James Legge believed that the Bamboo Annals was altered after its discovery in 279 or 280, sufficiently to render it useless for recovering exact dates. He made this mistake through his failure to read the text carefully enough to see that the intended beginning dates of Xia and of Zhou as shown in the dynastic summaries are implied in the chronicle, even though they are not what appear to be the beginning dates in the chronicle text. Professors Edward. L. Shaughnessy and David. W. Pankenier also believe that the text was rewritten after discovery, seemingly enough to make it useless for recovering exact dates (for Pankenier this is not quite true). Each has persuaded himself that the date of the Zhou conquest was different from the date(s) (1050 in the chronicle, 1051 in the Zhou summary) in the "modern text": for Pankenier it was 1046, for Shaughnessy it was 1045. Each thinks it obvious that the unaltered original recovered text must have had the correct date, so that if we only had it, his own view would be sustained.

I do not think that wishing something to be true is reason enough for believing it to be true, and I am sure these gentlemen would agree. They do have genuine reasons, which I have been unable to accept. It is my own view that the dates we find throughout the "modern text" are in almost all cases the dates that were in the recovered text (though I think that the "modern text" is not quite the recovered text, but is a copy of that text made before the work of reconstituting it was complete). In what follows, I will prove that the dating scheme for the conquest era in the "modern text" is the product of experts in the court of Xiang Wang of Wei (reign beginning 318 BCE), who were reworking an earlier chronicle that probably already had undergone politically motivated revision. (I think also that if we examine the "modern text" in this light, it is possible to recover from it an exact and complete chronology for kings in the entire "Three Dynasties"; but I have made my attempt to do this elsewhere, in Sino-Platonic Papers 93 (1999) 1-68: "The Key to the Chronology of the Three Dynasties: The "Modern Text" Bamboo Annals".)

A fragment of the lost Di wang shi ji (Chronicle of Generations of Emperors and Kings) by Huangfu Mi (215-282) reads as follows:

"In the 4th year of Wen Wang's receiving the Mandate, on the 1st day, bingzi (13) of of the Zhou 1st month, the Kun Yi attacked Zhou, and thrice within the day got as far as the Eastern Gate of the Zhou [capital]. Wen Wang retired and cultivated his virtue, not engaging in combat with them."

(Quoted in the Mao commentary to Ode 167, "Cai Wei"; see Xu Wanyuan, Di wang shi ji jicun, Beijing: Zhonghua Shuju, 1964, p. 85.)

That Huangfu Mi had a source (albeit an imperfect one) is shown by the following facts:

(1) The "Modern Text" Bamboo Annals (Jin wen zhushu jinian) says at Di Xin 34, = 1069, "In the winter, in the 12th month, the Kun Yi attacked Zhou."

(2) In the Annals the Mandate-conferring conjunction of the planets is dated Di Xin 32 = 1071, making 1070 the Mandate year 1, so that 1069 is Mandate year 2 (in effect, though not called that in the Annals).

(3) The conjunction actually occurred 12 years later than the date given in the Annals, i.e., in 1059, so that 1058 was really Mandate year 1, and 1057 was Mandate year 2.

(4) In the year 1057, the 1st of the post-winter-solstice month was day bingzi (13), 30 January.

(5) The post-winter-solstice month was, traditionally (as stated in the Shi ji) the first month in the Shang calendar; but at this time the Zhou court apparently was also using this calendar; see Nivison, "the Dates of Western Chou" (HJAS 43 (1983)), p. 518.

(6) According to Du Yu (222-284), "Zuo zhuan hou xu," the original Annals used the "Xia" calendar, taking the first month of spring, i.e., the post-post-winter-solstice month, as first month. (This appears to be true also of the "Modern Text.")

From (3), (4) and (5), the 1st day of the Zhou 1st month of 1057, Mandate year 2, was day bingzi (13). If the Kun Yi attack was (as in Huangfu Mi) "on the 1st day, bingzi (13), of the Zhou 1st month," then this must have been the day. But from (1), (2), (5) and (6) the Annals date for the Kun Yi attack was in effect the Zhou 1st month of 1068, which was Mandate year 3 in the "Modern Text."

