ sin α (thegrazing incidence foreshortens the spatial error periodsin the plane of incidence, making the errors appear as a



higher-frequency and increasing the scattering angle (seeequation (19))).

A more exact method of computing telescoperequirements and estimating performance is obtainedusing the scalar scattering theory of Beckmann andSpizzichino. (Vector scattering theory may also be usedfor modeling performance. Some references by E Churchare listed in the bibliography.) The output intensitydistribution is given as

dPd*total

= StrehldP

d*spec+

dPd*spec

⊗ dPd*scatt

. (20)

Strehl is the Strehl ratio and equals the right-hand sideof equation (18). dP/d*spec is the intensity distributionin the absence of scattering–diffraction, such as resultsfrom large-scale geometric figure and alignment errors,dP/d*scatt is the scattered intensity distribution and ⊗signifies a convolution operation. Using scattering theoryto determine dP/d*scatt and integrating over d*yields theencircled energy. After much algebra this can be reducedto a one-dimensional integral where the scattering is afunction of the Fourier transform of the optic axial figureerror power spectrum density (PSD). Thus, by measuringthe axial surface errors (i.e. the axial one-dimensionalprofiles), computing the error PSD and measuring thegeometric in- and out-of-plane contributors, we canestimate performance. Initially appearing complex, thisformalism is very useful because optic (and replicationmandrel) fabrication processes are essentially surfacefilters. Imaging performance is a function of surfaceerror PSD, and optical fabrication technology provides ameans to operate on the surface error PSD. This approachprovides much more fabrication guidance than merelyspecifying a single bandlimited rms amplitude as an opticrequirement. This approach also correctly takes intoaccount the impact of the frequency content of the figureerrors in estimating performance.

FabricationGrazing incidence telescopes are produced by one ofseveral methods: direct polishing of the optics, replicationof polished mandrels or forming of thin foils. Multilayertelescopes are manufactured using standard precisionoptical fabrication methods but require the application ofthe multilayer coatings.

Material considerations for x-ray optics are important.Since all x-ray observations are made in space, the opticalelements must be strong enough to survive rocket launchacoustic and seismic loads. At the same time, optic weightmust be minimized to reduce payload weight. Dependingon the complexity of the spacecraft an active temperaturecontrol system may or may not be present, so mirrorelements should have both a low coefficient of thermalexpansion (CTE) and a uniform one. This minimizes opticdistortion due to temperature variations and gradients inthe spacecraft as it orbits the Earth.

Typical glasses used for x-ray telescopes areZerodurTM, a glass ceramic produced by Schott with an ex-tremely low CTE, and ULETM (which stands for ultralowexpansion) produced by the Corning Glass Co, anothervery low-CTE glass. Fused quartz was used for the Ein-stein Observatory mirrors. Potential new materials in-clude silicon carbide which has an extremely high strengthto weight ratio offering the promise of very light weightnon-replicated optics.

Optical fabrication of grazing incidence opticsOptic blanks arrive from the glass supplier as roughmachined pieces, within 250–1000 µm of final dimensions(by contrast, final allowable figure error might be only afew hundred Ångstroms, rms, or 104–105 times better).A carefully scripted material removal schedule utilizingever-finer grinding grits and polishing compound is usedto coarse figure the optics and remove residual machiningstresses (subsurface damage, or microcracks) that candegrade figure stability or even lead to catastrophic failureunder load. The inside (optical surface), outside andends of the blank are polished for damage removal, toprovide a surface that is easily cleaned and kept freeof contamination, and to provide a controlled bondingsurface for the optic mounting system.

The optical surface is figured in (grinding and) pol-ishing using computer-controlled fabrication technologywhere a computer provides a set of commands to movethe polishing head in a controlled fashion over the sur-face of the optic. Typically, the optic is supported with itsaxis nearly horizontal, the optic is made to rotate about itsaxis at a controlled rate, and the polishing tool is drivenaxially along the optical surface to describe a fixed path(e.g. a spiral), but with variable path velocity. Since ma-terial removal is inversely proportional to the path veloc-ity, control of that velocity allows correction of the opticalfigure. The velocity commands result from deconvolvingthe polishing tool material removal profile from the op-tic surface error map. The fabrication process, along withthe associated measurements, is iterative with typical er-ror correction rates of 50–90% per iteration depending onthe optics manufacturer. For Chandra, the correction rateswere typically 85–95% depending on the error spatial fre-quency content.

Smoothing of the optic, critical for controlling scatter,is usually performed as a separate operation from figuring.ROSAT optics were smoothed to surface roughness levelsof about 3 Å, rms, and the Chandra optics were smoothedto about 2–3 Å, rms.

The combination of computer-controlled fabricationand smoothing function as spatial frequency filteroperations, making them readily linked to the PSDrequirements that can be generated from the encircledenergy requirement.

Because of the unusual optic geometry, mostfabrication and metrology equipment is custom designedby the manufacturer. For larger optics such as on ROSATand Chandra specialized handling equipment is alsorequired.



Fabrication of grazing incidence optics by replicationWhen extremely good imaging (less than a few arcsecdiameter) is not required, or when extremely largecollecting area is required, replicated optics can be asuperior alternative to individually fabricated glass optics.In this approach a set of replication mandrels that arethe inverse of the desired final figure are fabricated first.Multiple replicas, each the inverse of a mandrel andtherefore possessing the desired mirror figure, are thenformed.

The advantages of replication are several. First, whenmultiple copies of the same mirror are required, replicationis cheaper and less time consuming than individuallyfabricating each mirror. Second, many more shells canbe nested within one another (without unduly growingthe size of the outermost elements) because the replicasare typically much thinner than glass elements. Thisresults in higher telescope ‘throughput’. Third, becauseof their relative ‘thinness’ replicated optics may weighsubstantially less than corresponding glass elements,reducing payload weight. (As an example, the largest X-ray Multi-mirror Mission (XMM) replicated nickel opticsare less than 1/4 the weight per unit axial length ofthe corresponding size ZerodurTM Chandra counterparts.)These advantages mean that either the instrument cancarry multiple copies of the telescope optics and detector(such as XMM) or multiple identical satellites can be placedin orbit (such as planned for Constellation-X). Replicationis a way to achieve much larger collecting area and highertelescope throughput with lower weight per unit area atless cost than with non-replicated elements.

The disadvantage of replication (to date) is thelevel of figure quality which may be achieved, limitingimaging performance to ∼14 arcsec, half power diameter(or 50% encircled energy). The mandrels themselvesmay be fabricated with the same accuracy as large glasselements. The difficulty arises in attempting to removethe lightweight replica from the mandrel, maintain itsfigure against internal stresses created while producing thereplica and then support it for flight, without introducinglow spatial frequency distortions in the figure.

The strengths of replicated optics and individuallyfigured optics telescopes such as XMM and CXOare complementary. Replicated telescopes, with theirlarge collecting area and good imaging capabilities, aregenerally more useful for imaging spectroscopy. Forbright objects, however, telescope throughput is not alimiting factor and the full imaging capabilities of glassoptics can be brought to bear. Replicated optics are moresuitable for some observations of very faint objects ordeep sky surveys where the photon noise limit obviatesfiner resolution, but the limiting angular resolution mayalso result in an inability to resolve discrete sources.Alternatively, glass optic telescopes may be limited inthe number of deep surveys possible owing to theextended observing time needed to collect enough photonsnecessary to make full use of their imaging.

Several methods of replication are briefly describedbelow.

Nickel replicas An aluminum mandrel is coated with∼200 µm of electro-less nickel (Kanigen). The electro-less nickel surface is loose abrasive ground and computer-controlled polished or diamond turned to the nominalfigure (after diamond turning additional figuring may berequired). The surface is smoothed (polished) to about 5 Å,rms, roughness, to complete the mandrel. The replica isproduced by first depositing 100–200 nm of gold on theelectro-less nickel surface and then electroplating the goldwith nickel to the desired thickness (∼1 mm). The replicais separated from the mandrel by cooling the mandrel (thecoefficient of thermal expansion for aluminum is abouttwice that of nickel). The gold coating separates withthe nickel replica because its adhesion to the electroplatednickel is much greater than to the Kanigen. Multiplereplicas may be made from a single mandrel before themandrel needs to be refurbished. Replicas of this type havebeen used for the BEPPOSAX, JET-X and XMM. Replica sizesrange from 300 mm long, 68 mm diameter on BeppoSAX to600 mm long, 700 mm diameter on XMM. The half-energywidth of the point spread function at 1.5 keV is 13 arcsecfor XMM.

As previously mentioned, internal stresses in theelectroplated nickel will deform the replica when it isremoved from the mandrel. Potential solutions to the lackof structural rigidity of these replicas include the use ofstiffening structures fixed to the replica during the nickelplating process and the use of ceramics or silicon carbide(SiC) in place of the nickel replica substrate. In the latterapproach the SiC substrate is formed to near final shapeby chemical vapor deposition (CVD) to a second mandrelslightly larger (102 µm) than the first. The first mandrel iscoated with 100 nm of gold, and then the SiC substrateis positioned around it. The gap is filled with epoxy,and, after curing, the gold-coated epoxy/SiC replica isseparated from the Kanigen-coated aluminum mandrelby cooling. The CVD SiC has little residual internalstress and has a Young’s modulus approximately 3 timeslarger than that of nickel. The SiC replica can be madesubstantially thinner than the nickel replica. A similarprocess using a beryllium substrate instead of SiC has beenused on the Exosat program. An alternative approachin which aluminum oxide is plasma sprayed directly onthe gold-coated Kanigen/aluminum mandrel is also underinvestigation at the time this article is being written.

Epoxy replicas In epoxy replication an aluminum foilserves as the replica substrate. A glass (PyrexTM orZerodurTM) mandrel is coated with ∼100 nm of gold.A thin (∼100 µm) aluminum foil is pre-formed to theapproximate shape of the coated mandrel but slightlyoversized. Both the gold-coated mandrel and the (inner)surface of the foil are sprayed with epoxy and the twoare then mated. The epoxy film, several tens of micronsthick, is cured in an oven before the foil is removed.The foil maintains the smoothness of the polished glassmandrel. Epoxy replicas have not at this time achievedthe imaging capabilities of nickel replicas, but the replicas



Table 1. An abbreviated list of some of the more notable x-ray telescopes.

Effective Energyarea Resolution bandwidtha

Mission Type (cm2) (HPD, arcsec) (keV) Comments

Skylab GI 42 (collecting ∼2 arcsec 0.2–2 First x-ray telescope;(1975) area) resolution solar observationsEinstein GI ∼200 at 1 ∼15 0.2–4.5 First telescope observatory;Observatory keV discovered 7000+ sources(1978–81)RXRSb NI ∼50 ∼2 arcsec 17.3, 25.6 First NI solar telescope(s);(1987) (collecting resolution nm 6.3 cm diameter primary mirror

area)ROSAT GI 400 at 1 keV ∼5 0.1–2.4 4 Au coated Zerodur shells;(1990–9) discovered 150 000+

sourcesASCA GI 1300 at 1 keV, 174 0.5–10 Conical foil Al mirrors,(1993) 600 at 7 keV Au coat over lacquer,

4 separate telescopesBeppoSAX GI 330 at 1 keV 60 0.1–10 Nickel-replicated conical(1996) optics,

30 nested shellsTRACE NI ∼20 1 17, 20, 28 30 cm diameter primary,(1998) nm 2 m focal length solar telescopeChandra GI 800 at 1 keV 0.5 0.1–10 Highest resolution, 4 shells,(1999) largest mirror 1.2 m diameter

transmission gratingsXMM GI 4650 at 1 keV, 14 0.1–12 Nickel replicas,(1999) 1800 at 8 keV 3 telescopes, 58 shells each,

reflection gratingsConstellation-X GI 15 000 at 15 (<10 keV) 0.25–10, Replicated optics (type to be(200?) 1 keV 60 (>25 keV) 6–40 determined), ∼80+ shells,

1500 at largest 1.2 m diameter>6 keV grating

a For NI telescopes the tabulated values represent the approximate centers of the reflection bandpass(es).b Stanford University/NASA Marshall Space Flight Center Rocket X-Ray Spectroheliograph.

are substantially lighter owing to their much thinnerwall and less dense substrate material. To date limitingperformance is ∼1 arcmin.

Segmented foil replicas Segmented foil mirrors have beenused on the Broad Band X-Ray Telescope (BBXRT), ASCA

and the Sodart telescope on board SPECTRUM-X-GAMMA.Here the mirrors do not form the complete surface ofrevolution, but instead form (typically) only a singlequadrant. Aluminum foils less than a millimeter thick andoptically smooth are rolled into a 90 section of the desiredconical shape (note that the mirrors, being true cones,are only approximations to Wolter I telescopes). Plasticdeformation (rather than elastic bending) is employed toproduce a final shape that more closely approaches thenominal and does not introduce stresses into the supportstructure. After rolling, the foils are dip-coated withan acrylic lacquer to improve (reduce) scattering due tosurface roughness. Care is taken to avoid introducingvariations in lacquer thickness that would change theoptical shape of the mirror, degrading imaging. Mirrorsof this type are very inexpensive to produce but have notachieved better than a few arcminutes resolution. In part

this limitation is a result of the approximation to a Wolter Isystem, but, more importantly, it appears that the lacqueris ineffective at smoothing errors with spatial periodsgreater than a few microns. These errors significantlyaffect the ability to image at better than the 1 arcmin level.

