Review of Mattos: The stone drums of Ch'in


Gilbert L. Mattos, The Stone Drums of Ch'in (Monumenta Serica Monograph Series, 19). Nettetal, BRD: Steyler Verlag, 1988. Paperback, 497 pp.

Review in Acta Orientalia, 1990, 51: 241-256.

Donald B. Wagner

Web version 1999; stylistic revisions 2021 and 2024.

Introduction

The "Stone Drums" are ten inscribed granite boulders now preserved in the Palace Museum, Beijing, whose inscriptions are believed to be extremely ancient and which have inspired a tremendous amount of research and speculation since the seventh century A.D. The stones average about 80 cm in height and 60 cm in diameter, and probably weigh about a half ton each. Gilbert L. Mattos gives a thorough study of all aspects of the stones, with special emphasis on the history of their discovery and preservation and on the interpretation of the inscriptions.

It will be useful to give the reader of this review an initial orientation by stating some of Mattos's most important conclusions. The Stone Drums were discovered, about A.D. 600, near Fengxiang, Shaanxi. Their vicissitudes since then are described in written sources of various periods, and there is no doubt that the stones now in the Palace Museum are the same stones as those described in the seventh century. However the taking of rubbings through the ages has damaged the inscriptions severely, and the best material now available for the study of the inscriptions consists of rubbings taken in the Song period. Mattos reproduces the best set of Song rubbings as well as a modern rubbing and various other relevant material such as hand-copies.

Of an estimated 700 characters in the original inscriptions only 501 remained in the Song and only 272 remain today. The difficulties in interpreting what remains of the inscriptions begin at the lowest level, with the identification of the individual strokes of the characters in a rubbing of a damaged surface, and continue at all higher levels.

The conclusion of Mattos's thorough and detailed study of the inscriptions is that they were composed in the state of Qin, perhaps in the fifth century B.C. They consist entirely of four-character rhymed verses in the style of the Shi jing, and include a few lines which clearly are paraphrases of Shi jing lines. They are entirely concerned with hunting and fishing expeditions. "Running through these poems is one theme, namely, the perfect order of things - from chariots to rivers to the movements of the footmen. Along with this we have a secondary theme complementing the first, namely, the great abundance of things - hunters, footmen, prey, fishes, vegetation, waters, etc." (p. 330).


This book gives what appears to be an exhaustive survey of the voluminous literature on the Stone Drums and simultaneously considers all the problems afresh. Its conclusions seem to be about as solid as will ever be possible, given the poor state of the inscriptions themselves and the paucity of the sources concerning their provenance. I shall nevertheless take up in the following a few points on which the methodology of the study of the Stone Drums - not Mattos's alone but most of the research which has gone before - might be improved.

The discovery of the Stone Drums

The treatment of the written sources which relate to the provenance of the Stone Drums would benefit from a more rigorous and at the same time more imaginative methodology. I shall suggest here the outline of what might be a useful approach.

The basic sources appear to be the following. The geographer Su Xu (fl. 627-649) is quoted in a Tang work on calligraphy:[1]

It is generally said today that, of extant remains of [ancient] calligraphy, those of Li Si are the oldest. It is not generally known that remains of [the calligraphy of] Shi Zhou [Shi Zhou zhi ji] are nearby, [here] within the passes [guan zhong, the capital region].

Li Si (ca. 278-8 B.C.) was the presumed author and calligrapher of the many stone inscriptions of the First Emperor of Qin. Shi Zhou was the legendary historian of King Xuan of Zhou (trad. r. 827-782 B.C.). Two of the odes of the Shi jing, nos. 179 and 180, are traditionally taken to describe hunting expeditions of King Xuan south of Mount Qi, which is not far from the area where the drums were found. The Stone Drum inscriptions also describe hunting expeditions, and the first line of one is almost identical to the first line of ode no. 179.[2] The early attribution of the inscriptions to Shi Zhou was therefore almost inevitable.

