8/22/2019 Advances in Early Greek Astronomy
1/32
1One effective criticism of Dicks is found in Kahn 1970. I have tried to emphasize thescientific character of Presocratic philosophy in Graham 2006, but with a focus on methodologyrather than empirical success.
Advances in Early Greek Astronomy
Daniel W. Graham
Contemporary views of early Greek astronomy seem to stand largely where they were left by D.
R. Dicks in his impressive study, Early Greek Astronomy to Aristotle (1970). In summing up the
Presocratics, he observed,
it is difficult to resist the feeling that their ideas on the physical universe olent
lucerna, they smell of the lampthey are the dream children of the speculative
thinker in his study intoxicated by the novelty and daring of the new intellectual
atmosphere, and intent on applying the new methods of thought . . . in the widest
possible field. . . . [The Presocratics] were not, however, primarily scientists,
much less astronomers, and observation of actual celestial phenomena seems to
have played a relatively minor role in their thinking. (60, Dickss italics)
His general assessment was not particularly novel even when he wrote it, but, backed up by a
solid investigation of the methods and accomplishments of the Presocratic thinkers, it was, and
still is, an authoritative conclusion. So much so, that little has been done to reconsider the
contributions of the early philosophers to scientific astronomy.1 Indeed, the phrase early Greek
astronomy is often used to denote the theories of the fourth century B.C. (sometimes even the
8/22/2019 Advances in Early Greek Astronomy
2/32
Greek Astronomy 2
2E.g. Neugebauer 1972; Kalfas 1990. Goldstein and Bowen 1983 recognize an earlystage of Greek astronomy, but give it a diminished role. A recent history of astronomy that isquite good on modern developments is poor on early Greek contributions: Wilson 1997.
3E.g. Grant 2007; Evans 1998. Pedersen 1993 praises the Presocratics but cites noastronomical advances for them.
third century and later), as if the Greek thought of the sixth and fifth centuries contains no real
astronomy, but only, as scholars sometimes put it, speculative cosmology or natural philosophy.2
Dicks licensed another generation of scholars to patronize rather than study seriously the efforts
of Presocratic thinkers to understand the workings of the heavens.3
It will be my contention that this view is myopic and false. The Presocratic philosophers
made major contributions to astronomy that are so fundamental that we tend to take them for
granted and overlook them in a search for a different kind of contribution which they did not
make. I shall first consider the challenges for attributing a scientific understanding of the
Presocratics (I), then examine one area in which they made an important conceptual advance (II),
which led to a powerful explanatory model (III) that was widely accepted as a scientific advance
(IV).
I
The brief against the Presocratics as scientific students of the heavens is invariably the one
mentioned by Dicks. They are too speculative, and not sufficiently empirical. Dicks gives them
credit, as do virtually all historians of science, for developing a method that had the potential to
solve the problems of science: by abandoning mythological traditions and subjecting external
8/22/2019 Advances in Early Greek Astronomy
3/32
Greek Astronomy 3
phenomena to a process of logical reasoning, untrammelled by religious dogma, and even by
investigating the actual processes of thought, they opened up a whole new field of knowledge
which is virtually inexhaustible (ibid.). To put the point in the discourse of science, they were
good at advancing hypotheses and at discussing them critically in an open forum, but bad at
testing them against observational data.
Failures of observation are not, however, the only problem Dicks sees. In some sense,
surprisingly, their theories were not abstract enough. Their theories are all qualitative in
conception (ibid., Dickss italics). The theories lacked a mathematical basis that could provide
a truly scientific view of the heavens. Dicks seems to think that a close observation of risings,
settings, and the like would actually provide the mathematical background needed for a robustly
scientific conception of the heavens. Philolaus innovative scheme in which the sun, moon, and
planets circle a central fire is an example of bold thinking without sufficient empirical
grounding: it is very much the product of the study and bears little relation to the facts of actual
observation (e.g. it takes no account of the latitudinal and longitudinal variations of the planetary
bodies) (70). Like many other historians of astonomy, Dicks sees the beginnings of scientific
astronomy only in the measurements of Meton and Euctemon (ca. 430 B.C.), who made
observations of the solstices and equinoxes and the length of the year and the month, in order to
solve problems of constructing a practical calendar. Out of studies like those of Meton and
Euctemon, one can build a theory of the motions of the sun, moon, and stars, whereas the general
speculations of the Presocratics lead to a scientific dead end.
Beside this traditional challenge to Presocratic science, from the perspective of scientific
inquiry, there is a very different challenge from the outside. In recent years it has become
8/22/2019 Advances in Early Greek Astronomy
4/32
Greek Astronomy 4
increasingly common for thinkers to reject the claims of science to offer a reliable and
progressive path to knowledge. Some philosophers of science have argued that there is no
standard by which to judge one scientific theory against another, but only a very human
conception, which can be replaced from time to time. Cultural critics also reject the claims of
science to represent secure knowledge that is somehow better or more reliable than what people
can arrive at by ordinary processes of negotiation and consensus. If this is so, then the whole
project of the history of science, to track and illuminate the path from ignorance to knowledge,
and from random discoveries to reliable methods, is an illusion. The history of science is no
more than a certain kind of social history of a field that in the past has made inflated and
unjustifiable claims to truth and social utility.
This is not the place to engage in a large-scale debate about methodology and
historiography. Yet I cannot proceed with the present study without at least taking a stand on
these broad issues. I share the assumptions of traditional historians of science and wish to
criticize the status quaestionis from within that framework. My disagreement with current views
of early Greek astronomy is not with the method of traditional historians, but with a shortsighted
application of that method. It is more difficult to reply to revisionist philosophers of science
who call into question the very possibility of scientific progress. The most influential theorist in
the area is Thomas Kuhn (1996 [1962]), who allows for a kind of progress in solving problems
relative to a given paradigm in normal science, but sees no possibility of judging progress
between incommensurable approaches to science, which compete during revolutionary episodes.
