1. Regarding the key term Ru 儒, Lewis criticizes Hsu Cho-yun's understanding of that group (p. 73). In his Ancient China in Transition (Stanford: Stanford University Press, 1965), 100–103 Google Scholar, Hsu wrote that the Ru had emerged for four principal reasons: first, “to meet the demand for training the new administration experts and strategists”; second, to provide a “liberal education”; third, to calm the fears of “rulers who had reason to be uneasy about unethically ambitious students”; and fourth, to open up “a new and lasting path by which any low-born but able young man could gain high office by his own competence.” Lewis (p. 74) believes that “almost every proposition” in Hsu's account is wrong, and one need only think of Gongsun Hong公孫弘 to know that not all Ru were selected for their ethical probity. At the same time, Lewis leaves the reader wondering what he himself has in mind when he speaks of “exceptional cases distinct from the careers of most schoolmen” (p. 74). Lewis also approaches Hsu Cho-yun when he writes that “the schoolmen claimed superiority over those in state service through their condemnation of venal interests.” (p. 79). For an alternate understanding of the term Ru, see Nylan, Michael, “A Problematic Model: The Han Orthodox Synthesis, Then and Now,” in Imagining Boundaries: Changing Confucian Doctrines, Texts, and Hermeneutics' ed. Chow, Kai-wing, Ng, On-cho, and Henderson, John B. (Albany: State University of New York Press, 2000), 17–56 Google Scholar.

2. I myself am increasingly uncomfortable with the rubrics “Legalist” and “Daoist” (which are more useful, admittedly, than “proto-Daoist”), but I confess to finding no better way to refer to the thinkers involved, except by their individual names, which is how early thinkers in most cases referred to themselves and others, as is demonstrated in Mark Csikszentmihalyi and Michael Nylan, “Constructing Lineages and Inventing Traditions in the Shiji” (forthcoming). See also nn. 14,17-19, and 23 below. Lewis seems to equate “philosophical” with “text-based,” and to regard the technical traditions as experientially based. He argues that the “philosophical traditions' claim to superiority over the technical ones was based on their supposed possession of an encompassing, generalizing wisdom” (p. 279). However, Jiegang's, Gu 顧額剛 classic work, Handai xueshu shilue 漢代學術史略 (repr., Taibei: Qiye, 1980)Google Scholar, Kanaya Osamu 金谷治， Shin Kan shisō kenkyū 秦漢思想研究 (repr., Kyoto: Heiranku, 1960), and Anne Cheng, “Intellectual Self-awareness in Han Times” (unpublished paper delivered at the April 1995 Association for Asian Studies national meeting in Washington, D.C.), all challenge us to examine the continuities between the “philosophers” or Ru and the fangshi 方士 in ways that Lewis fails to do.

3. The seductive quality of any notion of early orthodoxy as robust as Lewis's is worrisome to the degree that it appears to obviate the necessity for historians to remain alert to alternate realities when faced with the steadily growing piles of liter-ary and archaeological data to be sifted through and analyzed. The fact that Mark Lewis comes close to making his grand construction is testament to his own solid scholastic achievements, which are evident on every page, but also to the reader's propensity to prefer manageable simplicity to the inconvenient messiness of historical realities.

4. Lewis seldom distinguishes between general critiques of the state and specific critiques of a particular regime or an unjust ruler. Many critiques of an evil regime or ruler ultimately celebrate the well-ordered state. In his discussion of the Odes, Lewis fails to account for the fact that many of the individual odes celebrate the state, so intent is he upon adopting the Han scholastic reading cited in the Mao commentary. particular ruler in time, it by no means queries the basic institutions and modes of operation of the state.

5. For the argument against our modern preoccupation with the unique personal voice, see Nylan, Michael, “Individualism and Filial Piety in Han China,” Journal of the American Oriental Society 116 (1996), 1–27 CrossRefGoogle Scholar. Of course, many early texts speak of words as the “expression of oneself,” but those same texts generally emphasize commonalities in human experience. See, e.g., Guoyu 國語, in Guoxuejiben congshu 國學基本叢書 (Taibei: Wenhua, 1968), vol. 381,11.141–42Google Scholar (Jinyu 晉語 5). Part of my argument on this particular topic has been informed by Connery, Christopher, “Jian'an Poetic Discourse” (Ph.D. diss., Princeton University, 1991)Google Scholar.

6. For anticipated blessings, see Poo, Mu-chou, In Search of Personal Welfare: A View of Ancient Chinese Religion (Albany: State University of New York Press, 1998)Google Scholar.

7. The modern scholar's search for such an intellectual space responds, of course, to Hegel (1770-1831), who argued that the ancient Chinese lacked “the free space of inner reflection.” See Harbsmeier, Christoph, Science and Civilisation in China, vol. 7, pt. 1, Language and Logic (Cambridge: Cambridge University Press, 1998), 25–26 Google Scholar.

8. I query the applicability of the phrases “prepolitical” and “protokingship” to “an age of Fu Xi,” who was a legendary figure. According to legend, Fu Xi ruled with a full complement of officials, so I wonder whether Lewis refers here to the (highly controversial) attempts made by some Chinese archaeologists to impose a Xia dynasty and the Marxist schema of historical stages upon early Shang and pre-Shang sites. By “dynastic state” (see below), for example, Lewis appears to mean, “hereditary slave society.” Here, as in a few other passages (e.g., p. 233 on Confucius), Lewis comes perilously close to adopting as “history” the multiple fictions of the works he treats. In fact, whenever Lewis speaks of the “prepolitical,” of “protokingship,” or of the birth of the “dynastic state,” it is unclear whether he speaks of actualities (known primarily through archaeology) or the stuff of legends. If the latter, it would be helpful to know why the Xia and Shang states (if not the pre-Xia) are not considered dynasties. Sometimes Lewis's well-meaning attempts to synthesize all available secondary literature on his chosen subjects continue and so compound earlier mistakes in the secondary literature.

9. Different Han commentaries do not all agree in their ascriptions to authors of the various layers of the Yijing, though many do follow the “Xici” in this matter.

10. Lewis writes: “Fu Xi is not only the first of the sages, but also an encompassing figure whose discovery of the trigrams contains the work of all the sages. He directly contemplated natural patterns, whereas all later inventions—including his own— were inspired by the hexagrams created in the primal revelation. … This primal discovery, the root of civilization, is ultimately identified as the beginning of writing” (p. 198). According to Lewis, Cang Jie is merely a minor double of Fu Xi, a figure whose chief interest rests in the fact that he is a minister rather than a ruler. (This ignores Cang Jie's central role in connection with writing, as attested from the Warring States period, in Xunzi yinde 荀子引得 Harvard-Yenching Sinological Index Series, Supplement No. 22 [repr., Taibei: Chengwen, 1968], 81/21/58.) Of course, Lewis's three-phase schema would be weakened by a fuller discussion of Cang Jie, as the existence of ministers certainly implies the existence of more than a “prepoliticai” organization in high antiquity. The negative associations of Cang Jie's invention (the ghosts weep at the invention) are also damaging to Lewis's hypothesis.

11. Lewis, following an observation made earlier by Edward Shaughnessy, tells us that “the Duke of Zhou does not figure prominently in the records left by bronze inscriptions, and his rise to prominence comes in association with the increasing importance of written texts” (p. 210). It is surprising how few texts until Eastern Han con-cern the Duke of Zhou, as the duke certainly becomes the chief figure in many later texts. As it is natural to assume that a regent will be tempted to keep all power for himself, perhaps the taint of treason lingered on in too many traditions. See Allan, Sarah H., The Heir and the Sage: Dynastic Legend in Early China (San Francisco: Chinese Materials Center, 1981), 118–21Google Scholar. The few interesting articles of recent date on the duke include Shaughnessy, Edward L., “The Duke of Zhou's Retirement in the East and the Beginnings of the Minister-Monarch Debate in Chinese Political Philosophy,” in his Before Confucius: Studies in the Creation of the Chinese Classics (Albany: State University of New York Press, 1997), 101–36Google Scholar; a series of four posthumous articles from Gu Jiegang, to be found in Wenshi 文史 22 (1983), 23 (1984), 26 (1986), 30 (1988), and 31 (1988)Google Scholar, the most provocative of which is “Zhou gong zhizheng Chengwang” 周公執政成王， Wenshi 23 (1984) 1–30 Google Scholar; and Mashima Jun'ichi間島澗一, “Shūkō hikyo setsu shōkō 〃周公避居說小考, Chūgoku bunka 中國文化 (Tsukubo daigaku) 56 (1998), 1-13.

12. For Lewis, this portrayal of ritual in the Zuo is generally confirmed by the Gongyang, though the latter text seeks to convince its readers that the Chunqiu's preferred mode is to assign meaning chiefly through word choice rather than through the choice of events narrated. From this, Lewis concludes that the Gongyang “implies that texts, above all the Chunqiu,” were “the sole locus of kingship and political author-ity in the Eastern Zhou” (p. 146). Lewis's propensity to use absolutes and dichotomies in analysis does not serve him well here, for it undercuts the important point he has just made: that correct ritual behavior is perceived as the chief way to attain and retain charismatic authority.

13. Of course, many of Lewis's observations linking poetic practice and assertions of personal virtue might extend to the Zhou bronze inscriptions, given their intertextuality and the bronzes' increasing regularities of rhyme and meter.

14. In placing early texts within an intellectual context, Lewis likes to contruct genealogies. See, for example, p. 21, where the legal texts of Yunmeng “inherit” the role of the documents cast on Zhou funerary bronzes and are “successors” of the Jin covenants. Most of these documents are categorized in terms of “schools,” often as “proto-Daoist” (a rubric never adequately explained). This becomes important when Lewis later defines part of the Yijing as a “ru response to proto-Daoist texts” on p. 241.

15. Even the legendary Duke of Zhou's writing “fails to achieve its goal” in combatting slanderous opponents (p. 214). As for Confucius, as Lewis himself demonstrates, Han authors tend to depict the Sage most often as brilliant official or potential official, rather than as scholar-author (p. 221). In Han, “none of the leading scholars from the text-based traditions gained a significant office” (p. 74). Scholarship, he avers, was “not primarily a road to office” (p. 75). (And classical learning seems to have been less germane than the attainment of regular bureaucratic positions when it comes to deciding how a man's life was to be presented for commemoration.)

