|<Previous Section>||<Next Section>|
Appendix V. Solar Eclipses during the Former Han Period
The results derived from an examination of the eclipses reported in the HS are here summarized. During this period of 230 years (including the reign of Wang Mang) there occurred 559 solar eclipses, of which 98 or 17.5% were visible in some part of China. Fifteen of these were very small eclipses or were invisible in the capital, so that we should not expect Chinese astronomers to have noticed them. Of the remaining 83 solar eclipses, 55 or almost two-thirds seem to have been recorded by Chinese astronomers. In order to examine this unusually good record, I present here a list of the solar eclipses visible in China from 206 B.C. to A.D. 23, with my conclusions. 1
As sources for these eclipses recordings, there is (1) the SC, which contains very few of these recordings, so that its value is almost negligible. (2) The "Annals" and ch. 99 of the HS constitute a second source, giving only the dates of these eclipses. (3) HS ch. 27, "The Treatise on the Five Powers," near the end, contains a list of solar eclipses, which partly duplicates the information contained in the "Annals" and partly amplifies it by noting a few additional solar eclipses, giving the heavenly location of many eclipses, sometimes other information, and often presenting their astrological interpretation. This list ends with the reign of Emperor P'ing, so that the eclipses during A.D. 6 to 23 are only mentioned in ch. 99 of the HS. (4) The Han-chi seems to be an entirely secondary source, and has copied its dates from the HS Annals. Since however it was written in the second century A.D., it constitutes an early check upon the text of the HS, and very occasionally furnishes additional information. The information found in these four sources is discussed in the appendixes of the various chapters concerning eclipses, and need not be repeated here.
SOLAR ECLIPSES VISIBLE IN CHINA 206 B.C. to A.D. 23
206 B.C., July 6. Not noticed.
From a study of the above material, there emerge several interesting conclusions:
1. The records of eclipses in Former Han times are predominatingly reliable. More than two-thirds of the recorded eclipses, some 38 in all, are recorded correctly in the present text. Considering the length of time since the HS was written in the first century A.D., and the many opportunities for mistakes, both by astronomers and annalists before the HS was compiled and the opportunities for errors in transmitting the HS text, this is an excellent record. Fourteen other eclipses can be fitted into the actual dates, usually by only slight changes in the text. Only at most three recordings are hopelessly erroneous; two of these are due to errors in the transmission of the data. When we consider how very easy it is to write mistakenly the number of a month or the cyclical day, the essential correctness of the HS is a marked evidence of the care that was exercised in compiling it and in preserving and copying faithfully its text.
It does not seem to have been the case in Han times, as was sometimes later the case, that the dates given in the history were those on which the emperor was informed of an event, rather than the dates on which events occurred. The eclipse of 15 B.C. is said to have been reported from outside the capital and to have been invisible in the capital, so that some days must have elapsed between its occurrence and its being reported to the throne. But it is dated correctly.
2. In the case of some eclipses that are not listed correctly, it is plain that the errors of dating occurred before the HS was compiled. In other cases the evidence is not so clear, but it looks as if the same thing happened. The description of the eclipse of 35 B.C. as being large and ending at sunset is such that the actual eclipse can be unmistakeably identified by calculation. In the chronological account of events in the Annals, this eclipse is put more than a year later than it actually happened; between the actual date and the date at which the Annals put it, several events are recorded. Hence it is plain that when Pan Ku composed the "Annals," this eclipse was already misdated. In the eclipses of 192 and 141, it also seems very likely that the error of dating occurred before Pan Ku prepared his HS. He or his sources probably had somewhat illegible astronomical records of eclipses, which were easily misread. As a whole, there seem indeed to have been made more errors of dating before the HS was written than have occurred in the transmission of the text since that time.
In other cases, minor changes have plainly been made since the HS was composed. With the eclipse of 192, there seems to have been later conflation between the account in the "Annals" and that in the "Treatise." A later correction of a date has possibly been made in connection with the eclipse of A.D. 1. For the eclipse of 154, three different cyclical dates are given by our three sources.
Dittography in Pan Ku's sources seems to have been responsible for the eclipse listed for 153, and almost certainly for that listed for 148. The reign of Emperor Ching (157-141 B.C.) was the worst period in accuracy of eclipse recordings. During this period there are five eclipses incorrectly dated, two dittographies, and only four correctly recorded eclipses. The "Annals" for this period are also the least satisfactory of all the "Annals" in the HS. The original sources for this reign seem to have been defective.
