bats using lower-frequency calls (longerwavelengths). Kingston and Rossiter suggestthat the range of echolocation calls in onespecies would generate ‘disruptive selection’because larger bats do not have the sameaccess to small prey as do smaller ones.Theirsis the first demonstration of how adaptiveevolution in bats, and so speciation, mighthave been driven through divergences inecholocation signals.

For their part, Siemers and Schnitzler(page 657)5 examined the behavioural con-sequences of differences in echolocation sig-nals used by similar species of bats to detectprey. In a portable flight-room, they chal-lenged flying individuals of five Europeanspecies of mouse-eared bats (Myotis species)to detect and attack prey sitting on or close tovegetation. This is presumed to be difficultfor the bats because echoes from prey couldbe masked by echoes — ‘clutter’ — from the background. Siemers and Schnitzlerstandardized the degree of clutter in whichthe bats operated, and documented theirbehaviour and foraging performance. Thefive species they used have similar huntingbehaviour and are placed in the same ‘foraging guild’ of bats (the ‘edge space aerial/trawling foragers’). The five speciesmight have been expected to perform at thesame level,but they did not.

In the tradition of Griffin and Spallan-zani, Siemers and Schnitzler controlled forother cues (vision, olfaction) and demon-strated a significant relationship between thedesign of echolocation calls and foragingperformance. Specifically, they showed thatforaging performance in clutter was pre-dictable from echolocation call design, par-ticularly from differences in calls that hadbeen considered minor. Their study is thefirst to provide empirical evidence thatseemingly minor differences in call designcan have real behavioural consequences. Incontrast to Kingston and Rossiter, Siemersand Schnitzler show that signal designs ofsimilar species can converge, reflecting foraging behaviour that is independent ofpresumed evolutionary relationships.

Individually and jointly, these two papersadvance our understanding of the diversity ofecholocation in bats.They have opened doorsto a better appreciation of the variety ofecholocation call designs, including the identification of cryptic species7 — that is,thediscovery that what had been considered asingle species really consists of two or more.Coupled with data on the enhanced echoesthat some flowers return to the bats that polli-nate them8, the new findings also allow betterinterpretation of insights into other pressuresacting on the evolution of bats. For example,another component of the echolocation storyis the listeners — other bats, or other animalsthat, like Griffin, eavesdrop on the calls9,10.Kingston and Rossiter’s work shows clearlythat changes in echolocation calls can affect

news and views

NATURE | VOL 429 | 10 JUNE 2004 | www.nature.com/nature 613

upwards by about 4 GeV/c2.The experimentalerror has been reduced by about 15%, sharp-ening our view of the underlying physics.

The role of the top quark in disentanglingthe fundamental principles of nature istwofold. On the one hand, its large massmakes the top quark a prime target in thesearch for new physics that might so far beunaccounted for. For instance, the long-hypothesized Higgs boson, which is the lastmissing ingredient of the standard model ofparticle physics, is predicted to interact withother particles with a strength that is propor-tional to their masses. So the physics of theheavy top quark would be significantly influ-enced by its interaction with the Higgs boson.On the other hand, the mass of the top quark is a key parameter in the predictions formany observable quantities.Small deviationsbetween measurement and prediction couldbe a signal of new physics, so the uncertaintyin the predictions that arises from the experi-mental error on the top-quark mass limits thesensitivity of experiment to new physics.

The values of several precisely measuredquantities, as predicted by the standardmodel, depend on the square of the top-quark mass, Mt ; their dependence is muchweaker on the as yet unknown mass of theHiggs boson (so far, experiment has exclu-ded4 any mass value below 114.4 GeV/c2).Therefore, in using a so-called global fit ofthe model predictions to all available data,animproved knowledge of Mt better constrains

not only bats’ views of the world, but also theability of one individual to communicatewith another. Siemers and Schnitzler’s resultsset the stage for examining the influence ofcall design on the ability of potential insectprey to detect and evade hunting bats9,10.

The two studies4,5 used species from bothsides of the bat echolocation fence. On oneside, mouse-eared bats, like most bats, sepa-rate call and echo in time (low-duty cycle);on the other, horseshoe bats separate them in frequency (high-duty cycle) (Fig. 1). Bothapproaches to echolocation are ancient,with fossil evidence indicating that they were present in bats some 50 million years ago11.The new data speak to the divergence of calldesign after the evolution of echolocation,but the early history of bats and echolocationremains unclear. There is plenty of oppor-tunity in this line of research: stay tuned forthe next chapter. ■

Brock Fenton is in the Department of Biology,University of Western Ontario, London,Ontario N6A 5B7, Canada.e-mail: bfenton@uwo.caJohn Ratcliffe is in the Department of Zoology,University of Toronto at Mississauga, Ontario L5L 1C6, Canada.e-mail: j.ratcliffe@utoronto.ca1. Griffin, D. R. Listening in the Dark (Yale Univ. Press,

New Haven, 1959).

