Thales's Prediction of a Solar Eclipse

The only detailed information is given by Herodotus (Histories 1.74): “In the sixth year of the war, which they [Medes and Lydians] had carried on with equal fortunes, an engagement took place in which it turned out that when the battle was in progress the day suddenly became night. This alteration of the day Thales the Milesian foretold to the Ionians, setting as its limit this year in which the change actually occurred.” I use G. S. Kirk's translation: Kirk G. S. Raven J. E. and Schofield M. , The Presocralic philosophers, 2nd edn (Cambridge, 1983), 81–82.

Martin Thomas-Henri , “Sur quelques prédictions d'éclipses mentionnées par des auteurs anciens”, Revue archéologique, n.s., ix (1864), 170–99; Neugebauer Otto , The exact sciences in Antiquity (Princeton, 1952), 142f. In fact, Martin's judgement was less severe.

Such was the reply to Martin's argument made by Tannery Paul , “Thalés et ses emprunts à l'Egypte”, Revue philosophique, March 1880, 303; idem, Pour l'histoire de la science hellène (Paris, 1887), 56.

Heath Thomas Sir , Aristarchus of Samos: The ancient Copernicus (Oxford, 1913), 15.

See ref. 1. For the ancient sources on Thales see Die Fragmente der Vorsokratiker, griechisch und deutsch, ed. by Diels Hermann , 6th edn, ed. by Kranz Walter (Berlin, 1951; hereafter DK), i, 67–81, no. 11.

Die Schule des Aristoteles: Texte und Kommentar, ed. by Wehrli Fritz , 2nd edn (Basel, 1967), H.8.

Some scholars believe that Thales was originally known as one of the Seven Sages, and only became a philosopher and scientist in the imagination of later authors. But the lack of direct evidence for such a conclusion is striking, whereas the evidence for early recognition of Thales as a philosopher and scientist is abundant. See Panchenko Dmitri , “Thales and the origin of theoretical reasoning”, Configurations, i (1993), 387–414, esp. p. 399, n. 30.

Mosshammer Alden A. , “Thales' eclipse”, Transactions of the American Philological Association, cxi (1981), 145–55, p. 151.

Ibid., 153. Cf. Demandt Alexander , Verformungstendenzen in der Ueberlieferung antiker Sonnen- und Mondfinsternisse (Akademie der Wissenschaften und der Literatur, Mainz; Abhandlungen der Geistes- und sozialwissenschaftlichen Klasse, 1970, no. 7).

This lack of involvement also distinguishes our case from such fictional predictions as the one ascribed to Anaximander: We are told ( Pliny , Natural History, 2. 191) that Anaximander foretold an earthquake, thus providing scholars with another reason for rejecting the tradition about Thales's prediction: Martin , op. cit. (ref. 2), 186ff.; Dreyer J. L. E. , A history of astronomy from Thales to Kepler, 2nd edn (New York, 1953), 12f. Yet we are told that Anaximander convinced the Lacedaimonians to leave their homes because of the coming earthquake ( Cicero , On divination, 1.50.112). This is the story of a sage who is capable of using his mind to penetrate hidden things; by means of this secret wisdom he is able to save the entire city. Nothing of this kind is present in Herodotus's account of Thales's prediction. The only real parallel is with Anaxagoras's prediction of the fall of the meteorite at Aegospotami ( Pliny , Natural history, 2.149; Diogenes Laertius, 2.10; Ammianus Marcellinus, 22.16.22, cf. 22.8.5; Philostratus , The life of Apollonius of Thyana, 1.2; and cf. Plutarch , Life of Lysander, 12, and Diogenes Laertius, 2.11–12). Yet the story of Anaxagoras's prediction lacks concrete detail. We are never told to whom he predicted the event, whereas Herodotus's account implies that Thales made his prediction publicly: He addressed it to the Ionians (the comparison with Herodotus 1.170 gives the impression that it was made at the Pan-Ionian meeting). Thales's real prediction could easily have provided a pattern for a number of fictional prediction stories.

