Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A study of the Shi Jing (The Book of Odes) manuscript excavated at Fuyang Rempel, Paul Edward 1999

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2000-0144.pdf [ 9.96MB ]
JSON: 831-1.0089418.json
JSON-LD: 831-1.0089418-ld.json
RDF/XML (Pretty): 831-1.0089418-rdf.xml
RDF/JSON: 831-1.0089418-rdf.json
Turtle: 831-1.0089418-turtle.txt
N-Triples: 831-1.0089418-rdf-ntriples.txt
Original Record: 831-1.0089418-source.json
Full Text

Full Text

A STUDY OF THE SHIJING WM. (THE BOOK OF ODES) MANUSCRIPT EXCAVATED AT FUYANG by PAUL EDWARD REMPEL B.A., The University of British Columbia, 1992 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 December 1999 © Paul Edward Rempel, 1999  In  presenting  degree  at  this  the  thesis  in  partial  fulfilment  of  University  of  British  Columbia,  I agree  freely available for copying  of  department publication  this or of  reference  thesis by  this  for  his thesis  and study. scholarly  or for  her  I further  purposes  Department  of  /fsiAfl/ Sllf0/££  The University of British Vancouver, Canada  Da,e  DE-6 (2/88)  O R r .  Columbia  that  be  It  gain shall not  permission.  requirements  agree  may  representatives.  financial  the  is  that  the  permission  granted  an  advanced  Library shall make  by  understood be  for  for  the that  allowed without  it  extensive  head  of  my  copying  or  my  written  Abstract In a general sense, this thesis examines several aspects of the Shijing HH (The Book of Odes) manuscript excavated at Fuyang #P£§, Anhui province, the People's Republic of China. The introductory chapter provides background on the Fuyang Shijing, as well as the transmitted versions of this Chinese classic. Further, it outlines the traditional scholarship dealing with the Shijing and offers a detailed summary of the other chapters. Chapter one introduces the methodology used in chapters two and three of this thesis. Part one of this chapter demonstrates the manner in which lexical and graphic variation is defined and analyzed, while part two shows how graphs may be analyzed based on a clear understanding of orthographic feature. Chapter one also provides background on the history of early Chinese writing forms and the evolution of Chinese script. Chapter two is primarily an orthographic study with two main objectives: (A) to distinguish the physical nature of the Fuyang Shijing; and (B) to date this manuscript. In the introduction to this chapter, I will also compare the Fuyang Shijing with other early bamboo and silk manuscripts. Chapter three deals with lexical variation involving the Fuyang Shijing and the various transmitted Shijing texts. It has three objectives: (A) to provide new readings of several odes in the Shijing; (B) to survey different types of lexical variation associated with the Fuyang manuscript; and (C) in each case determine which lexical variant, among the various possibilities, most likely represents the proximate original of the Shijing.  ii  Table of Contents Abstract  ii  Table of Contents  i tt  List of Abbreviations  v  Acknowledgements  vii  Dedication  viii  Introductory Chapter: Background on the Fuyang Shijing and Traditional Shij ing Scholarship; Thesis Arguments and Outline Chapter I:  A Survey of Methodology for Lexical and Graphic Analysis, and Orthographic Evaluation  Chapter IT.  Chapter III:  1  10  Introduction  10  Part One  10  Part Two  27  Conclusion  46  A Selective Graphological Study of the Fuyang Shijing  47  Introduction  47  Section 1  58  Section 2  73  Section 3  89  Conclusion  108  A Selective Lexical Study of the Fuyang Shijing Introduction  11.0 110  Lexical Variation Study no. 1  113  Lexical Variation Study no. 2  121  /'//  Lexical Variation Study no. 3  124  Lexical Variation Study no. 4  128  Lexical Variation Study no. 5  134  Lexical Variation Study no. 6  141  Conclusion  148  Thesis Conclusion  150  Bibliography  152  iv  List of Abbreviations BRI  Bronze inscriptions  DK J  Dai Kanwa jiten A ^ ^ D M ^ J I -  FH  Fenghuang shan MJffilll manuscripts, ca. 163-150 B.C.  FYSJ  Fuyang Shijing  GHYZD  Gu hanyu changyongzi  GSR  "Grammataca Serica Recensa"  GWZLB  Guwenzileibian  HYDCD  Hanyu da cidian  WSfeXWltfe  JGWB  Jiaguwenbian  EE|>pf^|§  ICS  The ICS Ancient Chinese Texts Concordance Series  JGWZGL  Jiaguwenzi  KMJS  Kokotsu moji jishaku soran  JM.TWCD  Jianmingjinwen  JSZD  Jinshi zidian  JWB  Jinwen bian  JY  Juyan f_3~$_ wood tablet manuscripts, ca. 125-104 B.C.  LZB  Lizi bian  MWD  Mawangdui J§_ift manuscripts, ca. 195-187 B.C.  J^|@  gulin  WM. '^Mm'Mffl^^Jft  zidian  ^#5C^|^#  cidian  ^  ^j^Jt^^Wi&W, __5Cf I M  &$tWb  v-  OBI  Oracle-bone inscriptions  QGG  Qin Gong gui  QJWZB  Qin jian wenzi bian ^1Sf  SD  Shigu wen 7SM.~5t stone inscriptions, ca. fifth-century B.C.  SHE)  Shuihudi H l ^ i f e bamboo manuscripts, 240-217 B.C.  SKZWB  Shi ke zhuanwen bian  SWJZ  Shuowen jiezi MXM^,  SWJZZ  Shuowen jiezi> zhu WtStM^fe  YQS  Yinque shan  ifUfeli! bamboo manuscript, ca. 140 B.C.  ZCW  Zu Chu wen  ti^ctjt  bronze inscriptions, Spring and Autumn period  ^EM^J(M ca. 100 A D .  stone inscriptions, ca. third-century B.C.  vi  Acknowledgements I am very grateful to my supervisor, Professor Ken-ichi Takashima, for his encouragement and support. This thesis would not have been possible without Professor Takashima's carefully considered guidance, always timely counsel and the valuable recommendations and corrections he offered on the drafts of my thesis. Over the years, I have also benefited a great deal from the advice of Professor Jerry Schmidt. Furthermore, a word of thanks is due Professor Daniel Overmyer who was of great assistance in pointing out problems of wording and making the chapters in this thesis work together as a cohesive whole. Next, I would like to thank Dr. Sun Jingtao for advising me on questions concerning Old Chinese phonology, and Dr. Colin Hawes for his views on good study habits and effective writing. Colin's maturity as a scholar and upstanding character have given me a great deal of direction during the course of my M A . program. Other colleagues and faculty members who have played a part in the realization of this thesis include: Dr. Susan Kong, Paul Crowe, Wu Keying, Norman Smith and Mary Yeung. Finally, I would be remiss not to thank Kathleen Wyma for proofreading this thesis and helping me maintain a sense of equilibrium through the years.  vii  This thesis is dedicated to the memory of WilhelmRempel  Surely here was charm beyond compare to view Maybe it was just that I was there with you  viii  Introductory Chapter: Background on the Fuyang Shijing and Traditional Shi jing Scholarship; Thesis Arguments and Outline Introduction:  The 1970s was a period of immense significance for Chinese archeology. During this time, archeologists discovered numerous ancient manuscripts that are of tremendous importance. For example, at Yinque 'Silver Sparrow' Mountain I S H l i l , Shandong province, numerous bamboo texts were unearthed in April, 1972 (Zheng: 23; Yates: 75-6). Of the manuscripts uncovered here, the most celebrated is the Sun zi bingfa J l i ^ f t (Sun zi's Art of War); however, several other documents related to important historical figures were also excavated from this site (Yates: 75-6). Moreover, at Shuihudi fHi^ife 'Sleeping Tiger Place', Hebei province, a large collection of bamboo inscriptions written prior to the Qin dynasty (221-207 B.C.) was unearthed in December, 1975 (Xu Fuchang: 1; Shuihudi hereafter referred to as SHD). The Shuihudi corpus contains texts written on subjects such as incarceration and corporeal punishment, arrest procedures and demonography, as well as Qin law in general (Xu Fuchang: 15-120; Harper: 459-98). Probably the greatest archeological discovery made in China during the 1970s occurred atMawangdui |§3ii|£, Hunan province, in December, 1973 (Zheng: 21; Shaughnessy, 1994: 47; Henricks, 1989: xii; Mawangdui hereafter referred to as MWD). At this site, documents of great importance to Chinese philosophy, literature, history, art, medical theory and scientific thought were unearthed — e.g., the earliest manuscript copy of the Yijing (The Classic of Changes), two copies of the Dao de jing MM-M. (The Classic of the Way and Its Virtue), an early version of the Zhan guo ze ^MW- (The Intrigues of the Warring States), and what may be the long lost Huangdi waijing ft ffj^f-M (The External Classic of the Yellow Emperor), etc. (Shaughnessy, 1994:46; Henricks, 1989: xiii-xiv). The research done on these excavated manuscripts has led to the publication of many books and articles of interest both to specialists and the general public (works in English include: Ames, 1993 and 1998, Henricks, 1982 and 1989, Silbergeld, Shaughnessy 1994 and 1996, Yates, Wu Hung and L i Ling; for works in Chinese see: Li Meili, Fu, Wu Jiulong). A discovery similar to those 1  above was also made in Fuyang county, Anhui province, 1977. In this case, the documents unearthed include an early compilation of Annals (nian biao  (Hu, 1989: 1-26), the Zhou  yi MM (The Zhou Book of Changes) and a fragmented bamboo manuscript containing the Shi jing  (Hu, 1988: 1). It is the latter manuscript which is the subject of this thesis. Background on the Fuyang Shi jing and Traditional Shi jing Scholarship:  Excavated artifacts and inscriptional evidence indicates that the Fuyang Shijing (hereafter FYSJ) belonged to the tomb of Xia Houzao WM'kk who was buried in 165 B.C. (Wen: 60; Hu, 1988: 1). Unfortunately, by the time Xia Houzao's tomb was unearthed by Chinese archeologists in 1977, it was already in poor condition. At some point the tomb's coffins collapsed, breaking open the decaying lacquer trunk in which the bamboo materials had been held. As a result, the documents were scattered bent and twisted inside the tomb, and many became black and rotten from exposure (Hu, 1988: 1). Hu Pingsheng'^^p _ and 1  Han Ziqiang ^ g 5K spent considerable time and effort sorting through the bamboo fragments and reconstructing the FYSJ (Hu, 1988: 1-2). In total, the FYSJ is comprised of more than one hundred and seventy bamboo fragments which vary considerably in length. Some of the shorter bamboo slips contain only one or two graphs, while the longest one has twenty-four. The vast majority of FYSJ slips contain verse from the Guo feng fjgJH ('Airs of the States') section of the Shi jing, but four odes from the Xiaoya yjvjl ('Lesser Ya Odes') section are also represented (Wen: 60,69). (According to Baxter, p. 355, the Guo feng "are generally regarded as folk songs collected from disparate geographic areas of the Zhou kingdom"; regarding the Xiaoya, Baxter writes: "These poems are generally assumed to be a product of the royal Zhou court. In content and style, however, there is some overlap between the Guo feng and Xiaoya sections.") Exactly how the FYSJ figures into the tradition of Shijing scholarship established in the Han dynasty (206 B.C.-220 A.D.) is of great interest to researchers. In the Han there were numerous schools of Shijing scholarship, each of which had its own primary text and exegetical commentary (Baxter: 357; Van Zoeren: 84). Two schools were named after the geographic areas they originated from: the Lu H school, which was the most influential Shi 2  jing school in the Western Han (206 B.C.-23 A.D.); and the Qi | f school associated with Yuan Gu H ® (Van Zoeren: 85; Wang Jingzhi: 20). Both of these traditions, with their texts and accompanying commentaries, all but disappeared during the Six Dynasties (222-587 A.D.) (Karlgren, 1942: 71; Wang Jingzhi: 20). Only traces of them remain in the form of citations located in various early texts. Another Shijing school in the Han dynasty was named after its founder, Han Sheng f$-£ of the state of Yan  (Baxter: 357; Van Zoeren: 85).  This school survived until the Tang dynasty (618-907 A.D.) and remnants of it can still be found in the Han shi wai zhuan ItH^ff^ (The Exoteric Commentary on Han's Odes) (Karlgren, 1942: 71; Wang Jingzhi: 20; Hightower). The Lu, Qi and Han schools are known collectively as the sanjia shi  or, as they will be referred to in this thesis, Three  School's Shi. We are much indebted to Qing scholars (1644-1908 A.D.) for extracting remnants of the Three School's Shijing from where they lay embedded in different textual sources (Karlgren, 1942: 71). The Mao ^school of Shijing (hereafter Mao Shi) scholarship also existed during the Han dynasty. This school likely takes its name from either Mao Chang  who taught the  Shijing at a provincial school during the end of the Western Han dynasty, or Mao Heng 3_ believed to have been a student of the philosopher Xun zi ^j^p (born ca. 312 B.C.) (Saussy: 51). Although it was not initially the most dominant school, by the time of the Eastern Han dynasty (25-220 A.D.) the Mao Shi had already became the most authoritative version of this classic (Karlgren, 1942: 71). In the beginning, the Mao zhuanflf(Mao Commentary) (ca. 150 B.C.) was the only exegetical text associated with the Mao Shi. However, several centuries later Zheng Xuan  (127-200 A.D.) wrote the jian H (The  Notes) which interprets the Mao Shi in a way that often differs from the zhuan (Baxter: 357: Karlgren, 1942: 72). Zheng Xuan's great reputation as a scholar played a pivotal role in ensuring the longevity of the Mao textual and exegetical tradition. For the glosses attributed to the Three School's Shi and Mao exegetical texts, I rely primarily on the works of several eminent Qing scholars — i.e., Wang Xianqian EE^cM, Ma Ruichen | § i i f § and Duan Yucai | § A t the same time, I will also consult Wen Xingfu 3t (61-91) and Hu Pingsheng (1988,2-22). 3  Thesis Arguments and Outline: In recent years, the FYSJ has received some attention from Chinese scholars (Hu, 1988; Wen). On the whole, however, the research on the FYSJ written in Chinese is incomplete and suffers from a lack of thoroughness. Since I delve into past FYSJ research later on in the second and third chapters, only a few words on this matter are required here. For example, the orthographic analysis previously conducted on the FYSJ refers to only a few graphs and fails to mention anything about the graphological methodology used (Wen: 97-8). The paucity of data and lack of methodological consideration seriously detract from the scholarly value of this research. As such, Wen's assertions regarding the date of the FYSJ and its overall orthographic make-up are tenuous at best. Further, while Hu Pingsheng's efforts to analyze lexical variation in the FYSJ has successfully shown that this manuscript occasionally uses different words from the transmitted versions of the Shi jing (Hu, 1988: 25-6,2-18, 36-88; Wen: 61-90), his research also fails to make methodology a concern, and no formal account of it is given. Since the study of lexical variation in pre-Han Chinese is largely reliant upon knowledge of Old Chinese phonology — which is notoriously complex and subject to different interpretations — the question of methodology is an important one (Baxter: 1-5; Pulleyblank, 1995: 5). As far as research on the FYSJ in English is concerned, the situation is more dire still: to date no meaningful study has been made of this manuscript in the West, and there is as yet nothing of consequence written on it in English. For example, William Baxter simply mentions the FYSJ in passing (357); whereas, Steven Van Zoeren writes only that the FYSJ fragments dealing with exegesis differ from any known Shi jing school and likely represent a previously unknown regional variation (86). To meet the need for a specialized and in-depth study of the FYSJ, I present this thesis. My analysis of the FYSJ will be in three parts: (A) Chapter one will provide background on the pre-Han writing system and survey early Chinese script. As well, it will outline the methodology that I employ in my study of the FYSJ and lexical variation; (B) In chapter two, I will conduct a graphological analysis of the FYSJ using a methodological approach which stresses orthographic feature and context. The objectives of chapter two's graphological study are to date the orthography found in the FYSJ, and discover the extent to which this composite manuscript is comprised of different source manuscripts; (C) In the 4  1  third chapter, based on a study of lexical variation, I will offer an alternative reading of several verse-lines in the Shijing. Moreover, I will analyze the types of lexical variation encountered when comparing the manuscript and transmitted versions of the Shijing, and attempt to address the question of the proximate original. All of these tasks will be undertaken using carefully laid out forms of lexical/graphic and textual methodology