16. Many who prided themselves on their mastery of the written word were accused of propounding “empty words and vain phrases,” even when their efforts, like those of Sima Tan, Sima Xiangru, and Yang Xiong 揚雄, were lucky enough not to be classed with those of the court entertainers. For the low position of fu writers whom Lewis portrays as authorities, see Kechang, Gong, Studies on the Han fu (New Haven: American Oriental Society, 1997), chap. 1Google Scholar. Sima Qian plainly says that he was “among the lower officials and participated in the lesser deliberations of the outer court” ( Hanshu 漢書[Beijing: Zhonghua, 1962], 62.2728)Google Scholar, and his father, Sima Tan “was certainly kept for the sport and amusement of the emperor, treated the same as the musicians and jesters and made light of by the vulgar crowd of the day” (Hanshu, 62.2732). Watson, Burton, Ssu-ma Ch'ien: Grand Historian of China (New York: Columbia University Press, 1958), 88 Google Scholar, lists a number of passages from the Liishi chunqiu 呂氏舂秋 and Shiji 史言己 which speak of “empty words” or “empty sayings”; to his list could be added mul-tiple references from the Hanfeixi 韓非子 and several from Zhuangzi 莊子. Christoph Harbsmeier, in the introduction to chap. 30 of his forthcoming translation of the Hanfei (Yale University Press), notes that, “Any philosophy was only seen as relevant… insofar as it could be applied meaningfully to … the consideration of the historical heritage.… The truth of a philosophical point was in its … relevance to a proper understanding of history.”

17. Lewis defines the Mohists by their attention to duty, righteousness, and warfare (p. 68) and the proto-Daoist (those that “ultimately emerged as Daoism”) at once as “textual traditions formed around writings” and “meditation-based schools” (p. 69).

18. Lewis's “schoolmen” are an ill-defined bunch, at any rate. At one point they consist only of the Ru and the Mohists, but in other passages the term seems to embrace any literate professional and perhaps other persons as well. Similar confusion attends a few other claims related to writing's unparalleled authority, for example: “Without the text, there was no master and no disciples (beyond the lives of the individuals involved)” (p. 58). While this seems obvious upon first reading, does it mean that Confucius was not a master until the earliest stratum of the Analects or the proto-Analects was composed?

19. Other shifting translations are offered by Lewis: he belatedly notes, for example, that the jing 經 (“canons” or “constants” that he has hitherto identified as “texts” may refer to “disciplines” as well (p. 233). For the definitions of liuyi 六藝and wen/shi 文實, see nn. 100 and 105 below. It seems that for Lewis, references to “study,” “learning,” “techniques” (shu 術), and shu 數 (which he translates in connection with the Yijing as “[written] numbers,” but which often represents the “cosmic regularities”) all describe written texts, as on p. 71, where “the methods [shu 術 ]of Lord Shang and Guan Zhong” are reduced to “legalist writings,” and p. 292, with an analysis of the Xunzi quote. If sayings or practices were ultimately transcribed in the course of time, does that constitute evidence in favor of writing's great authority? Does the author-ity of a text once written down significantly differ from what it was when it was only available in an oral version? Lewis dismisses such interesting questions, which have been taken up with remarkable force and vigor by students of classical Greece, Rome, medieval learning, and modern linguistics.

20. One recent study connecting the style and content of oral performance texts (the fu) to the style and content of prose writings is Taniguchi Hiroshi 谷ロ洋, “E'nanji no bunji ni tsuite” 淮南子の文辭について, Nippon Chūgoku gakkaihō 曰本中國學4 報 47 (1995), 17–32 Google Scholar. The centrality of ritual, rather than writing, seems to be attested by the Zuo: “Rites form the great warp of kingship. With one action that goes against the rites in two ways, the great warp is no more. Words should establish rules. Rules should make manifest the warp of rites. The warp of rites is forgotten behind many words. What use is it then to elevate records and documents?” (Chunqiu jingzhuan yinde 春秋經傳引得, Harvard Yenching Sinological Index Series, Supplement No. 11 [Peiping: Harvard Yenching Institute, 1937], 389/Zhao 15/fu 3).

21. Though it might be asked how such an educational program is related to an oppositional stance, Lewis is right to point out the oppositional features of much of the writing of the period, even if he exaggerates their importance.

22. Of xing ming 形名, Lewis writes, “These ‘names’ were originally oral, but since they were to serve as a standard for judgment in the future, they must have been set in writing” (p. 33).

23. On the other hand, Lewis's characterization of the Lushi chunqiu and Huainanzi 淮南子 as “state-sponsored compendia” (p. 302) is questionable, for the works can easily be read, in company with earlier texts, less as a “re-assertion of the intellectual hegemony of the state” (p. 303) than as ambitious assertions of charismatic authority by their sponsors, two persons near the throne. G riet Vankeerberghen, for example, details the attributes of perfect sagehood that Liu An劉安hoped his young nephew, Emperor Wu, would discern in him; see Vankeerberghen, , The Huainanzi and Liu Art's Claim to Moral Authority (Albany: State University of New York Press, in press)Google Scholar. Lewis discounts the suggestion that Lü Buwei 呂不韋 or Liu An plotted rebellion, insisting that both books were probably written to guide the conduct of a young ruler (p. 303).

24. Lewis inexplicably slots the writings of military commanders on campaign under this overly broad rubric of “oppositional texts,” in company with those of his surprisingly subversive “Legalists” (see n. 19 above). He also writes of musical performance as “an act of defiance” (p. 230), setting aside early Chinese notions of music as a centering device for the unsettled psyche.

25. On the subject of rhetorical ploys, one might name the numerous writings on pleasure in Warring States, which have been touched upon in Vankeerberghen, Griet, “Emotions and the Actions of the Sage: Recommendations for an Orderly Heart in the Huainanzi ”, Philosophy East and West 45 (1995), 527–44CrossRefGoogle Scholar; and Schaberg, David, “Social Pleasures in Early Chinese Historiography and Philosophy,” in The Limits of Historiography: Genre and Narrative in Ancient Historical Texts, ed. Kraus, Christina Shuttleworth (Leiden: Brill, 1999), 1–26 Google Scholar; and which are the central subject of Michael Nylan, “The Politics of Pleasure in Warring States and Han” (forthcoming).

26. On writing in general, see Goody, Jack, The Logic of Writing and the Organization of Society (Cambridge: Cambridge University Press, 1986)CrossRefGoogle Scholar; and Harris, Roy, The Origin of Writing (La Salle: Open Court, 1986)Google Scholar. My comments do not in any way endorse the views of certain archaeologists of China who would find “proto-writing” in clan emblems and such. Our earliest attested sample of writing in Chinese, the oracle bone inscriptions, is: (1) manifestly not the earliest Chinese writing, and (2) in all likelihood not the only writing of its time. I refer readers to Boltz, William G., The Origin and Early Development of the Chinese Writing System (New Haven: American Oriental Society, 1994)Google Scholar; and Robert W. Bagley, “The Earliest Known Chinese Writing: Bias of the Sample and Archaeological Context” (unpublished paper delivered at a conference on the origins of writing at Brigham Young University, April 2000). In the latter, Bagley notes that “the suggestion that writing was invented to serve ritual purposes has no basis apart from the claim that at Anyang it was confined to ritual, which is unlikely to be true,” Lewis, by contrast, dismisses the idea that writing in late Shang had important functions in the round of daily activities outside the kingly court and cult: “Some scholars speculate that the earliest script forms had been developed for use in daily activities, but that the evidence of this has vanished with the perishable materials to which such writings are committed. In fact, the early graph forms are clearly tied in form and significance to divination.… Whatever other roles writing played in Shang times, it was in the inscription of the religious activities of the rulers that the graphs found their definitive import” (p. 15). I am unsure why Lewis insists that one early use of writing was more definitive than another.

For the role of writing in other early civilizations, readers may consult Maisels, Charles Keith, Early Civilizations of the Old World: The Formative Histories of Egypt, the Levant, Mesopotamia, India, and China (London: Routledge, 1999)Google Scholar.

27. See Beard, Mary, “Ancient Literacy and the Function of the Written Word in Roman Religion,” Literacy in the Roman World (Journal of Roman Archaeology, Supplementary Series No. 3), ed. Beard, Mary et al. (1991), 35–58 Google Scholar. Two of the best contemporary descriptions of heterodox cults, both in Ying Shao's 應邵 Fengsu tongyi 風俗通義, fail to mention inscriptional materials, though they mention wu 巫 magicians, raising the dead, the slaughtering of sacrificial animals, and the indiscriminate mixing of men and women in cui tic songs and dances. See Fengsu tongyi tongjian 風俗通義通檢, Centre Franco-Chinois d'Études Sinologiques Index No. 3 (repr., Taibei: Chengwen 1968), 9.67-68 (“Chengyang Jingwang si” 城陽景王祠 and 9.70 (“Baojun shen” 鮑君神). Of course, the argumentum ex silentio is never strong enough to prove the presence or absence of such activities, but it is possible that, aside from rituals of the ruling house, inscriptionai materials were not commonly in use in connection with more ordinary cultic activities until calligraphy itself became “divine.” See n. 43 for further information.

28. Sociopolitical changes in Eastern Han are largely attributed by Lewis to the retraction of “bureaucratic control… with the abandonment of compulsory military service and of restrictions on ownership of land,” which made the state's influence in local communities increasingly dependent “on the loyalty of a new sort of elite” who “combined imperial service, large-scale landownership, and trade” (p. 10). The state's increasing dependence on cultural and literary models over and against institutions for war, which constituted a decided turn, presumably, away from sanctioned violence, occasioned an unprecedented “triumph” of the canon associated with Confucius (p. 10). See also p. 337, where Lewis talks again of the “transformation of the material bases of the Han state through the rise of landlordism and the abolition of universal military service.” The victory of Lewis's “Confucianism” then “abolished the tension between competing frames” (p. 337) and “justified the abandonment of the Warring States political legacy” (p. 338). In essence, “The tendency to compose and read texts as alternate or parallel states and to ground political programs in an imagined antiquity led to a total reevaluation of the past in the service of the new state” (p. 338). These are sweeping historical judgments, which are announced but not expanded upon or substantiated.