Therefore two things need to be explained, if we take Huangfu Mi and the Annals seriously. First, why does Huangfu Mi call the year Mandate year 4; and second, why does the Annals date the attack one year later than its "Mandate year 2," i.e., only 11 years earlier, and not 12 years earlier, than what appears to have been the correct date, the 1st of the Zhou 1st month of Mandate 2. This day was bingzi (13), 30 January 1057. Given the Annals' back-dating of pre-conquest Zhou dates by 12 years, the Annals entry ought to have been the 12th month of Di Xin 33, 1070, i.e., the 1st Zhou month of 1069 (back 12 from 1057).

The probable solution to the first problem is simply that Huangfu Mi's source was not the Annals itself (for he died only two or three years after it was discovered), but some later Warring States text that must have been partly based on the Annals. One can suppose that this text counted mandate years not from the year after the conjunction, but from the conjunction year itself, and this does make 1068 "year 4" (counting from1071, the Annals' date for the conjunction, back 12 from the correct conjunction year 1059).

The second problem is much more interesting, much more difficult, and far more important.

The astrologer-historians in the court of Xiang Wang of Wei, who succeeded in 318 BCE and reigned either to 299 or to 296, were probably working with an earlier chronicle that had undergone revisions possibly in the Zhou court as much as century earlier; but it would still have had the conjunction correctly dated to 1059. If so, it would have had the Kun Yi attack correctly dated to the (Xia) 12th month of 1058, probably with the day correctly specified, as day bingzi (13).

These experts deformed the chronicle to make it glorify the Wei state, and specifically to make it support Xiang Wang's father Huicheng Wang in the latter's claim to the title wang. They did this by making the year the claim was announced, 335 BCE, be exactly 700 years after the event that was the beginning of the ancestor-state Jin, i.e., the appointment of Tang-shu Yu as lord of Tang, which they put in 1035.. This event was believed (probably correctly) to have been in a year when Jupiter was in station 10, Da Huo (actually 1031 BCE). Probably to support this move, they also appropriated for Wei the Zhou-heralding conjunction of 1059. This lead them to move the conjunction back 12 years, from 1059 to 1071, and to make this year a year when Jupiter was (so they represented) in station 10, Da Huo (rather than in station 6, Chun Shou, its actual location).

Consequently, the date of the Kun Yi attack had to be the first day of the Zhou first month of 1069, i.e., the first day of the Xia 12th month of 1070. But they deduced, correctly, that this day could not be a bingzi day. What could be done? They also would have deduced, incorrectly, that the corresponding day in the next year was a bingzi day. The solution to their problem, then was to shift the event down one year; and this, I argue, is what they did.

How can we know that this was the way they saw their problem and that this was the way they worked it out? To be reasoning in this way, they would have to be using some method for determining the first days of months, in any past year. The system (I have shown that it was used in other parts of the Annals) was the intercalation schema of 19-year zhang and 76-year bu, each bu being identified by the ganzhi of its first day. This system was not quite accurate. The mean error was one day early after the first two bu (152 years) of back calculation, and one more day early for every four bu (304 years) of back calculation before that. There were different applications of the system, with different "zero" points in time taken as the basis for calculation.

The system of absolute dates for zhang and bu that they used may not have been the one used in the so-called "Yin Li" calendar (here I follow Shinjo Shinzo, "Shu-sho no Nendai," Shinagaku 4.4 (1928) pp. 542-543), but the Yin Li will serve to illustrate what they were doing. In this system 313 BCE was the first year of a zhang, that began with day dingwei (44). (the system supposes that a zhang starts with a winter solstice month in which the winter solstice is on the first day.) Ten bu (40 zhang) earlier was the year 1073, supposed in this system to begin with day dingchou (14), but actually beginning with day gengchen (17). So the system was giving an investigator near the end of the 4th century BCE dates three days early, for the time when Xiang Wang's experts wanted to put the conjunction. 1069 BCE actually began with months beginning bingxu (23), yimao (52), yiyou (22), which the Yin Li would give as guiwei (20), renzi (49) renwu (19); no day bingzi (13) here. But 1068 began with gengchen (17), gengxu (47), jimao (16), and therefore one would expect that in a Yin Li calculation 1068 would begin with dingchou (14, dingwei (44), bingzi (13).