Cf. Aaboe Asger , “Remarks on the theoretical treatment of eclipses in Antiquity”, Journal for the history of astronomy, iii (1972), 105–18.

The ancients give various dates though within narrow limits, ranging from the fourth year of Olympiad 48 = 585/584 (28 May 585 would belong instead to the previous year) to Olympiad 50 = 580/579-577/576 (an indiscriminate reference to a given Olympiad may also mean a reference to the first year of this Olympiad). Some scholars assume that c. 50 Ol was the Eudemian date, while a new date was introduced by Apollodorus (active in the latter half of the second century B.C.): Mosshammer , op. cit. (ref. 8), 146, n. 2; cf. Blanche Lenis , “L'éclipse de Thalés et ses problèmes”, Revue philosophique, clviii (1968), 184. The uncertainty occurs because both dates appear in sources ultimately dependent on Apollodorus. If a new date for Thales's eclipse was introduced by Apollodorus, it is not impossible that he received it from Hipparchus (see Pliny , Natural history, 2.53: The passage was convincingly explained by Neugebauer Otto , A history of ancient mathematical astronomy (Berlin, 1975), i, 319–21; cf. Tannery , Pour l'histoire de la science hellène (ref. 3), 38). Later sources (another of Apollodorus's novelties? — cf. Moshammer , op. cit. (ref. 8), 150f) change the name of the Median king and link the eclipse battle with Astyages, son of Kyaxares: See the texts of Hieronimus and Cicero in 11 A 5 DK and also P Oxy 2506 fr. 98 (The Oxyrynchus Papyri, Part 29, ed. by Page Denys (London, 1963), 21; Page's commentary, 45, n.l is misleading, see Huxley George , “A war between Astyages and Alyattes”, Greek, Roman and Byzantine studies, vi (1965), 201–6). The chronology of Kyaxares and Astyages remains a controversial issue; cf. Parker Victor , “Zur griechischen und vorderasiatischen Chronologie des sechsten Jahrhunderts v. Chr.”, Historia, xlii (1993), 385ff, esp. pp. 390ff.

Aaboe , op. cit. (ref. 11), 105. This truth immediately becomes obvious if one looks at the maps of solar eclipses, which can be found in: von Oppolzer Th. , Canon der Finsternisse (Vienna, 1887; English transl. by Gingerich Owen , New York, 1962); Ginzel F. K. , Spezieller Kanon der Sonnen- und Mondfinsternisse (Berlin, 1899); Mucke Hermann and Meeus Jean , Canon of solar eclipses −2003 to +2526 (Vienna, 1983).

Neugebauer , op. cit. (ref. 12), ii, 604.

Hartner Willy , “Eclipse periods and Thales' prediction of a solar eclipse”, Centaurus, xiv (1969), 60–71.

Hartner's conclusion is that Thales intended the eclipse of 18 May 584, while “the unexpected ‘Eclipse of Thales’ came as a surprise to the Sage exactly 12 months earlier”.

van der Waerden B. L. , Science awakening II: The birth of astronomy (Leiden, 1974), 120ff.

Cf. Hartner's criticism of van der Waerden (op. cit. (ref. 15), 71, n. 14): “The circumstance that a solar eclipse may follow a lunar eclipse after 23 ½ lunations does not of course suffice to make a prediction.”

Martin , op. cit. (ref. 2), 192.

See Hartner , op. cit. (ref. 15), 65–67, and Table 2 of the present paper.

The best attested case is the panic among Athenians during the Sicilian campaign (Thucydides, 7.50; cf. Plutarch , Life of Nicias, 23).

Calculations by Marina V. Lukasheva, Institute of Theoretical Astronomy, Russian Academy of Sciences (St Petersburg).

Laertius Diogenes , 1.22, with reference to Demetrius of Phalerum.

In 609/608 B.C. the Assyrian and Egyptian troops fought together against the Babylonian army of Nabopalassar. This effort to save the Assyrian kingdom was in vain, the allies having to retreat. But what happened to Ashur-uballit, the last Assyrian king, and the people around him? A natural supposition is that they took refuge in Egypt. See Chronicles of Chaldean kings in the British Museum, ed. by Wiseman D. J. (London, 1956), 19 and 63 (B.M. 21901, 66ff).

