Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Two philological studies on the Mawangdui Laozi manuscripts Herforth, Derek Dane 1980-03-17

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


831-UBC_1980_A8 H47.pdf [ 7.15MB ]
JSON: 831-1.0094912.json
JSON-LD: 831-1.0094912-ld.json
RDF/XML (Pretty): 831-1.0094912-rdf.xml
RDF/JSON: 831-1.0094912-rdf.json
Turtle: 831-1.0094912-turtle.txt
N-Triples: 831-1.0094912-rdf-ntriples.txt
Original Record: 831-1.0094912-source.json
Full Text

Full Text

TWO PHILOLOGICAL STUDIES OU THE MAWANGDUI jLAOZI MANUSCRIPTS by DEREK DANE HEBFORTH A.B.. The University of California, 1971 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Department of Asian Studies) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May 1980 Derek Dane Herforth, 1980 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department or by his representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department nf Asian Studies The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Date 2 5 March, 1980 Abstract This thesis focusses on some of the philological problems encountered in the study of two Chinese manuscripts of the Laozi dating from the late third century B.C.. The MSS. were recovered in late 197 3 from an early Western Han tomb unearthed at Mawangdui in the suburbs of Changsha, Hunan. The Introduction summarizes the details of the excavation of the tomb and discovery of the MS. documents. The main features of the MSS. themselves are then described with attention given to the graphic style, methods of dating and the arrangement of the text. The Introduction concludes with a discussion in the light of the new MS. evidence of some of the problems of the authorship and textual history of the Laozi. Part One treats substitution variation between the two MSS. as well as between the MSS. and the transmitted versions of the text. A typology of this sort of variation is proposed in which all substitution variants can be classified as one of the following: taboo graph, semantic variant, miswriting, simplified graph or loan character. The first four types are briefly discussed with examples.. The remainder of Part One is devoted to a detailed examination of the phenomenon of loan characters and the; analysis of some twenty problematic variants of this type from the two MSS. Part Two is a collation of the acknowledged quotations from the Laozi found in the Hanfeizi with the two Mawangdui MSS. and iii the transmitted versions of Laozi as recorded in a modern variorum. Because of the large number of variables involved, due especially to our lack of knowledge about the transmission of the Hanfeizi, it is difficult to draw firm conclusions on the basis of this collation. The evidence seems to suggest, however, the existence in pre-Han times of discrete textual traditions cf the Laozi and that some of the variation in the extant transmitted texts may derive from these separate pre-Han traditions. iv TABLE OF CONTENTS Abstract....,..,.,............................................ ii Table of Contents.............................................iv Abbreviations Used...........................,..y Note on Principal Sources and Bibliographic Conventions.vii Introduction. 1 Notes to Introduction.................................... 31 Part One 39 Notes to Part One.,,,,,,,,,,,,,,,,,...................... 87 Part Two.,,,.................................,,.,,..,,,,,,,.,,91 Notes to Part Two....................................... 137 Bibliography. ,,,,,,,,,,,,,,,,,,,,,,,,139 V Abbreviations Used AL the Laozi portion of Mawangdui MS. A ^ ) AM the miscellaneous texts following the Laozi on MS. A BL the Laozi portion of Mawangdui MS. B (Z, & ) BM the miscellaneous texts preceding the Laozi on MS. .B BMFEA Bulletin of the .Museum of Far_Eastern Antiquities BS Mawangdui Hanmu boshu (yi) (Mawangdui Hanmu boshu zhengli xiaozu 1974) FY Fu Yi FYY Fan Yingyun (jv/^Tu) GES Grammata Serica Recensa (Karlgren 1957) HF Han Fei (^^ ) HFZ Hanfeizi (Chen 19 58) HL hapax legomenon HSG Heshanggong (>n)'±/i^) JL •Jielaot (%•&), Chapter 20 of Hanfeizi KS Roshi kosei (Shima 1973) LC loan character ^) LZ Mawangdui Hanmu_boshu^ Laozi (Mawangdui Hanmu boshu zhengli xiaozu 1976) MC Middle Chinese, the language of the Guangy_un MWD Mawangdui (^±.±ft.) vi OC Old Chinese, the language of the latter half of the Zhou dynasty SSJZS Shisaniinq zhushu ("r^^ \%E.2%J) ST Mawanqdui boshu_Lapzi.iishitan (Yan 1976) TT transmitted text HB Wang Bi {X^&j) XE Xianger ($Hj) XS xiesheng (1^^) YL 'Yulao1 (o|j ^) , Chapter 21 of the Hanfeizi YZ Yan Zun ZD Zhonqwen dacidian (^^jfc) vii A Note on Principal Sources.and Bibliographic In preparing this essay on the Mawangdui (MWD) MSS. of Laozi I have made use of two transliterations1 of the text published by teams of Chinese scholars with direct access to the documents. 1) Mawangdui Hanmu boshu fyi) j|±:rfe *M % ^ gf hereafter BS (Mawangdui Hanmu boshu zhengli xiaozu 1974).. 2) Mawangdui Hanmu - boshu Laozi,^ ±±&_ %^^EJ & , hereafter LZ (Mawangdui hanmu boshu zhengli xiaozu 1976). BS contains photographs of the MSS.r transliterations into traditional kaishu and as an appendix a parallel printing of both MSS. with the Fu Yi text from the Daozang (•^jiM^f ^£ ) • T^e annotations to the transliterations are restricted to single character glosses. LZ lacks MS. photographs but provides a revised and well-annotated transliteration in simplified kaishu. Also included are the three parallel texts, incorporating the revisions in the transliteration, and three interpretative articles reprinted from Xuexi yu pipan 74.10, Wenwu 74.11 (revised) and Lis hi yan jiu 75.3. viii Citations from BS are made by reference to MS. A (^P ^) or B (7j&)r text L (Laozi) or M (miscellaneous), page and line number of the transliteration. Cross-checking of the MS. photographs is facilitated by the printing of MS. line numbers in the transliteration. In cases where the transliteration in BS has been revised in LZ, the latter only is cited by page number. In addition to BS and LZ, a photocopy of the MS. photographs and the transliteration in BS has been published in Taiwan. This includes a lengthy textual study by Yan Lingfeng which is cited by page number in the present study. 3) Mawanqdui boshu Laozi shitan hereafter ST (Yan 1976). The task of collating the MWD MSS. with the transmitted versions of the Laozi would have been much more onerous without the convenience of Shima Kunio's variorum, also cited here by page number. 4) Soshi kosei ^JJfetE , hereafter KS (Shima 1973).. References to pre-Han literature indexed in the Harvard-Yanjing series are to that text. In the case of the Shiji citations are by i uan and page number to the Takigawa edition. References to the Hanshu follow the same convention and are to the Zhonghua shuju edition.. Unless otherwise noted, reconstructions of Old Chinese are Karlgren's Archaic forms quoted from GSR. 1 Introduction Since the establishment of the People's Republic of China in 1949 archaeologists have excavated a number of ancient tombs in the area of Changsha in the central Yangzi basin.2 In 1972 work began on what has proved to be the most spectacular and significant of these excavations, a tomb complex dating from the early Western Han period.. Located at MWD in the eastern suburbs of Changsha, the complex has been identified as the burial site of the family of Li Cang ), the first Marquis of Dai(^^.).3 The complex is composed of three tombs of which the Marquis' own (Tomb No, 2) had undergone considerable despoliation at the hands of grave robbers. The other two tombs were found on excavation to be intact and to contain a large number of artefacts and written documents of surpassing historical interest. The MS. material which is the subject of this essay was discovered in Tomb No. 3 which contained the corpse of a man in his thirties. A wooden slip recovered from the tomb gives the date of burial as 168 B.C-/ midway in the reign of Han Wendi, Liu Heng d^J'H.) . . Chinese scholars have speculated that the man buried here was a son of Li Cang, although he cannot have been the successor to the Marquisate, Li Xi (jfcj^) who is known to have died in 165 B.C. (Hanshu 16.618).* The occupant of Tomb No. 1 has been identified as Li Cang's wife, thought to have died a few years after 168 B.C. 2 The MSS. themselves were found in the lower compartment of a rectangular lacquer box. . They were written on silk which was then pleated to fit into the container, thus causing considerable damage to the creased portions. Among the many texts copied onto these lengths of silk are some which we today possess in transmitted versions and others which bear some resemblance to extant works such as the Zhangupce or the Zuozhuan. Finally there are a number of entirely new documents, some broadly philosophical and others highly technical in nature. Each of the first two lengths of silk MS. cataloged by the excavators contains a copy of the Laozi -as well as other lost texts. Ms. A opens with the Laozi, untitled, followed by four heretofore unknown essays, also without titles.. Though the latter share some conceptual vocabulary with the Laozi, they are apparently not directly related to that work as it is traditionally understood.. On MS. B the Laozi follows four new philosophical texts: •Jingfa1 $5L&) , •Shidajing' (-r*J*SL), •Cheng' and ' Daoyuan• C^J^_) . The study of these texts has aroused great interest since it has been suggested by Chinese scholars that they may constitute the long-lost Four_Glassics_of the Yellow Emperor (•fr'rp' ^s^L) listed in the bibliography of the Hanshu (30.1730).5 Furthermore, the coupling in MS..B of the Laozi with texts associated with Huangdi has suggested to many that here may be found the documentary antecedent of the term ' Huanglao" (fr^), a compound used repeatedly by Sima Qian in the Shiji to designate a school of thought current in late 3 Zhanguo, Qin and early Han.6 The study and interpretation of these lost texts is certain to shed light on other questions in early Chinese intellectual history as well, particularly the problematic relationship between early Daoism as represented by the Laozi and Legalism. The dating-;-of the Laozi MSS. Obviously none of these texts can date later than 168 B.C.. when Tomb No._ 3 was sealed after the burial of Li Cang's son. In the case of the two Laozi MSS., a difference in calligraphic style helps to establish their relative chronology. AL was copied in a style transitional between the small-seal and clerical scripts rf M ¥ LZ 110); that is to say, the construction of the graphs retains many features of the small-seal style while the individual strokes have begun to lose the.roundness of the earlier script for the angular guality of the clerical style. 7 As these examples suggest, the graphic style of BL is closer to the clerical script than is AL. This is one piece of evidence which allows us to date BL somewhat later than AL. Another method of dating the two MS. copies of Laozi , this Small-seal MWD AL Clerical 8 MWD_BL -4 time in terms of absolute chronology, is provided by the avoidance of a taboo character in BL.. In every case where the graph -^-p is found in AL, it is replaced by WI in BL. . A taboo against the use of the character -^-jS went into effect on the accession of Han Gaozu, Liu Bang, in 206 B.C.; thus, AL must have been copied before that date.9 Gaozu's successor to the Han. throne was Huidi, Liu Ying (j§jl). Since this graph is not tabooed in BL, it seems reasonable to assume that BL was copied sometime between the accession of Gaozu and the accession of Huidi in 194 B.C.»o The arrangement of the MSS. . The earliest bibliographic information of any kind we have for the Laozi is contained in Laozi's biography in the Shiji where Sima Qian states that the sage composed a text of over five thousand words in two sections (_t T J| ) on the meaning of dao and de (63.6).. Both the MWD MSS. preserve the division into two sections; however, the order of the sections is reversed from that found in all transmitted versions. In the MSS., the section generally referred to as the 'Dejing' (i.e. Chapters 38-81 of the TTS) precedes the 'Daojing' (Chapters 1-37).. This has led Gao Heng and Chi Xichao to posit the existence in Warring States' times of two versions of the text: one similar to the TTS and property of the Daoists for whom the metaphysics of the Dao was paramount and the other in the arrangement of the MSS. and associated with the Legalists who were more concerned with the political phlosophy of the 'Dejing' (LZ 112). The only 5 surviving evidence on this point comes from Han Fei's commentaries on the Laozi , the 'Jielao' and *Yulao* chapters of the HFZ. The first quotation from the Laozi- discussed in the 'Jielao* is taken from the opening of the 'Dejing* section (i.e. Chapter 38). This has been treated as evidence that in the version of the Laozi known to Han Fei the 'Dejing' preceded the 'Dao jing*, as in the MWD MSS. As we might expect of a proponent of Legalism, Han Fei strongly favors the 'Dejing' as a source of quotations; however, the sequence of his citations does not preserve the order of passages in the MSS., nor does it respect even the boundaries between the two sections of the Laozi.11 Thus, it is quite possibly coincidental that we„find the openinq of the:'Dejing* cited first in the 'Jielao'._ One might just as reasonably hypothesize that Han Fei begins his commentary with this passage simply because it is the most general statement in the Laozi of the nature of de , a concept of great importance in the metaphysical underpinnings of Han Fei's Legalism. The MWD MSS. lack the clear-cut division into 81 chapters that characterizes all the TTS. Gao and Chi believe that the undivided state of the MSS. represents the original form of the Laozi and proceed to point out, as others have before them, that the chapter division in the transmitted versions often fails to respect the contextual units of the original (LZ 117). The statement that the MWD texts lack chapter divisions is disputed by Henricks who has noted the divisions marked by prominent dots placed in the center of the MSS. columns ( 1979: 49-51). Curiously, these dots are confined almost exclusively to 6 the 'Dejing' of AL.. Henricks' analysis of this part of the MS. has shown that the sections marked by these dots do, in fact, correspond to thematic units and in some cases represent a more reasonable division of the text than that found in the transmitted versions. Since it was not the practice of early bibliographers to record the subdivision of a work into units smaller than p_ian (Uj), it is difficult to ascertain when the current division of the Laozi into 81 (37+44) chapters was made.. A quotation, of questionable authenticity, from the .Cjilue (-fcc. 6 B.C.) of Liu Xin, cited in a Song work preserved in the Daozancj is relevant to this question (KS 6a).. The quotation describes an edition of the Laozi in two sections and 81 (37+44) chapters prepared by Liu Xin from an assortment of older texts totalling five sections and 142 chapters._ If this quotation is accepted as authentic, we then know not only that the present division of the text into 81 chapters originated with Liu Xin, but also that copies of the Laozi with various numbers of chapters were in circulation by the end of the first century B.C- Other traditions have been preserved describing early division of the text into 72 or 81 (36+45) chapters to tally with Daoist numerology (Chan 1963:75). The colophon of one of the Dunhuang MSS. of the Laozi (Stein #6453) records the latter division, though the MS. itself is in a fragmented state (KS 12a). Lu Deming's sound glosses on the Laozi (^;5"^.J^ ) make no mention of chapter divisions indicating that an undivided version of the text may have been current as late as the early seventh 7 century. From comments in the early Southern Song bibliography Junzhai dushuzhi G§KIM t$^^) . however, we can infer that by the middle of the twelfth century the division of the Laozi into 81 (37+44) chapters had become standard.12 if we impose this pattern of division on the MWD material, we find some discrepancies in the sequence of passages. ' Dejing 1 .' Dapjjing'. Chapter units 38-39 1-21 41 24 40 22 23 42 - 66 25 - 37 80-81 67-79 Gao Heng and Chi Xichao have argued that, in at least one case, the sequence of passages in the MSS..preserves a qreater sense of continuity than the chapter arranqement of the TTS. They point out that Chapter 41 interrupts the continuity between the end of Chapter 40 and the beginning of Chapter 42 where the common theme is the generation of the phenomenal world (^^%) (LZ 121-122). As for Chapter 41 itself, it seems to be an arbitrary lumping together of two very disparate passages, the first a description of the Dao's reception among knights (-i) of differing capacities and the second an evocation of concepts like shangde (-fc-^), dafanq )r etc. throuqh paradoxical similes and seemingly contradictory assertions. The suggestion that the sequence of these passages in the MWD texts better preserves the author's original progresssion of thought is 8 certainly plausible. On the other hand, it would be very difficult to argue convincingly for the superiority of the sequence of the MS. texts over that of the TTS in any other instance in which they diverge. It would seem that modern standards of continuity and logical progression are: of limited applicability in the discussion of the arrangement of a text as laconic as the Laozi. The derivation of the two Laozi MSS. As we have seen, the two MWD texts differ in style of script and date of copying.. Yet they share the.same sequence of passages and, to a large extent, the same wording.\3 Do they belong to the same textual tradition? This is a very difficult question to answer at the present stage of our knowledge of the early Laozi. Gao and Chi believe that BL was not copied from AL and this much seems fairly certain. (The case is less clear, however, if we consider the possibility that BL may have been transcribed from a dictation of AL.) They also assert that the two MSS. derive from different sources (LZ 119). This opinion is offered without supporting evidence or discussion and so seems little more than an impressionistic comment. My own view is quite the opposite, namely that AL and BL do share enough features to derive from a common archetype.. One possible way of uncovering evidence either to corroborate or refute these impressions would be to examine the two MWD MSS. .for what textual critics refer to as 'separative variants', i.e. disparate readings which reveal that BL was not copied from 9 AL. If only •conjunctive variation1 is found, we can assume a common archetype for both MSS. (The distinction between these two types of variation will be discussed and exemplified in Part Two below.) Another method of delineating the relationship between the two MWD MSS.. would be to collate them with quotations from the Laozi preserved in early texts like the HFZ whose relative freedom from corruption can be demonstrated by comparison with the TTS. If the collation revealed that one of the MSS. differed significantly from the other in its rate of agreement with the HFZ witnesses, we would then be justified in inferring that AL and BL belong to different traditions.. These results would suggest that the MS. which agreed in a significantly greater number of instances with the HFZ shares a common ancestor with the version of the Laozi known to Han Fei, while the other MS. is independent of that tradition. If, on the other hand, both MSS. differed but in contrasting ways from the quotations in the •Jielao' and 'Yulao', we could then posit three independent traditions of transmission.. Reference to a third 'control' text would in this way facilitate a more accurate judgement of the derivation of the two MWD MSS. In the preceding paragraphs I have attempted to describe the significant extra-linguistic features of the two Laozi MSS. recovered at MWD. Some of the many variations between these MSS. and the TTS at the level of word and phrase will be taken up in detail in the body of this study. It now remains to 10 discuss briefly some of the problems of the authorship and transmission of the Laozi. This is a vast subject which has been treated by generations of prominent Chinese, Japanese and Western Sinologists. A comprehensive review of the secondary literature alone would constitute a considerable scholarly undertaking. I propose instead to summarize critically three accounts of the origin of the Laozi each of which embodies a different approach to the problem of authorship. Laozi is one of those figures like Homer whose historicity is so uncertain that one suspects them of having been invented by Tradition. The gradual re-evaluation of the sources and methods appropriate to the discussion of authorship in early China has fostered the sort of radical skepticism implied by the last statement. Each of the following three hypotheses on the origins of the Laozi illustrates a different degree of skepticism vis-a-vis the received traditions. The first approach may be called 'traditionalist* because it adheres closely to the account of the composiion of the Laozi found in Shiji 63. Sima Qian attributes a work in two sections and some five thousand words on the meaning of dap and de to one Lao Dan 0^-) , a contemporary of Confucius (551-479 B.C.).. The Shiji biography also includes an account of a meeting between Lao Dan and Confucius. . Now the fact that this meeting is described several times in the Zhuanqzi has been interpreted by the traditionalists as conclusive evidence:that Lao Dan lived before Zhuang Zhou (middle of the fourth to early third c. B.C.). Even if the encounter with Confucius never took place, 11 they maintain, Lao Dan's book, the Laozi, must have been in circulation by the time of Zhuang Zhou.. As D.C. Lau has pointed out (1963: 149), this conclusion rests on the assumption that Zhuang Zhou is the sole author of the Zhuangzi. Authorship in early China is rarely so easily determined.. It can safely be stated that no extant work of pre-Han philosophy is entirely free of either corruption or interpolation. This is particularly true of the . Zhuangzi, whole sections of which are now recognized as dating much later than Zhuang Zhou himself. Thus, if one is going to date Laozi and his book solely on the basis that they are mentioned in the Zhuangzi, the dates of the relevant parts of the Zhuangzi will have to be established first. The traditionalist view can likewise be faulted for its assumption that Lao Dan composed the Laozi..This obviously is open to serious question, since cases of forgery and misattribution are not at all uncommon in Chinese literary history. Suffice it to say, the traditionalist view of the origin of this text has been superseded by a more sophisticated awareness of the difficulties of dating pre-Han literary material. Long before the discovery of the MWD MSS., most Sinologists had become dissatisfied with the traditional attribution of the Laozi and were pursuing a more critical examination of the evidence. One approach which reflects this increased distrust of traditional sources has been taken by Qian Mu in his study of the chronology of the pre-Han philosophers, Xiangin zhuzi^xinian (Qian 1956: 202-226). Qian's concern is to 'demythologize' the 12 figure 'Laozi'. To this end he performs a meticulous analysis of the traditions associated with the supposed author of the Laozi. He points out, no doubt correctly, that 'Laozi' is not, in fact, a name at all, but merely an epithet which could in principle be applied to any venerable man of learning. The several legends which by the time of the Shiji had grown up around a figure referred to as 'Laozi' can be separated into three strands, each traceable to a distinct historical individual, according to Qian. Thus, Lao Laizi ) is the figure mentioned briefly in Lunyu 38.18.7 who later became known as Hex iaozhangren (^ej/flj| . . Lao Tan (4/f) , keeper of the archives of Zhou, is described in Shiji 63 as having had an audience with Duke Xian of Qin 219 years after the death of Confucius. Most enigmatic of all is the figure of Zhan He {/feivj) , mentioned in the HFZ as a foreseer (338). Qian believes the figure 'Laozi' was popularized by none other than Zhuang Zhou himself who apparently delighted in inventing philosophical characters and then attributing words of wisdom to them. He even fabricated a number of disciples for his Laozi and arranged several meetings between him and the historical Confucius (Qian 1956:223-224). However, since none of the maxims attibuted to Lao Dan in the 'inner* chapters of the Zhuangzi are found in any extant version of the Laozi, Qian assumes that the latter did not yet exist when the *inner' chapters were composed.. In the 'Tianlun* chapter of the Xunzi, however, Laozi is characterized as 'discerning about bending (i.e« submission), but lacking insight into extending (i.e. assertive action)' 13 (^^IL^tifc^S/^Mt • Xunzi 64.17.51). Since this brief description accords well with the gist of the Laozi, Qian regards it as good evidence of the book*s existence. Of the three historical figures whose identities have become:conflated in the evolution of the Laozi traditions, Zhan He is the most likely to have authored a book in the period between the composition of the 'inner* chapters of the Zhuangzi and the Xunzi. Qian identifies him as a contemporary of King Xiang (^) of Chu (r. .298-262 B.C.).. The implication of Qian's theory is that Zhan He simply attributed his own book to the wise 'old man' of Zhuangzi's 'inner' chapters; thus, it became known as the Laozi. The obvious contrast between the traditionalists' and Qian's use of these sources lies in the thoroughgoing skepticism Qian brings to the Shiii biography of Laozi.. While declining to accept Sima Qian's statements at face value, Qian Mu nevertheless assumes that they contain certain historical truths which when purified of legendary elements are confirmed or, at least, not contradicted by other pre-Han sources. Thus, the primary object of his investigation is less the date of the Laozi text than the identity of the historical figures hidden behind the epithet 'Laozi*. One might say that he succeeds in rewriting the first part of Shiji 63 and then rests content with the very plausible suggestion that the composition of the Laozi can be dated between the 'inner* chapters of the Zhuangzi and tJie Xunzi. 