The Project's final Newsletter contained an overview of the accretional theory of the Dau/Dv Jing. It documents our priority, and it gives the basic statistical argument. Here is an updated transcript.
Newsletter 13 (10 Mar 1999)
The Gwodyen MatterNewsletter 9 (1 Sept 1998) reported three ways in which the Gwodyen texts confirm our Warring States text chronology. We here explore the first, adding a statistical argument not raised previously.
Our claim is that the three florilegia comprising the Gwodyen Dau/Dv Jing (DDJ), which imply a truncated DDJ text in the early 03c, agree with our theory, made public in 1990 and published in 1994 and again in early 1998, which predicted a truncated DDJ in the early 03c. I here review (1) our published statements, which are the basis for a claim of prediction, (2) the nature of the find, and (3) its formal and statistical implications.
Our TheoryIn the early 1980's, we were persuaded by linguistic arguments that placed the DDJ in the middle 03c. It also weighed with us that the Laudz genealogy in SJ 63 would reach back to approximately that period, but not much earlier. Internal and external evidence seemed to agree. We also felt that DDJ denounced rvn and yi, not (as is usually thought) in opposition to the Mencius, but in opposition to Sywndz, in whose writings other concepts attacked in DDJ 19 also occur. Sywndz being early 03c, that relation compels a mid 03c date for the DDJ, assuming it to be an integral text.
Further study convinced us that DDJ was not integral, but rather accretional. One point was the DDJ relationship with the Analects, whose successively later chapters seem to be aware of an increasingly complete DDJ, implying a growth process in DDJ also. Such a growth process accounts naturally for the well-known DDJ shift from a contemplative stance (in DDJ 1-37, the "Dau Jing") to a purposive one (DDJ 38-81, the "Dv Jing"). We first communicated this finding at the 3rd (1990) meeting of the New England Symposium for Chinese Thought, hosted by Dartmouth and devoted to the DDJ (Bob Henricks's study of the Mawangdwei DDJ had appeared the previous year), in a brief lecture which made something of a sensation. An outline of the accretional DDJ theory was included in Brooks Prospects (SPP #46, 1994). We ignored the fact that our new theory placed DDJ 19 in c0324, before Sywndz was active or even alive, on the assumption that we must have been wrong, or the idea which we imagined DDJ 19 to be responding to was actually pre-Sywndzian, or something. Further Analects researches raised no difficulty with the new accretional DDJ theory, which was accordingly incorporated in the notes to The Original Analects (1998; TOA) at points where the Analects seems to be aware of the DDJ (one is LY *14:34, TOA p168, which rejects the "return good for evil" maxim of DDJ 63 and the earlier DDJ 49; see further the index to TOA sv Dau/Dv Jing). The idea of an accretional DDJ beginning in c0340 but running until c0250 also seemed to resolve the ambivalent evidence for an 04c vs an 03c DDJ, which was weighed in Henricks Manuscripts (Chinese Culture v20, 1979). Henricks there decided in favor of the 04c option. If, as our theory holds, the text got its stylistic set in the mid 04c but was influenced by 03c usages in its later layers, just such an ambivalence might easily arise. To that or any similar ambivalence, an accretion theory provides an attractive answer, since it allows all the evidence, not just a preferred portion of it, to be incorporated in the solution.
So far, so good.
Early rumor about the Gwodyen text find, however, was highly adverse. We were told in December 1996 that a report made at a Dauism conference earlier that year in Peking revealed that the newly discovered DDJ text contained most of the DDJ, and was from a tomb precisely datable to 0368. Since we had posited that the DDJ nucleus was no earlier than c0350, this result was wholly fatal to our view. Only fools would have persisted after that news. We did however persist, and TOA as published in February 1998 continued to presume the accretional theory of the DDJ. As with the Sywndzian reference in DDJ 19, we assumed that the seeming refutation would somehow work out.
It did work out. The claim of an 0368 date has been abandoned by those who advanced it. Recent Chinese archaeological opinion is that in terms of tomb architecture and furnishings, Gwodyen 1 stylistically postdates the Baushan tombs, one of which is internally dated at c0316. The usually cited Chinese date for Gwodyen 1 is now "before 0300," which is late enough not to refute our theory, though depending on the final date, our theory might have to be adjusted in small ways. [The terminus ante quem for Gwodyen 1 is 0278, the year in which Chu moved its capital elsewhere].