Fabrication of normal incidence multi-layer telescopesNormal incidence x-ray multilayer telescopes are essen-tially manufactured using the same processes as conven-tional precision optics with two major exceptions: opticsmoothness (or microroughness), which affects both scat-ter and multilayer reflectivity, and the deposition of thex-ray multilayer coating.

As discussed above in the section ‘Reflection frommultilayer coatings‘ reflectance is a function of high spatialfrequency roughness (f > 103 mm−1). Super-smoothingof the surface roughness in this very high spatial frequencyregime to levels of 2–4 Å, rms, is necessary for goodreflectance.

Along with substrate roughness, the multilayercoating is the most critical element of the optics. To achievethe desired reflectance over the correct spectral bandwidthrequires uniformity of coating thickness both across theface of the optic as well as from layer to layer. Coating



density must also be uniform throughout the multilayer.In addition, interlayer diffusion must be minimized asmust thin-film stresses.

Telescope systemsOther parts of a telescope system are the detector, opticalbench and mirror mount or support structure. CHARGE

COUPLED DEVICES (CCDs) have supplanted position-sensitiveproportional counters. CCDs may be either front orback illuminated. Front-illuminated CCDs have the x-rays impinge upon the semiconductor gate side. Back-illuminated CCDs have additional wafer processing tothin the backside of the chip (the side away from thegates) and are then set so that the x-rays impinge upon thebackside. Front-illuminated CCDs provide better energyresolution and slightly higher quantum efficiency at highenergy than back-illuminated CCDS. Back-illuminatedCCDs provide substantially higher low-energy x-rayquantum efficiency. A wide range of pixel sizes areavailable: the Chandra CCDs have 24 µm pixels; the SolarX-ray Imager (SXI) on GOES N and O will have ∼15 µmsized pixels. The long focal length (∼10 m) of Chandracoupled with 24 µm pixels yields ∼0.5 arcsec angularresolution. CCDs provide excellent energy resolution,about 120 eV for the Chandra detectors. Several CCDs maybe arrayed to cover the full field of view of the telescope,and the detectors may also be arrayed so as to be alignedto the curved focal surface, thereby eliminating focus errorthat occurs with a flat detector and a curved focal surface(see equation (6)). Higher imaging resolution is achievedwith less energy resolution using detectors such as theEinstein, ROSAT and Chandra High Resolution Imager.This is a pair of stacked multichannel plates with anelectronic readout. Future developments include the useof microcalorimeters, which will be used for the first timeon Astro-E. This device has an energy resolution of∼12 eVat 6 keV but has a relatively large pixel size and so limitsspatial resolution. Spectroscopy with x-ray telescopes isalso accomplished by the use of transmission gratings thatmay be moved in and out of the focused beam, such ason Chandra, or reflection gratings fixed in the beam as onXMM.

X-ray telescope missionsSome significant x-ray telescopes and performance detailsare listed in table 1.

Future trendsIn grazing incidence telescopes future trends will continuethe development of increased collecting area, improvingresolution and increasing reflectance. The use ofceramics and SiC, including the use of monolithic supportstructures, will improve the limiting performance ofwhat used to be nickel mandrels, while also allowingdenser nesting of mirror shells. Researchers havebeen experimenting with the application of broadbandmultilayer coatings to grazing incidence optics, increasingreflectance at higher energies. With respect to normal

incidence telescopes future development will probablyfocus on achieving diffraction limited performance withincreased focal lengths to more fully utilize the improvedoptics.

BibliographyThere are a multitude of excellent articles in the literaturediscussing the technology of x-ray telescopes. For designissues of GI telescopes one is referred to

Van Speybroeck and Chase 1972 Design parametersof paraboloid–hyperboloid telescopes for x-rayastronomy Appl. Opt. 11 440

An excellent review is found in

Aschenbach B 1985 X-ray telescopes Rep. Prog. Phys. 48 579

Discussions of alternative Wolter-I-like designs withimproved off-axis performance can be found in

Nariai K 1987 Geometrical aberration of a generalizedWolter type I telescope Appl. Opt. 26 4428

Werner W 1977 Imaging properties of Wolter I type x-raytelescopes Appl. Opt. 16 764

Atomic scattering factors for computing reflectance arefound in

Henke, Gullickson and Davis 1993 At. Data Nucl. DataTables 54 (2)

This information is also available on the Wide World Webat URL

http://xray.uu.se/hypertext/henke.html.A second site,

http://www-cxro.lbl.gov:80/optical constantscontains the same data along with a reflectance ‘calcula-tor’. A general review of multilayer coatings can be foundin

Barbee 1990 Advances in multilayer x-ray/EUV optics:synthesis, performance, and instrumentation Opt.Eng. 29 711

Discussions of vector and scalar scattering theory arereviewed by

Church E 1979 Role of surface topography in x-rayscattering SPIE Proc. 184 196

Church E 1986 The interpretation of glancing incidencescattering measurements SPIE Proc. 640 126

For more detailed discussions of developments the readeris best referred to the Society of Photo-Optical Instrumen-tation Engineers (SPIE) annual proceedings for grazingincidence and multilayer x-ray optics. Some recent vol-umes are SPIE Proceedings volumes 3444 (published in1998), 3113 (1997), 2808 and 2805 (1996) and 2515 (1995).

Paul B Reid


X-rays E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

X-raysEnergetic electromagnetic radiation with wavelengthsin the range 10 nm (10−8 m) down to 0.01 nm(10−11 m). Although the boundaries are somewhatarbitrary, wavelengths shorter than 0.01 nm are calledgamma-rays and those longer than 10 nm extremeultraviolet (EUV). Cosmic x-rays are usually describedin terms of photon energies, a wavelength of 10 nmcorresponding to an energy of 80 electron volts (eV) and awavelength of 0.01 nm to 8× 105 eV (80 keV).

X-rays from cosmic sources are absorbed high in theEarth’s atmosphere. Even the most energetic (shortest-wavelength) x-ray photons fail to penetrate much closerto the ground than an altitude of about 40 km.

X-rays were discovered in 1895 by the Germanphysicist Wilhelm Konrad Rontgen (1846–1923).

See also: electromagnetic radiation, electromagnetic spec-trum, electronvolt, x-ray astronomy.


Yarkovsky Effect ENCYCLOPEDIA OF ASTRONOMY AND ASTROPHYSICS

The Yarkovsky effect changes the rotation and the orbitof a body orbiting in the Solar System by the asymmetri-cal re-radiation of thermal energy from the Sun. Theeffect constitutes a non-gravitational force which causesthe orbits of smaller, kilometer-sized asteroids to changeover time.

Yarkovsky Effect


Year E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

YearThe period of the Earth’s revolution around the Sun, orof the apparent motion of the Sun on the ecliptic. It maybe defined in a number of ways, each of which leads to aslightly different value:

Sidereal year. The time interval during which the Sunapparently completes one revolution of the celestial sphererelative to the stars (which, for this purpose, are regardedas being fixed in space). This is equal to the revolutionperiod of the Earth around the Sun as measured relativeto the stars, and is equivalent to 365.2564 mean solar days.

Tropical year. The time interval between twosuccessive passages of the Sun through the vernal equinox.Its length is 365.2422 mean solar days, about 20 minutesshorter than the sidereal year. The difference arisesbecause of the effects of precession. As this definition ofthe year is related to the recurrence of the seasons, the term‘year’, if unqualified, is generally taken to mean ‘tropicalyear’.

Anomalistic year. The interval between twosuccessive passages of the Earth through the perihelionof its orbit which, because of a slow change in the positionof perihelion, is not quite the same as the sidereal year. Itslength is 365.2596 mean solar days.

Gregorian calendar year. This is the value of the yearadopted for calendar purposes, and is equal to 365.2425mean solar days. For practical purposes it can be taken asequal to the tropical year (the difference amounts to 0.0003mean solar days).

See also: calendar.


Yerkes Observatory E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Yerkes ObservatoryYerkes Observatory lies 334 m above sea level in WilliamsBay, Wisconsin. It is a research branch of the Department ofAstronomy and Astrophysics of the University of Chicago.

The observatory was completed in 1897. Itwas financed by Charles Tyson Yerkes, a Chicagotransportation tycoon, but the inspiration behind itsconstruction was George Ellery Hale. The showpiece ofthe observatory was the 1 m refractor, the world’s largesttelescope in 1897, and still the largest refracting telescopeever built.

Until the mid-1960s, Yerkes Observatory housedall of the university department’s activities (includingmanaging the operations for MCDONALD OBSERVATORY inTexas from 1932–62). Today, the 77 acre site provideslaboratory space and access to telescopes for researchand instruction. A substantial fraction of the university’slibrary holdings in astronomy are housed at Yerkes.

The principal telescopes are the 1 m refractor, a 1 mRitchey–Chretien reflector which is used for adaptiveoptics studies, a 0.6 m reflector and an 18 cm Schmidtcamera for wide-field photography.

Recent research at Yerkes includes measuring thevelocities and distances of the furthest star clusterswithin the Milky Way to better determine the massof our Galaxy; spectroscopic measurements of lithiumabundances; spectra of the dust disk around BetaPictoris; and studies of the properties of distant galaxies.Yerkes astronomers are currently developing a camerafor the airborne Stratospheric Observatory for InfraredAstronomy (SOFIA).

For further information seehttp://www.astro.uchicago.edu/yerkes/.


Yohkoh E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

YohkohThe Yohkoh satellite was launched by Japan’s INSTITUTE OF

SPACE AND ASTRONAUTICAL SCIENCE (ISAS) on 30 August 1991for observations of solar FLARES and the solar CORONA in x-ray and gamma-ray wavelengths. Excellent observationshave been made over the last nine years (see figures 1and 2). The satellite has experienced the entire transitionfrom the maximum to the minimum of the last SOLAR CYCLE

and is proceeding to observations of the rising phase ofsolar cycle for the first time.

Yohkoh carries two X-RAY TELESCOPES: the hard x-raytelescope has imaging capability above 40 keV becauseof its utilization of modulation collimators. These hard x-ray images allow us to obtain new information on whereand how energetic electrons are accelerated in solar flares.The soft x-ray telescope is a grazing-incidence x-ray (5–50 Å) telescope to observe the solar corona and solar flareswith high spatial (3 arcsec) resolution. The soft x-raytelescope is sensitive to plasmas with temperatures from2 MK through 20 MK. Two other instruments onboardare the Bragg crystal spectrometer to observe the iron,calcium and sulfur lines from flare plasmas and the wide-band spectrometers to observe flare spectra from 5 keV to10 MeV.

The discoveries made so far by Yohkoh cover awide area of solar physics. In the following, we giverepresentative examples.

(1) Mechanism of solar flares. The Yohkoh data, for thefirst time, showed unambiguously that magnetic recon-nection is responsible for significant energy release in solarflares; intense heating occurs owing to magnetic reconnec-tion possibly with standing slow magnetohydrodynamicshocks.

Figure 1. X-ray image of the Sun taken with Yohkoh.

(2) Microflares and coronal heating. Yohkoh x-raymovies showed numerous small transient brighteningsfar smaller than flares in energy (microflares) as well asstructural changes of coronal magnetic fields. Magneticreconnection is again responsible for these brightenings.Although the heating mechanism of the persistent coronaremains unknown, these observations imply that anensemble of nanoflares, which are too faint to be resolved,maintains the persistent solar corona.

(3) X-ray jets. Yohkoh discovered x-ray jets with speedreaching several hundreds km s−1. X-ray jets may bedriven by the slingshot effect of the reconnected field linesor maybe due to chromospheric evaporation caused byheat flow from the reconnection site.

(4) Electron acceleration in solar flares. The Yohkoh hardx-ray telescope showed that hard x-ray flares essentiallyhave double-source structures, which are located at thefootpoints of the soft x-ray loop. This clearly indicatesthat a rich population of non-thermal electrons (up to10 MeV or higher) are created in flare loops in associationwith magnetic reconnection and emit hard x-rays withbremsstrahlung. A surprising discovery from the hard x-ray observations is the detection of an impulsive hard x-ray source located above the soft x-ray loop. The loop-topsource is due to bremsstrahlung of high-energy electronswith energy up to at least 50 keV. The energizationprobably occurs at the fast magnetohydrodynamic shockformed by the supersonic outflow from the reconnectionsite.

MAGNETIC RECONNECTION with associated slow magne-tohydrodynamic shocks appears to be a powerful engineto convert magnetic energy to plasma kinetic and ther-mal energies. The hard x-ray observations also show thatthe resultant supersonic outflow creates fast shock, whichconverts the flow energy to further heating and/or non-thermal acceleration of electrons and protons as a result ofFermi acceleration. In spite of the large magnetic Reynoldsnumber of the solar corona,Yohkoh has delivered unam-biguous evidence of magnetic reconnection in flares andin the corona.

