ElAnt v3n7 - Herodotus's Report on Thales' Eclipse
HERODOTOS'S REPORT ON THALES' ECLIPSE
Thomas Worthen ,
Department of Classics,
University of Arizona,
Tucson,
AZ 85721,
U.S.A.e-mail: tomwr@ccit.arizona.edu
Controversies surrounding Thales' prediction of a famous eclipse event and its dating stem from antiquity and continue into the present. The eclipse, Thales prediction of it, and the historical circumstances for which it was relevant are detailed only by Herodotos (1.74): After this, because Alyattes refused to surrender the Scythians despite the ongoing pleas of Kyaxares, a war ensued between the Lydians and Medes over a period of some five years. During this time on some occasions the Medes won a battle over the Lydians, while on other occasions the Lydians prevailed over the Medes. They even engaged in a battle by night. While they were still struggling for the upper hand in this war, during a battle begun in the sixth year, it happened that when the fighting had been joined, day suddenly became night. A prediction that this inversion of the day was going to happen, was made publically by Thales of Miletus in announcements to the Ionian people; he proposed exactly the same period favorable for it as the one in which the omen actually occurred. When the Lydians and Medes alike saw that it had become night in place of day, they broke off the battle and hastened on both sides with even more speed to bring about peace for themselves. ( 01 )
Historians of later antiquity dated the eclipse to the 49th or 50th Olympiad (585-577). Among these sources, Pliny (N.H. 2.53) gives the date (Ol. 49.1=585). Since nineteenth century science affirmed that there was a total eclipse visible in the Aegean and Asia Minor on 28 May 585 (a date that corresponds very well with Pliny's date), some modern historians seize upon this as the solar eclipse which Thales foretold and therefore the date of the battle. Just how the ancients arrived at these dates is puzzling. Was it done on the basis of the calculated acme of Thales? Was it by association with the names of dates of the rulers, the Median king Kyaxares and the Lydian Alyattes who, Herodotos says, quit warring with each other owing to this portentous apparition that occurred during one of their battles? Did they have a dated list of eclipse events from which they selected the one most nearly approximating the historical facts surrounding the battle? These issues are incidental to the purpose of this paper, which is to offer a novel set of considerations about the eclipse, Thales prediction of it, and Herodotos's misunderstanding of it . Nonetheless, historical considerations have become so enmeshed with the discussion of astronomical issues, that some discussion of them will become necessary during the course of this presentation.
A number of scientists have a-priori doubts about the ability of Thales to predict solar eclipses. ( 02 ) There was no eclipse prediction; hence what Herodotos's reports about Thales a myth. If the solution is that simple, then the year 585 can be of litte consideration for historians attempting to set these events into chronological sequence. Other scientists disregard historical factors altogether in their attempts to save Thales and his prediction. They speculate that, although Thales could not have used a methodology developed by Chaldean astronomers ( 03 ), he may have learned of some other principles upon which he based his prediction. Hence they defend this or that eclipse date based on their own view of what Thales' methodology must have been, without consideration of the problematic chronologies involved. ( 04 ) Herodotos's narrative of these events implies that Kyaxares continued to rule the Medes through the period of the war and the following negotiations. The historian confirms our inferences on this point in another place (1.103), declaring that Kyaxares fought the battle with the Lydians when the day was darkened, and it was he who united Asia beyond the Halys river. The unification of Asia, including the hegemony formed with Babylon to administer the coup de grace to the Assyrian empire is indeed attributed to the last two Median kings by Babylonian Chronicles. Herodotos's Kyaxares is to be associated with Umakishtar who led the Medes in 612 to join the Babylonians in sacking Nineveh. A later Chronicle that gives an account of the conquest of the Medes by Kyros II mentions the Median king's name as Ishtuwigu, clearly Astyages son of Kyaxares as named by Herodotos. The Median conquest is firmly dated through these records to 550. ( 05 ) Using the Babylonian terminus and Herodotos's length for Astyages' rule of thirty-five years, one can posit that the date of the king's accessionto ca.585. As was noted above, however, historians of later antiquity dated the eclipse to the 49th or 50th Olympiad (585- 577). ( 06 ) Among these sources, Pliny ( N.H. 2.53) gives a specific year (Ol. 49.1=585).