Public DiscussionIn May 1998, three months after publication of TOA, we learned in confidence from Bob Henricks the content and arrangement of the three Gwodyen DDJ florilegia. This was on the eve of the Dartmouth Laudz Conference, at which the facsimile volume of the Gwodyen manuscripts was to be released, and the contents of the texts finally made public. With Bob's agreement, I submitted to the Conference in absentia a short paper noting the fit of the Gwodyen DDJ data with our accretional theory. That paper was however not after all submitted to the Conference. The Conference instead only considered, and ended up deadlocked between, a Chinese theory of an integral 06c DDJ and an international theory of an integral but unwritten 03c DDJ. Archaeology Magazine (Nov/Dec 1998 p20f) also reported this deadlock as the final outcome of the Conference. (My attempt to introduce the third option to the Archaeology readers, in a letter published in Mar/Apr 1999 p10, failed through being reduced to inanity by the Archaeology editors). Though the attempt to introduce a third theory thus failed twice, the options which the Conference did consider, and between which they split, nevertheless do not stand. The 06c complete-DDJ theory fails because the Gwodyen florilegia ignore DDJ 67-81, whose governmental content should have interested a tutor of the Chu Heir Apparent. And the 03c amorphous-DDJ theory fails because the Gwodyen florilegia contain only future DDJ material, which is inexplicable if those maxims were, as the 03c theory holds, still part of an undifferentiated mass of oral wisdom, not yet crystallized into written form.
The official deadlock is thus between two untenable theories. Is there in fact a third theory?
The Gwodyen DataLet's examine it. The three Gwodyen florilegia and their contents are as follows:
A. DDJ 19 66 46BC 30AB 15 64B 37 63AC 2 32 25 5B 16A 64A 56 57 55 44 40 9 (total 20)
(in DDJ order: 2 5B 9 15 16A 19 25 30AB 32 37 40 44 46BC 55 56 57 63AC 64A 64B 66)
B. DDJ 59 48A 20A 13 41 52B 45 54 (total 8)
(in DDJ order: 13 20A 41 45 48A 52B 54 59)
C. DDJ 17 18 35 31BC 64B (total 5) plus the Tai-yi Shvng Shwei document
(in DDJ order: 17 18 31BC 35 64B)
The total is 33 passages, of which 32 are different (64B is present, with slight variation, in texts A and C). That one duplication precludes the inference that the Gwodyen materials constitute a single text of DDJ, but the existence of only one duplication raises the question of their independence. The chance that three random selections of these sizes from DDJ would so nearly avoid each other's choices is not tiny, but it is small (about 1 in 5 for B not duplicating A). Gil Mattos notes that except for Strips 5-18 of text A, all three are in the same hand, which also wrote the Tai-yi Shvng Shwei text (physically associated with DDJ text C), but that each of the three is on different length bamboo strips (personal communication, 9 Mar 1999). This looks like later copying, perhaps with the help of an assistant, of separate texts. The similar form of the three suggests mutual knowledge, with the repetition of 64B probably to be accounted for by local considerations (the variant in text A is slightly more "teacherish"). I will here consider them as separate selections, among which the later are aware of, and generally concerned not to duplicate, the earlier.
What DDJ Lies Behind the Gwodyen Florilegia?Discussion of the Gwodyen-implied DDJ has focused on its date, but the evidence for its character is more decisive. The integral theory holds that DDJ 1-81 existed, fully formed, behind the Gwodyen florilegia. Our accretion theory holds that the DDJ of that time may have been of no greater extent than the florilegia themselves attest. What then do the actual figures imply?
Their combined inventory is:
2 5B 9 13 15 16A 17 18 19 20A 25 30AB 31BC 32 35 37 40 41 44 45 46BC 48A 52B 54 55 56 57 59 63AC 64A 64B 66
Note that there is no gap greater than 5 between included DDJ chapters. For a sample of this size (33 selections) from a text consisting of DDJ 2-66 (the minimum suggested background text) of from the entire DDJ 1-81 (the maximum assumption), gaps of this size are not statistically remarkable. There is thus no strong implication that, for instance, DDJ 26-29 cannot have been present in the source text.
Can the Source Text be DDJ 1-81?
The case is far different with the unrepresented string of 15 chapters, DDJ 67-81, which on the integral theory were then part of the text, but which do not occur in the Gwodyen florilegia. The question is whether that long a sequence is likely to have been unrepresented in the three florilegia selected from the source text. Statistically, we ask: What are the odds that 33 draws from 81 items will include none of the last 15 items? This is the crux of the argument against a complete DDJ at this period. It may be helpful to work it out. For a step-by-step calculation, see the DDJ Problem page in the Statistics section. The following is a short version.
