Derived light and eclipses in the fifth century

+ mp nen + In: Journal of Hellenic studies, LXXXVIII - 1968 - pe 114 - 127 aes Asfr. Dao OBriea D by riciens] zul, Sciences : v Binen D., Derived light aud eelipses [chez les Pylhago Astrononiic. es in the Arch centr : JHS Ne O'Brien D., Derived light and eclips so Anaximeenes : id some that Thales, maxim claim he t deratiion of 7. | Consiiderat 14-127. 1968 114-12 of plaus Rens no is TRS There . e, eclips clipse an of n explanatio rect t exp . a correc Pythagorcans gave di onss 0 traditi in heyy In rent discrepanc i appa " The . <t goras 1 Anaxa e befor ant : claim ass ing by assum \ be ; resolved by tyy of Anaxa x gorass m may ; derived light and thee priori priorit es of derived light. Lreori in ing kindl and tion reflec à stimelion between Tanne sf, Helene Shut 443 hi Svs. CT” 943. O'BRIEN D., Derived light and eclipses in the fifth century: JHS 88 1968 114-127. Anaxagore, suivi en cela par Empédocle, semble être le premier à fournir une explication valable de l’éclipse. Contre cetle hypothèse, on’ : B 2 : x ERBEN . EN invoque, d’une part, le témoignage de Thalès, Anaximène et certains Pythagoriciens, et d’autre part, le fait que des prédécesseurs d’ Anaxagore ont saisi comment la lune empruntait sa lumière au soleil, cette dernière théorie impliquant à leurs yeux une juste explication de l’éclipse. Réponse à ces objections.

The Journal of Hellenic Studies, Vol. 88. (1968), pp. 114-127. Stable URL: http://links.jstor.org/sici?sici=0075-4269%281968%2988%3C114%3ADLAEIT%3E2.0.CO%3B2-2 The Journal of Hellenic Studies is currently published by The Society for the Promotion of Hellenic Studies. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/hellenic.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support@jstor.org. http://www.jstor.org Fri Jan 26 11:47:29 2007

IN a study earlier in this volume, “The Relation of Anaxagoras and Empedocles’, pp. 93-113, I listed the ancient evidence to the effect that Anaxagoras first gave the correct explanation of an eclipse, and that he was followed in this by Empedocles. A more extensive examination of the evidence raises certain difficulties. For what are, or might appear to be, Anaxagoras’ theories are attributed elsewhere to earlier thinkers. There are two principal elements in this contradiction, the one direct and the other indirect. 1. There is a direct contradiction when Thales, Anaximenes and some Pythagoreans are said to have given the correct explanation of an eclipse, at least if we suppose the Pythagoreans in question to have been earlier than Anaxagoras. 2. There has been thought to be an indirect contradiction when several thinkers before Anaxagoras are said to have derived the moon’s light from the sun. For a theory of derived light for the moon has been thought, whether rightly or wrongly, to entail the correct explanation of an eclipse. In what follows I shall attempt to solve these, and some other incidental difficulties. II 1. Thales. The ways in which Thales is associated with an eclipse vary. (i) Herodotus says that Thales forecast an eclipse of the sunt According to Diogenes Laertius the prediction ‘aroused the wonder’ of Xenophanes, and was spoken of by Heraclitus and Democritus.? (ii) One source for Eudemus repeats that Thales forecast an eclipse. Another source says that according to Eudemus Thales forecast eclipses, in the plural.4 A third source, Theo Smyrnaeus, says that according to Eudemus Thales eöpe zparos an eclipse.5 (iii) According to Pliny Thales investigavit primus omnium an eclipse. (iv) Later this appears in the form that Thales understood eclipses, kareiAndev.? (v) Finally, in Aetius and in a scholiast on Plato, Thales is said correctly to have explained the sun’s eclipse.® The possibility of Thales’ having predicted an eclipse of the sun must rise or fall with the state of contemporary Assyrian evidence. But even if we allow that Thales may in some sense have predicted a solar eclipse, still it does not follow that he will have been able to give the true explanation, or any explanation, of the phenomenon, since in the Assyrian world, at least before 609 B.c., the observation of celestial phenomena does not 1174 (DK 11A5, these references are to DielsKranz, Die Fragmente der Vorsokratiker, 5th edn. onwards). There are frequent references in later writers, cf. n. 61 below. 2 123 (DK 11A1), cf. Eudemus fr. 144 Wehrli. 3 Clem. Strom. i 65 (DK 11A5), fr. 143 Wehrli. 4 Diog. Laert. i 23 (DK 11A1), fr. 144 Wehrli. 5 Theo Smyrnaeus, quoting Dercyllides, Expositio rerum mathematicarum ad legendum Platonem utilium, 198.14-18 Hiller (DK 11A17), fr. 145 Wehrli. 6 N.H. ii 53 (DK 11A5). 7 Suda, s.v. Oalÿ (DK 11A2): repeated in Cedrenus, Historiarum compendium i 275.15-16 Bekker. 8 Aet. ii 24.1 (DK rrAr7a). In Remp. 600A (DK 11Ag). In Aet. ii 29.6 (cf. DK 59A77), Thales is listed with those who gave the correct explanation of the moon’s eclipse, an idea which is repeated in Simplicius, Cat. 191.5-7 and 194.12-13 (not in DK). This would have been a natural step, once it was thought that Thales had given the correct explanation of an eclipse of the sun.

seem to have been tied to attempts at explaining them. In the Greek world things were different: very possibly whoever first attributed the correct explanation of an eclipse to Thales supposed that the forecast of an eclipse, or still more of a number of eclipses, entailed an explanation, and since the forecast or forecasts, were, or were thought to be, successful, that the explanation must have been the correct one. It would not be difficult to imagine some such process of expansion underlying the passages cited above. Kirk and Dicks allow that Eudemus may have been responsible for this expansion of ideas.® Certainly Eudemus was capable of this kind of thing: elsewhere he argues that Thales ‘must’ have known a certain theorem, in order to calculate the distance of ships at sea.!° But in the present instance it seems fairly clear that Eudemus did not go beyond saying that Thales forecast an eclipse, for which Eudemus had the evidence of at least four fifth-century writers. The alternative version, in Theo Smyrnaeus, has fairly clearly arisen simply because a series of abbreviated extracts from Eudemus’ History of Astronomy has been listed under the general rubric, edpe mp@ros, which in the case of Thales leads to the fairly meaningless statement, eöpe mp@ros . . . OaAns 1jÀlov exAcufuy. At the same time, Theophrastus is probably not the immediate source of the report on Thales, (iv) and (v) above. For it is probable that Theophrastus is the source of the report in Hippolytus that Anaxagoras was the first to give the correct explanation of an eclipse.!? Probably therefore the expansion of ideas, if such it was, comes from Aetius, or an intermediate source, working on material from Theophrastus or from Theophrastus and Eudemus. 2. Anaximenes. The excerpts from Eudemus given by Theo Smyrnaeus also imply that the correct explanation of the moon’s eclipse was known to Anaximenes, Expositio 198.14-199.2 Hiller (DK 13A16 = fr. 145 Wehrli) : Eÿômuos ioropet Ev Taîs ’Aotpodoyiaıs, örı Oivoridns eüpe mp@ros . . . , "Avafınevms de dre n oeANvn ex TOO HAlov exer TO Pas Kal riva ékeirre TPÔTOV. I shall consider separately, in section III, the first part of this ascription, the question of the moon’s derived light. The second part of the ascription would at first sight most naturally be taken to mean that Anaximenes was the first to discover in what way the moon is in fact eclipsed. But there are two reasons why this is unlikely to be true. First, Anaximenes is also said to have had a fiery moon. A fiery moon would seem to be inconsistent with the correct explanation of the moon’s eclipse. For a fiery moon would not be likely to be extinguished the moment it entered the shadow of the earth. ® Kirk, The Presocratic Philosophers 80. Dicks, ‘Thales’ in CQ nis. ix (1959) 296. Both writers speak of Eudemus’ error hypothetically. Guthrie is even more tentative, A History of Greek Philosophy i 49. 10 Fr. 134 Wehrli (DK rrA2o). Cf. correct explanation to Thales among others, ii 29.6 (DK 59A77), continues in Stobaeus’ version with an additional half-entry, specifically referred to Theophrastus. Diels, Doxographi Graeci 217-18, supposes Dicks, that both parts of this kind of double entry are ‘Thales’ 303. 4 Boll rightly notes that the entry in Theo is simply an extension of the version recorded by Clement and Diogenes, but he does not make it clear that the genesis of the error lies in the formula edpe mpatoc, and his simplified form of quotation, epe np@tos NAlov EkAeupw, helps to obscure the point, ‘Finsternisse’, Pauly-Wissowa RE, vol. vi 2 col. 2341. 12 Ref. i 8.10 (DK 59A42). Cf. my earlier article, pp. 107 and 109. 13 Wehrli suggests that Aetius ii 24, epi ékhcipewc NAlov, is taken from Eudemus and not Theophrastus, Die Schule des Aristoteles, Eudemos von Rhodos 120. If this were so, it would perhaps be significant that the entry on eclipses of the moon, which attributes the taken, at different dates, from Theophrastus. On Wehrli’s view it could perhaps be argued that they show a collation of material from Eudemus and from Theophrastus. 14 Aet. ii 25.2 (DK 13A16). A reconciliation of this report with the report from Eudemus that ‘the moon has her light from the sun’ is attempted in section III, see especially 123 below. 15 It was possible to hold such a theory in later antiquity, but only in virtue of a theory of light that was essentially Stoic and Aristotelean, see 123 below. Anaxagoras’ moon was also described as Öudswpos and nenvpwuevog: I shall try to show that in his case this was not incompatible with the moon’s shining by reflection, 125-127 below.