29. The question “What was the main form of textual unit operating in pre-Han and Han?” has seldom been asked by classicists in our field; when it is, answers to that question will no doubt radically alter our perceptions of the ancient modes of transmission of knowledge. When Lewis supplies a date for a received text, he generally ignores the complex issues surrounding dating. It is nearly certain that most of the early texts he surveys, which he would call “philosophical,” first circulated in parts, which were only later made into the larger received texts that we now know. The original date of composition for each part, therefore, seldom coincides with the final date of compilation. Nor does Lewis alert the reader to the fact that certain writ-ten texts he reviews (e.g., the Shiji and Lunheng) did not circulate widely immediately after their initial compilation.

30. I am also concerned that texts from widely different areas, even long before the imperial period, are said to relate to “China,” an entity of modern construction, though I acknowledge the difficulty of continually reminding the reader of the fragmented state of politics and consciousness in “the geographic area that is today known as China.”

31. For reasons that elude me, Lewis would have the reader starkly contrast the function of the Odes, as presented in late Warring States disputations, with its use in mid-Western to Eastern Han. After all, throughout the period under review, the texts emphasize “the role of poetry as evidence of historical realities” (p. 165). And it is Lewis who tells us of the ongoing belief that the songmakers of old were driven to compose their texts by the moral and political decay they had witnessed—views that reappear time and time again in pre-Han and Han texts, including Sima Qian's “Letter to Ren An” (Hanshu, 62.2724–36) and the “Great Preface” to the Odes.

32. The most useful analyses of writing's authority in other early societies include Thomas, Rosalind, Literacy and Orality in Ancient Greece (Cambridge: Cambridge University Press, 1992)CrossRefGoogle Scholar; Harris, William V., Ancient Literacy (Cambridge: Harvard University Press, 1989)CrossRefGoogle Scholar, cited repeatedly below; Mary Beard et al., Literacy in the Roman World (cited in n. 27); Gledhill, John, Bender, Barbara, and Larsen, Mogens Trolle, eds., State and Society: The Emergence and Development of Social Hierarchy and Political Centralization (London: Unwin Hyman, 1966), pt. 3Google Scholar, “The Role of Writing and Literacy in the Development of Social and Political Power”; and Martin, Henri-Jean, The History and Power of Writing, trans. Cochrane, Lydia G. (Chicago: University of Chicago Press, 1988)Google Scholar.

33. Specifically, why did those who commissioned the bells seek to persuade a group of readers, as yet unidentified, of the perfect comparability between the Zeng state musical scale and those of Zeng's neighbors? See Bagley, Robert W., “Percussion,” in Music in the Age of Confucius, ed. So, Jenny F. (Washington, D.C.: Sackler Gallery, 2000), 60 Google Scholar.

34. See Kroll, Juri L., “Correlative Thinking and the Histories of Ssu-ma Ch'ien and Pan Ku,” Istoriya i arkheologiya Dal'nego vostoka K-70-letiyou E.V. Shavkunova (Vladi-vostok: Far Eastern University Press, 2000), 53–71 Google Scholar.

35. Dan Zhu, cited by Lu Chun 陸淳 (ca. eighth century) in Chunqiu jizhuan zuanli 春秋集傳慕例 (Siku quanshu 四庫全# ed.), 146.380-81. Dan is discussing the compilation of the Zuo from many early sources. He therefore talks of specifically “arranging them [the texts] by months and years,” but the passage broadly relates to all the main exegetical traditions.

36. Lewis's acknowledgement that the “texts mentioned … under the rubric jing were all recent compositions” (p. 301) tallies ill with his treatment of them elsewhere in the book as more ancient written texts. Note that when I speak of common wisdom, I consider Michael Loewe an exception. The acknowledged leader of Han studies, Loewe has, in every essay he has written, carefully distanced himself from the standard views in the field. Beginning at least with Crisis and Conflict in Han China (London: George Allen and Unwin, 1974)Google Scholar, he has avoided the notion of “schools”; eschewed talk of “Confucians” and “Legalists” as opposing advocates favoring ritual or law, respectively; and resisted the notions that the Han scholar-officials were uniformly “humanistic” (i.e., secular) in their orientation. This essay therefore owes a great deal to him, though he is in no way responsible for its errors.

37. Lewis also follows conventional wisdom on the Qin, emphasizing a few key events, including unification itself, the Burning of the Books, without mentioning an earlier biblioclasm attributed to Shang Yang 商鉄 in 350 B.C. by Hanfei, as cited in Wang Xianshen 王先愼, Hanfeizi jijie 韓非子集解 (Taibei: Shijie, n.d.), 13.67.

38. Lewis posits a shift “from an early reliance on face-to-face transmission from teacher to disciple outside the state sphere toward a transmission based primarily on texts, debates between philosophic traditions in the courts, and patronage by leading political figures” (p. 302). It is Lewis's belief that, “As texts became more numerous and available in the late Warring States period, commitment to a master or group was increasingly replaced by attachment to texts.” According to Lewis, by the mid-fourth century B.C., “the importance of direct teacher-disciple contact declined, and texts became the primary means of communicating doctrine” (p. 5). The supposition that masters became mere authors at one particular point in history, rather than repeatedly, not long after the death of each lineage master, seems to ignore the evidence. Lewis goes well beyond the present evidence, as I understand it, when he regards the existence of several distinct layers in some famous texts (especially the Zhuangzi 莊子 and the Mozi 墨子) as sufficient proof that “intellectual commitments were no longer defined by loyalty to the doctrine of a single teacher” (p. 6). I find no evidence of such a thoroughgoing shift. See n. 2 above for further information.

39. The longstanding account of textualization in Han often presupposes a surge in literacy rates in response to the throne's encouragement of literary learning. Typically, when historians suggest an increase in literacy in Han, they point to a single sentence reiterated in the Han records. The sentence says that when Gongsun Hong “became prime minister and enfeoffed marquis on account of his specialization in the Chunqiu” there and then “the learned wellborn men in the empire were to a man influenced by this” (Shiji, 121.3120; see also Hanshu, 88.3593). This phrase is usually taken to mean that men came to see that the best way to secure fame and fortune for their families was to give their sons a classical education. Presumably mastery of one classic and its interpretations could become the road to high rank and wealth for some, especially under those emperors who particularly liked elegant literary flourishes and harsh punishments for traitors. (The Gongyang masters for the Chunqiu were known to specialize in providing canonical justifications for executing suspected rebels among the aristocratic class.) However, the ability to read and write a classical text could gain a person great status in Han only to the degree that it was a highly unusual skill. Therefore the brief statement about Gongsun Hong hardly constitutes convineing proof of any dramatic rise in literacy. There is little evidence of state support for schooling (despite frequent pious talk about the local schools of ideal antiquity), and all the scanty evidence that exists concerns the state's support for what we might call tertiary education. Primary education, as far as we know—teaching young children the basics of reading and writing—remained the private responsibility of each family, and this means that literacy rates could not have increased much beyond the pre-Han figures, when literacy was generally limited to skilled workers and land-owning elites. The Qin and Han states, which so deftly used writing to increase their prestige, made few state-sponsored initiatives for the promotion of literacy. McKitterick, Contrast R., The Uses of Literacy in Early Medieval Europe (Cambridge: Cambridge University Press, 1990), 324 (Conclusion)CrossRefGoogle Scholar; and Yu Shulin 余書麟, “Liang Han sixue yanjiu” 兩漢私學硏究, Shida xuebao 師大學報 11 (1966), 1–39 Google Scholar.

40. Harris, Compare, Ancient Literacy, 89 Google Scholar.

41. One thinks of the First Emperor of Qin's 秦始皇 stelae; also of the stele set up on the peak of Mt. Tai 泰山 in A.D. 56 to commemorate Emperor Guangwu's 光武帝 achievements.

42. Harris, , Ancient Literacy, 11 Google Scholar.

43. See Nylan, Michael, “Calligraphy: The Sacred Test and Text of Culture,” in Character and Context in Chinese Calligraphy, ed. Ching, Dora and Liu, Cary (Princeton: The Art Museum, 1999), 142 Google Scholar.

44. Kern, Martin, The Stele Inscriptions of Ch'in Shih-Huang: Text and Ritual in Early Chinese Imperial Representation (New Haven: American Oriental Society, 2000), 191 Google Scholar, writes that the Qin state intended the University and Academicians to narrow the scope of authoritative learning, so as to do away with the troublesome baijia yu 百家語. Nathan Sivin and Geoffrey Lloyd, in their forthcoming The Way and the Word, argue that jiafa 家法 was designed to halt the proliferation of different interpretive traditions. Here I would emphasize that, if such institutions were so designed, they appear to have been ineffective.

45. Centuries before Han, a partial synthesis had emerged among prominent War-ring States thinkers typically depicted in modern studies as arch enemies—those people we now call (quite anachronistically) the Legalists, Daoists, Mohists, Confucians, and Yin-yang jia—though consensus was limited to certain key assumptions about cosmic law, the operations of the human body and the body politic, and definitions of the Good (see n. 51). It was basically this limited synthesis that would prevail through Han times. One searches in vain in the extant accounts of Han court conferences—in fragments collected in Ma Guohan's 馬國翰 Yuhan shan fang ji yi shu 玉函山房輯佚書 (Jinan: 1871; repr., Taibei: Shijie, 1967), as well as in Ban Gu's 班固 Bokutong 白虎通 (Po hu t'ung: The Comprehensive Discussion in the White Tiger Hall, trans. Tjan Tjoe Som [Leiden: Brill, 1949-52])—to find some new points of consensus. We can credit Han only with the greater incorporation of Five Phases theory into state discourse.

46. Shiji, 97.2692 (Liu Bang 劉邦 pissing in the hat); and 121.3127-28 (Dong Zhongshu). We know only that Dong was eventually pardoned by imperial edict, and after that, “Dong no longer dared to express his opinions on disasters and portents, but… devoted himself entirely to studying and writing texts.”

