Warring States Problems

E Bruce Brooks, University of Massachusetts at Amherst
ICANAS 36, Montréal, 1 September 2000

This lecture was delivered in six seconds under the 18-minute time limit at ICANAS 36. For the convenience of viewers, notes and references have been added to this on-line version. They are available by clicking on underlined keywords or on the bar Sample Bar at the end of a sentence.


In 1900, in a lecture at the International Congress of Mathematicians in Paris, David Hilbert proposed 23 problems whose solution he felt would significantly advance that science. To Note Mindful of his example, at this present meeting 100 years later, in the foremost French-speaking city of the New World, I should like to make a similar suggestion to Chinese text philologists.

In the study of ancient history, archaeology and philology are natural partners.To Note But in the last 50 years of this collaboration, the senior discipline of philology has taken the junior role, leaving archaeology to plot its finds along a time-line which rests ultimately on a perhaps unsound textual foundation. Without denying that much good work has been done, some of it by persons here present, I nevertheless suggest a renewal of effort in philology. In particular, I recommend that we take fuller account of the apparent diversity of text formation processes in the Warring States period.

Scholars in the past have usually been concerned to ask whether a given author wrote a given book, but there may be few Warring States texts to which this either/or "authentic/forgery" model properly applies. It appears to be more typical of the Warring States for an advocacy group, not an individual, to supply the motive for writing, and for the writing itself to be less a consecutive treatise than a gradually accumulating repository of examples used, or observations made, or positions taken, by that group over time.

For this situation and its variants, a different way of handling the evidence may be more fruitful.

In what follows, I will propose what I think is a more realistic typology of Warring States text formation processes, and will group the 23 problems as examples of those types. Further research will doubtless clarify the examples, and refine the typology itself. To Note Meanwhile, I give my best guess.

The first type I wish to mention is the one which the older methods tended to assume in all cases:

Authorial Texts

1. There is little doubt that the brief Gwodyen text Lu Mu-gung Wvn Dz-sz is the work of one person. Since the piece is a third-person story about Dz-sz, the author was probably not Dz-sz himself. And the defense of remonstrance in the piece differs somewhat from the way that issue is handled by the Mencian school, which sometimes invokes Dz-sz as an example, and which might otherwise be thought to be a likely origin for this separately circulation tract. The problem, then, is: What can be deduced from rhetorical or other features of this piece about its place in the debates of the 04c and early 03c? To Note

2. Too much time has been wasted on the question of Sun Wu, who in my view did not exist, To Note and it would probably be unwise to squander equal time on the almost equally fictive Wu Chi. But the Wu Chi text was clearly written after the Sundz text, by someone who knew the Sundz text. Determining the relation of the Wudz to the earlier work, and as far as possible its own date and place of origin, would be useful in itself. It would also help to set chronological limits on the "Sundz" perplexity.

3. It is widely agreed that Shan/Hai Jing (SHJ) 1-5, which as a group are sometimes called the Shan Jing, is the core of that work. It seems to be relatively free of supernatural details, not to mention human ones, traits which which mark it off from the following chapters. At the same time, its naturalistic details, as Vera will argue in a moment, do not make coherent topographical sense. To Note The indicated question, then, is: How exactly do economic geography and schematic geography interpenetrate in this work? And for whose benefit?

Collaborative Texts

4. No reader of the Shr Ji can long escape the impression that there are two major authorial voices in that work. Neither is much good as a historian, but they differ from each other. One is pedantic with sources, adulatory toward Han, undistinguished as a writer, and does not observe a taboo on the name Tan. The other is cavalier with sources, sometimes critical of Han, very style-proud and also notably vernacular in style; he does observe a taboo on the name Tan. Far be it from me to suggest identities for these people, but it might be good to separate their contributions, so that we know, at any one point, whether we are dealing with a pedestrian scribe or an irresponsible sensationalist. To Note

5. It is formally clear that the postface to the Lw-shr Chun/Chyou applies only to its first set of chapters, the Ji (LSCC 1-12), to which, and not to the whole text, it is physically appended. It seems also that there is a variety of viewpoints in these twelve chapters that may not be fully accounted for by its habit of quoting other works. How many people were probably involved, and what affinities did they have within contemporary thought? And what is the impact on the external evidence for the existence of works such as the Jwangdz, if only LSCC 1-12 are accepted as from the preface date of 0239? To Note

Layered Texts

6. Given the handling of a taboo on the First Emperor's name, I conclude not only (as D C Lau has also suggested) that the second and third series of LSCC chapters, the Lan and the Lun, were added later, but more specifically that they were added at two different points within the Chin dynasty. What can be reconstructed of the nature of that later extension of an older work? How did the resumption of state patronage take place? Were any of the same people involved in compiling the new chapters? Was the viewpoint of the text adjusted to the Chin Dynasty's mature idea of its place in the scheme of things? To Note

7. A harder problem of the same type, since there no seeming continuity of patronage, is to locate the later Shan/Hai Jing strata in time and space.