2. Galambos, R. & Griffin, D. R. Anat. Rec. 78, 95 (1940).

3. Thomas, J., Moss, C. & Vater, M. (eds) Echolocation in Bats and

Dolphins (Univ. Chicago Press, 2004).

4. Kingston, T. & Rossiter, S. J. Nature 429, 654–657 (2004).

5. Siemers, B. M. & Schnitzler, H.-U. Nature 429, 657–661 (2004).

6. Simmons, J. A. & Stein, R. A. J. Comp. Physiol. A 135, 61–84

(1980).

7. Barratt, E. M. et al. Nature 387, 138–139 (1997).

8. von Helversen, D. & von Helversen, O. Nature 398, 759–760 (1999).

9. Pye, J. D. Nature 218, 797 (1968).

10.Fullard, J. H. in Comparative Hearing: Insects (eds Hoy, R. R.,

Popper, A. N. & Fay, R. R.) 279–326 (Springer, New York, 1998).

11.Simmons, N. B. & Geilser, J. H. Bull. Am. Mus. Nat. Hist. 235,

1–182 (1998).

Particle physics

From the top…Georg Weiglein

The top quark is by far the heaviest elementary particle known. Ameasurement of its mass with higher precision has bearing on ourunderstanding of the fundamental interactions of nature.

The basic building-blocks of matter, asfar as we know, are quarks and leptons,together with the force-carrying parti-

cles that mediate their interactions. Quarksand leptons (the latter group including theelectron) are grouped in three generations;the particles in the second and third genera-tions seem a perfect copy of those of the firstgeneration, except that their masses aremuch larger. The top quark is the heaviest ofall quarks and leptons, and is central to someof the most pressing questions in particlephysics. For instance, why is the third-gener-ation top quark more than 300,000 timesheavier than the first-generation electron?Why are there two other quarks with precise-ly the same properties as the top quark butwith very different masses? And what is theorigin of mass itself?

Precise knowledge of the mass of the topquark and its interactions is a key ingredient intesting theory against experimental data. Onpage 638 of this issue1, the DØ Collaborationreport an improved measurement of the top-quark mass, using data taken at the Tevatronproton–antiproton collider at Fermilab, nearChicago. Combining this with previous mea-surements from DØ and its sister experimentCDF, the new world average2 for the mass ofthe top quark is 178.0�4.3 GeV/c2, where cis the speed of light (the mass of the protonexpressed in these units is about 1 GeV/c2).Compared with the previous world average3,the central value of the mass has shifted

10.6 n&v 611 MH 4/6/04 5:29 pm Page 613

© 2004 Nature Publishing Group

the likely value of the Higgs-boson mass. Infact, the 4 GeV/c2 shift in the central value ofMt has shifted the upper limit on the Higgs-boson mass by more than 30 GeV/c2, to 251GeV/c2 (at 95% confidence level)1. Thisupper limit has an important impact on theexperimental strategies used to search for theHiggs boson at present and future colliders.

Finding the Higgs boson in the predictedrange would be another triumph for thestandard model — as, of course, was the dis-covery of the top quark itself. Historically,the mass of the top quark had been predictedfrom a global fit to a wealth of precise measurements made at the LEP and SLCelectron–positron colliders (at CERN inGeneva and at SLAC in Stanford, respec-tively). The top quark was discovered at theTevatron in 1995,with a mass value in perfectagreement with the predicted range.

Although the standard model has passedmany experimental tests with great success, itcannot be the ultimate theory of the funda-mental interactions. This is evident from thefact that it describes only three of the fourknown interactions — namely, the electro-magnetic, weak and strong interactions, butnot gravity. It also has several theoreticalshortcomings and leaves many questionsunanswered. Perhaps the most attractiveframework for extending the standardmodel is supersymmetry. A supersymmetricextension of the standard model could be thelow-energy limit of a more fundamentalhigh-energy theory that would consistentlyinclude gravity and would describe all thefundamental forces in a unified way. Super-symmetric theories predict that there arepartners for all the known particles. Theminimal supersymmetric extension of thestandard model — the ‘MSSM’— comprisesone pair of superpartners for each quark andlepton, superpartners for the force carriers,and five Higgs bosons.