Secondly, the correct explanation of an eclipse would also seem to be discordant with the notion of stars moving around, and not under, the earth, ‘like a cap around our head’.1¢ The idea is presumably that the air on which the earth rests is too dense or buoyant for the stars to pass through it.17 The sun also moves around, and not under, the earth.® It would seem likely that the moon is intended to follow suit. But if neither sun nor moon passes under the earth, it is difficult, although perhaps not impossible, to see how the earth could block the sun’s light from the moon in an eclipse. Tannery’s suggestion seems very likely, that the yedders duces Ev raL rómur THY dorépuw ouurepihepouévas éxeivors, which Hippolytus and Aetius attribute to Anaximenes, were in fact the means which Anaximenes used to explain eclipses, whether of sun or moon.!? Tannery’s suggestion was accepted by Burnet, Boll and Heath.?° It has been rejected by Kirk, whose argument seems to have affected Guthrie.?! Kirk writes: ‘It is usually assumed that Anaximenes postulated these invisible celestial bodies in order to explain eclipses; but according to Hippolytus I, 8, 6 (DK 59A42) Anaxagoras, too, believed in them. Yet Anaxagoras knew the true cause of eclipses, therefore he cannot have postulated the invisible bodies for this purpose’. Curiously, Kirk’s argument is the reverse of the argument that was very properly adduced by Tannery in support of his thesis. For two sources affırm that Anaxagoras did in fact use these invisible heavenly bodies as an explanation of eclipses of the moon, in addition to the correct explanation. The one source is Hippolytus, who a few lines after the passage cited by Kirk, i 8.9, introduces what are evidently intended to be the same bodies, precisely as an explanation of lunar eclipses, alongside the correct explanation, the interposition of the earth. The same double explanation, interposition by the earth and interposition by other bodies, is attributed to Anaxagoras by Aetius, who specifically quotes Theophrastus.?? This combination of ideas is not in itself implausible, for it reappears in some Pythagorean accounts of lunar eclipses, which are considered in the section immediately following. The fact that Anaxagoras and some Pythagoreans used ‘dark bodies’ to explain the moon’s eclipse supports Tannery’s theory that similar bodies in Anaximenes were used for the same purpose.” A practical objection to Tannery’s reconstruction might be that both sources which mention the ‘earthy bodies’ in Anaximenes place them in the same region as the stars, so that, if Tannery is right, the stars must be nearer the earth than the moon is (or nearer than the sun and moon, if the ‘earthy bodies’ are to explain solar eclipses as well). But this positioning of the stars is not impossible. Anaximander and others thought that the stars were nearer to the earth than the sun or moon, and Leucippus thought that they were 16 Hippol. Ref. i 7.6 (DK 13A7). Aet. ii 16.6 (DK 13A14). 17 Cf. Aet. ii 23.1 (DK 13A15). Anaximenes’ earth ‘rides’ or floats on a cushion of air, Hippol. 22 Aet. ii 29.6-7 (DK 59A77). 28 Kirk, loc. cit., argues that Diogenes’ dpaveis Aidoı, used to explain meteors (Aet. ii 13.9 = DK Ref. i 7.4 (DK 13A7); [Plut.] Strom. 3 (DK 13A6); With Tannery’s suggestion there is no need to Aet. iii 15.8 (DK 13A20); ef. Arist. De caelo 294b13 ff. (DK 13A20). 18 Hippol. Ref. i 7.6 (DK 13A7). Cf. Arist. Meteor. 354a27-32 (DK 13A14). suppose that the evidence has been falsified. 1° Hippol. Ref. i 7.5 (DK 13A7). A slightly different form of words in Aet. ii 13.10 (DK 13A14). Tannery, Pour l’histoire de la science hellène (and edn. by Diès), 157-8. 20 Burnet, Early Greek Philosophy 78. Boll, ‘Finsternisse’, col. 2342. Heath, Aristarchus of Samos 44. 21 Kirk, Pres. Phil. 135 n. 1. 155-6. Guthrie, History i 64A12) have been read back into Anaximenes. Gershenson and Greenberg argue the other way round from Kirk, Anaxagoras and the Birth of Physics (New York, London, Toronto, 1964) 351 and 353. They assume that Anaximenes’ yebôeis pÜoeis were used to explain eclipses, but suppose that the passage in Hippolytus has wrongly attributed this, as well as the correct explanation, to Anaxagoras. But they do not mention that the same double explanation is attributed to Anaxagoras by Aetius, specifically quoting Theophrastus, nor that the same duality recurs in accounts of Pythagoreans. Cf. 124 below.