that is applied systematically and consistently. Below I offer a more detailed summary of the three main chapters comprising this thesis: In chapter one, I outline the different stages of the pre-Han writing system: e.g., the pictographic, multivalent and determinative stages. I also introduce the different types of characters that the non-specialist reader will need to know in order to follow my analysis of lexical variation: e.g., phonosemantic compounds, cognates, tomosomatic and monosematic graphs, as well as polysemous and paronomastic characters, etc. Some of these characters have a very salient role in chapter three; whereas others receive only passing mention, and are more important for the underlying context that they provide than the actual arguments themselves. In the first chapter, these character-types are all defined and their relation to the various stages of the early Chinese writing system explained. Moreover, I also engage in a detailed discussion of the phonetic criteria that are crucial for the identification and evaluation of lexical, graphic and cognate variation. This criteria is based on Karlgren's reconstruction of Old Chinese, which, although somewhat dated, is still widely referred to by phonologists, linguists and non-specialists alike (Baxter: 3-5; Pulleyblank, 1992: 365). Karlgren's Old Chinese reconstruction is especially useful to my study, since it is easily accessed in his monumental study of pre-Han language, The Grammataca Serica Recensa (Karlgren, 1957; hereafter GSR). The approach that I employ to utilize Karlgren's Old Chinese, however, is founded both on phonological studies by L i Fang-kuei $ 7 3 £ | and methods of application proposed both by L i and others such as William Boltz (1984,1994,1997). It is in this context that matters such as the significance of homorganicity in initials and finals are introduced. In the second part of chapter one, I overview the evolution of Small Seal (xiao zhuan ^MS) and Ancient Clerical (gu li ^j#) script, two forms of script which are central to my graphological analysis of the FYSJ. A basic knowledge of early Chinese script is essential for 5  appreciating the orthographic features embodied in the Small Seal and Ancient Clerical scripts; consequently, in the first chapter I also survey the development of Chinese script starting as far back as the Shang dynasty. (The traditional dates for this dynasty are 17661122 B.C.) This survey will incorporate an explanation of how scripts changed as writing in early China developed, touching upon topics such as simplified and complex writing forms, pictography and symbolification, linearization and the advent of vertical rectangularity. In chapter two, I also introduce the various sources used to analyze orthography in the second chapter — e.g., oracle-bone, bronze and stone inscriptions, as well as the silk and bamboo documents unearthed at the SHE) and JVTWD sites. To make my presentation of orthographic methodology more meaningful, in chapter one I apply it to an analysis of six sets of graphs written in different forms of script (a total of sixty-five graphs). The practice of dividing graphs into their component parts and examination of orthographic feature that I follow when treating these graphs, is also implemented in chapter two. The selective graphological analysis of the FYSJ that I conduct in chapter two is based primarily on the methodology outlined in the first chapter. However, in the introduction to chapter two, I also discuss several aspects of methodology in greater detail. For example, I overview the manner in which orthographic feature and style can be used to categorize and date graphs written specifically in Small Seal and Ancient Clerical script. Moreover, I delve into how two levels of context can be utilized to accurately ascertain the true physical nature of the FYSJ. Since I believe it is important for the reader to be familiar with the physical nature of early Chinese manuscripts in general, as well as the sources which play the most important role in the second chapter's orthographic analysis, the introduction also contains a brief comparison of SHE), MWD and FYSJ manuscripts. Several sets of examples are given to underscore the assertions resulting from this comparison. Chapter two contains three separate graphological studies, each of which is divided into four parts. The first part involves a detailed componential analysis of the primary graph featured in each of the graphological studies. This part is devoted to describing the various elements in the feature graph, so that the later application of orthographic and calligraphic criteria is more easily understood. After all, I believe clarity is the foundation upon which effective methodology is based. An orthographic survey of the feature character is given in part two of each study. This initial survey is required to establish reference points in the 6  form of orthographic features with which to date the feature characters. However, it should be noted that similar orthographic surveys will be conducted whenever it is necessary to procure a date for a specific graph, something which occurs frequently in chapter two. The third part also focuses on what I refer to as the feature character. But in this case, I compare orthographic and calligraphic variants of the feature character to ascertain the degree of formal variation that exists in the script of the FYSJ. This exercise is also integral to mapping out the textual space of the FYSJ, and formulating a time-frame in which to situate it. Finally, the primary objective of part four is to examine the feature character in its immediate and extended orthographic context. The goal here is to date the FYSJ bamboo slip on which the feature character is written, in addition to the slips which surround it. In the process of doing this, I hope to isolate several of the source manuscripts embedded within the composite manuscript of the FYSJ. The selective lexical study conducted in chapter" three of this thesis would not be possible without the groundwork laid out in chapter one. To fully realize the term 'lexical variant', it must be juxtaposed against other interrelated phenomena implicit in the same linguistic environment. That is, to clearly answer the question "What is lexical variation?" a distinction in terms of both definition and application must be made between it and other forms of cognate and graphic variation. The six separate studies comprising the third chapter assume that all these forms of variation are familiar to the reader. Each of these studies opens with a summary of the FYSJ slip containing the lexical variant that is to undergo analysis. Here I discuss which characters are written on the slip, as well as how many characters there are. All instances of graphic variation will be briefly elaborated on here, and the general state of the bamboo slip will be described. Following this I will compare the FYSJ lexical variant with its counterparts in the various transmitted versions of the Shijing. This preparatory work is essential because the status of a lexical variant is often reliant upon its immediate environment. After the lexical variation featured in each study is identified, I proceed with the main body of the study. Naturally, this entails the application of methodology discussed in the first chapter. But it also involves a philological component; that is, the citation of early Chinese textual sources to substantiate the claims that I put forth. At the conclusion of each lexical study in chapter three, I attempt to discern which of the lexical variants discussed is 7  the proximate original. According to some, the underlying premise of this exercise might entail the existence of a prototypical Shijing text some time in the distant past. If this were true, the FYSJ and the various transmitted versions of the Shijingmentioned earlier in this chapter — the Mao Shi and Three School's Shi could all be descendants of this hypothetical original. However, I do not address the question of a Mr-text for the Shijing in this thesis. Instead, when attempting to find the proximate original, I deal with individual words, not an entire text. On this topic, Blanford writes: "The proximate original is the one variant that has a greater probability of representing the original word than does any other variant, while recognizing that the determination of the original cannot always be achieved with full confidence" (188). Various forms of methodology for determining the proximate original of a given text have been devised by scholars working in the field of textual criticism. They include the lineage method and the principle of the "more difficult reading," which have already been applied to the manuscript and received versions of the Dao de jing and the Zhan guo zonghengjia^MW^MM  (Documents of the Strategists in the Warring States) (Blanford;  Boltz, 1997, 1984). The introduction to chapter three provides a more detailed profile of these forms of textual criticism. Because the analysis conducted in chapters two and three are completely separate from one another, each of these chapters has its own separate conclusion. My conclusions basically summarize the discoveries made in the chapter and the methods used to reach them. Moreover, the conclusions also give me the opportunity to explain how my study of the FYSJ contributes to the previous body of research conducted on this manuscript. To conclude this introductory chapter, a few words regarding my use of the term 'manuscript' in connection with the FYSJ are in order. Earlier in this chapter, I refer to the FYSJ as a composite manuscript and suggest that it is probably comprised of numerous different manuscripts collected together. As such, the FYSJ can hardly be considered a manuscript in the sense that it represents a single textual version, or one distinct Shijing school. At the same time, to refer to the FYSJ as a group of texts or manuscripts like the MWD and SHD would be inaccurate. As I demonstrate in chapter two, the FYSJ is comprised of numerous bits and pieces of different manuscripts written during different historical periods in early China. At best these fragments represent manuscripts which were at one time whole and complete. But the FYSJ can still be considered a manuscript in the 8  sense that it contains only Shi jing materials and was excavated from a single archeological site. It is in this sense that I hope the term 'manuscript' to be understood when associated with the FYSJ in this thesis.  9  Chapter One: A Survey of Methodology for Lexical and Graphic Analysis, and Orthographic Evaluation Introduction:  As mentioned in the introductory chapter, this chapter outlines the methodology needed to conduct a lexical analysis of the FYSJ and its transmitted counterparts, the Mao Shi and Three School's Shi. Whatever background is needed to understand this methodology and become familiar with the early Chinese writing system will also be provided. In addition, in the first chapter I will also study the background of early Chinese script and orthographic development. The effort expended in the course of this task will concentrate mainly on bringing together orthographic criteria that can be use to date the FYSJ, and analyze its physical nature. For the sake of clarity and convenience, chapter one is divided into two main parts: the upcoming first part deals with the study of words, graphs and phonology, while the second part is devoted to script, orthographic form and feature. Part One:  The earliest stage in the history of Chinese writing centered around the xingyizi 0 | g ^ 'pictographic character.' According to Qiu Xigui  this type of character conveys  meaning through its graphic shape or appearance (Qiu: 42). Since every pictograph represents one or more words, we may refer to early Chinese as a logographic writing system. It is important, therefore, to divorce Chinese characters from the notion that they functioned ideographically, as is often claimed; pictographs record actual words, not mere ideas (Keightley: 108). Although I will not offer a detailed account of such phenomena here, it warrants mention that the severe limitations implicit in pictorial script necessitated the development of other variant writing forms (Boltz, 1986: 424; Keightley: 108; Norman: 5960). Faced with the difficulty of conveying words for abstract concepts through pictographic representation, the early Chinese developed devices such as pictorial symbols. Two common examples of this are shang _L , used to represent the word for 'above', and xia f which stands 'below' (Norman: 59; Keightley: 108). Furthermore, along side pictographs and pictorial symbols, a relatively small number of early Chinese characters functioned as what 10  one writer calls "purely arbitrary signs" (Norman: 69). An illustration of this device is the graph for 'seven' qi -_ (Norman: 60). To acquaint the reader with early Chinese pictographs, I provide several examples of this writing form below. I feel that showing what archaic Chinese pictographs look like is preferable to describing them with words. However, just because a pictorial quality is implicit in the graphs below, does not mean what they refer to is always immediately clear. Some pictographs are easier for the modern eye to discern than others, as what may have been clearly pictorial to the ancient Chinese may not seem so to us. But even when the pictograph is not visually comprehensible at first, a simple explanation will usually suffice to evince the picture implicit in it. My examples are taken from the Jia gu wenbian ^ # 3 t $ f (A Compilation of Oracle-bone Characters; hereafter JGWB), a work comprised of characters from late Shang oracle-bone inscriptions (hereafter OBI). (The late Shang period is traditionally dated from the reign of King Wuding g;T ca. 1300 B.C.)  (1) Of all the examples above, this picture is probably the most abstract and, therefore, most difficult to discern visually. William Boltz believes this pictograph (JGWB: no. 411) is a representation of an animal 'pelt' (1986: 427). This view accords with Bernhard Karlgren, who writes, "The graph is [...] a drawing of fur (a tail?)" (GSR: no. 1066). But the semantic ambiguity associated with this pictograph allows other readings, as well. Example (1) has been interpreted as a drawing of a wild animal and, along with several other paleographers, Qiu Xigui asserts it was originally a pictographic depiction of a 'caterpillar' (Jia gu wenzi gulin W^W5C¥-ttW- no. 1540; hereafter JGWZGL). Regardless of what it may have originally depicted, the modern graphic equivalent for this pictograph is believed to be qiu 3j< (Kokotsu mojijishaku soran tP'&X^^fe&M'- no. 411; hereafter KMJS). (2) This pictograph (JGWB: no. 463) is the OBI form of mu g . The GSR refers to this graph as a drawing of an 'eye' and its appearance hardly permits a different interpretation (no.  1036). In the JGWZGL there is considerable consensus for this interpretation. For example, one scholar writes, "[This pictograph] resembles an eye which is wide open" fepffi  WSSBR  (JGWZGL: no. 601). (3) This pictograph (JGWB: no. 577) stands for the word 'winnowing-basket' (GSR: no. 952; JGWZGL: no. 2815). Until the first century A.D., the word 'basket' was still represented solely by the graph qi S ; however, afterwards the semantic determiner zhu was added to the character for this word (JGWZGL: no. 2815). I will explain the function of semantic determiners later in this chapter. (4) In regard to this pictograph (JGWB: no. 682), Karlgren writes, "The graph is a drawing of a high building" (GSR: no. 1129). Xu Shen's f f H (ca. 100 A.D.) interpretation accords with this view: "[The pictograph] resembles the tall shape of a viewing tower" (Shuowen jiezi zhu  WiStM^-^' 5:24b; hereafter  fk skW,lii£-ffi> 1  SWJZZ). Of the nine pictographs seen  above, gao ]tj is the only tomosomatic graph, or teknogram — i.e., a graph which is analyzable into graphemes (Boltz, 1994: 102). Such a character is traditionally known in Chinese as az/  The other eight pictographs featured here are monosomatic; that is,  irreducible graphemes "that typically function as graphic constituents in other [complex characters]" (Boltz, 1994: 101). Monosomatic graphs are traditionally known in Chinese as wen %, and were typically referred to as duli zi M f ! ^ 'single-body character' by Qing philologists (Boltz, 1994: 101-2). (5) According to Karlgren, this pictograph (JGWB: no. 698) is the archaic form of lai 'wheat plant' (GSR: no. 944). Not only is this view substantiated in the Shuowen jiezi WtJL (An Explanation of Simple and Complex Graphs; hereafter SWJZ) (ca. 100 A.D), but among paleographers studying OBI there is also widespread support for this interpretation (SWJZZ: 5:33b; JGWZGL: no. 1507). (6) As in the case of (3), the word originally represented by this pictograph (JGWB: no. 1151) is indisputable. The meaning of xiang  'elephant' is clearly conveyed through this 12  graph's depiction of a long trunk, tusks and a tail (JGWZGL: no. 1653). (7) This pictograph (JGWB: no. 1152) represents the word for 'horse' ma ,f§ (GSR: no. 40; JGWZGL: no. 1630). The SWJZ states: "[This pictograph] depicts the shape of a horse head, a tail of hair and four legs" &%WLW&fcZ&  (SWJZZ: 10: la).  (8) This pictograph (JGWB: no. 1356) represents the word 'rain' yu pjf (GSR: no. 100; JGWZGL: no. 1100). In the SWJZ, Xu Shen writes, "[This is a pictograph] of rain water coming down from the clouds. The — [uppermost horizontal line] resembles the sky; while the [component of the graph shaped like] n resembles clouds" j f e K ^ S T "fe~'^JKV\  fell  (SWJZZ: 11:10b). (9) This pictograph (JGWB: no. 1694) is a drawing of a head with hair and "a rudimentary body with two arms." It represents the word for 'child' zi =f- (GSR: no. 964; JGWZGL: no. 590). The manner in which I analyze the graphs in examples (1) to (9) is used for various purposes throughout this thesis. To my mind, whether for the purpose of orthographic analysis or the study of lexical variation, it is crucial that the graphs dealt with be clearly described, their various components labeled and explained. This exercise is intended to make the reader more familiar with the subject-matter itself — the graphs, as well as better understand the analysis and arguments I put forth in this paper. For example, this approach allows me to pinpoint parts of a graph that demonstrate a particular orthographic feature. Further, a careful graphic profile is also helpful if I want to apply graphic evidence to the discussion of semantic value. To make Chinese a more effective writing system, greater emphasis was eventually placed on the phonetic and semantic value of the words themselves. This transformation in the Chinese language occurred during the Shang dynasty and is referred to as the multivalent stage. During this stage individual graphs became "invested" with semantic plurality and could represent more than one word (Boltz, 1986: 426; 1994: 101). In examples (1) to (9) above, there are several Shang OBI graphs which manifest semantic plurality: yu p§ (8) 13  represents the noun 'rain,' as well as the verb 'to rain' (KMJS: no. 1356); and mu @ not only represents the word 'eye', but also the word 'to see' (JGWZGL: no. 601). Even in antiquity, however, Chinese graphs had a fixed phonetic value; therefore, it was imperative that a polysemous graph represent words that were either fully or nearly homophonous. Phonology is crucial to the textual analysis of the FYSJ presented in this study, and its significance in the overall structure and development of pre-Han Chinese cannot be overstated. Phonology plays an integral role in understanding the relationship between polysemous graphs and the words they represent. But what exactly are the phonetic constraints governing the match up of polysemous graphs and different words? Any attempt to answer this must be based on the phonological values that have been reconstructed for Old Chinese. (According to Baxter, p. 14, Old Chinese is "the Chinese language of the early and mid Zhou dynasty." Pulleyblank, 1995, p. 5, defines Old Chinese as the archaic language "relevant to a period terminating around -600 [B.C.]." For the above question to be properly addressed in linguistic terms, it is best divided into two parts. We must examine the pronunciation of polysemous characters Mkcyu M and mu @ in terms of their initials and finals. In regard to finals, for different words to adopt the same graphic representation, they must have identical main vowels and homorganic final consonants. Referring to two or more words as homorganic means they are articulated at the same point in the vocal tract — e.g., they are both/all bilabials, dentals or velars (Boltz, 1994: 91,93,100; 1997: 255). For example, since the [+voiced] [*g] and [-voiced] [*k] are both velar sounds produced from the soft palate at the back of the mouth, they can be considered homorganic. Similarly, because [+voiced] [*b] and [-voiced] [*p] are both pronounced through the closure of the lips, we also label them homorganic. The rhymes found in the Shijing can be used to test Old Chinese finals for homorganicity. This is possible because words belonging to the same Shijing 'rime group' yu bu HFIIS possess phonetically compatiblefinals(Boltz, 1994: 91). The practice of categorizing characters from early Chinese texts into Shijing rime groups started during the Qing dynasty. Initially, scholars believed fewer than ten rime groups existed; however, over time the groupings became increasingly refined so that eventually over twenty were recognized (Li Fang-kuei: 2). An example of how this categorization works can be given using 14  the character feng ]%. Since the Shijing indicates that feng JH, rhymes with words in the qin jg> rime-group such as xin  Un # and>7M  itf,  it can also be assigned to this group (Chen Fuhua  W$M'- 315-21). Although other methods of determining the Old Chinese rhyme suggest a different grouping for feng JH, conclusions based on the Shijing are usually deemed most reliable by virtue of this text's early origins (Li Fang-kuei: 3). Finding a phonetic principle to deal with initials is more complicated. Our present understanding of the Old Chinese pronunciation of initials is not as certain as it is for finals; therefore, any theory reconstructing their phonetic value must be applied with caution (Boltz, 1997: 255). Because of evidence which suggests placing constraints on the initials of words affiliated with a common polysemous graph is unnecessary, some recent studies have discounted any views to the contrary (Boltz, 1994: 93). Nevertheless, an analysis of writing in early China is impossible without some kind of methodology dealing with the phonetic value of Old Chinese initials. Consequently, I will utilize the principle stating that phonetic compatibility for word initials should also be based on homorganicity. This principle has the approval of numerous Chinese epigraphers and should prove suitable for the purposes of this paper (Mattos, 1988: 127-8; Boltz, 1997: 255). My treatment of polysemous characters based on the above phonetic principles is derived from phonological theory regarding xiesheng f|fl£ 'shared phonophoric' characters originally propounded by Bernhard Karlgren. Karlgren's beliefs regarding Old Chinese phonology stressed the importance of the "laws of phonetic alternation," which he based on an understanding of "bare" phonetic elements — i.e., what I have referred to as polysemous graphs (1963: 3-6; Mattos, 1988: 125-7). The approach I have adopted also reflects Li Fangkuei's later revision of Karlgren's work: both principles outlined above are implicit in Li's reconstruction system (9-12). Before going on, the bearing that the aforementioned phonetic criteria has on the study of lexical variation warrants some elucidation. It is true that in part one of the first chapter, lexical variation is mainly implied — i.e., if two characters are phonetically incompatible and semantically dissimilar, but both suit a common context, we may have a case of lexical variation and not something that is simply graphic in nature. As such, this chapter begins to treat lexical variation by defining it in juxtaposition with various forms of graphic variation. 75  But the phonetic criteria underlying most of the methodology in chapter one has a much more direct bearing on the arguments about lexical variation found in chapter three. Chapter three contains numerous instances where phonetic criteria involving initials and finals is crucial in explaining the lexical variation that occurs. For example: in Lexical Variation Study no. 3,ju and ju % are lexical variants, but they must be phonetically compatible to fit the rhyme sequence governing the verse-line in which they are situated; in Lexical Variation Study no. 4, consideration of phonetic criteria and semantic value is necessary to show that tongflfand tong % are lexical variants; and in Lexical Variation Study no. 6, phonetic compatibility plays a part in both establishing  H and // H as lexical variants, and showing why they are  matched up with one another. More examples such as these will be encountered in chapter three. How do the rules dictating phonetic compatibility apply to the polysemous characters mu @ 'eye'/'to see' and>-w M 'rain'/'to rain'? According to the GSR, the words in each pair are completely homophonous: the words 'eye' and 'to see' are both assigned the Old Chinese pronunciation *mjok; whereas, 'rain' and 'to rain' are both pronounced *gjwo (1036,100). (Throughout this study I have converted Karlgren's outdated form of phonetic transcription into IPA — see Pullum.) As such, both pairs satisfy the phonetic criteria regarding initials and finals outlined above. But there is more to the analysis of mu g and yu 35 than just pronunciation; the semantic qualities of these two polysemous characters also deserve our attention. The semantic affinity of the words represented by the graphs mu g and yu pff suggests the possibility of a cognate relationship. Providing we accept the existence of cognate word sets in pre-Han Chinese, both 'eye' and 'to see' certainly may be viewed as derivatives of a root-word pronounced *mjok. The words 'rain' and 'to rain' bear a similar relationship to one another; that is, both seem to be semantic derivatives of a root-word pronounced  *gjwD.  In the existing pre-Han corpus, there are innumerable examples of  characters possessing identical graphic representation, as well as compatible phonetic and semantic value. For example, Karlgren states that the graph fen ft stands for the verb 'to divide' *pjwan, in addition to the noun meaning 'a part, share' *b jwan. Once again, it is the h  phonetic value of these words which permits them both to be represented by the same graph. 16  Both pairs have homorganic initials andfinals— i.e., although [*b] is [+voice] and [*p] is not, they are still both bilabial; both words contain the main vowel [*a] and final consonant [*n]. Karlgren also relates how the graph du g£ represents the verb 'to measure' *d a:k, in addition h  to the noun 'a measure' *d a:g (1956: 2; GSR: nos. 471; 801). Both words have homorganic h  initials andfinals,although the former has a [-voice] final and the latter one which is [+voice]. The similar meaning, pronunciation and graphic representation is a strong indication that these pairs are cognate. To date, numerous scholars have expressed an interest in the subject of cognates and word families (Karlgren, 1956: 1-10; Pulleyblank, 1973: 111-25; Boltz, 1994: 95-101). Although all theories regarding the phonological properties of word families are still tentative, there is a consensus that they did, in fact, exist in Old Chinese (Pulleyblank, 1973: 111; Boltz, 1994: 97). The type of cognate related to polysemous characters has been dealt with previously by G.B. Downer whose work has resulted in the "surmise that nouns in Old Chinese can be derived from the corresponding verbs by suffixing an -s to the verb" (Boltz, 1994: 97-8). Such a conclusion is of considerable interest when dealing with cognates like mu g and yu  . However, since this study is not intended to be a morphological treatise, I keep  concerns such as the nature of cognates to a minimum. Of course, in the pre-Han Chinese writing system, polysemy did not always involve cognate words. Following the same principles of homorganicity in regard to initials and finals, words that originally lacked graphic representation of their own adopted graphs that already stood for other dissimilar words. This particular type of graphic borrowing is referred to as paronomasia and its use in the multivalent stage of Chinese writing was widespread (Boltz, 1994: 101; Norman: 59). Paronomastic borrowing was typically applied to abstract words which were harder to convey pictographically than simple objects (Boltz, 1986: 426; Keightley: 108-9). For example, in the pictographic characters introduced earlier, there are several indisputable examples of paronomasia. The pictograph originally representing the word 'elephant' (6) was eventually used paranomastically for the word 'image, appearance.' Their identical Old Chinese pronunciation .xz„«g<*dzjang was integral to this process of borrowing (GSR: no. 728). Similarly, the graph originally representing the word 'pelt' ^/w<*g jo:g (1) was paronomastically borrowed to represent the word 'to seek' gro<*g jo:g h  h  (Boltz, 1986: 426; GSR: no. 1066). Pictograph (5), which originally represented the word 17  'wheat' lai<*lag, was borrowed to write the verb 'to come' (Norman: 61; GSR: no. 944). The homophony of the words associated with graphs (1) and (5) means they comply fully with both of the applicable phonetic principles that I have adopted in this thesis. Paronomasia certainly did not apply solely to contentives — i.e., nouns, verbs, adjectives, and the like. It was also a factor in the graphic representation of pronouns, auxiliaries and various types of particles, or what are known as functors. It is easy to see how functors would have been nearly impossible to represent in pictographic form; consequently, it is not unexpected that they were usually conveyed in written form by borrowing preexisting graphs (Norman: 60: Karlgren, 1956: 1). For example, at some point in the Shang, the OBI pictograph representing the word 'winnowing-basket' (3) was employed to represent a homophonous functor. Although the exact function of the OBI functor qi M. is not known, numerous interpretations regarding this character have been put forth. The JGWZGL offers the following: 'modal particle,' a particle of hypothesis and/or futurity,' a functor expressing the meaning of 'ought to,' 'if,' and cause (no. 2815). Further, in Western Zhou (1134-771 B.C.) bronze inscriptions (hereafter BRI) the graph zhui -fii, which originally represented the word 'bird' *tjwar, was adopted by the particle pronounced we/<*djwar. When I stated earlier that archaic Chinese graphs had fixed phonetic values, I did not mean to suggest that they could have only one pronunciation. In pre-Han Chinese some graphs had more than a single pronunciation. This is a phenomenon that diverges considerably from the singular phonetic value inherent to the various kinds of polysemous characters. For example, based on our present understanding of the Old Chinese phonological system, it appears that the matrograph kou • could be pronounced polyphonically as both *k u: and *mjerj (Boltz, 1986: 426; GSR: nos. 110, 826). These two pronunciations of these h  this character could not be further apart: whereas the initial in the former pronunciation is producedfromthe velar region and belongs to the hou fg rime group; the latter has a [+nasal] bilabial articulation and falls into the geng ffi rime group (Chen Fuhua: 160, 282). Moreover, in Old Chinese each pronunciation of kou • carries a different semantic value: 'mouth' in the case of *k u: and 'name, call' in the case of *mjerj. Polyphonic characters may be more h  common in the writing system of pre-Han Chinese than specialists originally thought (Boltz, 1994: 102-6). Their presence in the Chinese of this period highlights the importance of 18  understanding the phonological structure for the analysis of early Chinese texts. Paronomasia represents a significant step in the evolution of Chinese writing, but as time progressed it did not adequately meet the needs of an increasingly more complex and evolved language system. The paronomastic use of graphs eventually became so extensive as to create considerable ambiguity in the meaning of what was recorded (Boltz, 1986: 426-9; Karlgren, 1963: 5). When too many graphs came to represent more than one fully or nearly homophonous word, writing was rendered inefficient and imprecise, and reading became overly reliant on textual context (Boltz, 1986: 426-7). Clearly, some sort of device was needed to distinguish the intended meaning of each paronomastic graph. The problem of semantic ambiguity had to be resolved for Chinese writing to operate as an effective and accurate means of expression. This problem was addressed during the formation of what William G. Boltz has dubbed the "determinative stage" of Chinese writing. The term "determinative" implies the addition of a specific graphic element to polysemous characters in order to mark the different intended meanings (Norman, 60; Karlgren, 1963: 5). Characters