14 The third approach to the authorship of the Laozi differs 14 markedly from Qian1 s in that it is scarcely concerned at all with the Laozi legends or the Shiji biography. In this approach, exemplified by ths: Japanese scholars Kimura Eiichi (see Hurvitz 1961) and most recently Shima Kunio (KS 1-7), the main focus is on the transmitted versions of the text itself and the many passages parallel to the Laozi found in other early texts. The Laozi is here treated not as the work of a single author, but as a patchwork compilation which evolved cumulatively into its extant form over a period of several centuries. In his study Shima first attempts to date the emergence of a core text and then to delineate the stages of accretion through which it passed as it came to assume its present form. The object of the investigation is thus the words of the text itself and their co-occurrence in other texts rather than the personality with which the text has been associated. . The discovery of two second century B.C. MSS. of the Laozi is of no immediate relevance to Qian Mu's or even the tradional theories of authorship since both posit a date of composition for the Laozi that is too early to be called in question by the existence of the MWD materials. On the other hand, Shima's argument, first published in 1973, comes into conflict with the evidence later unearthed at Changsha. The theory of multiple authorship has not been completely refuted, however, and still merits detailed discussion in the light of the MSS. Shima begins with the 'inner' chapters of the Zhuanqzi which, he agrees with Qian Mu, predate the composition of the Laozi. He regards the theory of universal change (ifej'itj) 15 enunciated in the 'inner' chapters, particularly in the •Qiwulun* as the intellectual antecedent of Laoism. He believes that the philosophy of Laozi emerged as an attempt on the part of Zhuangzi's school to apply his notions about natural change to the world of human affairs.. This pragmatic offshoot of Zhuangzi's metaphysics gradually gained currency so that in the later 'outer' and 'miscellaneous' chapters of the Zhuangzi we find not only passages identical with or similar to sections of the Laozi, but in the 'Gengsangchu• chapter locutions such as •the way/words of Lao Dan' (^^^-if^/).. This strongly suggests that by the date of these chapters there existed a body of thought associated with the name Lao Dan. Further evidence of this comes from the 'Tianlun* chapter of the Xunzi (cited above) where Laozi's ideas are briefly described along with those of other Zhanguo thinkers. Finally, in the 'Tianxia' chapter of the Zhuangzi, which contains a similar overview of the intellectual currents of pre-Qin China, Lao Dan is quoted and assigned a position in importance ahead of Zhuang Zhou himself.. From this evidence Shima infers that what he calls the •primitive text' of the Laozi had appeared by the time of the 'Tianlun' chapter of the Xunzi. In this he again agrees with Qian Hu, though Qian, of course, dates the complete Laozi to this period. Shima views the later and fuller account of Laozi's philosophy in the 'Tianxia* chapter of the Zhuangzi as the historical core of Laoism, a standard against which all statements later attributed to Laozi can be tested. Parts of the TTS which neither duplicate nor harmonize with the passage 16 in 'Tianxia' Shima regards as later accretions to this core (KS 3a). This is surely one of the most radical criticisms ever advanced of the authenticity of the TTS. . Shima marshals support for this view by listing a total of thirteen examples of phrases parallel to the TTS but attributed to quite different sources in pre-Han works. For instance, the phrase#"X^ 1 found in Chapter 2 of the TTS is attributed to Huangdi in Chapter 22 of the Zhuangzi. Or again, statements very similar to the 9&Zsp ®f^ti)$'4'z^t'f®^£of- iaozi 36 appear in both HFZ ('Shuolin') and the Zhanquoce where they are attributed to the Zhoushu (^^)» In each of these thirteen cases Shima assumes the wording was borrowed and incorporated into the Laozi from an earlier source. This then is the method by which the text of the Laozi was gradually put together from extremely diverse sources over a period of time extending from the late Zhanguo to the end of Western Han. As for the compiler of the most •primitive' layer of the Laozi, Shima proposes identifying him as Heshangzhangren (7»j_h. )LA ).1 5 The next stage in the growth of the Laozi, according to Shima, can be inferred from the many verbatim quotations from the text in the HFZ.. Still further augmentation had occurred by the time of the Huainanzi. This can be demonstrated, Shima maintains, by the fact that the opening sentence of Laozi 1 quoted in the 'Jielao' chapter of HFZ reoccurs together with the second sentence (i.e. ) added in the 'Daoying' chapter of the Huainanzi. Shima infers that ... was added 17 —O'J^ ••• some time betwen the composition of these two works.1* Another example of this gradual accretion of phrases to the Laozi comes from comparison of the 'Gengsangchu' chapter of the Zhuangzi. again with the 'Daoying'. In the former we find a string of statements attributed to Laozi that appear in Chapters 10, 22 and 55 of the TTS. Though there are points of identity between the two texts, the Zhuangzi passage includes a number of phrases not found in the Laozi. The 'Daoying* version, on the other hand, is identical in every respect with the opening of Laozi 10, including the phrase %L ^ which does not appear in the Zhuangzi. Shima's inference is that this four-character phrase was added to the Laozi sometime between the date of the 'Gengsangchu' and the 'Daoying*. Chiefly because of obscure chronology, Shima's hypothesis of the.patchwork compilation of the Laozi is, up to this point, difficult either to prove or disprove. He begins with a very minimal Or-Lapzi to which substantial layers were added during the late Marring States period.. This view of the origin of the text stands, in diametric opposition to the traditional assumption that Lao Dan described in Shiii 63 as an older contempoary of Confucius was the sole author of the Laozi. Yet Shima's theory agrees much more closely with what modern scholars' have inferred about the methods of compilation and transmission of pre-Han texts (see Lau 1963: 147-^174). Still, we may guestion Shima*s rather loose criteria for identifying quotations: vague similarity of content to some passage in the TTS of Laozi seems to be all that is required for him to claim a 18 phrase for his primitive Laoist core.17 A more fundamental methodological assumption of Shima's also requires examination: he recognizes as belonging to the primitive Laozi only those statements which have been quoted and attributed to Laozi in other texts. Prior to the.HFZ, if a quotation is not explicitly attributed to Laozi, Shima assumes that the borrowing went the other way, i.e. the phrase was adopted by the Laozi at a later date.. This methodology sounds rigorous; it is, in fact, too rigorous for the material to which it is applied. Just because Laozi 8 is not cited in a single pre-Han source, must we necessarily conlude that it did not become a part of the Laozi until Han times? This strict reliance on quotation, if applied to the dating of other pre-Han philosophical works soon results in absurdity. We know as little about Zhuang Zhou as we do about Lao Dan, yet no one relies solely on citations in other early texts in attempting to date the 'inner' chapters of the Zhuanc[zi. Admittedly, the lack of any historical reference in the Laozi and the terse, aphoristic style of the text makes it very difficult to spot possible interpolations. Would it be demonstrably wrong, however, to posit the existence of the text of Laozi, substantially as we know it today, in the middle of the third century B.C.? Identical passages contained in later works could then be explained as unacknowledged quotation of this fully-formed Laozi. Perhaps the major difficulty with such general hypotheses as these lies in the fact that we lack a rigorous chronology for the evolution of any pre-Han text in which parallels to the Laozi occur. 19 With the next step in Shima's argument we reach less speculative ground. Those chapters of the present Laozi not quoted in the Huainanzi, HFZ or Zhuangzi Shima treats as post-Huainanzi accretions to the text.*8 Though the Huainanzi may possibly incorporate some earlier material, it is extremely unlikely that its compilation predates the. maturity of its patron, Liu An (179-122 B.C.). D.C. Lau places it at about 140 B.C. (1963: 175). The disproof of Shima's hypothesis is, of course, in the existence of the MWD MSS. To reiterate, these two copies of the complete Laozi, corresponding in substance to the transmitted versions, were sealed in a tomb in 168 B.C., and one of the MSS.. was almost certainly copied before 206 B.C. Thus, we know now that whatever process of accretion and interpolation occurred in the formation of the Laozi, it had reached virtual completion by 206 B.C. The discoveries at MWD point up the difficulties in attempting to date a text on the basis of parallel passages found in other works, unless the direction of borrowing, i.e. the date of the parallel text, has been established. This is not to say that Shima has been proved completely mistaken in his treatment of much of the Laozi as a patchwork compilation; it merely means that overconfidence in the validity of his dating procedure led him to an erroneous conclusion. Kimura Eiichi, as summarized in English by Hurvitz, also adheres to the theory that the Laozi underwent gradual augmentation during the Warring States period.. However, he is less precise than Shima about the several stages of accretion 20 through which the text passed simply because he recognizes that short passages and locutions shared by two pre-Han works were not necessarily 'borrowed by either from the other, but rather the common property of all Chinese philosophers of the time' (Hurvitz 196 1: 315) . Shima next gives examples of some of the ways in which he believes the earlier versions of the Laozi were amplified and expanded: sections were supplemented to give a sense of completeness, the wording of some passages was dilated, maxims amd locutions were repeated in different sections of the work and transitional passages supplied. Much of this seems to make very good sense if we push the period of these interpolations back from before 10 B.C. (the approximate date of Liu Xiang's redaction of the Laozi) to almost two centuries earlier for the MWD texts. In one case, Shima's speculation about the late addition of epexegetical words to Chapter 71 is confirmed by the MSS. (Ks 7a, AL 7b.3, BL 7b. 11).. The version of the Laozi known to Sima Qian contained over five thousand characters, a figure which tallies well with both the length of BL (5,467 characters) and the several TTS examined by Fu Yi (4fj^) in early Tang times (Shiji 63.6 BL 9a. 5, 15b.9 and KS 5b).. The work thus described in the Shiji was almost certainly collated with other texts by Liu Xiang at the end of the first century B.C. Though they differ in many matters of detail, the lack of major discrepancies in style and content between the transmitted versions leads Shima to posit Liu Xiang's redaction as the archetype of all extant editions of the 21 Laozi. Although Liu Xiang's Laozi has not survived, it established in the early first century B.C. a textual orthodoxy which has continued unbroken, according to Shima.. The new evidence from MWD suggests that this orthodoxy may extend back as far as the late third century B.C. If we disregard the mistaken conclusion reached by Shima on the basis of the pre-rMWD evidence, is there any way of salvaging his theory of cumulative compilation of the Laozi by adjusting it to the new MS. evidence? If we retain Shima's date for the composition of the Or-Laozi, the process of accretion and augmentation he describes would have to have taken place between the early third century B.C. and the accession of Han Gaozu in 206 B.C., the latest possible date for the copying of AL. While it is not impossible to conceive of such a process occurring in less than one hundred years, this greatly shortened time-span seems to lend support to the single-author hypothesis. As mentioned above, the mere existence of the MSS..does not offer any sort of proof of either of these views; however, the plausibility of the.theory of cumulative compilation has, I believe, suffered in the light of the MS. evidence. The transmission of the Laozi The twenty-odd transmitted versions of the Laozi collated by Shima in KS are grouped into six textual traditions: 1) the Yan Zun {gfcg) text, 2) the Wang Bi (:EJD§5) tradition, 3) the Fu Yi (^f. |E) or guben - ) text, H) the Xianger texts from Dunhuang, 5) the Heshanggong (>«j_k/i\) tradition, and 6) the text 22 accompanying the commentary composed by the Tang Emperor Xuanzong (3^ ). With the aim of establishing the relative age and thus authenticity of each of these traditions, Shima traces their lines of descent through bibliographic records as well as official and Daoist historical writings.. This is a task requiring the command of a great number and variety of sources together with carefully considered speculation with which to fill the inevitable lacunae in over a thousand years of bibliographic tradition. The length and i intricacy of Shima's arguments fully reflect the complex nature of these problems. What follows is no more than a summary of his conclusions, some of which in the light of the new MS. material suggest fruitful avenues for future research. The Yan Zun (YZ) text A brief biography of Yan Zun is found in Hanshu 42 where he is mentioned as a mentor of Yang Xiong (53-18 B.C.), making him roughly contemporaneous with Liu Xiang.. Because this text does not differ significantly from the other transmitted versions, Shima concludes that Yan must have based his text on Liu Xiang's redaction. Yan is further credited with a commentary in two juan which disappears from bibliographic notice after Lu Deming's Jingdian shiwen xulu (£v§.of c. . 580. His text of the Laozi is known as Zhigui or Laozi zhigui (4*i^) and beginning with the Suishu bibliography is variously listed as having 11, 14 and 13 iuan. By Song times a commentary by Gushenzi (&/^3~) had become attached to this text and it is 23 this version with commentary, minus the first six juan corresponding to the 'Daojing', that has been preserved in the Daozang. The Hang Bi (WB) text Though or perhaps because Wang Bi (d. 249) composed an influential commentary on the Laozi, the present state of both his text and commentary is extremely corrupt, according to Shima. The work is listed in the bibliographies of Sui, Tang and Song times and is quoted by Yan Shigu (581-645) in his commentary to the Hanshu. By this time, however, an attempt was being made by supporters of the Heshanggong text to homogenize Wang Bi*s version with their own in the interests of religious orthodoxy. Though Wang Bi»s text survived in a mutilated state, there was evidently a period in the Southern Song when it all but disappeared from sight. The earliest edition of the text extant today is found in the Zhengtong Dapzanq of 1445 and here it is obvious that both the Laozi and its commentary have suffered heavy alteration in the direction of the Heshanggong tradition. All later redactions of Wang Bi's text stem from this corrupt version. The Fu Yi (FY) text Fa Yi (558-640) served as Grand Astrologer {JSJ&J^ ) in the early years of the Tang dynasty. His text of the Laozi together with a commentary, which may have been no more than a collection of sound glosses, was transmitted intact until the end of the 24 Song. By the time of the compilation of the Daozang in the first half of the fifteenth century, the commentary had been almost entirely lost. In addition to the Daozang edition, a slightly altered version of FY was the basis of a recension of the Laozi by Fan Yingyuan (i&Jj^Tu) of the Southern Song. Thus, Fan's text (FYY) is treated as belonging to the Fu Yi tradition in Shima's variorum. Shima quotes from a Song account preserved in the Daozang a passage describing Fu Yi's collation of various texts in the preparation of his recension (KS 10b). Mention is made in this passage of a text reported recovered from the tomb of Xiang Yu's concubine, Yu Ji (^xS)• If Fu Yi had access to such a presumably pre-Han text and actually made use of it in his collation, we might then expect to find in his redaction some earlier readings not preserved in the other TTS. This possibility will be investigated more closely when we examine passages from the MWD MSS. and compare them with the versions recorded in KS.. Shima's description of FY as having been heavily contaminated by other traditions does not bode well, however (KS 51b). The Xianqer (XE) texts This text represents considerable tampering with Liu Xiang's recension and together with the Heshanggong tradition has exerted great influence on the transmission of the Laozi. Though no mention of XE is made in the bibliographic chapters of the Sui and Tang histories, Lu Deming records a 'Xiangyu' (^§.^.) 25 commentary attributed to both Zhang Lu (^&j§, d. 216), the grandson of Zhang Daoling (6^.^^) an^ a certain Liu Biao (l^'J^O. The only other bibliographic entry for XE from the medieval period is found in the 'Guangshengyi xu* 3;J^/*fj~) DY the late nineth century Daoist Du Guangting Cfc±.7u/^)_. . Here the commentary is attributed to Zhang Daoling himself. Probably not long after this, XE disappeared from sight until the turn of this century when several MSS. belonging to this tradition were discovered at Dunhuang. Shima argues at some length in favor of identifying Zhang Lu as the editor of this text. The commentary apparently embodies the interpretation of Zhang Lu but was put into writing by his pupil Xu Laile (t$\fcft]) ; thus, it should properly be known as the 'Zhangxu' text.. However, as Shima explains, the surname Zhang of the patriarchs of the Daoist church was anathematized and in this case Xiang was substituted (KS 2 1a). Xiangxu (]§^) became, in the process of transmission, corrupted to Xiangyu (-£1.^) as found in the bibliography of Lu Deming's Jingdian_;shi wen. Still later, ^ was misread as the grass form of the graph (viz.^) or possibly as & , a frequently encountered graphic alternate for H| , and thus the referentless name was born. The Heshanqgonq ,(HSG) text Sima Qian makes brief reference to a Heshangzhangren (i^A~) in the Shiji (80.17) in a context that would place him chronologically in the middle of the Warring States period. A later legend concerning the transmission*' of Laozi's teachings 26 through HSG to Han Wendi is found in some versions of Ge Hong's Shenxianzhuan (^iXih^) but Shima concurs with other scholars in regarding this latter account as an interpolation (KS 26a, 53a.40). The earliest mention of the HSG edition of the Laozi in found in the Lianqlu (H^sL 520-526), a bibliography of the early sixth century, now known only through extensive quotations preserved in the bibliographic chapter of the Suishu of 656. The Lianqlu attributes a commentary on the Laozi to Heshangzhangren of the Warring States period, but this work had been lost by the.time the Sui bibliography was compiled,. Shima presents a long and intricate refutation of this traditional attribution and seeks to demonstrate that the HSG text and comentary are both forgeries of the late fifth century. No doubt the strongest evidence in his favor is the fact that the commentary makes extensive use of earlier commentaries on the Laozi, including works as late as that of Gu Huan (/§|| , f 1. . 483).. On the other hand, we find the HSG commentary itself quoted by Huang Kan (iL^iXv 488-545) in his scholia on the Lunyu (l^%l-;J|1JH5k). This allows us to date the HSG commentary quite precisely to the late fifth or early sixth century. The exact derivation of the HSG text is more difficult to determine because . of the likelihood of contamination between texts even after the transmitted traditions had become established.. Shima cites a number of examples which support his contention that the HSG text is a redaction of the XE (KS 32b). 27 This edition of the text was produced with commentary by the Tang emperor between the years 723 and 732., Xuanzong based his redaction on the XE and HSG texts, selecting readings from both. Thus, it represents a blending of two earlier traditions and must be used with care in collation. All later editions of the Laozi are likewise composite versions and so have: little value as textual witnesses. Shima*s:variorum_and recensions The history of critical recensions of the Laozi begins with Lu Deming's Daodejinq yinyi (ifj^&x' ) . Lu accepted WB as the most authentic text and compared it to other versions, particularly the HSG. Unfortunately, by his time the former had already become contaminated by the latter. Until Qing times little was accomplished in the area of textual anaysis of the Laozi other than the noting of variants. . Shima Kunio's purpose in preparing KS was to classify all the best extant editions according to the textual tradition to which they belong and then to produce a recension for each tradition. He has also included a large number of parallel passages and verbatim guotations from the Laozi found in a wide variety of sources extendng from the pre-Han corpus to the Taipinq yulan of 983.. He quotes in full the early commentaries, (i.e._XE, WB and HSG) as there are not a few cases where, in reiteratinq the Laozi text, the commentaries preserve readings which in the text itself have been corrupted through the influence of other traditions. For each of his six recensions, he provides notes in classical Chinese on the source 28 and extent of contamination between texts as well as the rationale for his choice among the variants. . Finally, he suggests that his recensions of the 'Daojing' of WB and the •Dejing' of YZ taken together form a definitive text CfcL^h ) of the Laozi. It remains for the present author to clarify the approach to the study of the MWD MSS. adopted in this essay.. If, for instance, the MSS. differed radically from the TTS of the Laozi, there would be cause for attempting a reassessment of early Daoist thought. This, however, is not the case. To be sure, there are a number of brief passages 'missing' in the MSS. and thus perhaps later interpolations.»9 Likewise there are a few individual words and a phrase or two unique to the MWD versions of the text.20 There are even cases of the additional particles of the MSS. disambiguating difficult readings in the TTS.21 The number and importance of these discrepancies, however, is not sufficiently great to urge a complete rethinking of the Laozi. The value of the evidence which can be gleaned from the silk MSS. is of a different kind. Ontil as recently as ten years ago the extant MSS. remains of pre-Han China were confined almost entirely to epigraphic materials, stelae, the Shang and Zhou bronzes and the Shang oracle-bone inscriptions.. The extensive pre-Han literary corpus was known wholly through the medium of TTS. It is this body of often copied and recopied data on which linguists have, faute-de-mieux, based their observations and hypotheses on the nature 29 of the: early Chinese language. The obvious philological value of early MS. material lies in its use as the best available touchstone for these descriptions and analyses of classical Chinese.22 In studying an early Chinese MS., the student is forced to become his own editor, applying what he knows of paleography, historical phonology and grammar to the decipherment and intepretation of the text before him. If it is a text which also exists in transmitted versions, his task will be lightened by the TTS themselves and the help afforded by a tradition of scholarly interpretation embodied in commentaries. But his investigation will also be complicated by the need to compare scrupulously the MSS. with the transmitted versions and to consider their inter-relationship.. Theories of the derivation of textual witnesses are necessarily based on the investigation of the nature and extent of variation between them. The two essays which follow aim to clarify what is meant by 'variation' in the. context of the MWD MSS. The first is a classification and discussion of the several kinds of variation possible between single graphs (i.e. substitution variants) in the MSS. and TTS. In the second, selected passages will be collated in an attempt to discover what the different types of variation reveal about the derivation of the two MSS. and the quotations from the Laozi found in HFZ. Finally, the heuristic nature of these studies must be emphasized. It would be overly sanquine to expect definitive evidence on the date of the Laozi or the course of its 30 transmission to emerge from this study. At this early stage in the author's acquaintance with the methodology of textual criticism, his objective must remain a humble one: to achieve a clear perception of some of the problems encountered in the study of early Chinese MSS. He takes heart, however, in the frequently encountered assertion that textual criticism is a skill learned more by grappling with a MS. than by the mastery of manuals. If this is so, then the pedagogical value of the exercise remains intact. The good critic aspires to the impartiality and clear reasoning enjoined in the critical writings of A. E. Housman and to the scrupulous attention to detail which can lead to the results claimed by Alphonse Dain: •On n'oubliera pas ... que tous les progres de la philologie ont 4te faits autour du probleme de l'edition des textes' (1964:159). 