3. Ch'ang-an, the imperial capital, was not the only place from which eclipses were observed. The remarks in the HS in connection with the eclipse of 15 B.C. establish by direct testimony the fact that eclipses were reported from outside the capital. The eclipses of 136, 135, 80, and 56 must also have been reported from outside the capital. It is hence dangerous to take the capital as the sole point of observation, as Hoang and others have done. Nothing less than the whole of China should be taken as the possible locus of observation. For Han times, there is however some evidence that eclipses were not reported to the capital from central and southern China; the eclipse of 2 B.C. was total in the present southern Szechuan, southern Hunan, and Foochow, but no statement of its totality got into the HS.
4. During long periods of years, all eclipses plainly visible in China (and some quite small ones) are reported, while during other periods, groups of eclipses are missed. The following table indicates the number of eclipses not mentioned in the HS and the number of years after the last unreported and plainly visible eclipse during which all plainly visible eclipses were reported:
It is especially remarkable that there were two periods of over half a century during which all plainly visible eclipses were recorded---from 166 to 115 B.C. and from 35 B.C. to the first eclipse correctly recorded in Later Han times (A.D. 26). What could have been the reason for such periods of complete success followed by periods of failure? For example, the five plainly visible eclipses between 174 and 166 B.C. were all missed, then all plainly visible eclipses during 51 years down to 115 B.C. were recorded.
Bad weather could hardly be the reason; eclipses were reported from outside the capital, so that local bad weather would hardly prevent an eclipse visible to the naked eye from being recorded. There is however a small correlation between the number of eclipses missed and the time of the year:
The missed eclipses seem to be concentrated during the summer months, especially May and July, more than half coming during four consecutive months. Yet the correlation between the number of eclipses missed in each month and the number recorded is only -0.34, with a probable error of +/- 0.17. There is thus only a very moderate degree of association between the number missed and the number recorded, and the number of instances is too low to make the correlation reliable. The correlation between the number of eclipses missed and the average number of overcast days is -0.39 +/- 0.17, a negligible amount. It however looks very much as if the interest taken in the observation of solar eclipses by the responsible observers determined whether eclipses were or were not observed. To be conspicuous (so that an eclipse could not be missed by ordinary people) an eclipse must reach a magnitude of 0.75 when the sun is high or a magnitude of 0.33 when the sun is near the horizon. Many of the eclipses recorded in the HS are much smaller than these magnitudes, so that they could not have been perceived unless astronomers were expecting them and used special means to observe them. Hence the attitude of the responsible observers or astronomers probably had much to do with the observation or non-observation of solar eclipses.
5. The Chinese used special means to observe eclipses and watched for them during the two or three days at the end and beginning of the Chinese months when solar eclipses were to be expected. Unless an eclipse is conspicuous, it is easily missed; under other circumstances, the diminution of light is too small to be perceived. Unless the sun is near the horizon or is covered by light clouds, it cannot moreover be watched with naked eyes. The necessary diminution of light can be obtained by looking through smoked mica or at the reflection of the sun in water or in a mirror. (Smoked glass may have been available at the time, mica was actually used as a screen, cf. 99 B: 16b.) Some of the recorded eclipses were so small that such special means were certainly used by Chinese astronomers. The eclipse of 178 only reached a magnitude of 0.20; that of 68 only a magnitude of 0.10; that of 16, only 0.08; that of 12, only a magnitude of 0.07 lasting only 67 minutes; that of 1 B.C., only a magnitude of 0.06. That such small eclipses were recorded demonstrates that eclipses were expected and watched for, using special means. There was an office at the capital entitled the Office for Watching the Heavens (cf. Glossary, sub voce), the members of which probably scrutinized the sun for eclipses. Probably this practise of watching for eclipses was however confined to the capital; the eclipse of 16 B.C. was not reported from outside the capital, although at the present Peiping it reached a magnitude of almost twice that at the capital. Thus astronomical activity reached high levels at the capital and was largely confined there.
