D. S. Nivison, Paper for Oct 1997 AOS-WB meeting, Boulder CO
 

"The Riddle of the Bamboo Annals"
 

     In late 1995 the State Council of the PRC launched a giant project to "research and determine the relative chronology and absolute dates" for the Xia, Shang and Zhou periods, a chronology that remains in dispute prior to 841 BC.  The official project description states that "this is an interdisciplinary project which encompasses the fields of history, archaeology, astronomy, dating technology, etc.  Therefore it is necessary for the experts in those fields to be open-minded and to co-operate with each other..."  Tasks are enumerated and scholars are named to be in charge of each, dynasty by dynasty.  Pertinent texts are to be reexamined, including the Bamboo Annals.  Astronomical phenomena are to be studied: conjunctions in relation to dynastic change; the eclipse in the Zhong Kang reign of Xia mentioned in the Shang shu; astronomical phenomena at the time of the Zhou conquest of Shang; the "double dawn" in the first year of King Yih of Zhou. One focus is to be the annual sacrifice system of late Shang as recorded in oracle bones and bronzes; another focus will be the problem of the exact date of the Zhou conquest.  At one point, it is said that the project will seek "to ascertain the absolute reign dates of the kings of the Shang and Zhou from the reign of King Wu Ding on"; but later this goal is stated less ambitiously, as "to ascertain relatively accurate reign dates for the kings before the first year of the Gong He Era" (841 BC).  For earlier than King Wu Ding of Shang (the 22nd king, by most counts), the hope is only for a "relatively detailed framework," or a "basic framework."  The whole project is to wind up with conferences and publications late in 1999.  I have placed in your hands copies of the article in the spring edition of Early China News containing the official project statement as translated by Prof. Sarah Allan of Dartmouth College.
 

     Early in 1995, on the basis of my research extending over 15

years, I drafted a book to which I gave the title I have given

this paper: The Riddle of the Bamboo Annals -- claiming, as you

have guessed, that I have solved this "riddle."  The next year

and a half required me to give most of my attention to ancient

Chinese philosophy; but early in 1997 I went back to The Riddle

and revised and corrected it thoroughly, copyrighting it.  Some

copies have been distributed, and I am putting parts on the

internet, asking only that due acknowledgement be given if it is

used.  I have not yet closed with a publisher, but will probably

publish it first in Chinese.
 

     My book -- I claim -- solves all of the chronological

problems addressed by the Chinese project, and more.  I have an

exact chronology of reigns from 1189 BC (my death date for Wu

Ding) to 841 BC (year 1 of Gong He).  I also put forward an exact

chronology for earlier reigns , back to 2037 BC, encompassing all

of Xia and Shang, and more.  (But I grant that I am uncertain

about my dating of middle Shang reigns, from 1474 BC, the

succession year of King Tai Wu, to 1250, the succession year of

King Wu Ding.)  I have given you copies, also, of this complete

chronology, without supporting argument.  Some of that argument I

want to give you now.
 

     My strategy has been to use the reign lengths and dates in

the Bamboo Annals as the basis of analysis.  In doing this, I

accept E. Shaughnessy's proof (HJAS 1986) that the "present text"

(jin ben) is essentially the text as given by the editors of the

text disenterred ca. 280 AD, while rejecting Shaughnessy's claim

that those editors modified what they found so as (probably) to

make it unusable for research on pre-Zhou dating; i.e., I hold

that the text (with inessential editing) is the text as buried

ca. 299 BC in the Wei state.  I then use D. Pankenier's research

on three dynasty-heralding conjunctions, dated February 1953 BC,

for Xia; December 1576 BC, for Shang; and May, 1059 BC, for Zhou.

He convincingly ties these to references or other events dated in

the Annals to 2029 BC, 1580 BC, and 1071 BC.  I then show how to

explain the discrepancies, and this enables me to begin

correcting other dates in the Annals.  As I proceed, I verify my

deductions by other data, both matter from the Annals and matter

from inscriptions and other early texts, that is (usually)

directly or indirectly astronomical.  Here are some examples:
 

     Xia:  In 1984 Pankenier showed (EC 9-10) that an obscure

text in the Mozi is a reference to a conjunction of the planets

in Pegasus in February, 1953; and through close analysis of the

Annals he shows that it marks the transfer of power from Shun to

Yu in the 14th year of Shun.  In an earlier argument (1983), he

had shown that the last year of Xia must be 1555 BC (for various

early texts say that Shang lasted 496 years; and the first de

jure year of Zhou ought to be the year after the conjunction of

May, 1059).  I then use Annals reign lengths for Xia, positing

two-year gaps between reigns for completion of mourning for the

royal predecessor (the Annals indicates gaps of varying lengths,

usually two years); i.e., the year of a king's death would count

as the first of the required three years (strictly, 25 months).
 