In supersymmetric models, as a conse-quence of the higher degree of symmetry, themass of the lightest Higgs boson can be pre-dicted directly (in contrast to the standardmodel, in which the Higgs mass is a free para-meter, allowing only an indirect determina-tion via a global fit). The predicted mass isvery sensitive to the mass of the top quark,scaling as Mt

4 — an even more pronounceddependence than in the standard-modelcase. Figure 1 shows the prediction5,6 for thelightest Higgs-boson mass in the MSSM: theeffect of the change in the top-quark mass, to178.0 ± 4.3 GeV/c2, is clearly seen. The directexperimental detection of the Higgs bosonwould enable its mass to be measured with anaccuracy below the 1% level. Thus, a preciseknowledge of Mt with an accuracy even better than presently available will be crucialfor Higgs physics in supersymmetric exten-sions of the standard model7.

Besides having an important impact onHiggs physics, the top-quark mass influences

many other predictions of the MSSM — forinstance, the masses of the superpartners ofthe top quark and the strengths of their inter-actions. The ultimate goal is to connect thepredictions of the MSSM, or other exten-sions of the standard model, with a morefundamental theory at a higher energy scale.This may provide evidence for the unifica-tion of all of the forces of nature into a singlefundamental interaction. Measurementsmade at the energy scales directly accessibleto us in collider experiments can be extrapo-lated to very high energy scales, but for this to be reliable a precise knowledge of Mt iscrucial7. If the extrapolation is sufficientlyprecise, it may even give us clues about thestructure of the unified force itself.

Further progress will require new experi-mental data — both the discovery of newparticles, such as the Higgs boson or super-symmetric partners, and more precise measurements of observable quantities thatallow a sensitive test of the underlying theory.Among these, improving the accuracy of themeasurement of the top-quark mass willcontinue to be of the utmost importance.From data taken during the present phase

news and views

614 NATURE | VOL 429 | 10 JUNE 2004 | www.nature.com/nature

tanβ

100101

110

120

130

140

Mas

s of

ligh

test

Hig

gs b

oson

(GeV

/c2 )

Minimal supersymmetric standard model

Uncertainty due to error on top-quark mass

Previous value for top-quark mass, 174.3 GeV/c2

New value fortop-quark mass,178.0 ± 4.3 GeV/c2

Figure 1 The mass of the lightest Higgs boson inthe minimal supersymmetric standard model(MSSM). The predicted value5,6 is shown as afunction of the parameter tan� , which relatesthe properties of the different Higgs bosons of the MSSM to each other (the other MSSMparameters are chosen such that they maximizethe resulting value of the Higgs mass). Thepredicted Higgs mass is sensitive to the value of the top-quark mass used in the calculation.The solid line indicates the prediction using thenew measurement of the top-quark mass fromthe DØ Collaboration1; the white band indicatesthe uncertainty of the prediction that resultsfrom the error on the top-quark mass. Thedashed line shows the situation before the newmeasurement (the previous experimental errorof �5.1 GeV/c2 is not shown). Based on the newvalue of the top-quark mass, an upper bound onthe mass of the lightest MSSM Higgs boson ofabout 140 GeV/c2 is established.

100 YEARS AGOAn interesting mathematical study of the conditions which probably obtained in the primitive solar nebula has beencommunicated to the Academy of Science of St. Louis by Mr. Francis E. Nipher…According to the equations developed by theauthor, it seems impossible that at the timewhen the planets were separating from theparent mass the nebula was wholly gaseous.The idea that the planets were formed fromcondensing swarms of meteorites is the onlyreasonable one which conforms with thenumerical results obtained. It also appearsthat at the times when the moon separatedfrom the earth, and Mercury from the sun,the respective parent masses must have beenin the solid state, the sun having fused andbecome vaporised since the separation ofMercury. Further, it seems unnecessary, andeven improbable, that the earth should everhave been in a state of fusion. By substitutingthe proper conditions in one of his generalequations, Mr. Nipher finds that the isothermal7000° C. is probably the one existing at thesun’s surface at the present time.From Nature 9 June 1904.