Precisely when change is most radical, rational grounds fail for choosing between competing
paradigms. Other theorists such as Imre Lakatos (1970) and Larry Laudan (1977) have
8/22/2019 Advances in Early Greek Astronomy
5/32
Greek Astronomy 5
suggested ways in which a more robust notion of progress within a research program or tradition
is possible, yet they still fall short of endorsing a notion of absolute progress in understanding
the world or getting at the truth of the matter. In the words of Xenophanes, even if [a man]
should completely succeed in describing things as they come to pass, nonetheless he himself
does not know (B34).
I could frame the present study within the bounds of a kind of local scientific progress,
and that would offer some kind of interesting result. Yet the long-term advances made by
science suggest that this notion of progress-within-a-tradition is unnecessarily weak. Science
has progressed not merely in solving its own parochial problems (just as, for instance,
philosophy has), but also in understanding phenomena in a deep way that allows their
manipulation and control in many cases. As Hilary Putnam famously remarked, realism is the
only philosophy that doesnt make the success of [a] science a miracle (Putnam 1975, Putnams
italics)*. In fact, I hope to offer the present study as a modest contribution to the cogency of
scientific realism.
I shall presume, in accordance with scientific realism, that it is possible to make scientific
discoveries. I assume the following minimal (necessary) conditions for scientific discovery: (1)
Scientist S puts forth hypothesis H to explain phenomenon P at a time t; prior to t no one had an
adequate account of P (as seen, perhaps, by the fact that there is no generally accepted account of
P). (2) S or a community C of his peers adduces empirical evidence E on the basis of theory T
that confirms H as an account of P. (3) After t and as a result of a consideration of E in relation
to H, C accepts H as an account of P. Now I do not claim that (1)-(3) constitute sufficient
conditions for a scientific discovery. Yet they seem to present at least a partial description of
8/22/2019 Advances in Early Greek Astronomy
6/32
Greek Astronomy 6
4Cf. the HippocraticAncient Medicine 1: I assert that [medicine] has no need of a vainhypothesis as do remote and bewildering phenomena, such as those in the sky or under the earth,concerning which it is necessary, if one is to explain them, to use a hypothesis. But if [readinga(\ ei/)fora)ei/with Littr] one should explain or claim to understand what these things are, itwould not be clear either to the speaker or to the audience whether his statements were true ornot. For there is nothing against which to test ones knowledge.
what we mean by a scientific discovery, and hence provide criteria for deciding in historical
cases whether an episode could count as a scientific discovery.
I began this section with recent objections to Presocratic science. But similar objections
go back to ancient times. Xenophon says about Socrates, he wondered why it was not plain to
[cosmologists] that it was impossible for men to discover [heurein] these things, inasmuch as
even the most advanced theorists did not agree with each other (Memorabilia 1.1.13).4 This
objection seems to point to the lack of evidence (2) as seen from the failure of consensus (3). It
seems that from the time of Socrates to the present, Presocratic theories have been rejected as
insufficiently scientific.
With this background, let us turn to early Greek theories of the heavens. Can we find
anything that would count as an astronomical discovery?
II
One phenomenon or set of phenomena that was recognized as an explanandum by early Greek
philosophers was the character of the moon, in particular its phases. Sixth century theorists
presented several theories to explain the changing appearance of the moon, but none of them
could be justified on the basis of scientific evidence. Thus the question of the moons light could
8/22/2019 Advances in Early Greek Astronomy
7/32
Greek Astronomy 7
5For Babylonia, see Hunger and Pingree 1999. For the cultural setting of Babylonianastrology and astronomy, see Rochberg 2004. For Egypt, see Neugebauer 1957, 80-96;Maravelia 2006, 134-35, 228-29, 428-429. For India, see Billard 1971. For China, see Needham1959; Heliophotism seems to have been known by the fourth century BC (227); by 20 BC someChinese astronomers had a correct understanding of solar eclipses (414).
6allotrion phs borrowed light ~ allotrios phs foreign man Od. 18.219, with Coxon1986, 245.
count as a problem of philosophy already in the sixth century, a problem that had not been
solved. Furthermore, a survey of astronomical lore of Babylon, Egypt, India, and China will
show that no culture had a scientific understanding of lunar light in this period, although
Babylonian astrologers, at least, had amassed extensive and detailed observations of the sun,
moon, and planets.5 At this time Greek philosophers seem to be unique in offering speculative
theories of heavenly phenomena often based on physical models of the world and its contents.
Our story begins with two fragments of Parmenides of Elea, found in the cosmological
part of his philosophical poem:
B14. [moon] shining by night, wandering around earth with borrowed light,
B15. ever peeking toward the rays of the sun.
Here we find a striking insight: the moon shines with borrowed light, as evidenced by the fact
that its shiny side is always facing the sun. There is no argument here. The first line presents a
clever word-play parodying a Homeric line.6 The second line recognizes an important
relationship, but does not elaborate. It is possible that Parmenides offered further considerations,
but we have no record of a scientific explanation, nor does the poetic setting lend itself to
8/22/2019 Advances in Early Greek Astronomy
8/32
Greek Astronomy 8
7Empedocles B45 with B43, B47, B48, B42, PlutarchDe fac. orb. lun. 929d-e.
8The learned essayist Plutarch also connects Parmenides B15 with Anaxagoras B18, Thesun imparts to the moon its brightness (De fac. orb. lun. 929a-b).
expositions such as one might expect from a prose treatise. Indeed, one might question whether
we are reading too much in to Parmenides remarks. Nevertheless, Empedocles virtually quotes
Parmenides line, and understands it as embodying a theoretical claim, one that he assents to and
elaborates on.7 If a successor, who had access to the whole poem and partook of its milieu, took
it as an explanatory hypothesis, we seem to be justified in reading it similarly. 8
The simple insight is that the moon gets its light from the sun, as seen by the fact that its
luminous face is always looking directly towards the sun. The insight gets the facts right, as
seen from hindsight, and marks the first explicit recognition of this relationship in the history of
the world. As presented, however, it is little more than an insight. It is embedded in no theory,
supported by no argument, justified by no further evidence than that stated in B15. Thus, by our
criteria, the insight does not constitute a discovery: there is something we can call a hypothesis
to account for a phenomenon, but no argument or evidence produced beyond what might support
an initial connection.