6. As is to be expected, the method of calculation given by Neugebauer shows itself remarkably accurate when its results are compared with Chinese records. In the eclipse of 181, Chinese records show that at Ch'ang-an the eclipse was total; calculation by Neugebauer's tables reaches that result, but Oppolzer and Ginzel both reach a different result. For the eclipse of 89, Chinese records give its time, late afternoon, which also checks, to a reasonable degree of accuracy, with the results of calculation by Neugebauer's tables. In this case, the eclipse seems merely not to have been noticed until some time after it began.
By Neugebauer's tables, the time of an eclipse may be determined to within fifteen minutes and the magnitude to within 0.03. 3 The eclipse of 136 may have actually been observed at the capital, although, according to Neugebauer's method of calculation, the capital was just outside the area of visibility.
7. The differences between the statements about eclipses in the "Annals" and in the "Treatise on the Five Powers" (HS ch. 27) indicate that the latter probably represents the list of eclipses kept by the astrologers at the capital, while the "Annals" also employ reports from outside the capital. In three cases, in the eclipses of 188, 80, and 28, the "Annals" state that the eclipse was total, while the "Treatise" says it was almost total. In each case, calculation demonstrates that at Ch'ang-an it was almost total, while at some other places in China it was total. Thus we can identify the "Treatise" with the reports of the astronomers at the capital.
Yet the astrologers at the capital sometimes utilized reports from elsewhere. The "Treatise" specifically mentions the observation of an eclipse in other parts of China, in connection with the eclipse of 15. It lists the eclipse of 135, which was not visible at the capital, as well as that of 136, which was close to the borderline of visibility there. The "Treatise" also records the eclipse of A.D. 2 as total, whereas calculation shows that at Ch'ang-an it was not quite total.
8. It is possible that the eclipse listed for 186 was a deliberate fabrication for the purpose of warning the ruler, the Empress of Emperor Kao, that Heaven disapproved of her acts. This listing seems hardly explicable by any other means. The eclipse discussed under the date 155 is also either a fabrication or else an actual eclipse dated a few years ahead. The eclipse of 201 is likewise doubtful. That only these three eclipses, all near the beginning of this period, should be doubtful, is a high testimony to the accuracy of Chinese recordings at this time. According to Chinese law, it was a serious and capital crime to report falsely a prodigy (such as an eclipse of the sun, cf. HS 100 A: 5b, to be translated in the Preliminary Volume of this series).
9. There is no evidence that the Chinese by calculation reached the dates of any eclipses recorded in the period. In the most promising cases, those of 201, 184, 155, and 145 B.C., I tried to calculate these eclipses by the methods that might have been used, and in each case reached negative results.
For the mistakenly dated eclipse of 201, I counted forwards from the date given for this eclipse to the end of the Former Han period by the Han eclipse period of 3986(2/3) days, by the Han chang of 6939 61/81 days, and by the saros of 6585.3 days, without coming upon any solar eclipses. Since it might have been possible for Chinese astronomers to know that there is a lunar period lasting one month, during which two solar eclipses and one lunar eclipse or two lunar eclipses and one solar eclipse may occur (although no evidence of this knowledge is found in Han records), I also took periods of one lunation from the dates found by this counting, but came upon no solar eclipse visible in China. This eclipse could thus hardly have been a calculation and its date must merely have been an error of some sort.
In the case of the eclipses listed in 186 and 145, I performed a similar calculation, with similar negative results. For the actual eclipse of 145 (not the mistaken date given in the "Annals," which does not correspond to any actual eclipse), I counted forwards and backwards for the whole Former Han period by the Chinese eclipse period of 3986(2/3) days. At each date thus found, an eclipse had occurred, but most of these eclipses were invisible in China. The second eclipse period after was the date of the eclipse of 123, the sixth was that of the eclipse of 80, the eleventh, that of 25, and the twelfth, that of 14 B.C. In addition, the eclipses of 167, 58, 47 B.C. and A.D. 19 occurred in this series, but these four eclipses are not recorded in the HS. The four recorded eclipses do not occur in any regular sequence, so that the probabilities are decidedly against this eclipse having been calculated by the Chinese eclipse period. I made a similar calculation, using the Han chang of 6939 61/81 days, but no recorded eclipse was found, only the missed eclipse of 164 and two invisible eclipses. Calculation by the saros of 6585.3 days, used by Greek astronomers, brought the eclipses of 181, 145, 127, and 1 B.C., the missed eclipse of 73, together with invisible eclipses. The saros does not however seem to have been known to the ancient Chinese. Thus even calculation of a difficultly visible eclipse from the actual date of that eclipse does not bring any results that would seem to encourage the computation of eclipses or to make probable the hypothesis that they were computed by the ancient Chinese.