Yohkoh is therefore revolutionizing our understand-ing of the solar corona and the behavior of magnetizedplasmas in general. Yohkoh shows for the first time howthe Sun dynamically relaxes its magnetic energy, buoyantas a result of the subsurface dynamo mechanism, by theprocess of magnetic reconnection. The formation of thesolar corona with frequent sporadic (10–100 Alfven transittimes) energy releases including solar flares, solar windand coronal mass ejections is a consequence and manifes-tation of the energy release process. This suggests thatmagnetic energy conversion through magnetic reconnec-tion is likely to be a common occurrence in the cosmos.

BibliographyFor details on the instrumentation on board Yohkoh, theinterested reader is referred to volume 136 of Solar Physics(1991).

Saku Tsuneta


Yohkoh E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Figure 2. Gigantic flare arch observed with Yohkoh.


Young Stellar Objects E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Young Stellar ObjectsYoung stellar objects (YSOs) are stars in the earlieststages of development. There are two principal kindsof YSOs: protostars and pre-main sequence stars. YSOsare intimately associated with fundamental by-productsof the star formation process such as bipolar outflows,circumstellar jets, masers, Herbig–Haro objects, andcircumstellar (protoplanetary) disks. YSOs are alwaysfound within or near interstellar clouds of gas anddust. Most YSOs populate giant molecular clouds anda significant number are members of fully or partiallyembedded stellar or protostellar clusters.

BackgroundStars are the fundamental objects of the astronomicaluniverse. They are self-gravitating balls of (mostly)hydrogen gas which convert the primary product of thebig bang (hydrogen) into heavier elements. For mostof its life a star maintains a state of stable equilibriumin which the inward force of gravity is balanced by theoutward force of pressure. This internal pressure isgenerated by the energy released in the nuclear reactionswhich burn hydrogen at the star’s center. These nuclearreactions are also the source of the star’s luminosity.During the time the star burns hydrogen in its core itmaintains a fixed radius and luminosity and consequentlya constant surface temperature. The exact value of theequilibrium radius, luminosity and surface temperatureof a star depends almost exclusively on one parameter,the star’s mass (see STELLAR MASSES). The more massivethe star, the greater its luminosity, size and surfacetemperature. Together, hydrogen burning stars of varyingmass form a well defined locus of points in the observableluminosity–effective temperature plane (the HERTZSPRUNG–

RUSSELL DIAGRAM). This locus of points is called the mainsequence and the hydrogen burning phase of a star’s lifeis known as the main sequence phase. Main sequencestars range in mass from about 0.08–100M (solar masses).Stars with smaller masses have insufficient weight to raisetheir central temperatures enough to enable hydrogenfusion. (Such objects are referred to as brown dwarfs.)Stars with larger masses are presumably too luminous tohold on to their outer atmospheres.

The formation and early evolution of starsSTAR FORMATION and early STELLAR EVOLUTION occur prior tothe main sequence or hydrogen burning phase of stellarlife. Most stars in the Galaxy originate in giant molecularclouds (GMCs). With temperatures seldom in excess of10 K and maximum dimensions of 50–100 pc, GMCs arethe coldest objects in the universe and the largest objects inthe Galaxy. With masses in excess of 105 M they also rivalglobular clusters as the most massive objects in the Galaxy.GMCs are composed primarily of molecular hydrogen andare characterized by mean densities of ∼100 cm−3. About10% of the mass of a typical GMC is in the form of dense

cores with n(H2) ∼ 104 cm−3. It is within such denseregions that stars are formed.

Although there is no complete theory of stellarorigins, basic astrophysical considerations result in thefollowing general picture of star formation and earlystellar evolution. Stars form through the gravitationalcollapse of dense molecular cloud cores. Before beingincorporated into a star, the interstellar material mustincrease its density by 20 orders of magnitude and collapseto a size nearly 7 orders of magnitude smaller than thedimensions of the original dense cloud core. Because thedust in the cloud is optically thin, internal energy gainedby its collapse is effectively radiated away. Thus the cloudmaterial remains isothermal and dynamically collapses.The collapse also proceeds in a non-homologous manner,with the inner regions becoming denser and collapsingfaster than the outer regions which are left behind.Eventually, the innermost infalling material becomesdense enough to be optically thick to its own radiationresulting in the development of a central quasistatic stellarcore surrounded by an infalling envelope. Thus is aprotostar born.

After the formation of its embryonic core, theprotostar enters the accretion phase of protostellarevolution. During this time the central stellar coregradually gains mass via the accretion of material from itsinfalling envelope. Before being finally incorporated ontothe growing stellar core, accreting material must dissipatethe gravitational potential energy lost in infall, giving riseto an accretion luminosity:

Laccretion = GM∗MR∗

where M∗, and R∗ are the mass and radius of theprotostellar core and M , the mass accretion rate. Thisaccretion luminosity can be a significant component ofa protostar’s luminosity. The mass accretion rate isgenerally expressed as:

M = m0a3

G

where a is the effective sound speed and m0 is aconstant which is sensitive to the initial conditions in thecollapsing core. Different theories of star formation predictsomewhat different values ofm0. In some modelsm0 varieswith time and is quite large (∼10) in the early stages ofcollapse, falling to near unity in the later stages. However,for most protostars, this constant must be near unity to beconsistent with the observed luminosities of protostellarobjects.

Initially the mass of the embryonic stellar core is verysmall, approximately 10−2 M and the central temperatureof the core is not sufficient to ignite nuclear reactions.As the embryonic core begins to grow its luminosity isdominated by accretion. The mass of the protostellar coreincreases asM∗(t) = Mt . Once a protostellar core reaches amass of roughly 0.2–0.3M, its central temperature reaches



106 K and deuterium-burning nuclear reactions are ignitedproviding the protostar with another source of luminosity.ACCRETION of new material and a fully convective coreenable the protostar to continue burning deuterium as itgrows. If placed on the HR diagram the protostellar corewould follow a trajectory parallel to and somewhat abovethe deuterium-burning main sequence which itself liessignificantly above and roughly parallel to the (hydrogen-burning) main sequence. A protostellar core can continueto gain mass by accretion until its central temperaturereaches 107 K and hydrogen fusion is ignited, at whichpoint the embryonic core reaches the main sequence. Thisoccurs when the cores attain masses around 7–8 solarmasses. Presumably, the protostar can still continue togrow and increase its mass. However, for reasons notyet fully understood, most protostars cease growing longbefore this point. In any event, stars which end up withmasses in excess of 7–8 solar masses have a very differentpost-protostellar evolution than stars which end up lessmassive.

The physical reason for this becomes apparent if onecompares the timescale for pre-main sequence evolutionwith that of protostellar collapse. The timescale for thegravitational collapse of a cloud core, the free-fall time, isdetermined largely by ρ, the density of the cloud:

τff =√

3π32Gρ

.

For the typical mean density (n ≈ 104 cm−3) of a cloudcore (of either low or high mass) the free–fall time isabout 4× 105 yr. The timescale for pre-hydrogen-burningevolution is the Kelvin–Helmholtz time:

τKH ≈ GM2∗

R∗L∗

which is very rapid for a high mass star (i.e. ≈104 yr forM∗ = 50 M) and relatively slow for a low mass star(i.e. ≈3 × 107 yr for M∗ = 1 M). More importantlyfor high mass stars τKH < τff and these stars beginburning hydrogen and reach the main sequence before thetermination of the infall or collapse phase of protostellarevolution. On the other hand, for low mass stars τKH > τff

and low mass stars have an observable pre-main sequencestage of stellar evolution.

The initial conditions for pre-main sequence evolu-tion are those which describe the mass, radius and lumi-nosity of an embryonic protostellar core at the the pointin time that infall and accretion cease. For this reason thelocus of points on the HR diagram which traces the initialstarting points of PMS evolution for all stars is called thebirthline. The position of a given star on the birthline is afunction of the mass it has acquired by the end of its proto-stellar accretion phase. Presumably, the protostellar evolu-tion of a given star is identical to that of all other stars untilthe time that the star stops accreting and reaches the birth-line. Prior to its appearance near the birthline, a young

stellar object is surrounded by an infalling protostellar en-velope which renders it invisible. Once the envelope is ei-ther mostly accreted onto the protostellar core or removedby some agent, the protostar becomes a visible PRE-MAIN

SEQUENCE STAR. Its initial luminosity and surface tempera-ture place it near the birthline which, for low mass stars,is itself basically coincident with the deuterium-burningmain sequence on the HR diagram. Because the abun-dance of deuterium in the young star is relatively low, itis burned up very rapidly. Without accretion to replen-ish the burned deuterium, nuclear reactions cease and thestar slowly contracts to the main sequence. The timescalefor this quasistatic contraction to the main sequence is theKelvin–Helmholtz time.

Observational characteristics of the low massyoung stellar objectsThe evolutionary status of a normal star is usuallydetermined by its placement on the HR diagram. Thisin turn requires measurements of two quantities: stellarluminosity and effective temperature. A star can bemeaningfully placed on the HR diagram provided it emitsa black-body-like spectrum that can be characterized by asingle effective temperature. For YSOs this is not alwayspossible. This is because throughout their formation andearly evolution stars are intimately associated to varyingdegrees with natal gas and dust. This circumstellar gas anddust can absorb and reprocess substantial amounts of theradiation emitted by a young stellar object, significantlyaltering its spectral appearance. The circumstellar gasand dust associated with a young stellar object has aspatial extent considerably greater than that of its stellarphotosphere. Consequently, emitting circumstellar dust,which is in radiative equilibrium with the stellar radiationfield of the buried star, will exhibit a wide range of(effective) temperature and the emission that emerges willhave a spectral distribution much wider than that of asingle temperature black body. In addition, at the opticalwavelengths the youngest objects are rendered completelyinvisible by the obscuration of opaque circumstellar dustand a significant fraction, if not all, of their luminousenergy is radiated in the infrared portion of the spectrum.

To determine the evolutionary status of a youngstellar object requires knowledge of its overall broadbandenergy distribution, particularly at infrared wavelengths.The shape of the broadband energy distribution of a YSOwill depend on both the nature and distribution of thesurrounding material. Consequently, the shape of thespectrum will be a function of the state of evolution ofa YSO. The earliest (protostellar) stages, during whichan embryonic star is surrounded by large amounts ofinfalling material, have a very different infrared signaturethan the more advanced (pre-main sequence and mainsequence) stages, where most of the original star-formingmaterial has already been incorporated into the youngstar itself. This is clearly apparent for low mass (0.1–2.0 M) YSOs where the majority of known sources canbe meaningfully classified by the shapes of their observed



λ (µm)

Log

(νF

ν )

Log (ν )

YSO ENERGY DISTRIBUTIONSYSO ENERGY DISTRIBUTIONS

Protostars:Protostars:

Pre-Main Sequence Stars:Pre-Main Sequence Stars:

Class 0 Class I

Class III

101001000

151311 12 14

Class II

blackbody

x 10

Figure 1. The classification scheme for the spectral energydistributions (SEDs) of young stellar objects. A vertical fiducialline is located at the infrared wavelength of 2.2 µm for reference.(Figure courtesy of C J Lada.)

spectral energy distributions (SEDs) into one of four broadclasses designated 0, I, II and III (see figure 1).

Embedded sources: protostarsClass 0 and I sources are characterized by SEDs thatpeak in the submillimeter and far-infrared portions ofthe spectrum indicating that the SEDs are dominatedby emission from cold dust (see CLASSIFICATION OF STELLAR

SPECTRA). These objects are the most deeply embeddedYSOs.

Class I SEDs are broader than a single black-bodyfunction. Longward of two microns these SEDs usuallyrise with increasing wavelength producing a huge ‘excess’of infrared emission compared to that expected from anormal stellar photosphere. They typically exhibit thesilicate absorption feature at 10 µm wavelength. ClassI sources derive their large infrared excesses from thepresence of large amounts of circumstellar dust. Thesesources are usually deeply embedded in dense molecularcloud cores and rarely exhibit detectable emission inthe optical portion of the spectrum. However, theyare detected at near-infrared (i.e. 2.2 µm) wavelengthsand frequently associated with small (infrared) reflectionnebulae. Indeed, a significant fraction, in some casesall, of the near-infrared emission from a class I sourceis scattered light. Roughly half of the class I sourcesexhibit atomic emission lines in the infrared. Otherwise,

the infrared spectra of these objects are found to befeatureless and heavily veiled. Class I sources are almostalways associated with very energetic, relatively massive,collimated outflows of cold molecular gas known asbipolar outflows. The luminosities of class I sources inregions of low mass star formation typically range between0.1–100L. Class I sources are relatively rare among YSOsin molecular clouds and statistical arguments suggest agesof these sources of order 1–5× 105 yr.