below the sun but above the moon. In the absence of evidence either way for Anaximenes, it is possible that he too thought that the stars were lower than the moon, or than the sun and moon. Tannery’s explanation of the moon’s eclipse would be compatible with the idea of a fiery moon, and with the notion that the stars, sun and moon pass round, and not under, the earth. At the same time, yewders púoeis as an explanation of the moon’s eclipse are perhaps sufficient to have given rise to the rather loose expression in Theo Smyrnaeus. As I have already noted, the items which Theo excerpts from Eudemus’ History of Astronomy are made to depend on an initial eôpe mp@ros. In the case of Thales this leads to the pretty meaningless statement, eöpe mp@ros . . . HAlov exAewsw, where two other sources show fairly clearly that what Eudemus said was that Thales forecast an eclipse. Very possibly, in the case of Anaximenes, Eudemus said simply that Anaximenes gave an explanation of the moon’s eclipse. Theo’s formula turns this into eöpe mp@ros . . . 1 oeAyvn . . . riva èkAeire rpörrov, with the consequent (false) implication that Anaximenes gave the correct explanation of the moon’s eclipse.?% 3. Some Pythagoreans. ‘Two entries in Aetius attribute the correct explanation of the eclipse of sun and moon to some Pythagoreans. The entry which deals with the moon’s eclipse adds the interposition of the antichthon as an explanation alongside the true explanation, the interposition of the earth.?7 Aristotle mentions an extension of this scheme, De caelo 293b21-25: éviois dé Soxet Kal mÀelw compara ToLadra Evdexeodu bepeodaı mepi To peoov, nuiv de ddnAa Sia THv Eemimpoohnow THs yfjs. 616 Kal Tas THs ceAnvns Erdeubeis mÄeiovs 7} TAS TOD HAlov yiyveodai dacw: TAY yap hepouévwr ékaoToy avrippatrew atrnv, AA” où póvov tiv yíjv. In the preceding lines Aristotle has considered a Pythagorean theory of an antichthon and a central fire. Fairly certainly the present scheme is also Pythagorean, for only on a Pythagorean scheme would there be bodies circulating around the centre and below the earth.?8 24 The position of Anaximander’s stars is determined by Hippol. Ref. i 6.5 (DK ı2Aıı) and Aet. ii 15.6 (DK 12A18). Stobaeus’ version of the entry in Aetius adds Metrodorus of Chios and Crates. For Leucippus, Diog. Laert. ix 33 (DK 67A1). Contrast the report on Democritus, Aet. ii 15.3 (DK 68A86). In a report on Parmenides, Aet. ii 15.4 (DK 28A4oa), the stars are apparently put below the sun; but the difficulties in the arrangement of Parmenides’ heavens are still unresolved, and it may be that this entry reflects misunderstanding of frr. 10-12, as Zeller argues, Die Philosophie der Griechen, Teil i Abteilung 1 (6th edn. by Nestle) 714 n. 2. It is significant that when Anaxagoras’ ‘invisible bodies’ are first introduced by Hippolytus they are specifically said to be carried around with the sun and moon below the stars, Ref. i 8.6 (DK 59A42). In the account of eclipses they are said to be below the moon, Ref. i 8.9 (DK ibid.), Aet.ii29.6-7 (DK 59A77). 25 The fact that Anaximenes’ stars un Oepuaivew Oud TO uújkos Ts droordoews, Hippol. Ref. i 7.6 (DK 13A7), need not show that they are further away from us than the sun, for their heat will presumably be proportionate to their size as well as to their distance. At Aet. ii 14.3 (DK 13A14), Anaximenes’ stars are said to be fastened to the firmament; but mention of td KpvotaddoewWés shows that the first part of the entry in fact belongs to Empedocles. 2° A more radical, and less likely, explanation would be that Eudemus or Dercyllides thought that the correct explanation of the moon’s eclipse followed from the fact of the moon’s derived light. 2” For the sun, ii 22.5 and ii 24.6 (not in DK), reconstituted from a single entry in Stobaeus by Diels, Dox. Gr. 352 and 354. For the moon, ii 29.4 (DK 58B36). 28 This is Zeller’s argument, Teil i Abteilung 1 532 n. 2. Simplicius writes, De caelo 515.25-26: Aeyeı de 6’ AdéEavdpos, 6tt d¥vatat Toüro Kai ds IIvdayopeiwy tidy éni tadtys pevouévav Tijs ÔdEns AKoveodaı. The dubitative note in övvaraıseems tome to tell against Guthrie’s suggestion, Historyi 283 n. 1, that Alexander is here drawing on Aristotle’s lost work on the Pythagoreans. For Alexander would presumably have accepted Aristotle’s authority on such a point as conclusive. On one occasion when Alexander disagrees with a point in Aristotle’s history of the Pythagoreans he is at pains not to blame Aristotle, but to suppose that Aristotle’s account uerayeypagdaı mäAÂov in tivo ap. Simpl. De caelo 392.12-32. Stocks’ Oxford translation, taken over by Guthrie, in his Loeb edition of the De caelo, seems to me to misplace the kaf in the first sentence quoted in the text above: ‘Some of them even consider it possible ... (my italics). The force of the kaf would seem to be intensitive, ‘quite a number of such bodies’, or additive, ‘several such bodies as well (as the antichthon)’.

Both Pythagorean theories include the correct explanation of an eclipse. Will they antedate Anaxagoras? Burnet, Boll and Guthrie suppose that they do, or may do.2* But, granted the unreliability of the attributions to Thales and Anaximenes, and in face of the evidence cited earlier for the priority of Anaxagoras, there will be no need to suppose that the Pythagoreans who included the correct explanation of an eclipse in their cosmology did so independently.5° I conclude therefore that there is no plausible claimant, directly for the correct explanation of an eclipse, whether of sun or moon, before Anaxagoras. III There remains the second and indirect element in our contradiction, namely the apparent discrepancy that, although Anaxagoras is said to have been the first to explain eclipses correctly, there is nonetheless a variety of evidence that attributes derived light for the moon to a number of thinkers before Anaxagoras. Modern scholars have supposed (whether rightly or wrongly remains to be seen) that a theory of derived light for the moon would have entailed the correct explanation of an eclipse, and would have been an essential element in Anaxagoras’ innovation. They have therefore either denied the reports of derived light before Anaxagoras, or denied Anaxagoras’ priority.?! I have already listed the evidence for Anaxagoras’ priority in my earlier article.?? The evidence for derived light before Anaxagoras is as follows: 1. Twice Plutarch says, or implies, that Parmenides made the moon derive her light from the sun. Each time he quotes a verse to illustrate the point: fr. 14: vurTipaës mepi yatav dAwuevor aAAÓTpiov das. fr. 15: alet manraivouoa mpds adyàs HeAioto. In particular, on one of these occasions, Plutarch refers to the theory that the moon derives her light from the sun as ro "Avagaydpevov, when a moment before he has attributed what would appear (to us) to be the same theory to Parmenides.5à 2. Eudemus, in the passage quoted earlier from Theo Smyrnaeus, is said to have attributed derived light for the moon to Anaximenes.54 3. Aetius attributes derived light for the moon to Thales, Pythagoras and Parmenides, as well as to Anaxagoras and Empedocles.55 4. There is something of the same implication in Plato’s remark in the Cratylus, that the 29 Burnet, EGP 177 n. 1 and 272. Guthrie tentatively proposes this reconstruction in the second volume of his History 196-7, although in his first volume, 285-7, he supposes that the Pythagorean explanation of eclipses is later than Empedocles and Anaxagoras. Boll is inconsistent in another way. At one point he uses the evidence for Anaximenes to reject Anaxagoras’ priority over the Pythagoreans, ‘Finsternisse’, col. 2343: ‘Die Annahme eines Einflusses des Anaxagoras auf diese Pythagoreer . . . ist angesichts des Anaximenes unnötig’. But in the preceding column he had rejected the evidence for Anaximenes: ‘Denn eine Erklarung der MondFinsternisse durch den Erdschatten ist bei des Anaximenes Vorstellung von der Erde als breiter Platte und von der nur seitlichen Bewegung der Gestirne kaum denkbar.’ 30 This is Zeller’s view, Teili Abteilung 1 532 n. 2. 31 References 119 and 120-121 below. 32 Pp. 106-109 above. 33 Fr. 14, Adv. Col. 1116A (partly quoted DK 28B14). Fr. 15, De facie 929A-B (partly quoted DK 28B15). In these two places the implication is quite clear that the moon depends for her light upon the sun. In another place, Quaest. rom. 282B (not in DK), fr. 15 is quoted with a slightly looser implication: donep n oelivn npoogyew eOéder tat Kpeittove (Le. Tt NAloı) Kal devtepedew (fr. 15), Kata tov ITapueviônv. 34 Fr. 145 Wehrli = Theo Smyrnaeus, Expositio, 198.18-199.2 Hiller (DK 13A16). The report of derived light for Anaximander in Diog. Laert. ii 1 (DK 12A1) is clearly false, since it conflicts with the account of the moon as ‘an opening in a wheel of fire, Aet. ii 25.1, ii 29.1 (DK 12A22), cf. Diels, Dox. Gr. 167, Parmenides Lehrgedicht (Berlin, 1897) 111. 35 Jointly, ii 28.5. For Parmenides alone, ii 26.2 (DK 28A42); cf. ii 30.4 (DK 28Baı).