31 N^ Introduction 1. The term 'transliteration' is here used in the specialized sense given it by Alphonse Dain (1964:125): 'the operation [or result] of transcribing a text from the earlier graphic style of the exemplar to a more modern style'. . The MWD MSS. themselves are written in the small-seal and clerical scripts; the versions of these MSS..published in periodicals and book form are all transliterations into kaishu. 2. Principal sources for this brief summary on the MWD excavations and recovery of the silk MSS. are as follows: Hunansheng 1974, Xiao 1974, Tang Lan et al. .1974, Gao Heng and Chi Xichao 1976, Buck 1975, Loewe 1977 and Li Xuegin 1979. 3. Li was probably already Chancellor (2M^) to King Wuhui (^Q) of Changsha when the Han central government enfeoffed him as Marquis of Dai in 193 B.C. He died seven years later in 186 B.C. (Shiji 19.6, Hanshu 16.618, Ma 1972:16). His name is found written^']on one of the seals recovered from his tomb, in the Shiji and in the;Hanshu. 4. In his position as Chancellor, Li was apparently succeeded by a certain Yue (??J&) who in 184 B.C. was enfeoffed as Marquis of Lilinq (gtpfe. Shiji 19.6, Hanshu 16.624). , 32 5. The. importance of the figure of Huangdi in the 'Shidajing' has led other scholars to identify this text with the Yellow Emperor's (Book of ) Lord and Vassal (^ ) or the Li Mu (#$t-) , both listed in the Hanshu bibliography (30. 1731). 6. Steve Dickinson of the University of Washington has written an exploratory study of this problem, 'Historical Implications of the Ma Wang Tui texts: Huang-lao Taoism Reconsidered *. . 7. Concerning the evolution of these styles of script we have the testimony of Xu Shen (30-124) in the preface to the SW (Duan 1815:15.20) and Ban Gu (32-92) in the bibliographic chapter of the Hanshu (30.1721). The small-seal represents a modification of the style which had developed in the state of Qin during the Zhanguo period. On the unification of China by Qin Shihuang in 221 B.C. it was made the standard. The clerical script began as a simplification of the small^seal, a kind of shorthand for use: in record keeping, and later established itself as a standard script. 8. Examples of the small-seal and clerical graphs are copied from Akai 1 974. 9. The Chinese scholars who have written on the dating of the Lao zi MSS. differ among themselves on the question of when an emperor's personal name became taboo, Xiao Han (74:41) 33 dates AL between 206 and 195 B.C., implying that^-p became taboo only on the death of Liu Bang.. Gao Heng and Chi Xichao (LZ 111,120), on the other hand, assume that the taboo against began in 206 B.C.. when Liu Bang became emperor; hence, they date the copying of AL to before 206 B.C. Similarly, they state that BL must have been produced by 194 B.C. when Huidi assumed the throne.. Although the locus classicus for the tabooing of names states that the taboo went into effect only after the completion of funeral rites (^7^./5"l4 . Liii SS JZS: 5. 58b) , this probably refers to Zhou practice. In the Han it was customary to avoid the personal name of the reigning emperor as is shown by an amnesty proclaimed for violators of the current taboo by Han Xuandi, Liu Bingyi (^ B), in 64 B. C. (Hanshu 8. 256, quoted in Chen Yuan 1928:129).. Both syllables of Xuandi*s first name (recorded in Hanshu 8.238) are very common words in Chinese. This had led to frequent violation of the taboo which, in the end, prompted the Son of Heaven to give himself a new name, Xun (1&J) , thereby releasing and for general use once more. This bit of evidence supports the earlier dates for the Laozi MSS. proposed by Gao and Chi. 10. Xiao Han extends the possible date of BL down to the end of the Empress Lu*s regency in 180 B. C«_ He does not explain why he thinks the taboo against Huidi's name did not go into effect until eight years after his death. Incidentally, the fact that we find such uniform avoidance of the graph 4|3 in 34 BL casts suspicion on the frequent assertion that tabooing practice in the Han was much less strict than in later times. Hu Shi has written at some length on this topic citing many instances in the transmitted versions of the Shi ii , Hanshu etc.,where taboos were not followed (Hu 1943, 194 3a), The evidence from Eastern Han memorial stelae makes it clear that the observance of taboos in that period was indeed lax. The only Western Han evidence available, that from transmitted literary and historical documents, is much less reliable since the likelihood of the later restitution of once-tabooed graphs is quite high. The very early Han date of BL suggests that the scribe may have been following a stricter Qin practice of tabooing which was gradually relaxed as the Han dynasty progressed. Futher study of early MS. materials will be necessary before the details of Western Han tabooing can be clarified. 11. The sequence of passages (identified by their later chapter numbers) cited in the 'Jielao' is as follows: 38, 58, 59, 60, 46, 14, 25, 1, 50, 67, 53, 54.. For the 'Yulao': 46, 54, 26, 36, 63, 64, 52, 71, 64, 47, 41, 33, 27., Chapters 59, 60 and 47 are quoted in their entirety.. Citations from Chapters 46 and 54 are found in both the 'Jielao' and •Yulao'; when taken together, these quotations form the whole of the transmitted versions of these two chapters. 12. In his annotations on an edition in 13 juan of the Laozi 35 zhi qui ), Chao Gongwu observes that its chapter division differs slightly from the other editions (i^i^) , implying by this expression that it is nonstandard (Chao 196 7 : 2.6 9 3). 13. Graphic divergence between the two MSS. is common, but only infrequently does it point to a difference in meaning.. In other words, one often finds the same morpheme given two different graphic realizations in the MSS. The reason for this is the lack of graphic standardization in earlier stages of Chinese. (See Part One below on loan characters.) 14. Qian dates the death of Zhuang Zhou to 289 B.C. or a little later (1956:564). Xunzi died in the middle of the third century B.C. (Lau 1963:175); thus, according to Qian's hypothesis, the Laozi was composed some time in the first half of the third century B.C. This late Zhanguo date is corroborated by the rhyming evidence in the text which has been analyzed by Karlgren (1932) and reassessed in the light of the new MS. evidence by Pulleyblank (1979 :1-12) . 15. Qian Mu regards •Heshangzhangren' as an agnomen of his candidate for 'the real Laozi', Zhan He (1956:266). Elsewhere (KS 7a) Shima suggests as a possible author of the 'primitive' Laozi the figure Gengsangchu (1^7$$) i a character Sima Qian denies ever existed (.Shiji 63.10). Given the fragmented nature of the evidence and the 36 number of variables involved, one can only remark that this sort of speculation is of little help in achieving a better understanding of the text. 16. The first sentence is quoted alone elsewhere in the Huainanzi ('Fanlun'), as noted by Shima (KS 4a). This can be explained by the existence of two versions of the Laozi at the time the Huainanzi was compiled, Shima suggests, one more 'primitive* and the other incorporating later interpolations. 'Fanlun* quotes from the former, 'Daoying' from the latter. . 17. This leads to seme rather far-fetched parallels, e.g. the identification of A. % S & ^ , atttributed to Lao Dan in 'Tianxia* with Laozi 22 * i'J AE# or the claim that the meaning of J-vA lso from 'Tianxia' is somehow related to^Jl>^^H^ of Laozi- 59, or that .^Qg^F) ^4#^ ^^7^7^. from 'Qiwulun' is the ultimate source of ^qfe^J^ of Laozi 1 (KS 3a, 6b) . 18. Shima's listing of the chapters of the Laozi not quoted in the Huainanzi (KS 4b) is marred by inconsistency. Chapters 17, 29 and 66, for example, contain phrases also found in the:Huainanzi. 19. In addition to the two sentences in Chapter 71 discussed in Shima's preface, I have.noted the following passages in the 37 TTS which lack reflexes in the MSS. Reference to KS will show that many of these passages are also omitted from one or more of the transmitted versions.. Chapter 2 nri * ^ 15 fa 23 # * ;L i| £ *>ft 24 u$ % X 25 JIUT* J£ 27 % 1 29 2^ ^ s& °^ 30 * ¥^^7flJil^ 39 3fciHD-fcX£.J&3$Q 7i f 20. An obvious example is the opening of Chapter 37 where the MSS. have the sentence ^llfu^ {AL 14a.10, BL14b.7). The TTS all show ^JJ^^^^^^^KS 130). AL 8b. 9). Though BL is defective for part of this passage, it retains the 38 particle -ifej in both sentences. All the TTS lack these two occurrences of the particle-thi ; hence the punctuation of this passage has long been a point of controversy.. The presence of in the MSS. means thatmust be understood as object of the verbs /fy and thus translated '...always rid yourself of desires in order to observe its secrets;/But always allow yourself to have desires in order to observe its manifestations' (Lau 1963:57). The alternate interpretation which originated in the Song dynasty punctuated after fa and^, yielding '...let there always be non-being, so we may see their subtlety (i.e. the subtlety of the ten thousand things),/ And let there always be being, so we may see their outcome' (Chan 1963:97, also 99.5 for a discussion of this controversial reading) . 22..This is not to claim infallibility for any single early MS.; but, at the very least, one can be certain that such a source will not contain errors from a later age. That the MSS. do provide clarification of moot points in the grammar of Old Chinese is shown in the case of ffa . . Several conflicting descriptions of the distinction between these , two negatives have been proposed; the evidence from AL and BL confirms the validity of one of those:descriptions for the.period of the MSS. The controversy and the new evidence have been summarized by Pulleyblank (1979:17-22).. 39 Part One The editing of MS. material generally proceeds from decipherment and transcription/transliteration to collation with other witnesses and finally interpretation. In other words, discussion of the relationship of a MS. to other versions of the same text must be founded on a clear understanding of the nature and extent of variation between the versions being compared. In the case of the MWD MSS. of Laozi, the work of decipherment and transliteration has largely been completed by a team of Chinese scholars who have direct access to the MS. remains. To date they have published four transliterations of the MSS., each incorporating revisions of its predecessor.! Of the two used as sources for this study, one (LZ) is annotated and contains much helpful information on the meaning of rare graphs and the reconstruction of corrupt passages.2 However, there has yet been no general classification of the ways in which the TTS vary from the MSS. at the level of single graphs. In the first part of this section we will attempt such a classification in the belief that it will facilitate later discussion of the connection between the MSS.,and other versions of the Laozi. Once a graph from the MS. has been deciphered and transliterated into its modern kaishu equivalent, the next step is to determine its relation to the corresponding graph or graphs in the TTS of the Laozi. At the graphic level this 40 relationship will be one of identity, which, of course, creates no difficulties, or non-identity which can be accounted for in one of the following ways. I. Taboo characters This category was referred to in the Introduction as a means of dating the copying of these two MSS. In addition to the alternation noted there between in AL and in BL and the vast majority of TTS, the following graphs usually occur in the TTS in place of the taboo characters found in the MSS.: >i&3 for it , personal name of Han Huidi (r..194-188 B.C.) ^ foryl-S , personal name of Han Wendi (r. 179-157 B.C.), always in TTS fWl tor personal name of Han Jingdi (r. 156-141 B.C.)., always in TTS ^ for % , personal name of Han Zhaodi (r. 86-73 B.C.). /v and K/ alternate frequently and quite unpredictably both between MSS. and TTS and between the various TTS themselves. Since both qraphs appear in the MSS., the confusion between them in the TTS is probably older than the tabooing of it^, personal name of Tang Taizong (r. . 627-650), though the Tang taboo may well have aggravated the confusion. Just because a TT does not observe one of the above taboos cannot be taken as proof that its prototype dates from before the time the taboo went into effect. It was common practice to 41 restore the original graphs to TTS after a taboo became inoperative, though, of course, there was a great deal of variation in the degree of thoroughness with which this was done. It appears that sometimes it was not done at all as in the cases of and in the TTS of Laozi. In accounting for this phenomenon we have to assume that the function of these graphs as substitutes for taboo words in the Laozi was forgotten. II. Semantic Variation This is a kind of variation in which neither of the graphs involved was ever tabooed, yet both readings make good sense as they stand. The variants are often synonyms reflecting a difference in phrasing rather than in basic meaning.. 1. 5; M & fin & BL 8b.7 Si most TTS (KS 216, Ch. 77) 2. _^i«I^Sn AL 5b.7 ] '2\ all TTS (KS 184, Ch. 62) 3.i|_^^^£^-& BL 5b. 12 /3t ]1H all TTS (KS 184, Ch. 62) This sort of variation between the MSS. and the TTS is, on the whole, relatively rare. In comparing a single TT to the MSS. one will find a number of these variants; however, as there are, in most cases, other textual traditions which will be found to agree with the MSS., the MWD material offers little previously unattested data of this kind. 42 Since the above two types of variation are semantic in nature they are not treated in the published transliterations of the MSS. One uncovers them by collating the MSS. with the transmitted versions of the text. The remaining three kinds of variation however, represent deviation from modern graphic standards and hence are 'corrected' or at least noted in BS as follows: x (y). This simply means that MS. graph x should be understood as graph y. In LZ this convention has been refined so as to distinguish between MS. graphs which are patently incorrect (i.e.. miswritings, graphs which represent words inappropriate to the context) and non-standard graphic forms (i.e. allographs) of words appropriate to the context (i.e. abbreviated forms and loan characters). In LZ the former are emended between wedge-shaped brackets while the standard forms for the latter are provided in parentheses. III. Graphic confusions (Miswritinq) There are a few examples in the MSS.. where miswriting is the only possible explanation for a given graph.. This conclusion has been reached only after careful consideration of the meaning, phonetic shape and graphic form of the word in question., The rationale for emending the MS.. reading is as follows: a) The meaning of the MS. graph precludes treating it as a semantic variant; it cannot be made to fit the context. b) Phonological disparity between the variants precludes the treating of one as a loan character for the other. 43 c) The graphic similarity of the variants suggests scribal error. In such cases emendation is by reference to a superior witness, the other MWD MS. whenever possible. 4. #^4lj§_?& AL 7b. 11 J^J BL 8a.7 The word /t_ in the context of the above passage simply does not make good sense. The superior reading is preserved not only in BL, but in all the TTS as well (KS 210, Ch. 74).. The wide phonetic disparity between these variants corroborates the impression that AL at least was transcribed by sight from another MS., rather than taken down from dictation. To write /<_ for i><_ involves a visual, not an aural, confusion. One occasionally encounters examples of variation which are amenable to more than one interpretation. Either semantic variation or miswriting seems to provide a possible explanation for the;next example. 5. AL 9a.? Itffl BL 9b. 7 sw >jf >E>^M;JOfj;f The syllable written >|ff *siok is found 7^ *'iwen ('watery abyss') in all the TTS (KS 60, Ch. 4) as well as in BL. . Though no early textual examples of the AL graph have survived, it is recorded in the SW with a definition ('deep and clear') which permits a perfectly sensible reading of 5: '(The Dao is) deep and clear like the great ancestor of the phenomenal world' (reading~&t)' as^A ) . I have counted two other occurrences of the graph ^fj in AL (LZ 20, 28) where BL and the TTS (KS 68, 128; Ch. 8, 36) record /Ml , and one in the unique AM (24b. 4)., In all three of these cases yfft does indeed appear to be the superior reading because of the presence of rhyming words in the environment.3 The question is whether)^ in 5 should be considered an error for >^ff along with the majority of occurrences of >)§) in AL. There are two complementary principles which the Western textual critic might invoke at this point: utram^in alterum abiturum erat --'which reading would be more likely to have given rise to the other?'— and the answer, difficilior lectio prpbipr --'when presented with two readings, one easier to understand than the other, the editor generally accepts the latter as the more reliable or original, on the assumption that the scribe would tend to interpolate or simplify the text he was copying in the easiest possible manner' (Kleinhenz 1976:285). If the yfift in BL is explained as the interpolation of a relatively common synonym (a facilior_lectio) for the rare >i§) , then AL can be accepted as the sole witness of an earlier textual tradition.. There is, to be sure, an argument from the principle of Ockham's razor which favors the uniform treatment of all occurrences of in AL and thus the emendation of 5 to . This is persuasive but not conclusive because it discounts the genuine possibility that the variant in 5 is an 45 occurrence of the word attested in SW. In favor of the reading in AL it is worth remarking that because a scribe is capable of mistaking^ for>ifl three times does not mean he is incapable of copying >/fi correctly. . The single general principle of textual criticism is that 'each case is special' and deserves its own special solution.* In the present case one is hard put to choose between the alternatives. IV. Simplified Graphs This type of variation is best described as a scribal shorthand in which enough of the graph is retained to allow recognition while the rest of the strokes are abbreviated.. Its use is ubiquitous in both MSS. Some of these simplified graphs are themselves hapax leqomena (see below) and may reflect scribal idiosyncrasy. ^'L> for|)Li BL 2a.5, 9b.4, 11b.? for fj? , unique to BL where it occurs at least six times, 3b. 7 passim S-J for || , AL 3b. 12 The contrastive distribution of these abbreviated graphs would suggest that AL and BL were transcribed by different hands. V. Loan Characters Loan characters (LC) make up by far the largest category of single-graph variant in the MWD Laozi MS.. Every LC gloss 46 (signalled x (y) in BS and LZ) implies the following relationship: MS. graph x is to be pronounced and understood as the less ambiguous (y) designated by the commentator. There are many hundreds of passages in the TTS of the early Chinese literary corpus which without recourse to this simple device of textual criticism would today remain impenetrable or at least invite serious misreading. . The reasons for this are simple: standardization of the Chinese script was first promulgated in 221 B.C. by Li Si, prime minister of the First Emperor of Qin. Hence, any written material which predates this graphic reform will reflect the heterogeneous scribal practices of Zhou times. This is most emphatically demonstrated by the multiplicity of graphic forms found in the early bronze inscriptions. Furthermore, since a relatively high degree of graphic uniformity among scribes at the far-flung reaches of the Chinese empire cannot have been achieved for many centuries, it should come as no surprise that we continue, to find large numbers of LC and 'non-standard1 graphs on original documents which postdate the reform, such as the MSS. recovered at MWD. It seems reasonable to speculate that Li Si's standardization of the script had immediate effect on scribal practice only at the short-lived Qin court and perhaps its major provincial outposts. It was the continued maintenance of a single graphic standard at the court of the: Han and later dynasties that over the centuries fostered the dissemination of a uniform script throughout the empire. The ease with which a single LC equation can disentangle a 47 difficult phrase of classical Chinese has, quite predictably, led to the frequent misapplication of LC glosses in exegesis. The equating of graph x which no one can understand in the context with a more easily explicable (y) soon becomes the unskilled commentator's solution to almost all textual difficulties, particularly if he is prepared to ignore that certain degree of phonetic similarity supposed to obtain between x and (y).If x is to be pronounced and equated semantically with (y) in a given context, the assumption is that the normal pronunciation of x should not differ too widely from that of (y) , otherwise graph x would not have been pressed into service by scribes to write the syllable elsewhere commonly represented by graph (y) . In other words, x when used as an LC is a purely phonetic symbol; its usual semantic associations are irrelevant. For example, >^ in 5 above could never function as an LC for vHrf simply because *siok is phonetically incompatible with *'iwen; it is inconceivable that one could do duty phonetically for the other.. Hence, they must be treated either as semantic variants (i.e. and /f read as independent words) or as a case of miswriting. Clearly, the phonetic affinity between x and (y) must be demonstated in terms of the reconstructed values for Old Chinese (OC). For this reason the examination of LCS to discover the precise degree of phonetic divergence permissible between the pronunciations of x and (y) could not begin until the phonology of OC had been reconstructed in detail. Bernhard Karlgren's reconstruction of the sound system of 48 • OC was first published in 1940 and then, in a revised and expanded version, as GSR in 1957. . Armed with this battery of systematic phonological data, he proceeded to examine critically the language of the TTS of the Zhou classics and in particular to scrutinize some of the thousands of glosses embedded in the commentaries to these early texts.. The results of these labors can be found in a long series of articles extending from •Glosses on the Kuo Feng Odes' (1942) right up to 'Moot Words in Some Chuang Tse Chapters' (1976), the last paper he published before his death in the autumn of 1978. His work on LCS takes the form of a long, critical review of over two thousand such glosses proposed by Chinese commentators from as early as the Eastern Han down to the present (Karlgren 1967). In the body of the article he dismisses many hundreds of LC proposals as 'phonetically impossible' on the grounds that the reconstructed normal pronunciation of x differs too widely from the reconstructed pronunciation of (y). For Karlgren, an LC equation is legitimate only when the phonetic disparity between x and (y) does not exceed the phonetic bounds of any single xiesheng (XS) series as Karlgren has defined and reconstructed them. »XS series' is given to a group of graphs all of which employ the same phonetic 'speller' in their construction, e.g. the element in GSR #324 which includes compound graphs with such disparate modern pronunciations as rui^Kj, yue and shuo BTLI . The fact that a single graph with a meaning and pronunciation of its own (modern dui) was used repeatedly as a 49 phonetic element in the construction of compound graphs can only-mean that at the time of their first appearance in the Zhou dynasty the pronunciations of , ^ andlJL/ must all have been quite close to the pronunciation of Ki • . This is one of the fundamental assumptions on which the work of reconstructing OC proceeds. Karlgren, incorporating additional evidence from other sources, reconstructs the pronunciation of these three graphs as *diwad, *diwat and *siwat respectively. Because the members of a XS series all share a common phonetic element, it goes without saying that such graphs can always, in principle, function as LC for each other. Karlgren, in fact, refers to members of the same, phonetic series as 'authorized' LC. By tabulating the degree of phonetic variation found among the syllables of each XS series in OC and then grouping initials and rhymes separately, Karlgren provides a convenient array of the criteria by which he judges the validity of LC glosses (1967:10-17). These basic points can easily be illustrated by a few examples from the MSS.. Whenever we find, for instance, ^ in the MSS. for &1 in the TTS, we can immediately accept the former (lit. 