Since Herodotos takes pains to give the length of rule for each of the Persian and Median kings, however, one can count the historian's contiguous regnal lengths backward from the sixth year of Xerxes ( 07 ) to Kyaxares. By that method, the death date of the latter must be about 595. . There is a ten-year discrepancy. Furthermore, there is alternate tradition that names Astyages as the Median king at the critical battle with Alyattes. (08, 09) These inconsistent data from antiquity need resolution. Herodotos's own data do not match the date later antiquity gave to the eclipse he says marks a crucial battle led by the king Kyaxares. The crux involves the reputations both of Herodotos and of Thales. We may discard Thales' prediction entirely, thus freeing the chronological debate from any date set by astronomy. At the other extreme we might completly readjust the Median chronology to suit the eclipse date of 585.
The speciously broad path to a solution supposes that astronomical dating via an eclipse of record must correct Herodotos's chronology or even override it. Efforts designed along this line, however, have been unconvincing. Eduard Meyer, for example, proposed that Kyaxares' death date could be brought to 585, and thus allow him to have been king at the time of the date of the eclipse battle supposedly established by sci ence. Meyer accomplished this end only by manipulating Herodotos's text and discounting the historian's usual procedure. ( 10 ) It would be convenient to Herodotos's credibility if a date could be found for the eclipse that accords with his Median chronology without correction. If an Anatolian eclipse occurred between 615 and 586, before Kyaxares' death in 585, we could dispense with the alternate tradition about Alyattes leading the battle. There are two solar eclipses during the period in question that were visible from Asia Minor: the morning eclipse of 30 September 610, and the morning eclipse of 18 May 603. What about Herodotos's description of the battle eclipse makes it unique and so assignable to one eclipse more than another? Herodotos says that the event occurred near the onset of the battle. ( 11 ) If so, then choosing the eclipse of 585 entails an anomaly unless the battalions began their skirmish late in the day. That eclipse did not begin until 4:20 (3:52) p.m. local time. ( 12 ) The historian also records that day became night and that night came about in place of day. If we take this to refer to a solar eclipse, it must have been a total solar eclipse. As seen from the region of the Halys river (assumed site of the battle), however, the eclipse of 28 May 585 covered no more than magnitude 0.6 (.9). ( 13 ) Such an eclipse might even go unnoticed should it occur when the sun is high in the sky. Herodotos's description purports to be true to the occasion, but 'suddenly' could not be true to observation surrounding the eclipse of 585.
A sudden conversion from day to night is a possible description of very few observed solar eclipses. First of all, the eclipse would need to be completely total. Conditions approaching true darkness are rare even so; for when there is an extensive solar corona, it continues sending down enough light to read by. ( 14 ) An unexpected total eclipse, lacking corona, occurring sometime near noon might be observed to occur suddenly, for in this case the gradual diminution of the sun's light and heat might go unnoticed until magnitude >0.95; then it becomes cool and dusky suddenly, as if night were coming on. (15, 16) Many eclipses, therefore, would match Herodotos's description as well as or better than the eclipse of 585. Alas, however, neither eclipse mentioned above matches the description any better. The eclipse of 30 September 610 is more likely than that of 585 only in that the onset of the battle would have to be placed, as is logical, in the morning instead of late afternoon. This eclipse, however, darkened the skies over the Halys less than that of 585. The track of totality for the morning eclipse of 18 May 603 was farther to the South of the supposed battle zone than the other two were to the North, and therefore it was no more striking. The only advantage these candidates may have is that they both occurred prior to either date (585, 595) calculated for Kyaxares' death.
An eclipse of 21 September 582, singled out by some modern commentators, offers some advantages in visibility; but it occurred after the both the death dates for Kyaxares. ( 17 ) At this point we need to scrutinize antique traditions about the eclipse passage and Thales' prediction of the event. The historical tradition reports that Thales did predict the eclipse of 585 and celebrated the scientist as a rare genius. ( 18 ) It is noteworthy, however, that the doxographical tradition about Thales makes no such claim for his prediction of a solar eclipse, only that he understood the phenomenon. ( 19 ) Knowing the causes of solar eclipses and being able to predict them are miles and ages apart. Therefore, the disparity between these two traditions reveals that Herodotos was likely the sole source of the report that Thales predicted the solar eclipse marking the war's end. Indeed, from the evidence, the only set of conclusions which follow are these: 1) Thales was not able to predict solar eclipses; 2) Herodotos reports falsely that Thales did so on one occasion; 3) this led to confusion among subsequent historians, some of whom recast Herodotean chronology to suit some recorded (calculated, in the case of modern historians) eclipse date; 4) scientists, however, were not drawn into this misinterpretation; having read Thales or his epitomizers, they report only about his knowledge of the causes for eclipses and nothing about prediction.