A passage in the geographical treatise Yuanhe jun xian tuzhi (by Li Jifu, 758-814), completed in 813,[3] in its description of Tianxing District (near modern Fengxiang, Shaanxi), describes the stones and then quotes another passage by Su Xu:[4]

The stone drum inscriptions are located approximately 20 li [ca. 11 km in this period] south of the district [i.e. the district seat of Tianxing]. In shape the stones resemble drums. They are ten in number. [Their inscriptions] record a hunting [expedition] of King Xuan of Zhou, and thus their inscriptions are remains of [the calligraphy of] Shi Zhou [Shi Zhou zhi ji]. In the Zhenguan period [627-649] Su Xu, Attendant of the Board of Civil Service, wrote of them: "Yu [Shinan], Chu [Suiliang], and Ouyang [Xun] all acclaimed their antiquity and excellence. Although the years have been long and [the inscriptions] have deteriorated, what remains is still worthy of note; it is a great pity that the compilers of geographical treatises through the ages have not recorded them."

(We shall return further below to the three persons named here.)

A commentary on the Hou Han shu, compiled by order of the Crown Prince Li Xian in the period 675-681, commenting on a difficult sentence, proposes the reading yuanyuan 元元, "common people", for the enigmatic yuan er 元二 and explains:[5]

In ancient writing, when a character was to be written twice, a small "two" [er] was written under the first character to indicate that this character was to be read twice. In later times this was not understood, and therefore [this yuanyuan] was read as yuan er. . . . In the Stone Drum inscriptions of Qizhou, whenever there is a repeated character, this character er is used; this is a clear confirmation.[6]

The place-names mentioned in these passages are Guan zhong, Qizhou, and Tianxing. Guan zhong, "the region within the passes", corresponds roughly to the southern part of the modern province of Shaanxi. Qizhou, a prefecture (zhou) in the Tang period, is roughly modern Fengxiang County, Shaanxi. Its name was changed to Fengxiang in 742. Tianxing (named Yong before 765)[7] was a district (xian) in Qizhou Prefecture.


The first thing to be done with sources like these is to ask the questions: Who was each author? What was his source for the information he gives? How are the sources related?

The persons to be considered are the following ten, most of whom have biographies in the Tang histories:[8]

Ouyang Xun, 557-645, calligrapher, co-author of Yiwen leiju.
Yu Shinan, 558-638, calligrapher, author of Beitang shuchao.
Chu Suiliang, 596-658, calligrapher.
Su Xu, fl. 627-649, geographer, co-author of Kuo di zhi.
Li Xian, 651-684, Crown Prince.
Four co-authors of Li Xian's commentary on the Hou Han shu: [9] Li Jifu, 758-814, geographer, author of Yuanhe jun xian tuzhi.

The first nine on this list were near-contemporaries, and a check of their biographies and other parts of the Tang histories indicates that they were connected in various ways: Ouyang Xun and Yu Shinan were friends; Chu Suiliang studied calligraphy under Yu Shinan and was acquainted with Ouyang Xun; Su Xu, Chu Suiliang's father Chu Liang, and Yu Shinan were among the "Eighteen scholars" appointed as Scholars of the Office of Literature on a single occasion;[10] Yu Shinan was the Crown Prince Li Xian's teacher; it was Li Xian who assembled the group which produced the Hou Han shu commentary; and one of this group, Xu Shuya, seems to have had some sort of connection with Yu Shinan and Chu Liang.[11]

It is also interesting to note that Li Xian, in the period when the work on the Hou Han shu commentary was in progress, had among his titles Prince of Yong (Yong wang) and Governor of Yong (Yong mu). I have not found a clear indication that he actually had his domicile in Yong (i.e. Tianxing, see above), but this might well have been the case.

The two quotations from Su Xu above are likely to have been from a single original text, and this text was almost surely the source for Li Jifu's description of the drums. Li Jifu even uses the phrase Shi Zhou zhi ji, which occurs in Su Xu's text. It is therefore probable that the location, ca. 20 li south of Tianxing, was also in Su Xu's text, and there is no reason to suppose that Li Jifu had any direct knowledge of the whereabouts of the drums in his own time.

Did Su Xu himself have direct knowledge of the Stone Drums? Perhaps so, but this is not a necessary conclusion. It is just as likely that he had heard or read of them from the three calligraphers whom he names.