'stomach') as LC for the latter ('call, refer to') because they both belong to the same XS series. This sort of LC, ubiquitous in the MWD material, is indicative of the the low level of graphic standardization at the time the MSS. were copied. . All cases of this type of alternation between members of the same phonetic series (^ for/3&. , ^ for^l , for ^ t and 4 u forM,f$ for^f etc., etc.) are unarguable as LC 50 and hence not further dealt with in this study.. In some cases, the LC graphs belong to different phonetic series but are homophonous as far back as we can reconstruct them (e.g. A£ for 1^1 and J°l for-$f~ ). Finally, there is a large number of instances in which the graphs are not related and the syllables they represent are only imperfect homophones: ^- for^W , £1 for 'ia, ^ f or ^, etc. The use of /Z^ to write'1^ , however, is still acceptable since both syllables belong to the same rhyme group (x_) while their initials, though not identical, do co-occur within single phonetic series. (Jr., a homophone except for tone of ^_ is the phonetic in'r-% .) Likewise, in the MSS. is admissible as an LC for in the TTS because the phonetic variation between these two syllables can be found within individual XS series. ('F- rhymes with || and is phonetic in^, a word which rhymes with'Vo ; the initial consonants of %. and are identical.) The fact that appears in BL for Ifl can be explained by pointing out that words homophonous (except for tone) with It occur within the XS group to which f^E, belongs (viz. /jfiL ). Having laid aside this sort of orthodox LC, we will in the remainder of this section, examine loan variants from the MWD Lao zi which do not seem to fit these basic criteria. As a heuristic guide in these analyses we will continue to use the phonetic contours of the XS series as best we can determine them, although further study of MS. and epigraphic sources may reveal that the LC practice of different periods and areas of pre-Qin China was not always congruent with what we know of XS phone tics. 51 As has been suggested above, Karlgren, on the basis of the same working hypothesis, was led to reject many LC glosses proposed by Chinese commentators who lacked his knowledge of the phonology of OC. To refute these dubious equations, he found it necessary to advance a set of criteria by which the validity of all LC glosses could be assessed. Quick to realize that the XS series of OC constitute: a fossilized record of early LC practice, he adopted ad hoc the phonological contours of each of these series as a set of limits within which all valid LC proposals ought to be confined. Even as little as twenty years ago, Karlgren was unfortunately not in a position to infer the criteria governing LC use from a careful study of the occurrence of LC in early MSS. since at the time of his researches there were no such MSS. available. He seems occasionally to have made use of evidence from the bronze inscriptions, but, on the other hand, he did not undertake a thorough empirical study of LC practice in this important source.. It is certainly true that, even when working with early MS..material, one ought to begin by testing the phonetic variation between each pair of LC against the variation found within XS series since each of these constitutes a cluster of primordial LCs; however, until sufficient data has been compiled, it is clearly a priori to claim that for all time and all place LC practice was limited by the phonetic bounds of these series.. For one thing, this claim reguires the concomitant assumption that, prior to standardization, all users of the written language had either innate or learned command of those phonetic bounds to guide them 52 in their orthography. In their transcriptions of documents, by sight and perhaps by ear, and in the. composition of original material, it is the actual practice of scribes, observable in their MSS., which must be treated as the primary data from which we may then proceed to infer the general principles governing the use of LC in OC. Working from TTS, the victims of centuries of copying, editing and type-setting, Karlgren had no choice but to attack the problem of LC in reverse by advancing a set of 'principles' long before all the evidence is in. In a more constructive vein, the quantities of early MS. and epigraphic material now becoming available for study permit us to turn our attention from the TTS to a heuristic, empirical examination of precisely what LC practice meant to those who themselves composed and first copied the early documents. The gradual accumulation of LC data from these sources will, at some future date, either confirm our working hypothesis or point up its need for revision. There is, of course, another level at which Karlgren's LC procedure can be called in guestion. This is the validity of his OC reconstructions themselves. For nearly every step of the argumentation in his LC article, the point d'appui remains his own hypotheses concerning the phonetic value of the syllables of OC. In the 1940's when his reconstuction first appeared his theories had few contestants; still, on general principles, it should be obvious that if the reconstructions themselves can be shown to be faulty, then his system of LC criteria will be invalidated. Over the years certain inconsistencies in his 53 treatment of the XS series and. flaws in his methodology of reconstruction have become apparent. . In many matters of detail,, where the evidence is so often exiguous and/or equivocal, Karlgren's theories have evoked criticism and counter-evidence or at least, counter-interpretations of what evidence exists. A number of linguists have been motivated to fashion their own reconstructions of OC, witness the systems of Dong Tonghe, F.K. Li, Zhou Fagao and E.G. Pulleyblank. While this study cannot attempt an investigation of all the points of divergence among these reconstructions, reference will be made to them where relevant to the LC evidence discussed below. Hapax L eqomena Hapax legomena (HL) do not form a separate class of textual variant; they are rather graphic forms unattested in the extant lexica of the Chinese script.. In the present study, the term HL is used to refer to graphs not recorded in the largest lexicon of the modern kai shu script, the Zhongwen. dacidian. Dictionaries of graphic forms found in bronze inscriptions have not been scrupulously searched. Yan Lingfeng provides a convenient list of the HL in the two Laozi MSS. . (ST 41-44).. Most of these unattested graphs form a sub-class of LC since through their obvious XS affiliations they can be linked phonetically to the graphs in the TTS. AL 8a.7 ^ ]i$ • lM\ all TTS (KS 214, Ch. 76) ^EJ is merely an unattested member of the phonetic series 54 which uses *th as its speller, any of whose members can stand for any other. The standard graphic form of this syllable has A as signific. on the other hand, represents a syllable now written with a graph of a different XS series. This is still valid, however, since the . series spelled with ^ includes^, homophonous except for tone with OC /h$ . . 7a.*t*fr^$biL 6a. 7 ^ J "JI BL 6b. 1, all TTS (KS 190, Ch. 64) 7b.%$®r$ii^ AL 14a. 10 BL 14b.7, all TTS (KS 130, Ch. 37) In these two HLS =kz) has been used as phonetic in place of the speller ^ of the standard graphs found in BL and the TTS. The loan of ^ for 4tJ is phonetically acceptable since both words belong to the same rhyme group and differ only slightly in kind but not place of initial articulation.. This alternation between spellers is still found in the modern script where both tr^nj and e^b are accepted graphic forms of the word e meaning 'falsehood'.. Some problematic single graph variants 8. ^^1^ $ AL 9b. 12 i^lilfc • ^ ] BL 10a.10 The transmitted versions of this passage show three different readings: one identical to BL, and two slightly 55 expanded versions, viz. 3% 5& ^bij_^^_ and &>^i|k \(KS 70, Ch. 9). In his translation W.T. . Chan rejects the longer versions with their mention of 'fame accomplished' as incompatible with Daoist detachment from the quest for worldly success (196 3:115).. In any case, it is clear that the graphs iTJ^and.'&J in AL correspond in some way to 1^ (or possibly S^<y ) and in all other versions. What is the nature of the correspondence? Tj^l 'complete, accomplish' is a synonym of -i^ and thus should be treated as a semantic variant. The two belong to completely different rhyme groups (^f and^l^xespectively) and thus are phonetically too distant to be considered LC. The interpretation of-l^tl as LC for^l:, however, is quite acceptable despite Karlgren's objections that the correspondence is 'phonetically strained' or 'somewhat poor' (1967:#1551). These two words belonq to the hekou category of the rhyme which means that both syllables have a medial -w-followed by a mid-vowel and closed by a dental consonant.. Karlqren reconstructs them in Ancient Chinese as *d^'iuet and *zwi- respectively. (His Archaic reconstruction of these initials is not reliable.) The co-occurrence in XS series of the initials represented by dz'- and z- in Karlgren's system can be exemplified by the pairs >"f! /i^, /j^. and the double reading for ^ , •dz'isk and *zi-. The final of the yin rhyme of this group has been reconstructed in various ways, as *-d and *-r by Karlgren and Dong, as *-d by Li, as *-r by Zhou and as *-l >-j by Pulleyblank. The validity of these reconstructed values does 56 not concern us here, the point to note being that there are contacts within XS series between the and ru subdivisions of the rhyme (e.g.^/^t, §| and^/^i). Seen in the light of this evidence, it should be clear that is an acceptable LC forij^ . Furthermore, both Karlgren (1967:#1549) and Shirakawa (1974:2,146) note that on early Zhou bronzes 1^ often has the meaning of 'consequently, thereupon', indicating that this loan has a long history. These two words share the root meaning of 'follow a course' and may possibly be cognate. as a full verb means 'follow through, complete, accomplish,' but as a connective in narrative it introduces the last stage in a sequence of actions, •thereupon, consequently', ways functions as a full verb in OC usually with the sense of 'transmit, carry on' implying strict adherence to an earlier example or orthodoxy. This is one possible implication of Confucius' description of himself as a care-taker of Zhou tradition, a faithful transmitter of the way of the sage-kings, not an innovator o^tfn^fc (Lunyu 11.7.1) .s The sentence ^ ^ /L_ from the Zhonqyonq (SSJZS:5.885b) also exemplifies this basic meaning of i& : •(Wenwang's) father founded (the dynasty) and his son carried on'. The second LC equation in 9 above,^ used to writeis more complex. Karlqren reconstructs the two syllables *niwad and *t'wad showing discrepancy both in the type of initial and the height of the nuclear vowel.. Correspondence between A- and t'- is relatively rare but not unknown IH^/^t ^fi /^k and 57 "2[ phonetic inJJ^v 7) . As for the finals, several of the members of the phonetic series ^3 belong in their Guangjun and modern readings to a different Middle Chinese (MC) rhyme (fj|) from that of the speller T^l (|^) . Because none of these aberrant syllables (viz. tK| , ^ ,^F3) is found in rhyming position in the Sailing, phonologists have had little choice but to project the MC vocalic value back to OC and assign these words to the rhyme in spite of the fact that TAl with its mid-vowel belongs to the group.. This incongruence between the XS series and rhyme group has led Pulleyblank to hypothesize a secondary origin for the MC low vowel of Vi , i.e. a shift from an earlier mid-vowel shared with ^ to MC *-iajVl by the time of the Guanqyun (1979:25). Turning now to the paleographic evidence, we find some indirect confirmation of this hypothesis.. In the first place, i^itself is a clerical-script simplification in which the earlier graphic form has been distorted beyond recognition. (Vf^certainly has no phonetic connection with £c *k©n, in spite of the appearance of the modern graph.) The syllable we today write is found in SW written with the graph ^(Jt.and glossed&P 'decline; withdraw'. Xu Shen says it also means 'walk slowly' and then lists what he considers its three semantic components: % 'small steps', £| 'sun' and 'walk slowly'. Though Xu does not identify a phonetic element among these, Pulleyblank (1979:26) has proposed that Q may serve that function here. SW also records the variant graph and the form -il^L is preserved in the Yupian of the sixth 58 century. Assuming that functions as phonetic in both these forms, it is then easy to see how vg r a member of the same phonetic series, could have been used as a loan for V^. (Another example of this LC correspondence occurs at AL 7a.8.) Furthermore, the existence of these graphic variants and of the loan of ^3 f or i& in AL suggests that Karlgren's •fi'iwad n may be incorrect for ilL which probably still had a mid-vowel and rhymed with T^l in the second century B.C. . The reconstruction of OC is complicated by the fact that, on the one hand, it has as its phonetic a rusheng syllable ending in -p (X), while on the other, it is found rhyming with words of the group (Karlgren*s *-d) in the Shijing.. To account for this discrepancy between XS speller and rhyming behavior, Karlgren reconstructs with a final *-b (homorganic with the -p of X ) which, by the time of the composition of the Shijing had dissimilated to *-d permitting rhyming with $&sjfords. This hypothesis satisfies the demands of both XS construction and rhyming and has been adopted by Dong and F.K. Li. It is, however, highly speculative in terms of general phonetics. Pulleyblank has proposed a different solution to this problem. He regards the MC departing tone as the reflex of an earlier *-s suffix and thus reconstructs OC "fcl with a final *-ps. Assuming assimilation of -p to -s, he sketches the following phonetic development: *-ps >*-ts >*-s >MC . By the time of the Shiiinq, ^9 would have ended in *-ts permitting rhyming with qu sheng words of the group (*-ls in Pulleyblank* s reconstruction). 59 9.*fe$L{l,J&gl 13a.3 ] 7^ BL 13a. 12, KS 110, Ch. 28 |, a word meaning 'canopy*, occurs four additional times in AL (10b.12, 13b.6 and 14a. 11 twice) where BL and the TTS (KS 82, 120, 130, Ch. 15, 32, 37) consistently have i5^ or the homophonous. The remaining three instances of this word in BL and the TTS (Chapters 19, 28 and 57) occur in parts of AL which are defective. ^y§. , as its use of the phonetic element /3L suggests, has a velar initial while vf^. begins with a bilabial an alternation quite unusual within XS series, but not altogether unknown.. One example is found in the pair "^^/>^# *pian and *'wan in Karlgren's Ancient reconstruction. . Xu Shen's analysis of ^ , the speller in , also points to a possible connection between these two types of initials: "tr* *kiung is identified as the phonetic in ^ *b'uk. (These two syllables belong to the same rhyme group, and the alternation between final -k and -ng is attested within the phonetic series spelt with ~tt by the word %r *kiuk.) To account phonetically for this correspondence between initial velars and bilabials is not easy.. It may point to an early cluster incorporating both of these stops, but until more evidence is available, this explanation remains very speculative. In any case, the variant for , confined to AL, is highly idiosyncratic. 60 10. ^"^6 >f jl^^iJ_BL 2a. 9 ] "ft • ^4 Jli-a11 TTS (KS 136' ch--39) Despite differences in XS affiliation pointing to different pronunciations in OC, in Han times the two syllables written ^ and ^ became homophones so that 4* is often found in TTS as an LC for ^ • It would be anachronistic to assume that ^ was here a loan for-4fr , however, since the MWD MSS. appear to preserve the modal distinction between these two syllables. This suggests that they may still have been distinct in pronunciation as well.. Though commonly called a 'negative imperative', this description of -ffi is misleading, since, as the present example shows, the word can occur in a subordinate clause and have a third-person subject. Thus, the range of its usage exceeds the boundaries of what is normally understood as the imperative mood in Indo-European grammar (Lii 1955:24-35). A more accurate description, 'subjunctive negative', has been suggested by Pulleyblank (1979:23). The difference in finals between a|_ -n and ^ »t is not recognized by Karlgren in his stated principles as acceptable in LC practice, even though examples of this type of phonetic alternation can be found within XS series (5./4-S., °^0/^n I^A^I ^"/^ etc.). Further comfirmation of the validity of the loan in this example is provided by four rhyme words, all from the rusheng division of the ^ group, %%r1jfrlri£& andfjd.. Finally, another example of a syllable in final -n, writteni^_, loaned for a syllable in final -t (f/J) is found at BM 40b. 7 where the rhyme word is ^. . Although in the present 61 cases^ and »^ were clearly used to write a syllable ending in -t, this sort of alternation is quite uncommon and presents unsolved problems of phonetic interpretation. One might surmise that the two MSS..graphs actually had unattested readings in -t and proceed to look for further evidence of this phenomenon in dateable MS..material. Such evidence might allow us to assign it to a restricted time period or possibly a dialect area.6 11. 1% IK*. AL 8a.9 BL 8 b. 6 sw %j1r[l^'ki The alternation of ||&_ in AL for i> in BL occurs three times in this section of the text. appears thrice : again at BL 3b. 3 where AL is defective. All transmitted versions have the graph $\ in each of these six instances (KS 216, 156; Ch. . 77, 4 8) . is HL but obviously belongs to the phonetic series whose speller is || • ^ has readings in both the 7Li and ;>C__rhyme categories and its XS affiliates include syllables in either rhyme. X\ is homophonous with 1| in its rhyme reading. (These two graphs, :t\ and fj , form an LC pair elsewhere in the MWD MSS., AM 12a.4.) The difficulty comes in relating the MS. graphs to the^fl of the TTS._ Though SW clearly states that || is phonetic in^l , the disparity between initial *g- and s- precludes for Karlgren (GSR #435a) , Todo (1965:695) and Shirakawa (1974: 12.75) treating the variants \% as LC. Alternation between velar and sibilant initials within a 62 single XS series is certainly not unknown; F.K.Li provides eight examples of this sort of contact and incorporates the evidence •a into his reconstruction, proposing *gwjsn for g /:&. and *skwanx for (1975:241).. Yan's suggestion (ST 22) to construe 5, as LC for'lose', or, according to the SW definition, 'to have lost something', works phonetically, but seems to fit less well than-?! 'reduce1 into the same context with •superabundance' and^ 'supplement'. 12.^^'J&^ffAL (LZ 5) T§ - &^]^ • BL 4b.7 (cf . . KS 170 , Ch. . 55) though unattested, is no doubt a graphic variant for *lat found in the Tangyun. The Guanqya preserves the form ^ and glosses it 'scorpion*. This word may well be related to *liad 'stinging insect' as well as ^ *t'ad 'scorpion'. The MS. graph transliterated ATI. in LZ is, according to the accompanying note (LZ 15.22), equivalent t an archaic form of Ttlsl . These three graphs are members of XS series which rhyme in the category, and in addition, all three have velar initials; hence they are incontestable as LCs.. The interesting thing to note is that for a graph which Karlgren reconstructs • xiwar in AL we find in BL. The tradition that 7L*S and ^\ are merely graphic variants for the syllable *xiw©r has been disputed by Karlgren (GSS #1009a) who reconstructs i\ as • d'iong, phonetic in^ *d* iong, *di8ng and *d'ong. The existence of these phonetic compounds is persuasive evidence for 63 Karlgren's hypothesis; on the other hand, the alternation between $,Kand ^ here in the MWD MSS. would seem to confirm the traditional reading of *xiw3r for jfe. . . One possible way to resolve the conflicting evidence would be to recognize that 3&3v may not be a phonetic compound graph at all. Just as characters such as^,|| , oa , and-^*f do not include phonetic elements but are themselves primary graphs, so the pronunciation of jj^ need have no connection with the phonetic value of lk\ . . In the case of such obviously compound graphs as ifi^ and is indeed phonetic, but, as noted in SW (s.v, ^1^) , it is surely an abbreviated version of , and thus visually identical with iS\. In the evolution of the Chinese script the : gradual abridgement of phonetic elements is a common phenomenon. Many examples could be quoted in which the archaic phonetic has undergone such simplification as to become identical in the modern kaishu reflex with a completely unrelated graph: with its phonetic (not ij!- ), ^phonetic in 7^, ^| (unrelated to 13) phonetic in ^ , and ^ whose original speller was J$- , not The use of jj<k , abridged to , as phonetic in and ^fvlfe\ can thus be regarded as simply a further example of this well-attested phenomenon. This allows us to retain the traditional eguation betweenTL^J and , a correspondence corroborated by the graphic variation between AL and BL. Finally, it should be noted that the alternation between ^ and in the passage above exemplifies the coexistence of palatal and velar initials within a single XS series, a phonetic correspondence often ignored by Karlgren and others. _ 64 13. ^l§^g BL 7a.2 ^LJ$jj3 AL 6b.7 ^ is HL; however, its phonetic value can be reconstructed through XS connections and its function here as an LC. It occurs again at BL 11 a. 9 where all the TTS showf!^ (KS 82, Ch. 15). (AL is defective at ths point.) The graph belongs to the phonetic series spelt with X. most of whose members have initial m-. One member of the series, the word written , however, is found with initial 1- in the Guangyun and modern dialects, necessitating the reconstruction of an initial cluster in OC, *ml-. The present LC equation confirms this reconstruction since the syllable which ->t_ writes also has an 1-initial. As for the finals, although and^p belong to two distinct rhyme groups in terms of Shijing rhyminq (^LandJ.), the two had coalesced completely by Han times, making this sort of loan not only possible but quite common (Luo and Zhou 1958:36). The following example provides further evidence that the breakdown in distinction between the and jjjL groups had already occurred by the the time the MWD MSS. were copied. • 14. ivJfi/Ly- KS 188 (YZ) (Ch. 64) >fa ] 4— AL 6a. 4, -^j- BL 6a. 10 In view of the merger of the two rhyme groups "5C_ and by the time of the MSS., the use in AL of i~ to write 4^ seems unobjectionable. Relating the initial affricate of -4" iu BL to the initial nasal of kft is more problematic. SW provides some 65 suggestive etymologies. GSH #365a + *ts*ien i5-t!L**- t ^ ^ * GSR #555a ^ *ts*i©r XA K — ^ GSR #564a ~ *niar GSR #364a 4f *nien f£ Ik ^ AA ^ 4"/|f Xu Shen's graphic analyses of , and 4j- , if valid, point to the co-occurrence within single:XS series of the OC homorganic initials *n/- and *ts'- (Pulleyblank 1962: 133). Though this phonetic alternation is, no doubt, a valid one, in the interests of accuracy, let us reexamine Xu Shen's graphic analyses in the light of some pre-Han script forms. Taking I/x7 first, Xu's treatment of as phonetic appears sound and even suggests a cognate relationship between the two words since 'come second, be inferior* and 'order, sequence* are among the common meanings of ^fc. Shirakawa cautions against this etymology for the graph, however, by demonstrating that many of its earliest uses can be related to the meaning of !£. 'yawn* (1974:8.200-203).. He is not clear on how the two dots (sometimes three in oracle bone forms) to the left of %L ought to be understood, but he apparently thinks that the use of this graph to write a word meaning 'secondary; sequence' and other unrelated words represents early loan practice. Though this may be a valid observation, it does not of itself refute the proposed cognate relationship between the words *ts'i^r and *nia r; it merely sugests that the analysis of the graph into and !Z may be incorrect on etymological grounds. Xu Shen's treatment of 4~ as phonetic in ^ is motivated by 66 the small-seal form he quotes as headword i" in kaishu transliteration. . Shirakawa notes that the graph ^ first appears on late Zhou bronzes; earlier forms, including the many occurrences in the oracle bone inscriptions, consistently show as the lower element (1974: 7. 93). Though he prefers to explain this primitive graph as a picture of a grain-carrying celebrant in a harvest rite, the A here may well function as phonetic. The fact that a graph with speller A came by the late zhou to be written with phonetic 4" could be interpreted as further evidence of the affinity between the initials of the two words represented by these graphs, a relationship reiterated in A A-the analysis of T itself. Finally, although^ , adduced by Xu as an 'older graphic form' (£>C_ ) of , is not attested in the pre--Han epigraphic corpus, there remains the possibility that -4" here . functions a s phonetic in a graph homophonous with . Beassessing the evidence in this way with a view to greater historical accuracy has not significantly changed the nature of the case; on balance, the paleographic data supports the acceptance of \ as LC for 15. A^^ti^^AL 7a. 4 i$_]4cK-all TTS (KS 197, Ch. 67) ^>yg-fra>/vBL 7a. 12, H J*. HFZ 379, >>^ all TTS iS. ] i%j HFZ, TTS. LZ (17.45) explains the MS. as a copyist's error for a variant form of ; in fact, £^ ('establish') is 67 unobjectionable semantically and, furthermore, it rhymes with ;£§, the MS. variant in the second clause of this passage. For this reason it seems preferable to regard the MS. reading as independent of the TTS rather than proposing a loan relationship bet ween ±1 a nd ^fldr.9 At this point we must digress from the consideration of LC to discuss a syntactic problem raised by this example. The MS. versions of this passage exhibit what appears prima facie to be an inversion of the two clauses of a condition, the protasis (^T/^D X^£\±-£-) following the apodosis. This sort of inversion is virtually unattested in the TTS of the Zhou classics, except in a small number of cases where the sentence has exclamatory force as in Mengjzi 4. IA. 7 and 19.3A.3. The present example from the Laozi MSS., however, is a declarative statement; so to construe it as •Heaven intends to establish someone [as sovereign], if it walls him around with compassion* violates the known rules of OC syntax. In reading the transmitted versions of this sentence, the problem never arises since the troublesome words /jfl/Ja have in every case been deleted: 'What heaven succours it protects with the gift of compassion* (Lau 1963:129). Unless we propose on the basis of this example in the MSS. and HFZ to abandon the syntactic rules for conditional sentences in Chinese, the solution to this problem lies in rejecting the prima facie interpretation of -fa and % as words meaning •if* and proposing other semantic values for these graphs. In his Jjngzhuan shice , a compendium of particle usage in the Zhou and early Han corpus, Wang Yinzhi 68 lists examples of ^ used like 73 in texts chronologically not far removed from the Laozi, such as the Guanzi and Mengzi (1798:7.23-24). This usage is also found in one of the other MWD texts, the Shidaiing (BM 24b. 1, 27b.2,9, 30b.6). Wang Yinzhi quotes a passage from Laozi 13 in which he glosses the graph T£ as §iJ . . This passage appears in AL as 16. -^t4^4^-T "£^*5L AL 10b. 3. •Hence, he who would rather be in control of himself than control the world can (then j£ ) be entrusted with the world.' (BL 10b.12 varies insignificantly.) Both the FY and HSG texts show gij for ^ here (KS 78). This solution is applicable to the occurrence of ^ in the HFZ reflex of Laozi 67 quoted above, but what of \ in the MSS.? If we return to Chapter 13, we find another example of this curious use of He in the sentence parallel to and followinq 16: 17. ^>^|-^^T ;*3f£j.AL 10b. 4, (BL 11a. 1 has final ^ .) 