III
Of what happened next we have no detailed information. But consider what a follower of
Parmenides, who took his proposals seriously, as we know Empedocles did, might glean from
Parmenides insight. If the light from the moon is reflected from the sun, then the phases of the
8/22/2019 Advances in Early Greek Astronomy
9/32
Greek Astronomy 9
9DC291b17-23. Cf. Popper 1998, 133 n. 63.
10Cf. PlutarchDe fac. orb. lun. 929b-d.
moon are plays of light and shadow on the face of a body. The patterns presented in the phases
of the moon: thin crescent, thicker crescent, half-circle, gibbous shape, can be produced only by
a spherical or spheroid body, as Aristotle explicitly noted.9 Hence we can infer that the moon is
a sphere, or approximately so.
Furthermore, the moon must be opaque. Otherwise it would never appear dark, but
would always glow as a translucent body partially illuminated by the penetration of the suns
light, even on the parts not facing the sun. Since part of the moons surface is dark part of the
time, the body must block the suns rays where it does not reflect them.10
Furthermore, the moon must be closer to the earth than the sun, for otherwise it would
never be completely dark, as it is during the time of the new moon. For if the moon were farther
from the earth than the sun, then the sun should illuminate it when it is approximately in
conjunction with the sun, as its luminous face would be oriented toward the earth. As it is, the
moon is full only when it is in opposition to the sun, and dark when it is in conjunction.
These three features allow one to construct a model of the heavens in which the moon
moves around the earth relative to the sun. The interplay of position and illumination provide a
basis for anticipating the appearances of the moon. Indeed, the model provides a theory for
predicting in detail the phases of the moon. At the conjunction of the moon and sun, the moon
will be invisible. At a few degrees distant from the sun (at the latitude of Athens, for instance),
the moon will appear as a thin crescent, above and to the left (east) of the setting sun. At about
45 distant the moon will appear as a crescent occupying about one-fourth of the lunar disk. At
8/22/2019 Advances in Early Greek Astronomy
10/32
Greek Astronomy 10
11For the term, Mourelatos 2002, with reflections on Xenophanes influence; for the roleof Parmenides, see Whrle 1995; Graham 2002. Cf. O'Brien 1968. Popper 1998, 70-73, 84-88,96, 100, 136, sees Heliophotism as a great discovery of Parmenides, one which inspired both hisontology and his cosmology.
90 it will appear as a half moon. At 135 it will appear as a convex shape covering about three-
fourths of the disk. And at 180 a full moon will appear. After the full moon, the phases will
repeat in mirror image, with the shiny face directed towards the rising sun in the east rather than
the setting sun in the west.
Thus the model provides for detailed predictions that correlate the apparent shape of the
moon with its position relative to the sun, assuming only the features of the model and that the
moon is traveling westward more slowly than the sun. Nothing about these predictions is
striking to us now. Yet prior to Parmenides, none of the several theorists of lunar light and lunar
phases so much as noticed a correlation between the phases and the position of the moon relative
to the sun. Once we formulate the model, we have precisely correlated the position of the moon
with that of the sun. We can observe the progression of the phases night by night for lunar
month, and see all the predictions borne out faithfully. We can watch for another month and
observe them repeating day by day all over again. The model makes predictions which are
invariably confirmed. It reduces the question of lunar phases to a problem of geometry, and
solves it.
We have then an initial insight about the source of the moon light: the sun. This
explanation, which has been called heliophotism, seems to originate with Parmenides.11 It
gives rise to a model, which supports a general theory of lunar phases, which in turn can be
verified empirically. As I shall demonstrate later, this theory is espoused by both Anaxagoras
8/22/2019 Advances in Early Greek Astronomy
11/32
Greek Astronomy 11
12See Lakatos 1970.
and Empedocles, two near contemporary cosmologists who were a generation younger than
Parmenides. One of them, we have seen, tacitly acknowledges the influence of Parmenides, and
both of them clearly respond to philosophical doctrines of Parmenides.
Is this then a solid scientific advance? Let us take the story one step farther. The theory
of Heliophotism suggests explanations of other heavenly phenomena apparently unconnected
with lunar phases. If we can show a link between heliophotism and other explanations, we will
have a clear case of what Lakatos called a theoretically progressive problemshift; if it can show
how the theory was supported by new evidence, we will have an empirically progressive
problemshiftand perhaps more than that.12
The model of lunar light and its associated theory have implications for another problem
of Presocratic cosmology. What causes lunar and solar eclipses? The case of the lunar eclipse is
perhaps the easiest to address within Heliophotism. The source of the moons light is the sun. If
the moon ceases to shine, it must be because in some way or other the suns light is failing to
illuminate it. The model itself suggests one time when the suns light might fail to reach the
moon. At the time of the full moon the sun and moon are in opposition, which means that they
are on opposite sides of the earth. The earth, in most early theories, is relatively large. It is,
then, a feature of the model that the earth mightintervene between the sun and moon, causing the
suns light to fail and the moon to go dark. In effect, the moon would be passing into the shadow
of the earth. Here a potential explanation grows out of a model and its associated theory.
The model suggests another account for solar eclipses. Whereas the moon gets its light
from the sun and so is subject to being blocked by an intervening body somewhere between the
8/22/2019 Advances in Early Greek Astronomy
12/32
Greek Astronomy 12
sun and the moon, the sun generates its own light. Hence only a quenching of its light or a direct
interposition between the sun and the earth could explain a solar eclipse. But here the model
again suggests a possibility. According to the model the moon remains as a dark body even
when it disappears from view at the time of the new moon. Indeed, the theory dictates that
whenever the moon disappears, it is near the sun, in conjunction with it. We can observe at
times when both bodies are present in the sky, that they are approximately the same size,
subtending about of the ecliptic. We also know that the moon is below the sun, closer to
earth than it. Hence during every new moon, there is a dark body in the vicinity of the sun that
could, under the right conditions, interpose itself between the earth and sun. If this happened,
the moon would occult the sun.
I shall call the complex theory that accounts for both the darkening of the moon and that
of the sun by an interposition or antiphraxis (a term of art used by Aristotle and later
philosophers). Now in fact both Anaxagoras and Empedocles put forth the Antiphraxis theory.
They were (it will be argued later) the first cosmologists to do so. It is instructive to consider
Anaxagoras general theory of the heavens, as reported by Hippolytus.