Class 0 sources are considerably more extinguishedand embedded than class I sources. Their energydistributions peak at submillimeter wavelengths andmost are not detected at wavelengths shortward of20 µm. Unlike class I sources, their energy distributionshave widths similar to single-temperature black-bodyfunctions however, they are characterized by extremelylow temperatures, 20–30 K! All are associated with bipolarmolecular outflows which are typically more energetic andmuch better collimated than those associated with class Iobjects. On the other hand, as a whole class 0 sources arenot significantly more luminous than class I sources. Inaddition, some observations suggest that class 0 sourcesemit significantly more submillimeter radiation on smallspatial scales than do class I objects. In particular, thecircumstellar mass traced by submillimeter measurementswithin 1000 AU of a class I source is usually found toamount to a fraction of a stellar mass, while for a class0 source this mass can be comparable to that of thecentral star. Class 0 sources are relatively rare makingup roughly 10% of embedded sources. This suggests thattheir lifetimes are only of the order of ∼104 yr, which isconsistent with estimates of the dynamical ages of theirassociated outflows.

Asubset of the class I sources, known as flat-spectrumsources, merit additional comment. Flat-spectrum sourcesare embedded YSOs that have SEDs intermediate betweenclass I and II sources. Unlike the standard class I source,these objects are often visible stars which display extremeT Tauri characteristics. That is, they are optical emission-line variable stars, whose optical spectra display littlein the way of photospheric absorption features and areheavily veiled (most likely by excess continuum radiationfrom accretion shocks at the stellar surface). Indeed,the most well known member of this group is T Tauriitself. Like class I sources, these objects are also veiledat infrared wavelengths, but not as strongly. Roughly halfthe flat-spectrum sources display atomic and molecularabsorption lines at 2.2µm. This is significant because theseare the only embedded objects for which the inner stellarcomponent can be spectrally classified and placed on theHR diagram. These objects are found to be characterizedby late-type (M) photospheres whose luminosities placethem near or above the birthline. Flat-spectrum sourcesappear to be objects in transition between the class I andII phases and can be considered ‘optical protostars’.

The embedded sources are generally considered tobe protostars. Operationally, we define protostars asYSOs in the process of accumulating into a stellar-like



HOT

COLD

Kinematic Signature of Protostellar InfallKinematic Signature of Protostellar Infall

NGC 1333-IRAS2

LSR Velocity (km/s)

Tem

perature (K)

Figure 2. The expected kinematic signature of infall from gas surrounding a protostellar source. Oval curves represent lines ofconstant infall velocity. The thick dashed line represents blue-shifted gas, the thick solid line red-shifted gas. Along a singleline-of-sight the radio telescope beam intercepts the τ = 1 surface in the outer, colder regions of the cloud at redshifted velocities andthe inner, hotter regions of the cloud at blueshifted velocities producing an asymmetric emissionline profile. The observed spectrumof carbon monoxide emission from the protostellar source NGC 1333 IRAS2 in Perseus is also shown. (CO spectrum adapted fromWard-Thompson et al 1996 Mon. Not. R. Astron. Soc. 314 625.)

configuration the bulk of the material they ultimatelywill contain as main sequence stars. Evidence for aprotostellar nature of the embedded sources is derivedfrom the following considerations.

First, theoretical models of collapsing, rotating cloudcores predict the density and temperature structure ofprotostellar objects and these models produce SEDs whichclosely match those of known class I sources.

Second, the only viable source for the enormousenergies of the bipolar outflows generated by class 0and I sources is gravity. Specifically, it is the release ofgravitational potential energy by material falling deepinto the potential well of a protostellar system. That is,material falling all the way down to the surface of theembryonic stellar core. Exactly how this energy frominfall and accretion is tapped to drive a protostellar windis unclear. However, observational relations betweenaccretion and outflow diagnostics suggest that the mass-loss rate characterizing the outflows is related to the massaccretion rate as Mwind = f Maccretion, with 0.01 ≤ f ≤ 0.1.

Third, the observed veiling of class I spectra in theinfrared requires the existence of dust very close (i.e.1 AU) to the protostellar surface. Theoretical modelspredict that dynamically infalling protostellar envelopescontain enough such material close to the protostellarsurface to account for the degree of observed veiling.

Fourth (and most significant), kinematic evidence forcollapse has been observed toward a number of mostlyclass 0 sources. Under favorable conditions sub-millimeterand millimeter-wave spectral lines from collapsing cloudswill display a kinematic signature of infall motion (seefigure 2). This signature takes the form of an infall

asymmetry in the line profile in which the redshiftedportion of an optically thick emission line is depressedrelative to the corresponding blueshifted portion. Surveysof embedded sources have revealed the infall symmetrytowards a substantial number (∼ 1

3 ) of class 0 sources. Suchkinematic signatures provide the most direct evidencefor infall and for a protostellar nature for these heavilyembedded sources.

Revealed sources: pre-main sequence starsClass II SEDs peak at visible or near-infrared wavelengths.Like class I sources the class II SEDs are broader than asingle black-body function. However, longward of twomicrons class II SEDs fall with increasing wavelengthusually in a power-law-like fashion. This results in aninfrared excess which, though significant, is much smallerthan that exhibited by class I sources. The infrared excessindicates the presence of circumstellar material associatedwith the star.

Theoretical calculations predicted more than twodecades ago that T TAURI STARS would display such energydistributions if they were surrounded by luminousaccretion disks. Consider an optically thick and spatiallythin disk that surrounds a young star and radiateseverywhere like a black body. Imagine the disk to becomposed of concentric annuli with radial dimension Rand area 2πRR (see figure 3). Each annulus radiates as ablackbody of temperature T (R). The emergent spectrumof the disk will then be the superposition of a series ofblack-body curves of varying T (R). Now if T (R) ∼ R−n,the Wien law tells us that the frequency of maximumemission scales as ν ∼ T (R) ∼ R−n. The luminosity



Figure 3. Schematic diagram of a spatially thin, optically thickdisk and its emergent (class II) spectral energy distribution,which is composed of a superposition of black-body functions.

radiated in each annulus is given by:

Lν dν = 2πR dR σT (R)4 ∼ R2−3n dν ∼ ν3−(2/n).

Therefore, if the temperature gradient in the disk ischaracterized by a radial power law, the emergentspectrum will also be characterized by a power-law slopein frequency or wavelength. In an SED, νLν ∼ ν4−(2/n) orα = (2/n)−4 where α is the slope of the SED when plottedas a function of log λ.

Thus the power-law shape of the infrared portion ofclass II SEDs strongly suggests that the infrared excessarises in an optically thick circumstellar disk. The slopeof the SED is directly related to the temperature gradientin the disk. The slopes of class II SEDs longward of2 µm wavelength are observed to have values in therange between −0.7 and −1.3, corresponding to a rangein n, the index of the disk temperature gradient, of 0.6–0.75. A viscous accretion disk is predicted to produce atemperature gradient characterized by n = 0.75 whichcorresponds to α = −1.33. This also turns out to be thesame temperature gradient and spectral slope predictedfor a flat purely passive disk which derives its luminosityfrom the reprocessing and re-radiation of light it hasabsorbed from the central star. The majority of class IIsources have shallower slopes (typically α ≈ −0.7) whichsuggests that they are surrounded by flared (passive)disks.

The excellent agreement between the predictions ofdisk models and observations suggests that the most likelyinterpretation of the nature of class II sources is that theyrepresent young stars surrounded by circumstellar disks.They differ from class I objects in that they lack large,massive (infalling) envelopes of gas and dust (see DUSTY

CIRCUMSTELLAR DISKS). However, it is interesting to note thatthe infrared to millimeter excess emission from class IIsources is sufficiently large that, if the emitting materialwere spherically distributed and not confined to a highlyflattened structure, such as a disk, the star would suffersignificantly more extinction than is observed. Indeed,disk masses derived from detection of optically thin

Figure 4. Resolved HST image of a circumstellar disk observedin silhoutte aganist the Orion Nebula. (Courtesy of the SpaceTelescope Science Institute.)

continuum millimeter-wave emission range between 0.01–0.1 solar masses. Compelling evidence confirming thedisk interpretation of class II SEDs has been provided byresolved images of YSO disks obtained by interferometricobservations at millimeter wavelengths and dramaticoptical images by the Hubble Space Telescope (seefigure 4).

Class II sources can be observed at optical as wellas infrared wavelengths. Therefore, considerably moreis known about the nature of these objects than is knownabout class I or class 0 sources. When observed opticallyclass II sources typically exhibit the characteristics ofclassical T TAURI STARS (CTTS). Conversely, most all CTTSstars possess class II SEDs. Classical T Tauri starsare low-mass, pre-main sequence, emission-line variablestars. In addition to excess infrared continuum emission,these stars also exhibit excess emission at ultravioletwavelengths. The optical spectra of CTTS containhydrogen emission lines and frequently various forbiddenemission lines as well. The forbidden lines are believed toarise in stellar winds originating near the surface of thestar. Typically these lines are observed to be blueshiftedand the absence of redshifted emission is interpretedas strong supporting evidence for the presence of anocculting disk close to the stellar surface. The origin ofthe Balmer emission lines is more mysterious. Analysis ofthese lines provides evidence for both mass loss and massloss in these objects. However, both the mass loss andultraviolet excess are believed to be consequences of diskaccretion onto the stars. The accretion rates are typically∼10−8 M yr−1 and, though significant, are relativelylow compared to the typical infall rates encountered in



protostellar evolution.Because they are visible, CTTS can be placed on

the HR diagram, and they are found to lie betweenthe birthline and the main sequence, consistent withtheoretical expectations for pre-main sequence stars.The existence of CTTS near the birthline indicates thatcircumstellar disks form in the protostellar phase ofevolution as part of the collapse and infall process priorto PMS evolution. When their positions are compared totheoretical PMS calculations (which predict trajectories ortracks of early stellar evolution on the HR diagram) onederives ages for CTTS usually between 106 and 4 × 106

years, although such comparisons are hampered by thefact that most existing calculations assume diskless starsand the process of accretion of disk material can havean effect on the evolution of a star’s luminosity andtemperature. Class II sources are relatively common in starformation regions where they typically outnumber class Isources by more than 10 to 1.

Knowledge of the duration of the class II or CTTSphase of evolution has interesting implications for theformation of planetary systems. Analogy with thesolar system suggests that planets are likely formed incircumstellar disks. The duration of the class II phase ofevolution therefore sets limits on the duration of planetbuilding around a young star. The typical lifetime of a classII source deduced from observations of young clustersis about 3–4 million years. The initial stages of planetformation must occur during this interval. However, itis important to note here that the duration of the class IIphase may vary significantly between stars of differingmass and even stars of similar mass formed in differentenvironments or with different initial conditions. Forexample, in the Trapezium cluster in Orion, ionizingradiation from a massive O star appears to be evaporatingthe circumstellar disks of other nearby cluster stars. Thisresults in abbreviated disk lifetimes (≤106 yr) for thosestars and may present difficulties for planet formation intheir short-lived disks.

Class III SEDs typically peak at visible and infraredwavelengths for low mass stars and decrease longwardof two microns more steeply than class II sources.Since their shapes are more or less similar to single-temperature black bodies, the energy distributions of classIII sources are readily interpreted as arising from extinctedor unextincted photospheres of young stars. By definitionthese stars display no infrared excess. However, their lightstill could be substantially extinguished by foregrounddust.

Class III sources can be readily placed on the HRdiagram. Class III sources are found to lie above themain sequence and can be thought of as ‘classical’ pre-main sequence stars in the sense that their positionson the HR diagram can be unambigously compared topredictions of theoretical PMS tracks (for diskless stars).Comparison with such theoretical tracks shows that theages of class III sources range from roughly 106 yr to morethan 107 yr. Although most class III sources have ages

>5 × 106 yr, and are likely candidates for post T Tauristars (PTTS), a significant number have ages which overlapwith those of classical T Tauri Stars (CTTS). This indicatesthat many stars may evolve through the CTTS or classII phase of evolution very rapidly (<106 yr). Class IIIsources are relatively strong (but variable) x-ray sourcesand can be identified in x-ray surveys. In the Taurus regionsuch observations suggest that the population of class IIIsources is at least comparable in size to that of class IIsources. Older class III objects are also found to extendwell beyond the boundaries of star formation regions in x-ray surveys. Class III sources typically produce little or noHα line emission. All class III sources are therefore weak-lined T Tauri stars (WTTS). (Optical astronomers classifyPMS stars with Hα equivalent widths less than 10 Å asWTTS and PMS stars with Hα equivalent widths greaterthan 10 Å as CTTS.)

Bipolar jets, outflows and protostellar evolutionAlthough the SED classes discussed above correspondto distinct physical classes of YSOs, the variation in theshapes of the energy distributions from class 0 to IIIis quasi-continuous. It corresponds to a sequence ofthe gradual dissipation of circumstellar gas and dustaround newly formed stars and represents a continuoussequence of evolution from protostar to main sequencestar. To evolve from class 0 to I to II requires the removalof the circumstellar material contained in a protostellarenvelope. In principle, the clearing of circumstellargas and dust could be accomplished by accreting allthe surrounding material onto the star itself. However,this possiblility conflicts with the observation that starformation is an inefficient process. The cores which formstars contain considerably more mass than the stars whichthey produce. Thus, the removal of circumstellar gas anddust would appear to require some active physical agent.This agent is most likely the energetic bipolar outflow thatis ignited early in the protostellar phase of evolution (seefigure 5).