'He who loves to treat himself as the whole world can be given custody of the empire' (cf. Lau 1963:69). In the transmitted versions of this sentence one finds in place of He the graphs (XE, WB), @»J (FY) or ^ (HSG).. The ^ , of course, should be understood as 7b as in Wang Yinzhi's examples. But is there any possibility of construing He as a connective particle here? Wang quotes examples of -jfp used to mean 75 and |1>J (1798: 7. 12-13) and perhaps the neatest solution would be to treat these examples of Ht in the MSS. as loans for-^u in this sense. There remains another possibility, however, suggested by the editors of LZ. They regard He here and at AL 11 a.1, BL 69 11a.11 (KS 82, Ch. 15) as a graphic error for ^ (LZ 63.14). ^ is occasionally found even in TTS as a connective with the sense of 73 or 75* (Wang 1798:2.11-12). In the Laozi MSS. this use occurs four times in Chapter 17 (AL 11a. 8, BL 11b.6) and 18 (AL 11a.11, BL 11b.8) where some of the TTS retain the synonymous 3| (KS 86, 88).. Thus, if ~\ is accepted as a plausible scribal error for ^ , the difficult MS. readings in Chapters 13 and 67 can be resolved in a manner which accords with the stylistic features of this version of the text (viz. the. use of ^ as an initial conjunction). Pulleyblank has further suggested that this may be phonetic in ^ and hence an orthodox LC for that word. If the solutions either by assuming scribal error or an LC connection between ~^ and ^ seem too speculative, here can be understood as LC for "^Q meaning 73 or |'J as proposed above. In either case, a reading is produced which does not violate the rules of OC syntax. to.&%&t&*W\W*BL 13a.2 AL 12b. 5 iHir ]^ llfL most TTS (KS 106, Ch. 26) <g ] all TTS The interpretation of this brief passage has exercised the ingenuity of many commentators and translators.. It is preceded by the statement that 'the gentleman when travelling all day never lets the heavily laden [baggage] carts out of his sight' (Lau 1 963:83). ^ ^jin the next line has been variously 70 translated as 'magnificent view' (Waley 1958: 176 and Chan 1963: 146) and as 'camp and watch-towers' (Duyvendak 1954 :67, reading ^ for^) while the editors of LZ read the MS. graphs 3-^TB7as ^ /|^g' and paraphrase 'rest houses for travellers' (33.44). One of the reasons for these widely divergent interpretations is to be found in the use of the concessive 'although' in BL and all the TTS. The context is not well defined here and none of the readings makes for a smooth and obvious extension of the meaning of the previous sentence. Ten years before the MWD MSS. were unearthed D.C. . Lau proposed understanding^; as1^ 'it is only.,,' (1963:189). This seems to produce the least difficult reading and is confirmed by AL: •Only when there are enclosures and look-out towers does he dwell at ease and hence is detached'. This is the most satisfactory interpretation because it reiterates from the previous sentence the gravity of the gentleman's concern in arriving without loss or mishap.. The variation between S?L in the two MSS. and^ in all the TTS allows two interpretations.t At first sight, an LC relationship appears unlikely since J.^c.belongs to the Tu rhyme and ^ to the^t rhyme. Furthermore, sense can be made of the MS. version as it stands by reading 1%L § as 'enclosures and watch-towers' ("g7 LC for ) or possibly 'encircling watch-towers'. This interpretation would mean that the graph ^ (for ^ meaning 'encampment') in the TTS is a semantic variant rather than an LC forJJ<_or vice-versa. It is, however, possible to account phonetically for the presence of /f? in the transmitted 71 •2? ± versions. The speller in the graphite is an abbreviated ^ (GSR #256), a TLA rhyme word as are nearly all of the members of its phonetic series. The graphs and'x^c, however, belong with r^> to the category as shown by their rhyming behavior in the "Ji. ^ Shijinq (discussed in Karlgren 1964:#299) . . •Jft which shares the phonetic *-~> with ^ is a graphic variant of ^ ; thus, it is possible to construe iKas an LC f or'JhT . In the present state •on 20 of our knowledge of OC rhyming, and J^are. distinct and do not inter-rhyme. The fact that they share the same phonetic is puzzling since members of the same XS series generally all fall within the same rhyme group. One other point worthy of mention in connection with 18 is the use of \7a as an adverbial suffix. . Though examples of this usage can be found in literature as late as Eastern Han, this function of ^ , dating back to the Odes, seems gradually to have been taken over by the particle f£ as in the TTS. 19.§&l*f AL 8b.12 ^ J^/j-BL 9a. 10, all TTS (KS 56, Ch. 2) It is, of course, possible to understand ^ in AL as an orthodox LC for itb ; if, however, it is construed as a loan for the of BL and the TTS, then it calls into question Karlgren's reconstruction of *-iar for the final of phonetic series^tJ , GSR #358. The series has traditionally been assigned to the OC ;>L rhyme whose reconstructed value consists of mid-vowel and a velar final. The rationale for Karlgren's splitting of the OC 72 rhyme and reconstructing *-iar for series Jtb is to be found in a few anaomalous rhymes in the Odes, one of which includes a graph with phonetic (1954: 304) . 1 ° The words collected here belong to OC rhyme groups 4&Sc_(4a) • % , and (fvf) whose established values Karlgren is then required to homogenize in order to produce a reconstruction which will preserve the rhyming. The question to be asked here is how heavily these cases of hedge rhyme ought to be weighed in phonetic reconstruction. There certainly exists good evidence against treating series 3ttJ separately from the rest of the OC rhyme group. In the first place, fc<M. rhymes with three other words of the \_ group in Shiiinq 46.197.5. Secondly, the OC demonstratives itb and *Sxeg are no doubt cognate and, unless there is reason to believe their pronunciations diverged very early, reconstruction ought to confirm this etymological link. It seems clear that in positing OC *-iar for the entire n phonetic series on the basis of a putative rhyme between , V^and.^f- • Karlgren has allowed the exceptional hedge rhyme an importance out of proportion with the bulk of the evidence. F.K. Li's more conservative reconstruction of *tsjig for-^Ir and *sjig for agrees more closely with the rhyming evidence reviewed above.. It should be noted, however, that the alternation between s- and ts- within the same XS series is not common and that § as LC for is on phonetic grounds less likely than ^ as LC for , a member of its own series. a 73 20. .J2.^7r5j ftJEI^Si ^} &Ll0b.6 1b+ J~to a11 TTS <KS 80' Ch. .14) 3] BL 11a.3r T^i all TTS •These three things are incalculable/cannot be scrutinized; thus they meld into a unity. ' The MS. reading with %~\ is semantically quite acceptable, although the transmitted version is perhaps to be preferred because so makes a better rhyme with — . Rhyming between the <ju and rjl categories of the £|g group is not unknown in the Odes (43.191.5, for example) is attested throughout the Han and as late as the sixth century (Pulleyblank 1 973:371). Its occurrence here between \\ and — suggests that the sibilant final of was still audible in early Han times. One could even go a step further and speculate that the revision in later texts of st to %u might represent an effort to retain the rhyme after the final *-s of the former syllable had disappeared. (fl and are both HL. Yan (ST 42) reads © as IS which permits interpretation as a loan graph for the homophone commonly written yH? or 5txias in the TTS. LZ (31.23) links ® to ^found in SW meaning 'carry-all' and pronounced *g'wan. 's phonetic % places it in the same rhyme group (-5^) with the above words with an initial of the same type. LZ (63.15) reads this unattested graph as a loan for 'cord' but there is no need to take this literally as the graph is, of course, an acceptable substitute for ytL. . 74 21.&&Hr4ceU4il& AL 1 1b. 1 1§ J 7£ BL 11b. 10, all TTS (KS 90, Ch, .19) Here again there are two ways of dealing with the variant in AL. If we wish to bring the variant *g into line with the of all the other versions, we must posit an LC relationship between the two words. has a number of readings all of which are found together with in the ^ rhyme. One of these readings (t^ij&JO differs from the phonetic value of 73- {v%-%J%$ only slightly in the that it has a medial glide -i-; this sort n of alternation, however, is characteristic of some phonetic series in this rhyme group; "s^t *t'og with its phonetic *tiog (abbreviated to ) and^^ *d'iog/7J5j *d' og are two examples. As an LC for 7J. then, ^ is quite acceptable. There is a possibility that these two words are related etymologically. The connection between filial piety and fostering, caring for one's aging parents is implicit in the structure of the graph 7j- composed of an abbreviated character •^•aged' and the graph for 'offspring* -J- . . This connection is later explicitly drawn in Confucian ethical writings like the Liji (Tj ||-tfjj, SS JZS : 5. 830b) . One of the common meanings of is 'to rear, foster*, though in the extant Zhou corpus my impression is that this meaning of the word is applied most frequently to the raising of domestic animals. In the language of the: bronze inscriptions, however, % is often found in a sacrificial context suggesting that the word's earliest associations were with the nurture of the ancestors' spirits by means of offerings (Shirakawa 1974:8.129).. The syllables 75 written ^ and f§ are thus not only close enough phonologically to be loaned for each other, they appear to be related etymologically as well. If there is any value in applying standards of consistency to the sayings of Laozi, then our LC solution to the. problem raised by the reading in AL must seem rather dubious.. In the passage just preceding the sentence quoted in this example, Laozi inveighs against 'filial compassion' (^^.) as the result of discord amonq family members: A ?r~ fc\ ^ (BL 11b.8).. To claim that the people will again be filial (as if that were their original, uncorrupted state), if the ruler •exterminates benevolence and discards rectitude' (Lau 1963:75) plainly contradicts the disparagement of filiality in the previous passage. A way out of this dilemma is provided by reading AL as it stands: 'Exterminate benevolence, discard rectitude and the people will again foster compassion', i.e. simple, basic sympathies will find their natural expression when people are no longer constrained to act out the imposed social virtues of ffi (and 7^. too, no doubt, if this interpretation is correct). . This reading would mean treating 73- in BL and all the TTS as LC for the j| of AL rather than the reverse. 22. .7JL!)§^3&AL 14a. 8.. 7^]f^BL 14b. 6, all TTS (KS 128, Ch. 36) The graph in AL transliterated ?X_ actually looks something like 32 . (ST 42 transliterates it .) Earlier forms of the 76 modern graph 7X. consist of a pair of X. , appearing in the oracle bone script as ^ and asfl on the bronzes. Admittedly, none of these bears a very close resemblance to the MS..graph here, but LZ (35. 64) relates this form to the cjuwen graphs found in SW under fi , viz.^ and^j. The note goes on to say that this 7*_*giug should be regarded as LC for the ^ *niog of BL and the TTS, but at first sight this appears unlikely.. The two words belong to different Shijing rhymes, ^ and^il , and the disparity between the two initials, velar and nasal, seems irreconcilable.. On the other hand, if on these:phonetic grounds we reject 7iL as LC for and read the line as it stands ('The amicable and weak overcome the strong'), we have then introduced into the Laozi a new concept 7^, a word used nowhere else in the entire text.. The compound ^11 , however, occurs repeatedly, three more times in AL (8a.6,7 and 8) and again at the beginning of Chapter 78 (KS 218) in the TTS where the MSS. are defective. Thus, in the context of the entire Laozi, the evidence would seem to urge accepting i%_ as LC for ^ , in spite of our assumptions about what should constitute an orthodox LC.. On the other hand, it is certainly true that the equation of the MS. graph with 7R_ may be incorrect, though I am unable to suggest an alternative interpretation at this time. At the present state of our knowledge this problem admits no easy solution. 77 23. %p\k.-ft^M^h AL (LZ 23) Lva* BL 11b*8f most TTS (KS 88» Ch..18) While is unarguably a member of the ^ rhyme, the precise status of St and its XS derivatives is a matter of controversy. Neither ^ nor ,3 is used as a rhyming word in the Shiiinq. , however, rhymes with words of both the (/£. and *§<9, 55.222.2) and ^i^groups ( , 46.197.4).. Inter-rhyming between these two groups is already attested in the Shi-jinq. but the status of another of the rhyming words in Ode 197.4 (y$-) is also in dispute. Dong places it in the group while Karlgren reconstructs it as a member of the rhyme (•p'iad). Thus, the bulk of the early Zhou evidence, scanty as n it is, would seem to argue for the reconstruction of ^ with the nuclear mid-vowel -9- characteristic of both the H% and ^jjrjrhyme groups. . The data from Han dynasty rhyming is not plentiful either, but some of the evidence cited by Luo and Zhou (1958:169, 171, 255) suggests that 41. and its derivatives, all of them qusheng words, may have shifted in Han times to the group.. From the early Han there is an example of rhyming within and , both rushenq syllables, the former of the £ff category, the latter of the fla category. (As mentioned previously, rhyming contacts between gu and rushenq. syllables of the same rhyme are rather common, cf. Pulleyblank's reconstruction of final *-ats >-as for the rhyme and *at for its rushenq analogs.) J£ and its derivatives apparently still rhymed in both the and categories in the Later Han.. Luo and Zhou cite an example in which /^fojfords (these two 78 categories had merged completely by this time) inter-rhyme with $S and a rushenq word from the group. Finally, there is another example from the.Later Han which shows rhyming between and*^, the latter a member of the ^ group. The occurrence of 'fA as LC for ^ in AL would seem to support the hypothesis that ^ changed rhyme categories from^^lS^to ^ , though it would be difficult on the basis of the present evidence to date this change. The same apparent shift from an earlier nuclear mid-vowel C^CJ-hyme) to "the lower -a- (^ rhyme) was cited above as a possible explanation for the discrepancy in MC rhymes between the members of the XS series with the phonetic ^3 (see discussion of example 8).. 24.^i^ip^ AL 8a.9, BL 8b.6 ip] #0 most TTS (KS 216, Ch. 31) The MS. graph i.J) looks remarkably like the modern ip MC *ian- 'seal' and is glossed by the editors of both BS and LZ (12, 47) as 4^ **i^k 'press down'. (Cf. the synonym variants in two of the TTS: ^ in YZ and in one XE.) In fact, what is now written ip is found in the small-seal script as ^ while *'iak was composed of the same two graphic elements aligned horizontally rather than vertically,^ . Xu Shen describes the graph as 'turned around' and notes that ^P, a vulgar form, is accompanied by the hand radical.. The MS. graph here is simply the SH form transposed, as Yan points out (ST 24). What we 79 today write should properly be constituted, as found on the Xixiaspng (i§t 2&'A§}) , a memorial stele dated 171 (reproduced in Alcai 1974:648). The modern kaishu form with as its •phonetic' is a patent misspelling since *'i«k has no phonetic or semantic connection with the XS series Karlgren reconstructs •ngang. The likelihood of a cognate relationship between and ip is much greater. To begin with the phonological side of the problem, Karlgren's reconstruction of *'isk for OC is open to guestion. To be sure, the word is found in the MC *-i3k (^%) rhyme in the Guanqyun; however, in the Shijing it rhymes exclusively with words in final -t (viz./pyk andE5 ). Dong Tonghe treats this as compelling evidence and reconstructs a final *-t for ^p in OC. (gp, which shares the phonetic P with ^p, is another word that has shifted from a Shijing rhyme in -t to a MC rhyme in -k.) Pulleyblank (1960:61-65) has collected eleven examples of this type of final alternation and concluded on the basis of Tibetan cognates that the velar final may have been the original form but that the dental final had by the time of the Shijing already developed in a prestigious dialect through the assimilation of -k to the high front vowel characteristic of this rhyme. More recently (1980:11), he has advanced a different solution to this problem by proposing the reconstruction of palatal finals for the ]£ rhyme group (i.e. *-c for the ^ category, the rushenq analog of fL ) which yielded dental stops (*- c >- t) in most cases but occasionally velars (*- c >- k). F.K. Li hypothesizes that this unusual phonetic 80 development cam be explained by dialect divergence, a process whereby -t after -i- changed to -k or, in some dialects, the reverse assimilation of -k to -t when -k followed -i-(1975:273). These examples of diachronic correspondence between -t and -k are analogous to the development of -ng readings for syllables rhyming in -n in the Shi jing (e. q. ^ , 'oft) . . Precisely how these changes came about has yet to be agreed upon.. Once we accept, on the basis of the Shi-jing rhymes, final -t for OC then it remains to be shown that alternation between -t and -n is well attested among cognate words. There are examples available in the very rhyme group (j=L/1f) to which in and belong: 'real, substantial' and ^ 'solid, substantial'. Among other likely cognates showing this type of final alternation flfl- 'tough, sinewy','speak haltingly'/|^ 'glue', ^J/?u , ifc/fifli and ^'/^ might be cited. The basic meanings of ip and ^fl are surely close enough to suggest cognation. . 25. fr£ it !k AL (LZ 21) Ml^-^BL 10b.3, i^/^ff_all TTS (KS 74, Ch. .11) %iC and belong to the same rhyme group so the only point of phonetic divergence between the variants is in the initials, Karlgren's *n- and *s-. There are a few XS series in which these initials co-occur; the series spelt with illi , for example, includes ^ *niat and %• *siad. Other instances have motivated the reconstruction of an initial cluster *sn- (Li 1979: 24 1). With homorganic initials and identical finals, there 81 can surely be no objection to reading fe-/i^'as LC for^^. . 26. A±*^t,m t^^fito AL 8a.6 'li^jlffe H BL (LZ 47) None of the TTS have the graphs /qfik and 4% /{% or anything corresponding to them (KS 214, Ch. 76).. The graph used as a gloss for ^3L in LZ,ii§-, (18.58) is recorded in SW together with its quwen form It .. This latter graph means 'extend across, spread out*. Both and k\ are acceptable LC for ^ which means 'stretch, extend'. , (For an example of a syllable in i^-loaned for a word in k- see Ex. 25.) Thus the MS. passage might be rendered 'Man at birth is soft and weak; at death, stretched out, rigid and stiff. 27. #± a5j 7&BL 11a. 11 $i J all TTS (KS 82, Ch. .15) The HL transliterated H±. in both BS and LZ appears on close examination of the MS. photographs to bear a marked resemblance to the clerical script form of -JtPL . For one thing, the top horizontal line is unbroken and lacks the short vertical strokes usually found in radical 140 as written in BL, viz. tyJ or . Furthermore, the wavy horizontal lines of radical 90,^ in the MS., are not in evidence. Though it is impossible merely on the basis of photographs to deliver a verdict on this matter, an LC correspondence between *g'iwang and *k'wang is much more 82 plausible than one between $± *tsiang (?) and *k'wang.„ If the MS.. graph does, in fact, have the phonetic fi , then its affricate initial, incompatible with the labio-velar k'w-, will reguire a new interpretation for the passage. One possible solution might be to treaty* as LC f or }0j *dz'ang 'concealed', an -appropriate epithet for a valley. . 28 . yt 7u £ ^ AL (LZ 4) ]4^1 most TTS (KS 160, Ch. 50) Yan treats 7r% as LC for 4^ (ST 47). At first sight this seems unlikely since members of the XS series with phonetic % fall into either the ^"K. or the TU rhyme while and £ji (the variant found in three of the XE texts) both belong to the {%_rhyme. However, there is some evidence from other XS series which seems to point to an affinity between the two groups % and . The word "7\SL , for example, is a member of the T^rhyme, but its phonetic IL belongs to the -ik group. The graph writes a syllable belonging to the rhyme but appears to have fu as its phonetic. Furthermore, one encounters use of the phonetic fn of the ^ group to write words belonging to the 7L/rhyme (e.g.^H , ifn and^f?) while ^ itself and >\% both have readings in the TLA rhyme in addition to their 4% -rhyme pronunciations. . Karlgren (GSE #134, 238) treats these double readings as having arisen from a confusion of the two phonetics H t^) and |D (fu) • . This would explain the phenomenon adeguately if it were not for the semantic connection which also 83 obtains among these words: 3^ , , (all ^ ) and ^ , and $k (all 7u ) share a semantic core 'pliant, weak' and so would appear to be related more intimately than a simple graphic confusion would suggest. One could further adduce the double readings for the character (GSE #131p) where there is no possibility of graphic confusion.. This sort of evidence would suggest that the word Mn from the %J rhyme may at one time have had a reading in the rhyme. It is also possible to explain this variant as semantic rather than as an LC. The editors of LZ gloss the MS..graph as , but the usual meanings of the latter, 'measure, estimate; move', do not seem to fit the context here,. A more satisfactory interpretation of this graph is suggested by the word recorded in the Fangyen of Yang Xiong (53-18) glossed as 'drill' and pronounced likeiffii, according to Guo Pu's commentary (Zhou and Wu 1956:55. 5). Unfortunately, Yang Xiong does not identify the geographical origin of this dialect word, so we do not know if it was current in the Changsha region. Semantically, however, the meaning 'drill, bore' fits the present context even better than ^JL: 'the rhinoceros will have no place to bore his horn'.. The word written IBfri is no doubt related to 'point, tip'.. 29. i?in ^&Bt BL 5a. 10 i\-t»£]44" all TTS (KS 176, Ch. 58) The use of ^"^L in BL f or in the transmitted versions is 84 best not considered an LC at all. Although alternation beween the. stopped finals of rushenq syllables and gusheng words has been noted before both in LC pairs and within single XS series, the difference in vowel, a/fe , between these two words presents a serious difficulty. It is precisely this alternation between high and low nuclear vowels which suggests treating these words as synonym variants rather than LC. In several published expositions of his theory of the close/open ablaut in Sino-Tibetan, Pulleyblank has provided numerous examples of this alternation which he regards as distinguishing very broadly •extrovert' from 'introvert' forms (1963:220-221, 1965 and 1973a). Such a semantic or grammatical distinction is not readily apparent in the present example, but the hypothesis does allow us to relate these two syllables by means of a well-attested phonetic correspondence. Readingas the homophonous the . common semantic core of these ablaut syllables seems to be- 'spread (out), extend outward'. (The connotation of is •leak' or 'ooze out', whereas -i^f implies 'set out, display'.) Perhaps the meaning of the text here is the one suggested by the WB and HSG commentaries: '(the sage) is straight but does not extend outward' (cf, Chan 1963:203), i.e. he keeps his rectitude to himself without displaying it or extending it as an example toward others. 