(6) The sun and moon and all the heavenly bodies are fiery stones carried around
by the revolution of the aether. . . . The moon is below the sun and nearer to us.
(8) The sun exceeds the Peloponnesus in size. [H] The moon does not have its
own light, but gets it from the sun. The revolution of the stars carries them under
the earth. (9) [A] The moon is eclipsed when the earth blocks it, or sometimes
one of the bodies below the moon; the sun is eclipsed when the moon blocks it at
8/22/2019 Advances in Early Greek Astronomy
13/32
Greek Astronomy 13
13Hippolytus testimony is supported by Anaxagoras B18 (see n. 8* above).
14This claim was controversial in antiquity, as we shall see later. PlutarchNicias 23.2-3 =A18 attributes the first theory of Heliophotism to Anaxagoras; Plato Crat. 4091-b = A76attributes Heliophotism to him, but suggests an earlier discovery of the theorybased on playfuletymology. Dicks 1970, 59, accepts Anaxagoras as the first to give the true explanations of themoons phases . . . , and allows that Empedocles (54) and Anaxagoras (59) understandeclipseswithout managing to notice how these admissions undermine his general assessment ofPresocratic astronomy. Thoren 1971, 24, observes more generously, If there is anything thatcan truly be called a datum in pre-Socratic astronomy, it must surely be that Anaxagorasprovided, if not for the first time, then at least to the first general satisfaction, an explanation ofthe phases of the moon. Closely related to this tradition is the report that he also explained thecause of eclipses. See also Heath 1913, 78-80; O'Brien 1968; Curd 2007, 211-12.
the time of the new moon. The sun and the moon make their turnings when they
are deflected by the air. The moon makes frequent turnings because it cannot
overcome the cold. (10) He first correctly explained eclipses and illuminations.
He said the moon was earthy and had in it plains, and valleys.
(HippolytusRefutation of All Heresies 1.8.6-10 = DK 59A42)
The sentence marked [H] asserts Heliophotism.13 The heavenly bodies are stones, and hence
solid, three-dimensional bodies. The moon is earthy, and hence opaque. It lies below the sun
and nearer the earth. Anaxagoras gives, almost as a corollary, the correct account of both lunar
and solar eclipses, namely Antiphraxis [A], the first to do so according to 10 of the report. 14
We see from this report that Anaxagoras acknowledges all features of the model constructed
above, and treats the theory of eclipses as an extension of the model.
From Hippolytus and presumably his doxographical sources we get no indication of how
Anaxagoras came to his astronomical views or how they were connected in his mind. It seems
plausible, however, to see Heliophotism as the insight that inspires the whole theory. The model
8/22/2019 Advances in Early Greek Astronomy
14/32
Greek Astronomy 14
of lunar light provides the bridge from Parmenides original insight to an explanation of eclipses.
We find the same general account in Anaxagoras contemporary Empedocles, though not so
neatly juxtaposed as here. It does not matter for present purposes which of the two cosmologists
originally connected all the dots in the theory, though I believe it is Anaxagoras for reasons I
will allude to shortly. In any case, the model suggested by Heliophotism provides the basis for a
theory that explains not only the phases of the moon but both lunar and solar eclipses, and does
so correctly, as determined by subsequent research. There is then, a progressive articulation of
astronomical theory from Parmenides to Anaxagoras and Empedocles which can plausibly
understood as an instance of theoretical progress.
What of empirical evidence? This, according to critics of Presocratic theory, both
ancient and modern, is one of the weak links in the Presocratic program. The early cosmologists
are ingenious in their theorizing but quite unable to adduce empirical evidence to decide between
one theory and another. Yet as we have seen, Heliophotism suggests a model that makes precise
predictions for phases of the moon, which are confirmed infallibly. Furthermore, the model
makes predictions about lunar and solar eclipses. A lunar eclipse caused by the earths shadow
will happen only at the time of the full moon. Anaxagoras allows for lunar eclipses to be caused
also by other dark bodies. But as we now know, the earth causes all lunar eclipses.
Furthermore, the moon enters into the shadow of the earth from the right (west) and exits to the
left (east), just as the model would require, even with a moving sun and stationary earth.
A solar eclipse will, according to the model, occur only at the time of the new moon. A
circular disk must be seen to pass over the sun as the sun overtakes the moon and passes behind
it in its westward course. Today we would describe the moon as passing in front of a stationary
8/22/2019 Advances in Early Greek Astronomy
15/32
Greek Astronomy 15
15 Graham and Hintz 2007.
sun as the moon travels eastward in its orbit, but the anticipated effect is the same. Every solar
eclipse conforms to the model. In fact there were two solar eclipses visible in the region of
Greece during the period of activity of Anaxagoras and Empedocles. One in 478 BC passed over
the Peloponnesus and Attica and proceeded across the Aegean to Ionia. One in 463 passed over
central and northern Greece. Neither would have been visible (except as a partial eclipse) in
Acragas, Empedocles homeland in Sicily. The former could have been visible to Anaxagoras
whether he was his native Clazomenae in Ionia or in Athens (according to different accounts of
his life). Elsewhere I have argued that the otherwise inexplicable remark in 8 of the Hippolytus
passage quoted above (the sun is larger than the Peloponnesus) only makes sense if Anaxagoras
inquired into the extent of the umbra of the earlier eclipse, and the information from the
Peloponnesus would have been available to him only if he were in Athens at the time. 15 The
timing and location suggest that Anaxagoras was the first to understand the solar eclipse, and
that he also tested it empirically by establishing the extent of the shadow cast (assuming,
incorrectly but plausibly, that the shadow was about the size of the body that cast it).
If this is so, then Anaxagoras not only first hypothesized that solar eclipses are
occultations of the sun by the moon, but also tested his hypothesis empirically when an
opportunity presented itself. Thus (1) Anaxagoras offered an explanatory hypothesis for a
mysterious phenomenon, one that had been first offered by Parmenides but not defended by him.