Hartner , op. cit. (ref. 15), 60f.

The data are based on Kudlek Manfred and Mickler Erich H. , Solar and lunar eclipses of the ancient Near East from 3000 B.C. to 0 with maps (Neukirchen, 1971).

As we know from their letters to the Assyrian kings, Mesopotamian observers of the seventh century B.C. were careful about various peculiarities of observed celestial phenomena, allegedly so as to be able to draw lessons as to what this or that particular feature signified as an omen. The eclipse of 11 January 689 was likely to be studied especially carefully because it occurred shortly before the destruction of Babylon by the Assyrian king Sennacherib, and this eclipse is a member of both the 27- and the 54-year series.

Ginzel , op. cit. (ref. 13), 14; Demandt , op. cit. (ref. 9), 6f; Hartner , op. cit. (ref. 15), 71, n. 17.

Based on work by Kudlek and Mickler; the time indicated for the maximum occultation is true local time in hours and fractional parts thereof.

Calculated by Marina V. Lukasheva, Institute of Theoretical Astronomy, Russian Academy of Sciences, St Petersburg.

The table is based mostly on the analogous table in Hartner's article. However, in Hartner's table nos. 7 and 13 are missing, and the data for nos. 9–15 have been replaced in accordance with calculations by Marina V. Lukasheva and Liana I. Rum'anzeva of the Institute of Theoretical Astronomy. The value for Δt assumed in these calculations is based on Stephenson F. R. and Morrison L. V. , “Long-term changes in the rotation of the Earth: 700 B.C. to A.D. 1980”, Philosophical transactions of the Royal Society of London, ser. A, cccxiii (1984), 47–70.

If there had been datable records of solar eclipses observable in Egypt and if Thales had had someone to provide him with the necessary information, he would have established that both major recent eclipses, 30 July 607 (0.86 in Memphis) and 18 May 603 (0.69 at 08.21), had two consecutive predecessors in the exeligmos: 27 June 661 (0.74 at 15.94), 5 May 705 (0.71 at 18.15), and 15 April 657 (0.95), 14 March 711 (0.76), respectively. (Data from Kudlek and Mickler).

Parker Richard A. and Dubberstein Waldo H. , Babylonian chronology 626 B.C.—A.D. 75 (Providence, 1956), 1.

According to Hartner's calculations for Miletus, the magnitude of the eclipses of 19 August 636 and 12 February 635 reached 7.6 digits (at 07.22) and 10.6 digits (at 10.04) respectively.

This is not surprising from the point of view of source criticism. Herodotus's information on the prediction is concrete, not vague. He gives the addressee of Thales's prediction (the Ionians) as well as the manner in which Thales defined the term for the predicted event. It seems that what we have in Herodotus goes back to an even more detailed account, which could be found in one of the Ionian writers, such as Dionysius of Miletus.

Is it possible that Thales could have taken into consideration all the eclipses referred to in Table 2, including those of small magnitude? The answer is affirmative. First of all, the Milesian could have received information from Milesian colonies spread over a wide area, and from other places in touch with Miletus (for instance, the eclipses of 30 September 610 and 13 February 607 must have been very impressive in the Black Sea area). He also was in position to observe all these eclipses himself if the weather was not overcast. He would have had to know, however, that solar eclipses can occur only at the time of the new moon (it is impossible to watch for an eclipse every day!), but this knowledge is testified for Thales by Aristarchus of Samos (P Oxy 3710, col. II 34–43; The Oxyrynchus papyri, liii, ed. by Haslam M. (London, 1986), 97). Now the method of observation of solar eclipses adopted by the Greeks and apparently already by Thales was to watch the reflection of the sun in water ( Plato , Phaedo, 99d; Diogenes Laertius, 7.146; see also Boll Franz , “Finsternisse”, Realencyclopaedie der classischen Altertums-wissenschaft, vi (Stuttgart, 1909), cols 2329–65, esp. col. 2349). This method would allow an observer anticipating a solar eclipse to detect one of very small magnitude (as I anticipated, and confirmed during the solar eclipse of 10 May 1994 in North America).