47. Shiji, 28.1395. Emperor Wu admired the Ru chiefly for their elegant airs and fancy phrases; in court rituals, he “wished rather to choose from the Ru methods in order to make them [the proceedings] more elegant” (Shiji, 28.1397). The Ru, for their part, were unable to produce any detailed information on the feng 封 and shan 禪, because they insisted upon confining themselves to precedents mentioned in the Odes and Documents; hence, the charge that they were incapable of coming forward with any worthwhile suggestions (Shiji, 28.1397). Apparently, Qin and early Western Han emperors were not intimidated by this growing class of learned advisors. They did not forbid the Yantielun's 鑒鐵論 account, which shows one or more wenxue 文學 battling it out with government spokesmen—and winning. To cite a later example, that of Emperor Zhang 章帝 of Eastern Han (r. A.D. 76-88), less than a decade after he convened “all the classicists” at the White Tiger Pavilion in A.D. 79, he taunted the scholastics with a classical allusion conveying his annoyance that their protracted deliberations never seemed to spark any constructive activity: since “a single [Music Master] Kui was enough for [the sage-ruler] Yao,” he wondered why so little had been accomplished by the large band of ritual experts in his own service; see Hou Hanshu 後漢書 (Beijing: Zhonghua, 1965), 35.1023 Google Scholar. The list of Han emperors actively hostile to classical and pseudo-classical models must include Emperors Wen 文, Jing, Xuan 宣, Cheng 成, Zhang (during the later part of his reign), and An 安 as well as most of the late Eastern Han emperors after Emperor Shun 順 (r. 125-44).

48. One might even include as “working under a cloud at court” Dong Zhongshu, whose writings were attacked by Emperor Wu and by his own student; see Hanshu, 56.2524-25.

49. Hou Hanshu, IB50, 66-67. Compare Cambridge History of China, 619. Guangwu did execute several officials for submitting fraudulent registers, but this had no appre-ciable effect on the process.

50. Yang Xiong criticizes a preference for primitivism in Fayan 法言 1.2, for example, but this idea had been taken up by many.

51. Note that it is not entirely clear whether Lewis would see the Five Classics as the common intellectual heritage of many thinkers. On the partial pre-Han synthesis, see Yates, Robin D.S., “Body, Space, Time, and Bureaucracy: Boundary Creation and Control Mechanisms in Early China,” in Boundaries in China, ed. Hay, John (London: Reaktion Books, 1994), 56–80 Google Scholar; Sivin, Nathan, “State, Cosmos, and Body in the Last Three Centuries B.C.,” Harvard Journal of Asiatic Studies 55 (1995), 5–37 CrossRefGoogle Scholar; and Michael Nylan, “Boundaries of the Body and Body Politic in Early China,” to be published by Princeton University Press, the Carnegie Council, and the Ethikon Institute.

52. See Xunzi yinde 15/6/6-16/6/20 for Xunzi's excoriation of Mo Di 墨翟 and Song Bing 宋鉼, Shen Dao愼到, Tian Bing 田駢, Hui Shi 惠施, Deng Xi 鄧析, Zisi 子思， and Mencius, among others.

53. Of course, to some degree talk about the state's desire to promote “broad learning” and its announced intention to “weave a tight net [to capture] omissions and lacunae” that had arisen over the course of centuries was offset by the state's practical need to cap the number of interpretive traditions accorded formal support. For “broad learning” as the indispensable condition of the professional classicist, see Liji 禮記, “Ru xing” 儒行,in Liji yinde 禮記引得, Harvard-Yenching Sinological Index Series No. 27 (Peiping: Harvard-Yenching Institute, 1936), 41/13. For “weaving the net,” see Hanshu, 88.3621; and Hou Hanshu, 79A.2546.

54. At the beginning of Han, after centuries of war, Han classicists bemoaned the faulty nature of their two sources of authority: key texts were missing or nearly undecipherable, while key portions of the ancient systems of “rites and music,’ were irrecoverable. Over the course of the Western and Eastern Han dynasties, Han classicists were to see further dramatic losses of respected “traditions” in the wars attending the restoration and downfall of the Han, losses that made it all the harder to ascertain, let alone adhere to the Way of the Ancients. In response to these perceived gaps in classical knowledge, the process of gathering, compiling, interpreting, and inventing new “classical” materials proceeded apace in Han, but always with referenee to the evolving sociopolitical, economic, and cultural scene.

55. Nylan, , “A Problematic Model,” esp. 28–32 Google Scholar.

56. Hanshu, 56.2523; trans, after Loewe, Michael, Divination, Mythology and Monarchy in Han China (London: Cambridge University Press, 1994), 16 Google Scholar, with minor modifications.

57. Fengsu tongyi, 4.31.

58. Shuowen jiezi 說文解字 (Hong Kong: Taiping, 1966)Google Scholar, 15A.4a (postface), as trans, in Thern, K.L., Postface of the Shuo wen chieh tzu, the First Comprehensive Chinese Dictionary (Madison: University of Wisconsin, 1956), 17 Google Scholar. Compare Xu Fang's 徐防 famous contemporary statement, from a memorial of A.D. 103, which insists that even the “Academicians and students in the Academy… all speak from their own minds, with none following the authority of a school's teaching” (Hou Hanshu, 44.1500). For similar comments, see Hou Hanshu, 35.1213, the appraisal of Zheng Xuan 鄭玄； and Hou Hanshu, 79A.2547, giving a general overview of Eastern Han Ru studies.

59. Fengsu tongyi, 5.35. Zhouyi yinde 說文解字, Harvard-Yenching Sinological Index Series, Supplement No. 10 (Peiping: Harvard-Yenching Institute, 1935), 46/Xi B/3, talks of “the same goal, [reached by] different paths; the single focus, [reached] by a hun-dred thoughts.” See also Shiji, 121.3124, where the different interpretations all “return to one”; and Hanshu, 22.1027, for a variation on this with respect to the Six Classics. Xu Gan徐幹in Eastern Han is typical of those who offer praise for those who can see what is common to different categories (Zhonglun 中論, chap. 8). Compare Huainanzi, in Wendian, Liu 劉文典， Huainan honglie jijie 淮南子鴻例集解 (Beijing: Zhonghua, 1989), 2.56 Google Scholar: “The hundred jia each have different theories.… Now in terms of their way of governance, Mo, Yang, Shen [Buhai], and Shang [Yang] each have one rib of the carriage canopy or spoke of the wheel. If you have them all, you possess the correct number, but if you are missing one, the whole thing does not become use-less” (trans, after Mark Csikszentmihalyi, private communication, following Wang Niansun 王念孫).

60. See Hanshu, 1.313, in connection with the taixue 太學and the boshi 博士. See also Hanshu, 88.3598, which tells us that the classicists' job is to lun tong yi 論同異 (“assess similarities and differences”), in hopes of tracing lines of thought back to their common source.

61. This information can be gotten by comparisons of Wang Guowei 王國維, Han Wei boshi timingkao 漢魏博士題名考, in Xueshu congbian 學術叢編 (Taibei: Yiwen, 1971), vol. 3 Google Scholar, with Yan Gengwang嚴耕望, Liang Han taishou cishi kao 兩漢太守刺史考 (Shang-hai: Shangwu, 1968)Google Scholar, and the three chapters devoted to the “Rulin” 儒林 (Forest of classicists) in the Shiji, Hanshu, and Hou Hanshu. Shiji, 112.2949, the biography of Gong-sun Hong, shows how lowly placed and easily dismissed the boshi were, at least at his time. Gongsun Hong after the age of sixty was twice appointed boshi, and each time asked to give advice on foreign affairs. Also, the placement of the “Rulin” chapters in the Shiji, Hanshu, and Hou Hanshu is highly suggestive, for the chapters follow others devoted to harsh officials, eunuchs, and such. Finally, if a man of learning held any important office, he was given a separate biography in another chapter, based on his bureaucratic rank; official standing consistently trumped classical learning in the offi-cial histories, as is clear from a listing of Shen Pei's 申培 students who held ranks as boshi and as provincial governors, commanders, and palace secretaries (Shiji, 121.3122).

62. See Fukui Shigemasa 福井重雅, “Rikukei rikugei to gokei: Kandai ni okeru gokei no seiritsu” 六經六藝と五經：漢代における五經の成立, Chūgoku shigaku 中國史學 4 (1994), 139–64Google Scholar; “Shin Kan jidai ni okeru hakushi seido no tenkai: gokei hakushi no setchi o meguru gigi sairon” 秦漢時代における博士制度の展開：五經 博士の設置をめぐる疑義再論, Tōyōshi kenkyū 東洋史研究 53 (1995), 1–31 Google Scholar. Fukui uses the argumentum ex silentio to argue that Emperor Wu never established or restored the office; I conclude instead their relative unimportance to grave matters of state. The date of 51 B.C., of course, is the date of the Shiqu 石渠 court conference convened on the classics, rites, and music; the Shiqu conference, known only from fragments, seems to have been very much like the White Tiger (Bohu) conference of A.D. 79 in its focus on rites and music.

63. Shiji, 121.3126; said of Xu Yan 徐延 and Xu Xiang 徐襄, grandsons of Master Xu of Lu, both of whom were reportedly experts in “demeanor” (rong 容). We are told that it was in this capacity that they were appointed to office; and they began a whole school of interpretation that focused on demeanor rather than textual interpretation. Hanshu, 88.3614, makes them experts not in rong but in song 頌 ‘praise-songs” as song were used in ceremonial toasts, by analogy with performances and inscriptions once dedicated to the ancestors.

64. Shiji, 121.3121.

65. Talks with Martin Kern have brought home to me the significance of calling Shusun Tong “the father of the Ru.” What we once took for irony we now take for an accurate, if despairing comment on the professional Ru.

66. See, e.g., Shiji, 121.3119; Hanshu, 88.3593-94. Hanshu, 88.3619, shows us that one successful test-taker at the Academy went on to be appointed zhanggu 掌古 assistant to the taishi 太史 “senior archivist.”

67. See Shiji, 121.3118-20, for the original proposal for the Academy students by Gongsun Hong. Note that the proposal suggests that students be tested on their recognition of Chinese characters; nothing is said about their understanding of the text's content. Unfortunately, we know little about the content of the tests administered at the Academy. It is not even certain that the tests were always written. For the word count for the Odes, see Wilkinson, Endymion, Chinese History: A Manual (Cambridge: Harvard University Asia Center, 1998), 468 Google Scholar, which also notes some 510 alternate characters. (For comparison's sake, the longest inscriptions known on the oracle bones are some 200 characters. The longest bronze inscriptions are no more than 500 characters: e.g., the “Mao gong ding” in the Palace Museum in Taiwan, which is now considered authentic by most specialists aside from Noel Barnard, has 497 characters; and the inscriptions on the Zhongshan 中山 ding and fanghu bronzes of ca. 314 B.C. have 469 and 450 characters respectively.) The Lunyu 論語has altogether only 15,883 characters in the whole text; the Documents, some 16,000. The Shiji is nearly as long as all Thirteen Classics put together (with 589,283 characters), and is thirty-three times longer than the Lunyu ( Wilkinson, , Chinese History, 22 Google Scholar). Shiji, 121.3122, says that Emperor Wu liked to have fine literary phrases justify his harsh policies; hence the rise of Gongsun Hong.