(2) He adduced evidence for the hypothesis, which confirmed his hypothesis. That is, he was
able to confirm that the eclipse of 478 took place at the time of the new moon. A circular disk
was seen blocking the suns light (this was an annular eclipse in which a thin ring of fire was
8/22/2019 Advances in Early Greek Astronomy
16/32
Greek Astronomy 16
visible at the circumference of the sun). Finally, he seems to have been able to confirm that the
maximum eclipse was visible in a limited area about the size of the Peloponnesus. On no earlier
hypothesis was the eclipse a shadow with a limited radius, so the test would have effectively
eliminated all earlier hypotheses. Anaxagoras could have found out the extent of the eclipse by a
basic procedure consistent with the simplest form ofhistoriasking travelers at the port of
Piraeus what they had seen recently. What was important in this case was not a highpowered
scientific methodology, but an effective way of checking an otherwise meaningless piece of
information: was the eclipse seen over a limited area? The theory made the question relevant,
and the answer provided an important piece of empirical evidence in favor of a bold new
hypothesis.
IV
One large question remains: did Anaxagoras manage to convince any of his contemporaries and
successors? Did he achieve the kind of discovery that won over his peers? One of the major
criticisms of Presocratic cosmology, even in ancient times, was that its theories were
unverifiable. The heavenly bodies are remote and the true causes of their motions and actions
are thus inscrutable. One piece of evidence of the failure of early cosmological theories was the
very proliferation of such theories, without any apparent way to eliminate rivals and achieve
anything approaching consensus. In modern terms, the theories were, given the limited
empirical means available, not only unverifiable but (what is worse) unfalsifiable.
Here I propose to turn to a late account of cosmological theories, the doxographical
8/22/2019 Advances in Early Greek Astronomy
17/32
Greek Astronomy 17
tradition preserved in Atius. The very scheme embodied in doxography is adversarial: alternate
theories of the same phenomena are presented, often in contrasting groupings. Doxography
feeds on disagreement and often preserves such disagreement as a kind of fossil record, dissent
frozen in time. Three chapters from Book II of Atius Placita deal with topics of direct
relevance to the present inquiry: 28 On the Illumination of the Moon, 29 On the Lunar Eclipse,
and 24 On the Solar Eclipse. These chapters allow us to glimpse competing opinions and to
infer what sort of relation the several accounts had to each other.
We can survey opinions about the moons light in Atius:
Ch. 28. On the light of the moon
1. Anaximander, Xenophanes, Berosus [say] it has its own light (idion . . .
phs).
2. Aristotle says it has its own light, but it is somehow rather tenuous.
. . .
4. Antiphon [says] the moon has its own light, and that the light that is
hidden around it is darkened by the impact of the sun, as the stronger
naturally darkens the weaker; this happens to the other heavenly bodies
also.
5. Thales first said it is illuminated by the sun. Pythagoras, Parmenides,
Empedocles, Anaxagoras, Metrodorus likewise.
6. Heraclitus says the sun and moon manifest similar effects. For since the
heavenly bodies are shaped like bowls, when they receive rays from the
8/22/2019 Advances in Early Greek Astronomy
18/32
Greek Astronomy 18
moist evaporation, they shine to our view, the sun more brightly, for it
travels in purer air, while the moon travels in more polluted air, and for
this reason it appears dimmer.
As is typical in doxographical collections, the list is not laid out chronologically but according to
some dialectical scheme. The impression we are given is that there were two parallel traditions:
Anaximander, Xenophanes, Heraclitus, and Antiphon held that the moon generates its own light,
while most of the other early philosophers held that it is illuminated by the sun. It so happens
that the moon is illuminated by the sun, so some Presocratics got the explanation right. They
also got it right for the first time, since no other culture we know of ever came to this conclusion
independently. But was it a lucky guess, or a scientific discovery? Of course a scientific
discovery can begin with a lucky guess, but to count as scientific it must be confirmed in some
systematic way.
For the purposes of this inquiry, there are two contrasting opinions. Either the moon gets
its light from the sun or it generates its own light. The first theory is Heliophotism; the second I
shall callIdiophotism. Let us put our protagonists in historical order, as far as is possible. Then,
according to Atius, and abbreviating the two theories by their initials:
Philosopher Theory
Thales H
Anaximander I
Pythagoras H
8/22/2019 Advances in Early Greek Astronomy
19/32
Greek Astronomy 19
16For a more extended argument on this point, see Graham 2002, 352-58. See also Dicks1959; Stephenson 1997, 342-44. For Thales as originator of the theory, see Panchenko 1993;Panchenko 1994; Panchenko 1996. O'Grady 2002, 126-46, thinks Thales may have made somesort of prediction based on repeated cycles, but also thinks he may have understood the nature ofeclipses, based on a new fragment from Aristarchus in POxy. LIII. 3710. ii (c).36-43 (seeLebedev 1990). This may, however, reflect no more than a reconstruction by Aristarchus. Notethat one can make predictions on the basis of repeating patterns of eclipses with nounderstanding whatsoever of the cause of the events.
Xenophanes I
Heraclitus I
Parmenides H
Anaxagoras H
Empedocles H
Antiphon I
Metrodorus H
Aristotle I
No pattern emerges from this survey.
There are, however, good historical reasons for questioning several of Atius
attributions. In the case of Thales we actually have no reliable astronomical theory to go on. It
seems likely that he left no written record of his theories. We get a reference to the earth as like
a raft floating on a primeval seaa view that is probably incompatible with a correct
understanding of lunar light. One report ascribes earthy heavenly bodies to himbut on what
authority we do not know. Aristotle himself seems to have had no written record of Thales
views to consult.16 Although I cannot go into detail here, it seems likely that the legend of
8/22/2019 Advances in Early Greek Astronomy
20/32
Greek Astronomy 20
17 Burkert 1972, 299-350; Huffman 1993, 231-88.
18 Pendrick 2002, 296-97, claims that the impact of the suns light is meant to explainwhy the moon appears dim in daytime; but the suns light does not just fall on the moon duringthe daytime, nor is it plausible that Antiphon could have thought so in the late fifth century.More likely the matter concerns the secondary or residual light seen during a lunar eclipse.Anaxagoras seems to have had a theory about this earlier. See OlympiodorusMeteorology67.32-37, Plato Cratylus 409a-b, PlutarchDe fac. orb. lun. 933f, with Boll 1909, 2343-44;O'Brien 1968; Panchenko 2002. Grgemanns 1970, 36 n. 68, thinks this theory may originatewith Atius; but note the Plato passage.