Clearly one of the Hou Han shu commentators must have seen the Stone Drum inscriptions: either the actual stones or a copy or a rubbing. The location given here is Qizhou; this may merely mean somewhere in this prefecture or, more probably, it may mean the administrative seat of the prefecture, which seems to have been in Yong (Tianxing).[12]

Thus the most likely location of the Stone Drums when they first appear in the written record is ca. 11 km south of modern Fengxiang, Shaanxi. Archaeology in this area has made rapid progress in the past few years. The walls of the Qin city of Yong, which was the capital of Qin from 677 to 383 B.C., have been located in Fengxiang. South of Fengxiang, approximately at the location specified here, are the Qin royal tombs.[13]

It is a fair guess (but only a guess) that the Stone Drums, assuming that they are not forgeries, were found either in a Qin royal tomb or in the remains of some ceremonial structure associated with the tombs. We have however no record of the actual discovery. The implicit assumption made by most scholars, including Mattos, that the drums were not discovered significantly before they appear in the sources, in the early seventh century A.D., appears not to be well founded. The silence of the sources before this time has no significance; and indeed they are not quite silent. It is odd that Mattos does not mention, in his otherwise very thorough review of the sources, that a mountain named Shigushan, "Stone Drum Mountain", in Chencang, 25 km southwest of Yong (Tianxing), is mentioned in sources as early as the Liang period (502-557).[14] It may be that this fact is irrelevant, but it is not obviously so, and it would have been best to give an explicit argument on the point.


In the eighth century and the beginning of the ninth a number of other authors, including the poets Du Fu (713-770) and Han Yu (768-824), mention the Stone Drums in various contexts, sometimes giving a location for them. They mention three different locations: Qiyang, Yong (Tianxing), and Chencang. There was a district named Qiyang in Qizhou prefecture, but here the term is probably used to designate vaguely "the area south of Mount Qi", alluding to King Xuan's hunting expeditions. The mentions of the other two places, Yong and Chencang, are more of a problem. Li Zhongcao[15] appears to believe that every writer who mentioned the location of the drums actually had correct knowledge of the location in his own time, and therefore he finds it necessary to postulate that the drums were moved at least twice between Chencang and Yong. Mattos seems to make the same assumption about the sources, but he rightly finds it difficult to believe in all this moving about of five tons of granite. Therefore he postulates that, while the drums remained at one location, some writers chose to describe this location as near Yong and others as near Chencang.

This problem disappears when one realizes that some writers do not know what they are talking about, but merely quote others uncritically. I have argued above that in the early ninth century, when Li Jifu gives the location south of Yong, he is likely to be relying on sources from the early seventh century. The location in Chencang is first mentioned in the middle of the eighth century (Mattos, p. 39 fn. 7); again many of the writers who give this location may merely be quoting others. There is nothing extremely unlikely about a single move of the drums from the place where they were found to another place, for example Chencang, for safekeeping or display or for some other reason; especially after they had come to the attention of a Crown Prince. Mattos's explanation seems to me less likely, but it is certainly a possibility. It might also be that some writer simply was mistaken in giving Chencang as the location, and that other writers copied this mistake. Such a mistake could for example have come about because of the confusing series of changes of names and boundaries of districts in Qizhou Prefecture which were made in the eighth century. At one point Chencang was named Fengxiang, and at another time this name was applied to Yong (Tianxing).[16]

Guo Moruo (writing in 1933, quoted by Mattos, pp. 100-104) gives an interesting speculation concerning the original context of the Stone Drums. The geographical treatise Kuo di zhi, completed in 624, of which Su Xu was a co-author, states that the "Plain of the Three Sacrificial Sites", San zhi yuan, was located 20 li south of Yong (Tianxing),[17] i.e. at the same place where Li Jifu located the drums. The three sites in question were established by Dukes Xiang, Wen, and Ling of Qin (r. resp. 777-766, 765-716, 424-415 B.C.). From this and other evidence Guo Moruo concludes that the drums were related to sacrifices performed at the sacrificial site of Duke Xiang in 770 B.C. Numerous later scholars have refuted this theory; among other things there are better sources which indicate that the sacrificial sites were located elsewhere.