An unanticipated phenomenon of fundamentalimportance for star formation, bipolar molecular outflowswere only discovered by millimeter wave observations inthe late 1970s and early 1980s. Bipolar outflows are veryenergetic flows of cold molecular gas generally consistingof two spatially separate lobes moving diametrically awayfrom an embedded YSO at hypersonic velocities. Theseoutflows are very massive, often containing considerablymore mass than the central YSO which drives them.This indicates that the molecular outflows primarilyconsist of swept-up material and not ejecta from theembedded driving star itself. Bipolar outflows are themanifestation of an underlying primary wind generatedby the embryonic protostellar core.

Other manifestations of this primary driving windinclude Herbig–Haro objects, circumstellar jets and watermasers. Herbig–Haro objects are clumps of shock-excited gas created by the collision of the primarywind with dense ambient cloud material. Herbig–Haro



Figure 5. HST image of a bipolar outflow at optical wavelengths. The two blobs of emission are the Herbig–Haro objects 1 and 2. Thevisible emission originates in shock-excited gas interacting with a hidden high-velocity bipolar wind generated by an invisible class Iprotostar located in the dark cloud core between the two lobes. (Courtesy of the Space Telescope Science Institute.)

Figure 6. Infrared HST (NICMOS) image of one of thecircumstellar jets that drives the bipolar Herbig–Haro 1–2outflow system shown in figure 5. The class I protostellar sourcewhich produces the jet is so deeply buried that it is not visibleeven in this deep infrared image. It is located near the upperright corner of the image in a region devoid of emission.However, light from the protostar is visible in this image. Theextended infrared reflection nebula, which surrounds the jet, isproduced by light from the obscured protostar which has beenscattered into our line of sight by dust in the cavity created bythe outflow. (Image courtesy of John Bally and Bo Reipurth andthe Space Telescope Science Institute.)

objects emit primarily emission-lines and are brightest atoptical and infrared wavelengths (see HERBIG–HARO OBJECTS

AND EXCITING STARS). They are known to be interactingwith very fast (vwind > 100 km s−1) winds and oftendisplay proper motions of similar magnitude. Similarto molecular outflows, Herbig–Haro objects often exhibit

bipolar morphologies and can extend over distances (i.e.≈1 pc) comparable to the largest molecular outflows.Water maser sources arise in very small but dense regionswhere the conditions are such that certain microwavetransitions of H2O become nonlinearly amplified bystimulated emission and as a result extremely bright.Like Herbig–Haro objects, maser sources often displaysignificant proper motions and likely represent materialinteracting with and swept up by the fast primary wind.

Very close to the surface of the protostellar object,the primary wind is most often manifest by a highlycollimated, circumstellar jet (see figure 6). These jetscontain sufficiently hot and ionized gas to emit at opticalas well as centimeter wavelengths. Although such jetsappear to originate very close (≤50 AU) to the protostellarcore, they can also extend to large distances from thecentral object. Such jets are frequently observed toterminate at Herbig–Haro objects which have the shapes ofextended bow shocks. Such jet–bow shock systems oftenexhibit bipolar morphologies similar to bipolar outflows.

Bipolar outflows are individually energetic enoughto disrupt a protostellar envelope as well as an entiredense cloud core. Indeed, the masses of some outflows(e.g. MonR2) are known to be comparable to those of thecloud cores in which they originate. Bipolar outflows aretherefore capable of driving the evolution of a protostarfrom the embedded class 0 to the revealed class IIphase. Moreover, in the process of removing surroundingcircumstellar material, outflows play a significant role indeterming the final mass of the central star. Exactly howthis mass is determined and how the initial spectrum ofstellar mass (i.e. the initial mass function or IMF) originatesare open questions.

The high frequency of association with embeddedsources indicates that such outflows are ignited veryearly in protostellar evolution and have a durationcomparable to the lifetimes of the protostars that drivethem. This raises the apparent paradox that protostars aresimultaneously sources of both infall and outflow. Theresolution to this paradox, as mentioned earlier, is thatoutflows must be accretion driven.



Bipolar outflows are almost always associated withembedded class 0 and I sources and very rarely with classII or III stars. There is some evidence to indicate thatoutflows decline in energy as a source evolves from a class0 to class I stage. Without massive infalling envelopes,accretion rates in class II sources are not robust enoughto continue to drive energetic outflows. To evolve from aclass II to III stage likely results from the accretion of theremaining circumstellar disk material onto the star.

Disk accretion in the class II (or even class 0 and I)stages may not be steady. There is a subset of YSOs,called FU Ori stars, which appear to be characterizedby non-steady accretion histories. These stars displayeither episodic brightness variations of large amplitudeor spectral signatures associated with unusually high diskaccretion rates. The prototype of this class, FU Orionis,experienced an intense outburst in which its luminosityincreased by a factor of 100 in less than one year. Prior tothis outbust, FU Ori appeared to be a typical CTTS. Theaccretion rate necessary to power its outburst luminosityis ∼10−4 M yr−1 which is significantly in excess of theaccretion rates characteristic of protostellar objects (2–6 × 10−6 M yr−1) and class II sources (∼10−8 M yr−1).FU Ori outbursts are believed to be short lived (∼100 yr)and for most sources not as energetic as that experiencedby FU Ori itself. However, such bursts of accretion mayoccur frequently during the early evolution of a YSO anda star could accrete a significant portion of its final massin such episodes.

Young stellar objects of high massDespite their somewhat different history of formationand early development, high mass (M∗ > 2 M) starscan sometimes display characteristics similar to low massYSOs. This is particularly true for intermediate mass stars(2 ≤ M∗ ≤ 10 M) whose formation and early evolutionis most similar to that of low-mass YSOs. Intermediatemass pre-main sequence stars are known as HerbigAeBe (HAEBE) stars. More massive analogs of CTTS,the HAEBE stars are emission-line stars which typicallydisplay class II SEDs and therefore possess circumstellardisks. They range in age between 0.5–5 × 106 yr, similarto CTTS. Luminous (L > 102 L) class I sources are thelikely precursors to HAEBE stars.

For more massive stars (i.e. M∗ > 8 M) there isno pre-main sequence phase. When a protostar growsto a mass in excess of about 10 M, it begins to emitcopious amounts of ultraviolet radiation. This results inthe dissociation and ionization of all the hydrogen in theimmediate circumstellar vicinity of the protostellar core.A small, dense region of hot (104 K) ionized gas, knownas an ultra-compact H II REGION, is then produced. Becausethe pressure in the ionized gas can be as much as threeorders of magnitude higher than that in the surroundingmaterial, the H II region expands rapidly, at the speed ofsound (10 km s−1) characterizing the ionized gas. It quicklyevolves into a compact H II region and then ultimatelya fullblown H II region. In the process the protostar

quickly evaporates and disrupts its circumstellar disk andinfalling envelope and perhaps even those of neighboringstars as well. The lifetime of such massive protostars islikley very short (<104–105 yr). Yet, high-mass protostarsalso manage to produce bipolar outflows and masers.Little more is known about the protostellar evolution ofmassive stars. Their brief protostellar lifetimes, coupledwith the fact that massive stars are intrinsically rare, makesit extremely difficult to find and investigate examplesof such protostars. Such objects can be expected to belocated at very large distances from the Earth making theirdetailed study difficult. Moreover, massive protostars areso energetic that they significantly alter the conditions inthe surrounding natal material making the processes ofprotostellar and early post-protostellar evolution difficultto distinguish.

BibliographyBachiller R 1996 Bipolar molecular outflows from young

stars and protostars Ann. Rev. Astron. Astrophys. 34111–54

Beckwith S V W and Sargent A I 1996 Circumstellar disksand the search for neighbouring planetary systemsNature 383 139–44

Lada C J and Kylafis N D (eds) 1991 The Physics of StarFormation and Early Evolution (Dodrecht: Kluwer)

Lada C J and Kylafis N D (eds) 1999 The Physics of StarFormation and Early Stellar Evolution II (Dodrecht:Kluwer) (in press)

Levy E H and Lunine J (eds) 1993 Protostars and Planets III(Tucson, AZ: University of Arizona Press)

Shu F H, Adams F C and Lizano S 1987 Star formation inmolecular clouds: observations and theory Ann. Rev.Astron. Astrophys. 25 23–81

Charles J Lada


Young, Thomas (1773–1829) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Young, Thomas (1773–1829)English physicist who developed the wave theory of lightto explain interference. He explained the aberration ofstarlight by suggesting that the Ether was not disturbedby the motion of the Earth through it.


Yunus, Abu’l-Hasan ibn (950–1009) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Yunus, Abu’l-Hasan ibn (950–1009)Astronomer, born in Egypt, calculated trigonometricfunctions for use in astronomy and wrote an astronomicalhandbook, al-Zij al-Hakimi al-kabir, the Great Tables of Caliphal-Hakim, which contained observations made by Yunus,including 30 lunar eclipses used by SIMON NEWCOMB in hislunar theory. Yunus was also an astrologer, predictingthe date of his own death in seven days’ time. He madepreparations, locked himself in his house and recited theKoran, dying on the day predicted (see also JEROME CARDAN).


Z Andromedae (Z And) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Z Andromedae (Z And)Z And is the prototypical SYMBIOTIC STAR. Discovered as amundane M-type VARIABLE STAR with a 2 mag range in visualbrightness, Z And rose to prominence in the 1920s whenJOHN STANLEY PLASKETT reported a peculiar class A spectrumwith strong nebular lines on spectra near maximum light.Frank Scott Hogg later noted TiO bands on Plaskett’sspectra. The simultaneous presence of A-type featureswith the TiO bands observed in much cooler M-type starsled Hogg to speculate that Z And and a handful of similar‘stars with combination spectra’ might be a new type of(binary) stellar system. Roughly a dozen of these systemswere known in the 1940s, when PAUL MERRILL coined theterm symbiotic star for the class.

Basic propertiesAs in all symbiotic stars, the behavior of Z And canbe divided into two states, eruption and quiescence.During quiescence, the optical spectrum resembles an M-type giant, with a strong red continuum and deep TiOabsorption bands (figure 1). This continuum peaks in thenear-infrared, where CO absorption bands dominate thespectrum. In addition to the M-type stellar photopshere,optical data show a weak blue continuum and intenseemission lines from a variety of ionized species. The H IBalmer lines are the strongest optical lines; He II, He Iand occasionally [Ne V] and [Fe VII] are also intense. Thestrong Balmer emission jump at 3646 Å indicates that anionized nebula produces much of the optical continuum.This ionized nebula is roughly 104–106 times denser thanthe low-density gas in a typical planetary nebula or H IIregion. In most planetary nebulae the [O III] and [Ne III]forbidden emission lines are strong, and the He I tripletlines are usually three times more intense than the He Isinglet lines. In Z And, the [O III] and [Ne III] lines areweaker than intercombination lines from C III], N III] andO III]; the He I singlet lines often rival the intensities of thetriplet lines.

Ultraviolet spectra of Z And confirm the hot stellarsource suggested by the prominent He I and He II emissionlines. The blue continuum from this component risessteadily into the far-ultraviolet and must peak shortwardsof the Lyman limit. The shape of the UV continuumsuggests a temperature exceeding 105 K for the hot star.The emission line spectrum is particularly intense, withHe II, C IV and N V as the most prominent features. Theselines are fairly broad and indicate material motions inexcess of ∼200 km s−1. Despite the intense far-ultravioletcontinuum and the presence of high-velocity gas, Z Andis not a bright x-ray source due to strong hydrogenabsorption between the binary system and the Earth. Thesystem is a modest far-infrared and radio source due tothe large amount of dust and gas in the BINARY SYSTEM.

The quiescent photometric and spectroscopic varia-tions of Z And are remarkably complex. The main varia-tions in the system are phased with the orbital period of759 days. Orbital motion of the hot component has not

4000 5000 6000 7000 80000

10

20

Wavelength (Å)

Flu

x (1

0-13 e

rg c

m-2

s-1

Å-1

)

Figure 1. Spectra of Z And in quiescence (lower curve) andoutburst (upper curve). The red end of the spectrum isdominated by TiO absorption bands and a strong Hα emissionline. The blue continuum is weak in quiesence but rises by afactor of 3–10 in outburst. Several strong emission lines fade inoutburst, while other increase in intensity.

been detected. The small amplitude of the red giant or-bital motion, ∼7 km s−1, implies masses of ∼2M for thered giant and 0.5–1.0M for the hot companion if the ratioof stellar masses is 2–4 as in other symbiotics. The in-tensity of the blue continuum and many strong emissionlines varies by factors of 2–3 every orbit. The continuumvariations are usually smaller, ∼0.2–0.4 mag in the opticaland up to 1 mag in the ultraviolet. The system is brightestwhen the giant lies behind the line-of-sight to the hot starand is faintest when the giant lies in front of the hot star.This behavior suggests that the RED GIANT occults the emis-sion line region. The giant does not eclipse the hot staritself; variations in the bright ultraviolet continuum andthe polarization of several emission lines indicate an or-bital inclination of ∼45. This inclination is much smallerthan the 70 inclination needed for eclipses of the hot star.