19SeeA Po. 90a15-18, 93a29-b7;DC291b17-23, 292a-6, 297b23-30;Mete. 367b19-30;GA 777b24-26.
Thales prediction of a solar eclipse led to the uncritical attribution to him of a complete theory
of lunar light and solar eclipses, by astronomers writing after Aristotle.
As for Pythagoras, he left no writings and his society seems to have maintained a code of
silence for the sixth and early fifth centuries B.C. Recent scholarship has shown that much of
the cosmology and astronomy attributed to Pythagoras or the Pythagoreans originates from
Philolaus, who was a Pythagorean, but was active in the second half of the fifth century. 17 He
wrote a treatise containing cosmological and astronomical speculations. If Pythagorean views
originate with him, then they do not originate with Pythagoras, and they fall after many of the
other figures in the chronological list. In the case of Antiphon, a closer reading of the evidence
shows that he thinks the moon has some secondary light of its own, which is obscured by the
primary light of the sun falling on it; in other words, what we usually see is a reflection of the
suns light.18 As for Aristotle, Atius is just wrong, unless he refers to some missing account of
secondary light.19 So let us reexamine the list:
Philosopher Theory
8/22/2019 Advances in Early Greek Astronomy
21/32
Greek Astronomy 21
Thales ?
Anaximander I
Pythagoras ?
Xenophanes I
Heraclitus I
Parmenides H
Anaxagoras H
Empedocles H
Philolaus H
Antiphon H
Metrodorus H
Aristotle H
A different picture emerges. As far as we can see, Idiophotism does not alternate with
Heliophotism; first comes Idiophotism, then Heliophotism. The turning point is Parmenides, and
after him there is no going back.
We get another survey of opinions on the subjects of solar and lunar eclipses from Atius
(II.24 and 29, respectively):
Ch. 24. On the solar eclipse
1. Thales first said the sun is eclipsed when the moon moves directly under
it; and the moon is seen to be earthy . . .
8/22/2019 Advances in Early Greek Astronomy
22/32
Greek Astronomy 22
20See n. 16* above.
2. Anaximander [says it happens when] the mouth of the breathing-hole of
fire is covered.
3. Heraclitus, with the turning of the bowl, so the concave part is turned up,
the convex part down towards our sight.
4. Xenophanes, with quenching; another sun in turn appears toward the east .
. .
5. Some [say] a condensation of clouds coming over the disk [of the sun].
6. [The Pythagoreans say] when the moon comes under [the sun].
7. [Empedocles,] when the moon comes under [the sun].
Here we see that Thales, certain Pythagoreans, and Empedocles give the correct account of solar
eclipses. But we must remove Thales from the list for reasons similar to those given for the
theory of lunar light.20 As for the Pythagoreans, some of them wrote after Empedocles, so we
cannot on the vague reference given, attribute a breakthrough to them. The anonymous theory in
5 has been plausibly referred to Anaximenes, though the attribution is not certain.
For the lunar eclipse we get the following reports:
Ch. 29. On the lunar eclipse
1. Anaximander [says a lunar eclipse occurs] when the opening on the wheel
[of the moon] is blocked.
. . .
8/22/2019 Advances in Early Greek Astronomy
23/32
Greek Astronomy 23
3. Alcmaeon, Heraclitus, Antiphon with the turning of the bowl-shaped
container and its inclinations.
4. Some of the Pythagoreans according to Aristotles history and the
statement of Philip of Opus [say it occurs] sometimes by the blocking [of
the suns light] by earth, sometimes by a blocking by the counter-earth . . .
5. Xenophanes [says it and?] the monthly disappearance [occur] by
quenching.
6. Thales, Anaxagoras, Plato, the Stoics, in agreement with the mathematical
[astronomers, say] the moons conjunction with the sun and the fact that it
is illuminated by the sun cause the monthly disappearances of the moon,
and its moving into the shadow of the earth cause its eclipses (as the earth
lies between both bodies) . . .
7. Anaxagoras, as Theophrastus says, [holds that] sometimes bodies below
the moon cause its occlusion.
8. Chrysippus [says that] the moon is eclipsed when the earth causes its
occlusion and it enters into the earths shadow.
Those who understand lunar eclipses as the blocking of the suns light by the moon are Thales,
certain Pythagoreans, Anaxagoras, Plato, and the Stoics, including specifically Chrysippus. The
attribution of the correct eclipse theory to Thales is dubious for reasons similar to those for
denying to him a theory of heliophotism. As far as the Pythagoreans, the counter-earth seems to
be a unique feature of Philolaus cosmography, and Philolaus wrote one or two generations after
8/22/2019 Advances in Early Greek Astronomy
24/32
Greek Astronomy 24
21Atius 2.7.7, 3.11.3, 3.13.2, Huffman 1993, 239.
22See n. 18* above.
Anaxagoras.21 Antiphon the Sophist is grouped with the Heraclitean theory; yet his account of
lunar light as reported in chapter 28 is incompatible with Heraclitus theory, so I omit the alleged
theory of Antiphon from chapter 29 as highly problematic.22 As for Plato, we have his complete
dialogues, which advance no theory of eclipses.
Thus we get the following scheme, with A for a theory based on Antiphraxis, O for
other:
Solar Eclipses Lunar Eclipses
Philosopher
Thales ? ?
Anaximander O O
Anaximenes? O
Xenophanes O O
Alcmaeon O
Heraclitus O O
Anaxagoras A
Empedocles A
Philolaus A
Chrysippus, Stoics A
Mathematical astronomers A
8/22/2019 Advances in Early Greek Astronomy
25/32
Greek Astronomy 25
23 Boll 1909, 2343-44; Bicknell 1967.
24 Toulmin 1967 views the theory as problematic and primitive. For other reports ofBerosus theory, see Lucretius 5.720-28, Vitruvius 9.2.1-2.
With both lunar and solar eclipses, we observe the introduction of a new theory, and then a shift
to that theory by all successors.