But it seems that it was the third month that began with bingzi. The ("Xia") 12th month of 1069 was the preceding month, was it not?

Perhaps not: The solstice day for 1068 was wushen (45), and at this time in the Shang and Zhou calendars the observed solstice day was 91 days after the autumn equinox, instead of the actual solstice day, only 89 days after. Thus the observed day was two days late, and would be gengxu (47) rather than wushen (45). So in the actual 1068 calendar in Zhou and Shang the first month of this year was the second month in Zhang Peiyu's Zhongguo xian-Qin shi li biao (which I have been using), i.e., the month that actually began with day gengxu (47), and for a later investigator using the Yin Li it therefore began with day dingwei (44). The solstice month is month 11 in the "Xia" calendar system; therefore the "12th" month was the month beginning with day jimao (16), read by the investigator as bingzi (13) -- just as my problem requires.

Here, the alert reader will have noticed a problem. We cannot suppose that the Yin Li system as used in the 4th century BCE somehow incorporated a "standard time" convention used in the 11th century BCE, i.e., of dating the solstice two days late, when that system was applied to determining 11th century dates. On the contrary, the system's error for the solstice was greater than its error for first days of months. E.g., while the actual first day of 1073 was three days later than the Yin Li day -- which was also the Yin Li solstice day -- the true solstice day was five days later, being day renwu (19); and so by 11th century convention it would be day jiashen (21). A Yin Li user was not simply ignoring the problem; what he was doing was using the Yin Li itself to tell him what the solstice day was, as is suggested in Mencius (4B 26).

But how, then, can we suppose that the user would have (mis-)identified as the post-solstice month of 1068 the month whose first day he (mis-)identified as bingzi? I see two alternatives:

One: We must suppose that in applying the zhang-bu system there was a specified articulation in detail for a bu and its four zhang, for identifying long (30-day) and short (29-day) months, intercalary months (seven in each zhang) and intercalary days (producing two long months in succession). We could work out this system and apply it to the zhang beginning with the year 1073, thus discovering what, probably, the year 1068 looked like to the investigator. Knowing that the investigator was looking for a month he could call the post-solstice month, beginning with a day he could call day bingzi, allowing for the possibility of a one-month and/or a one-day error at the time, we would certainly find a month and day that would satisfy him, enabling him to say "Aha! That's it!" I leave this as an exercise (for myself as well as others), because I think that another surprising possibility is more likely.

Two: Perhaps Xiang Wang's experts had at hand a table -- like Zhang Peiyu's -- that gave ganzhi for first days of months, year by year, back to and through the 11th century BCE, even with major events written in. They would, then, be using state-of-the-art science (the zhang-bu system) to check this table's accuracy. Finding, by using the system, that 1073 began not with the table's gengchen (17) but with dingchou (14), they then turned to the table for 1068 and revised it accordingly, reducing every day number by three. In the table (as argued above) the supposed post-solstice month of 1068 was the month beginning with jimao (16), which they "corrected" to bingzi (13). But is there any evidence that such tables existed?

Yes. Three centuries later, Liu Xin had a piece of one, which he called the "Chunqiu Li." It didn't continue far enough for Liu to correlate it with known dates, so he was not able read off from it its intended absolute dates, and of course (using his own calculations) he got these wrong. Nonetheless, his table told him, correctly, that the 12th month of Wen Wang's 42nd year began with day dingchou (14). (Unaware of this, Qian Daxin in the 18th century decided that this "Chunqiu Li" was Liu's own invention.)