Beyond these tentative conclusions, there are indications in the text that allow us further argument. First, we notice that Herodotos's language describing the effects of this eclipse sounds like a standard literary formula for the description of a solar eclipse: day suddenly becomes night. Herodotos expressed this idea twice in the battle passage, using slightly different language each time:
. . . the day suddenly became night ( hemeron nukta genesthai ) (1.74.2)
. . . they saw night coming on in place of day ( nukta anti hemeres genomenen ) (1.74.3)
Furthermore, Herodotos reuses this description ( 20 ) at 7.37 to dramatize the portentousness of a solar eclipse that, in fact, did not take place. First in a long list of unlucky portents surrounding Xerxes' march into Europe Herodotos mentions a solar eclipse. It occurred the moment the Persians departed from their winter camp in Sardis, a departure prompted by the news from Athos that the canal was finished (7.37.1): . . . as he was setting forth the sun, leaving its seat in the heaven was invisible, though there was not a cloud in the sky but rather clear for the most part, and night came to be in place of day. This description, with its report of a cloudless, bright sky above which night replaced the day, seems like an eye- witness account. ( 21 ) Yet, no solar eclipse occurred in the spring of 480. There was, however, a lunar eclipse. Was this simply a blunder on the part of Herodotos? Once it was determined, a century ago, that no eclipse occurred, there were many attempts to mitigate the gaffe, most of them shielding Herodotos from error and laying the burden of fault on his sources. ( 22 ) The error seems to go deeper than this, however; either Herodotos deliberately fabricated the event of 480, or he was simply ignorant about eclipses and was unable to discriminate between solar and lunar when he encountered a reference to eclipses in his sources. The language Herodotos uses to describe the eclipse of 480, which did not occur, is very similar to that he uses in the passage under discussion. One is tempted to conclude that Herodotos's language about the battle eclipse is not derived ultimately from any eyewitness's account of the event he records. ( 23 ) Rather, having learned that there was an eclipse, he drew on his imperfect understanding of such events and invented or borrowed stock language to describe an eclipse as he conceived one to be.
Despite the similarities between these accounts, however, which point to stock language, there are several peculiarities in the description at 1.74 that require comment. At 7.37 the historian personifies the sun as leaving its seat in heaven and becoming invisible; at 1.74 he makes no mention of the sun at all. Furthermore, in his account of Thales' description is a phrase which is indicative that Herodotos is following some source--an account of an actual prediction as reported in some (Milesian?) tradition. The phrase is: ouron prothemenos eniauton which is not usually translated as I have done above (he proposed . . . the . . .period favorable for it) but with something like 'proposing the propitious year for it to happen'. This phrase has been used to 'disprove' any prediction by Thales. Prediction of eclipses never centers about solar years, but rather complicated cycles that follow the precession of the nodes in the moon's orbit, the shortest and simplest of which is 177 days long. ( 24 ) Hence this phrase involving the time period of a year is proof the Herodotos and/or his sources knew nothing about eclipses. ( 25 ) This argument is, however, rendered untenable by cursory research into uses of eniautos using LSJ. From its earliest uses in the Epic, the word is occurs in contexts where it is contrasted with the usual word 'year', etos ; it is clear from such contexts that the word refers to recurrent, seasonal, time periods of various lengths. (26, 27) Since eclipses occur at the moment of new moon (solar) or full moon (lunar), any prediction about them by the ancients would likely not prescribe the day and hour of their occurrence, but the lunar month, first or fifteenth, when they would occur. Hence 'favorable period' is terminology so perfectly apt to prediction that Herodotos is likely quoting it from an astronomically sophisticated source. The adverb suddenly , as we have seen, is inappropriate to the experience of eclipses; it is not, however, part of his stock language for eclipses, for it is not used at 7.37. Perhaps Herodotos heard or read an actual report about 'suddenness' associated battle eclipse that he did not fully comprehend. Therefore, the features of his language dependent upon that report preserve some features of a remarkable event, while other parts, invented out of Herodotos's misunderstanding, so garble the account that his successors have been unable to rationalize it. The invented language has to do with solar eclipse phenomena: the sun disappears and night replaces day.