Z And also varies on extremely short time scales.Some emission lines, notably O III λ3444, vary markedlyfrom one night to the next, while other lines appearconstant. Some emission lines, such as H I and He I,have velocity fluctuations that are several times larger thanthe orbital motion of 7 km s−1, but these have not beeninvestigated in much detail.

In addition to these quiescent variations, Z Andundergoes occasional outbursts. Figure 2 shows acomplete light curve for the system that includes all knownmajor outbursts. Most outbursts last ∼7 years, and beginwith a slow increase in brightness, V ∼ 10.5–11.0 to V ∼9.5, accompanied by a periodic 0.5–1.0 mag oscillation.This phase is followed by a more rapid rise to visualmaximum, V ∼ 8.5–9.5, where the optical colors decreaseand the UV colors increase. Visual maximum is followedby a slow decline with a continuing 0.5–1.0 mag oscillation.Within this train of oscillations, the interval between



successive minima is 10–15% shorter than when the systemis quiescent. During a complete outburst, there is exactlyone more minimum present than in a corresponding timeperiod during quiescence. This behavior is observed in atleast two other symbiotic stars, CI Cyg and AX Per, andresembles the ‘superhumps’ observed in the SU UMa classof cataclysmic variables.

The fine details of these phenomena change fromoutburst to outburst, but the overall behavior isremarkably repeatable. During the rise to maximum,the high ionization emission features fade as the opticalcontinuum brightens. Some high ionization emissionlines, such as [Ne V] and [Fe VII], appear to weakenconsiderably during the rise in brightness, but the He IIlines often remain strong throughout the eruption. Atmaximum light, the optical spectrum resembles an A-type or F-type supergiant. Bright H I lines with strongabsorption cores are prominent, and He I emission linesmay also have weak absorption features. These absorptionand emission lines fade as the system declines from visualmaximum. The TiO bands and high ionization emissionlines reappear as the system fades and re-establish theirpre-outburst levels once the optical continuum returns tonormal.

The history of outburst observations at otherwavelengths is shorter than at optical wavelengths.The infrared brightness of the giant remains constant,indicating that the giant does not participate in theeruption. In the ultraviolet, the continuum longwardsof ∼1400 Å appears to rise in step with the visualbrightness. The continuum at shorter wavelengths hasnot always followed the optical variations, but the dataare sparse. The low-ionization emission lines track theoptical variations; higher-ionization emission lines trackthe short-wavelength ultraviolet continuum. At longerwavelengths, the radio continuum flux declined as theoptical brightness increased in the most recent eruptions.The light curve for these last two eruptions, however, doesnot resemble the prolonged outbursts in the 1930s, 1940sand 1950s. Observations of future outbursts will be neededto see whether or not this behavior is ‘typical’.

A new feature of Z And’s quiescent behavior isthe discovery of a persistent 28 min oscillation in theoptical continuum. The oscillation has an amplitude of0.002–0.005 mag and is visible during quiescence andthe most recent small outburst. The short period of thecoherent oscillation suggests an association with the hotcomponent. Although this variation could be due topulsations in the hot star, a rotational period is the mostlikely explanation. If true, the period demonstrates thatthe hot component must be a WHITE DWARF instead of a mainsequence star.

Finally, two of the strongest optical emission lines inZ And occur at 6830 Å and 7088 Å. These lines remainedunidentified with any known atomic or moleculartransition until Hans Schmid proposed an association withRaman scattering of the O VI 1032 Å and 1038 Å lines offneutral hydrogen atoms surrounding the binary. With

24000 25000 26000 27000 28000

11

10

9

8

29000 30000 31000 32000 33000

1110

98

34000 35000 36000 37000 38000

1110

98

Vis

ual M

agni

tude

39000 40000 41000 42000 43000

1110

98

44000 45000 46000 47000 48000

1110

98

49000 50000 51000 52000 53000

1110

98

Julian Date (2,400,000 +)

Figure 2. Optical light curve for Z And using the 10 day meansof observations from the AAVSO International Database.

high-quality optical spectropolarimetric observations,Schmid and colleagues confirmed this hypothesis andderived an orbital inclination of roughly 45 from thevariation of polarization position angle with orbital phase.They followed up these measurements with a similaranalysis of the symbiotic AG Draconis. The optical Ramanlines at 6830 Å and 7088 Å are strong on the opticalspectra of many symbiotic stars, which should allowthe measurement of accurate orbital elements for manysystems.

InterpretationFigure 3 shows our current picture of the Z And binarysystem. The two stellar components are a red giant with abolometric luminosity of roughly 3000L and a low-masswhite dwarf with a luminosity of roughly 1000L for an



xw1

yw1

Figure 3. Schematic picture of the Z And binary. The binaryconsists of a red giant star (right) and a small white dwarf star(left). The red giant loses material in a stellar wind indicated bythe arrows. In the orbital plane, the white dwarf deflects windmaterial into an accretion disk. This gas spirals inwards throughthe disk until it encounters the white dwarf magnetosphere.Disk material then falls onto the white dwarf along magneticfield lines, indicated as dashed lines. High-energy photons areproduced by accretion through the disk and by material fallingthrough the white dwarf’s magnetosphere. These photonsionize the red giant wind. The wind heats up as it approachesthe white dwarf and accretion disk. (Adapted from Kenyon(1986) and Sokoloski and Bildsten (1999).)

assumed distance of 1.5 kpc. The high luminosity of thered giant drives a low velocity stellar wind, which forms anextended envelope surrounding the binary system. Someof this material feeds an accretion disk around the whitedwarf. This gas drifts inwards through the disk and fallsonto the hot component. The energy generated by massinfall—either through the accretion process itself or bynuclear burning of the accreted material—produces theblue continuum; the high-energy end of this spectrumionizes some of the surrounding nebulosity. The densestportion of the ionized nebula lies within an outflowingwind near the photosphere of the red giant. This gas hasa density of at least 1010 cm−3. The density in the gassurrounding the binary system as a whole is much lower,108 cm−3 or less.

The simple picture is deceptive, because some ofZ And’s behavior remains unclear. Despite relativelygood knowledge about the system geometry, the massof the white dwarf is uncertain by a factor of two. Therelative contributions of the disk and the hot white dwarfto the total luminosity are not known. The origin ofsome emission lines is also poorly understood. In currentmodels, low-ionization emission lines, such as H I andHe I, form close to the red giant; higher ionization lines,such as He II and C IV, form close to the hot white dwarf.The line fluxes and profiles generally fit this picture, butthe radial velocities of the lines do not. This failureprobably reflects our poor understanding of wind-drivenmass transfer in a close binary system.

The origin of Z And’s eruptions is also a mystery.In the simplest model, the outbursts of Z And and othersymbiotic stars are due to thermonuclear runaways similarto those that produce classical novae. These eruptionsoccur when hydrogen from the accretion disk collectson the surface of the white dwarf. Nuclear reactionsbegin when this material reaches a critical mass. Thesereactions cause the white dwarf to expand in radius bya factor of 10–100, thus producing the large observedincrease in optical brightness. This dramatic increasein radius occurs at roughly constant luminosity, so thetemperature of the white dwarf cools from over 105 Kto roughly 7000 K. Aside from the apparent constancyof some high ionization emission lines, much of theultraviolet and optical spectroscopic behavior is consistentwith this picture. However, the rapid recurrence time of10–20 years is much smaller than has been achieved inany calculation of an outburst in conditions appropriatefor symbiotic stars. Most calculations imply a recurrencetime of 100 years or more.

Additional study is needed to unravel the natureof Z And’s eruptions. Further work on repetitivethermonuclear runaways may reveal a mechanism thatallows outbursts every 10–20 years. The main alternativeto the thermonculear picture, disk instabilities similarto those responsible for dwarf NOVA eruptions, naturallyproduces such short recurrence times, but cannot explainthe apparent constancy of the bolometric luminositythroughout an eruption. If future observations show thatthe luminosity is not constant, a disk instability mightprovide a better explanation for the eruptions of thisprototypical symbiotic star.

BibliogaphyFernandez-Castro T, Gonzalez-Riestra R, Cassatella A,

Taylor A R and Seaquist E R 1995 The active phaseof the hot component of Z Andromedae Astrophys. J.442 366

Kenyon S J 1986 The Symbiotic Stars (Cambridge:Cambridge University Press)

Mikołajewska J and Kenyon S J 1996 The inscrutable hotcomponent in the symbiotic binary Z AndromedaeAstron. J. 112 1659

Schmid H M and Schild H 1997 The polarimetric orbit ofZ Andromedae Astron. Astrophys. 327 219

Sokoloski J L and Bildsten L 1999 Discovery of a magneticwhite dwarf in the symbiotic binary Z AndromedaeAstrophys. J. 517 919

Scott Kenyon


Zollner, Johann Karl [Carl] Friedrich (1834–82) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zollner, Johann Karl [Carl] Friedrich(1834–82)German astronomer, born in Leipzig, was a pioneersolar astronomer, classifying solar prominences. He firstsuggested that the spectral types of stars represent anevolutionary sequence, starting hot and cooling. This ideawas taken up with variations by H C VOGEL and NORMAN

LOCKYER.


Zach, Baron Franz Xaver von (1754–1832) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zach, Baron Franz Xaver von(1754–1832)Hungarian astronomer, became director of the SeebergObservatory (Gotha), organizer of the ‘celestial police’who took it upon themselves to search for the planetmissing, according to BODE’s law, between Mars and Jupiter.Recovered Ceres according to GAUSS’s prediction, when ithad been lost behind the Sun following its discovery byPIAZZI.


Zeeman Effect E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zeeman EffectThe splitting of a spectral line into two, three or morecomponents, that occurs when the source of that linelies within a magnetic field. This phenomenon is namedafter the Dutch physicist, Pieter Zeeman (1865–1943), whodiscovered the effect in the laboratory, in 1896. Theseparation of the components of a line is proportionalto the strength of the magnetic field and the number ofcomponents, and the polarization of the light in eachcomponent depends on the orientation of the field to theobserver’s line of sight. The Zeeman effect enables thestrength and orientation of magnetic fields (for example,the magnetic fields in sunspots) to be measured. Wherethe components are too close together to be resolved intoseparate lines, the line appears broader than would be thecase in the absence of a magnetic field (this phenomenonis called Zeeman broadening).

The Zeeman effect occurs because each of an atom’sorbiting electrons has a small magnetic field (or magneticmoment). When the atom is placed in a magnetic field, theelectrons can align themselves at certain discrete angles tothe magnetic field (the orientations are quantized), eachof which corresponds to a marginally different energylevel. Consequently, each energy level of the atom issplit into two or more closely spaced sub-levels, and moretransitions (movements of an electron from one level toanother) are then possible, each transition correspondingto a spectral line (or a component of a line).

See also: absorption spectrum, atom, emission spectrum,polarization, quantum mechanics, quantum theory,spectrum, sunspots.


Zeeman, Pieter (1865–1943) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zeeman, Pieter (1865–1943)Physicist, born in Zonnemaire, on the isle of Schouwen,Zeeland, Netherlands, Nobel prizewinner for physicsin 1902 with HENDRIK ANTOON LORENTZ ‘in recognitionof the extraordinary service they rendered by theirresearches into the influence of magnetism upon radiationphenomena’. Became professor at Leiden University,where he discovered the splitting of spectral lines bya strong magnetic field, indicating the quantization ofthe spin of the electron, their negative charge, and theunexpectedly high ratio of their charge and mass (e/m).Zeeman predicted that Zeeman splitting should be seen inthe magnetic field of the Sun, and this was completelyverified by GEORGE HALE, at Mount Wilson Observatory,even to the correct interrelationship between the directionsof polarization and the magnetic fields.


Zeiss, Carl (1816–88) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zeiss, Carl (1816–88)Optician and industrialist, born in Weimar, Germany,established at Jena the optics factory noted for theproduction of lenses, microscopes, and telescopes.


Zel’dovich [Zeldovich, Seldowitsch], Yakov Borisovich(1914–87)


Zel’dovich [Zeldovich, Seldowitsch],Yakov Borisovich (1914–87)Russian physicist, worked at the Institute of ChemicalPhysics in Leningrad (later in Moscow), played asignificant role in the development of Soviet nuclear andthermonuclear weapons. In the 1960s he worked onastrophysics and cosmology, including the theory of blackholes, the formation of galaxies and clusters, and the large-scale structure of the universe. He identified the Sunyaev–Zel’dovich effect of a ‘shadow’ in the cosmic microwavebackground caused by intervening electrons in clustersof galaxies. He developed astroparticle physics in thecosmological theory of the Big Bang and started to developa quantum theory of gravity.


Zenith E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

ZenithThe point on the celestial sphere that is vertically above anobserver on the Earth’s surface. It is 90 distant from anypoint on the horizon. The point 180 opposite the zenith,directly underfoot, is the nadir.