68. Crespigni, Rafe De “Political Protest in Imperial China: The Great Proscription of Later Han, 167-184,” Papers on Far Eastern History 11 (03, 1975), 1–36, esp. 15-17Google Scholar. According to de Crespigny, “this change in the significance of university exami-nations reflected a general decline in the university itself and in the orthodox tradi-tions of scholarship” (p. 16). Only briefly, after Emperor Shim's death in A.D. 144, did Dowager Empress Liang 梁后 order all officials of 2000-picul to 600-picul rank to send their sons to the Academy and introduce special examinations, so that some fifty to sixty students each year could gain entry directly from the Academy into the ranks of palace courtiers without nomination by their local authorities. The Proscrip-tions beginning in A.D. 167 soon made attendance at the Academy a liability, rather than an asset, in attaining high office. Bielenstein, Hans, The Bureaucracy of Han Times (Cambridge: Cambridge University Press, 1980), 140 CrossRefGoogle Scholar, by contrast, presumes no change in this respect from the Western Han system.

69. For the numbers of students at the Imperial Academy (taixue), see Tianlin, Xu 徐天麟, Xi Han hui yao 西漢會要 (Taibei: Shijie, 1971), 25.219–21Google Scholar; and Tianlin, Xu, Dong Han hui yao 東漢會要 (Taibei: Shijie, 1971) 11.111–14Google Scholar. One of many unanswered ques-tions concerns the exponential growth in Academy numbers, which could never have been contained in the buildings on their sites (see Hou Hanshu, 79A.2547). Most “students” must have been clients of the great families. It seems doubtful that the Academy could have offered anything like classes to all these men. Possibly the growth in numbers reflects, in this time of rapid reinfeudation, the desires of talented young men not in command of impressive resources to become the clients of capital officials.

70. See Hans Bielenstein, Lo-yang in Later Han Times, repr. The Museum of Far Eastern Antiquities 48 (1976), 69 and 71 Google Scholar, citing Hou Hanshu, 79A.2547.

71. Bielenstein, , The Bureaucracy of Han Times, 140 Google Scholar, says that “It looks as though even at this early time [i.e., 124 B.C., the time of the founding of the Academy] the examina-tions were set at three levels of difficulty, leading to corresponding levels of official employment.” By a refinement of A.D. 4, the forty best students were sent to serve in the imperial palace; the next twenty, to the heir's household; and the next forty were sent to the commanderies and kingdoms. “Only a fraction of students entered the bureaucracy from the Academy. All others had to strive for appointment as junior staff in the central and local administrations, for nomination as Filially Pious or Abun-dant Talent, or for recommendation by a friendly official, unless they were prepared to abandon hope of official employment” (The Bureaucracy of Han Times, 140—41). Regarding the literacy of the lowest level of the bureaucracy, Loewe, Michael, Records of Han Administration (Cambridge: Cambridge University Press, 1967), vol. 1, 16Google Scholar, notes that the vast majority of the Juyan strips inscriptions “were probably made by low-grade clerks and officials, whose education had been scanty and whose practice at writing may have been somewhat limited. It is therefore not surprising that in many cases, routine reports tend to follow regular well-established patterns and that a mini-mum of initiative was displayed or expected.” Some strips specifically mention mili-tary officers in their thirties who are described as “capable of writing, keeping accounts, administration both within the houguan and over civilians; [having] good knowledge of statues, ordinances and military matters,” and the like ( Records of Han Administration, vol. 2, 179 Google Scholar). This review does not go into the complicated question of legal documen-tation, discussed well in Hulsewé, A.F.P., “A Lawsuit of A.D. 28,” in Studia Sino-Mongolica: Festschrift für Herbert Franke, ed. Bauer, Wolfgang (Wiesbaden; Franz Steiner Verlag, 1979): 23–24 Google Scholar.

72. Hou Hanshu, 48.1606. The Supreme Commander Zhao Xi 趙喜 dissuaded Emperor Ming from this course of action, but it is suggestive that the foremost patron-emperor of the Ru thought the functions of the Academy could be integrated and absorbed into the new ritual center. Under Emperor Shun, a teacher famed for his “knowledge of rites and music” was appointed to oversee the taixue (Hou Hanshu, 79A.2557).

73. In A.D. 131-32, six years after Emperor An's death, a reconstruction force of 112,000 convicts took ten months to restore or build 240 buildings on the grounds under Emperor Shun.

74. For some unknown reason, however, the Analects was printed in the Stone Classics in place of the Odes.

75. That the registry of students (mingdie 名牒 )played an important role in Eastern Han is clear from Ying Shao's Fengsu tongyi, where references to the practice occur in chaps. 3-5. Unfortunately, we know very little else about the practice, except that any official recommending a student was required to take legal responsibility for that student's conduct, by analogy with a family head's taking legal responsibility for the conduct of his family members. See Tianlin, Xu, Dong Han hut yao, 377 Google Scholar. Certainly, such registrations played a major role in the repeated Proscriptions of the period. See Bielenstein, , The Bureaucracy of Han Times, 137–41Google Scholar. Note in A.D. 146 the state's first demand that mature candidates for the Academy formally register their allegiance to the throne via jiafa (presumably allegiance to a textual tradition), rather than resting content with shifa 師法 (modelling oneself on the teacher more generally), which was undoubtedly a push (probably ineffective) for greater social control (Hanshu, 6.281).

76. Hanshu, 10.310, commenting on the events of 30 B.C., says, for example, that Liu Xin was appointed to collate “central [palace?] secret texts”; see also Hanshu, 100A. 4203. We know that Ban Gu worked in the Lantai 蘭臺 (Orchid Terrace) for some twenty years ( Wilkinson, , Chinese History, 752 Google Scholar). Sima Qian is said to have consulted texts in the Stone Chamber (Shiji, 130.3296); there is no reference in the Shiji to the Shiqu 石渠, which is mentioned in Hanshu, 36.1929n.8, as north of the Weiyang 未央 main audience hall, but that palace library must have existed by 51 B.C., during the reign of Emperor Xuan, since a court conference was convened there. The only Hanshu reference to the Tianluge 天祿閣 is to Yang Xiong working there (Hanshu, 87B.3580). On the “secret books” (mishu 秘書), however, there are repeated references. See Hanshu ji buzhu zonghe yinde 漢書及補注綜合引得, Harvard-Yenching Sinological Index Series No. 36 (Peiping: Harvard-Yenching Institute, 1940)Google Scholar, V:06010. For Em-peror Hui惠帝as the first emperor to start up libraries, see Xu Tianlin, Xi Han huiyao, 222.

77. Hanshu, 30.1701; Hanshu, 36.1967. (Hou Hanshu, 79A.2548, reminds us that many of the documents in the libraries were “secret texts,”) That the imperial libraries were not generally open to the public, or even to other members of the imperial bureaucracy, presents a strong contrast with libraries (1) under the Ptolemies in Alexandria; (2) possibly under the Attalyd rulers of Pergamum (Attalus I-III); and (3) certainly under Caesar Augustus in Rome which served as “public libraries” open to scholars who were sufficiently well-connected. I am indebted to Geoffrey Lloyd and Corey Brennan for this information; see also Casson, Lionel, Libraries in the Ancient World (New Haven: Yale University Press, 2001)Google Scholar. At the Han court, as is clear from one anecdote, it was a very great honor to be admitted to work in the imperial libraries (for a scholar, it was like ascending to Penglai, the isle of immortality), and the post was viewsd as a step-ping stone to higher rank. The story told of Ma Rong 馬融 (79-106) is that he irritated Dowager Empress Deng 都后 by writing a satirical poem, and “he was stuck in the Eastern Lodge for ten years and not allowed to transfer” (Hou Hanshu, 23.822; 60A.1954). That story suggests that the post of imperial librarian was not high in the imperial bureaucracy. In a related area, Nathan Sivin and Geoffrey Lloyd, in their forthcoming The Way and the Word, draw our attention to the public character of foren-sic debate in Greece, contrasting it with the court location of debate in early China. But the contrasts are not absolute: we know that a few famous texts (such as the Liishi chunqiu) were hung in the marketplace for public perusal and evaluation. We have also the text of the Xunzi, which openly criticizes Mencius and at least eleven other thinkers of the pre-imperial era, and the Hanfeizi, which openly criticizes Hanfei's teacher, Xunzi.

78. For example, we know that scholars smuggled their own versions into the classics by bribing the librarians of the Orchid Terrace. See Bielenstein, , Luoyang in Later Han Times, 69.Google Scholar (By the way, this account is not necessarily aimed against the eunuchs, since many scholar-officials were apparently involved in the scandal.) We know also that many important texts were unaccountably lost, including the original guwen 古文 documents associated with Kong Anguo 孑1安國. See Nylan, Michael, “The chin wen/ku wen (New Text/Old Text) Controversy in Han,” T'oung Pao 80 (1994), 83–145 CrossRefGoogle Scholar; and Nylan, , “The ku wen Documents in Han Times,” T'oung Pao 81 (1995), 1–27 CrossRefGoogle Scholar.

79. The Meng Xi 孟喜 tradition for the Yi claimed to derive its authority from a deathbed transmission from the Yi Master Tian Wangsun 田王孫, and the tradition was honored though the story of its transmission was widely regarded as a lie. Even more astonishing is the honor accorded a pseudo-canonical Shangshu version in one hundred juan 卷, after it was known to be a pastiche by Zhang Ba 張霜 of several genuinely early texts—and this, though its contents departed from the canons stored in the imperial libraries (Hanshu, 88.3599, 3607). On the subject of “authorship,” Liu Xiang's name was probably affixed to many texts that he only edited, annotated, or compiled. Certainly, Dong Zhongshu's name was affixed to texts such as Chunqiu fanlu 春秋繁露， which were only “in the school of Dong,” and so on. Christoph Harbsmeier, in the general introduction to his forthcoming translation of Hanfei (Yale University Press), observes that prior to Western Han, what we deem “personal authorship” was “the result of textual accretion long after a master's death”; also that “the collections made in the name of ancient Chinese [i.e., Warring States] authors typi-cally included material other than what was literally composed and written by an author.”