of this kind usually combine two matrographs into one tomosomatic character and are referred to either as xiesheng 'shared phonophoric'  f g i f  characters, or phonosemantic compounds. Once the two components are  combined, they function as a phonophoric and semantic determiner within the newly formed character. In more specific terms, the phonophoric is the component of the xiesheng character which "bears the sound" (Boltz, 1994: 101). Moreover, since phonophorics were originally independent polysemous characters such as those discussed above, they may be viewed as the primary element upon which the phonosemantic compound is constructed. We may therefore refer to the phonophoric as the 'core' graph. Conversely, the semantic determiner is the element added to the phonophoric, its function being to disambiguate the meaning of the phonosemantic compound. As Boltz writes, it is "a secondary graphic constituent of a character used to specify which of two or more possible meanings is intended by a particular use of that character" (Boltz, 1994: 102,104). A second form of cognate word often appears in sets of phonosemantic characters. The cognate which appeared during the determinative stage of the Chinese language is fundamentally similar to the earlier multivalent form. Obviously, its graphic representation is distinguished by the use of a semantic determiner; however, this form of cognate is subject to 19  the same phonetic constraints governing cognates represented by multivalent characters. Specifically, cognates represented by the same phonosemantic compound must have homorganic initials and finals belonging to the same rime group. To illustrate what I mean, several sets of cognate words represented by xiesheng characters are provided below (Boltz, 1994: 95: GSR: no. 746): (la)  an< *?a:n 'security, stability, settled'  (lb) an < *?a:n 'to press down, hold down, make secure' (lc) j | < *?an 'quiet, at rest, settles' In terms of semantic value, all of the above words stem from the root sense of being 'settled' (GSR: no. 746; Boltz, 1994: 95-6). Phonetically, all three words have homorganic initials — the glottal plosive [*?] — and belong to the yuan ju rime group (Chen Fuhua: 300-1). (2a) „ you < *djo:g 'draw forth, proceed from' (2b) }ft chou < *t jo:g 'pull out' h  (2c) f& xiu < *dzjo:g 'sleeve' Although the initial pronunciation given (2b) differs slightly from (2a), and (2c) stands out from both (2b) and (2a) because of its initial affricate [*dz], all three initials are produced from the same articulation point and are, therefore, homorganic. Moreover, these words all belong to the yiu |£| rime group (Chen Fuhua: 142). Consequently, we know that on a phonetic level (2a), (2b) and (2c) are sufficiently compatible to constitute a cognate series. According to Boltz, (2a) to (2b) are probably sufficiently compatible in terms of meaning to meet the semantic requirements of a cognate set; that is, (2a) and (2b) are definitely both related to the root-sense of "conducting, follow from or along" (Boltz, 1994: 99). As for (2c), Boltz concludes that a sleeve functions as a kind of "conduit" which makes it akin to the same root-sense as (2a) and (2b). However, Boltz admits, "It quickly becomes a matter of opinion, and sometimes of spirited debate, how much semantic latitude one ought to allow in identifying word families" (1994: 99). Unfortunately, the semantic value of proposed cognate sets cannot be subjected to the same kind of set criteria that is applicable to their phonetic value. 20  In addition to the two sets above, there is also: (3a)%-ya< *rp 'tooth' (3b) 5?ya< *rjo 'a sprout' (3c) "fftya < *rjrj 'to welcome, receive' (3a) to (3c) are all fully homophonous (Karlgren, 1956: 2; GSR: no. 37). These words are all pronounced with a [+nasal] velar initial and belong to the yu ^ rime group (Chen Fuhua: 172). But in terms of semantic value, (3c) has nothing in common with the other two characters. Although, as already mentioned, the semantic latitude between items in a word family is largely a matter of subjective judgment, such is not the case here (Boltz, 1994: 99100). As such, examples (3a) to (3c) demonstrate how lexical items in xiesheng sets do not necessarily bear a cognate relationship to one another; indeed, there are innumerable xiesheng sets which contain non-cognate lexical items. For examples of non-cognate phonosemantic compounds, refer first to one of the OBI pictographs discussed earlier. The teknogram ft (example 5) gao<*kog, which belongs to the xiao H" rime group, is usually used as a phonophoric in xiesheng characters (Chen Fuhua: 152; GSR: no. 1129). Combined with ^ 5/<*g i:eg, the semantic determiner for silk materials, the character h  meaning 'white silk, undyed silk'  gao<*kog is created (GSR: nos. 876,1129). Since both  the initial and final for gao |jf are phonetically identical to gao ft, the principles permitting phonosemantic compounds are clearly adhered to here. The phonetic compatibility of these two characters can be further substantiated by the Shijing rime categories, which show that gao |g also belongs to the xiaoflfrime group (Chen Fuhua: 153). Another example of non-cognate xiesheng characters can be found in the set seen below, which is based on the phonophoric # fei < *pjwar 'it is not,' not,' 'without' (GSR: no. 579). In GSR, no. 579, Karlgren states that this graph may originally have been a drawing of two wings and represented the word 'to fly'. Subsequently, this graph may have been borrowed paronomastically for use as a negative. Clearly, in (la), (lb) and (lc) the meaning of the semantic determiner is unambiguously related to the semantic value of the character as a whole. (The number allotted to semantic classifiers is based on the Kangxi index.) 21  (la) with cl. 63, 'a door':  Mfei< *pjwar 'wooden door, door leaf  (lb) with cl. 173, 'rain':  9^fei< *p jwar 'heavy snow-fall'  (lc) with cl. 140, 'grass, weeds':  f£ fei < *p jwar 'name of a plant (radish?)'  h  h  As previously stated, Old Chinese pronunciation is as important as semantic value in the makeup of a xiesheng series. The full homophony of (lb) and (lc) naturally means that the phonetic interchangeability of these three words is beyond dispute, at least as far as we know. Further, the [-aspiration] initial in (la) does not impair homorganicity with the other three characters in the list. In discussing writing forms such as polysemy and polyphony, I have dealt with graphs for which a basic understanding of structure may be asserted. Sometimes, however, a clear understanding of the nature of a compound graph is more elusive. In many graphs the phonophoric is so difficult to distinguish that its very existence is questioned. The term huiyi ifM. 'semantic compound' is traditionally used to categorize combinatory graphs which appear "entirely semantic in their inspiration"; that is, the product of combining two semantic elements, rather than a semantic and phonetic determiner as is the case with xiesheng characters (Keightley: 190). For example, the syssemantic graph ming B J 'light, bright' has been assigned the Old Chinese pronunciation *mjarj which is comprised of the bilabial initial [*m] and a final corresponding to theyang |§ rime group (GSR: no. 760; Chen Fuhua: 274). The graphemes ri Ej 'sun' andyue ft 'moon' could both function as the semantic determiner forming B J , in which both are present. Yet, neither ri E] or yue ft bears a suitable phonetic value to function as the phonophoric for ming B J . A lack of homorganicity makes both ri 0 'sun' and yue ft incompatible, for while ri 0 is pronounced *njet, which has a [+nasal] retroflex initial and a final corresponding to the rime group zhi J f , yue ft has been given a [+nasal] palatal initial and assigned to the yue ft rime group (GSR: nos. 404, 306: Chen Fuhua: 244, 238). In order to fully comprehend the structure of semantic compounds such as ming BJ further research is clearly necessary. It has been suggested that once our knowledge of Old Chinese pronunciation improves we may discover that one of the elements in semantic compounds actually does function phonophorically. Of course, if such is the case, huiyi characters would amount to nothing more than an elusive form of 22  phonosemantic compound (Keightley: 191). The last part of this section will deal with the phenomenon known since the Qing dynasty as 'phonetic borrowing', or as jiajie  Although phonetic borrowing is in a sense  the antithesis of lexical variation, this form of graphic variation plays a significant role in the analysis of chapter three. For example, in Lexical Variation Study no. 51 assert that chang % is a phonetic loan character for chang ft. Moreover, in my examination of FYSJ bamboo slip S041, found in Lexical Variation Study no. 2, phonetic borrowing is also mentioned. The writing system in pre-Han China was not like it is today when most words are conventionally associated with certain characters. By contrast, until the implementation of a 'principle of unique association' sometime after the commencement of the Qin dynasty, there was a high degree of arbitrary graphic representation in Chinese (Boltz, 1997: 254-7). The phenomenon underlying this arbitrariness may be interpreted as the reverse of polysemy; that is, instead of graphs being adopted by more than one different word because of their phonetic qualities, as a result of jiajie the same word was commonly represented by more than one homophonous or phonetically similar graph. Although Karlgren discusses paronomasia in the context ofjiajie, on the basis of this characterization I would like to draw a distinction between these two types of phonetic borrowing (1963: 3-4). The negative implications of phonetic borrowing on a wide scale are not be difficult to imagine. Textual variation based on arbitrary graphic representation will inevitably result in semantic ambiguity. And the extent of phonetic borrowing in a text will undoubtedly be matched by the need for graphic analysis and contextual interpretation, as opposed to more straightforward reading. Fortunately, phonetic borrowing is contingent upon the pronunciation of pre-Han Chinese; consequently, just as an understanding of Old Chinese phonetic structure is helpful in distinguishing sets of xiesheng characters, it also enables us to identify phonetic loan characters. Bernhard Karlgren was the first to propose that the rules of phonetic compatibility for phonosemantic compounds also apply to the process of phonetic borrowing. Karlgren wrote, "In the thousands of Hs [xiesheng] characters of the classical script we thus have a material of paramount value for verifying what degree of phonetic similarity was demanded for permitting of a Kt [jiajie] employment of a character, or, to turn it the other way, what degree of phonetic dissimilarity made a Kt employment unallowable" (1963: 5). Thus, forjudging phonetic interchangeability in loan characters, we will refer to the 23  same two phonetic principles outlined earlier: i.e., homorganicity for discerning phonetic latitude in initials, andfinalswhich conform to the Shijing rime groups (Boltz, 1997: 255). I will refer to Karlgren's seminal study "Loan Characters in Pre-Han Texts" for examples of phonetic borrowing (1963: 1-10). The first set found below — (la) to (3b) — represents Karlgren's type B jiajie which involves one graph being replaced by another fully homophonous graph. The second set represents his type C jiajie which involves a borrowing process that is identical to B, except the loan character is merely phonetically interchangeable and not completely homophonous (1963: 3-4). (la) gong  *ku:g 'prince, public'  (2a)yu j!j *njo 'prison' (3a) su  borrowed for  (lb) gong Tft * ku:g 'merit'  borrowed for  (2b)yu H *njo 'to withstand'  *sju:k 'to lodge the night' borrowed for  (3b) su  *sju:k 'reverent'  (4a) tion ^ *d i:3n 'to destroy'  borrowed for (4b) tion Ift *t i:3n 'ample, good'  (5a) bao ^ *po:g 'respond'  borrowed for (5b) fu _ . *p ju:g 'hastily'  b  h  h  (6a)you ^ *djo:g 'to proceed from' borrowed for (6b) you $f *zjo:g 'equal to' According to the criteria used in this study so far, most of the examples ofjiajie given above are permissible. Clearly, the completely homophonous loan characters need not concern us here (examples la-lb, 2a-2b, 3a-3b). Moreover, since the finals found in examples (4a) and (4b), as well as (6a) and (6b) are completely homophonous, the only consideration for phonetic compatibility lies in their respective initials. Examination of the initials in both pairs shows indisputable homorganicity. In fact, [*d-], [*t-] and [*z] are all sounds produced from the alveolar ridge. By contrast, while the final consonants in (5a) and (5b) are identical, their main vowels are not; consequently, strictly speaking these two characters do not comply with the necessary phonetic constraints. Referring to the Shijing rime groups, the phonetic incompatibility of the vowels [*o:] and [*u:] is clear: bao $g belongs to the you  group,  while/a M: falls into the wu M group (Chen Fuhua: 149,215). However, the evidence Karlgren cites for this case ofjiajie borrowing is irrefutable: in a phrase from the Liji „ | B 24  (The Record of Ceremony) which reads, "hastily go"  „ bao wang, the character ^ bao is  clearly used in place of $tfu (1963: 2). This piece of evidence shows that we cannot rely completely on Shijing rime groups to discern phonetically compatible finals. Fortunately, further analysis of examples (5a) to (5b) may suggest an alternative rule with which to treat such cases. Closer scrutiny of the finals in the characters bao  and fu jtfc.  indicates they are produced from similar places: the head vowels [*o:] and [*u:] both possess [+back] and [+round] features, and while the latter is [+high], the former is higher-mid. It appears, therefore, that even if the characters are not in the same rime group, the borrowing process may still be allowable if the rime groups involved are similarly pronounced. To test for borrowing of this kind, the characters in question must be examined on the level of phonetic feature. Below, I offer a cognate series compiled by Karlgren which contains a great deal of jiajie borrowing (1933: 68). Anyone familiar with pre-Han texts will be accustomed to encountering different forms of graphic variation. For one thing, a survey of textual materials from this period reveals rampant graphic variation within cognate word sets or word families. This involves a form of graphic representation for cognate words that is fundamentally different from the two other types of cognates previously discussed. The first type of cognate discussed involves a common polysemous graph which represents numerous cognates. In this case, deciphering the meaning of the character depends largely on context. The second type of cognate involves a common phonophoric and a determiner which is added to elucidate exact meaning and usage. The third cognate type involves different graphic representation for most or ail cognate variants in a word family. A common phonophoric is not characteristic of this type; however, the semantic determiner this form of cognate depends on to be understood in written form corresponds with other xiesheng characters. This third type of cognate indicates the absence of systematic graphic representation in the pre-Han writing system. Although the semantic value implicit in these ten examples suggests the possibility of lexical cognation, there is solid phonetic evidence refuting such a conclusion. Examples (la) to (le) are characters which comply with the phonetic criteria I have adopted for this paper; their initials are all velar — either plosive or fricative — and the finals all belong to the xiao W rime group (Chen Fuhua: 149-151). A closer look at the points of articulation used in the 25  final positions of examples (la) to (le) results in the following: in all cases the main vowel [*o] is [+round], [+back], [+mid] and [+lax]; the final consonant is the [+voice] plosive velar [*g]. By contrast, examples (2a) to (2e) all represent words which phonetic criteria prohibits from being associated with the same word root. In terms of initials, they are not consistently compatible in phonetic terms. Moreover, for the most part they belong to different rime groups, ranging from^ow ^ to duo M (Chen Fuhua: 142-144, 213, 222, 267). In this group, (2e) is the most strikingly dissimilar: its main vowel [*a] is [+front], [+low] and [-round] which places it in the duo f$ rime group, and its final consonant is a [-voiced] [*-k]. (It is worth noting that in pre-Han Chinese, the final [*-k] could function as a type of archemorpheme referring to a previously mentioned subject. In this sense, [*-k] was