85 This section began with a preliminary analysis of single-graph variation between the two MSS. or between the MSS. and TTS. We proposed and illustrated a general classificatory scheme by which every instance of substitution variation can be identified as belonging to one of five types: taboo character, semantic variant, miswriting, simplification or LC. These five categories seem to exhaust all the possibilities of substitution variation in the MWD Laozi and will perhaps prove generally applicable to other OC epigraphic and MS. remains as well. The reason for cataloging types of variation in this manner is not purely descriptive, however. As will become apparent in the next section, the process of collating texts and then interpreting the results of collation requires that a strict distinction be maintained between significant and insignificant variation. The delineation of these five classes of variant greatly facilitates general discussion of their relative significance prior to collation and thus expedites the process of collation itself. It was with this purpose in mind that the classification was undertaken. The remainder of this section was devoted to a discussion of the category LC and an empirical examination of twenty-odd examples of problematic single-graph variants, most of them LC, from the MSS. As Karlgren has attempted to deal systematically with the phenomenon of LC, his approach was scrutinized. In the following heuristic treatment of individual variants, we tested his working hypothesis which predicts that the phonetic disparity between any two LC will never exceed the degree of 86 phonetic divergence found within a single XS series. The validity of this hypothesis depends, of course, on how the XS series are defined. The evidence analyzed here would suggest that Karlgren's hypothesis can be retained as valid, if, at the same time, we reject his own arbitrary treatment of certain XS series in GSB. Some of the LC data . was found to be highly anomalous and to permit at present only very tentative phonetic interpretation. The object of this part of the essay, however, has not been so much to correct Karlgren and settle difficult issues as to give a clear and balanced account of the linguistic evidence which can be gleaned from the study of some of the substitution variants in the Laozi MSS. Having completed this part of the investigation, we can now proceed to the different task of collation which will require that we not merely describe variants but that we form a judgement as to their relative significance for the tracing of textual history. 87 N°te§^ to_.Part Ong 1. The Chinese team published their first transliteration of the Laozi MSS. in Wenwu 74.11.8-20. This, of course, has now been superseded by subsequent revisions. There also exists a second edition of BS, in simplified kaishu, published in Beijing in 1975. I have been able to consult this transliteration briefly and it appears to have been the basis for LZ, published a year later., The rationale for citing the earlier edition of BS when it can be assumed that LZ represents a better understanding of the MSS. lies in the fact that LZ is printed in simplified characters whereas the earlier BS uses traditional kaishu. As noted above, whenever these two sources differ, LZ is cited as more authoritative, its simplified graphs retransliterated into the traditional forMS. 2. One characteristic of these two commentaries on the MSS. is thsir almost exclusive preoccupation with paleography.. One can only admire the command the editors demonstrate of the sources and techniques of wenzixue ( >C ) , but their analyses invariably fail to explore the phonological implications of the problems raised by the graphic variation in the MWD material. Since any significant problem in Chinese paleography is also a problem in historical phonology, this study attempts to redress the imbalance 88 found in the notes in LZ (and ST) by treating more fully the phonological side of these questions. . 3. LZ 20 shows rhyme between , 7^ ,/f= . m LZ 28 and AM 24b.4 the rhyme is between >$jf and A.. 4. A.E. Housman makes this point with his usual incisiveness (1921 :68-69). ...textual criticism is not a branch of mathematics, nor indeed an exact science at all. It deals with a matter not rigid and constant, like lines and numbers, but fluid and variable; namely the frailties and aberrations of the human mind, and of its insubordinate servants, the human fingers.. It therefore. is not susceptible of hard-and-fast rules. It would be much easier if it were; and that is why people try to pretend that it is, or at least behave as if they thought so. Of course you can have hard-and-fast rules if you like, but then you will have false rules, and they will lead you wrong; because their simplicity will render them inapplicable to problems which are not simple, but complicated by the play of personality. A textual critic engaged upon his business is not at all like Newton investigating the motions of the planets: he is much more like a dog hunting for fleas. If a dog hinted for fleas on mathematical principles, basing his researches on statistics of area and population, he would never catch a flea except by accident. They require to be treated as individuals; and every problem which presents itself to the textual critic must be regarded as possibly unique. . In debating whether or not to emend a problematic MS._ reading, the rule of thumb enunciated by Eugene Vinaver provides some guidance: ' is right to preserve a reading as long as it is possible that it comes from the original, and it is wrong to replace it by what is merely probable.' (1939:369). In example 5 above, the SW entry for 89 the reading in AL possible while the rhyming renders the emendation probable. . 5. This phrase appears in Mozi as ^ "^J 'M^ 'follow without innovating' (63.39.19).. Compare Xu Shen's definition of iTJ^as^lf a word meaning 'follow a preset course' and related to words like-L^,iil andNfi. 6. In the case of a language like Latin for which a large body of MS. material has been closely studied, it is possible for th=: medieval period to identify scribes' vernacular languages and to trace diachronic changes in their pronunciation through mistranscriptions of the phonetic values of the original Latin. Since data of this kind can be assembled only from a large range of dateable MSS. all emanating from a single geographical area, it would be premature to hypothesize about dialect distictions on the basis of the MWD Laozi texts alone.. However, it is not unreasonable to expect that close analysis of LC practice in the many MSS. now being recovered in China will, at some future date, provide us with good evidence of OC dialect phonology. 7. The XS relationship between these two graphs is not recognized by Karlgren despite Xu Shen's graphic analysis of 1| as phonetic in 5f'^ (AA^^^p. >)$i makes evidence 90 8. I quote Xu Jie's redaction of SWr SW xizhuan (%% %^4A% ) • In Xu Xuan's slightly earlier version A is not treated as phonetic but as a second sematic element in the graph 4 (Ding 1960: 952b). 9. There is a strong likelihood that ±^ and /ff-r are, in fact, cognate words (Pulleyblank 1973a: 121). 'ring-wall' is no doubt closely related to words like 'garden wall*, 'courtyard wall*,^sL 'ring-shaped jade insignium', rt^'ring', etc. A^y means 'protect by circular patrol' as suggested by the early graphs which depict a crossroads with feet pointing in opposite directions circulating around a central object interpreted by Shirakawa as a settlement of some kind (1974: 5.138). Though iH and *giwad belong to different rhyme groups, Tu and /P. respectively, the former can be regarded as the 'nasalized counterpart' of the latter (F.K. Li 1975:266) since they share the same initial consonant and nuclear vowel.. The precise reconstruction of the final of the group is a point of controversy among linguists, but all agree it was an alveolar sound and thus homorganic with the final -n of the 7u group. 10. Karlgren follows Duan Yucai (1930: 18. 82) in treating%xti as a rhyme word in Shijinq 45. 195. 2. Whether in fact "sit] rhymes with %<. , .^JL^etc. is open to question. Neither Jiang Yougao (1966:2. 2.20a) nor Lu Zhiwei (1948:76) concurs with Duan. 91 Part Two •Collation' is, literally, the 'bringing together' of textual witnesses for comparison. This stage of the critical process is ordinarily a preliminary to recension, 'the reconstruction of tie earliest form or forms of a text which can be inferred from the surviving evidence.'* In the present study, however, since there is no question of attempting the complex task of reconstructing an archetype, collation will be used as a method of describing a textual witness through contrast with other witnesses. . The lemma in this collation will be the acknowledged quotations from the Laozi found chiefly in the 'Jielao' (JL) and 'Yulao' (YL) chapters of the HFZ. These citations will be collated with the MWD MSS. and the TTS found in Shima Kunio's variorum. Strictly speaking, the citations in the HFZ are themselves transmitted versions of the Laozi and thus, unlike the MSS., subject to both the internal corruption characteristic of a long vertical transmission and to horizontal corruption (i.e.. contamination) from the other TTS.. Is it possible to discover by collation the extent to which these kinds of corruption have occurred? In their original form, the Laozi quotations in JL and YL date to less than a century earlier than the MWD texts and it seems worthwhile to attempt to characterize the state in which they have survived.2 The immediate aim of any collation is the discovery of 92 variant readings. Once isolated, however, not all types of variation can be considered egually significant. For our present purposes 'significant variation' can be simply defined as discrepancy between the substantives of the textual witnesses. The clear distinction made in Western criticism between 'substantives, the actual wording of a text', and 'accidentals, the transient [graphic] forms' of the words, seems eminently applicable to the criticism of Chinese texts.3 In the present collation it would be mistaken to regard as significant variation in accidentals between the MS. witnesses and the HFZ citations because the best available version of JL and YL is a modern critical edition of the HFZ, a version from which the original accidentals (i.e..simplified forms and LCS) have long vanished through graphic standardization. This principle of discounting variation between accidentals when the sources of the variation are separated by many centuries has been stated by Fredson Bowers (1975:359):•.,,the distance between the classical holograph and the inscription of the preserved examples makes it impossible, usually, for any of the accidentals to derive from the holograph with enough fidelity for any question of authority to inhere in their forms.. follows that the establishment of such a text does not include the variant forms of the accidentals.' The lack of an early MS. version of the citations in JL and YL thus restricts us to the:investigation of substantive variation; however, we may still be able through contrastive analysis to form an accurate impression of the HFZ version of the Laozi.. 93 The relevant data for this study must be further limited to passages in the JL and YL explicitly acknowledged as quotations from the Laozi by the use of the introductory locutions , 7^\%> 7ti-3-^ / etc. As a cursory examination of Shima's variorum will show, not only JL and YL but other chapters of the HFZ as well contain passages parallel to Laozi, but since their source is not identified, we cannot assume that they represent faithful citation of the version of the Laozi known to the author. It is just as likely that they are. paraphrases of material from the Laozi or perhaps, as suqqested by Kimura Eiichi, common maxims current in Zhanquo times and thus do not derive directly from the Laozi at all. The use of etc. by HF to introduce a citation suqgests, on the other hand, that he is guotinq verbatim from a text of the Laozi. We must immediately insert a caveat here, however, and cite two instances where 9 introduces quotations which include emendations of the Laozi apparently made by HF himself. HF was a skilled rhetorician, not a professional scribe, and on occasion his eagerness to drive home a point may well have overcome any scruples he felt about the faithful citation of his sources. This is surely the best way of accounting for the variation in the following two cases. (l.HA±*^-k >A.£*^i Jb^^-tki HFZ 403, KS 204, Ch. 71 £*faU£^ . •^^•*i]^3ah BL 7b. 11 AL is defective for most of this passage. Significant divergence between YL and BL is confined to the variant 94 ^^/^^ . However, as noted by Gu Guangqi (HFZ 404.6) , there is good reason for supposing that HF modified the original wording of this passage to make it better conform to the historical illustrations he cites. After referring briefly to two incidents from early history he states in conclusion, "The reason the King of Yue became hegemon was that he was willing t^^) to serve. . The reason King Wu became king is that he was indifferent (^-^) to an upbraiding. Therefore [the Laozi ] says, 'The reason the sage meets with no difficulties (^^) is that he does not consider them difficulties (^=Jsi ) ; hence, he is without difficulties (-fe^).V' (cf. Lau 1963:133).. The reading ^ -J* for the second Tr- ^ of the YL, though preserved in every other witness of this passage, is not appropriate to the context created by HF's historical illustrations.. In fact, it expresses the meaning opposite to that intended by HF. There are thus good grounds for assuming the variant is an emendation from the hand of HF himself.. 2. jtjlj^il^ .... tl^^^lSt, HFZ 414, 416; KS 122, Ch. 33 BL 14a. 7 The variant |^ is unigue to the YL. Chen Qiyou (HFZ 416) explains it as a deliberate alteration of the text by HF to reconcile Laozi's maxim to the story cited in illustration. The maxim and HF's apologue both point up the difficulty of self-knowledge. In the case of HF's illustration, however, the message is expressed in a simile linking wisdom and the eye: wisdom consists not in the breadth and distance.of one's vision, 95 which imply ambition in this context, but in the ability to see oneself. Therein lies its difficulty. Chen argues convincingly that Laozi's -^Q was changed to fjj in order to accord with this imagery. If we accept this explanation of the origin of this fj[y# the variation then loses significance because it does not represent a traditional reading, but rather an emendation. The T| following %Q in all the TTS was probably omitted accidentally from BL. AL is unfortunately defective at this point, but ^ does occur in both MSS. after the subject in each of the. other eight parallel sentences which make up this chapter. The syntax of the YL citations of this passage differs considerably from that of the MSS. and TTS. The latter are sentences consisting of a nominalized verb-phrase and a noun predicate. (The final -till of the MSS. is required by this grammatical structure; note its preservation in the FY tradition.) The citations in YL are transformations by object exposure of verbal sentences. They can be parsed as exposed direct object (the noun phrases ^ IL /l^tffe) followed by an anteposed resumptive pronoun ^ (antecedant ^^s/^kftj^), the verb (^J} ) and its denominative object (aJ^/5^) • Different as they are in terms of structure, the semantic value of these variants is practically identical (excepting, of course, the discrepancy between Hi and -&a). . It seems impossible to determine whether HF altered the syntax here by introducing the verb %q or whether he was quoting accurately from a version of Laozi rather different from the one reflected in the MWD MSS..and the TTS. 96 These two examples may raise the suspicion that HF, his use of the quotative expression ^3 notwithstanding, in fact quoted from his version of the Laozi rather loosely, altering it freely to suit his own arguments.. If this could be shown to be the case, any attempt at collation with the MSS. would by the same token become meaningless as the variants in JL and YL would lack any traditional (i.e. . pre-HF) validity. That most of the variants in the Laozi citations represent HF's own emendations is certainly one possible hypothesis. I do not believe, however, that it is of much help in explaining the bulk of significant variants.. It is one thing to be able to show with a fair degree of plausibility that a particular variant can best be explained as an emendation of the original reading to suit HF's argument. However, to propose that every instance of significant variation can be accounted for in this same way is clearly a priori. This explanation must not be applied mechanically, but kept in reserve for cases of variation which suggest it, i.e. cases like the above in which we observe a close congruence between a variant unique to the HFZ citations of Laozi and the gist of HF's commentary. In the following exploratory examination of the three witnesses, we must keep clearly in mind all the possible types of corruption which might explain substantive variation. Generally speaking, once a variant can be accounted for as the result of obvious scribal error as in Ex. 7, 15, 29, 31, 46, 47, 80 and 81 below, it loses significance, i.e. it no longer supplies evidence of the separate derivation of the 97 witnesses. On the other hand, an unaccountable variant may conceivably constitute evidence of a separate textual tradition. 3. i§.£.»ii!L ^l!_-tkj HFZ 369, KS 54, Ch. 1 i|,;L=l4L]i|L«j4MLi •. ^M-^AL 8b-8 Though the final 4u is attested in the Huainanzi and Wenzi reflexes of this passage, none of the TTS retains either -tJu. . This is evidence of what has already been pointed out by other students of the MWD MSS., namely that all the.TTS have been heavily edited of particles, particularly final -4b and Jj^. . One can only hypothesize about why and when this came about.. One theory suggests that the early versions underwent a bogus archaization at the hands of editors in order to make the text sound more arcane and oracular (Pulleybank 1979:13-14). In any case, that the particles were deliberately excised from the text seems almost certain. There are simply too many of them for us to assume that they could all have disappeared through accidents in transmission.. Furthermore, their highly consistent absence from all the transmitted traditions, with the exception of FY and occasionally HSG, suggests an early date for their removal. It is perhaps not unreasonable to associate this sort of tampering with the text with its use in the popular religious movements of the Eastern Han.. One might make the further inference that all the TTS, again excepting FY, derive more directly from such a popular text, already shorn of particles, than from the MWD versions. Since the variation in syntactic structure : between the MSS. 98 and HFZ versions of this sentence can hardly be attributed to corruption or emendation, it would seem to point to separate textual traditions.. The presence of ^ in the HFZ permits two interpretations. Since the original citation probably had , ^ may be the result of contamination from the;TTS all of which have the taboo-substitute. It may, on the. other hand, conceivably be an intrinsic emendation this text during the time the taboo against 'V~ was in effect. The six-character phrase /r" found in JL is cited twice by Shima in his variorum, once at Chapter 3 and again at Chapter 64. In the present study this phrase, which occurs only once in the HFZ, is collated with Chapter 64 (Ex. 79). 4. .^t; jfc4^£.^i|^HFZ 368, KS 80, Ch. 14 There is no discrepancy between this quotation in JL, the two MWD MSS. (the last graph in 4L is illegible) and any of the TTS. . The final -tlu found in the Wenzi version of this passage and the ^ -tj^ in the Huainanzi can both be explained as grammatically determined by the nominal sentences which incorporate these two noun phrases from the Laozi.. These eight words form the predicate of a verbal sentence in the MSS. and most TTS, hence final -t?U is not required. The main verb appears to have been dropped from four of the XE texts.. The reflexes of Laozi 25 (noted by Shima in KS 104) are not prefaced by and hence not collated here. 99 5.4^j$i^f||M^ HFZ 391, KS 106, Ch.,26 jyij^fl AL (LZ 25), BL 13a. 1 In this.passage there is no significant variation between any of the witnesses. The FY text shows a marked preference throughout the Laozi for the graphic form-^-3 while the other TTS almost invariably use^. This sort of graphic idiosyncrasy could, of course, have been imposed rather late in the transmission of the text; still, it is one of the features which distinguish the FY tradition from all the other transmitted versions of the Laozi. "tol] 3%>A AL 12b. 4, BL 13a. 1.M]^ AL, ^ BL. M± J^ii^ AL, ij_ & BL. . f -ttj ± AL, BL. , The initial ^ €) in the lemma is quite possibly a copyist's duplication of previous occurrence of this locution a few words earlier and nas perhaps replaced the Jb-^ found in both MSS. and all the . TTS. The variant -i&A in WB and HSG for % tj- elsewhere appears to be a relatively late emendation attributable perhaps to the influence of popular reliqious Daoism. BL alone preserves the reading where all other texts have ^fi , a synonym variant.. This is one of the few instances of significant variation between the MSS. The pronominal A appears in both MSS. as well as in the FY text. My intuition is to regard the version with ^ as an older reading.. Because its presence here seems not to be required by the sense or the syntax, it is easier to conceive of its being lost or deleted in 100 transmission than to propose its later interpolation.. As the sentence which precedes the final -thi in the HFZ reflex of this passage does not involve noun predication, the presence of -iki is not obligatory. T.M^^X^^^^^T^fl^EitS'l^HFZ 391, KS 106, Ch. 26 j] i AL 12b. 5, BL 13a. 2. ] 5. AL. .j? ] ^ AL, BL The reading with i is unique to the MWD MSS. The phrase X i. is not attested in any of the major pre-Han texts for which indices have been compiled..^ / -f" / i§ ^ ^- iy^/IH/ ^ are, on the other hand, common in such Zhanguo texts as Mengzi , Mozi , Xunzi and HFZ. This suggests that the MWD variant may be a scribal error for iL . The fact that both MSS. show the same error is evidence for their common derivation, though, of couse, the variation here with other texts is not necessarily separative. The alternation between Etl and ^ presents a more complex problem. There is little doubt that this is a very early variant since it is mentioned by YZ in the fragment of his commentary to the Laozi preserved in the Daodezhenjing xuande zuansu (^^&i&^%.2jfc>) of 964 (KS Thus there is evidence for regarding the ££_ in the YL as a genuine early reading rather than the result of contamination from the HSG tradition. The occurrence of ^ in the XE texts leads Shima to conclude:that XE follows 'YZ's alteration' of the reading S_ and hence that YZ's text was the basis for Zhang Lu's redaction (i.e. the XE text). The presence of ^ in both MSS. proves that 101 this variant is considerably earlier than YZ; hence, XE may not have been based on YZ at all. Likewise, the reading 3_ of the HSG tradition is not an 'emendation' of YZ's 'emendation' of HFZ, but, more plausibly, a descendant of the genuine pre-Han reading preserved in HFZ.. Shima's hypothesis about the late origin of the HSG text leads him to ignore this possibility and to impute corruption to HSG more readily than the evidence seems to warrant. e.?t*fo**#*5ft***-4*™*> HFZ .,7. KS 108. Ch. 27 % AL 12b. 11- AL. -kv]$P$ AL, BL 13a.8. AL. . 3b>]B>f. AL, BL M~h. in AL is an LC for the superior Sf4. of YL, BL and all the TTS: '...though it seems clever, betrays great bewilderment' (Lau 1963:84). Both MSS. have the graph 4 after-^0 . This particle occurs most commonly in classical Chinese as a locative preposition similar in function to or as a final marker of interrogation. Its use here is not particularly well attested and can be described only somewhat imprecisely as a 'pause particle'. Fortunately there is an example from the Shiji (86.22, quoted in Yang 1955:164) whose syntax parallels the Laozi passage rather closely :^(j^j ykfc-Jsfc ->fig A.4- ^J^A* >fcrJ£#l| •Though Jing Ke fraternized with tipplers, he was of a pensive and studious character'. One might also adduce a similar use of 4f" in the frequently encountered locution 75^ IL^ 'at this point, then' in the language of the Zuozhuan and Guoyu. I am inclined to regard the MSS. reading ^hn as more original, but the lack 102 of 4t in the YL reflex can hardly be taken as good evidence for separate derivation.. The order of the final rhyming binomes found in HFZ and all the TTS is reversed in both MWD witnesses, another bit of evidence in favor of a common derivation for the two MSS., but again, not necessarily a separative discrepancy.. 9m$^*JV®&^&&%^®&£- HFZ 394, KS 128, Ch. 36 m ALl4a. 7, ^ BL 14b. 4. ® ] £ AL, BL The variants between YL, the MSS. and the TTS are all LCS and thus lack any significance for the tracing of textual derivation. The agreement of many of the graphic forms found in the TTS with those in the YL is best attributed to the independent graphic standardization of these witnesses rather than to direct transmission between them. Because the process of standardization was occurring pari passu with vertical transmission of the texts, it is impossible to assign archetypal authority to any one of these loan variants.. Note how even the two MSS. differ in one case in their choice of graphs to write what is presumably the same syllable. lO./ft^^M^®^*- HFZ 394, KS 128, Ch..36 AL 14a.8, BL 14b. 5.