80. The Simas, father and son, quite explicitly use the term jia to group particular orientations towards political rule, rather than a devotion to a particular body of texts. Basically, the Shiji employs the term jia as a way of framing an argument that has little to do with texts and their transmission; the argument concerns a perceived falling away by contemporary classicists from ancient practices and attitudes. According to Sima Qian, these few practices and attitudes were to be upheld insofar as they reflected the careful transmission of specific “techniques” for applying the knowledge of the distant past to current policy-making dilemmas. These practices and attitudes, many of them preserved in the rites and music, supposedly survived—unlike most texts—from the Western Zhou period or even before. Texts to Sima Qian, even the Odes and Documents, were important mainly insofar as they provided descriptions or contexts by which to explain these antique usages; writing in the sense of “literary composition” was in many cases an auxiliary art. Note that Sima Tan's quite restricted use of the term jia (later revised by Liu Xiang) tallies perfectly with that of other classical masters in Han times, especially Yang Xiong. See Csikszentmihalyi and Nylan, “Constructing Lineages and Inventing Traditions.”

81. Only the Mohists constituted a well-defined and cohesive group (if not “school”) in the pre-Qin period; and in the closing days of Eastern Han, those trained atjingzhou 荆州 and by Zheng Xuan seem to have defined themselves by their school affiliations.

82. The phrase “can talk it” is said of Master Gaotang髙堂, a specialist in the rites texts, in Shiji, 121.3126; and of Fu Sheng, a specialist in the Documents, in Hanshu' 88.3603. See also Shiji, 121.3125; Shiji, 121.3118, which describes all of Ru learning in terms of the masters who can “talk” (yan) the Five Canons; and Shiji, 101.2746n.2, which only makes sense if we understand the canonical passages as being read out in Qi dialect. Many passages also talk of the masters chanting (song 誦) the texts and traditions to their students. For example, the description of two famous teachers, Dong Zhongshu in the mid-second century B.C. and Ma Rong in the late second century A.D., is very nearly the same; both taught by chanting the texts, which were then transmitted through recitation by their chief pupils to the relative beginners. See Shiji, 121.3127; Hou Hanshu, 35.1207; and Guanzi (Guoxue jiben congshu 國學基本叢書 ed.; Taibei: Shangwu, 1968)Google Scholar, pian 篇 59, “Dizi zhi ״弟子職 (Duties of students). The impression that teaching was oral is strengthened by a remark in Hanfeizi, 14.71: “Now even if the ruler of men does not teach the officers orally, nor with his own eyes spy out the traitors.… “In a famous pictorial brick from Sichuan that shows a master expounding on the classical traditions (see Lim, Lucy et al., Stories of China's Past. Han Dynasty Pictorial Tomb Reliefs and Archaeological Objects from Sichuan Province [San Francisco: Chinese Culture Foundation, 1987], 122)Google Scholar, the master (unlike his disciples) holds no bamboo bundles or silk scrolls in his hand. Contrast the famous painting of Fu Sheng 伏生 teaching the Documents (attributed to Wang Wei 王維； probably of Song date; now in the Osaka City Museum), where the master holds a text prominently in his hands. Interestingly enough, Hanshu, 30.1710, seems to regard the loss of traditions under pre-Han as a loss of texts, not of practices—in direct contrast to the Shiji accounts.

83. See the story of Anguo, Kong in Shiji, 121.3125 Google Scholar; also that of Ouyang Xi 歐陽歙 in Hou Hanshu, 79A.2556, whose family for eight generations had served as Academicians. It was said of Dong Zhongshu that he was unusual in that he made scholarship his only business (Shiji, 121.3128): “to the end of his life, he did not engage in any money-making venturies; he made improvement of his learning and writing books his business.” Ca. A.D. 100, we perhaps see a change, in that more writings are mentioned in connection with the scholar-officials who also had successful careers, and writings are increasingly listed in the position where children would be, as “offspring” of the family. But many “big families” made landed wealth and high office their main sources of revenue and status.

84. The first book market was said to be in Luoyang in the Western Han ( Wilkinson, , Chinese History, 253 Google Scholar). Wang Chong visited bookstalls in his own day in Luoyang (ca. A.D. 95), according to Hou Hanshu, 49.1629. Guangwu is said to have brought to the capital of Luoyang 2,000 carts of books altogether, but supposedly in A.D. 196, when the emperor in his haste fled the capital, he carried out only some seventy carts, some part of which were ruined by rain in transit or when the silk was used for other purposes. The rest of the imperial libraries' contents went up in smoke when Dong Zhuo 董卓 burned the palaces. See Fengsu tongyi, yi wen 佚文, 2.99.

85. I have been trying to find the average weight for a bamboo bundle, but as bundles varied so widely, with the typical bundle having between thirty and sixty bamboo strips, I have not yet had much success. For more information, see Wilkinson, Endymion, Chinese History: A Manual, Revised and Enlarged (Cambridge: Harvard University Asia Center, 2000), 446 Google Scholar. Given the weight of bamboo bundles, the notion of the “big book” designed to include numerous pian comes relatively late; even in Eastern Han, the typical bamboo-bundle texts are small, as can be seen from the depictions of Confucius's disciples at the Wu Family shrines.

86. Those said to have owned libraries include Confucius, Mozi, Lü Buwei, Liu An, and the King of Hejian 河間王. Among “cruder polemics” I include the statement in Hanfeizi, 32.196, that “half the people of Zhongmou” have left their fields to “engage in literary pursuits.” Hanfeizi, 49.347, rants that private families all possess copies of the penal code (“the laws of Shang Yang and Guan Zhong”) but the state is getting poorer and poorer, as it lacks farmers; further, that private families all have copies of the books of Sunzi and Wu Qi, but the armies are getting weaker. “Thus in the state of an enlightened ruler, there are no belles-lettres on inscribed bamboo slips; the instrument of education is the law. There is no talk of former kings and the people's models are the police” (trans, slightly modified from Christoph Harbsmeier's forthcoming translation of Hanfei [Yale University Press]).

87. See Hanshu, 30.1720, for a summary of the works classified as “elementary learning.”

88. This is the proper context into which to fit legends about the First Emperor's reading through many pounds of documents per day. As Xinyu 新語 (Sibu beiyao 四部備要 ed.), A/10b, insists: “To work hard at learning, intoning the Odes and Documents— that's something any ordinary man can do”; trans. Mei-kao, Ku, “A New Discourse on the Art of Government: Being a Translation of Hsin yu of Lu Chia (? —178 B.C.) of the Western Han Dynasty” (M.A. thesis, Australian National University, 1974), 96 Google Scholar.

89. For slaves as “office workers,” see T'ung-tsu, Ch'ü, Han Social Structure (Seattle: University of Washington Press, 1972), 145 Google Scholar. For one anecdote showing that even the slaves of a certain Ru master were literate, see Yi-ch'ing, Liu, Shih shuo hsin yü: A New Account of the Tales of the World, trans. Mather, Richard (Minneapolis: University of Minnesota Press), 94 Google Scholar; for learned eunuchs, adept in framing the language of edicts, overseeing the correction of imperial editions, or submitting elegant memorials full of classical citations, see Hou Hanshu, 78.2513, 2519, 2528, 2533. Note in addition that Yang Xiong in Fayan 法言 ( Xinbian zhuzi jicheng 新編諸子集成 ed.; Taibei: Shijie, 1978), 2.4 Google Scholar, denounced as “children's amusements” the carving of seal characters. The Hongdumen xue 鴻都門學 (Academy of the Vast Capital Gate) was established by Emperor Ling 靈帝 in A.D. 178 so that young men might be trained in the arts of writing government documents, which often required poetry and calligraphic forms. Many shi, including Cai Yong 蔡虽, objected to the establishment of the Academy, charging that its students were “petty persons” being trained in trivial pursuits (Hou Hanshu, 60B.1992). Only in the late second century A.D. do we see elites beginning to take great pride in putting brush to paper or silk. For further information, see Nylan, “Calligraphy.”

90. This observation draws upon Bagley, Robert W., “Meaning and Explanation,” Archives of Asian Art 46 (1993), 6–26 Google Scholar. As it would be hard to imagine a situation where a text-centered culture (e.g., Shang) evolved into a less text-centered culture (Warring States and Han), which then became more text-centered again, it is interesting to see that two recent papers by David Keightley seem ready to query the assumption that Shang oracle bone inscriptions represent documents gathered for state archives. See Keightley, “Making a Mark: Reflections on the Sociology of Reading and Writing in the Late Shang and Zhou” (paper prepared for the “Sociology of Writing” conference at the University of Chicago, November 6-7, 1999); and “The Diviners' Notebooks:

Shang Oracle-bone Inscriptions as Secondary Sources” (paper prepared for the Paris conference celebrating the 100th anniversary of the discovery of oracle bones, September 24,1999). In my view, it is entirely possible that writing somehow “fixed” the divination rituals and that the inscriptions, even when ritually housed, were never consulted after their carving. (Rosalind Thomas, Literacy and Orality, chap. 3, esp. p. 94, discusses the crucial distinction between making records and making documents, where there is the expectation that documents will be consulted later.)

Turning to the Western Zhou bronze inscriptions, some inscriptions were written across several different bronzes, which were then separated, or were written in dif-ferent directions across bronzes. Others, as Keightley notes in “Making a Mark,” 22, were “found deep in the dark interior of certain Western Zhou ritual bronzes,” in places where “it would have been virtually impossible for those who used the vessels in the rites to read.” (This subject is intelligently discussed by Lothar von Falkenhausen in Early China 18 [1993].) Such facts suggest that their inscriptions cannot be “docu-merits” as part of an “archive” in any modern sense belonging to those two words; the inscriptions, like the wine and food, were for the ancestors to savor. (With the inscribed bells, by contrast, the inscriptions appear on the outside surface, where they can be carried on the wind.) They suggest also that the semantic meaning of certain proclamations and incantations was conveyed chiefly through oral recitation, rather than through reading. See also Keightley, David, “Art, Ancestors, and the Origins of Writing in China,” Representations 56 (Fall 1996), 68–95 CrossRefGoogle Scholar.