8. It is hard to imagine the Dzwo Jwan, considered as a commentary on the chronicle of Lu, being first undertaken in any state but Lu. For one thing, Lu court rank may have been a precondition for access to the chronicle text. At the same time, what look like late additions to the Dzwo Jwan seem to have been written with the Chi ruler as their audience, and have the effect of forecasting a great destiny for Chi, a prediction which would presumably not have been well received at the court of Lu. When, and under what auspices, did the work relocate itself from Lu to Chi? Mencius's return to Chi from his mother's funeral in Lu occurred at just about the right year, and his outlook on contemporary issues is at many points reminiscent of that of the Dzwo Jwan. Was he involved in any way? To Note

9. The Jung Yung is quoted in the 03c (Tswei Shu) layers of the Analects and also of the Mencius, but only its lower-numbered sections are attested in this way. The higher-numbered ones have been shown to reflect Chin or later administrative usage. As with the Lw-shr Chun/Chyou, it seems that work on this finished text was resumed under the Empire. Why was it in the category of works liable to be so treated? Who did the treating?

Assembled Texts

10. The Gwodyen texts titled Yw Tsung ("Collectanea") give us early 03c examples of a type previously known only from later times, but long suspected for the Warring States. What can be learned about the type from close study of these four specimens? 

11. It has been maintained by many including Knoblock that Sywndz 27 is at least in part a work by, or validly reflecting, Sywn Ching. A reappraisal in the light of the preceding model might reach a different result.

Conflated Texts

12. Here is a nifty little problem. It is clearly stated by Lyou Syang that he produced our Jan-Gwo Tsv by conflating six named works, all now lost. Some contrasts of style and ideology among its hundreds of tales are obvious. Others are less so. Who would like to separate this jumble into its six original components? The continued use of JGT as though it were a single, let alone a reputable, historical source does not display our science in a very grown-up light. It also limits our knowledge of the Shr Ji. Which of the six original constituents were known to the Shr Ji perpetrators, and which were not? These seem to be largely uncharted waters. To Note

Accretional Texts

13. I hope I may be forgiven for thinking that the Analects is now a fairly well understood text, but there are other possible examples of that classic school accretion type. One is the Dau/Dv Jing. Like the Mician anecdotal chapters, if less consecutively, it seems to have maintained a dialogue with the evolving Analects. This pattern of mutual connection leads to a prediction about the evolution of the DDJ text which has been strikingly confirmed by the Gwodyen DDJ florilegia. Those florilegia draw only on portions of the DDJ which our Analects theory had predicted would be found to be in existence as of the most probable date of the Gwodyen 1 tomb. To Note The growth of the DDJ may then give evidence for the nature and evolving political concerns of the group which presumably sponsored it. What kind of group was it, what were its political concerns, and why did those concerns evolve?

14. The Mencius text stems mostly from the post-Mencian school, which added gradually to the original core of interview transcripts now found in MC 1. But this is not a case of simple linear accretion. At one point the school seems to have divided, and what we now have is the parallel accretion sequences of both schools. To Note The subtle dialect differences between them are of interest to historical linguistics, and the blatant philosophical differences between them may be relevant to the history of thought.

Accumulative Texts

15. The previous accretional type featured the slow building of an essentially single work. There is a less integral mode, in which various material produced by, or of interest to, the sponsoring school is tossed into the drawer together, but is not all of the same character, and does not organize itself as a quasi-homogeneous entity. Such apparently is the Gwandz, whose late commentary chapters show a school-type awareness of the earliest chapters, but which as a whole possesses no organizing historical thread. What sort of institutional continuity can be inferred from a careful analysis of this material? How did the text process, whoever sponsored it, survive into and beyond the disturbances of the Chin dynasty? To Note

16. The Mwodz school, apparently more disciplined than the Gwandz group, preserved its ethical and military writings in separate compartments of its corpus, so also the logical and anecdotal writings. But the geographical affinities of the sponsoring group, and the latest date at which its text-production process was still active, are even less clear than with the Gwandz. This is an intolerable situation for one of the major thought traditions of the period, and elucidation is urgently called for. To Note

17. Even if we reject the prodigy myth of Sywn Ching, as I think we must, his productive life was long. But it would be naive to regard everything in the present Sywndz text as his own composition, and it seems probable that even during his lifetime there was other input into the school repository, doubtless subject to his general oversight. Sywndz lost his court patronage at the end, there is a notable lack of information about first generation disciples, and the Sywndzian writings eventually needed to be put in order by Lyou Syang, a further indication of posthumous disorganization in the school. How many hands can be detected in the present mix? What history do they imply?