idea which Anaxagoras put forward of the moon deriving her light from the sun was in fact an ‘ancient’ idea.’ 5. Plato’s remark is enlarged upon by Cleomedes, De motu circulari corporum caelestium, 200.24-202.4 Ziegler: nıdevav de Kal of malaıdraroı TÔv dvoık@v Te Kai doTpoAdywv ti amd Tod úÀlov ú oep TO dads Eye ws ÔAdv Eorı mp@rov pev ek THs erypodoyias Tod dvdpaTos aurns, ovTws wvouaouérns ex Tob aédas adryv dei veov Eyew (this is the explanation given by Plato in the Cratylus) kai èk rod émididoc8ar daiSas Tots eis Ta "Apreniora elovodor: TodTo yap ovpBordv Eorı TOD EEwdev Exeıv Tijv cednvynv To das. 6. Diogenes quotes a tale similar to Plato’s, ix 34 (DK 59A5): PaBwpivos Gé brow ev Jlavrodarnı ioropiaı Aéyew Anuörpırov epi 'Avafayópov, ws oùk elnoav adrod ai dd£a1 ai Te mepi HAiov Kai veAnvns, dÀAa dpyatar: Tov S’idyipjobar. The usual response of modern scholars has been to deny this evidence piecemeal, in order to safeguard the priority of Anaxagoras, or of Anaxagoras and Empedocles. Thus Tannery questioned the authenticity of one of the verses which Plutarch quotes as from Parmenides, and explained away the other as insufficient confirmation of Plutarch’s statement.3” Kirk assumes that Eudemus is speaking of reflected light, and says that this must be ‘another backward projection, this time of a belief common to Xenophanes, Parmenides and Empedocles’.8* (The inclusion of Xenophanes must be a mistake, since he is reported as having believed that the moon has her own light.)?® Guthrie describes the reports in Eudemus and Aetius as ‘isolated claims in late writers’, and concludes, as does Heath, that there can have been no notion of derived light before Anaxagoras or Empedocles.*® But the tradition of derived light before Anaxagoras cannot be so easily suppressed. Out of their context the verses quoted by Plutarch from Parmenides are not, it is true, conclusive evidence for the moon’s derived light. But this is not unnatural. Plutarch’s quotations are intended as a reminder, and as a literary adornment, not as a demonstration of something in a work that had become rare or unfamiliar, which is the purpose, for example, of Simplicius’ quotations (cf. Phys. 39.20-21). As it is, even with the verses out of their context, an obvious reason why the moon’s light is aAAdzpiov, and why she is always looking in the direction of the sun, would be because her light depends on the light of the sun. There is no intrinsic reason to doubt Plutarch’s evidence that this was in fact the intention of the verses. Parmenides’ poem seems to have been available to Plutarch, for twice he includes Parmenides in a list of authors whose works, he seems to say, would have been familiar reading for an educated person of his day.“! There is no discernible reason 36 4ogA-B (DK 59A76). The passage has been quoted in my earlier article, 107 above. 87 Science helléne 215-19. Boll, ‘Finsternisse’, col. 2342, remarks against Tannery that the evidence for derived light before Parmenides is ‘absolut sicher’. priority of Empedocles or Anaxagoras is to be breached by Parmenides, for whom there is good evidence of derived light, then there is no longer any reason, on grounds simply of priority, for refusing to allow the notion of derived light to other Diels’ attempted reconciliation of the evidence on thinkers, as well as Parmenides, who are earlier Parmenides seems to me unsatisfactory, Parmenides Lehrgedicht 110-12. In effect Diels simply applies to Parmenides what we are told of Anaxagoras and Democritus by Aristotle and Olympiodorus (roughly than Empedocles and Anaxagoras. the passages cited below 126-127). 38 The Pres. Phil. 156 n. 2. 39 Aet. ii 28.1 (DK 21A43). 40 Guthrie, History i 286, Aristarchus 19 and 75-77. cf. ii 66. Heath, On the question of derived light, Guthrie speaks of ‘the doubtful exception of their (i.e. Anaxagoras’ and Empedocles’) near-contemporary Parmenides’. But ifthesupposed 41 At De recta ratione audiendi 45A-B (DK 28A16), Parmenides is listed with Archilochus, Phocylides (then still extant, see Pauly-Wissowa, vol. xx I col. 505), Euripides and Sophocles. At Quomodo adolescens, 16C (DK 28A15), he is listed with Empedocles (whom Plutarch knew well), Nicander and Theognis. In a quite different context, De Pythiae oraculis 402E (DK 11Aï), designed to flatter Serapion by the antiquity of the tradition he follows of the poet philosopher, the list includes Orpheus and perhaps Thales.

why, if Plutarch knew Parmenides’ poem, he should twice have falsified the context of his quotation. Eudemus cannot properly be lumped with Aetius as a ‘late writer’. Aetius quotes a sentence from, or based on, Anaximenes, which is presumably taken from Theophrastus :*? it is reasonable to suppose that Theophrastus’ contemporary, Eudemus, whose researches were hardly less extensive, will likewise have had some independent knowledge of Anaximenes. Aetius’ inclusion of Thales is probably unreliable. In general, it is unlikely that much, or any, detailed knowledge of Thales’ astronomical ideas, if he had any, was available to Theophrastus. In particular, Aetius’ inclusion of Thales conflicts with the claim, quoted by Theo Smyrnaeus from Eudemus (115 above), that Anaximenes was the first to discover that the moon has her light from the sun. But Parmenides’ work was clearly available to Theophrastus for the writing of the De sensu, 3-4: and Parmenides is not simply added to a list, as Thales is; in addition Aetius writes specifically, ii 26.2 (DK 28A42): ITapueviôns tonv Tat nÀlwe (sc. THY ceAjvny elvai bow): kat yap dm’ adroô dwrilerau. 43 This whole complex of evidence therefore, two quotations from Parmenides, and three direct reports from Plutarch, Eudemus and Aetius, supported by the looser tradition from Plato and apparently Democritus, is unlikely to be simply wrong.# Boll takes the other choice. He denies the evidence on Anaxagoras’ priority. But this tradition is hardly less well authenticated than the other. Direct reports in Plutarch and in one of the sounder parts of Hippolytus are again supported by the looser evidence from Plato and Democritus.“ 42 i 3.4 (DK 13B2). 43 Tdp is in Stobaeus and Eusebius, xv 27.2; it is lacking in Pseudo-Plutarch and Galen, see Dox. Gr. 357a6, bro, 627.19. Diels, following Karsten (Parmenidis Eleatae carminis reliquiae [Amstelodami, 1835] 248 n. 80), dismisses the inclusion of ydp as absurd, Prol. 62-63, and cites as ‘omnino contraria’, 62 n. 1, the Stoic idea that the sun is wetfw dé tig vgs ta. näcav in adtoù pwrilecba, Diog. Laert. vii 144. But the connexion of thought implied in yáp seems to me perfectly possible for the fifth century; and the argument of the Stoics, far from telling against it, seems to me to indicate a similar attitude of mind. (Cf. ‘Anaximander’s measurements’, CQ, n.s. xvii extent and style of his quotations from Empedocles show that he must have had access to a large part, if not the whole, of the physical poem: see Empedocles’ Cosmic Cycle (Cambridge, 1968) 150-1. I would not claim to distinguish between ‘available’ and ‘readily available’ in the case of Theophrastus and Eudemus. Apollodorus found Anaximander’s work, Diog. Laert. ii 2 (DK 12Ar1); and Simplicius records a direct quotation from it, which is taken presumably from Theophrastus, Phys. 24.18-21 (DK 12A9 = Theophr. fr. 2, Dox. Gr. 476). In this instance Dicks’ judgment has perhaps been coloured by his belief that of the series of three pairs of phrastus and Eudemus . . . are extremely small’. figures offered by Tannery in his reconstruction of Anaximander’s system, ‘only 27 in the series has any textual authority’, JHS 36. This is a thoughtless simplification of Kahn, Anaximander and the Origins of Greek Cosmology 88 n. 2, cf. 68. In fact 27, 28 and 19 are recorded in the doxographical tradition, if Diels’ text is to be trusted, Dox. Gr. 348a3-4, b2-3, 351a8, 355a18-19, b16-17. These figures, and the report that ‘the sun is equal to the earth’, Aet. ii 21.1 (DK 12A21), cf. Diog. Laert. ii 1 (DK 12A1), very possibly explain Eudemus’ claim that Anaximander was concerned zepi meyelöv Kai dnootnudtwv, if we suppose that in Simplicius’ rather ragged context, De caelo 471.2-11 (DK 12A19), this has become attached to planets instead of to sun, Again, CQ 301, “There is, therefore, no justification whatsoever for supposing that very late commentators, such as Proclus (5th century A.D.) and Simplicius (6th century A.D.), can possibly possess more authentic information about the Pre-Socratics than the earlier epitomators and excerptors...’ But Simplicius in fact quotes what appears to be a nexion with planets, Dicks finds the report from Eudemus ‘nonsensical’, FHS 30. For Tannery’s reconstruction of Anaximander, see ‘Anaximander’s Measurements’, CQ n.s. xvii (1967) 424-7. 45 ‘Finsternisse’, col. 2343. 46 Cf. 106-107 above. [1967] 426. The point slightly affects the present argument, in that without ydp the additional sentence is perhaps the more likely to be a simple repetition of the general attribution of derived light to Parmenides and others at ii 28.5. 4 I have added these observations on the quality of the evidence, partly in order to off-set the generalisations recently put forward by Dicks in CQ ns. ix (1959) 294-309, and to a lesser extent in FHS Ixxxvi (1966) 26-40. Dicks writes, CQ 299, ‘. . . it can confidently be said that the chances that the original works of the earlier Pre-Socratics were still readily available to his (sc. Aristotle’s) pupils, such as Theolarge part of Parmenides’ Way of Truth, and the moon and stars. By insisting on the formal con-