The reports of Atius assign Empedocles to solar eclipses and Anaxagoras to lunar
eclipses. They also point out that Anaxagoras admitted other bodies that could cause lunar
eclipses by blocking the suns light. I shall call these by their modern name, asteroids. There is
an reasonable explanation for his extension of the theory to asteroids, which I shall not pursue
here.23 Other ancient testimonies assure us that Anaxagoras understood both solar and lunar
eclipses and we have a fragment of Empedocles to demonstrate that he understood lunar eclipses.
There is no record of Parmenides having had anything to say about eclipses. Thus a corrected
reading of Atius, with figures put in chronological order, indicates the appearance of
Heliophotism with Parmenides and the appearance of Antiphraxis with Anaxagoras and
Empedocles. As we have seen Parmenides account of lunar light and Anaxagoras and
Empedocles theory of eclipses are closely related, as parent and offspring. Further, the present
exercise show that once these related theories were introduced, no other competing theories were
proposed.
There is one exception: Berosus offered an alternative account of the moons light
(Atius 2.28.2). But Berosus was a hellenized Babylonian priest writing in the late fourth
century BC. His theory is heavily influenced by the now-standard theory of Heliophotism and
seems like just a rather ad hoc variation of it, by (dare we say?) a barbarian.24
The Epicureans
8/22/2019 Advances in Early Greek Astronomy
26/32
Greek Astronomy 26
25Lucretius 5.705-14, 751-70.
26See note 19* above.
27Vitruvius 9.2.3-4.
28AlmagestVI.
withheld assent from Heliophotism and Antiphraxis, in accordance with their policy of allow
plural explanations of obscure phenomena, yet even they given pride of place to the two
Presocratic theories.25 Aristotle embraced both Heliophotism and Antiphraxis (pace Atius), and
indeed used the later as a paradigm of successful scientific theory.26 The influence of his
cosmology on posterity was immense. Significantly, the mathematical astronomers also
embraced the new theory and built it into their mathematical models of the heavens. As an
example, we have the account of Aristarchus, who (even in his heliocentic theory) attributes the
moons light to a reflection of solar light.27
After Atius time Claudius Ptolemy would use the
theory as the basis of his study of eclipses.28
Of course what we get from the doxographical tradition is a mere abridgement of
elaborate theories and complex arguments that renders them into their lowest common
denominators. Yet from what we can see, new theories of lunar light and of eclipses were just
not offered after the early fifth century BC. They seem to have become superfluous because the
phenomena they addressed ceased to be problematic. Whereas other astronomical and
cosmological phenomena continued to be hotly debated, lunar light and eclipses ceased to be
mysterious.
What we seem to have, then, in Helophotism and Antiphraxis is two instances of
immediate and long-lasting consensus. That most elusive feature of scientific progress is found
8/22/2019 Advances in Early Greek Astronomy
27/32
Greek Astronomy 27
at the heart of Greek cosmology and at the beginnings of Greek astronomy. Anaxagoras
systematized Parmenides insight about lunar light, applied it to the problem of eclipses (and
other phenomena), provided empirical confirmation of his predictions, and convinced peers
(starting with Empedocles) that he had proved his point. Indeed, by Aristotles time his theory
had come to occupy the position of a Kuhnian paradigm, and by Chrysippus time it was perhaps
de rigeurin astronomy. Thus (1) Anaxagoras advanced a novel hypothesis, (2) adduced
evidence for it, and (3) proved it to the satisfaction of his peers and successors. The grand idea
that started the process came from Parmenides, but, as Hippolytus reported and Plutarch
explained, the systematic development and proof of the idea came from Anaxagoras.
More could be said about the reception of this theory, including how it got projected back
to Thales and Pythagoras. But perhaps enough has been said for nowexcept for one last point
that may seem too obvious to make: the consensus that began in the fifth century BC has
continued to the present day. The model of lunar light developed by Anaxagoras is that still
found in astronomy textbooks and manuals, as is his theory of eclipses. To be sure, we
understand now that the earth is spherical rather than flat, the earth orbits the sun rather than vice
versa, and the sizes and distances of the heavenly bodies are different from Anaxagoras. Yet for
all that, he got the source of lunar light and the cause of solar and lunar eclipses just right. He
made a major scientific discovery, one that paved the way for more discoveries, and one that was
so robust that it survived the transition to theories of concentric spheres, deferents and epicycles,
equants and eccentrics, and ultimately the Copernican Revolution itself. In view of Anaxagoras
accomplishment it seems perverse to hold that no Presocratic philosopher ever got beyond airy
speculations or made scientific progress. Scientific astronomy began, if not in the sixth century
8/22/2019 Advances in Early Greek Astronomy
28/32
Greek Astronomy 28
as some hold, then surely in the early fifth century BC. If it began with mere speculation, it
quickly transcended its beginnings to offer the first paradigms of successful problem-solving and
models for future astronomical inquiry.
There is one further lesson suggested by this story. In the absence of meaningful dissent
from Heliophotism and Antiphraxis over a period of 2,500 years, it appears that they have ceased
to be problematic theories and morphed into scientifically indubitable descriptions of the world.
One of many competing hypotheses spawned by Presocratic philosophyone of the dream
children of a speculative thinkerhas passed all its tests and settled down into the textbooks
and handbooks of the twenty-first century. A scientific realist could not ask for a happier ending
to the story.
8/22/2019 Advances in Early Greek Astronomy
29/32
Greek Astronomy 29
REFERENCES
Bicknell, Peter J. 1967. "Lunar Eclipses and 'Selenites.'"Apeiron, 15: 16-21.
Billard, Roger. 1971.L'astronomie indienne: Investigation des textes sanskrits et des donnes
numriques. Paris: cole Francaise d'Extrme-Orient.
Boll, F. 1909. "Finsternisse." In Paulys Realencyclopdie der classischen Altertumswissenschaft,
edited by Georg Wissowa, vol. 6, cols. 2329-64. Stuttgart: J. B. Metzler.
Burkert, Walter. 1972[1962].Lore and Science in Ancient Pythagoreanism. Translated by E. L.
Minar Jr. Cambridge, Mass.: Harvard University Press.
Coxon, A. H. 1986. The Fragments of Parmenides. Assen: Van Gorcum.
Curd, Patricia. 2007.Anaxagoras of Clazomenae: Fragments and Testimonia. Toronto:
University of Toronto Press.