91. Zhonglun, Preface, probably by Xu Gan himself or an immediate disciple (trans, after John Makeham's forthcoming translation [Yale University Press]). The passage continues: “Yet there are many who are only too happy to look up to them [the book-worms and pedants] and only a few who argue against doing so.” Prior to the passage cited, the preface speaks of the great scholar who commits “sayings of excellence” to memory: “During the day he meticulously studies the classics and their apocrypha, and at night one by one, he observes the constellations. He investigates the primordial chaos … [and] he makes long-term considerations. … What time does he have to make a clamor about philological matters or to work at securing a hollow reputation by joining the ranks of vulgar fellows?” See also chap. 10 in the same work on the importance of dress as exemplifying virtue.

92. Sima Qian traces contemporary moral decline to two factors: (1) the imperial bureaucracy under Emperor Wu tested candidates for public office on their mastery of written forms, rather than their commitment to the values reflected in classical literature; and so (2) the Ru ranks had lately been filled by careerists willing to promote the “wealth and power” agenda of Emperor Wu, ever since Emperor Wu made it plain that such service would be richly rewarded. Lured by such rich rewards, candidates for office, though trained in Ru techniques, saw their primary goal as the acquisition and maintenance of prestige and power, for themselves and for the emperor. They showed little interest in preserving the spirit or even the letter of the past, except insofar as a posture of preserving the past would further their personal ambitions. For further information, see Csikszentmihalyi and Nylan, “Constructing Lineages and Inventing Traditions,”

93. Hou Hanshu, 79A.254546. An interesting passage in the Liji lists all the things to be borne with the left hand held up, including writing tablets, stalks of dried flesh, flutes, pillows, bows, and so on. No special distinction seems to have been accorded writing or writing tools (Liji yinde 17/20).

94. Shiji, 121.3124; Hanshu, 88.3603, 3610.

95. Shiji, 121.3119, shows, for example, that the Imperial Academicians' “disciples” were to be appointed and later advanced on the basis of their (1) knowledge of rites and music; (2) loyalty to the throne; and (3) good reputations. While at the Academy, they were to be tested on their acquisition of learning in the “arts” (a term that can refer to either the Six Canons or the Six Polite Arts). For the canons as aides to memory, one might compare the instructive work on medieval Europe by Carruthers, M.J., The Book of Memory: A Study of Memory in Medieval Culture (Cambridge: Cambridge University Press, 1990)Google Scholar. Building on the work of Francis Yates, Carruthers show convincingly that medieval writers never considered writing a substitute for memory, but rather an aid to memory. No medieval text, she points out, makes a distinction between the act of writing on the memory and writing on some surface. Writing in itself was of no particular significance; significance was attached to important teachings which had been preserved in memory, thought about and commented on, and which thereby had become an important cultural resource. The purpose of reading, reciting, or listening to a text was to see through the text to the spirit of the author, then to internalize that spirit in one's own self. This corresponds closely to Yang Xiong's attitudes toward reading and writing in Han China.

96. See, e.g., Zhou yi yinde 44/Xi A/12, a passage treated by Lewis in his chap. 6, pp. 254-55; the remark attributed to the Documents canon in Hanshu, 36.1966; Fengsu tongyi, 2.9; Hanfeizi, 34.238, which considers the problem of “empty words”; and Hanfeizi, 32. 209, which criticizes over-credulity with respect to texts. For stories that treat past traditions with skepticism, see Nylan, Michael, “Han Classicists Writing in Dialogue about Their Own Tradition,” Philosophy East and West 47.2 (1996), 133–88CrossRefGoogle Scholar.

97. See, e.g., Mencius 7B/2; Hanfeizi, chaps. 32-33, which contain numerous anecdotes designed to demonstrate the unreliability of writing; Hanfeizi, 47.328, which speaks about the problems when “writings are too concise or written laws too elliptical”; and Hanfeizi, 48.336, which says, “It is the nature of statements that they become true because many believe them.” Certainly, the main point of Sima Qian's interspersing the full texts of the First Emperor's stele inscriptions with Sima Qian's descriptions of the emperor's conduct is to remind us that texts can lie. Of course, many thinkers railed equally against the pernicious oral persuasions that “create disorder and anarchy in the world” (e.g., Xunzi yinde 15/6/1-2).

98. Xunzi yinde 11/4/52. See, for example, Mencius 7A/21, which characterizes the noble man in terms of his ability to “render his message intelligible without words”; Xunzi yinde 5/2/38, 24/8/105, 87/23/6; Fayan 1.1-2; and Hanshu, 36.1971, which insists that the Way resides in worthy humans with great aspirations. Yang Xiong, ever mind-ful of the proliferation of competing ethical visions propounded in a broad spectrum of written texts purporting to be authoritative, compared texts to more mundane goods whose real value would have to be “weighed in the balance.” Not texts, but teachers, would distinguish good from bad. And the quality of a man's teachers, not the number of texts at his command, would determine a person's ethical character. So while Yang in his own works set a standard for elegant writing, he was quick to express his utter disdain for literary techniques that relied upon memorization and regurgitation—the mere trappings of culture. See also the story told of Mozi, who felt no compunction to consult the many books in his library once he had truly understood the principles of the cosmos, as reported in the “Guiyi” 貴義chap, of the Mozi. This emphasis on the importance of a teacher or tutor who will lead others through the difficult parts of the ritual of reading may go back to Shang times, Keightley speculates in “Making a Mark,” 37.

99. Note that I do not believe that Confucius composed one or more texts, but traditions recorded in various Han accounts, including his Shiji biography, credit him with the composition of the Chunqiu, the “Great Commentary” (“Xici”) to the Changes, and several other texts; see Shiji, 47.1935-37.

100. I have only just begun a search of pre-Han literature, but the earliest appearance of the term liuyi 六藝 (Six Arts) known to me seems to come from the Shiji, though Analects 1/6 mentions the “arts” (yi). (Lewis's talk of “Sima Qian's curriculum” in Shiji, 130.3297, is odd indeed.) The young prince in the Guoyu is to study the histories of the state, rhetoric, precedents, and the legal code, according to Guoyu, 17.191; see also Liji yinde 12/52. It is noteworthy that as late as the Hanfeizi, 10.46, the standard expression for kingly rule is to tingzhi 聽治 “listen (and so attend) to good rule” or tingyan 聽言 “listen to proposals.” That this literally refers to “listening” to a reading of bureaucratic accounts and policy papers is made quite clear by a later anecdote in the same work, Hanfeizi, 35.259, where the king “listens to the accounts 聽計),though there are more than he can listen to in a single session.” See also Guoyu, 1.3, where the ruler hears many sorts of different oral presentations when attending to policy matters; also the numerous Zuo references to the “duke being pleased” {gong yue 公悅) with good speeches (e.g., Chunqiu jingzhuan yinde 136/Xi 28/23, Zuo; 202/Xuan 14/fu; 223/Cheng 6/fu 2). The Hanfeizi at the same time speaks of several rulers who can read, including King Zhao of Wei (r. 295-76 B.C.; Hanfeizi, 32.212); also, of written submissions made to the states' highest ministers, including their prime ministers (Hanfeizi, 32.208). (By comparison, in Europe, it was not before the Renaissance that kings and princes began to make it a practice to read documents for themselves!)

101. I am indebted to Lai Guolong, a Ph.D. candidate at the Department of Art History at the University of California at Los Angeles, for the foregoing statement; and for supplying the additional tabulations. By Lai's count, twenty Chu tombs from the Warring States period contain writings on bamboo or silk (with six of those tombs containing fewer than ten bamboo strips); five tombs from the Qin dynasty; and seventeen from the Han (excluding the frontier military stations). Many other tombs contain a variety of valuable items, including ritual bronzes and jades; admittedly, many of the tombs of the nobility had been looted, but the grave-robbers of the contemporary anecdotes were looking for precious metals and stones, not texts.

102. For example, the court of the Lord of Chunshen 春申君, according to some legends, included literary men, as well as other sorts of entertainers (Shiji, 78.2395; 74.2346-47). The specialties of the figures at Jixia 稷下 (customarily, if erroneously identified as the Jixia “Academy” in many modern studies, including Writing and Authority, pp. 33, 75-78) were equally mixed. For a definitive review of the material on Jixia, see Sivin, Nathan, “The Myth of the Naturalists,” in his Medicine, Philosophy and Religion in Ancient China (Great Yarmouth: Variorum, 1995), 1–33, esp. 19-26Google Scholar.

103. The old formats greatly facilitated memorization; as late as the Hanfeizi, for example, whole chapters of the text are rhymed. For the implications of the language crisis and its relation to writing, see Olson, David R., The World on Paper: The Conceptual and Cognitive Implications of Writing and Reading (Cambridge: Cambridge University Press, 1994), chap. 4, esp. 71–76 Google Scholar. Gongsun Long's 公孫龍 work and some of the early chapters of the Zhuangzi seem still to be making sense of a world in which words have become things in themselves, subjects of philosophical reflection as well as objects of definition. Also notable in fourth century B.C. texts is the use of huo yue 或日 “another says” for variant versions, for “traditional cultures treat alternative expressions of the same sense as being ‘the same’ while literate ones only consider the stricter criterion of verbatim repetition as the same ( Olson, , The World on Paper, 87 Google Scholar, citing Finnegan, R., Oral Poetry: Its Nature, Significance and Social Context [1977])Google Scholar.

104. Music and the ear's training to perceive fine music dominate discussions of cultivation in the early texts, but the ear's domination slowly yields to that of the eye. Most have assumed that the introduction of Buddhism is largely responsible for this change, but a growing dependence on reading is an equally plausible factor. For the changing role of music in early Chinese culture, see von Falkenhausen, Lothar, Suspended Music: Chime Bells in the Culture of Bronze Age China (Berkeley: University of California Press, 1993)Google Scholar; deWoskin, Kenneth, A Song for One or Two: Music and the Concept of Art in Early China (Ann Arbor: University of Michigan, Center for Chinese Studies, 1982)CrossRefGoogle Scholar; Bagley, Robert, “Percussion”; and Kern, Martin, Die Hymnen der chinesisehen Staatsopfer: Literatur und Ritual in der politischen Repräsentation von der Han-Ziet bis zu den Sechs Dynastien (Stuttgart: Franz Steiner Verlag, 1997)Google Scholar.