^1 ] $ AL, BL The MWD reading with ^ is found in all TTS except two of the XE MSS. from Dunhuang and Fan Yingyun's recension. The two Dunhuang texts show which is no doubt a loan for 'to appropriate, help oneself to'. Since the XE commentary uses 103 ^ in its paraphrase of this sentence, Shima rightly regards H^as a late corruption. FYY selects the reading 4>^_on the grounds that ^ is not an early variant and Shima infers from this that the substitution of ^ for^*- originated with XE. The occurrence of ^ in both MWD MSS. shows that this variant was current long before the XE tradition was established by Zhang Lu. The only other citations of this passage in the pre-Han corpus, both of them attributed to the Zhou shu (Jf] ^) rather than to the Laozi, have 3pv as does the quotation in the Shiji sup yin (^'kp. . Barring the assumption that the variant originated with the MWD texts, this sort of significant divergence between the MSS., the HFZ and other pre-Han sources again implies independent derivation, i.e.. the existence of distinct textual traditions even in pre-Han times. and^| in the sense of 'give* are perfect homophones in OC and often foun.d as textual variants because of their synonymity. The TTS all show -^f as in YL. 11.&*»Ffl&#>£5Jfil ff A^IL^^AHFZ 392, KS 128, Ch. . 36 °\Hfj] fl&AL (LZ 28) BL 14b. 6. V^AT)] ^fl AL, >HTJ BL AL and the Daozang edition of HSG both lack the first 5j , reading 'The fish does not get free of the abyss1. The parallelism with the next sentence and the fact that no other pre-Han reflex of this passage lacks •») suggests that the reading in AL and the HSG text is defective, the result of probably independent and accidental omission of the graph. The 104 word is found only in the YL citation. In a case like this it is impossible to determine with any degree of certainty whether the variant is intrinsic and thus separative or extrinsic, the result of interpolation.. One can only remark that the preservation of such unique readings in the TT of the YL implies that the text has avoided pervasive contamination in the direction of the TTS of the Laozi.. On the variation between and >)ff see Part One, Ex. , 5. The MSS. are unique in lacking the possessive marker between •fP/il and i'Jg'. . This may point to textual corruption as it is unusual in OC to find two nouns, the first a genitive and the second modified by an adjective, without occurring between them. The variant is probably conjunctive. Note the retention of ^ in the FY tradition and in three of the XE texts. The passage from JL first found collated with the opening of Laozi 37 in Shima's variorum (KS 130) has been miscopied.. It is correctly reproduced in KS 132 and is treated as a reflex of Chapter 38 in the present collation. The sentence occurs only once in the HFZ. . U. %&>RFZ 326 f KS 132f ch. 38 There is no variation with the MSS. or any TT. 13. ^^-^4?-^ HFZ 328, KS 132, Ch. 38 7f J >'k AL 2a. 2, BL 2a. 1 105 This is yet another case where there appear to have been two textual traditions as far back as the witnesses can take us.. The version with ^ , preserved in the YZ and the FY tradition, means: 'The highest power is inactive and yet there is nothing it does not accomplish'. The version in both MSS. with instead of ^ is reflected in the WB and HSG textual traditions: 'The highest power is inactive and without a motive [lit. 'a means'] to act' (cf. Chan 1963:167) There is unfortunately no definitive evidence in the context of chapter 38, or, indeed, in the entire Laozi which gives the lie to either of these variants. The evidence of the MWD MSS., if anything, complicates the issue by showing that the reading -V\ is considerably older than the XE texts, the earliest evidence of it heretofore. Thus it seems safe to treat this variant as further evidence of a derivation for the MSS.,distinct from that of the version of Laozi guoted in the HFZ. ^^%^n HFZ 329, KS 132, Ch. 38 The final -t!zJ found in JL and both MWD MSS. . has not survived in any TT. 15. ^ ^ >^4-tizJ HFZ 330, KS 132, Ch. 38 -f^]-fe BL 2a. 2 This variant is no doubt a copyist's error since /'A»> is mentioned and described in the same MS. column a short distance above this passage. This sort of error, the.obvious result of 106 scribal inattention, does not constitute evidence of an independent derivation for BL. Since we can provide a plausible rationale for the error in the textual environment, the divergence loses significance. Again, final -tkj has disappeared from all the TTS. 16>l>Afe'^t!LHFZ 331 Though introduced by this sentence is completely unattested in any Laozi witness. As suggested by Song Gaoyuan sflE* HFZ 332. 6), probably introduces a reiteration of HF's own definition of A-g_ found at the beginning of this section of JL: i&k ft y^&jfaL 17. *!|^7UffTJ ^X,^ HFZ 331, KS 132, Ch..38 3 ^fe^ BL 2a-3 <&L defective) The final -%J, unique to BL, is not obligatory in this sort of verbal sentence. 18. ^i/V ^ft^-HFZ 331, KS 132, Ch. 38 4^.Jl,J^i_BL 2a. 3 (AL def ect ive) . yfo ] 7b AL 2a. 3, BL AL is legible once again from . The m-J found in BL is retained in all the TTS. Its absence in JL is perhaps due to the presence of HF's commentary between these citations (i.e.. Ex. 17 and 18).. With the exception of the WB tradition and FYY which have ^75, the TTS all show Mb as in JL. .. 107 19. *4L^*fe*^**^ HFZ 331, KS 132, Ch. .38 AL (LZ 1).**Lfctf|_BL 2a. 3 ^^%3/a^iAL, BL. AL, BL Initial absent from JL but attested in both MSS. and all TTS. AL repeats and adds , the result a longer and perhaps more original version of this passage: 'Therefore he loses the Dao. Having lost the Dao, he loses virtue'. Both Ja and ^] are frequently encountered in the MWD MSS.. for the standard HFZ 331, KS 132,, Ch..38 % AL (LZ 1),^|^BL 2a.3.^&3&]£)*i_ BL The many citations of this passaqe extendinq from the •outer' chapters of Zhuangzi to the Taipinq yulan, the TTS of Laozi and the MWD MSS. all agree in omitting the second occurrence of ^ in each of these four parallel sentences. As in the case of the unique to YL (Ex. 7) , the origin of this variant can only be guessed at; hence there is no justification for treating the discrepancy as separative. 21 . .^1 ifciHFZ 335 This brief statement, though introduced by c&Jrjr has no reflex in any other text of the Laozi. Wang Xianshen (HFZ 336.1) suggests it is HF's own abridgement of the passage from Laozi he quotes next.. 108 22 At % M&^Mj%%% HFZ 335, KS 134, Ch. 38 %-1L]tX$iL BL 2a.5. ^ BL 4f ] 4LAL (LZ 1), BL In AL the first seven characters of this passage are illegible. The absence of ^ from JL can perhaps be accounted for by the presence of commentary between the end of Ex..20 and the beginning of the present passage. (These are contiguous passages in the MSS. and TTS; all of the latter, except for YZ, retain ^ .) HF's intervening exegesis vitiates the pivotal function of and provides a reason for its omission, and are OC homophones. The first 4L in JL and BL is unattested in the TTS, though the second is preserved in FY and three of the HSG witnesses. The final^f-, unique to the JL, cannot have its usual function as an interrogative marker here. Yang Shuda (1955:163) lists several examples of used as a final exclamatory particle, but the presence of final -tbj in the JL version of the parallel sentence which follows this one suggests that ^f* here may be a corruption. 23 .^^^ilA^Nb^ ^/L-! 4l\ HFZ 338, KS 134, Ch. 38 Again, though the first -tbl is not peserved in any of the TTS, the second is retained in the FY tradition and two of the HSG texts. The tendency for final particles to be dropped from all but the FY text has been noted earlier. Though JL matches the MSS. word for word in this passage, the variant~5Co for ^ is found in all the TTS. The only instance cited by Shima of H in this position occurs in a Jin dynasty work Laozi yiwen fanxun 109 ki*L) by Sun Sheng (3^§§> c. . 302-373). _ Unfortunately, this text is listed neither in Shima's bibliography (KS 35-48) nor in Yan Lingfeng's (Yan 1965), nor in the Harvard-Yanjing index to the Daozanq. One can only assume that Shima has quoted the passage from some third work which he fails to identify. . Perhaps the only inference one can make on the basis of the distibution of this variant is thatHta has replaced the earlier word to eliminate the repetition of it which occurs at the end of the previous sentence. 2ti.*tAfetM*A£^fef^ HFZ 340, KS 134, Ch. . 38 TJL]/^ AL 2a.5, BL 2a.6. /|^ ] Jf fin * AL (BL defective). ] ^ifo* BL and y^- , synonym variants, are frequently found alternatinq in the TTS. The MSS. show a marked preference for ^throughout the Laozi. The first occurrence of j£ in the MSS. is reflected only in the Wertzi citation of this particular passage. A number of transmitted witnesses show M in the second sentence, however. The Si between^ and Tr- has perhaps been omitted in AL; it is attested not only in BL but also in the citation of this second sentence in the Wenxuan commentary. 25. *\4^3M:t_iHFZ 340, KS 134, Ch. .38 i\] "&jk AL 2a.6, BL 2a. 6. $L AL, BL All TTS except YZ show re. W\j is LC for 110 25.*!|19&5i^E>7<fi^f HFZ 413, KS 142, Ch. 41 B^Li] &j BL 2b. 5 (AL defective) No significant variation in any witness. 27.7\T7^jt_^n^S7>A^'tb HFZ 359, KS 152, Ch..46 %~]% AL 3a.8, BL (LZ 38) Neither the.MSS., nor the YL version of this sentence (HFZ 387), nor any TT retains final -tiu here.. Since this is a verbal sentence, final -kj would seem to be optional. . FY preserves the variant 3nk f or J| , probably a scribal error induced by the similarity of the two graphs. . 28. 7\T-fe".dt/£ B) '^y? ^|>t.HFZ 360, KS 152, Ch. 46 ^"Mfe]^r BL 3a.10. tf£ AL 3a.8, BL The omission of TN I- in BL is quite likely a copyist's error. Final ^ is unique to the JL reflex; it is absent from the citation of this passage in YL (HFZ 387) possibly because the two clauses of this sentence are there separated by commentary, attenuating the aspectual relationship between them and thus increasing the likelihood of the final particle's omission in the course of transmission.. HF*s copy of the.Laozi possibly had here, a significant variation from the MS. witness es. 111 29.^^7v.7j^ HFZ 361, KS 152, Ch. 46 3*§l]lfjl AL 3a. 8, BL 3a. 10, YL (HFZ 387).. 7\M*] X, BL The JL variant 7K§I for the IfjL of YL, the MSS. and TTS is undoubtedly a copyist's error. ^Si has apparently been raised from the beginning of the following sentence, hence the discrepancy is not significant. The omission of 75^ in BL is probably also a scribal error. This sentence is unattested in the WB tradition.. 30. ^8i1£ HFZ 361-2, 387; KS 152, Ch. 46 &8i]JSj&.AL 3a.? BL is defective for all but the first character of this sentence. Significant variation between the witnesses is lacking. . 31. ^1|.'®^$M%HFZ 387, KS 152, Ch. . 46 4^ ] JL (HFZ 362) BL is defective here. . The 'l-^ of AL, JL and YL is preserved in the FY tradition. Two of the XE texts show ^ , a rhyming synonym variant, while the other TTS have 7v . Both of these variants may well be facilioreg.lectipnes. The JL reflex of this sentence shows the unique variant for ^rf- . This can be accounted for as a transposition of the last word of HFZ's commentary immediately preceding this citation:i§^&&"^ i^^^'J . It is not inconceivable that the occurence of ^'-l in Sima Qian's 'Letter to Een An' (Hanshu 62. 112 2727) is a reflex of this early scribal error. 32 ,-kQ ^X^j^J^ HFZ 388, KS 152, Ch. 46 " 4j ] 'ML AL (LZ 3) Because of lacunae we cannot be sure about the MS., version of the last sentence of this chapter, but the occurrence of 4-S-/c_£in AL strongly suggests that the MSS. . parallel the TTS: "^u>iL£J^^/cL-^ . There are at least two possible explanations for the variation between this and the: YL reading.. Wang Xianshen (quoted in HFZ 390.20) assumes corruption and would restore a second to the YL version between and ^j. . Equally plausible is the hypothesis that HF has cited only the first clause of this final sentence, a clause which, with or without the word is itself a grammatical sentence. Though there is no conclusive evidence in favor of either of these possibilities, the presence of in YL suggests that the variant is separative. The final found in all three of these witnesses is retained in all the TTS except the XE tradition. 33. T&f*\ >A£s HFZ 409, KS 154, Ch.,47 ^)>VJ >VA AL 3a. 11, BL 3a. 10 The jfi found in these three and other early witnesses has not survived in any TT. Of the TTS the FY tradition alone preserves "3] >yA , perhaps reflecting the HFZ version of the Laozi. The >A of the MSS. is retained in the HSG tradition. , Both "^J and are lacking in all other TTS. 113 34.^^3^^ «JJ>A^a ^V|__ HFZ 409, KS 154, Ch, 47 ] 7x£iAL 3a. 11,£1J BL 3a. 12. "q >A AL 3a. 11 (BL defective) ^SJ is an acceptable LC forjlftj. Again ^ is absent from the TTS, Sj >/v is preserved in the FY tradition and >VA. only in all the HSG texts but one. Many of the TTS show the variant Hi for -£tj. |JL occurs in the acknowledged citation of this sentence in the 'Daoying' chapter of the Huainanzi. The same passage appears unacknowledged in the 'Zhushu' chapter where the reading 7\Qis preserved. This strongly suggests that the original has become ILi in the 'Daoying' through contamination from the TTS. Many of the acknowledged quotations in the Huainanzi appear to have been contaminated in this way while the unacknowledqed witnesses have survived unscathed. 35. ^^^^ ^>3^^> HFZ 410, KS 154, Ch. .47 iB ] iti^ . ] AL 3a. 11 AL is defective following the second 3t. . , Here the two MSS.. show significant variation. Whether the variation is separative or not is difficult to determine since the topic markers and ^ are not obligatory here.. The HFZ citation agrees with BL. 36. *%** X&^W HFZ 410, KS 154, Ch. 47 afl BL 3a. 1. 7r ] % BL (AL defective throughout) This passage is not well preserved in either MS. Though the Jfa of BL is reflected in all the TTS, Shima restores 9^ to 114 his recensions of the WB and the YZ traditions on the basis of the wording of the relevant commentaries. /$a belong to different rhymes and thus are best treated as semantic variants. The TTS all have 7r~ for the ^7 of BL. The other two negatives are illegible in the MSS. The fact that % was a taboo character during the Former Han suggests that the variant %, the official ersatz for % , is not original. Incidentally, the frequently encountered description of 7^ as subsuming a direct object does not seem appropriate to this example. jj^ here is surely intransitive and thus this % cannot be explained as a fusion of ^ and ^ . 37. ifcii ^ > TU HFZ 371, KS 160, Ch. 50 Both MSS. are partially defective here. What is legible shows no variation with JL or any TT. 38a. yi £_^_r& — % HFZ 371 (1.3), KS 160, Ch. 50 b. '±H&- HFZ 371 (1.4).. This passage is quoted twice in sliqhtly different form in JL. It cannot be collated with the MSS. because of their defective condition. The TTS all agree with 38b. The particles -^2j and ^ , because they are not reflected in any other version of the Laozi, have been treated as interpolations by Chinese commentators on the JL (HFZ 373. 5). 39. f\L HjLltM— HFZ 371, KS 160, Ch. . 50 115 The legible fragments of the two MSS. show no variation with JL or the TTS. . "ffi? is omitted from the beginning of the: JL version because the quotation is interrupted by HF's exegesis; hence, we cannot be certain whether or not HF's version of the Laozi contained this particle. It is preserved in the YZ and FY texts. With the exception of the FY text, all the TTS exhibit wording apparently derived from the shorter version of the MSS. The complete agreement of the two MSS. is again worth noting. It seems impossible to determine with any certainty the nature of the relationship between the longer version of this passage found in JL and the shorter form appearing in the MSS. They are semantically equivalent and the MS. version may conceivably be the product of early editing of the more original longer version, but there is no way of substantiating this. There are other cases of variation between the slightly fuller version of a passage in the HFZ citations and the more concise MS. versions. (Cf. Ex. 52, 72 and 73 below.) The MSS. and JL preserve a reading in the last half of this passage which is unattested in any TT. Stl is omitted from all of the transmitted versions but FY, three of the HSG texts and a quotation in the Wenxuan commentary. The third occurrence of HFZ 371, KS 160, Ch. 50 AL 3b.6, BL 3b. 10 BL : differs from AL here only in the character 4"iL, an LC for 116 appears as in FY, FYY and two WB witnesses only and is omitted elsewhere. Both Chen Qiyou (HFZ 374.13) and Shima (KS 161) explain this occurrence of ^L. in JL as a graphic error for (The miswriting of /C. for in certain versions of Ch. . 60 was noted as early as Tang times by Cheng Xuanying, ^ ^ , as guoted in KS 11b.) However, the fact that both MSS. also show Tt. would seem to authenticate the JL reading. Furthermore, although the two graphs and <fo could perhaps be mistaken for each other in grass-script (see examples in Akai 1974:19-20, 40), they are guite distinct in both MS. script-styles and much less likely to be confused (cf. ^ , ^ in AL and i , 3^ in BL). Thus % ^.^ti i-dL^_+ M •=-can be defended as a genuine pre-Han reading which has undergone partial omission and or emendation in its transmitted reflexes. It is easy to surmise why such faulty transmission may have occurred: as it stands in the MSS. and JL the string of eight graphs is not easily interpreted. almost surely means 'thirteen' rather than 'three put of ten' as WB and other commentators have felt constrained to interpret it, but it is impossible to say with any certainty what 'thirteen' things are being referred to.. HF explicates 'thirteen' as the four limbs and nine orifices of the body but this interpretation does not seem to fit the instance of in Ex. 40. The meaning of the last occurrence of ^ is also problematic. Construing it as the subordinating particle of possession, the string reads 'all go to the thirteen places of death'.. If, on the other hand, this is read 'go', as suggested by the punctuation of Gu Guangqi (HFZ 373. 12), the 117 result is even less satisfactory: 'all go to the place of death, go to the thirteen'. It is, of course, possible that the reading in the MSS.. and JL is itself defective and that the corruptions found in the transmitted versions represent attempts to make:sense of what was already in pre-Han times a corrupt passage. There appears to be no ready solution to this puzzle. 41.4i^iHFZ 372, KS 160, Ch. 50 %h ]#_lAL 3b.7, BL 3b. 11 This citation is introduced by the guotative verb and is reflected in all the TTS. The occurrence of *tiap in the MSS. is best treated as a synonym variant for 4% *sriiap. . 42. p^^t^iS. 92.^ A^L ¥ ^ HFZ 372, KS 160, Ch. . 50 ^E] ^ AL 3b.8, BL 3b. 12. . BL. 7^ J 4k AL, §L BL. ^^J^^aL, j&^BL The variant f^. for ^ is shared by the MSS..and is probably a graphic error. The compound not attested in early texts, of BL (AL is defective) is retained in the YZ text writteni^_ . This probably represents a textual tradition distinct from that of JL and the other TTS rather than a miswriting ofi^, as Shima proposes (KS 161). 7\ is an LC and ^ an alternative: graphic form for 'rhinoceros'. The reflex in the TTS of the second sentence parallels the version in AL graph for graph. .yr7fl , a rhyme word and;^, a rhyme word, are not LC but substantive variants.. 118 43. %kft^&kfa&ift4&tAllFZ 372, KS 160, Ch. 50 AL 3b. 8 (BL def ective) . Ji\ ] AL, BL 4a. 1 On the variant ^ see Part One, Ex 28.. ^ is occasionally found as LC for /'\ in the TTS of the pre-Han corpus (examples in ZD, s.v.tic def. 3.1). 44. £ TVr^^g HFZ 372, KS 161, Ch. 50 The MSS., defective for most of this passage, show no variation with JL or the TTS. The H of the JL witness has apparently been miscopied as 77 by Shima. 45. ^^^ H HFZ 372, KS 161, Ch. 50 Again, there is no variation between AL (BL is defective) and JL. The final ^ is preserved in the FY text only. 46. £'^"3^ %^t!3&HFZ 400, 403; KS 164, Ch. 52 Both MSS. are partially defective at this point. No variation occurs between the legible graphs and YL. Two of the XE MSS. show corruption here. One repeats the first ^3 and has replaced ^f" with ffl , the latter apparently raised from the beginning of the following sentence.. This second scribal error is also found in another Dunhuang MS. Curiously, Shima (KS 165) has selected as the original reading in the XE tradition, 47 . 7\ ^ %j 75^Li HFZ 380, KS 166, Ch. 53 119 ^TTo^Tv^^^Wl BL 4a. 11 and LZ 40 (AL defective) Since these two expressions are both introduced in the JL by the locut ion °j?, it seems worthwhile to collate them with the other witnesses. A^is attested in every text but the word ^uis unique to the HFZ. The commentators (HFZ 381.2) offer speculative explanations for this graph including the hypothesis that it is a corruption. With the exception of two XE MSS. where ^ , apparently raised from the following sentence, has replaced 7^ , the TTS all agree with BL. (A similar scribal error was noted in the same two Dunhuang MSS. in the previous example.) is, of course, an 'authorized' LC for the 7^11 of the TTS. 48.^7^ HFZ 380, KS 166, Ch. 53 This is another two-character phrase introduced by 75f i^U . The word is reflected in all the TTS in the sentence 7\4LT& % "TJ ^id^(FY) . A. is obviously not associated with ^ in the TTS. *S -X is defined by HF as test which is in turn glossed as X-'Ti'f 'the forking of a by-path (from the main road) '. Gao Heng (HFZ 381.4) attempts to substantiate this etymologically, but the binome M ^is not attested elsewhere with this meaning. Wang Xianshen and Chen Qiyou, on the other hand, understand literally as 'splendid' and construe ^Tp-^l'x/Tr figuratively as 'the point of (moral) deviation from the great road (i.e. the Dao) ». For Laozi and HF, a preference for the show and luxury implied by would surely constitute deviation from the Daoist principles of frugality and 120 simplicity. The MWD texts here read A^,^ % % IL (AL 4a. 7, BL 4a. 11 has for ) , showing no apparent reflex of the word 'fit at all.. The unique and problematic Ipr/^ of the MSS. is explained by the commentators as a loan foriM^f 'dale' or as an LC for ^a-itself (LZ 14.20).^^? is not attested in pre-Han texts and ^^L *kieng with its final -ng is perhaps less likely to have been written ?$ *keg than is the word S^k. *g'ieg which means •footpath' and differs phonetically from only in terms of medial and tone. Thus the variation with^L of the TTS appears to be semantic. The next few phrases from Laozi 53 collated by Shima are not introduced by a quotative verb such as 13 or . Two and three-character fraqments have here been woven into the texture of HF's prose in a manner similar to that used in the epexegetical commentaries (&&) to the Confucian canon by Kong Yingda (3t^Hl|j. They are omitted from the present collation. H9.*fMM HFZ 380, KS 166, Ch. . 53 Variation between the witnesses is lacking. 50. HFZ 380, KS 166, Ch. 53 Only the first graph of AL and the first two of BL are legible. ^ @ is reflected in three of the XE witnesses. The other TTS all show % |$ (FY tradition) or i^fj (YZ, one XE, WB 121 and HSG).. No TTS reflects the W\ (LC for) of BL. . The remainder of this chapter is too defective in the MSS. for collation. JL cites the words M 3- which are reflected in all the TTS, the latter graph in a variety of XS derivatives, viz.^ , ^,#5,^ etc. 51.lhjt_*-££ 4M&*%»HFZ 390, KS 168, Ch. 54 ^Jj^tJ^ BL 4b. 2 (AL defective) Both MSS. are again defective for most of this passage.^ is retained after both 3tl and 4tL in the great majority of TTS. 52.3-3!h-^$# ^cL-Bt-at*^!!HFZ 390, KS 168, ch. 54 >V.^] !<%k BL 4b. 2. .i*-tr*^« ] 7r-BL (AL defective throughout) The JL reflex of this passage (HFZ 384) is not preceded by a quotative expression but is woven into the flow of HF's discourse. Curiously, JL parallels the MSS., except for the omission of >VA occasioned possibly by the assimilation of Laozi's phrases to the wording of HF. is preserved in a minority of the transmitted versions, viz. all the WB texts and tiree out of four of the HSG witnesses. Its omission in the other TTS changes the syntax of the sentence so that the function of shifts from noun to verb. Neither the MSS. nor the TTS contain any reflex of the adverbial"^*1 in YL. 122 #***A*£4***-F*ftag HFZ 384, KS 168, Ch. 54 ^ ] H BL 4b.3. SJ^|| BL (AL defective throughout) Most of the TTS, notably WB and HSG, show 25* after 4% in each of these five parallel sentences. Presumably because the verb phrase ~M is bisyllabic in contrast to the monosyllabic stative verbs -|| /~f*L#jp i and in the other four sentences, 75 is emitted from the second sentence in both YL and the MSS. 7^ occurs even here, however, in the Wenzi reflex as well as in over half of the TTS, leading Shima (KS 169) to infer that this instance of 73 was added sometime in the Later Han. For the occurrences of 75 throughout this passage most of the XE texts show iftL , quite likely an early LC for 75 . flo has a reading homophonous with 73 and often rhymed in *-ag in OC (GSR #885). The taboo graph is reflected only in the FY tradition. LZ revises the transliteration'!^ in BL to 5^-. . The latter is found in the FY text. 4^ /y% and H are synonym variants. >A^r%j^ ^^fe^T HFZ 384, KS 168-9, Ch. 54 Except for the omission in one XE MSS. of the phrase >*^rfe*^ana the usual alternation of (JL, AL and FY) with §, variation between JL, the MWD texts and the TTS is lacking. 123 55 JP^*^*^^*^ HPZ 384/ RS 169R CH>54 ^ BL (LZ 40). BL (AL defective throughout) ^ and'/Sj are synonyms; the former is retained in the FY tradition only. All other TTS show^i . . is an LC for theo^, of most TTS. The final -t!?j in JL is unattested in the transmitted versions. 56. ^^*1* 3^ ^* '-5 7n?iX7^T^ HFZ 341, 342; KS 176, Ch. 58 AL lacks both instances of the particle which appears in all the TTS except the XE tradition. The first of these two sentences appears to have been omitted in the copying of BL (LZ 50.23) while the second is only partially legible. 57. "fe^^/fc^ HFZ 343, KS 176, Ch. . 58 BL 5a. 9 (AL defective) A large majority of the TTS have the Jftj of BL. HF seems to understand literally: "S&jttuAH ^^X^ "Thus he instructs people 'Be thoroughly familiar with its extremes'". If HF had construed 4& as an interrogative (i.e«. LC for fft^ ), it is unlikely he would have used the verbl^}..f& is also found in two XE texts. 