As Mattos notes, it is nonsense to rely on a sixth-century A.D. source for the location of a site established in 770 B.C.; but the statement of the Kuo di zhi can nevertheless be interesting if we ask a different question: not whether it is true or false but what led its author (or the author of his source) to believe that it was true. A likely answer is that there were visible on this plain 20 li south of Yong the remains of three ancient structures of one sort or another (for example structures related to the Qin royal tombs), and that there was some reason to believe that these were associated with Qin and with sacrifices. It would then have been straightforward to conclude that these three structures were the Three Sacrificial Sites. The Stone Drums fit nicely into this hypothesis, for their inscriptions are plainly about hunting, and Duke Wen established his sacrificial site after a hunting expedition.[18] The discovery of the drums may thus have been the primary reason for identifying the ruined structures on the plain as the three sacrificial sites of Qin. Since this plain was in fact, as noted above, the location of several Qin royal tombs, this hypothesis suggests again an association of the drums with one of the Dukes of Qin who had their capital at Yong in the period 677-383 B.C. The hypothesis is not itself very strong as it stands, however, for it is not until the Song period that any extant source associates the Stone Drums with Qin rather than with King Xuan of Zhou; but a study of Tang geographical writings concerning the Fengxiang region might possibly reveal related traditions about Qin which could fit into the hypothesis.

The interpretation of the inscriptions

The most important part of this book is the interpretation of the inscriptions on the Stone Drums. The interpretation is closely argued, character by character and line by line, and in this undertaking Mattos's methods seem by and large to be those accepted by specialists in Chinese epigraphy. It is these methods which I shall discuss here.

The study of the drum inscriptions naturally relies heavily on the study of bronze inscriptions. H. G. Creel has given an excellent short discussion of the problems involved in this study,[19] from which I quote:

Chinese scholars have been doing serious work in this field for at least a thousand years, and the greatest strides have undoubtedly been made in the twentieth century. Yet it is improbable that any really competent scholar would claim that he could read every character of any very long inscription, without any possibility of error. . . . The study of bronze inscriptions is necessarily a highly specialized field. In the entire world only a small number of scholars are really expert in it. They have necessarily developed a rather complex apparatus and jargon. There is a large literature on the subject, but even to read this effectively requires a good deal of study, for which only a few have the time and the inclination. It is difficult to present one's arguments concerning bronze inscriptions fully, in a manner that will be wholly convincing to others who have not made such a study.

It must be said immediately that Mattos seems to have a solid command of this "rather complex apparatus and jargon", and that I am not competent to evaluate his work in detail. A historian who wishes to use the bronze inscriptions as sources will normally rely on the conclusions reached by specialists like Mattos, skipping over the detailed reasoning on which the conclusions are based. This sort of short-cut is often necessary when a specialist in one field needs to use the work of specialists in another field: but it has its dangers. For example, the reasoning behind an interpretation of an inscription may include implicit historical assumptions which the historian who wishes to use the interpretation would not be willing to accept, if he were aware of them. A good basic rule for the historian using the results of other specialists - bronze inscription studies, radiocarbon dates, bronze alloy analyses, and so forth - is that he should acquire a general knowledge of the methodology of the field in question and be sure (1) that he finds this methodology acceptable and (2) that the specialist whose results he is using follows this methodology correctly. To demand more than this is a counsel of perfection; to accept less is a dangerous game.

In attempting to follow this basic rule I can note at the start that (2) Mattos's argumentation does generally seem to conform to the methodology of the field of Chinese epigraphy. I am however, less sure of (1): some assumptions are generally made in this field which I feel uncertain about accepting. These are discussed in the following.


The total corpus of extant Chinese texts from before about 500 B.C. consists of the Shi jing, the genuine chapters of the Shu jing, and the various bronze inscriptions. (Oracle bone inscriptions can be ignored in the present discussion.) The first two total no more than about 45,000 characters in all; the third, perhaps as much as a fifth of this figure. Taken together, these texts no doubt constitute a larger corpus of well preserved and reliably dated texts of such high antiquity than is available for any other archaic language; yet in absolute terms the corpus is rather small. Its size corresponds perhaps to that of this volume of Acta Orientalia, with perhaps 100,000 words. Only a small part of the vocabulary of twentieth-century academic English will be found in this one volume, and it is also doubtful that anywhere near all of the possible grammatical forms of English will be found exemplified here. --And this comparison is generous, for the archaic Chinese corpus is not homogeneous in age, but is spread out over a period of many centuries, so that the coverage of any particular time is sparser. A more apt comparison might be with the Oxford book of English verse.

Because of the small size of the extant corpus and for other reasons we should a priori be prepared, in any long archaic Chinese inscription, to meet:

* Graphs not found, or used in ways not found, elsewhere in the corpus.
* Words not found elsewhere in the corpus.
* Grammatical constructions not found elsewhere in the corpus.
* Idiosyncrasies or mistakes of the author or the calligrapher.