105. Somewhat confusing is Lewis's shifting account of the wen/shi 文赏 relation (pp. 101, 141-44, 236, 272-73, etc.). For more methodical observations, see Martin Kern, “Ritual, Text, and the Formation of the Canon: Historical Transitions of wen in Early China,” forthcoming in T'oung Pao, which shows that wen 文 more often referred to social and cosmic pattern than to “writing” until mid-Eastern Han; also that wenzhang 文章 did not come to refer primarily to literary writings until mid-Eastern Han. Kern's hypothesis is confirmed by remarks in Zehou, Li 李澤厚 with Gangji, Liu 劉綱糸己, Liang Han nteixue shi 兩漢美學史 (Taibei: Jinfeng, 1987), 198 Google Scholar; also, by a careful reading of the texts on wen gathered in Zhongguo meixueshi ziliao anbian 中國美學史資料案編 (Taibei: Mingwen, 1983)Google Scholar.

106. Compare Hanshu, 88.3592, and Hou Hanshu, 79A.2545. (A second version of the second story found there has these candidates for public office “carrying on their backs” Confucius's very own ritual vessels as they go to profess their allegiance.) Readers may be amused to learn that at the end of Eastern Han, the Ru looking to throw their lot in with Liu Biao 劉表 at jingzhou supposedly carried their ritual vessels and their charts and texts to his camp, according to the Jingzhou wenxue jiguan zhi 荆州文學記官志 (Official record of the college in Jingzhou), attributed to Wang Can 王藥: “Those who came from afar carrying texts on their backs and ritual implements on their shoulders numbered over 300”; see Yiwen leiju 藝文類聚 (Shanghai: Shang-hai guji, 1965), 38.693 Google Scholar. The verbal precedents for such passages go back at least as far as the Zuo, where the dianji 典籍 “statutes and records” of the reigning Zhou dynasty are transferred to Chu. See Chunqiu jingzhuan yinde 421/Zhao 26/7 Zuo; see also Liishi chunqiu (Xinbian zhuzi jicheng ed.), 16.189-90.

107. See, for example, Liu Xiang's essay on wen in chap. 19 of the Shuoyuan 說苑, “Xiuwen” 修文 (Cultivating pattern).

108. There has been considerable debate over the date of the first paper in China. Silk and hemp rag can be found long before A.D. 105, the traditional date of paper's invention; the earliest excavated examples go back as far as 141 B.C. See Xianlin, Ji 季羡林, Zhongyin wenhua guanxi shilun wenji 中印文化關係史言侖文集 (Beijing: Xinhua, 1982), 44–50, 484–86Google Scholar. But a comparison of two plates in Weichao, Yu, A Journey into China's Antiquity (Beijing: Morning Glory Press, 1997), vol. 2,232–33Google Scholar, shows examples of paper suitable for writing on that date to Eastern Han; in Western Han, paper was made from fibrous hemp bundles “unfit for writing on” (p. 232). Note, however, that T.H. Tsien, writing in vol. 5, pt. 1, of Needham's, Science and Civilisation in China (Cambridge: Cambridge University Press, 1985)Google Scholar, shows that paper only very gradually—long after the Han—replaced wooden strips as the main medium for texts. Jixing, Pan 潘吉星， Zhongguo kexue jishu shi: zao zhi yu yinshua 中國科學技術史：造紙與印刷 (Beijing: Kexue, 1998)Google Scholar, shows many examples of Han paper, but I have been unable to secure a copy of Pan's book.

109. The Erya 爾雅 (at least the first part) dates to the third century B.C., according to Bernhard Karlgren and South Coblin; Yang Xiong's Fangyan 方言, to ca. A.D. 5; and the Shuowen, to A.D. 100. (Nothing more than the title is known of the dictionary attributed to Sima Xiangru.) See Early Chinese Texts: A Bibliographical Guide, ed. Loewe, Michael (Berkeley: Society for the Study of Early China and Institute of East Asian Studies, University of California, Berkeley, 1993), 96 and 429 Google Scholar. Tradition also ascribes the new developments in the conventions for the punctuation of texts to Ma Rong, but those traditions are too hazy to serve as firm evidence. According to Goody, Jack, The Interface between the Oral and the Written (Cambridge: Cambridge University Press, 1987)Google Scholar, passim, only when the written script is taken as model for language can there be writtCT\ explicit logics, grammars, and dictionaries. According to Olson, The World on Paper, 88, such developments reflect the growing awareness that the logographic script cannot completely capture the full meaning that can be imparted by oral speech. “The problem for writing then becomes that of inventing devices … which can com-pensate for what is lost” (p. 111). “Texts written to circumvent the limits of simple transcription are, by definition, texts written to be read” (p. 113).

110. According to Olson, , The World on Paper, 95 Google Scholar: “To make a written text clear, this must be ‘edited in’ by descriptive commentary and by the elaboration of a meta-language indicating how the text is to be taken. In the absence of such commentary, the reader is faced with the task of determining how the text is to be taken. A naive reader, faced with a text, may be tempted, indeed was tempted … to ascribe whatever effect the text had on him or her as being the meaning intended by the writer,”

111. Liu Xin, of course, was a key figure at the court of Emperor Ai 哀帝 (7-1 B.C.), when the wrangling over the Archaic Script edition of the Zuozhuan began; see Hanshu, 36.1967-72.

112. The Western Han Fuyang 阜陽 Odes provides a good example of a standard text showing many variant characters, whose phonetic differences do not reflect a local dialect. See “Fuyang Hanjian Shijing” 阜陽漢簡詩經， Wenwu 1984.8,1–12; Hu Pingsheng 胡平生 and Han Ziqiang 韓自強, “Fuyang Hanjian Shijing jianlun” 阜陽 漢簡詩經簡論, Wenwu 1984.8, 13-21.

113. The transcribed oral transmissions, of course, were the so-called Modern Script canons sponsored by the Han court and the written texts that had survived minus their oral interpretations were the so-called Archaic Script texts. For further information, see n. 78.

114. See Wang Chong, Lunheng, esp. chaps. 82-85, whose contents as they relate to writing have been summarized in Nylan, “Calligraphy,” 41-44. Ge Hong, Baopuzi 抱樸子, “Shang bo” 尙薄, asserts that written texts are at least as important as virtuous conduct in the assessment of a man's talent. See Mingzhao, Yang 楊明照, Baopuzi waipian jiaojian 抱樸子外篇校箋 (Beijing: Zhonghua, 1997), 107 Google Scholar

115. These long written texts in new formats included many very long commen-taries to the Classics and explications of the penal code. The Hou Hanshu biography of Ying Shao shows us his phenomenal production of written texts, with some 136 pian attributed to him (Hou Hanshu, 48.1615). My working hypothesis depends upon this sort of evidence. The ״Rulin״ chapter in the Shiji (Shiji 121) tells us which canonical texts the masters specialized in (though some of these texts were possibly thought of as oral). The Hanshu “Rulin” chapter (Hanshu 88) shows an increasing number of masters writing their own texts in response to the texts they teach (e.g., Hanshu, 88. 3597, the example of Ding Kuan 丁寬； Hanshu, 88.3604, the example of Zhou Kan 周勘 writing on Five Phases theory). By the time of the compilation of the Hou Hanshu, the biographical traditions relating to the classical masters are more preoccupied with their written compositions, and writings are treated as the “progeny” of their authors.

116. See the edict of A.D. 57 by Guangwu, on the zhangju 章句 being too long; citation in Xu Tianlin, Dong Han huiyao, 134.

117. Wang Chong (27-97) is one of the first we know of to roam the capital's book-stalls in search of texts to read on his own, and to base his own authority on years of solitary reading and writing, after schooldays spent under the direction of Ban Biao 班彪, father to Ban Gu. See Drège, Jean-Pierre, “La lecture et l'ecriture en Chine et la xylographie,” Études Chinoises 10.1-2 (1991), 77–112 CrossRefGoogle Scholar. On p. 103, Drège suggests that “the domination of the eye over the ear in the practice of individual reading” began only in Song.

118. See Zhuangzi yinde, Harvard-Yenching Sinological Index Series, Supplement No. 20 (Cambridge: Harvard University Press, 1956), 7/2/90; trans, after Watson, Burton, The Complete Works of Chuang tzu (New York: Columbia University Press, 1968), 49 Google Scholar.

119. Davidson, James, Courtesans and Fishcakes: The Consuming Passions of Classical Athens (London: Harper Collins, 1997), xvi Google Scholar.

120. In most Han texts, speech and writing are related, as in one passage cited by Lewis on Huang-Lao, p. 341.

121. The Shuowen definition of du 讀 is to read out loud. Yan Shigu's 顏師古 (Tang) preface to the Hanshu commentary clearly defines du as to recite out loud. Wang Su's 王肅 commentary to the Analects defines xi 習 “practice (the text)” as du “read out loud.” Full citations to these and many other examples may be found in Drège, “La lecture et l'écriture,” esp. 79; and Mingren, Zhang 張明仁, Gw jin mingren dushu fa 古今名人讀書法 (repr., Beijing: Shangwu, 1998)Google Scholar.

122. Lewis largely equates the Six Arts with the “literary remains of earlier rulers,” ignoring a possible reference to the six polite arts of the nobility, only one of which is writing (p. 234). On p. 288, Lewis appears to redefine liu yi as “major texts of the ‘arts’” but it is difficult to think what major texts covered charioteering, for example. For the probability that the term liu yi represents the official Qin designation of the canon, in contrast to earlier usages, see Baoxuan, Wang 王裸玄, Xi Han jingxue yuanliu 西漢經學原流 (Taibei: Dongda, 1994), 15–20 Google Scholar.

123. Chunqiu jingzhuan yinde 379/Zhao 12/9 Zuo.

124. Manguel, Albert, A History of Reading (Harmondsworth: Penguin, 1996), 8–9 Google Scholar, with the last sentence citing Stevenson, Robert Louis, “My Kingdom,” A Child's Garden of Verses (London, 1885)Google Scholar. Over recent years, prior to the writing of this review, I have had extensive discussions touching upon writing's authority with a number of scholars, including Robert Bagley, Donald Harper, Martin Kern, Waiyee Li, Geoffrey Lloyd, and Nathan Sivin, which have revolutionized my thinking, and I would like to express my gratitude to those scholars here. Of course, the views expressed here are my own.