Perhaps, however, the event in question was not the total solar eclipse we have assumed it to be from his description. Short of a solar eclipse, what eclipse event involving a sudden change could be dramatically portentous? The text gives us a clue in the form of a curious little sentence that precedes the description of the eclipse: 'In this war they brought about a battle by night; . . .' ( en de kai nuktomachien tina epoiesanto ). Some editors excise this sentence from their text. ( 28 ) Those who remove it do so on the grounds that it is a blunt digression, and hence a gloss that has intruded. Why such a gloss would be inserted here, however, no one can explain. Hude keeps it in the text, but How and Wells feel compelled to explain, 'This one is not the eclipse battle'. The issue needs revisiting.
If we take the night battle seriously, there may be an explanation. First we must notice in the text that the night battle is the first mention of a specific battle in this war of five years' duration; then the so called eclipse battle, carefully excluded by scholars from being the night battle, is introduced without further adieu. If we were not so certain that there are two battles mentioned here, the only natural way to take the second mention of the battle, with the definite article tes maches , is as a reference to the first mention of a battle, namely to the night battle. The night battle indeed would be the one and only battle in our minds, if we were not so given to an interpretation which presupposes that Herodotos is talking about an event set in the daytime by a solar eclipse. If it was possible to come to Herodotos 1.74 without a prejudice built up by millennia of commentary as to what it all means, one might translate:
After this, when and because Alyattes refused to give up the Skythians to Kyaxares despite his continual petitions, a war took place between the Lydians and the Medes over a duration of five years during which often the Medes got the best of the Lydians and often the reverse. In this war they brought about a battle by night; and the engagement came about in the sixth year when they were still contending with each other at war on an equal basis, when it happened, as the battle was beginning, that day suddenly became night. (Thales the Milesian predicted to the Ionians this change [of day to night] would come about, setting beforehand the favorable period in which the ominous event did indeed happen.) When the Lydians and Medes saw it become night instead of day, they quit the battle and rather made haste on both sides that peace came about.
Interpreting the passage without prejudicing it to refer to a total solar eclipse sets us free to examine just what other kind of phenomenon could be the subject of the report; it leaves us the problem, however, of explaining the passage in terms of its astronomy, of dating it within the regnal years of Kyaxares, and of justifying how Thales may have predicted the reported phenomenon. Supposing that it was a night battle. Then, at its onset, day became night. The night battle began in the evening, and day became night. The ordinary exchange of night for day could not be the issue here, however. The phenomenon must be extraordinary, and therefore ominous, to justify Herodotos's singling it out. After sunset, day changes to night more or less gradually depending on the season. The summer campaign season is a time of long twilight, but even the expectation of a two-hour twilight would not allow a general to plan to begin a battle at sunset or to extend an ongoing battle into the darkness of night. At the full of the moon and for a few nights prior, bright moonlight extends visibility from sunset and beyond sufficiently for some activities to take place all night long. Could battle be one such activity? Epipolai, a notable night battle reported by Thucydides, took place at the bright of the moon (7.44). During the waxing moon just before it was full, an army on bivouac would notice the increasing illumination of the moon during the evening and into the night, and anticipate its even more glorious rise on the following evening. If one side or the other, or both, had planned the evening battle, but it grew dark suddenly and contrary to expectation, the battle would have to be suspended. Herodotos remarks on the suddenness of nightfall that intervened upon the battle. A nightfall more sudden than usual would have to be one that arose contrary to the expectation that there would be natural lunar illumination for the planned battle at night.
The only event that could frustrate such an expectation for a full moon, other than an overcast sky, is a total lunar eclipse intervening at, or shortly after, sunset-moonrise. Judging from an unhappy event reported by Thucydides, lunar eclipses were greatly ominous. Since Nikias called off a battle owing to the lunar eclipse of 27 August 413, we should not be surprised if such an event caused the Medes and Lydians to interrupt their battle, lay down their arms and seek peace. Expecting to continue their battle into the night under the light of a full moon, the Medes and Lydians were surprised when the moon failed to appear, as expected, to light their ongoing battle. Indeed, the portent was so startling that not only did they break off the battle, they concluded the war.