Dicks, D. R. 1959. "Thales." Classical Quarterly, N.S. 9: 294-309.
. 1970. Early Greek Astronomy to Aristotle. Ithaca: Cornell University Press.
Evans, James. 1998. The History and Practice of Ancient Astronomy. New York: Oxford
University Press.
Goldstein, Bernard R., and Alan C. Bowen. 1983. "A New View of Early Greek Astronomy."
Isis, 74: 330-40.
Grgemanns, Herwig. 1970. Untersuchungen zu Plutarchs Dialog De facie in orbe lunae.
Heidelberg: Carl Winter.
Graham, Daniel W. 2002. "La lumire de la lune dans la pense grecque archaque." In Qu'est-ce
que la philosophie prsocratique? edited by Andr Laks and Claire Louguet, 351-80.
8/22/2019 Advances in Early Greek Astronomy
30/32
Greek Astronomy 30
Villeneuve d'Ascq: Presses Universitaires du Septentrion.
. 2006. Explaining the Cosmos: The Ionian Tradition of Scientific Philosophy . Princeton:
Princeton University Press.
Graham, Daniel W., and Eric Hintz. 2007. "Anaxagoras and the Solar Eclipse of 478 BC."
Apeiron, 40: 319-44.
Grant, Edward. 2007.A History of Natural Philosophy: From the Ancient World to the
Nineteenth Century. Cambridge: Cambridge University Press.
Heath, Thomas. 1913.Aristarchus of Samos: The Greek Copernicus. Oxford: Clarendon Press.
Huffman, Carl A. 1993. Philolaus of Croton. Cambridge: Cambridge University Press.
Hunger, Hermann, and David Pingree. 1999.Astral Sciences in Mesopotamia. Leiden: Brill.
Kahn, Charles H. 1970. "On Early Greek Astronomy."Journal of Hellenic Studies, 90: 99-116.
Kalfas, Vassilis. 1990. "Criteria Concerning the Birth of a New Science: The Case of Greek
Astronomy." In Greek Studies in the Philosophy and History of Science, edited by
Pantelis Nicolacopoulos. Boston Studies in the Philosophy of Science, vol. 121, 171-85.
Dordrecht: Kluwer.
Kuhn, Thomas S. 1996[1962]. The Structure of Scientific Revolutions, 3rd edition. Chicago:
University of Chicago Press.
Lakatos, Imre. 1970. "Falsification and the Methodology of Scientific Research Programmes." In
Criticism and the Growth of Knowledge, edited by I. Lakatos and A. E. Musgrave,
91-196. Cambridge: Cambridge University Press.
Laudan, Larry. 1977. Progress and Its Problems: Toward a Theory of Scientific Growth.
Berkeley: University of California Press.
8/22/2019 Advances in Early Greek Astronomy
31/32
Greek Astronomy 31
Lebedev, Andrei V. 1990. "Aristarchus of Samos on Thales' Theory of Eclipses."Apeiron, 23:
77-85.
Maravelia, Amanda-Alice. 2006.Les astres dans les textes religieux en gypt antique et dans les
Hymnes orphiques. Oxford: Archaeopress.
Mourelatos, Alexander P. D. 2002. "Xenophanes' Contribution to the Explanation of the Moon's
Light." Philosophia, 32: 47-58.
Needham, Joseph. 1959. Science and Civilisation in China, Vol. 3:Mathematics and the
Sciences of the Heavens and the Earth. Cambridge: Cambridge University Press.
Neugebauer, Otto. 1957. The Exact Sciences in Antiquity, 2nd edition. Providence, R.I.: Brown
University Press.
. 1972. "On Some Aspects of Early Greek Astronomy." Proceedings of the American
Philosophical Society, 116: 243-51.
O'Brien, Denis. 1968. "Derived Light and Eclipses in the Fifth Century." Journal of Hellenic
Studies, 88: 114-27.
O'Grady, Patricia F. 2002. Thales of Miletus: The Beginnings of Western Science and
Philosophy. Aldershot: Ashgate.
Panchenko, Dmitri. 1993. "Thales and the Origin of Theoretical Reasoning." Configurations, 3:
387-414.
. 1994. "Thales' Prediction of a Solar Eclipse."Journal of the History of Astronomy, 25:
277-88.
. 1996. "Thales' Theory of Solar Eclipses and the Birth of Theoretical Science in Early
Sixth-Century Ionia."Hyperboreus, 2: 47-124.
8/22/2019 Advances in Early Greek Astronomy
32/32
Greek Astronomy 32
. 2002. "Eudemus Fr. 145 Wehrli and the Ancient Theories of Lunar Light." In Eudemus of
Rhodes, edited by Istvn M. Bdnar and W. W. Fortenbaugh, 323-36. New Brunswick,
N.J.: Transaction Publishers.
Pedersen, Olaf. 1993[1974]. Early Physics and Astronomy, revised ed. Cambridge: Cambridge
University Press.
Pendrick, Gerard J. 2002.Antiphon the Sophist: The Fragments. Cambridge: Cambridge
University Press.
Popper, Karl R. 1998. The World of Parmenides: Essays on the Presocratic Enlightenment.
Edited by Arne F. Petersen and Jrgen Mejer. London: Routledge.
Putnam, Hilary. 1975. "What is Mathematical Truth?" In Philosophical Papers, vol. 1.
Cambridge: Cambridge University Press.
Rochberg, Francesca. 2004. The Heavenly Writng: Divination, Horoscopy, and Astronomy in
Mesopotamian Culture. Cambridge: Cambridge University Press.
Stephenson, F. Richard. 1997.Historical Eclipses and Earth's Rotation. Cambridge: Cambridge
University Press.
Thoren, Victor E. 1971. "Anaxagoras, Eudoxus, and the Regression of the Lunar Nodes."
Journal for the History of Astronomy, 2: 23-28.
Toulmin, Stephen. 1967. "The Astrophysics of Berossos the Chaldean."Isis, 58: 65-76.
Wilson, Robert. 1997.A History of Astronomy Through the Ages. Princeton: Princeton
University Press.
Whrle, Georg. 1995. "Wer entdeckte die Quelle des Mondlichts?"Hermes, 123: 244-47.