Now here is a great problem. The correct original meaning of an inscription may not be obtainable without invoking one or more of these exceptions; but if the invocation of any of these exceptions is freely permitted, the floodgates are open: any interpretation, no matter how eccentric, can be justified.

In general the deepest problem of archaic Chinese epigraphy, and the one least discussed, is that for any given inscription a great many interpretations are normally possible and justifiable. In order to limit the number of interpretations to be considered, certain rigid rules are applied. These rules are absolutely necessary to avoid chaos, but one can be sure, a priori, that they will often also rule out the correct interpretation.


Mattos states three rules which "ought to be strictly observed" in proposing a reading for a character (p. 130):

1. Co-occurring graphs in the texts which differ in shape and structure represent different words.
. . . [I]t hardly seems tenable that the calligrapher by some whim would have been so inconsistent as to write a word in two entirely different ways within the same text . . . This certainly would run contrary to the rules of writing which rest on consistency as a fundamental principle. Granted some leeway must be made for human error (e.g. inadvertently leaving out a stroke or some such thing) but certainly not to the extent of construing graphs differing in shape and structure as standing for one and the same word. In our opinion, explanations which would allow this are unacceptable and more often than not are quite unnecessary.

2. The texts presumably are grammatical.
We assume that the author of these texts composed them in accordance with the prescribed [sic] [20] rules of classical Chinese grammar and that every line in the text is explicable in terms thereof. In the event that a proposed reading would yield an ungrammatical structure - one for which there is no grammatical precedent in the ancient Chinese literary language - it shall be rejected. . . . [W]e normally will rely on parallel examples found in the classical literature, particularly Shih-ching.

3. The texts presumably are coherent.
We assume that in each of the [Stone Drum] poems the subject concerned was developed in a logical way and that each, from beginning to end, represents a coherent unit. Proposed readings which would detract from, or destroy, the logical unity of the piece shall be rejected.

The third rule seems sensible enough, but with an important caveat: the inscriptions originate in a culture very alien to ours, and what is coherent or incoherent will not always be obvious to us today.

The second rule assumes that we have complete knowledge of the grammar of the language in which the Stone Drum inscriptions were composed. In fact we do not. An example is the occurrence of the character 殹 in two of the inscriptions, in the lines:

Qiān yì miǎn miǎn 汧殹沔沔, "The Qian X flows amply"

Qiān yì jì jì 汧殹洎洎, "The Qian X is brimming over"

(Mattos, pp. 172-176, 195, 249, 257). Karlgren (GSR[21] 589 b, e), citing only these two lines, defines as "a particle", written 繄 in the Zuo zhuan, and suggests that it is cognate to the particle 伊 , which is found in the Shi jing used with the same function as wéi 維. [22]

The more usual interpretation of , first proposed in the eleventh century, is that it stands for 也 (Mattos, p. 172). There was not a great deal of evidence for this interpretation until the discovery in 1975, at Shuihudi in Yunmeng, Hubei, of a collection of Qin legal texts of the third century B.C. is used throughout these texts with the function of, and does not occur.[23] There is therefore no longer any room for doubt that was used for in Qin in the third century B.C. There is, however, no evidence for this usage before this time.

Either of these two interpretations could be applied to in the above lines, and the lines would be translated in the same way in either case. Mattos, however, applying his Rule 2, rejects both. His reason is that in the Shi jing neither nor ever occurs in a line of the form ABXX, where XX is a reduplicative. There are 324 such lines in the Shi jing: this may seem to be a great many, but is it really a sufficiently large sample of the language in which the Shi jing was written? occurs for example in the lines[24]

Zhù chéng yī xù / Zuò Fēng yī pí 築城伊淢  作豐伊匹
"The wall he built was moated / The [city] Feng he built matched it."

It is useful to be reminded that reduplication may well have had a grammatical as well as a rhetorical function. Nevertheless we do not have a sufficiently detailed knowledge of the function of reduplication to reject with any confidence the proposition that the of the Stone Drum inscriptions is the same as the of the Shi jing. Nor can the interpretation of as be confidently rejected.

In this particular case the problem is not great, for there seems to be no other viable interpretation, and Mattos simply translates as "X". If, however, some other interpretation had been possible - for example if were attested as a loan for some word which plausibly could fit in the line in question - then Mattos would probably have confidently accepted that interpretation as the only one possible (cf. e.g. p. 150). This confidence, I claim, would have been unwarranted.