affixed to pronominal particles such as shu fHi, 'which', ge # 'each', huo ^ 'some' and mo ~% 'none'. For more, see Pulleyblank, 1973: 122, 1995: 92.) (2a) also stands out from the set in that its initial [*{-] is [+retroflex]. Another incompatible final is found in (2c): its main vowel possesses phonetic features such as [+round], [+back], [+high] and [+long]; its final consonant is the [+nasal] ENG [*-n]. It falls into the dong ^ rime group. (la) Syao < *djog 'kiln'  (2a) i | zhu < *rjo:g 'to cast metal  (lb) % yao < *djog 'jar, vase'  (2b)  (lc)  xiao < *sjog 'to melt'  (2c) | § rong < *di:u:rj 'to steam'  (Id)  chao < *tsog 'to fry, roast'  (2d) M diao < *ti:o:g 'fade, whither'  (le) ^ zao < *sog 'to dry'  shuo < *sjak 'to melt'  (2e) JJf xi < *sjak 'dried meat'  It warrants noting that even Karlgren was not certain that every cognate in each of his word families actually belonged. In regard to his proposals, Karlgren writes: "So each small 'word family' has to be considered merely as a kind offrame, containing materialsfromwhich a choice will have to be made in the future. I am very far from affirming that all the words in each group are cognate; I only mean to say that they may be suspected of being cognate. In a few cases the affinity is absolutely obvious and certain and in many more it is strongly probable. In the rest it is only possible and at least worth discussion" (1933: 59). Subsequently, any revision conducted in regard to the above cognate set should not be 26  understood as a criticism of Karlgren's judgment; instead, it constitutes a continuation of his work on word families. Part Two:  The dramatic transformation occurring in the Spring and Autumn (722-484 B.C.) and Warring States (403-222 B.C.) periods had a profound impact on how Chinese characters were written. Broad social, political and cultural changes which took place in China during this time resulted in quick and widespread economic development. Increased productivity in agriculture and handicrafts caused writing to spread beyond the ruling class to other sectors of society where it was used more frequently than ever before (Qiu: 51-2). David Pankenier echoes the opinion of other scholars when he writes, "In the wake of growing agricultural production commerce and handicrafts began to develop [sic], the use of writing became daily more widespread and newly created characters, abbreviations and variant forms circulated in great numbers" (30; Qiu: 52). The unprecedented transformation of script referred to here was manifest mainly in orthographic simplification. As such, we may link the aforementioned conditions to the gradual formation of the simplified liu guo zi f\Wfi£- 'Six States script'; however, before examining this important type of Warring States script, one crucial aspect of Chinese orthography in general warrants discussion. I refer to the interrelationship of complex and simplified writing forms which is a recurring phenomenon in Qiu Xigui's survey of early Chinese orthography, the Wenzixue gaiyao  (An Outline of the Study of  Writing). Qiu Xigui discusses the complex forms of writing found throughout early Chinese history in terms of zhengtizi lEft^ "orthodox script." This refers to a writing form that 1  would be more appropriate for formal occasions or ceremonial use. By contrast, the simplified form, or what Qiu calls suti zi flrfl^ "vulgar script," refers to a kind of script 1  which can be conveniently written and is suitable for daily usage. (Although the terms "orthodox" and "vulgar" are common in the accounts of Chinese script by Qiu Xigui and other writers, the ethical and political implications of these terms make them unsuitable for academic work; therefore, I do not use them in my own writing.) The demand for a quicker and easier way to write always necessitated a simplified writing form, which is really nothing 27  more than an abbreviated version of the complex script contemporaneous with it. In the early development of Chinese orthography, these two forms of script sometimes coexisted temporarily in a relatively static state; however, it is the nature of orthographic evolution that the complex script of the day eventually underwent a process of simplification (Qiu: 43-4, 58). Discussing the relationship between these forms of script, Qiu Xigui writes, "Actually, in the course of orthographic development, the function of vulgar script was extremely important. Sometimes a new kind of orthodox script developed out of a vulgar form from an earlier stage. A more frequent situation [in the evolution of early Chinese script] is when writing techniques belonging to the vulgar form later become absorbed by orthodox script, or when it [vulgar form] clearly expedited the evolution of orthodox script" ^ f f > %E5[.¥-]& 8BM j  mm - (44). For example, in the Shang dynasty complex script is found in BRI, while the more practical and simpler oracle-bone script constitutes its simplified counterpart. As was typically the case throughout the first millennium B.C. and before, the simplified script of the Shang dynasty was born out of necessity. Since Shang rulers carried out oracle-bone divinations on a frequent basis, a large number of characters had to be incised on turtle plastrons and scapulae. Engravers found themselves in a position where they had to develop a quick and efficient way to write on these hard surfaces; consequently, they began using a more simplified script characterized by relatively thin strokes, as well as round shaped graphs instead of something more square. Moreover, whereas complex BRI characters tended to be hollowed or gauged out {tion shi IftH) during the Shang, oracle-bone characters were instead engraved in profile (gou kuo ffaM) (Qiu: 42). The combined effect of this simplification of script was a diminishment of pictorial quality. Naturally, complex script existed in the Warring States period. In fact, a complex script resembling the writing of the preceding Spring and Autumn period was common throughout China from the fifth to third centuries B.C. The complex script of this time is characterized by a high degree of uniformity and linearization (xian tiao hua i M ^ b ) , as well as the tendency to straighten lines which were previously curved or bent to facilitate 28  pictography (ping zhi hua  SMb) (Qiu: 46-8). However, the popularity of simplified script  grew to an unprecedented degree during this period. The variation visible in the writing of China's six eastern states — Qi ^ , Chu $t, Yan ^ and the three states of Jin ff, which has been called Six States script, differed considerably from the traditional complex form. Because the geographical base of the Six States script was divided between a number of politically and culturally distinct areas, the orthographic diversity implicit in it is extremely broad (Qiu: 578). Along with a higher degree of simplification and a marked reduction in pictorial quality, this diversity constitutes a distinguishing feature of Six States script (Norman: 62; Pankenier: 31). Below, 1 offer eight examples of Six States script from the Warring States period (Gao Ming: 107,60, 89,495: Zeng: 226,237; Zhang: 79, 83). Since I have already commented on the characters ma H, gao ft and qi S. in relation to their OBI pictographic form, (14), (16) and (17) require no further comment. I will, however, explain the orthographic components which comprise the characters in examples (10), (11), (12), (13) and (15). A clearer idea of the structural content of these characters will make the calligraphic and orthographic comparisons easier to follow.  (10) dao H 'way, road, method': This teknograph is composed of two main components: (a) shou ~i§i 'head' and (b) chuo 3lL 'to go over, pass over' (GSR: nos. 1048, 1102). Component (a) is likely the phonophoric for this character, since it is nearly homophonous to dao in Old Chinese (Chen Fuhua: 142, 145) — e.g., compare dao<*d o:c_ and shou<*cj\o:q. The SWJZ h  states that (a) is a pictograph of a head with hair on top (SWJZZ: 9:15a), while Karlgren writes, "The graph is a drawing of the head of a horned animal" (GSR: no. 1102). According to Karlgren, the latter component can further be divided into two separate elements: the chi 29  if 'street' element at top and the zhi Al. 'foot' element below (GSR: no. 1258). Other forms of this character are found in examples (18), (28), (33), (38), (50) and (58) forthcoming. (In the following explanations, I will use square brackets to indicate the corresponding examples of each character.) (11) de f H 'virtuous, virtue': This character is conventionally comprised of three main graphic components: (a) zhi _ 'straight, right', which is composed of a verticalized eye graph and a straight line; (b)xin /_ 'heart'; and (c) chi ^ (GSR: no. 919). However, as example (11) is an abbreviated form of this character, element (c) is absent. [(19), (29), (34), (39), (59)] (12) % wei 'to do, to be, to make': This character is comprised of (a) zhua /ft 'claw' and (b) wei  which Xu Shen believes to be a pictograph of 'a female monkey' (SWJZZ: 3:14b). In  regard to (a), Karlgren writes, "In compound characters the element mostly means 'hand'" (GSR: no. 11). The SWJZ states that (a) depicts the cover of a finger or a 'nail' (M¥0/TO (SWJZZ: 3:14b). However, there is good reason to disagree with Xu Shen's understanding of (b). According to Yu Xingwu J ^^ 1  and other paleographers, this graph depicts an elephant  being led along by a human hand and Xu Shen's mistake is the result of referring to corrupted Small Seal graphs (JGWZGL: no. 1654). [(20), (30), (35), (40), (52), (60)] (13) ji R 'come to, reach': GSR, no. 681 states, "The graph shows a 'hand' grasping a 'man'." In this teknograph, the lower element ^ is the pictograph of a 'hand'. [(21), (31), (35), (40), (52), (60)] (15) wu  'not have, not, no': GSR, no. 103 states, "The graph is [... ] a drawing of a man  with dancing plumes in the hands." Although we can only speculate on what the dancing figure is actually holding, on the whole Karlgren's understanding of this character is sound. In support of this view, Sun Haibo &M$. writes, "[This graph] depicts the shape of a person dancing with an ox-tail in each hand"  feAM¥^JlffnteP  (JGWZGL: no. 226). [(23),  (32), (42), (55), (63)] As mentioned, in the OBI pictographs discussed at the outset of this chapter, we have 30  already seen earlier forms of (14), (16) and (17). Referring back to OBI pictographs (3), (4) and (7), the comparatively diminished pictographic quality of the Six States script is obvious. Whereas (7) clearly offers a visible depiction of a horse, the same cannot be said of (14) — i.e., compare (14) with example (7)  Example (14) would not look like a horse to anyone  reading this graph without prior knowledge of what it represents. And although (16) still bears a remote resemblance to a 'high building,' it is not nearly as pictographic as its counterpart, OBI graph (4) (ti. In (16), the graph's curved and shortened lines make it appear less like a building. Further, (17) does not resemble a ' winnowing-basket' as much as the OBI pictograph (3)  . The Six States graph for qi 5£ abbreviates the appearance of a basket  containing lines that signify interwoven strips of bamboo. (The graphic element comprising the bottom half of example 17 will be discussed in connection with example 49 forthcoming.) Examples (10) to (17) all contain characteristics which typify Six States script. In addition to displaying minimal pictorial qualities, they also show a markedly high degree of simplification. But since orthographic forms used during the Warring States period are bound to be less pictographic than OBI writing, it may be more meaningful to compare Six States script with scripts contemporaneous to it. Comparison of this sort is forthcoming. However, before proceeding any further, a comment should be given here concerning the approach taken in the analysis of examples (1) to (17). I have taken care to describe all the various components found in each of the graphs featured in order to arrive at a clear and accurate understanding of their orthographic features. An awareness of these features allows the graphologist to study and compare Chinese graphs on a level that is more precise and nuanced than that of script. As the present chapter continues, I will utilize orthographic feature to discuss graphs written in different scripts. Moreover, this methodological approach will continue to play an important part in chapter two, where I endeavor to date the FYSH script. Among the states established during the Spring and Autumn and Warring States period, the state of Qin was in many ways unique. Moreover, its distinctiveness can largely be attributed to location. Since the state of Qin was situated where the Zhou dynasty was originally located, the script used there was particularly traditional in form. That is, in contrast to the states located in eastern China, Qin writing preserved the pictographic quality of the earlier Zhou script. In addition, because the state of Qin was relatively remote and isolated — located as it was to the distant west of the other states, the script used there 31  evolved more slowly than in eastern China. That a simplified orthography did not appear in the Qin until the middle of the Warring States period is one indication of this (Qiu. 52; He: 164-5). However, not only is Qin script of special interest to the study of Warring States orthography in general, it is also of particular interest to the present study of the FYSJ. After all, this manuscript version of the Shijing is written entirely in script originating from the state of Qin. To fully understand Qin script we must view it as encompassing two visually distinct forms: i.e., Small Seal and Clerical script. Small Seal script is viewed by some as the parent of Clerical; therefore, I will proceed with a discussion of it first. In the Han dynasty, a distinction was made between so-called Large Seal or da zhuan -jxM. script, the official script of Warring States Qin, and Small Seal, which is traditionally believed to have been created during the Qin dynasty. Drawing such a distinction can be misleading if it causes us to overlook the fact that these Seal scripts share enough in common to be considered part of the same orthographic lineage (Qiu: 65-6). To better understand how these scripts relate to one another, we should view Large Seal as the complex form of Seal script and Small Seal as its simplified descendant (Qiu: 63). However, this observation does not tell the whole story of Small Seal script and there are several more points regarding this script that warrant mention. Although Small Seal was clearly not "invented" by court ministers during the Qin dynasty, it was the end product of a conscious modification of pre-existing complex Seal script that occurred during this time (Qiu: 65). The modification resulting in Small Seal script likely involved some innovation; however, its formation was mainly the result of "a thorough unification and arranging of the Qin script along the same lines as its natural developmental tendencies" (Pankenier: 33). Moreover, the central aim of this modification was to produce a standard Seal script with which to replace the various simplified scripts deemed incompatible with the emperor's efforts to consolidate power and unify China (Pankenier: 30-33). From a larger perspective this 'unification of script' (shu tong wen zi Ulfjjt^y ) was merely one facet 1  of the "spirit of standardization" implicit in the world-view at the time (Boltz, 1997: 257). Other manifestations of this world-view include the Qin emperor's standardization of weights and measurement, as well as currency (Norman: 63). Below are some examples of the complex Seal script antedating the Small Seal script of the Qin dynasty (Shi ke zhuanwen bian ^M^ScM', hereafter SKZWB: 1:22; 2:27; 3:17; 32  6:17; 7:24; 10.1; 11:21). The orthographic characteristics embodied in these characters already resemble the script used in parts of the FYSJ. Examples (18) to (23) are the equivalent of Six States graphs (10) to (17); therefore, they do not require componential analysis. The other four graphs (examples 24-27) have not previously been featured in this study; therefore, a brief account of their orthographic structure is forthcoming. Once again, it should be noted that the method I use to describe orthographic makeup will be followed throughout the second chapter.  (18)  (23) (24) yu fe 'fish': As Xu Shen notes, (24) is a pictograph of a fish in which "the fish tail resembles the tail of a swallow"  _&^&SH5$M*B'I&1 (SWJZZ: 11:17b). [(44)]  (25) cao J|L 'grass, plant, herbs': This particular character is found in GSR, no. 1049. It is composed of three main components: (a) unlike the later conventional form of this graph, in (25) there are two sets of cao pty 'grass, plants' (GSR: no. 1052); (b) ri 0 'sun'; (c) shi -j'ten' (GSR: no. 686). Components (a) to (c) are all pictographic in nature. [(45)] (26) yuan J | 'circle, circumference, round': According to GSR, no. 227, this teknograph contains two graphic constituents: (a) wei P 'circle' in the upper section; (b) ding ^ 'cauldron' in the lower part. Karlgren mentions that: "In the modern graph the ting [ding] cauldron has been corrupted into pei cowry [bei m] (GSR: no. 227)." As will be evident shortly, this example of graphic corruption is already apparent in the SWJZ. [(46)] (27) min _ 'toad': Xu Shen wrote the following in regard to (27): "[This graph] is pictographic: a toad's head is the same as a snake's head" 33  _If |WJ (SWJZZ:  13:10b). Subsequently, we may view (27) as containing the following components: (a) min H which is a pictograph of a 'toad' (GSR: no. 1252); and (b) xue /X which is a drawing representing the word 'cave, pit, hole' (GSR: no. 409). By the time of the SWJZ, component (b) was already absent from this character. [(47)] Examples (19) and (23) are from the Spring and Autumn bronze container called the Qin Gong gui ^ 5 „ _  ('A Bronze Vessel of the Duke of Qin'; hereafter QGG) (Jinwen bian  nos. 87, 323; hereafter JWB) which has been dated approximately to the reign of Qin  Jinggong^-J|£ (576-537 B.C.) or Qin Gonggong fj|4±„  (608-604 B.C.) (Qiu: 60). The  remaining graphs are from the Shi gu wen ^$M.yi ('Stone Drums'; hereafter SD) traditionally ascribed to King Xuan __t of the Zhou dynasty (r. 827-782 B.C.). However, they were more likely produced by one of the dukes of Qin (Mattos: 1988, 2). Whereas Qiu Xigui dates the SD to somewhere between the late Spring and Autumn period and early Warring States period (approx. 480-475 B.C.), another source dates them more exactly to 475 B.C. (Qiu: 59; He: 159-60). Among examples (18) to (27), only a single character can be used to compare Seal script with the OBI pictographs seen earlier. Although (22) may still be considered pictographic in comparison with later types of script, it does not possess nearly the same degree of pictorial quality as (7). Whereas the OBI pictograph 9 clearly depicts the head and mane of a horse, as well as its legs and a tail, (22) has converted these elements of the original drawing into an oval, five strokes on a downward angle and three horizontal lines. Clearly, unlike Shang OBI pictograph (7), the SD seal graph (22) is more symbolic in shape than pictorial, something which reflects a common tendency in the evolution of Chinese orthography. In the upcoming orthographic analysis of the FYSJ, this tendency will be observed on a regular basis. The transformation from a pictographic form to one which is more symbolic plays a key role in dating the various Small Seal and Clerical scripts which comprise the FYSJ. A comparison of OBI graphs.(3) and (4) with Small Seal forms is still forthcoming. Unfortunately, since (3) and (4) do not appear in the QGG and SD, we cannot compare them with Seal characters from the Western Zhou period. I have already mentioned how vastly different Six States script is from other forms 34  contemporary to it. To substantiate this claim, I will compare Six States script and early Seal script. Due to the stark contrast between these two forms of script, a limited comparison of only three sets is sufficient for the task: Six States (10) :: SD (18) [the character dao >_]: The Six States form of shou ^ 'head' has an abbreviated number of strokes and has lost its original pictorial form — i.e., both the 'head' and the 'hair' or 'horn' element in l § is indistinct compared to 1§ in SD (18). The chuo 3^ component in (10) is also an abbreviated form of its (18) equivalent. While (10) contains this component in the form of >, SD graph (18) contains 9 (S , which is an older and more 1  pictographic form of the graph representing the intersection of a thoroughfare (JGWZGL. no. 2289). Six States (11):: QGG (19) [the character de $!]: The Six States form of this character is also more simplified and less pictographic than its QGG counterpart. Firstly, with its reduced number of strokes, the xin  'heart' grapheme in (11) manifests a diminished pictographic  quality. While the 'heart' in (11) is written  , in (19) it is rendered more pictographically  in the form of HP (GSR: no. 663). Further, the chi tf 'street' component found on the left side of QGG graph (19) has been completely struck from (11). Six States (15):: QGG (23) [the character wu fia]: The simplification (15) undergoes is even more drastic than that manifest in either of the above sets. In (23) the form of a dancing figure T with ox-tails %% in both of its hands is clearly discernible, something which shows this graph was highly pictographic even in the Spring and Autumn period. Conversely, as the result of extreme simplification in regard to the ox-tail element and abbreviation of the dancing figure, (15) no longer resembles somebody dancing. In each of the above sets, Six States script is much less pictographic than the Seal script found in the QGG and SD inscriptions. Diminishment of this quality is caused by widespread graphic simplification and abbreviation, which includes reduction in the number of strokes and the elimination of certain graphemes or graphic components. As will become 35  clearer as this study progresses, both of these phenomena figure prominently in the process of orthographic evolution in ancient China. As such, the reader should prepare to witness similar transformation when this paper turns to an examination of Qin and Han Ancient Clerical script. At the same time, by juxtaposing Six States script with early Seal script, a clearer impressions of their respective orthographic identities is attained. While Six States script is not one of the focal points of this study, Seal script certainly is; therefore, the detailed orthographic profile of Small Seal script given in the upcoming survey may be viewed as an extension of this initial impression. Examples (28) to (32) below post-date the SD inscriptions by roughly 150 years (SKWZB: 2:28-9; 3:17: 3:19; 6:9). They are from the Zu Chu wen  ('Text of  Renouncing Chu'; hereafter ZCW) stone inscriptions which have been variously dated to either 312-310 B.C. or 328 B.C. (SKZWB: 1; Qiu: 59). Examples (28) to (32) correspond to the BRI and SD Seal graphs (18), (19), (20), (21) and (23). Unfortunately, the ZCW does not contain graphs that correspond to SD examples (22), and (24) through to (27).  A comparison of this ZCW set and the one above shows a change in the proportion and shape of Seal script. For example, the horizontal m«@ 'eye' matrograph in (29) is more horizontal and flat in shape; overall, it is less circular than in QGG (19) — e.g., compare C I P (29) to © (19). This represents a change that would eventually be incorporated into the Small Seal form of this character (see examples 34,39). More drastic structural change resulting from graphic abbreviation is evident in a comparison of ZCW (28) and SD (18): the graphic element <l positioned at the far right of (18) is completely absent in (28). Moreover, the 'foot' grapheme found in the SD form of dao M has been moved. Whereas this element is located at the bottom of the graph in SD (18); in ZCW (28) it is situated on the left-hand side, something which also becomes conventional in the Small Seal form of this character (see examples 33, 38). None of these alterations have a major impact on the pictographic quality of the ZCW Seal script. In this sense, therefore, the three sources discussed so far — QGG, 36  SD, ZCW — show a great deal of orthographic similarity. The ZCW graphs differ from the earlier QGG and SD graphs in at least two other ways. One aspect of difference is crucial to this historical examination of Seal script, while the other is probably inconsequential. The verticalized rectangular shape resulting from the repositioning of the 'foot' grapheme in ZCW (28) is of special significance in the development of Small Seal orthography. The same vertical rectangularity is also manifest in ZCW examples (29) to (32). As will soon be evident, one of the most salient characteristics of Small Seal script is its vertical rectangular shape (Wang Fang-yii: vii). However, the ZCW script also exhibits a trait which clearly does not represent movement towards Small Seal script and is probably attributable solely to calligraphic style. In contrast to both the preceding SD Seal graphs, and the standard Small Seal graphs to be featured next, the strokes comprising the ZCW script have sharpened ends (Wang Fang-yii, vii). But just because calligraphic style lies outside the orthographic development of script, and is not generally of value for determining date, does not mean it has no place in the analysis of the FYSJ. A great deal can be learned about the physical makeup of a composite manuscript like the FYSJ from a study of its calligraphy. Examples (33) to (37) below all represent Small Seal graphs from ceramic inscriptions dating to the Qin dynasty (Yuan Zhongyi: 142-5). Overall, these graphs possess lines which are not as curved as those in examples (18) to (32). Instead, they exhibit the slightly angular lines more characteristic of Small Seal script. Examples (33) to (37) are contemporary to the Qin dynasty stone tablet inscriptions (hereafter STI) seen in (38) to (42) (SKWZB: 2:26; 2:28; 3:10-9; 6:9). Qiu Xigui states that STI are probably the best materials presently available for the study of Small Seal script (Qiu: 59). Despite being from different regions of China, the STI graphs are all dated to 221 B.C. which is the year that emperor Qin Shihuang (259-210 B.C.) established Small Seal as the official script of the Qin dynasty (SKWZB: 1: Wang Fang-yii: vii). Finally, (43) to (49) are from the SWJZ and representative of Small Seal script dating to around 100 A.D. (Gao Ming: 192, 214, 300, 218, 396, 89,405). Containing over nine thousand characters, the SWJZ is another invaluable resource for the research of Small Seal script — albeit the form of Small Seal written during the Eastern Han. This graphological dictionary contains many Small Seal characters that simply do not exist anywhere in earlier materials (Qiu: 63). For example, although Spring and Autumn BRI 37  and the early Warring States SD contain the Seal script characters (24) to (27), neither the ZCW, the ceramic inscriptions, or the STI from the Qin dynasty offer corresponding graphs. Moreover, for Small Seal equivalents of the Six States characters (16) and (17), once again we must rely on the SWJZ.  In each of the following three sets, I refer to the SWJZ to compare graphs written in Small Seal script with OBI pictographs: OBI pictograph (7) :: Small Seal (37), (43) [the character ma H]: Basically speaking, the Small Seal forms of ma % 'horse' resemble SD graph (22) already discussed. Perhaps the most significant difference between SD (22) and its corresponding Small Seal forms is the latter's more rectangular shape. I have included (37) in this study because it is an interesting example of a graph exhibiting both Small Seal and Clerical features. Although the lower part of this graph — the horse's legs and tail — still roughly match Small Seal graph (43), its squarer upper portion — representing the head and mane — is distinctly Clerical in appearance (see examples 54, 62). OBI pictograph (3):: Small Seal (49) [the character qi S ] : Both graphs contain the component representing a 'winnowing-basket'. In the case of OBI (3), qi 3=L is written while in SWJZ (49) this character is written almost identically in the form of 1Sf. The main differences in the two characters are: 1) the Small Seal form possesses a horizontal line at the 38  top of the U-shaped basket; 2) there is a component beneath the picture of a basket in (49) which Karlgren refers to as "two legs and a stroke crossing their upper part" (GSR: no. 487). There is a consensus among etymologists that this is a pictograph of a kind of stand, presumably for the basket to sit on (Wilder: 539; SWJZZ: 5:22a). OBI pictograph (4):: Small Seal (48) [the character gao ft]: Whatever minor differences exist between these two forms of gao ft, they are obviously variant orthographic forms of the same character — e.g., compare SWJZ (48) with (it in OBI (4). The above examples prove that Small Seal script was still relatively pictographic for its time. As such, this evidence reflects what was mentioned earlier regarding the conservative nature of Seal script in general. Jerry Norman writes, "The Qin script is directly descended from the bronze inscriptional script of the late Western Zhou dynasty. In the course of its development the seal script had taken on a more regular and balanced appearance without, however, changing to such a degree that its ultimate pictographic origins became totally obscured" (63). Such a view is further substantiated if we consider the form qi % and gao ft take in Six States script; both (16) and (17) are much less pictographic than Small Seal graphs (48) and (49). The difference between Small Seal and its ancestral Large Seal script is certainly not as great as the contrast between Small Seal and Six States script. The various forms of Seal script have several common characteristics, such as strokes with both "rounded corners" and ends (Wang Fang-yii: v-vii). However, some distinct differences in the orthography of Small Seal and Large Seal have been observed. For example, David Pankenier states that (A) "[in Small Seal script there] was a change in the brushstrokes making them tend to be straighter and more uniform, with the result that the pictographic quality was diminished"; (B) [in Small Seal script there] was the abbreviation or simplification of the radical [i.e., semantic determiner] or phonetic [i.e., phonophoric]" (32). Qiu Xigui's opinion regarding complex Seal and Small Seal script is essentially in accord with these two points (65). At the same time, Wang Fang-yii 3i7f ^ makes the following observation concerning Small Seal characters: (C) "Each character is vertically rectangular in shape" (vii). Early signs of this last orthographic feature are already apparent in ZCW examples (28) to (32), something which marks the 39  gradual orthographic development from Warring States Seal script to standard Small Seal. Pankenier writes that Small Seal graphs (44) yu ft and (47) min  are good examples  of point (A) (32). There is no disputing that these graphs are less pictographic than SD (24) and SD (27). For example, comparing SWJZ (44) with SD (24) H, not only are the side fins of the fish in (44) no longer distinct, but the square graphic constituent at the center of this character distorts the shape of the fish altogether. It is unlikely that a reader lacking prior knowledge of this graph and without suitable context could conclude that (44) represents a fish by looking at it. The xue )X component in graph (47) resembles a cave or hole about as much as the corresponding element in SD (27); that is, at least to a modern day reader. On the other hand, the object inside the confines of xue /% in SWJZ (47) does not resemble a creature to the same degree as the toad in SD (27); the SWJZ matrograph <H lacks the arms and feet which are so distinct in SD ^P, and its depiction of a tail is not as clear either. The examples Pankenier offers for point (B) are Small Seal characters (45) cao ^ and (46) yuan ^ . Example (45) lacks the lower semantic element W present in the corresponding SD graph (25). Moreover, compared to the SD graph (26), the lower graphic component in (46) is also abbreviated. As previously mentioned, the change seen in this component transpired when the graph for 'cowry' W was used to replace the original 'cauldron' It graph. Other examples supporting point (B) are Small Seal (33) and (38), which lack the far right grapheme i | contained in SD (18). More than any other trait, the upright rectangular quality of Small Seal characters separates them from their ancestral forms. In the QGG and SD inscriptions, the Seal characters are typically wider, filling a space which is more of a broad square than a vertical rectangle. Examples (18), (19), (21), (23) and (25) are perhaps the clearest illustrations of this orthographic tendency. For example, the fingers of the  'hand' element dominating (21) are  excessive in length, making this graph shorter and broader than its Qin dynasty counterpart ^) seen in (41). Furthermore, the ma MJ graph found in (37) and SWJZ (43) are clearly higher than SD graph (22). And the arms of the dancing figure in QGG (23) stretch to the sides much farther than in Small Seal graph (42). However, it would be a mistake to assume that 40  this particular trait of Small Seal script was nonexistent before the Qin dynasty. ZCW examples dao  (28), de $g (29), wu $& (32) and especially ji S. (31) are already narrower in  a vertical sense than their antecedent Seal forms. But on the whole, the tendency towards rectangular form becomes even more pronounced and consistent among standard Small Seal characters — e.g., examples (38), (39), (44), (45), (46), (48), (49) all exhibit this trait. Whereas Clerical script may be considered the simplified script of the Qin dynasty; Small Seal constituted the traditional and more complex state script. However, one can also view these two scripts as belonging to one graphic lineage; Small Seal being the simplified offspring of Large Seal script and Clerical script the still more simplified grandchild (Pankenier: 35; He: 166; Qiu: 63,68). Several Han sources state that Clerical script was "invented" by members of Qin Shi huang's court in a manner similar to Small Seal (Pankenier: 33-4). But based on recently excavated bamboo and ceramic inscriptions, we now know Clerical script already started taking form in the late Warring States period (Zhang: 32). This does not rule out the possibility that certain Qin officials were involved in shaping and standardizing Clerical script (Pankenier: 34). Both traditional sources and contemporary scholars suggest Clerical script was at least partially formed in a deliberate manner to meet the needs of scribes, government clerks and other people requiring a script easier to write than Small Seal. It is important to keep in mind that the demand for a more practical form of writing was not unique to the Qin; rather, this phenomenon was common throughout China during the Warring States period. The same motivation underlies the development of the Six States scripts (Zheng: 32). Because the Clerical script of Warring States Qin is not fully mature, it still bears some resemblance to the Seal script from which it originated (Qiu: 69). A more mature form of Clerical script, which resembles Seal script less, gradually took form during the reign of emperor Wu of Han H J W (r. 140-87 B.C.) (He: 167; Qiu: 77). Eventually, however, the Clerical written during the Qin and the Western Han dynasties became called Ancient Clerical guli  script; whereas the Clerical that evolved in the late Eastern Han (25-220 A.D.)  became known as Eight Parts ba fen  script (Qiu: 79). Even in the very beginning, the  significance of Clerical script was monumental. One scholar believes its formation effectively led to the abandonment of the "pictographic factor' in Chinese script, and set in motion a 41  trend towards orthographic "symbolification" (fu hao hua ffi^fcik) (Pankenier: 34; Zheng: 25); that is, a writing system based on abstract symbols. This claim is open to debate. For instance, it is difficult to prove that pictography in Chinese script was ever really discarded. Further, when discussing orthography in general one must always be cautious making absolute statements. However, there is no doubt that diminished pictography is a vital orthographic feature of Clerical script, and an important means with which to date graphs written in early manuscripts such as the FYSJ. Examples (50) to (57) below represent Ancient Clerical script from the Shuihudi bamboo inscriptions (hereafter SHD) (Qin jian wenzi bian ^ H f X ^ H ; hereafter QJWZB: 108, 759, 188, 719, 436, 394, 507; LZB: 23). The tomb from which these bamboo documents were excavated belonged to a person active during the reigns of Qin Zhaorang ^ H Q I S and Qin Shihuang, a period roughly spanning 240 to 217 B.C. (Zheng: 33). Therefore, the appearance of Clerical script in the SHD proves that this form of writing did not originate in the Qin dynasty as traditionally believed. Without a doubt, the SHD inscriptions are the best source presently available for the study of Clerical script in its formative stage (Qiu: 69).  At first glance, Ancient Clerical and Small Seal script appear quite different; however, they share several common traits which expose a strong orthographic affinity. In the analysis below, I will discuss examples (50) to (57) in terms of the following orthographic features: (A) the verticalized rectangular shape characteristic of Small Seal which is also evident in Ancient Clerical script; (B) the similarity of Ancient Clerical and Small Seal script in terms of orthographic structure — e.g., the position of graphic components within characters; (C) the use of strokes in the writing of Ancient Clerical and Small Seal characters, as opposed to dots that distinguish later forms of Clerical script. Unlike the comparisons of orthography given 42  earlier in this section, which were primarily intended to provide a clear understanding of the various early Chinese writing forms, the following juxtaposition of Small Seal and Ancient Clerical scripts has a more direct bearing on the upcoming orthographic analysis of the FYSJ. As chapter two illustrates, much of the FYSJ is written in Small Seal and Ancient Clerical script; therefore, the orthographic features used in the analysis given below will prove quite essential when I work on dating the FYSJ, and mapping out its physical nature. The above Ancient Clerical graph dao j _ (50) is vertically rectangular, something • which typifies Small Seal script. Other examples from the SFTD also exhibit this feature, such as: wu M (55), gao ft (56) and qi S. (57). Of these three examples, (55) probably best embodies orthographic feature (A), since it stands upright; both (56) and (57) lean slightly to the right. In regard to orthographic feature (B), the semantic determiner \- chuo 'walking' found in (50) corresponds to Small Seal characters (33) and (38): in both scripts this graphic element is positioned completely to the left of the graph. Clerical script characters (55) to (57), as well as wei fe (52) are also orthographically similar to their Small Seal counterparts (40), (42), (48) and (49). Actually, the structural similarity between SHD (56) and (57) and SWJZ graphs (48) and (49) is self-evident. In (55), the position of the dancing figure and plumes is comparable to what is seen in STI (42); in both cases the plumes are positioned to either side of the dancer. Unfortunately, the SHD does not contain an Ancient Clerical equivalent of the character de | § ; however, it does have examples of the 'heart' matrogram. The shape of the matrogram  used in SHD (51) bears a fairly strong resemblance to the ^  grapheme found in Small Seal graph (39). Turning our attention to point (C), Ancient Clerical graphs (52) and ma ,H (54) are featured. The part of graph (52) depicting the legs of an elephant is written using strokes of a length that match Small Seal graph (40). By contrast, the legs of the horse in SHD (54) are written partly in the form of strokes and partly using abbreviated dots; therefore, there is some indication in this character of the orthographic change embodied by more mature Clerical script. Certainly, this component of SHD (54) lacks the same stroke length found in SWJZ (43). At the same time, however, neither is it comprised of the kind of abbreviated dots featured in example (60) from the Clerical script of the early Han period. As much as some orthographic affinity is inevitable among scripts belonging to the 43  same orthographic lineage, the different motivation underlying their creation and use can also be expected to manifest some degree of orthographic dissimilarity. In the case of Small Seal and Ancient Clerical script, this dissimilarity can be defined by certain orthographic features that also play a part in the upcoming graphological study of the FYSJ. For example, although SHD (50) to (57) mostly retain a vertical rectangular shape which betrays their origin in earlier Small Seal script, the horizontal rectangularity characteristic of Clerical script is also evident in them. SHD (52), ji  (53) and (54) are all distinctly wider than Small Seal graphs  (35), (37), (40), (41) and (43). But the most prominent difference between the Small Seal and Ancient Clerical scripts is the relatively high degree of simplification embodied by the latter. By simplification I refer to orthographic features such as an indistinct and cursory manner of writing, reduction in the number of strokes, and characters which lack the same balance or sense of proportion usually found in Small Seal script. Taking all of these orthographic traits into consideration, it should be clear that the diminishment in pictorial quality is the end result of orthographic simplification. For example, the 1~ component in SHD graph (50) is written in a cursory and indistinct fashion compared to the much more clearly articulated 1 in STI (38). And compared to Small Seal STI (40), the hand element in SHD (52) has lost its pictographic quality entirely — i.e., the grapheme (ART' in STI (40) has been replaced by a few horizontal lines. On the whole, SHD (54) resembles a horse much less than Small Seal (43): the horse head in (54) has lost its oval shape and the three long horizontal lines no longer resemble a mane; the legs, tail and body depicted in (43) are reduced to a few short downward strokes. Further, the dancing figure in SHD (55) is less defined than in STI (42). Due to graphic simplification, the dancing figure's head and feet have been eliminated — i.e., compare ^ (55) with ^ (42), and the objects in its hands are more poorly formed. Finally, although SHD (56) and (57) structurally resemble SWJZ (48) and (49), both SHD characters lack the balance and regularity of form which characterize the Small Seal graphs. For an examination of Clerical script from the period post-dating the SHD, we turn our attention towards the MWD bamboo inscriptions (Lizi bian W^^M' 650,471, 937, 23, 1473, 99, 163, 1480; hereafter LZB). Strictly speaking, the manuscript found at MWD is still considered to be written in Ancient Clerical, although it was produced some thirty to forty 44  years after the SHD. Some believe that the MWD inscriptions were written sometime during the reigns of Han emperors Wen SC (r. 179-156 B.C.) and Jing J | (r. 156-140 B.C.) (Zheng: 21). William Boltz, on the other hand, dates the MWD to about 195-87 B.C. (1994: 160-1).  In terms of overall shape, the MWD graphs still reflect some influence from Small Seal script. Most noticeable is the vertical rectangular shape of MWD graphs, such as (58), (59), (64) and especially (65). By contrast, graphs (62) and (63) are both more square in shape. As well, (61) also fills a space that is square in form, as opposed to vertically rectangular. Consequently, it would seem that the Ancient Clerical script used in the Warring States SHD bamboo inscriptions and the MWD documents shows about the same degree of vertically. This may suggest that until Clerical script evolved into its more mature ba fen form at about the time of Han emperor Wudi, all Ancient Clerical script showed signs of verticalization to an equal extent. Naturally, such a general claim can only be made with confidence if more graphs from SHD and MWD are compared, something outside the scope of the present study. However, MWD Clerical script definitely shows a higher degree of simplification than the form of this script seen in the SHD. On the whole, the MWD script is written in a more cursory style, displays a greater reduction in the number of strokes, and exhibits a tendency towards converting lines to dots. We can make the following comparisons: SHD (50) :: MWD (58): Although the orthographic structure of the two graphs is similar, (58) is written in a more cursory fashion — e.g., compare the X component in (58) with \~ in (50). Cursoriness also appears in the graphic representation of a 'head' found in MWD (58) — i.e., in (58) the two horizontal lines have become shortened and less horizontal. SHD (51):: MWD (59): Compared to the relatively pictographic 'heart' matrogram in (51), 45  the same element in (59) has been drastically abbreviated, losing its pictorial quality completely — e.g., the  component in SHD (51) is written as  in MWD (59).  SHD (52):: MWD (60): Graph (60) has fewer strokes than its SHD counterpart and the 'feet' are written using small dots instead of strokes as seen in (52). Consequently, (60) has become simplified to the point of losing much of its pictorial quality. SHD (55) :: MWD (63): Graph (63) has undergone a higher degree of simplification than (55), making it much less pictographic. In fact, the simplified structure of (63) closely resembles Six States graph (15). In both cases, the graph is divided into an upper part formed by two long horizontal lines and four short vertical lines, and a lower portion comprised of matching twin graphemes — e.g., compare MWD graph (63) with Six States graph (15) SHD (57):: MWD (65): Although these graphs are largely the same, an increased simplification is manifest in (65), with its use of dots in the writing of the wu % component. Conclusion: This chapter was designed to introduce non-specialists with a background in Chinese to the analysis of graphic, lexical and orthographic variation. It is intended to give the reader a broad understanding of how Old Chinese phonology, semantic value and graphic representation interrelate in pre-Han Chinese. Further, chapter one provides the methodological background required to distinguish different types of characters and the various orthographic features that comprise Small Seal and Ancient Clerical script. Hopefully, this detailed introduction will enable the reader to follow the arguments proposed in chapters two and three more easily. Ideally, it should make the reader familiar enough with textual criticism and the fundamentals of the study of script to challenge some of the assertions put forth in this thesis.  46  Chapter Two: A Selective Graphological Study of the Fuyang Shi jing Introduction:  Chapter two continues where the first chapter ended. Having provided a functional linear outline of the orthography of Small Seal and Ancient Clerical, in this chapter I put this understanding of these scripts to the test, using them to evaluate the graphs and orthographic forms comprising the FYSJ. Whenever necessary I draw upon the general principles of orthographic evolution presented in the first chapter. At the same time, in the course of conducting graphological comparison and analysis in the second chapter, our understanding of phenomena such as the relationship between graphs and their orthographic forms will doubtlessly be enhanced. This chapter focuses on the following objectives: (A) I attempt to distinguish the physical nature of the FYSJ and show what the implications of my findings are. We already know that the FYSJ is comprised of different source manuscripts (Wen: 978); however, the full extent to which the FYSJ is divisible into pieces has yet to be determined. (B) I date select graphs, and the context they are part of, to determine the full chronological range of the various manuscripts in the FYSJ. While a rough date based on cursory analysis has already been assigned to the FYSJ (Wen: 97-8), a more detailed evaluation of this matter is definitely needed. Before the graphological material provided by the FYSJ can be used for further research, it is imperative that we know exactly when it was written. In addition to this introduction, chapter two is divided into three graphological studies, each dealing with one a particular 'feature' character and its various orthographic forms. These studies are themselves divided into four parts. The first part involves componential analysis in which the various graphemes or structural components found in the feature character are examined. I believe that performing this kind of analysis at the outset facilitates whatever orthographic analysis comes next. Labeling the graphemes that I work with enables me to render more lucid descriptions and provide cogent explanations. Moreover, with a clear picture of how each character is constructed, the reader should be able to follow the identification and comparison of its orthographic forms more easily. For example, the componential profile given the graph dao H in chapter one allows discussion of its orthographic forms to be more easily understood. Below are six different forms of dao Hf 47  from the first chapter:  # $ « 1 1  #  Comparison of these graphs is greatly facilitated if what is referred to by the terms 'head', 'street' grapheme and 'foot' grapheme is known. The phonetic and semantic information related to the words represented by FYSJ graphs will generally not be included in the componential