This interpretation of Herodotos 1.74 allows us to retain his report of a night battle and of an eclipse that ended it. Lydian sources about the battle, the omen, and the consequences, we suppose, must have found their way to Miletus where Thales himself, or a follower, correlated the timing of the event to a lunar eclipse foretold by the father of astronomy. The report, as passed on to Herodotos then 1) claimed Thales' prediction of an eclipse which coincided with the battle eclipse; 2) reported that the light failed unexpectedly, thus abrogating the battle; 3) recounted the political consequences of the omen. Herodotos then garbled this account owing to his assumption that the reported eclipse phenomenon was a solar eclipse. He retained the first and third notices, but misunderstood the notice about the unexpected failed illumination. Instead, using language he had learned to describe solar eclipses, he interpreted the Lydian report as an event in which broad day suddenly turned into night.
The year 585 must be forever forgotten as important to Thales and his science. Indeed, we must declare, with others, that there was no such thing as Thales' prediction of a solar eclipse. After due consideration of possible heuristic methods (to be reported in a separate paper), this author concurs that prediction of solar eclipses is beyond the ability of anyone during a single lifetime of unaided observation. (30, 31) Prediction of lunar eclipses, on the other hand is quite another matter. Thales may well have developed a heuristic method for prediction of lunar events. Working with eclipse cycles of 177 days Thales could have predicted that a lunar eclipse would occur during full moon of the next cycle, thus setting not only the period during which it occurred but the calendar date of the eclipse within a day. But there were no striking eclipses of the moon in 585.
If 585 is no longer acceptable as the date for the phenomenon Thales predicted, a search for the right one seems incumbent upon us. The period 615 BCE to 596 marks the most certain overlapping reigns of Kyaxares the Mede and the Lydian Alyattes. During this period, we find three candidates for total lunar eclipses that might have taken place during campaign season (eliminating eclipses in December, January, and February) : 1) 16 November 613 BC: a twilight eclipse; the moon rose already eclipsed; 2) 3 September 609: a twilight eclipse (invisible eclipse, over before local moonrise) ( 12 ), the moon rising eclipsed and remaining obscured through twilight until 10:30PM local time; 3) 4 July 606: late evening eclipse, moon rising uneclipsed near sunset at 8PM and beginning to be eclipsed near the end of twilight at 10PM and reaching total eclipse at 11:30PM.; These evening events (7/4 606, 9/3 609 and 11/16 613) seem best to fit the circumstances seemingly set forth by Herodotos for the night battle. Two also occurred during prime campaign season, And one of them ( 9/3 609) provides an exact match with scenario for a night battle developed above. Some may want to extend the time frame (615- 585) and so to include an event of 4 July 587: a twilight eclipse; the moon rose eclipsed with sunset at 8:10PM and remaining eclipsed for one hour to 9:07, reemerging to become fully illuminated again by 10:30.
In any case, armies fighting at twilight, expecting the full moon to light their continuing onslaught were astounded not to see the moonrise as expected during the half-hour after sunset. As the bright sky faded, the growing darkness of night that had been delayed on the previous three or four nights with its fully splendid presence revealed the moon's face occulted by a shadow. The men, we imagine, were in a panic and the soothsayers non-plussed. Their fearsome report led the generals to abort their plan to continue the battle into the night. On that occasion, indeed, ignorant armies clashed by night. Herodotos, it seems, re ported a battle from annals for the summer of 587 shrouded in omens; the report was compiled from accurate sources, but Herodotos did not fully understand the nature of the omen. His confused report led to a specious date determined from solar eclipse records known in antiquity, and extended to Thales powers he could not have had. Thales' skill with eclipses became legendary thereafter, but it was not mythical, as some have alleged, merely misunderstood.
FOOT NOTES 05. Cambridge Ancient History , 2nd ed., volume 4, p.17.
06. Cicero De div. 1.49.112, Eusebios Chron. 2, (p. 187 Karst, p. 101 Helm) and Solinus (15.16).
11. tes maches sunesteoses (1.74.2).
14. Author's personal experience at Cabo San Lucas, Baja California, 11 July 1991.
25. So argues Neugebauer, 1952, 142 and following.
28. Herold, as reported in Hude's apparatus.
29. Otto Neugebauer, The exact sciences in Antiquity (Princeton, 1952), 142 ff.
30.See the author's 'Thales' Method for Predicting Lunar Eclipses', forthcoming.
31. Mosshammer (see note 3), 155.
Thomas Worthen
e-mail: tomwr@ccit.arizona.eduElectronic Antiquity Vol. 3 Issue 7 - May 1997 edited by Peter Toohey and Ian Worthington antiquity-editor@classics.utas.edu.au ISSN 1320-3606