It is however interesting that the first eclipse listed for Later Han times, dated on Feb. 16, A.D. 25, was almost certainly calculated. It is not found in the list of eclipses in the relevant "Treatise" of the History of the Later Han Dynasty, which was compiled by Szu-ma Piao, who lived ca. 240-306. This eclipse is listed in the Ku-chin Chu, written about A.D. 300. Calculation shows that this partial eclipse was invisible in Asia, and was confined to Alaska, western North America, and the eastern Pacific Ocean (cf. p. 509, n. 3). This date cannot thus have been the result of any Chinese observation; since the Chinese listing agrees with the date of the actual but invisible eclipse, it must have been calculated. Both Hoang and Chu Wen-hsin however include it in their lists of Chinese eclipses. This eclipse shows how dangerous it is to accept an eclipse listing without scrutiny.
Perhaps the most peculiar eclipse is that of 145 B.C., which was visible only at sunrise at the tip of the Shantung peninsula. I have pointed out that for it calculation was quite unlikely; since it was visible to the naked eye, it was probably observed and reported to the capital.
10. The court astronomers seem to have kept lists of lunar eclipses as well as of solar eclipses, although no such lists have come down to us. One such lunar eclipse got into the SC (that of 178), probably because Szu-ma Ch'ien or his source misread the word for "lunar" as "solar" in his source. Eclipses of the moon were not believed to portend anything important, so were not recorded in the histories.
11. Where the dates in the Chinese text do not correspond to any actual eclipse, slight emendations nearly always enable us to show what was the original date. The eclipse of 35 B.C. shows that such emendations actually reach the original dating. In this case, the eclipse is described in such a way that there is no doubt which eclipse is denoted by the recording. To reach the correct date, it is necessary to change a 5 in the year to a 4, a 6 in the month to a 9, and a jen-shen 壬申 in the date to a ting-ch'ou 丁丑, all of which are mistakes easily made in copying records. These mistakes occurred before Pan Ku compiled his History.
In other cases, slighter or greater changes are necessary. They have been noted in the appendices dealing with those eclipses. Altogether the dates of thirteen eclipses were rectified with a considerable degree of probability; those of 192, 176, 161, 149, 145, 141, 136, 135, 123, 108, 35, 1 B.C., and A.D. 6. The eclipses of 201, 151, and 155 also present difficulties.
12. The calendars of correspondences between Chinese and European dates are essentially correct. Hoang, who prepared the best of the calendars dealing with this period, based his tables largely on previous Chinese studies and partly on the correctly recorded eclipses, as is quite proper. He also published a table of Chinese eclipses, indicating the correspondence between astronomically calculated and Chinese dates, which he took as the basis for his calendar. In those eclipses for which I have been able to identify the original of a now incorrect date, and in which the correct date was unknown to Hoang, it is possible to test Hoang's calendar. In every case it is shown to be reliable (within close limits) for the Han period.
The most interesting case is the eclipse of 96, in which a change in the intercalary month from one year to the next (which is required by a contemporary calendar for this year discovered by Stein in the desert; cf. Chavannes, Documents chinois, p. 71) furnishes, unknown to Hoang, a date from his calendar correct within one day to that obtained from astronomical calculation. Thus modern science justifies Chinese calendrical calculations.
There are however slight discrepancies. These never amount to more than three days (except for the intercalary month mentioned above), so that the essential accuracy of Hoang's table is maintained. More recent calendars are no better; concerning the eclipse A.D. 6, which is misdated in the text by two months, and is said to have occurred on the first day of the month, both Hoang and Chen Yuan's Comparative Daily Calendar (Chung-hsi-hui Jih-li) equate the actual date with the last day of the month, so that both these calendars seem to be a day in error here. Because of inaccuracies in Han calculations, it is quite possible that months which our present calendars calculate as having 29 days actually had 30 days, so that modern calendrical computations may in some years be a few days in error. Such seems actually to have been the case. Hoang's calendar requires minor reworking.