Boll’s choice carries another disadvantage. Boll accepts the reports of derived light for philosophers before Anaxagoras, including Anaximenes and Parmenides, and he supposes that derived light means reflected light. But successive entries in Aetius say that Anaximenes’ and Parmenides’ moon was fiery, srúpwos 47 A longer entry says that Parmenides’ sun is an expiration of fire, while the moon is a mixture of fire and air.48 A fiery moon, even a partially fiery one, would seem to be inconsistent with the moon’s deriving her light from the sun, if derived light means reflected light.*? The whole difficulty is therefore as follows. 1. Anaxagoras is the first person who is reliably reported as having given the correct explanation of an eclipse, and he is said to have been the first to do so. 2. But there is a wide complex of evidence to the effect that thinkers before Anaxagoras held a theory of derived light for the moon: derived light has been thought, whether rightly or wrongly, to entail the correct explanation of an eclipse. 3. Two thinkers before Anaxagoras who are said to have held a theory of derived light for the moon are also said to have had a fiery moon: a fiery moon would seem inconsistent with the moon’s deriving her light from the sun, if derivation means reflection. A partial solution would be to break the assumption that reflected light would have entailed the correct explanation of an eclipse. Anaximenes and Parmenides, it could then be argued, thought that the moon’s light was reflected from the sun; while Anaxagoras first used the idea to explain eclipses. But this solution does not fully explain the evidence. 1. It does not give Parmenides and Anaximenes a fiery moon. 2. It does not fully account for the form of words in Plutarch and Hippolytus, quoted above, 106-107 that Anaxagoras was the first to give a clear account specifically of the lighting of the moon: Plutarch, Nic. 23 (DK 59A18) : repi cedjvns karavyaou@v Kal akıäs. Hippolytus, Ref. i 8.10 (DK 59A42) : oöros adwpice mp@Ôros ra mepi Tas exAetipers Kal bwriouoÿs. Cf. Plato Crat.409A-B (DK 59A76): 6 Ekeivos vewort Aeyer, örı 1) oeÂÿyn ar Tod HALov exe To bs. 3. And so clear-cut a distinction between Anaxagoras and his predecessors does not altogether explain the controversial note in Democritus and Plato, as cited above 118-119.50 49 Anaxagoras’ moon is also described as didzupos 47 ji 25.2 (DK 13A16); ii 25.3 (DK 28A42). Theodoretus makes a composite entry, and writes, more emphatically . . . êk pdvov ovveordvar nupés, and xenvpœuévos. I shall try to show that in his case this was compatible with a theory of reflection, Graecarum affectionum curatio iv 23 (Dox. 356). 125-127. 48 ii 7.1 (DK 28A37). A later and shorter entry, ii 20.8a (DK 28A43), says that the sun comes from the rarer and hotter part of the Milky Way, while the moon comes from the denser and colder part. Zeller, Teil i Abteilung 1 715 n. 1, and Guthrie, History ii 66, more or less despair over the dis- Empedocles’ moon is made of air solidified by the action of fire: Plut. De facie 922C (DK 31A60); Aet. ii 25.15 (DK ibid.); [Plut.] Strom. 10 (DK 31A30); cf. the account of the formation of the crepancies in these two reports and in those cited composed of fire: that it is an inference is shown in the preceding footnote. They can best be reconciled if Theodoretus is taken as a rewriting of the (original of the) more specific entries in Stobaeus (the only representative here of Aetius), ii 25.2 and 3, and if these and the report at ii 20.8a are taken as by dove. 50 The position outlined in the last paragraph but one is more or less that taken by Taran, Parmenides, a text with translation, commentary, and critical essays (Princeton University Press, 1965) 245 nn. 39 and 40. But to achieve this position Taran has to deny the reports of a fiery moon for Parmenides (in fact he offers the hesitant formula, that the moon is ‘most probably composed entirely or predominantly of air (Night)’, 244-5, my italics); to equally one-sided versions of the exceptionally full statement at ii 7.1, where the moon is a mixture of fire and a colder element. It is not unnatural for the moon to be colder than the sun, if her light is derived light: ef. n. 54 below. firmament, Aet. ii 11.2 (DK 31A51). There is no need to accept the inference of Aetius that it is