50 YEARS AGOA variety of cells and tissues of mammalssurvive for long periods at the temperatureof ‘dry ice’, �79� C., when frozen in mediacontaining glycerol. At the other extreme,Andjus’s work on the whole animal showsthat, by special methods of cooling andrewarming, rats can be revived from deepbody temperatures of about �0.5� C….Hamsters, chosen because of their knownadaptability to body temperatures between2.5 and 38� C., were cooled by the methodrecently described by Andjus and Smith forrats… The animals stiffened progressivelyduring this process until they were wood-like to the touch, and presumably consistedof a hard frozen shell surrounding anunfrozen or only slightly frozen interior…The extremities intimately exposed to thebath fluid at �4� C. to �7� C. were severelyfrozen — the ears, for example, attained the consistency of cardboard, and may wellhave undergone crystallization of 80 per centof their water content. It is remarkable,however, that damage obviously due to this cause has been seen in only two of the twenty-one animals revived completely,some of which have been kept for manyweeks afterwards in apparently normalhealth. A. U. Smith, J. E. Lovelock, A. S. Parkes

From Nature 12 June 1954.

10.6 n&v 611 MH 4/6/04 5:29 pm Page 614

© 2004 Nature Publishing Group

news and views

NATURE | VOL 429 | 10 JUNE 2004 | www.nature.com/nature 615

of operation at the Tevatron (known as ‘Run II’), the experimental error on Mt willbe reduced to 2–3 GeV/c2; at the LargeHadron Collider8, currently under construc-tion at CERN,this accuracy will be improvedfurther to 1–2 GeV/c2.

The ultimate precision on Mt, however,will be achieved at a linear electron–positroncollider. Such a machine is currently in theplanning phase and could go into operationaround the middle of the next decade. Datafrom the linear collider could improve theaccuracy on the top-quark mass by about afactor of ten9–11.Only then will the uncertaintydue to the experimental error of the top-quark mass be well enough under control forthe information gleaned from the LHC in the next decade — on the Higgs boson (orbosons), supersymmetric partners or othernew physics — to be fully exploited. ■

Georg Weiglein is at the Institute for Particle Physics

Phenomenology, Department of Physics,University of Durham, Durham DH1 3LE, UK.e-mail: georg.weiglein@durham.ac.uk1. The DØ Collaboration Nature 429, 638–642 (2004).

2. The CDF Collaboration, the DØ Collaboration and the

Tevatron Electroweak Working Group. Preprint at

http://arxiv.org/abs/hep-ex/0404010 (2004).

3. Hagiwara, K. et al. Phys. Rev. D 66, 010001, 271–433 (2002).

4. The ALEPH, DELPHI, L3 and OPAL Collaborations and the

LEP Working Group for Higgs Boson Searches Phys. Lett. B

565, 61–75 (2003).

5. Heinemeyer, S., Hollik, W. & Weiglein, G. Comput. Phys.

Commun. 124, 76–89 (2000).

6. www.feynhiggs.de

7. Heinemeyer, S., Kraml, S., Porod, W. & Weiglein, G. J. High

Energy Phys. JHEP09(2003)075 (2003).

8. Beneke, M. et al. in Standard Model Physics (and More) at the

LHC (eds Altarelli, G. & Mangano, M.) 419–529 (CERN,

Geneva, 1999).

9. Heuer, R.-D. et al. Preprint at http://arxiv.org/abs/

hep-ph/0106315 (2001).

10.Abe, T. et al. Preprint at http://arxiv.org/abs/hep-ex/0106056

(2001).

11.ACFA Linear Collider Working Group. Preprint at

http://arxiv.org/abs/hep-ph/0109166 (2001).

On page 636 of this issue, David C.Knauth and colleagues1 claim thefirst detection of molecular nitrogen

(N2) in interstellar space. This simplediatomic molecule, made of one of themost abundant elements in the Universe, isthe most common constituent of Earth’smodern atmosphere. It is also a major com-ponent of the atmosphere of Saturn’s moonTitan, and has been detected in traceamounts in the atmospheres of Venus andMars. But it has proved surprisingly diffi-cult to find N2 in any environment beyondthe Solar System.

Chemical models of dark interstellarclouds (whose densities are usually in therange of 103 to 105 particles per cm3) suggestthat N2 should be the most abundant form ofnitrogen in these regions. This leads to theprediction2–4 that the ratio of N2 to hydrogenshould be about 10�5. In contrast, models for diffuse interstellar clouds, which aretransparent and have densities of about 102 particles per cm3, predict a much lowerN2 abundance, in the range between 10�9

and 10�8 that of hydrogen2,5.Both predictions suggest that N2 might be

observable, but searches for this molecule ininterstellar space had been fruitless untilnow. One of the difficulties in detectinginterstellar N2 arises from the fact that thesymmetric diatomic molecule has noallowed rotational or vibrational (dipole)transitions. Thus, N2 — unlike most of the

120 or more species now detected in darkinterstellar clouds — cannot be detectedeither through millimetre-wavelength obser-vations of rotational emission lines orthrough infrared spectroscopic detection ofvibrational bands (absorption or emission).