Then there is the famous astrological text in Guo yu "Zhou Yu" 3.7. I have argued (in a communication to the American Oriental Society, annual meeting in 1992) that it is based on a record, not of celestial data, but of dated events in the Zhou conquest campaign beginning in January 1040 and concluding on 18 April, expressed in the "Xia" calendar. I argued that someone, in the middle of the 5th century, was applying the zhang-bu system to the tabular data before him, i.e., here too checking an old text by applying the latest science. This showed him that the Zhou victory on day jiazi (01) near the end of the 2nd month would have to have been on the 1st of the 3rd month. His calculation indicated day dates two days early, making him read jiazi as renxu (59), so that jiazi, which he knew to be the victory day and in the second month, appeared to be the first of the next month. So he concluded that the calendar must be the Zhou calendar with the winter solstice month as first month, and shifted all dates back two months, i.e., 59 days, with the result that the campaign, for him, started (as before) near the end of a month, but in late autumn. His invented astrology for the beginning of the campaign precisely reflects this conclusion. If I am right, he must have been using a table like Liu Xin's, perhaps the very same one. But for his account of the conquest Liu used the Guo yu account, trying to reconcile it with his own astronomy. That Guo yu account had put Jupiter in Chun Huo at the start of the campaign, because in the early-middle 5th century BCE someone counting back by Jupiter's supposed 12-year cycle from an observation at the time would have concluded, wrongly, that Jupiter was in Chun Huo in late 1041.

Notice, that this one-year down-dating of the Kun Yi attack was required as a result of the 12-year back-dating of the conjunction. So that back-dating operation on the original correct chronology must already have been done before the Kun Yi calculation I have just explained was done. Further, all of the "modern text" dates for the conquest era are determined by the 12-year back-dating of the conjunction and its relocation from Chun Shou to Da Huo.

This needs demonstration. The rewriting of the text must have been done in Wei in the late 4th century BCE, with the object of showing that the appointment of Tang-shu Yu, and the earlier conjunction, were in years when Jupiter was in station 10, Da Huo. The persons doing it had before them material that dated and located the conjunction correctly, in 1059, Di Xin 28, and in Jupiter station 6, Chun Shou. The first move was to identify 1059 as a Da Huo year, making 1063 the Chun Shou year, still Di Xin 28. This pushed all earlier Shang first-year dates back four years, and identified 1050 as a Chun Huo (station 7) year, so that it, as in the Bamboo Annals, must be the year of the conquest -- following the invented account we now have in the Guo yu.. But this would appear to be impossible, because after the conjunction, now in Da Huo but still in 1059 (become Di Xin 32), there had to be at least a nine-year "Mandate" reign of Wen Wang before his death (he actually died in 1050). The remedy, necessarily, was to suppose that the Da Huo conjunction year Di Xin 32 was one Jupiter-cycle earlier, i.e., 12 years earlier, in 1071. This moved Shang dates back another 12 years, so that Di Xin's first year, actually 1086, was put a total of (4 + 12 =) 16 years earlier, 1102, as in the Annals. But since the conjunction was tied to Zhou, Wen Wang's death became 1062 instead of 1050, and other earlier and later pre-conquest Zhou dates were accordingly shifted back 12 years.

Furthermore, post-conquest Zhou chronology does not allow the conquest to be in 1050 without also a lengthening of Wu Wang's post-conquest reign by three years. So the relocating of the three-years slip's worth of text that Shaughnessy shows to have been shifted from a hole in the chronicle for Cheng Wang back to the end of Wu Wang's chronicle must be a feature of the text as perfected ca. 310 BCE. The slip move might have been done at this time, or (more likely, I think) much earlier, in a doctored version of the chronology done in Zhou or Lu. (It was not done, as Shaughnessy thinks, by the Jin court restorers). There was such a text, as shown by features of the Annals that boost the prestige of Zhou and of Zhou Gong; and one can reasonably suppose that Xiang Wang's experts used everything.

So all of the dates that Shaughnessy and Pankenier want to see as the work of Six Dynasties restorers must really have been worked out before the text was buried. Let them not suggest that the re-dating of the KunYi attack might have been done as the text was being restored, ca. 280 CE. By that time, the Yin Li system yielded day dates three or four days late; so its inaccuracy would be obvious. We would have to suppose that the Jin court scholars were consciously and cleverly trying to fake a mistake that might have been made in Warring States China; and it is not reasonable to ascribe such a motive to them.