     Three things emerge:
 

(1)  The 5th year of the 4th Xia king Zhong Kang turns out to be

1876 BC, and in that year the Annals records a solar eclipse

(also mentioned in the Shang shu and the Zuo zhuan) in the 9th

month, necessarily the first of the month.  As my colleague Kevin

Pang has discovered, there was a ring-form eclipse in the morning

of 16 October 1876, the 1st of the Xia 9th month, totality close

enough to the Xia capital so that it would have been reported.
 

(2)  The year 1555 turns out to be not the last year of Di Gui

(Jie), the reputed last Xia king, but the last year of Fa, who

supposedly was the next to last king.  I tentatively concluded

that Jie is an imaginary addition to Chinese history, and I

quickly found confirmation of this guess.  Granet, a generation

ago, had noticed that traditional accounts of Jie seem to be

based on accounts of the last Shang ruler Zhou Xin.  The Annals

record for Jie can be counted off into exactly 8 slips' worth of

text, indicating that it was added later (perhaps late 5th

century BC); other Xia kings have accounts of irregular length.
 

(3)  The first year (after mourning) of the 14th Xia king Kong

Jia turns out to be 1577.  As I showed in a paper for our Society

in Portland Oregon in 1994, the first day of that year in the Xia

calendar (starting the year with the pre-spring-equinox month)

was jiazi, first day in the 60-day cycle -- explaining the king's

name: it must mean "Great Jia."  (The preceding king, Jin, had

another name, "Yin Jia," i.e., "Next Jia" (or "Succession Jia"?);

and his succession date begins with jiaxu, day 11, the second jia

day in the cycle.  And the (actual) last king, Fa, had a reign

that began on a gui day; so apparently he is the true "Di Gui.")
 

     Next, Shang (Yin, if you prefer):  Pankenier's first year of

Shang 1554, with the Annals' last year of Tang, the Shang

founder, imply that the first year of his successor (and

grandson) Tai Jia ought to be 1542; and analysis of the day date

for a sacrifice to Tang in that year, given in the "Yi Xun"

chapter of the Shang shu as quoted in the Han shu (so I am not

leaning on a spurious text) confirms that the year was 1542.  But

1542 did not begin with a jia day (we now ought to use the Shang

calendar, beginning the year with the post-solstice month; but

the rule is not strict).  1542 actually began with a ren day; the

next year began with a bing day; and one waits until 1539 for a

jia day as first day.  What was going on?
 

    Yi Yin, Tang's prime minister, traditionally said to be a man

of selfless virtue, was trying to make himself king, taking

advantage of the forced inactivity of the legitimate heir during

mourning.  The Annals says he actually did make himself king.

(Pious commentators avert their eyes in horror.)  Also -- with

most chronologies -- the Annals has two kings preceding Tai Jia:

Wai Bing, 2 years; and Zhong Ren, 4 years.  Mencius agrees, and

says that Yi Yin temporarily deposed Tai Jia in order to "reform"

him.  I would argue that what actually happened was this:  After

Tang died, Yi Yin was in conrol.  first, in 1542, he named Tai

Jia's uncle Zhong Ren as stand-in functioning king while Tai Jia

observed mourning.  Then he went further, and exiled Tai Jia in

1541, putting another uncle or brother (and puppet), Wai Bing, in

his place as mourning king, for two years.  In 1539, mourning

completed, Wai Bing was dispensible, and Zhong Ren's proper reign

began.  In four years, 1536, Tai Jia escaped from detention (his

7th year, according to the Annals -- as indeed it actually was);

he then returned to court and killed Yi Yin.  Tai Jia's 12 years,

given him in the Annals, actually began (de jure, after mourning)

with 1539 (which gave him his jia name) and ended in 1528.
 

     I must now review the rules that one can infer from Shang

material:  No two successive reigns have the same gan name (thus

these kings' names function as nianhao).  If a king's succession

year would dictate the same gan, the accession (post-mourning)