The proper solution, I suggest, lies in breaking the other assumption: that derived light means reflected light. This is in fact a modern assumption, which was not shared in later antiquity. For a theory that the moon derives her light from the sun, not by reflection, but by a mixing and joint kindling of the moon’s own light with light from the sun, is given at some length by Plutarch’s Lucius in the De facie and by Cleomedes.5t 1. Lucius introduces kindling and absorption as together a pair of alternatives to the theory that the moon shines by reflection, 929B-C: dwrilecbar roivuv Tv oeAnımv (i) oùk ws adov 7) Kptoradov due Kai Siadavoer Tod mAiou mıdavov Eorıv (ii) od’ ad kard oAAapbir Twa Kal ovvavyaopòv worrep ai Saides adopevov TOD bwrós. oùrws yap obdev Troy Ev vouumviais 7 Öexomunviaus €orar mavoeAnvos Hiv, el ui oréyer und avrıdparreı Tov 1jÀvov aAAd (ia) Sunow bro mavóryros (iia) 7) Kara ovyKpaow exAdpume Kal ovvekdrrrer mepi aörijv ro das.22 The same idea is mentioned again briefly towards the end of Lucius’ speech where he concludes, in favour of the theory of reflection, that the behaviour of the moon mravròs pâÀAov Emideikvvow où ovykpaow GAN’ eradnv oùdè oÛ\aupu Aa repilauu abrns dvta Tov dwriouov, 931A-B. The context immediately preceding shows that the ovyxpacis and oúAAamis would be a conjunction of the light of sun and moon. 2. Cleomedes describes the view that the moon shines by mixing her own light with that of the sun immediately after the view that the moon shines by reflection, De motu, 182.20-184.3 Ziegler: rpirn éoriv aipeoıs 1 Aéyovoa Kipvaobar adrijs (sc. THs oedúvms) Td das Ex re Tob oikelou Kai Tod nAıarod dwrös Kal Tovoûrov yiveodaı oùk drraboûs evovons abrns oùdé . . . [by reflection] . . . ddd’ dAMoioyuevns Úrrò Tod HAıarod dwrds Kal karà ToLadTyY Tv Kpôow iov ioxovons TO bs, où mpdtws, GAAG KaTa peroy#v, ds Ötdrrvpos atSnpos karà weroyijv loxe TO bs oùk dmrabijs dv, AMA Terpappévos bm’ adroû. This is the view which Cleomedes himself adopts, 188.4-7 Ziegler: eöòndov ws où karà dvdkhaow ékméure (sc. % ceAtvn) 7d pds, GANG du’ dAns md TOV HALaKdv Arrivwv Aaumpvvouem Kal KEeKpapévov Exovoa TO Pas oùrw Aaumpúver Tòv dépa. Cleomedes’ later discussion, 190.1-16, makes it reasonably clear that this was in fact the theory of Posidonius, for Posidonius is credited with a defence of the objection that a moon which shines by kindling could not block the light of the sun in an eclipse. Posidonius’ reply is that the airy nature of the moon has sufficient BdOos to prevent the rays of the sun passing right through the moon, so as to be visible to us. Plutarch attributes the same defence to Posidonius, 929D, although from Plutarch’s account alone it would not be clear whether the theory that Posidonius defends is one of kindling or absorption. Cleomedes raises the equivalent objection, 188.8-28: if the moon has a light of her own, mixed with or kindled from that of the sun, how can she be darkened instantaneously ignore the report of a fiery moon for Anaximenes; to relate Plato’s remark in the Cratylus exclusively to the explanation of eclipses; and to ignore, or overlook, the specific mention of lighting in Plutarch and Hippolytus. 51 Ferguson suggests that derived light in PseudoPlutarch’s report on Empedocles, Stromateis 10 (DK 31A30), could mean kindled light and not reflected light, “Two notes on the Preplatonics’ in Phronesis ix (1964) 99. But he rightly rejects this idea, without considering whether the distinction would explain the evidence for thinkers other than Empedocles. 52 The views are marked (i) and (ia), (ii) and (iia), to make it clear that there are two, and only two, principal theories, each of which is once repeated. Cherniss, in the Loeb edition of the De Jacie, gives separate attributions for glass and ice in (i), for (ii), and for (iia), thus suggesting that there are three (or four) different theories involved, and so confusing the structure of the piece. Two of Cherniss’ attributions are questionable. View (i) cannot be taken simply as Empedocles’ view, for it is the theory that light shines through the moon, whereas Empedocles is introduced a moment later, 929E, to illustrate what is clearly intended to be a different view, that light is reflected from the moon. View (ii) is not adequately paralleled by the Pythagoreans mentioned in Aetius ii 29.4 (DK 58B36), for there the moon which is kindled and extinguished is not said to be kindled from the sun, which is an essential element in the theory which Plutarch speaks of. The use of torches to describe kindling is interesting. Cleomedes, quoted 119 above, uses the same image to describe the ‘ancient’ theory of the moon’s light.

by the shadow of the earth in her own eclipse? Cleomedes’ answer is the Aristotelean and Stoic theory that lightness and darkness are instantaneous affections of air and other rare bodies.55 Now if we strip off these two more or less sophisticated accretions offered by Posidonius and Cleomedes, then the simple theory of a moon whose light is kindled from the sun will at once resolve the difficulties in the evidence for the fifth century. For derivation by kindling, as distinct from reflection, is not inconsistent with, in fact it demands, a fiery moon. Equally, derivation by kindling, in a simple sense, does not entail, in fact it excludes, the correct explanation of an eclipse. For derivation by kindling, except on a specifically Aristotelean theory of light, makes it unlikely that the moon would lose her light the moment she entered the shadow of the earth; and similarly one would not expect a moon which shone by kindling to be able to conceal the light of the sun, except at the expense of revealing her own. Hence I suggest that two pairs of ideas should be distinguished, in order to resolve the apparently discrepant traditions on the question of derived light and the priority of Anaxagoras. 1. Empedocles and Anaxagoras derive the moon’s light from the sun by reflection. In accordance with this, they both give the correct explanation of eclipses. 2. Anaximenes and Parmenides are said to have derived the moon’s light from the sun: they are also said to have had a fiery moon; and they are not (reliably) reported to have given the correct explanation of eclipses. This, I suggest, is because their moon derives its light from the sun not by reflection but by kindling.*4 These two pairs of ideas are sufficiently similar, in that they both derive the moon’s light from the sun, for it to be possible to allege, as did Plato and apparently Democritus, that Anaxagoras was revamping an ancient idea, and for Plutarch to write, in the Life of Nicias 23 (DK 59A18): 6 yap mp@ros caploraróv re mévrwv Kal Oappalewraror mepì cedjvns karavyaouÓv Kal okiâs Adyov eis ypadiy karabéuevos *Ava£ayôpas . . . A theory of the moon’s being lit from the sun by kindling, and not by reflection, with ‘invisible earthy bodies’ or some other means to explain eclipses, would fit nicely the notion of something less ‘clear’ and less ‘bold’ than the correct explanation. However, I must at once admit that there is no direct evidence for the fifth century of a moon whose light is kindled from the sun. What was familiar in the fifth century was the idea that heavenly bodies were kindled and extinguished, either in eclipses or by their rising and setting. Xenophanes and Metrodorus of Chios thought that the stars and sun respectively were kindled and extinguished.5 Diogenes thought that the sun was extinguished.°® Some Pythagoreans thought that the moon was kindled and extinguished.5? Later the idea was taken up in the Epicurean school;58 and Dercyllides, in the fourth century, mentions as a false, but apparently common, assumption that the ‘divine’, i.e. the heavenly, bodies rose and set because they were kindled and extinguished.5® It would 53 Apart from the passages in Plutarch and Cleomedes, the idea of kindling seems to make a fleeting appearance in Martianus Capella, De nuptiis Philologiae et Mercurii ii 118, ‘quis Lunam flammet vel minuat radius’, if flammare has a literal sense, and possibly in Philo, De prov. ii 95 Aucher, and accensa should be taken literally in Cicero’s brief “iubar Lunae . . . secundum perceptionem et redindescriptions of the moon at De nat. deor. i 8.7 and De Rep. vi 17. 54 This would explain why, although Parmenides’ moon is fiery, it is apparently not as fiery as the sun, cf. n. 48 above. 55 Xenophanes: Aet. ii 13.14 (DK 21A38). tegrationem lucis solaris’, although on p. g2 the idea Metrodorus: [Plut.] Strom. 11 (DK 70A4). is of reflection. Reflection seems to be the usual view, although the phrasing is not always explicit: Geminus, El. ta astr jae, chapters 9-11; Macrobius, In somn. Scip., i 15.10-12, cf. i 19.8-14; Pliny, NH ii 8.11; Basil, Hexaemeron 189A. Possibly incensa 56 Aet. ii 23.4 (DK 64A13). 57 Aet. ii 29.4 (DK 58B36). 58 Diog. Laert. ix 92. 59 Ap. Theo Smyrnaeus, Expositio 199.21-200.2 Hiller.