The odds that a first draw will be from the range 1-66 is 66/81, or 0.8148. Since each single text, and with one exception each successive text, observes a rule of nonduplication, the second draw will be without replacement, thus the odds of drawing a second item between 1 and 66 are 65/80, or 0.8125 (that is, slightly lower), and so on successively. The probability at each step is defined by a fraction whose numerator and denominator are each reduced by 1 from the previous step. Only on the 33rd draw does duplication occur, so that for this one step only we should not reduce the denominator, hence the final fraction is not 34/49 but 35/50.
Multiplying these fractions together (successive probabilities in a complex event multiply) gives us an overall probability of 0.0001354, or 1 chance in 7,384, that 33 consecutive draws from a text comprising DDJ 1-88 should represent only DDJ 1-66.
There is not a chance in a thousand, not a chance in seven thousand, that the DDJ text on which the Gwodyen florilegia drew was a full 81-chapter DDJ.
This is sufficiently unlikely that we may confidently reject the hypothesis that the background text contained all of DDJ 1-81. It was in all probability, as the accretion theory predicts, an incomplete DDJ. More specifically, it was a DDJ that was incomplete at its later end. The facts imply not just an incompleteness theory, but specifically an accretion theory.
Is The Date of Gwodyen 1 Consistent With The Accretion Theory?As noted above, the date of Gwodyen 1 must be later than Baushan 2 (c0316), and earlier than the abandonment of the site as the Chu capital in 0278. Li Xueqin, in an October 1998 lecture at Dartmouth (which we attended in person), suggested that the Heir Apparent to whom the Gwodyen 1 occupant was tutor was the later Kau-lye-wang (accession 0262). We don't dispute the plausibility of that argument. But note that the earliest date at which the future Kau-lye Wang could have been named Heir Apparent, and a tutor appointed for him, is the accession of his father Syang-wang, which was in 0298. An immediate appointment of a tutor followed by the sudden death of the tutor is perhaps an extreme supposition, hence 0298 may be too early a date for Gwodyen 1. On the other hand, there is a style change in the last Gwodyen tombs in which Gwodyen 1 does not participate, so that the years just before 0278 are archaeologically precluded. We thus have a possible range 0298/0278, of which both extremes are a little unlikely. The least risky conjecture, the one which minimizes possible error, will then be the midpoint: c0288.
What did our previously published theory predict about a DDJ in c0288? That theory is fairly precise as to the earlier chapters, but less so with the later ones. One reference we consider reasonably firm is the DDJ 80 line "make small the state, make few its people," which is an unlikely conquest agenda (earlier chapters speak of dominating the world from a position of weakness). It may instead be trying to put a favorable spin on the loss by Lu of half its territory, and half its population, as a result of the Chu attack of 0255/54. In general, extrapolating from the published version, our theory predicts that a DDJ text of c0288 would be likely to reflect anything in DDJ 1-55, and might include chapters up to DDJ 65 or so, but should not include anything from DDJ 70 onward. That is, a DDJ text sampled as of c0288 might, on our theory, consist of DDJ 1-65, more or less, depending on details such as rate of accretion and the exact date of Gwodyen 1.
The source text actually implied by the Gwodyen 1 florilegia is DDJ 2-66.
The agreement, to put it mildly, is fairly good.
BonusAnd as a bonus, the Gwodyen texts resolve our DDJ 19 dilemma. That version of DDJ 19 does not have the terms rvn and yi, but instead chyau "cleverness" and li "profit." Our present DDJ text of that chapter thus appears to be a later emendation (perhaps indeed an anti-Sywndzian emendation; note the anti-Sywndzian additions to the mid-century layers of the Analects; TOA 193) which was made sometime between the Gwodyen text and the Mawangdwei text (the latter already has the familiar wording rvn/yi).
Postscript 2000 1. The Dartmouth Conference volume contains no mention of our paper (which had been submitted by prior agreement in absentia, but was not after all distributed), although in absentia remarks from regular conferees who left early are included in that report. The only mention of our view of the Gwodyen DDJ in the volume is in a footnote reference by Carine Defoort (Leuven) to the originally distributed form of this and a previous Newsletter (see p39). Making the contents of the Newsletter available here may be of use to readers of the Dartmouth volume who would otherwise be puzzled by the reference.
2. Bob Henricks's translation of the Gwodyen DDJ (Columbia 2000) considers and rejects three theories of the DDJ text, but makes no reference to the present theory, a theory with which Bob has been personally acquainted for a decade (he was in the audience, fourth row center, at the original 1990 lecture, and of course he himself saw the contents of our 1998 in absentia paper).
From these seeming efforts to avoid its discussion in the scholarly and general media, we can only conclude that the present theory may indeed have something going for it.
Original Newsletter Copyright © 1999 by the Warring States Project
19 May 2004 / Contact The Project / Exit to Publications Page