The eclipse of 1 October 583 would not, of course, have been regarded as a fulfilment of Thales's prediction. This eclipse was too far from the area of Greek settlements to give rise to rumours (the remote city of Cyrena on the Libyan coast was the only significant Greek settlement where this eclipse could have been impressive enough for this). If the partial occultation of the solar disk before setting was observable in Ionia, it is not necessarily the case that the ordinary people would have considered this phenomenon as similar to a sudden darkness during daytime.

So Beloch Julius Karl , Griechische Geschichte, 2nd impression, i/2 (Berlin, 1926), 354.

According to the Babylonian chronicles, there were attacks by the Assyrian and Babylonian armies on each other in March 615 (B.M. 21901, 11); Jerusalem was seized by Nebuchadrezzar on 16 March 597 (B.M. 21946, Rev. 12); Neriglissar returned to Babylon in the month of Addaru (February–March 556) after a campaign in Western Cilicia, on the border of Lydia (B.M. 25124, Rev. 26). See Wiseman , Chronicles (ref. 24), 55, 73, 77.

It seems to follow from P Oxy 2506 fr.98 (cf. ref. 12) that either Alcaeus or his brother Antimenidas participated in the war between Astyages and Alyattes. (We are told, by the way, that Alcaeus praised Thales in one of his poems — 11 A 11 a DK).

According to the calculations by Marina V. Lukasheva, the maximum magnitude of this eclipse was 0.75 at 08.18 for 39°N, 35°E (in the vicinity of ancient Hattusa). Such an eclipse, given the low altitude of the sun in the morning, must have been quite impressive, at least as an omen. However, Herodotus's account of the war between the Lydians and the Medes suggests that the battle took place somewhere to the south or southeast of Hattusa, because we are told that the war was carried on with equal fortune on both sides and that the kings of Babylonia and Cilicia mediated in making peace. The magnitude of the eclipse for a plausible place of the battle must then have been c. 0.80; the brightest planets, and even stars, could be seen. In any case, one should not give undue weight to the Herodotean expression, “the day suddenly became night”. In contrast to his account of the prediction, Herodotus gives no real detail about the phenomenon (or about the battle). The artistic image he uses fits, strictly speaking, no reality at all; it is just a topical expression (cf. Demandt , op. cit. (ref. 9), 10ff) repeated almost word for word when Herodotus refers to the fictional eclipse at the beginning of Xerxes's army's march (7.37). Moreover a characteristic indication that is present in the latter case, that “the sun disappeared”, is lacking in the former. It is worth noting, indeed, that Herodotus makes Thales predict “the alteration of the day” and not “the disappearance of the sun” (which is the original meaning of what became later a technical term). It seems that Thales consciously avoided limiting his prediction to a total eclipse.

One should consider the possibility that Pliny's testimony, usually understood as referring to the date of the eclipse, could in fact record the date of the prediction. Pliny , Natural history, 2.53: “apud Graecos investigavit primum omnium Thales Milesius olympiadis XLVIII anno quarto praedicto solis defectu qui Alyatte rege factus est urbis conditae CLXX.”

See the data in Ginzel's Spezieller Kanon (ref. 13). I give just two examples. The eclipse of 21 September 582 had a successor in the exeligmos, but its magnitude was 4.2 digits in Athens (Ginzel), with the sun quite high above the horizon; 0.10 in Hattusa (Kudleck and Mickler). The eclipse that happened on 18 April 527, the successor in the exeligmos of that of 16 March 581, was central in southern Indian and China and of negligible magnitude, if any, in Ionia (0.16 in Hattusa). As to Helicon's prediction, if he predicted the eclipse of 29 February 357 (which fits perfectly the existing evidence) and not that of 12 May 361 (which is now generally assumed without any real support from the evidence), then his prediction could be based on the recurrences of solar eclipses in the exeligmos. A note on this is in preparation.