13. In 37 cases, the position of the sun at the time of a solar eclipse is given. By the use of Neugebauer's Sterntafeln, the positions of the stars mentioned in the Chinese sources have been calculated for the date of these eclipses. The comparison of these records with the calculated position of the sun does not furnish any convincing proof that these positions were the result of any observation. In partial eclipses, the stars could not have been observed; almost all of these eclipses were observed as partial. For the total eclipse of 181, when the stars could have been observed, the position is 14° in error. For erroneously dated eclipses, the position, if given, is grossly in error, and corresponds more to the (erroneous) date than to the actual position of the sun at the time of the actual eclipse. For the five eclipses whose dates are corrected and in which positions are given, the errors in the dates are 3 yrs. 9 mo., 3 yrs. 4 mo., 1 yr. 7 mo., 10 mo., and 1 yr. 8 mo., while the errors in the position are 103°, 130°, 66°, and 105° respectively. Towards the end of the Former Han period, these positions are no longer given exactly in degrees, only the constellation being given. It looks as though someone had calculated the position of the sun from the date in the calendar at some time towards the end of the Former Han period, possibly about 27 B.C. (Liu Hsiang?), at which time the dates of many eclipses were already in error.
Yet more of the positions are fairly correct. Of the 37 eclipses for which positions are given, 24 are correct to 8° or less in R.A.; seven more are in error only 10 to 14°. (Since Chinese astronomical observations were always made with reference to the equator, not the ecliptic, celestial positions are reduced to right ascension for purposes of comparison.) The other six are grossly in error, from 42° to 105°. These are the eclipses of 201, 192, 176, 161, 141, and 80. The first five of these are incorrectly dated in the HS. For the others whose dates are correct, there possibly has been some corruption in the original names of the Chinese constellations.
These positions cannot have much significence and may be neglected. Their use in Han times was astrological; the various parts of the sky were taken to correspond to various localities on earth, hence an eclipse in a certain part of the sky indicated something in the corresponding locality on earth. It is possible that the heavenly location of the eclipse was falsified in order to make the eclipse interpret some earthly event. In however only six cases in the whole period are we given an interpretation of the eclipse in terms of its heavenly location. Much more common was an interpretation in terms of its month or day. Of the eclipses whose location is grossly erroneous, only those of 201 and 141 were interpreted in terms of their location. Hence it is not likely that many of the erroneous heavenly locations were deliberate falsifications.
In conclusion: The outstanding impression left by the Chinese recordings of eclipses in the Former Han period is their high degree of fidelity to fact. The Chinese were not to any great extent interested in fabricating eclipses as portents and it was dangerous to do so. They had not yet begun to predict eclipses. They watched for eclipses, at times with great pertinacity, and succeeded in observing eclipses that were quite small and required the use of special means to be seen. It is but natural that the original records should have suffered errors of transmission; as a whole they are surprisingly correct. This fact constitutes an unimpeachable testimony to the fidelity of the HS to fact. (Reproduced by permission and with alterations from Osiris, vol. 5 , pp. 499-522.)
1. The visibility of these eclipses has been computed by the method discussed in HFHD I: app. III. Both P. Hoang, Catalogue des éclipses de soleil et de lune relatées dans les documents chinois et collationées avec le Canon de Th. Ritter v. Oppolzer, "Variétés sinologiques," no. 56, and Chu Wen-hsin, Li-tai Jih-shih K'ao are not altogether reliable, since they do not discuss these eclipses in detail and sometimes list an eclipse that was invisible in China as the one referred to by the Chinese historian.In dating these eclipses, capital roman letters refer to the year of a reign or year-period, small roman letters to the (lunar) month, and arabic numbers to the day of the month. European dates are given by the Julian system, as in Oppolzer, but years are B.C., not astronomical years. (Hoang uses the Gregorian calendar for dates B.C., but the Julian for the first millennium A.D.) In references, arabic figures followed by a colon denote chapters of the HS, app. denotes appendices of those chapters, and small roman numbers particular eclipses in those appendices.
2. The data for overcast days are for the years 1924-1936 and have been very kindly reported to me by Fr. E. Gherzi, S.J., Director of the Siccawei Observatory.
3. A. Pogo, "Additions and Corrections to Oppolzer's Canon der Mondfinsternisse," the Astronomical Journal, 1938, no. 1083.
|<Previous Section>||<Next Section>|
|Published by The Institute for Advanced Technology in the Humanities, © Copyright 2003 by Anne Kinney and the University of Virginia|