See also: nadir.


Zenith Distance E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zenith DistanceThe angular distance, measured along a great circle on thecelestial sphere, between the zenith and a celestial object.The zenith distance of a celestial object is equal to 90minus the object’s altitude.

See also: altitude, celestial sphere, great circle, zenith.


Zenithal Hourly Rate (ZHR) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zenithal Hourly Rate (ZHR)Ameasure of the activity of a meteor shower that takes intoaccount various observational factors. It is defined as thenumber of meteors that would be seen by a single ‘ideal’observer in a cloudless, perfectly dark sky if the radiantwere at the zenith. The ZHR is always greater than theobserved hourly rate. It is calculated as

ZHR = (N/t)× R × L× C

whereN is the number of shower meteors observed in timet (in hours), R is the radiant-elevation correction factor, Lis the limiting-magnitude correction factor and C is thecloud correction factor. The simplest expression for Ris 1/ sin α, where α is the mean elevation in degrees ofthe radiant over the time t ; more complex formulae giveslightly better results, particularly for small values of α.The next correction factor, L, is given by r6.5−LM, wherer is the population index and LM the observer’s limitingmagnitude, 6.5 being the assumed limiting magnitude fora perfectly dark sky. Values of ZHR become unreliablewhen LM is worse than about 5. The population indexis a measure of the magnitude distribution of a shower.Older meteor streams are depleted in smaller meteoroidsand produce a lower proportion of faint meteors; their rvalues are higher (e.g. for the Lyrids, r = 2.9). Youngermeteor streams yield a more even distribution of meteormagnitudes and have lower r values (e.g. r = 2.1 forthe Quadrantids). The third correction factor is given byC = 1/(1 − x), where x is the cloud cover expressed as afraction; if x exceeds one-fifth the calculated ZHR will beunreliable. The error associated with the overall result isobtained by dividing it by

√N . Asimilar formula (without

the R term) can be applied to observed rates for sporadicmeteors to give the sporadic hourly rate (or corrected hourlyrate). For sporadic meteors, r = 3.42.


Zeno of Elea (c. 490–c. 425 BC) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zeno of Elea (c. 490–c. 425 BC)Philospher, born in Elea, Lucania (now southern Italy),formulated Zeno’s Paradoxes, identifying inconsistenciesin the linguistic formulation of the mathematical theoryof infinitesimals. Diogenes Laertius reports that Zenoproposed a universe consisting of several worlds,composed of ‘warm’ and ‘cold’, ‘dry’ and ‘wet’ but no voidor empty space.


Zero Gravity E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zero GravityA term sometimes used to describe the state ofweightlessness or free fall. Weightlessness is the sensationexperienced by a body falling freely under the influenceof gravity, in other words, experiencing no resistance to itsacceleration. An astronaut in a spacecraft which is coastingin a gravitational field experiences no sensation of weightas both he and his surroundings are ‘falling’ at the samerate. Zero gravity does not imply that there is no gravityacting. A person in a freely falling lift will be acceleratingat the same rate as the lift itself; therefore, there will beno relative acceleration (and therefore no force) betweenthe floor of the lift and his feet. He feels no sensation ofweight, but both he and the lift are falling in the Earth’sgravitational field.

See also: weight.


Zhang Heng [Chang Heng] (78–139) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zhang Heng [Chang Heng] (78–139)Mathematician, astronomer and geographer, born in Nan-yang, China, became chief astrologer and minister underthe Emperor An’ti, corrected the calendar to bring it intoline with the seasons. Invented the first seismoscopefor measuring earthquakes, essentially a series of finelybalanced balls that dropped and made a noise when therewas a tremor. He constructed a rotating celestial globe as amodel of the universe and described the 320 stars that canbe named, out of the 11 520 very small stars, apparentlyvisible to the naked eye (it must have been possible to seebeyond magnitude 6.5 in China).


Zhu Xi [Chu-hsi] (1130–1200) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zhu Xi [Chu-hsi] (1130–1200)Philosopher, classical commentator, scientific thinker, andhistorian, born inYu-hsi, Fukien Province, China. He had atheory explaining fossils and realized that mountains hadonce been under the sea. He visualized the Earth’s originsin condensation from cosmic matter, and perceived theuniverse as evolving and spinning from elemental force.


Zodiacal Light E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zodiacal LightA faint cone-shaped glow in the night sky stretchingalong the ecliptic, alternatively known as the counterglowor gegenschein. Given a dark sky and the absence ofmoonlight, it is visible at all times from the tropics. Fromtemperate latitudes it is best seen about an hour and a halfbefore sunrise in the fall or the same time after sunset inthe spring, for at these times the ecliptic makes its greatestangle with the horizon. The zodiacal light is caused bysunlight scattered by interplanetary dust particles in theplane of the ecliptic.

See also: interplanetary dust.


Zodiacal Stars E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zodiacal StarsStars located within 8 of the ecliptic, i.e. within the zodiac.There are 3539 stars listed in the Zodiacal Catalog (ZC), ofapparent magnitude 8.5 and brighter. They are the onlystars that can be occulted by the Moon: observations oflunar occultations are valuable as a check on the Moon’sposition.


Zond E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

ZondSeries of eight Soviet deep space missions. Launched1964–1970. Zond 1 was a failed Venus flyby. Zond 2was a failed Mars mission. Zond 3 (launched July 1965)conducted a lunar flyby. Zonds 4–8 were part of the testprogramme for a Soviet manned lunar mission. Zond5 (launched September 1968) was the first spacecraft tosuccessfully circumnavigate the Moon and return to Earth.Zond means ‘probe’.


Zucchi, Niccolo (1586–1670) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zucchi, Niccolo (1586–1670)Born in Parma, Italy, became a Jesuit, and in 1608, orperhaps 1616, used a lens to observe the image producedby a concave mirror, the first reflecting telescope. Hedescribed it in a book Optica Philosophica, in 1652. Hewas the first to observe the spots on Jupiter, in 1630. Inabout 1640, he is reported to have examined spots onMars, as discovered by Fontana, but this must be regardedskeptically, unless his telescope was better than is believed.


Zwicky, Fritz (1898–1974) E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

Zwicky, Fritz (1898–1974)Swiss physicist, born in Varna, Bulgaria, became professorat the California Institute of Technology. He researchedgalaxies and produced a comprehensive catalog of them.He had an all-inclusive approach to astronomy, whichsuggested that if something was physically possible thenit existed somewhere in the universe—he called this‘morphological astronomy’. In 1934 he predicted theexistence of neutron stars and black holes, formed bysupernovae (a word he coined). His studies of thedynamics of galaxies showed the existence of dark matterdecades before this was generally accepted.


ZZ Ceti Stars E N C Y C L O P E D I A O F A S T R O N O M Y AN D A S T R O P H Y S I C S

ZZ Ceti StarsZZ Ceti is the generic name for pulsating WHITE DWARFS

of type DA, which have a pure hydrogen outer layercomposition. This name is equivalent to DAV (for variableDA white dwarfs) frequently used in the literature.

At the end of their evolution, medium mass stars(M 6–8M) become white dwarfs. After trans-formation of their central hydrogen into helium on themain sequence, and of their helium into carbon andoxygen during the subsequent evolutionary phases—giant branch, horizontal branch (HB) and asymptoticgiant branch (AGB)—the stars evolve towards the whitedwarf stage. The white dwarfs are the naked degeneratecarbon–oxygen (C–O) core surrounded by a tiny heliumlayer (in ≈ 20% of cases—the DB white dwarfs), itselfsurrounded by an even thinner hydrogen outer layer (inthe remaining ≈ 80%—the DA white dwarfs). This lateevolution proceeds either through the PLANETARY NEBULA

phase, during which the stars expel their outer layersinto the interstellar medium, or through the subdwarfsequence (sdO, sdB). The way a star becomes a whitedwarf depends on whether enough hydrogen mass is lefton top of the degenerate core on the HB, allowing the starto increase its luminosity from shell hydrogen burning toclimb the AGB in the HERTZSPRUNG–RUSSELL DIAGRAM (HRD);if not enough hydrogen is left at this stage, the star evolvesdirectly towards a white dwarf through the subdwarfsequence.

As no more nuclear fuel is available once the starshave reached this stage, their subsequent evolution isgoverned by the release of the internal thermal energystored in their degenerate core, and of the gravitationalenergy available during the final contraction of the outerlayers onto the degenerate core. This latter source ofenergy may still contribute to the stellar luminosity duringthe early phases of the white dwarf lifetime but rapidlybecomes negligible as the stars reach their final degeneratestructure.

White dwarfs constitute accordingly the graveyard ofmost of the stellar population in our Galaxy, and in othergalaxies, except for those most massive stars which endas SUPERNOVAE. So they offer astrophysicists a way to testtheir current knowledge of STELLAR EVOLUTION, some phasesof which are still poorly understood.

Along the two currently known channels leadingto white dwarfs, the stars may become pulsationallyunstable: these are the Variable Planetary Nebula Nuclei(PNNV) and the variable sdBs (also called EC14026pulsators). Once on the white dwarf sequence, the starscross three more instability strips as they cool down:(1) the pulsating PG1159 STARS (also called GW Vir stars),which are the nuclei of planetary nebulae after the nebulaehave been diluted in the interstellar matter, at an effectivetemperature between 150 000 K and 80 000 K; (2) thevariable DB white dwarfs (DBV) at about 25 000 K; and(3) the variable DA (ZZ Ceti stars) at about 12 000 K. Eachof these groups of pulsators offers a way of studying

many interesting features of their internal structure atvarious stages of evolution through the powerful tools ofasteroseismology.

The first white dwarf pulsator discovered was theDA HL Tau 76, in 1968. But it is the one discoverednext, in 1971, R548 or ZZ Ceti, which gave its name tothe subsample of pulsating DA. The number of pulsatingwhite dwarfs may seem quite small: only four PG1159,seven DB and 29 DA white dwarfs are known to pulsateat present. Not surprisingly, the ZZ Ceti form thelargest group because (1) the hydrogen atmosphere whitedwarfs (DA) are the most numerous and (2) the coolingtime scale increases with decreasing luminosity and theZZ Ceti define the coolest instability strip, so more starswill be found in a given range of temperature. Butconsidering that we have been able to discover only thosepulsating white dwarfs that are close enough, because oftheir intrinsic faintness, the population of pulsating whitedwarfs may in fact constitute the largest group of VARIABLE

STARS in our Galaxy.White dwarfs have been discovered in two main

ways: from surveys of PROPER MOTIONS and from surveysof objects with a blue color excess. Faint blue stars witha large proper motion have a high probability of beingwhite dwarfs; systematic proper motion studies such asthose conducted by Giclas, Luyten and the Bruce ProperMotion survey have led to the discovery of many faint bluestars later identified spectroscopically as white dwarfs.More recently, systematic surveys aimed at discoveringquasar candidates among UV or blue color excess objectshave also produced new white dwarfs: i.e. the Kiso,Palomar–Green, Hamburg Quasar, Edinburgh–Cape andMontreal–Cambridge–Tololo surveys. White dwarfswith appropriate colors, or atmospheric parameters, tobe candidates for pulsators are subsequently observedthrough fast photometry techniques to search for STELLAR

PULSATIONS. Efforts to search for new pulsating whitedwarfs are still ongoing and new pulsators are regularlydiscovered, mainly among the ZZ Ceti group.

General properties of ZZ Ceti starsZZ Ceti stars are found in a narrow instability stripin the HRD. The hot (blue) and cool (red) edges arebeginning to be well defined since 29 ZZ Ceti stars areknown. However, the precise observational location ofthe instability strip still needs to be improved. Thiscan be achieved partly by improving the precision inthe determination of atmospheric parameters (effectivetemperature Te and surface gravity log g) and partly byfinding more ZZ Ceti stars. The boundaries of the ZZ Cetiinstability strip depend on stellar parameters like Te andthe total mass of the star, but also on less well understoodphysics entering into the models like the efficiency ofconvective transport in the envelope. Accordingly, thelocation of the blue edge of the ZZ Ceti has been debatedfor years and is still a matter of discussion. The rededge is even more poorly understood because convectionbecomes an ever more important conveyor of the energy



radiated by the stars as they cool across the instabilitystrip from blue to red. At the red edge, most of theenergy is transported by convection and there is as yetno satisfactory description of the interaction between thepulsations and the convective motions. Both the blueand red edges depend on the total mass of the stars,the instability strip being wider for more massive whitedwarfs. The observationally determined instability stripis about 1200 K wide for the average ZZ Ceti population,whose mean mass is the same as the mean mass of whitedwarfs: 0.6M. It extends from about 12 400 K on theblue edge to about 11 200 K on the red edge. Because theinstability occurs at the effective temperature for which thehydrogen Balmer lines have their maximum equivalentwidths, the method for determining Te based on fittingthese lines is unfortunately not very sensitive. The methodbased on fitting the satellites of the strong Lα absoptionobserved in the UV spectra of ZZ Ceti stars at λ = 1400 Åand λ = 1600 Å is more sensitive but requires UV spectrawith a high signal/noise ratio, i.e. from the Hubble SpaceTelescope (HST).