have been a simple step from this to the idea that the moon’s light was kindled from the sun, which is essentially Cleomedes’ view and the view which appears in Plutarch. But this extension is not directly attested for the fifth century, and the reconstruction I have offered must therefore remain a speculative one. IV To simplify the exposition of the foregoing argument, I have delayed consideration of some incidental difficulties, real and imaginary, in the evidence on Anaxagoras’ moon. 1. In a recent study of Anaxagoras, two American writers, Gershenson and Greenberg, deny that Anaxagoras gave the correct explanation of an eclipse. They write that Anaxagoras ‘held that it (sc. the moon) does not give off its own light but shines instead by the reflected light of the sun. On the basis of this view he (Anaxagoras) would have been able to give a fairly complete explanation for the phases and eclipses of the moon provided he assumed also that the moon was nearer the earth than the sun. Unfortunately we do not know if he made this assumption nor have we any reliable reports from which we could conclude that he constructed a theory of lunar phases or eclipses’, Anaxagoras and the Birth of Physics (New York, London, Toronto, 1964) 41. The idea that the moon is ‘lower’ than the sun, and so nearer the earth, is in fact explicitly attributed to Anaxagoras by Hippolytus, i 8.7 (DK 59A42): eva de tiv ceAjny karwrepw Tod mAiov mÄNOLWTEPOV pv. Anaximander and Empedocles thought the same. There is, I think, no indication that anyone in the fifth century thought otherwise.60 The denial that there are any ‘reliable’ reports that Anaxagoras gave the correct explanation of an eclipse of the moon in effect dismisses the evidence already noted from Plutarch, Hippolytus and Aetius. Gershenson and Greenberg offer no specific arguments against the reliability of Hippolytus and Aetius, and they do not even mention Plutarch.* 2. The size of Anaxagoras’ sun, larger, or many times larger, than the Peloponnese, could be an obstacle to the correct explanation of an eclipse. For if the sun is smaller than the earth, then the shadow cast by the earth should extend outwards in an ever increasing cone, and the moon would be eclipsed night after night.62 a eyes lA € m But even if Anaxagoras did intend the sun to be smaller than the earth (and this need not have been his intention), it is clear that he did not draw the conclusion that the earth’s 60 Anaximander: Aet. ii 20.1, ii 21.1, ii 25.1 (DK 12A21 and 22); Hippol. Ref. i 6.5 (DK 12A11). Empedocles: Aet. ii 31.1 (DK 31A61). Cf. ‘Anaximander’s Measurements’, CQ n.s. xvii (1967) 423-9. There was of course in the fifth century a difference of opinion over the position of the stars, see 116-117. 61 They do of course include thé passage from Plutarch in their translation of the testimonia, 128, but Plutarch is not mentioned in the note to the passage quoted above, 474, nor in the brief discussion of the point later in the work, 351 and 353. Plutarch and Hippolytus are cited above, in the discussion of Anaxagoras’ priority, 106-107. For the passage from Aetius, ii 29.6-7 (DK 59A77), cf. 116 n. 22 above. In the second half of this entry the «al in “Avafayópas, &s gnor Oedypactoc, Kai Tor dnoKdtw Tic oeAnvns é00’ 6te owudrav ènunpoodotvrwv may imply that the correct explanation as well as the additional bodies were attributed to Anaxagoras by Theophrastus. On the nature of this double entry cf. n. 13 above. There are of course less ‘reliable’ reports in later writers, who associate Anaxagoras with the forecast or the explanation of an eclipse, e.g. Valerius Maximus, Facta et dicta memorabilia viii 11, ext. 2 (not in DK), Philostratus, Vita Apollonii i 2 (DK 5946). 82 ‘Larger’: Hippol. Ref. i 8.8 (DK 59A42); Diog. Laert. ii 8 (DK 59A1); Theodoret. Graec. aff. cur. iv 22 (Dox. Gr. 351). ‘Many times larger’: Aet. ii 21.3 (DK 59A72); Eus. Praep. Evang. xv 24.2 (cf. Dox. Gr. 351). Galen’s statement, that the sun is larger than the earth, Hist. philos. 63 (Dox. Gr. 626), although it may be true in substance, is probably in Galen simply a random error, arising from the fact that the preceding sentence, on Anaximander, twice makes a comparison of sun and earth. (The corresponding statement, that the moon is the same size as the Peloponnese, quoted by Plutarch, De Jacie 932B, is missing from Diels-Kranz.)

shadow covers a large part of the heavens. For, according to Aristotle, Anaxagoras explained the Milky Way as the light of stars screened from the rays of the sun by the shadow of the earth.® The shadow of the earth must therefore be a fairly narrow band, which would only occasionally obscure the light of the moon. 3. Burnet allows that Anaxagoras gave the correct explanation of an eclipse, but he gives a particular twist to the question whether he was the first to do so. He writes, EGP 272: ‘As to the moon’s light and the cause of eclipses, it was natural that Anaxagoras should be credited at Athens with these discoveries’. Burnet’s explanation is an expansion of a remark by Zeller, to the effect that Anaxagoras was ‘one of the first’ to bring the idea of reflected light for the moon to mainland Greece.* Burnet’s idea might possibly account for Plato’s remark in the Cratylus, if it stood in isolation. But it is difficult to see why a local Athenian tradition should determine both Plutarch’s testimony and Aristotle’s, if we suppose Hippolytus’ testimony to derive ultimately from Aristotle’s school. It is fairly evident that Burnet and Zeller write as they do simply by way of providing covert defence for their own (false) belief that Empedocles wrote before Anaxagoras. 4. I turn to a more serious point. It is not explicitly stated that Anaxagoras’ moon shines by reflection. Plutarch’s words, accounted a verbatim quotation (fr. 18) by Diels, are not in themselves decisive, De facie 929B: os évriômor THe oelÿvm To Aaumpôr. This sentence shows that the moon’s light is derived light, but not whether it is derived by kindling or by reflection. (The adjective Aaurpés might perhaps suggest reflection, but it is hardly conclusive evidence.) Moreover, Anaxagoras’ moon is described in the doxographical tradition as yíjv TETUpWLEVNV orepépviov Exovcay Tip, or as oTepéwua Sidmupov, and the unevenness of the moon’s appearance is said to be the result of a mixture of ro mupoadés and ro Cod@des.** If the body that ‘reflects’ light itself becomes hot or fiery, this would seem to be a process of mixing and kindling, and not of reflection. These difficulties, although serious, are not, I think, insurmountable. In the first place, it is reasonably clear from the context of Plutarch’s discussion in the De facie that the words quoted, or summarised, by Plutarch were intended to describe a theory of reflected light. For Lucius remarks, 929B: 6 pv odv Eraipos Ev rit SuarpiBie roûro 81) ro *Ava£ayéperov ämodewvds ds “ijAvos Evrißmor THe oeh TO Aaumpov’ médokiunoer. There is a slight break in the dramatic sequence of the dialogue at this point, but the most natural implication of Lucius’ remark would be that ‘our comrade’ won the applause of the company by his exposition and defence of Anaxagoras’ thesis. Now throughout the dialogue the trio composed of ‘our comrade’, Lucius and Lamprias is evidently intended to present a common front, by way of providing the scientific foundation for Sulla’s myth. The thesis which Lucius, their spokesman on this point, himself approves is that the moon shines by reflection (and so has an earthy rather than a fiery nature) as opposed inter alia to the view that she shines by mixture and kindling. Probably therefore the theory which ‘our comrade’ defended was likewise a theory of reflection. It is no inconsistency that after referring to ro ’Avagaydpevov, Lucius puts forward the theory of reflection as 76 to ’EumedorAeovs. In the rich display of erudition which Plutarch likes his characters to offer it is entirely natural that the two theories should be, at least in essentials, the same. The ideological structure of this part of Lucius’ speech will then be as follows: 88 Meteor. 345a25-31 (DK 59A80), discussed more fully below, 126. % Teil i Abteilung 2, 1241-2. Zeller notes without comment, 1241 n. 3, the evidence for the priority of Anaxagoras’ explanation of an eclipse. 65 See 93 n. 3 and 94-96 above. 66 The three statements come respectively from Achilles, Isag. 21 p. 49.4-7 Muller (DK 59A77); Aet. ii 25.9 (DK ibid.); and Aet. ii 30.2 (DK ibid.).