58. A^i^L-^^ S ^--^^^ HFZ 344, KS 176, Ch. 58 -^Ljy^. BL 5a.10. ^-»y-J Wi BL (AL defective throughout) FY retains the topic marker -th\; FY, FYY and YZ retain final . The of JL is found, written , in FYY only. 124 59. Tafin*^! &^*Jfe»l J[^*^£iM&FZ 345, KS 176, Ch..58 Tlj^ BL 5a.10.$'j Jfrj BL.££ BL. . A] S^&BL (AL defective throughout) ^'-^ 'cut' and jfj'J 'prick, stab' are synonym variants. . The latter is unigue to BL; most TTS show^'J or 4^ both of which rhyme with^il . If we assume the rhyme is original, then ^ij is best explained as a scribal error for^l] since the MS. graph is clearly a clerical-style form of^l] . On for3^-, see Part One, Ex. 29. *t'iog is an acceptable loan for%M *diog. 60. >DA?-7^ HFZ 349, KS 178, Ch. . 59 -^a ] BL 5a. 12 (AL defective throughout) Though most TTS collated by Shima reflect the ^ of BL,-£n is found in YZ and Lu Deming's Jingdian.ghiwen citation of WB as well as WB's commentary. The of the WB texts can therefore be attributed to contamination. All but one XE text show 3^V for Unfortunately, the XE commentary to the 'Dejing' has not survived, so we do not know whether 5§ was interpreted literally ('model, pattern') there. *siak is most likely a semantic emendation rather than an LC for *si-3k. 6lM%%£*&m&W%.~*%lL*il{&*EFZ 350, 351 ; KS 178, Ch. . 59 S^]*^ BL 5a. 12 (AL defective throughout) The TTS all reflect the ^ of BL, the FY texts showing the homophonous'Mi. . The^ of JL is not a loan for these words, but 125 more probably a scribal error.. is preserved in YZr most of the XE texts and the FY tradition; other TTS appear to have replaced this by. . ^IL appears as & in four XE witnesses, an emendation attributable to XE, according to Shima. Another possibility would be to regard the homophonous & and flUL as early LC each of which came with the passage of time to be interpreted literally.. Lu Deming cites this passage from WB with ^ and WB's commentary uses in its paraphrase of this sentence. The extant WB texts, however, show here. FYY (quoted by Shima KS 179) regards fas WB's own emendation. The ^Lin the textual witnesses of the WB tradition would then be attributable to contamination. . flfiL and 41 belong to different rhyme groups ( ^Land^^ respectively) and could not be loaned for each other. All the TTS show X. for the/t_a*| found in JL and BL. . The final character ^A3>is defective in BL. 62.i^W#.*^*.**.W«*»4MfiHFZ 351 , 352 ; KS 178, Ch. 59 AL is defective. throughout and BL for most of this passage. The second pi'J has been omitted from YZ and four XE wit ne sses. 63. £*2t §'J=JX^ M HFZ 352, KS 178, Ch. 59 None of the legible graphs in either MSS.. show variation with JL. . Both the MWD witnesses are defective where fi'] would occur; this conjunction is preserved in the FYY text only. The presence of in AL here proves that HJ in the TTS represents 126 the original reading and is not a taboo-substitute for •. Incidentally, the wide-spread substitution of Ml for^-jS in the TTS of the pre-Han corpus has no doubt obscured the semantic distinction between these two words.. Even Xu Shen, who lived, of course, after the tabooing of ^ , merely glosses each in terms of the other, A worthwhile investigation of the distinction would obviously have to be restricted to pre-Han epigraphic and MS..materials. 64.7q 1^"JX^^ HFZ 353, KS 178, Ch. 59 No variation occurs between any of the witnesses. BL is defective at ->X--p< . 65. >^£*Sc©£*4^.%*, 4S.*4L»hj HFZ 354, KS 178, Ch. 59 AL (LZ 7) , ^fc BL 5b. 2. ] AL, BL This passage is preceded in the MSS. byS-s^i which may have been omitted from JL as redundant. On the other hand, YZ and one XE text also lack these two words. The three Dunhuang MSS. show^^ here. Shima incorrectly regards the resulting string as ungrammatical. It is not, if one punctuates after Mb. In fact, the final -tbj of the JL and the MWD witnesses seems to urge treating the last six characters in the TTS as an independent nominal sentence. The TTS all reflect the MWD MSS. in lacking IK before *3candM. & is LC for^ft... The ^ found in YZ, XE and HSG is probably cognate with^. Though they belong to different OC rhymes (^ and ), fifa has an MC reading 127 4£ homophonous with and the syllables are synonymous, meaning •base of a plant stalk'.. No TT has preserved final-th-u . 66 . £ M% 7^3 XA Hl# HFZ 355, KS 180, Ch. 60 Ml^jEl BL 5b. 4 (AL defective throughout) The nominalizing particle is preserved in Wenzi, YZ and FYY. Most XE texts show the synonym variant •fresh meat' for FYY has||$j, perhaps originally a scribal error for.t^- . . 67. XVJ_^£ 7VF^|^^^ HFZ 356' KS 18°' Ch* 60 >j£ ] il BL 5b. 4 (AL defective) -ii. in BL is most probably an LC for its XS derivative A^L which appears in all the TTS. It is difficult to perceive from Karlgren's reconstruction the phonetic affinity between ±u *gli-sp and 4^ *liad. He does not group them in the same phonetic series and, furthermore, treats A$L *giw-ad as a third, unrelated entity. It can be shown, however, that and 4£ both have 3A. as their phonetic, aven though Xu Shen regards A^L as a semantic compound (Ai As/ ). SW lacks an entry for Aii. and the homophonous zAii; it does, however, define another homophonous graph as S»a 'approach, oversee'. This same gloss applied to 'lil in the Guanqyun and the use of for AiL in the reflex of this very passage from the Laozi in the commentary to the •Chuzhen' chapter of the Huainanzi (quoted in Ding 1960:6870b) confirm the fact thati^and ^ are graphic variants for the same syllable. In his graphic analysis of Xu Shen treats as 128 signific and ^ (Karlgren's *d'-ad/*diad) as phonetic, itself means 'come up to, reach' (U^'$L ) and appears to be related etymologically to . The phonetic connection between $L and ~±L is paralled by the use of ^ as an abbreviated phonetic (^^) in ^ *d'sp, a word synonymous with ^ (fMj^ls]), according to Xiaoxu edition ('h^/fc) of SW. The phonetic correspondence between gushenq syllables reconstructed with final palatal glides *-jtl by Pulleyblank and rusheng syllables in -p is further borne out by the following cognate pairs, many of them related graphically as well: ft /&\. IV/j^, tt/^ » •fr/H and (Pulleyblank 1962:233-234, cf. Dong 1967: 241 and Li Fang-kuei 1975:257-268). This pattern of evidence justifies treating both i and as phonetic in idfc ( is, of course, signific as well) and grouping Ml and together with i. in the same XS series. FY and all but one of the HSG texts show % after^T.. 68.5||f$ %^^^ %A[%A-& HFZ 356, KS 180, Ch. 60 The TTS all lack both occurrences of -thv in JL and the MSS. 69 . $A # * 4% ^ HFZ 357, KS 180, Ch. . 60 Z^^%%A%^% AL 5a. 11, %A%4>$-L 5b.5 AL becomes defective after the graph ^ in the MSS. is best treated as a contraction of the negative vF- and an object pronoun that has the A of the previous clause, 7f,^A.~& f as its antecedent. JL has 7f which here does 129 not subsume an object; thus the text shows E^/ in the usual post-verbal object position. Incidentally, there is a curious shift in the JL from the use. of A as object of ^"fb in the previous sentence to the use of ^/ in this passage. This surely suggests corruption but it is impossible to say which variant was original. \%/ is preserved only in the FYY recension while the MSS. and all ether TTS show A . I'J J ^ BL 5b.6 (AL defective) The §•] of JL is reflected in the FYY text.. Shima seems to regard this as a late emendation. YZ lacks any word here, but the remaining TTS all have . Final ^ is found in all transmitted versions save the XE tradition. 71.©'i&&£^-tk^*7#£&a-tkjHFZ 396, KS 186, Ch. .63 Hfefr] AL (LZ 8) (BL defective) ~K7&]-k$ BL (LZ 43) (AL defective) Though both MSS. are defective for most of this passage, variation between ^ and 7o^ can be discerned. .7j^ seems to be by far the most common locative particle in the MWD witnesses of the Laozi; it even appears later in the same chapter in AL (5b.11 and 6a.1; BL is defective) and throughout Chapter 64 in both MSS. It is difficult to account for the sudden appearance of here. Both and 7s4 are curiously preserved in the FY tradition, producing the reading®f] ^A^7J^#&3 . Both 130 occurrences of ^ are omitted in YZ and XE. 72. ^"F^I&!-^'^#>1} HFZ 396, KS 186, Ch. 63 HM')*] $1. AL 5b. 11 (BL defective) The longer YL version of this passage is reflected word for word in the FY tradition. Most of the other TTS shorten this wording by omitting -k. . YZ omits ^.T^. and >yt* , but no witness agrees perfectly with the MWD version. 73. XT 4^***© HFZ 396, KS 186, Ch. 63, Tv^'^1] AL 5b. 11 (BL defective) The variation between YL, AL and the TTS parallels that of the previous example, except that TvT is here omitted from five XE texts as well as YZ. These XE witnesses all show 'K for&® . 7U.:£'3r H^T-dlJ 1) HFZ 400, KS 188, Ch. 64 AL is legible for the first sentence, including -tbi, and the characters H 1^. only. BL is defective throughout the passage. No TTS retains either instance of final attested in YL. 75.^*^^*flMn;tfi HFZ 404, KS 190, Ch. 64 % n% AL 6a.7 On this variant see Part One Ex. 7a.. is lacking in all TTS. 131 76. "^At 405, KS 190, Ch. 64 ^^J^^Hfi AL 6a. 8. Jit Af] ^ AL, BL 6b. 1 The TJT) of AL (lacking in BL) is found reflected as >X. in the FY text, appears to have been emended to in all but one of the XE witnesses. jl^ and the final ^ of YL are unattested in any other version of this passage. The locution ^(1^ occurs six times elsewhere in the MSS. .and TTS of Laozi, once each in Chapters 14, 16, and 52 and three times in Chapter 28. YL may possibly preserve the original reading here, theA^ of which must have been dropped very early from certain seminal witnesses. The variant is not semantically significant, but does provide another bit of evidence of the lack of contamination from the TTS in the HFZ witnesses._ 77. # HFZ 407, KS 190, Ch. 64 'l-^T ]3<L%fi AL 6a. 8, BL 6b.2. . * ] $ AL, BL. ^ -til] ^ AL, BL The YL variant'!-^ is unique as is the in the MSS.; all TTs show >yN^i. 7f in YL and the TTS is probably a taboo-substitute for^ . In this case % incorporates an object whose antecedent is . The optional final iiu of YL is retained in the FY tradition.. Three of the HSG texts show llj here which also probably refers back to ^*^0 78. >^ ^OT^M-^HFZ 861, KS 192, Ch. 65 T£] %V -fcv AL 6a. 10 , BL 6b. 4. ,M) ] AL This quotation appears in the 'Nansan' chapter of HFZ, but 132 as it is introduced by the phrase "^<£H , it is included in this collation. The second fa in the MSS. seems amenable to two interpretations; it may be an LC for >o or Laozi may possibly be punning deliberately here.on the use of £0 to mean >u . (There are examples of this usage in ZD, s. y. £p , def.. 17.) The of AL is unattested elsewhere, all texts showing the substitute ^. The final -tb of 'Nansan« and the MSS. is found in FY only. 79.£°^^ ^rjjH HFZ 952, KS 148, Ch. 44 This citation in the 'Liufan' chapter of HFZ is also explicitly acknowledged by the introductory 13 . BL is defective for all but the first two characters of Chapter 44, but the other witnesses, including AL, show no variation with the 'Liufan* version. 80.^^—^^^^^. HFZ 379, KS 196, Ch. . 67 ^ ]3x.4I AL7a.1, BL 7a.8.^ ]A% AL, BL. 44 ] BL The AL version of these two sentences reads which is a solecism. This corruption can easily be explained by assuming that the copyist's eye jumped from the AfL at the end of the first sentence to the same graph at the end of the second sentence, omitting the two intervening graphs ^ and "Si The ^ of JL finds its unique reflex in the FY text. >Y- is unique to the two MWD MSS. In five of the XE texts Is? has been transposed and appears as^l^^ . YZ, WB and all but one HSG text sh ow fa for the homophonous^f . ^ji is an LC for 133 81 HFZ 376, 377; KS 196, Ch. 67 ^J7\^ BL 7a.9 (AL defective). BL Initial 7v may have been omitted from the JL because this citation has been isolated from its context in Chapter 67 where 7^- functions as a pivot between the enumeration of the 'three treasures' and the explanation of their efficacy. (Cf. a similar case of the likely omission of in Ex. 22.) in BL for elsewhere is no doubt a copyist's error, Jj7^_ having been raised from the following sentence. 82.*^^*T>^^^^^^^ HFZ 378, KS 196, Ch. 67 The second occurrence of ^ is preserved in the FYY recension alone. The omission of this ^ in the other TTS changes the function of the following tfy from adjective (•completed, mature') to verb ('become'). AL agrees with JL in having ^ for the of BL and all the transmitted witnesses. This is another case where the two MSS. diverge significantly. 83 . HFZ 379, Ks 197, Ch 67 ^fffe] BL 7a. 12 (AL defective) . ^] ^ BL Again, initial 7v appears to have been omitted by HF as in Ex. . 85. , It is retained in all the TTS. 7& is unigue to JL and is probably an error for >A . Wang Xianshen remarks in his commentary on this passage (HFZ 379.2) that a pause should come after (7^)^. This sets ^ off as the topic of both parallel 134 four-character comments which follow: 'Through compassion, one will triumph in attack and be impregnable in defence' (Lau 1963: 12 9). Four XE texts and the FY tradition show ^ forlf^; one XE text has jj^p , a graphic variant of ^ . fjff *d'ien and •tian are synonym variants, ^ 'battle array' here used as synecdoche for^tf 'offensive warfare'. JE. appears in the FY texts and one XE witness for the found elsewhere. Another XE text hasiE^. Shima regards both ^/^i. and jE/5E^c_as emendations originating with XE which have contaminated the FY tradition. The remaining phrases in the JL parallel to the end of Laozi 67 are not acknowledged citations and hence.have been omitted from this collation. . As was noted repeatedly in the discussion of the data presented above, it is often impossible on the basis of the available evidence to determine the relative authority of two variant readings. The text chosen as a lemma in this collation is not itself an early MSS. but a witness whose earliest extant version is a printed edition dating from the Qiandao period (1165-1173) of the Southern Song, some thirteen centuries after the death of its reputed author. Because:of this enormous lacuna in our knowledge of the transmission of the HFZ, we can hardly insist on the hypothesis that all of its unigue variants, except those clearly attributable to scribal error, date from pre-Han times. It is certainly possible that none of them do.. On the other hand, the high incidence of agreement between JL, YL and the MWD MSS. in the use of particles would suggest that 135 the HFZ citations of the Laozi have escaped pervasisve corruption and survive today in a relatively unadulterated state, at least as far as substantives are concerned. Though we cannot determine with any certainty whether any of the unique variants in HF's quotations of Laozi are authoritative, the very fact that these unique readings have survived at all is surely further evidence that JL and YL are relatively free from contamination. The collation has failed to uncover any evidence of homogenization of the citations with the TTS such as can be discerned in the acknowledged quotations from Laozi found in the Huainanzi. The inference that can, I believe, be drawn from the collated data is that the instances of significant semantic variation found between HFZ and the MWD witnesses in Ex. 7 r 10 13 {^k/7r-) and 42 fc^/Bfc) as well as the cases of add-omission in Ex. 52, 72 and 73 and possibly the syntactic variation of Ex. 2 and 3 point to the existence in pre-Han times of separate textual traditions of the Laozi. It is possible then that HF was citing from a version of the text substantially different from that preserved in the MWD MSS. A second inference that it seems to me can be drawn from the data concerns the derivation of the TTS. In each of the four cases of semantic variation between the HFZ and the MSS. we found both variants reflected in the TTS.. This implies that, whatever the nature of the recension of the Laozi produced by Liu Xiang in the first century B.C., it may not have been the archetype of all the transmitted versions. Obviously, if likely pre-Han variation is preserved in the TTS, one can assume that 136 the lines of derivation for at least some of the TTS reach back beyond Liu Xiang to the discrete, pre-Han textual traditions which I believe the collation allows us to posit. The collation has also thrown some light on the special nature of the FY text, mentioned in the Introduction to this study. The many instances of this text's retention of particles (Ex. . 22 , 23, 58, etc.) and its occasional reflection of readings otherwise unigue to JL (Ex. 33, 40, 55 and 80) fully substantiate the Song description of FY as a recension based in part on pre-Han witnesses. . (The relevant passage, quoted in KS 10b, mentions YL, but curiously not JL, as one of FY's sources.) Though the result is a composite version and thus by no means infallible, the evidence makes it abundantly clear that FY is the most reliable of the complete transmitted witnesses.. Finally, some inconclusive evidence concerning the derivation of the two MWD MSS. has also been noted. Ex. 6, 19, 42 and 82 all contain what may be separative variants. Cases of conjunctive variation between the two MSS. and all other witnesses were also found (Ex. 7, 8, 11, 40, 41 , 77 and 80)., However, it would clearly be premature to attempt an interpretation of this data without collating AL and BL in their entirety. . 137 Notes to .Part. Two See: Encyclopedia Britannica, 15th ed., s.v. 'Textual Criticism' by E. J. Kenney for a clear exposition of the critical process. On the assumption that JL and YL were composed by HF himself, they can be dated to before 233 B.C-» the:year of his death according to the Shiji (63.15). The attribution of these two chapters as well as other Daoist passages in the HFZ to HF has been disputed by Rong Zhaozu (1927:99-101) who argues that because in the genuine 'Wudu' chapter •abstruse sayings' {^C^ieT ) are described as unsuitable for use in governing (HFZ 1058) , it is unlikely that HF would have written a commentary on the Laozi. That •abstruse sayings' necessarily refers to the Laozi is, of course, an arguable assumption on the part of Rong.. He recognizes only two of the extant 55 chapters of HFZ as genuine and suggests that JL and YL were interpolated by early Han Daoists.We have, on the other hand, Sima Qian's description of HF's thought as deriving from the doctrines of Huanglao ( Shiji 63. 14) . Since the Shiji was completed c. 91 B.C., to save Rong's hypothesis requires the assumption that Sima Qian was completely unsuspecting of the extensive and, from his standpoint, recent Laoist interpolations in the HFZ. The authorship and precise date of JL and YL is not, however, immediately relevant to our 138 present purpose, the collation of three early witnesses of the Laozi. The terms 'substantives' and 'accidentals' were first used by Walter Greg in his essay 'The Rationale of Copy-Text' (1951). For the concise definition of these terms I have followed the phrasing of Bowers (1950:13, 1962:1). 139 Bibliography Akai K iyomi -ifc ft *A % 19 74 Shotai_jiten. Tokyo: Tokyodo. Bowers, Fredson 19 50 'Current Theories of Copy-Text, with an illustration from Dryden'. Modern Philology 48. 12-20. 1962 'Established Texts and Definitive Editions'. Philological Quarterly 41. 1-1,7. Buck, David D.. 19 75 'Three Han Dynasty tombs at Ma-wang-tui •. World Archaeology 7.1.3 0-45. Chan Wing-tsit 1963 The Way of Lao Tzu,/Tap-teaching}. Indianapolis: Bobbs-Merrill., Chao Songwu ^"^5^ 1967 Junzhai dushuzhi. (Reprint of Wang Xiangian's 1884 edition) . Taibei: Guangwen shuju. , Chen Qiyou f$l»f«<£, ed. . 1958 Hanfeizi jishi. (Expanded and revised edition in 2 vols.). Shanghai: Zhonghua shuju.. Chen Yuan ±* 1928 Shihui juli. (196 2 reprint). Beijing: Zhonghua shuju. Dain, Alphonse 1964 Les Manuscrits. (Nouvelle edition revue)Paris: Societe d'e'dition 'Les Belles Lettres'. Ding Fubao J r ed. 1960 Shupwen jiezi gulin. (Reprint with Supplement of 1928 edition, 12 vols.). Taibei: Shangwu yinshuguan. Dickinson, Steve 1979 'Historical Implications of the Ma Wang Tui texts: Huanglao Taoism Reconsidered'. Unpublished seminar paper. . Dong Tonghe 'JfJ^]'^-19 67 Shangqu yinyunbiao gao. (Reprint of Gupli zhpngyanq yanjiuyuan lishi yuyan,yanjiusup,jikan 18.1-249.). Taibei: Lianguofeng chubanshe. Duan Yucai ^sL ±: *X (19 30 Shuowen jiezi zhu ( Wany.ou_wen.ku ed., facsimile of 1815 printing). Shanghai: Shangwu yinshuguan.. 140 Duyvendak, J.J.L. 19 54 Tao_Te-Ching; The Bpgk.:pf the Way and Its Virtue. London: John Murray. Encyclopedia Britannica. 15th ed. (1979). S.y. •Textual Criticism,' by E, J. Kenney. Housman, A. E. 1921 'The Application of Thought to Textual Criticism'.. Proceedings of the, .Classical. .Association 18, 67- 84. Hu Shi o^i^L 1943 'Lianghan linwen puhui kao'. Tushu iikan, new series 5. 1 1943a •Du Chen Yuan Shihui juli lun Han hui zhutiao'. Tushu iikan, new series, 5.1 Both of the above reprinted in Hu..Shi_ xuanjji:: kap ju. Taibei: Wenxing shudian, 1966. Hunansheng bowuguan, Zhongguo kexueyuan kaogu yan jiusuo yfy&Jt^tyife 1974 'Changsha Mawangdui er-, sanhao Hanmu fajue jianbao'., Wenwu 1974.7.39-48. Hurvitz, Leon 1961 *A recent Japanese study of Lao-tzu'. Monumenta Serica 20.311-367. Gao Heng T§I 5 and Chi Xichao Mb 1976 'Shitan Mawangdui Hanmuzhong de boshu Laozi '. Reprinted, revised from Wenwu 1974.11.1*7 in Mawangdui hanmu boshu zhengli xiaozu, Mawangdui Hanmu boshu Laozi, pp. 109-128. . Beijing: Wenwu chubanshe. Greg, Walter W. 19 51 'The Rationale of CopyText'. Studies; in., Bibliography 3.19-36. Henricks, Robert G. 1979 'A note on the guestion of chapter divisions in the Ma-wang-tui manuscripts of the Lao-tzu '. Early_Ghina 4.49-51. Jiang Yougao >x>^ 19 66 Jiangshi vinxue shishu. (Photoreprint of 1814 edition). Taibei: Guangwen shuju. Karlgren, Bernhard 1932 'The poetical parts in Lao-Tsi •. Gotebprgs Hogskolas Arsskrift. 38. 3. 1-45. 19 54 compendium of Phonetics in Ancient and_Archaic Chinese. Reprinted from BMFEA 26. 141 19 57 Grammata Serica Recensa. Reprinted from BMFEA 29.. 1964 Glosses-on the Book.of Odes. Reprinted from BMFEA 14 ("l942),~16 (1944)", 18 (1946). 1967 Loan Characters in Pre-Han..Texts. Reprinted from BMFEA 35-39 (1963-1967)". Kleinhenz, Christopher, ed. 19 76 Medieval manuscripts.,and textual criticism. Chapel Hill: University of North Carolina Department of Romance Languages. . Lau D. C. 1963 Lap Tzu.-Tap Te Ching. Harmondsworth: Penguin Books. Li Fang-kuei 1975 'Studies on Archaic Chinese'..Translated by G.L. Mattos. Monumenta Serica 31. 219-287., Li Xuegin ^ ^ife 19 79 'Mawangdui boshu jianyao jieshao'. Unpublished workshop paper. . Loewe, Michael A.N. 1977 'Manuscripts found recently in China: a preliminary survey'. T'ounq Pao. 63. 2-3.99^136. Lu Zhiwei jS^'*^ 1948 Shiyun-:pu. , Yanjing xuebao monograph 21. Lu Shu xiang E^K^ 1955 'Lun wu J yu wu £ ^ ]'. Reprinted in Hanyu yufa lunwenji, pp. 12-35. Beijing: Kexue chubanshe.. Luo Changpei^ an)j zhou Zumo 7°] Tit 19 58 Han Wei Jin Nanbeichao;yunbu_yanbian yanjiu. Pt.1. Beijing: Kexue chubanshe. Mawangdui Hanmu boshu zhengli xiaozu >i. ^ ^ % %L '^^S. 1974 Mawangdui Hanmu boshu_-Jyij,- Shanghai: Wenwu chubanshe. 1976 Mawangdui Hanmu boshu Laozi. Beijing: Wenwu chubanshe. Ma Yon g ^ 1972 'Daihou he Changsha chengxiang'. Wenwu 1972.9. 14-21. 1933 Gushu xuzi iishi (1962 reprint) Taibei: Guangwen shu j u. _ Pulleyblank, E.G. 19 62 'The consonantal system of Old Chinese'. Asia Major N.S. 9.59-114, 206-265. 142 1963 'An interpretation of the vowel systems of Old Chinese and of written Burmese*. Asia Major N.S. 10.200-221. 1965 'Close/Open ablaut in 14. 230-240. 1973 'Some further evidence regarding Old Chinese -s and its time of disappearance1. Bulletin of .the School of Oriental and African Studies, Uniyersiy..of_London 36. pt. 2. 36 8-373. 19 73a 'Some new hypotheses concerning word families in Chinese* . Journal of Chinese .-Linguistics 1. 1. 111-125. . 1979 'Linguistic notes on the Laozi manuscripts'. Unpublished workshop paper. Qian M u A% 19 56 Xiangin zhuzi xinian. Hong Kong: Hong Kong University Press. Bong Zhaozu ^ ^ 19 27 'Han Fei de zhuzuo kao' Zhong shan..da xue, yuyan ^lishi yanjiusuo zhoukan 1. 4. 94-106. Shen Qianshi >^Zi 1945 Guangyun shenqxi. (1960 reprint) .. Bei jing: Wenzi Gaige chubanshe. Shima Kunio 1973 Roshi kosei. Tokyo: Kyuko shoin. Shirakawa Shizuka ^ ''I'^fih 1974 Setsumon shinqi. Kobe: Hakutsuru bijutsukan. Shisanjing zhushu. (Photoreprint of 1816 woodblock edition.) Taibei: Iwen yinshuguan, 1976. Tang Lan M ^ et al. _ 1974 'Zuotan Changsha Mawangdui Hanmu boshu'. Wenwu 1 974.9-45-57.. Vinaver, Eugene 1939 'Principles of Textual Emendation'. Published in Studies in French Lanquaqe..and^.Medieayal_Literature presented to Professor.iMildred K. Pope_bY_Pupilsr  Colleagues and Friends, pp. . 351-369. . Freeport, N. Y.: Books for Libraries Press. Waley, Arthur 19 58 The_Way_and_It_s Power: A Study of -the Tao Teaching and Its Place in Chinese Thought. (Paperback edition). New York: Grove Press, Inc. 143 Wang Yanzhi .± §I^L 1798 Jinqdian shici. (1931 reprint). .Shanghai: Shangwu yinshuguan. . Xiao Han B^ 1974 'Changsha Mawangdui Hanmu boshu gaishu1. Wenwu 1974.9.40-44. Yan Lingfeng Ml If ^ 1965 Lap.Lie Zhuang sanzi zhi jian.shumu_ Xghang. pian_. Taibei: Zhonghua congshu pianshen weiyuanhui.. 19 76 Mawangdui boshu Laozi shitan. Taibei: Heluo tushu chub an she. Yang Shuda 19 55 Ciguan. Beijing: -Zhonghua shuju. Zhou Fagao J°l *i i»l 19 73 Hanzi. gu.jin yinhui: A pronpuncing di c ti ona ry of Chinese characters in Archaic and-Ancient_Ghinese#  Mandarin and Cantonese. Hong Kong: Chinese Oniversity of Hong Kong. Zhpngwen dacidian. (1st revised edition). Yangmingshan: Zhonghua xueshuyuan, 1973. Zhou Zumo J^l^aH and Wu Xiaoling ^ ^^k, eds.. 19 56 Fanqyan jiaojian ji, tonqjian. Beijing: Kexue chubanshe. . 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items