The only viable approach to findinginterstellar N2 is to search for the spectrallines created by electronic transitions in themolecule. These lines are found exclusively at far-ultraviolet wavelengths (shorter than100 nm), for which space-based telescopesare required because the Earth’s atmosphereblocks such radiation. For technical reasons,however, most ultraviolet telescopes havenot covered the far-ultraviolet spectralregion where the N2 bands lie. For example,the Hubble Space Telescope cuts off at about115 nm, well above the wavelength neededfor an N2 search. The Copernicus satellite —a small mission that was developed and ledby the late Lyman Spitzer and operated from1972 until 1980 — was the first orbitingspectroscopic observatory capable of far-ultraviolet searches for N2 in interstellarspace,but no detection was achieved6.

The best chance for astronomers tosearch for interstellar N2 has been affordedby the Far Ultraviolet Spectroscopic Explorer(FUSE) mission, now in its fifth year ofoperation. FUSE was designed specifically to extend ultraviolet spectroscopy to theshorter wavelengths that are not accessible to the Hubble Space Telescope, including

the spectral region where the electronicbands of N2 lie. Knauth et al.1 have takenadvantage of FUSE’s far-ultraviolet sensiti-vity to search for N2 — and apparently theyhave found it.

In a classic example of spectroscopicsleuth work, Knauth et al. have detectedabsorption by N2 in the line of sight towardsthe star HD 124314 by sorting through andeliminating other features that are blendedinto the spectrum. These other features arisethrough the absorption of radiation by thestar’s own atmosphere, by foreground inter-stellar gas (mostly molecular hydrogen) andby N2 in the outer vestiges of Earth’s atmos-phere. The detection of interstellar N2 wasaided by the fact that several individual N2

lines are accessible to FUSE and also becauseFUSE covers the N2 wavelength region withtwo separate detectors, which means thatinstrumental artefacts in the data can beeliminated.

The line of sight towards HD 124314 doesnot intersect a dark molecular cloud; rather,this is a long pathlength,probing one or morediffuse clouds. So, according to model calcu-lations, the ratio of N2 to hydrogen should becloser to the 10�9 to 10�8 level that is predictedfor diffuse clouds than to the 10�5 level pre-dicted for dense clouds. Knauth et al. havefound an intermediate value, with N2 repre-senting about 10�7 of the total hydrogenabundance in their observed line of sight.This abundance of N2 does not fit either thedense-cloud or diffuse-cloud models.

Among the possible explanations is thatthe line of sight towards this star containsone or more ‘translucent’ clouds, which arereckoned by astronomers to be intermediate(or possibly transitional) between dense anddiffuse clouds7. Alternatively, the models fordiffuse clouds might be incorrect, or thedetection claimed by Knauth et al. is wrong.The first and third of these options can prob-ably be eliminated, as the line-of-sight dustextinction to this particular star is too smallto include a translucent cloud, and theclaimed detection of N2 seems secure. So wemust surmise that the chemical models forN2 in diffuse clouds are inadequate.

Normally it is assumed that, with theexception of hydrogen, molecules in diffuseclouds form through gas-phase chemicalreactions2,5.But in dense clouds an additionalprocess, that of molecule formation on grainsurfaces, is probably important8,9.The detec-tion of N2 by Knauth et al.1 suggests thatgrain-surface reactions might contributemore to diffuse-cloud chemistry than previ-ously thought. This conclusion is consistentwith earlier searches in diffuse clouds forNH, another simple diatomic moleculefound to be more abundant than expectedfrom gas-phase chemistry alone10,11. Ifgrain-surface reactions are required toexplain the measured abundances of N2 andNH, it is possible that other surface reactions

Interstellar chemistry

Molecular nitrogen in spaceTheodore P. Snow

Astronomers have found evidence of molecular nitrogen in the cloudsof gas between the Earth and a distant star. The chemistry involved inthe formation of these diffuse clouds might need to be rethought.

10.6 n&v 611 MH 4/6/04 5:29 pm Page 615

© 2004 Nature Publishing Group