The major problem, however, is Rule 1, quoted above, that the same word will not be found written with two different graphs anywhere in the ten Stone Drum inscriptions. Mattos invokes this rule quite often in rejecting alternate interpretations (e.g. pp. 153, 155, 173, 213), but he gives nowhere any real argument for its validity. The passage quoted above says in effect no more than that this rule is so obvious that I am a fool if I doubt it. I dislike being called a fool, and therefore have put some effort into disproving Mattos's contention.

Noel Barnard states, and argues cogently for, a superficially similar principle, which he calls "the principle of constant character structures":[25]

Repeated characters, or elements of characters, in any one document, [are] always written on the same structural principles - numbers of strokes, position and intrinsic features of stroke combinations, always accorded in each occurrence of the same character written by the same writer.

This principle can be mistaken for Mattos's Rule 1, but in fact it concerns graphs, not words; it is useful in detecting forgeries, but is not applicable in the interpretation of graphs as words. I do not know whether Prof. Barnard agrees with Rule 1, but many scholars of bronze inscriptions do.

In pre-Han transmitted texts there are numerous counterexamples to Rule 1. Curiously, Mattos gives three of these himself, in other contexts, without taking notice of their significance for his rule (pp. 150, 160, 264):

敺、驅, "to gallop"; both graphs are used for this word in Meng zi and in Zhou li.
樸、朴, "to simplify", both used in Xun zi.
shuài 率、帥, "leader", both used in the Zuo zhuan (cf. GSR 499 a-c).

A cursory search in Grammata Serica recensa (GSR) for pairs of different graphs used in the Shi jing for the same word in the same meaning turned up the following (I arbitrarily stopped looking when I had found ten):

yuè 閱、說, "pleased", GSR 324 p, q
ěr 爾、邇, "near", GSR 359 a, c
yìn 田、朄, "a small drum", GSR 362 a, 385 l
均、鈞, "even, equal", GSR 391 c, e
wèn 聞、問, "fame", GSR 441 f, g
suì 卒、崒、碎, "to break", GSR 490 a, c, n
suì 㒸、璲, "jade insignium carried at the belt", GSR 526 a, j
zhǐ 旨、耆, "to effectuate", GSR 552 a, l
fēi 匪、斐, "ornate", GSR 579 c, i
zhān 詹、瞻, "to see, to look at", GSR 619 a, c

In arguing against Rule 1 we may also look at modern parallels. The consistency of modern English spelling is a product of the efforts of "schoolmasters and governesses" in the eighteenth century; one need only look in any seventeenth-century English printed book to find numerous examples of the same word spelled in different ways, often on the same page or even in the same sentence. Even in the modern regularized English script a word can be seen in several structurally distinct graphic representations within the same text: in Roman or Italic, and with capital or small initial letter. And when is a word the same word? The three modern Chinese graphs 他、她、它, respectively "he", "she", and "it", represent the same word of the spoken language, arbitrarily divided into three under the influence of Western languages. We can hardly begin to guess how often, if ever, it occurred in ancient China that a single word was distributed over several graphs, perhaps according to non-linguistic criteria that are entirely foreign to us today.

The counterexamples given above will perhaps be rejected as being relevant to transmitted texts but not to inscriptions, and the modern analogies may also be rejected as irrelevant. But it cannot in fact be my task to disprove Rule 1: no proof has been offered by those who invoke it.

I conclude that Rule 1 is arbitrary, unjustified, and probably incorrect. Its function is to reduce the number of competing interpretations for an inscription, thereby producing a specious impression of reliability.

More generally, I conclude that historians must be very careful in using bronze inscriptions as sources. One cannot in general rely uncritically on the conclusions of epigraphers; it is necessary to study carefully the possible alternate interpretations and the grounds on which these have been rejected.

Conclusion

I shall end this review by emphasizing that Mattos's book is a giant step forward in the study of the Stone Drum inscriptions. He surveys and discusses all of the previous work which has been done, and applies a methodology which in general is more critical than that of his predecessors. Though I have expressed disagreement with a part of his methodology, and though I also disagree with a few details in its application, my conclusions are not different from the selection of Mattos's conclusions which is quoted at the beginning of this review.

[1] Chinese text given in full by Mattos, p. 37, fn. 1. Here and elsewhere I use my own translations; often, as here, my translation differs considerably from that of Mattos.