(i) Reflection is introduced, as ro ’Ava£ayöpeıov, 929B. (ii) Two alternative theories, kindling and absorption, are then disposed of, in part with the help of illustrations from Empedocles, 929B-D. (iii) Lucius concludes by endorsing reflection, drodeiweraı rotvuv ro Tod ’EpmeSoxdéovs, 929E. But if Anaxagoras’ moon shines by reflection, how can it be sremvpwpévos and Sudaupos? The answer, I suggest, is to be seen in terms of a comparison with Anaxagoras’ stars. In the Meteorologica Aristotle says that Anaxagoras’ stars have a light of their own, but that this is ‘prevented’ or ‘held back’, kwAveodaı, by the light of the sun, except for the stars of the Milky Way, whose own light is visible, because it is screened from the sun’s rays by the shadow of the earth.°” Olympiodorus makes explicit the parallel with the moon: the moon has a light of her own which is visible as a ‘smouldering’ light in eclipses; at other times we see the sun’s light reflected from the moon, 67.32-68.2 Stüve: rpirn Sdéa ’Avafayópov Kai Anuorpirov. obtoi dacı tov yalakiav efvar 76 iSvov das T@v dorépov TÔv wij dwrılouevwv bo Tob HAiov. éyovor yap, Paci (Stüve, bor codd.), ra dorpa ro iSiov dds Kai emikrnrov TO darò Tod úÀlov. Kat ÔmAot m ceArjvyn. Taúrys yap Erepov pev TO iSiov dds, Erepov SE 76 darò Adv: TO yap idtov adriÿs Pas avOpax@dés eorw, ws ÖnAoî quiv 1 eAAeubis adrijs. GAN’ où mavra, dace (Stüve, dnoı codd.), 76 érikrnrov déxyovrau: Tà obv wij Sexdpeva Ereiva Tov KUKAov Tod yalafiov amepyalovrar.®8 Olympiodorus, it is true, does not usually show any knowledge of the Presocratics which he could not have obtained from Aristotle, or earlier commentators, or from the commonplaces of doxographical tradition. In any case, the parallel with the moon seems to be offered as Olympiodorus’ own illustration, as distinct from the views said to be held by Democritus and Anaxagoras, which are marked by pact. It would be wrong therefore to take Olympiodorus’ words as positive evidence for Anaxagoras. Nonetheless in this instance Olympiodorus’ idea seems to have a good chance of representing Anaxagoras’ view. For sun, moon and stars, according to Anaxagoras, have the same origin: they are stones, or earth, caught up into the sky and heated by the force of rotation.®® But the stars and the moon are colder than the sun: the stars because they move in a colder part of the sky; and the moon because of coldness in her composition.?® It would seem reasonable to expect that the same kind of origin for stars and moon will mean the same kind of light for both. When a similar theory of ‘natural’ light which is overwhelmed by the light of the sun is credited to Antiphon, it is specifically applied to the moon as well as to the stars. The same theory for Anaxagoras would explain both Plutarch’s evidence, 67 Meteor. 345a25-31 (DK 59A80). Repeated in Aet. iii 1.5 (DK 59A80); Diog. Laert. ii 9 (DK 59A1); Hippol. Ref. i 8.10 (DK 59A42). The idea that the light of the moon, or of the stars of the Milky Way, shines out brightly in the darkness, but is, or would be, overwhelmed by the light of the sun became fairly common. In one form or another it is attributed to Democritus, Arist. Meteor.345a25-31 (DK 68Agı); and to Antiphon, Aet. ii 28.4 (DK 87B27); and it reappears in Plutarch’s De facie 933C-E; in Cleomedes, De motu 180.2-13 and 216.28-218.6 Ziegler. For the general theory of weak and strong lights, Arist. De caelo 305a9-13. 68 The smouldering light of the moon when she eclipses the sun is described in Larousse Encyclopedia of Astronomy (revd. edn. 1966) 159-60, with a colour photograph, facing p. 166. 6 Hippol. Ref. i 8.6 (DK 59A42). Aet. ii 13.3 (DK 59A71): I take tie de edtovias tic nepiôwmoewc to explain karapdéfavra as well as dvapzdoavra. Diog. Laert. ii 12 (DK 59Aı). For the motion of the moon as earth, see Plato, Apol. 26D (DK 59A35), and Achilles, Isag, 21 p. 49.4-7 Muller (DK 59A77). 70 For the stars: Hippol. Ref. i 8.7 (DK 59A42). For the moon: Aet. ii 30.2 (DK 59A77), cf. Hippol. Ref. i 8.9 (DK 59A42). 71 Aet. ii 28.4 (DK 87B27). There is a somewhat similar theory of natural and derived light for the moon in Gregory of Nyssa, Hexaem. 117 B-C. But here the moon’s own light is suppressed not by the stronger light of the sun, but by the density of the moon’s own nature: Tv uêv yap olkodev Aaunnddva N tho obolag naydtns anjuBlvve.

that the moon shines by reflected light, and the doxographical tradition, that she is sremvpwuévos and udarvpos. "2 This theory would help to explain, and I think derives some confirmation from, the continuation of Plato’s account of Anaxagoras’ moon in the Cratylus 409B (the earlier part of Plato’s account is quoted 107 above): Hermogenes: més 57; Socrates: To pev mov ‘oéhas’ kai TO ‘ps’ Tadróv. Hermogenes: vai. Socrates: veov Òé mov Kai Evov dei dott mepi tiv oeAnvnv Toûro TO dads, elmep GANnOA ot Avafayopeıoı Aéyovow: KUKAwı yap mov del aùrijv epumv (sc. 6 mos) vedv dei érBala, Evov dé ÜTdpyet TO TOU mporépov umvös. Hermogenes: srdvv ye. > / , , 4 2 A 7 AN x ee A > A 9 Plato's words apparently mean that according to Anaxagoras, or his disciples, a new moon retains some light from the previous month. This idea, that the moon stores or absorbs light, fits well with the idea that Anaxagoras’ moon does not simply reflect light, but is herself, at least to some extent, a fiery body, as the doxographical tradition claims. 7% Such a theory, of residual natural light for the moon, would be a plausible step from what I have suggested was the ‘ancient’ theory of mixing and kindling to the pure theory of reflection in Empedocles. D. O’Brien. Gonville and Caius College, Cambridge. 7 The bodies ‘below the moon’ (cf. 116 above) may have been retained (from Anaximenes) to explain eclipses when the occulting body did not show any light of its own, or to explain the greater frequency of lunar over solar eclipses. Gershenson and Greenberg, Anaxagoras 338-40, argue that the tradition of a fiery moon is wholly false, and arose ‘by generalizing from the sun to all the heavenly bodies’. The trouble with this argument is that it makes the process of generalisation go back at least to Aristotle, since Aristotle says that the stars of the Milky Way had a residual light of their own: but a generalising tradition of this kind would more naturally have arisen in a doxographical compilation, where the original work was no longer known, and where there would be pressure to make Anaxagoras’ views conform to other accounts of a fiery moon. 73 Ferguson, Phronesis ix (1964) 100, notes the ‘very strange picture’ which Plato gives of the Anaxagoreans; but he appears to think that Plato’s account is incompatible with a theory of reflection. It is interesting that the moon is in fact now thought to have some light of her own, see the report in The Times, 20 Nov. 1967, p. 6.