year determines the gan instead.  Also, gui as gan is taboo.  So

whenever the rules would call for gui, the next day, jia, is used

instead.  Further, in Shang the inter-reign interval for mourning

is usually 3 years rather than 2; it was not that mourning was

longer; probably it was simply that the deceasing king's year of

death counted as in his calendar only if he lived through most of

it.  These rules neatly explain the names of the next four kings

(and in fact all Shang kings' names).  Wo Ding, 3 + 19 years (not

19 as in the Annals: 1527 began with a jia, avoided; 1524 began

with a ding.  Xiao Geng, 3 + 5 (not 5): 1505 began with a ding,

avoided; 1502 began with a geng.  Xiao Jia, 3 + 17 (not 17): 1497

began with a geng, avoided; 1494 began with a gui, resolving to

jia.  The next king according to received chronologies was Yong

Ji, followed by Tai Wu; but oracle inscriptions reveal that

actually Tai Wu came first.  The Annals gives him an impossible

75 years; I think his reign was probably actually 3 + 29: 1477

began with a jia, avoided; 1474 (taking the year as beginning

with the first month of winter) began with a wu.  The reason why

Yong Ji was traditionally made to precede Tai Wu seems to be that

it was thought that his reign, in the fifth generation after the

founder, must (like King Mu of Zhou) begin exactly a century

after the de jure first year of the dynasty, i.e., in 1475.  The

experts in the late fourth century BC recognized mourning

intervals for Xia and pre-Xia eras, but not for Shang and Zhou.

When these intervals were edited out, a 12-year gap was created

before Tai Wu, and Yong Ji was moved back to fill it.
 

     Jumping ahead:  From 8th king Tai Wu to 22nd king Wu Ding my

reconstruction of chronology must rely on gan names (my theory is

new and needs criticism) and reign lengths in the Annals, which

cannot be completely correct; so one must work out something that

makes the minimum number of ad hoc assumptions.  Wu Ding's death

date I take to be 1189, partly on the basis of Wu Ding oracle

inscriptions that mention lunar eclipses; and I find that if he

had a reign of 3 + 59 years (rather than 59), his first year

(1250) began with a ding day.  From then on dating is complicated

by a radical change in the rules of succession.  Keightley's

analysis of oracle material shows that in each of the six

generations after Tai Jia there were two brother kings; and I

have assumed that the reason was the danger (dramatized by Yi

Yin) that during mourning an ambitious minister might usurp the

kingship.  To forestall this, a king would designate a brother as

functioning successor king while his own son and eventual heir

did the mourning; when the son succeeded, he would merely have to

mourn an uncle.  But in the generation beginning with Yang Jia,

there were four siblings in succession, probably because royal

brothers were attempting usurpation -- in the end successfully,

because Wu Ding in the next generation was the son of Xiao Yi,

the last of the four.  After that fraternal succession of the

original kind ceased.  Each king from Zu Jia on sought to keep

the succession in his own line by giving his son royal status

before his own death.  Close analysis of late Shang ritual cycle

inscriptions reveals that each of the last three kings apparently

gave his son a calendar that began before his own death -- this

being something that recorded history has missed entirely.
 

     I give now an example of how I fix dates of reigns for

Western Zhou.  First (as I showed in a paper for our Society six

years ago in Eugene, Oregon), by many independent proofs 18 April

1040 BC is the date of the Zhou conquest; most memorably, the

date in the Zhou calendar was the first day of the Qing Ming

solar period, the major festival in the ancestor cult (and in

1040 a jiazi day); this is echoed in the "Da Ming" Ode, in praise

of the Zhou royal ancestors, down to King Wu, who (as this Ode

ends) "then attacked Great Shang; this was in the morning, Qing

Ming [Day]."  King Wu died two years later, and the Zhou Gong

Regency was the first seven years of King Cheng's reign of 2 + 30

years (not, as usually held, the seven years preceding the 30

years).  The Annals does begin Cheng's 30 years with 1037,

getting the date right but omitting the mourning interval.  King

Kang's reign was 2 + 26 years (not the Annals' 26); and King

Zhao's reign was 2 + 19 years (not 19).  We should therefore

expect the Annals'first year for King Mu, 962, to be six years

too early; and bronze inscriptions of Mu's era show that in fact

the date was 956.
 

     I have given examples enough to demonstrate that there is a

way to recover exact dates back to the beginning of Xia and even

beyond:  One uses the reign lengths and dates in the Bamboo

Annals, and corrects them by using astronomy and inscriptions.

But this will work only if one also recognizes that normally a

reign length as given in the Annals was preceded by a mourning

interval.  There are mere vestiges of these intervals in the

Annals' chronicles for Xia, and in the Shang and Zhou chronicles

they have been deleted entirely, with resulting distortion in the

chronology.
 

     I see no indication in the Chinese scholars' plans for their

chronology project that they are cognizant of this problem; and

in fact they seem to have assumed that there is no hope for a

complete and exact chronology for history earlier than Wu Ding.

I would have to say that unless they recognize the impact of the

ancient three-years mourning custom on dating, they are going to

get almost nothing right before Gong He in 841 BC.
return to home page  nivison: Chronology, Huang Di through Western Zhou  Fully dated Western Zhou Bronze Inscriptions (1997)