[2] Bernhard Karlgren, The book of odes: Chinese text, transcription and translation, Stockholm: Museum of Far Eastern Antiquities, 1950, pp. 123-125. Ode no. 179 has Wǒ chē gōng gōng 我車既攻, "Our carriages are well-worked"; the inscription has wú chē jì gōng 吾車既工, with no apparent difference in meaning.

[3] On the history of this work see especially the introduction to the modern critical edition, Yuanhe jun xian tuzhi, ed. by He Cijun, Beijing: Zhonghua Shuju, 1983.

[4] Ibid., j. 2, p. 41; Mattos p. 37.

[5] Hou Han shu, critical ed. Beijing: Zhonghua Shuju, 1965, introduction p. 8.

[6] Ibid., j. 16, pp. 614-615. Cf. Mattos, p. 21, fn. 1.

[7] Yonghe jun xian tuzhi, j. 2, p. 41.

[8] Jiu Tang shu, Beijing: Zhonghua Shuju, 1975, j. 189a, 72, 80, 86, 70, 189a, 68, 189a, 148, pp. 4947, 2565-2571, 2729-2739, 2831-2834, 2541, 4956, 2507, 4953, 3992-3997: Xin Tang shu, Beijing: Zhonghua Shuju, 1975, j. 198, 102, 105, 81, 102, 198, 89, 198, 146, pp. 5645-5646, 3969-3973, 4024-4029, 3590-3592, 3969, 5657, 3756, 5654, 4738-4744. Su Xu has no biography.

[9] Jiu Tang shu, j. 86, p. 2832; Xin Tang shu, j. 81, p. 3590. There were seven co-authors of Li Xian's commentary, but for three of these, Cheng Xuanyi, Shi Cangzhu, and Zhou Baoning, no further information is available.

[10] Jiu Tang shu, j. 72, p. 2582; Xin Tang shu, j. 102, p. 3976.

[11] Xin Tang shu, j. 198, p. 5653.

[12] Jiu Tang shu, j. 38, pp. 1402-1403; Xin Tang shu, j. 37, pp. 966-977; Yuanhe jun xian tuzhi, j. 2, p. 41. The information given on this point is quite confusing, however, and will require more careful study.

[13] Han Wei et al.: "Qin du Yong cheng zuantan shijue jianbao", Kaogu yu wenwu, 1985.2.: 7-20; "Fengxiang Qin Gong lingyuan di-er-ci zuantan jianbao", Wenwu, 1987.5: 55-65; "Qin du Yong cheng kaogu fajue yanjiu zongshu", Kaogu yu wenwu, 1988.5/6: 111-127.

[14] See Li Zhongcao, "Shigu zuichu suozai di ji qi keshi niandai", Kaogu yu wenwu, 1981.2: 83-86 + 82. This article is also cited by Mattos.

[15] Ibid.; cf. Mattos, p. 39.

[16] Jiu Tang shu, j. 38, p. 1403; Xin Tang shu, j. 37, p. 966; Yuanhe jun xian tuzhi, j. 2, p. 40.

[17] Quoted in the Zhengyi commentary to the Shi ji, Beijing: Zhonghua Shuju, 1962, j. 5, p. 180.

[18] Shi ji, j. 28, p. 1358; Édouard Chavannes, Les mémoires historiques de Se-ma Ts'ien, vol. 3, 1898, repr. Paris: Cathasia, 1967, p. 420.

[19] Herrlee G. Creel, The origins of statecraft in China, vol. 1: The Western Chou empire, Chicago & London: University of Chicago Press, 1970, pp. 464-475.

[20] Presumably he means "accepted". There was of course no prescriptive grammar of classical Chinese.

[21] GSR = Bernhard Karlgren, "Grammata Serica recensa", Museum of Far Eastern Antiquities, Bulletin, 1957, 29: 1-332.

[22] Karlgren, "Glosses on the Ta Ya and Sung odes", Museum of Far Eastern Antiquities, Bulletin, 1946, 18: 1-198, gloss 859, p. 59.

[23] Shuihudi Qin mu zhujian, Beijing: Wenwu Chubanshe, 1978, passim.

[24] Ode no. 244; Karlgren, Book of odes, pp. 198-199.

[25] Noel Barnard, "A recently excavated inscribed bronze of Western Chou date", Monumenta Serica, 1958, 17: 12-46, pp. 37-39.