18. Yet more finely divided than the four-part Mwodz corpus is the complex Jwangdz, whose solution is made more difficult by the fact that our text is only a popular abridgement of the longer Han one. It seems that clusters of accretional texts came together, as in the Mencius example above, and that these in turn were edited into a larger unit at least once, and possibly several times. The latest material in the work is of Han date. Without getting distracted by sentimental Jwang Jou questions, what can be said of the nature and location of the "little traditions" which are rolled up together in the present hodgepodge? To Note

Parasitic Texts

19. All the above are examples of continuous text processes, or intermittent ones reopened under similar or compatible auspices. With the Shang-jywn Shu, however, we may enter a category where an old but small bit of writing is used as a nucleus for accretion by a substantially different sponsoring group. At least one early portion of this text seems to be its military chapters, which may conceivably date from near the time of Lord Shang, its completion seems to have taken place centuries later, in Han. What can now be done to continue and complete the analysis of this text begun by Duyvendak? To Note

20. There is no credible tradition that Han Fei had disciples. Not even the lurid tales invented about him by the Shr Ji give a hint of disciples. To Note There is thus no plausible structure by which any body of writings by him could have been accumulated and preserved (compare the Sywndz case, where we can plausibly infer a well organized text-producing and disseminating effort). The only document in this large corpus which at the outset may plausibly be attributed to Han Fei is his loser memorial, the whining and tedious HFZ 3. It would appear that his persona was revived in Han by statecraft theorists who found him convenient, as a victim rather than an architect of the hated Chin, to shelter their own work under. Around that memorial as a nucleus, these Han theorists then seem to have added the large bulk of the present Han Feidz. That this and the Shang-jywn Shu evolved together at one point is shown by the overlap between SJS 13 and HFZ 53. As a five-finger exercise, some enterprising beginner might establish the directionality of this two-chapter relationship, and then go on to clarify the general relationship between the two text sequences. To Note It belongs to a perhaps more advanced level to establish the connection between these texts and the political conditions of the first few Han reigns.

Imitative Texts

21. The prestige of the Chun/Chyou of Lu in the 04c was evidently sufficient that a commentary on it could be successfully repackaged as a Chi statecraft text, the Dzwo Jwan. This is hard to conceive if Chi, and thus also Jin (which is a major focus of Dzwo Jwan interest), had had a chronicle of its own. The implication is that our Bamboo Annals is not based on an earlier chronicle of Jin, but is an imitation of the Lu Chun/Chyou, some of whose features, and even later interpolations, it appears to echo. Besides the Chun/Chyou, the Bamboo Annals compilers evidently had access to inscribed bronze vessels, plus some recently forged documents of Shu type. How many more sources can be identified?

22. It is widely agreed that the Gwo Yw is in some way related to the Dzwo Jwan, and that, of the two, the Gwo Yw is the later work. David Pankenier dates it astronomically to the last years of the 04c or the first ones of the 03c, thus establishing a terminus ante quem for the Dzwo Jwan. To Note The GY might be thought of as the paperback edition of the DJ, with the labored commentary on the CC left out, and the exciting anecdotes left in and augmented. But it also has directions of its own. How can those directions best be characterized, and for what audience was the work written? As with Wu Chi and the Sundz, the less enshrined Gwo Yw text may be our best handle on the ever-vexed question of the Dzwo Jwan.

23. The Dzvngdz is quoted, as such, in the Chywn-shu Jr-yau of 631. From that fact, and those quotes, it would seem that the Han text of the Dzvngdz is completely preserved in Da Dai Li 49-58, and it is evident from a study of this material that it was partly compiled by assembling Han anecdotes of Dzvngdz, on the model of the Analects. When those borrowings are eliminated, does a core remain? The whole question of spurious traditions arises in strong form with Dzvngdz, whose historical role as an early head of the Confucian successor school in Lu is attested by LY 8:3, but who seems to have been rejected by that school, along with the rest of its early disciple leaders, when the Kung family took over the enterprise in the early 04c. When Dzvngdz reappears in the Analects, more than fifty years later, he represents a different, more family-centered, line of thought that is also a concern of one of the Mencian schools. Did the Dzvng family have a role in this repackaging of a still usable ancestor? How much Dzvngdz material embedded or echoed in other texts might stem from this source? To Note


Many other excitements, among them the accretional Shr and the forged Shu, await the intrepid Warring States philologist, but I must not exceed the numerical model of my distinguished predecessor, or the time limit allowed by our hosts. I will thus end with this 23rd question.

At the web site mentioned on the handout, there is a link to the 23 problems of David Hilbert. To Note It will there be found that after a century, a majority of Hilbert's problems have been either solved or clarified. It is to be hoped that our science may give an equally good account of itself in the century which is soon to open. Or, to conclude with the words with which Hilbert himself ended,

"May the new century bring it gifted prophets and many zealous and enthusiastic disciples."

Calligraphic Separator

Postscript. Our own best current conclusions about the texts in question, in continually updated form, may be consulted at the on-line reference publication Classical Chinese Texts.


All lectures and abstracts posted on this site are Copyright © by their authors.

 Back to Lectures Page

2 Sept 2000 / Contact The Project / Exit to Lectures Page