Within the uncertainty in the determination of theatmospheric parameters, and the detection limit of thepulsations, there are no stable stars in the ZZ Cetiinstability strip. All DA white dwarfs crossing theZZ Ceti instability strip should become pulsators. Asa consequence, the properties derived for ZZ Ceti fromasteroseismology are presumably applicable to all DAwhite dwarfs.

The pulsations observed in ZZ Ceti stars are non-radial oscillations excited by the κ–γ mechanism andby the response of the surface convection zone, bothresulting from the recombination of ionized hydrogenin the stellar envelope. Theoretical models show thisto happen at the Te corresponding to the blue edge.Additional complications occur at cooler temperatures asthe radiative flux and the convective flux, both perturbedby the oscillations, start interacting nonlinearly, makingthe theoretical modeling much more complex. Most of ourpresent understanding of pulsating white dwarfs comesfrom linear pulsation theory, in which only informationderived from the frequency of the oscillations is used.

Non-radial oscillations in stars manifest themselvesin the form of waves of two types: (a) pressure modes (orp-modes) for which pressure is the restoring force; theyare acoustic waves; and (b) gravity modes (or g-modes),for which buoyancy is the restoring force. In ZZ Ceti stars,p-modes would have periods of the order of 1 second orless, and the corresponding motions of the gas, mainly inthe vertical direction, would have to fight the high gravityof the star, resulting in a very small amplitude. Thesemodes, while predicted to be unstable, have never beendetected. The g-modes have periods two to three ordersof magnitude longer, from about 100 s to about 1200 s.The motion of the gas induced by the pulsations becomesdominantly horizontal close to the surface and the highgravity is no longer an obstacle to the propagation ofthe waves. g-modes propagate as waves at the surface

of the stars, producing temperature fluctuations whichtranslate into flux variations detectable by very sensitiveinstruments like photomultiplier photometers and, morerecently, CCD photometers. They also propagate insidethe stars, but they can do that only in those regions ofthe star where their frequency is lower than the Brunt-Vaisala frequency (N ) and the Lamb frequency. N isthe characteristic frequency with which a particle of gastaken away from its equilibrium position in the radiativepart of the star returns to its original location throughoscillations. The Lamb frequency characterizes the localsound frequency corresponding to the degree of the non-radial mode . This criterion defines cavities inside thestar in which the waves can propagate.

Because of the spherical symmetry of the star, andassuming that the rotation is small enough to preserve thatsymmetry, the non-radial oscillations follow a geometrydefined by spherical harmonics in the horizontal direction.This horizontal structure is characterized by the degree ofthe spherical harmonics, which may be understood as thenumber of lines of nodes of the gas motions on a sphericalsurface, and by the azimuthal number m, which is thenumber of those particular lines of nodes passing throughthe poles defined by the pulsation axis of symmetry. Inthe vertical direction, a given mode has a structure whichis the solution of the eigenvalue problem obeying theboundary conditions at the surface and in the center of thestar. The order k (or n depending on conventions adoptedby various authors) is the number of nodes of the radialcomponent of the eigenfunction. In white dwarfs, becausethe core is degenerate, the g-modes do not propagate deepin the interior as N tends to 0 in degenerate matter. A g-mode of frequency f is consequently reflected at the depthwhere f equals the local value of N . So, contrary to thecase of normal stars, like the Sun and other low-mass mainsequence stars, where the expected detection of g-modeswould allow study of their deep internal structure, in whitedwarfs the observed g-modes allow study of the structureof the outer regions lying above the degenerate core.

An interesting property of the g-modes is that, for agiven degree , and for high order k, modes of successiveorder become regularly spaced in period: this is called theasymptotic regime and the difference in period betweensuccessive orders is the period spacing. Departures fromregular spacing are induced by quasi-discontinuities inthe chemical composition found in white dwarfs at theinterface between the outer H and He layers, and deeperbetween the He and the C–O core: this is referred to asmode trapping.

Modes of same degree and order, but with differentvalue of m (where − ≤ m ≤ ) have the same frequencyin a non-rotating, non-magnetic star (mode degeneracy).Rotation breaks the degeneracy, and in the limit of slowrotation (rotation period long compared to the pulsationperiods) a mode of degree is split into 2+1 modes equallyspaced in frequency. From the frequency separationbetween the components of such multiplets, it is possibleto derive the rotational period of the star. Detecting triplet



or quintuplets also allows an independent check of theiridentification as = 1 or = 2 modes. Similarly, themagnetic field breaks the degeneracy, each mode of degree being split into + 1 components, the frequency shiftbeing proportional to B2.

While theoretical calculations predict a large numberof non-radial modes to be unstable in ZZ Ceti stars, onlya small number are generally observed. The first reasonis trivial: because we see the star as a point source,only large-scale perturbations of the stellar surface can bedetected, i.e. small values ( ≤ 3) because of geometricalaveraging. Another reason is that most of the ZZ Ceti showamplitude variations. Some modes are not found whentheir amplitude is below the detection limit. In most cases,many observational campaigns are necessary to recover amore complete frequency spectrum. But even then, themost complete frequency spectrum is far from the expectedlarge number of non-radial modes that can be producedby the combination of various , k and m. A very efficientselection mechanism must be invoked. The first selectionmechanism that has been invoked is mode trapping: incompositionally stratified white dwarfs, it has been shownthat those particular modes which have a node of theireigenfunction close to the composition interfaces haveminimum pulsation kinetic energy. As a consequence theyhave smaller growth rates. Their amplitudes grow linearlyuntil they saturate the energy available to the pulsations.This simple explanation is rather unsatisfactory since, asthe amplitudes of the modes grow, nonlinear effects couplemodes and the energy stored in the trapped modes can goback and forth in other modes, stable or unstable. Butthis could still act as an efficient selection mechanism.Another selection mechanism implies coupling betweenthe pulsations and convection. Pioneering works showthat convection may damp very efficiently the pulsationsin large frequency domains. But the predicted instabilitystrip in this case does not fit with the observational one.

With the increasing number of known ZZ Ceti stars, itbecomes possible to determine some global properties. Aclear relation between pulsation periods and Te emerges:pulsation periods increase as Te decreases across theinstability strip. This is a consequence of the fact that theunstable g-modes should have periods comparable to orlonger than the thermal time scale at the bottom of theconvection zone. As the ZZ Ceti cool from the blue tothe red edge of the instability strip, the outer convectionzone induced by hydrogen recombination becomes deeperand the thermal time scale at the bottom increases. Thereis also a clear tendency for ZZ Ceti of longer periods tohave larger amplitude, except close to the red edge whereamplitudes fall abruptly to small values. This shows thatthe pulsation excitation mechanism stops being efficient ata given Te (for a given total mass) and this defines the rededge.

Asteroseismology of ZZ Ceti stars: internalstructure and evolutionThe comparison of the observed frequencies withtheoretical calculations based on the linear theory of NON-

RADIAL STELLAR PULSATIONS allows one in principle to derivemany basic parameters about the structure of the STELLAR

INTERIOR and the evolution of ZZ Ceti stars. A necessaryrequirement, however, is to be able to unambiguouslyidentify the pulsation modes. In the case of multiperiodicstars this is not a trivial task. The aliases introduced inthe frequency spectrum by gaps in the data, unavoidablea fortiori in single-site observations, have encouragedthe development of international ground-based networks.The Whole Earth Telescope (WET), operating since 1988,has been quite successful in the asteroseismology ofPG1159 stars and the DBV. For ZZ Ceti stars, which havemuch fewer modes, the safe identification of the pulsationmodes requires additional constraints. Such constraintsmay be provided by time-resolved spectroscopy, usingthe capability of the HST to get UV spectra, or ofthe Very Large Telescope. The wavelength-dependentvariations of the ZZ Ceti spectrum induced by thetemperature changes during the pulsation cycle havedifferent observational signatures according to the degree of the mode, because of the different limb darkeningeffects on the stellar surface averaging. The simultaneouscombination of ground-based fast photometry and time-resolved spectroscopy seems a promising avenue for thefuture of the asteroseimology of ZZ Ceti stars.

From the period spacing, one can in principle deducethe total mass. But this requires that enough modes canbe identified, which is not the case for most ZZ Ceti.Additional constraints from spectroscopy are necessary toget Te and log g. When PARALLAXES are available, the radiusand the mass may be derived and comparison of calculatedg-modes in white dwarf models may solve the ambiguityin the identification of the degree of the observed modes.

The departure from regular period spacing predictedby mode trapping could also be used in principle to derivethe mass of the outer hydrogen layer. But for the samereason that it precludes unambiguous identification, thesmall number of observed modes makes mode trappingvery difficult, and often impossible, to detect. The mass ofhydrogen is then derived from best fits with stellar models,but the solution is generally not unique. Determining themass of the hydrogen layer left on top of the degenerate C–O core is one of the main issues in white dwarf astrophysicstoday because it is the main uncertainty in the white dwarfmodels used in other calculations: i.e. cooling time scaleand calibration of the white dwarf luminosity function,effect of crystallization etc.

Rotational splitting has been measured in someZZ Ceti. The resulting rotation periods, found in the rangeof a few hours to more than one day, are an indicationthat most, but not all, of the angular momentum is lostduring stellar evolution. The exact physical mechanismsare not completely understood, but from asteroseismologyof ZZ Ceti and other white dwarfs we know the final state.

Only upper limits to the magnetic fields in pulsatingZZ Ceti have been derived. No ZZ Ceti is known with amagnetic field greater than a few kG. None of the knownstrongly magnetic white dwarfs has been found to pulsate.



In spite of the fact that the ZZ Ceti have typical coolingtimes of the order of 109 years, the great stability of theirpulsation frequencies should allow us to measure the effectof evolution on the pulsation modes. The cooling timescale depends mainly on the total mass of the degeneratequasi-isothermal core and its chemical composition. Asthe star cools, the internal structure changes, and thespectrum of the eigenmodes changes accordingly. Afew ZZ Ceti stars are regularly observed in a follow-upprogram to ultimately measure the rate of period changesinduced by their evolution (P ). Measuring P would be aunique check that the theory of stellar evolution correctlypredicts the composition of the white dwarf core. The bestcase to date is G117-B15A, for which one has only an upperlimit for P = 2.8±1.7×10−15 s s−1. This is consistent withthe core being a C–O mixture.

In addition to the poorly determined hydrogenenvelope mass in ZZ Ceti, the other major uncertaintyin the calculation of the cooling sequence is the roleof the crystallization phase. The release of the latentheat of crystallization slows the cooling. This is animportant effect when determining the age of the oldestand coolest white dwarfs in the solar neighborhood, whichin turn may be used to determine the age of the galacticdisk. The predicted effect of crystallization has neverbeen confronted to observational tests. Massive DAwhite dwarfs should be in the crystallization phase whilecrossing the ZZ Ceti instability strip. Most DA have thesame mass of 0.6M and are predicted to crystallize ata Te lower than the red edge of the ZZ Ceti instabilitystrip. The only exception is BPM37093, which has a massof 1.1M and is currently the only massive ZZ Ceti onwhich testing the efficiency of the crystallization may beundertaken. Unfortunately, the effect of crystallization onthe oscillation frequency spectrum is small and the presentuncertainty on the hydrogen mass layer hides its signature.

Progress on both the theoretical side of non-radialpulsations and the observations of ZZ Ceti stars shouldallow us in the near future to use the calibratedcooling sequence of white dwarfs to get an independentdetermination of the age of the galactic disk, of globularclusters etc, and to better understand the evolution of thestars whose lives end as white dwarfs.

BibliographyThe essential reference book for understanding the theoryof non-radial pulsations in stars is:

Unno W, Osaki Y, Ando H, Saio H and Shibahashi H 1989Nonradial Oscillations of Stars (Tokyo: University ofTokyo Press)

A recent IAU Symposium dedicated to helio-asteroseismology contains a review on pulsating whitedwarfs in general:

Provost J and Schmider F-X (ed) 1997 Sounding Solarand Stellar Interiors, IAU Symposium 181 (Dordrecht:Kluwer Academic)

The more recent European Workshops on WhiteDwarfs contain many contributions concerning ZZ Cetistars; these are published in:

Koester D and Werner K 1994 White Dwarfs, SpringerLecture Notes in Physics (Berlin: Springer)

Isern J, Hernanz M and Garcia-Berro E 1997 White Dwarfs(Dordrecht: Kluwer Academic)

Solheim J-E and Meistas E G 1999 11th European Workshopon White Dwarfs (ASP Conf. Ser. 169) (San Francisco:Astron. Soc. Pacific)

The original description of the Whole Earth Telescopeconcept may be found in:

Nather R E, Winget D E, Clemens J C, Hansen C J andHine B P 1990 Astrophys. J. 361 309

Recent results on ZZ Ceti asteroseismology may be foundin the proceedings of the WET workshops; the latestpublished one is:

Meistas E G and Moskalik P (ed) 1998 The Fourth WETWorkshop Baltic Astronomy 7 nos 1/2

Gerard Vauclair