study. Since the analysis in chapter two is graphological in orientation, information on the possible or known meanings of lexical items lies outside our consideration here. However, on occasion exceptions will be made: if the meaning of a word affiliated with a particular graph helps substantiate conclusions derived from componential evaluation, it may be included for reference. Following the componential analysis, in each graphological study found in chapter two I briefly survey the orthographic evolution of the feature character. Sometimes the graphs in this survey aid the preceding delineation of graphic components. But this is not what is most important. By providing historical background on the orthography of the feature characters, I hope to better explain the manner in which they embody Small Seal and Ancient Clerical orthographic traits. If the graphs in the FYSJ are to be successfully dated, it is necessary to establish reference points in the form of specific traits marking early and late Ancient Clerical orthography. For this purpose, 1 primarily utilize examples from the SHD and MWD texts, although other materials will also be referred to occasionally. As mentioned in chapter one, the script seen in the SHD bamboo inscriptions is thought to have been written around 240-217 B.C. (Zheng: 33). Conversely, the Ancient Clerical script found in the MWD texts represents a form which was current from 195 to 187 B.C. (Boltz: 1994: 160-1). Our knowledge of script is still too limited to date particular Ancient Clerical graphs by the year or even by the decade; however, the SHD and MWD provide benchmarks with which to determine the earliness or lateness of FYSJ graphs. Part three of each graphological study starts to address the first primary objective of chapter two; that is, to demonstrate the degree to which the FYSJ is actually a highly fragmented patchwork of numerous different manuscripts. Unfortunately, the two pioneers 48  of FYSJ research, Hu Pingsheng and Han Ziqiang, completely ignore this aspect of the FYSJ in their evaluation of it (Hu, 1988: 23-35). It is doubly unfortunate, therefore, that they still proceed to speculate which school of Shijing scholarship the FYSJ represents (Hu: 28-31). Surely this question can only be approached after a better understanding of the physical nature of the FYSJ is realized? Wen Xinfu has also conducted preliminary work on this important facet of FYSJ research; however, Wen's analysis is far too approximate and limited in scope. The data derivedfromhis calligraphic analysis is not accurate because it is based on the consideration of entire groups of odes lumped together. And his orthographic analysis is severely compromised by the paucity of examples examined (97-8). More on this later. In order to properly discern the miscellany of manuscripts in the FYSJ, I believe the examination should focus on specific graphs and individual bamboo slips. Moreover, the accuracy and reliability of the findings depends on taking into account a large number of graphs. One can view part three as responding to the first objective on a comparative basis; whereas, as will be explained shortly, part four performs the same task concentrating on individual cases. In each of the three graphological studies in this chapter, the feature character selected recurs numerous times within the FYSJ. The various forms of the feature character are compared in part three as a means of distinguishing the manuscripts in the FYSJ. In this regard calligraphic style is perhaps secondary in importance to orthographic criteria. Yet calligraphy should still be considered an integral part of the graphological methodology used in this chapter. The initial examination of orthographic forms in part three will also involve a dating process, albeit on a somewhat cursory level. I will date the various orthographic forms of each feature character independently; that is, without verifying the dates I give them against the immediate context. The intrinsic value of this exercise lies in formulating a general time-frame in which to place the contents of the FYSJ as a whole. It will give us a relatively clear idea of the chronological range to be found among the manuscripts that comprise the FYSJ. However, the cursory dating process discussed here also highlights and affirms the significant disparity between the orthographic forms uncovered for each feature character. It should noted that Wen Xingfu has already dated the FYSJ based on evidence drawn from the different scripts found in it. But once again Wen's research is rough and approximate 49  at best; in fact, it essentially amounts to a single statement: "The style of writing in the fragmented bamboo slips of the Fuyang Shijing likely belongs to early Clerical script, but some can be called Eight Parts script and it [this manuscript] probably still contains traces of seal script" ^nmn^sc^mm,  mwmmzAfrw,  mmw-m&m,  mnm^^mm  iii (Wen: 97). Once again my criticism of this approach to graphology concerns the unit of analysis employed by Wen Xingfu. Rather than characterize the FYSJ in terms of different scripts (shu  a category simply too broad to be very meaningful, in my opinion this  particular manuscript should be examined on the level of orthographic form. Failure to do so ensures that the physical nature of the FYSJ will only be superficially understood, something which may be acceptable from the vantage point of Wen's general research, but not in a fairly specialized study such as this thesis. To portray the physical nature of the FYSJ in more precise terms, the varying orthographic forms implicit in early Clerical script must be taken into account. For example, although both the SFTD and MWD use Ancient Clerical script, comparison between the two clearly shows that they differ orthographically. I will refrain from citing examples of this here; however, in the course of the present graphological study this point will be proven irrefutably. Further, neither of these excavated materials contains just one form of Ancient Clerical script; to some extent the different texts contained in them are all distinguished by a different orthographic form. A cursory glance through the Lizi bian (A Compilation of Graphs in Clerical Script; hereafter LZB) and Qin jian wenzi bian (A Compilation of Graphs from Qin Bamboo Inscriptions; hereafter QJWZB) suffices to confirm this claim. Not acknowledging the nuances of orthographic form hinders accurate evaluation of the physical nature of the FYSJ. In passing, it warrants mention that Wen Xingfu's statement regarding the make-up of the FYSJ contains an important error. It is highly doubtful that the FYSJ — buried in the fifteenth year of the reign of Emperor Xiaowen  (165 B.C.) — contains  characters written in the so-called Eight Parts variety of Clerical script (Hu: 23; Wen: 60). As mentioned in chapter one, according to Qiu Xigui this type of Clerical script did not really take form until the Eastern Han dynasty (Qiu: 79). At this juncture, I will briefly compare the SHD and FYSJ. I do this for a number of reasons. Firstly, the basic physical nature of the FYSJ will become more apparent as a result of being juxtaposed against the SHD. This will give the reader a clear impression of the FYSJ 50  composite manuscript before further substantiation of its exact form is provided in the upcoming graphological studies. Further, this comparison exposes the reader to the type of methodology utilized in chapter two, as well as the main resource materials referred to there. As mentioned, the SHD contains disparate orthographic forms of Ancient Clerical script. Some of the forms represent a less mature form of Ancient Clerical script; while others are written in a fashion indicative of a mature form closer to the type found in the MWD. This dissimilarity can be attributed to the fact there are twenty-five different texts within the SHD written at different times in the state of Qin (Qiu: 61). For example:  a ) t _ ^  (  2  )  ^  (  3  )  ^  (  4  )  ^  (  5  )  i _ .  (6)0 • (7)^m*. The orthographic disparity between most of the graphs above is self-evident. Examples (1) to (3) (QJWZB: 448) all show the character chu ft 'go out, come out' (GSR: no. 496). Karlgren believes this character "shows a 'foot' going outfroman area indicated by a curved stroke" (GSR: no. 496). In this case, the area referred to may be a pit or hole of some kind. However, others think the OBI form of chu {£} depicts a foot inside some form of footwear (KMJS: no. 775; for another interpretation see SWJZZ: 6:3b) — e.g., see OBI graph (4) (JGWB: no. 775). Of these examples, (1) is from the SHD text Xiao Iii $ # ; (2) is from SHD text Feng zhen shi  and (3) is from the writing found in the text entitled Xiao  These graphs  are not distinguishable on the basis of calligraphy alone, that is, their disparity does not simply reflect a personal or regional writing style. A distinction based on orthographic feature is undeniable. Differences such as the horizontal rectangularity in (2), the increased linearity in (3) and the relative height and roundness in (1) are all examples of orthographic criteria discussed earlier in chapter one. Examples (5) to (7) are all different orthographic forms of the graph zhi;_ 'go to' written in Ancient Clerical script (GSR: no. 962). According to Li Xiaoding 51  the  original OBI form of zhi  likely depicts a foot positioned above or on top of the ground (fe  A M f t ^ - t ^ - ^ ) (KMJS: no. 771; for a second view see SWJZZ: 6:2b). These examples are from the following SHD texts: (5) is from Wei li zhi dao fe^iZM.; (6) is from Xiao lit; and (7) is from Nei shi za p ^ H . The orthographic features distinguishing them include: (A) the linearized strokes implicit in (6), which I think evolved as part of a process of simplification from the more rounded lines seen in (5); (B) the conversion of the fork-like orthographic configuration distinguishing (5) and (6) to the abbreviated form lacking a straight vertical stem seen in (7). It is probably safe to conclude that the orthographic form in each SHD text is relatively consistent. While some texts are on the whole written in a comparatively early form of Ancient Clerical script, others are written in a form reflecting a later date. I do not intend to embark on a lengthy examination of the SHD Ancient Clerical forms here. Such an undertaking would require a study all its own. However, the three sets of examples given below certainly seem to support this claim.  (8)  (12)  (15)  ±_ ± * (13)  f €B (16)  ¥  (17)  (14)  iS  I  (18)  ¥  £p (19) f  (21)^^^(22)^^(23)^^(24)  __p (20)  /  fS^  As is evident, the chu ft graphs in SHD (8) to (10) (QJWZB: 448) are all very uniform in shape. Moreover, they all closely resemble graph (1) seen above, exhibiting the same shallow 52  U-shaped strokes and height. These four graphs are all from the SHD text entitled Xiao and together they substantiate the claim of orthographic consistency among the different SHD texts. The same can be said for the zhi 2. graphs (11) to (14) (QJWZB: 438). There is considerable consistency among these four graphs, all which are from the SHD text entitled Cang Iii  Only (13) seems to stand out somewhat from the other examples of zhi ;_, as  it is slightly narrower in form. In the second row, SHD (15) through to (20) (QJWZB: 1007) are all examples of the graph jia ^ 'shell' (GSR: no. 626). Although OBI specialists interpret this character differently (KMJS: no. 1674), the SWJZ shows that in the past jia ^ was thought to depict a human skull ( A § B _ - ^ ^ P ) (SWJZZ: 14:19b). There is a high degree of orthographic consistency in the two sets of graphs seen here: the square 'head' in the first three graphs from the text called Xiao lu is not as rectangular as in the second set, which is from the Fa lu da wen. In the third row, examples (21) and (22) (QJWZB: 376), which are from the text Cang Iii, are again very consistent with one another orthographically. The character shown here is shi ^ 'eat, eclipse, food' and according to Karlgren it is "a drawing of a food vessel with lid " (GSR: no. 921; also see Li Xiaoding in KMJS: no. 662). Most significantly, both (21) and (22) are distinguished by their vertical rectangularity. The same orthographic consistency is apparent in the set comprising (23) and (24) from Qin Iii za chao  as well. Graphs  (8) to (24) certainly suggest that orthographic form in the SHD is to a large extent particular to each text. However, while this type of orthographic consistency appears to be applicable as a general rule for the SHD, it should be noted that this consistency can be influenced by calligraphic style and to some extent the structure of individual graphs. Like the SHD, the manuscript of the FYSJ also encompasses different orthographic forms of Ancient Clerical script; however, by comparison the degree of orthographic disparity in the FYSJ is much greater. Overall, while the FYSJ contains orthographic forms of Ancient Clerical script that are as early as those in the SHD, at the same time, there are also FYSJ forms which represent the Ancient Clerical script written in the early Han dynasty. The greatest difficulty a graphologist will encounter examining the composite manuscript of the FYSJ is its lack of any clear demarcation with which to discern the various source manuscripts. With the SHD we know the orthographic forms can be approached and 53  analyzed text by text; however, when dealing with the FYSJ the graphologist must approach the subject matter by examining graphs and bamboo slips individually. In the second chapter, FYSJ graphs are first evaluated in part three of each section, while the broader graphological context provided by bamboo slips does not receive treatment until part four. The two sets of graphs below indicate how different the calligraphic and orthographic forms for any specific character can be in the FYSJ. The first set depicts five graphic forms of the character zhi 2. (S006, S023, S079, S085, S093; the numbers assigned FYSJ slips in this study accord with Hu: 2-18 and Wen: 61-91), which has already been discussed in connection with its SHD forms. The second set shows five different forms of the character wo  'first-  person pronoun' (GSR: no. 2) (S001, S030, S059, S081, SI42).  (30) The first set above presents several distinct orthographic forms of the character zhi;_. For example, FYSJ graphs (28) and (29) clearly resemble one another orthographically, although the quality of the strokes — i.e., length and boldness — indicate calligraphic disparity. As such, we may conclude that while FYSJ (28) and (29) were likely written during roughly the same time period, they still represent different manuscripts. I will not evaluate these graphs in any detail here. Suffice it to say that (28) and (29) can be examined in connection with SHD (7); they exhibit the same orthographic features as seen earlier in the description of (7). Examples (25) to (27) are orthographically further removed from (5), the earliest SHD form of this graph, than the other FYSJ graphs. This can be attributed to the fact that they represent a later orthographic form. By most accounts the character wo  is a pictograph of some sort of weapon with a  saw-like blade (KMJS: no. 1456). In the second row, we see tremendous orthographic disparity among the examples of wo  provided. For example, FYSJ (32) is much sparser 54  than any of the other forms of wo 3% found in this manuscript. In fact, this graph is remarkably similar to the OBI form of wo  seen in example (35) (Gao: 350) and the BRI  graph form the Spring and Autumn period in (36) (Gao: 350). The wo  in (35) and (36)  probably represents the oldest orthographic form of this character (Gao: 350), and the fact that it is still found in the FYSJ should be of considerable interest to anyone studying the history of orthography in early Chinese. Moreover, FYSJ (30) is distinguished orthographically by its short lines, an additional horizontal stroke and a vertically rectangular shape. FYSJ (33) is distinguished by its width and an additional short stroke in the upper right hand corner which is absent in the other examples. The brief examination of FYSJ (25) to (34) demonstrates the same kind of analysis I conduct in part three of each graphological study in chapter two. But in the main body of the second chapter — as opposed to this overview — I will make a greater effort to date each of the orthographic forms analyzed. To accomplish this I will compare the FYSJ characters in their various orthographic forms with corresponding SHD and MWD graphs. An example of this dating process will now be provided using the graphs given below:  (37) -?~(38) T * ( 3 9 )  (40)«^(41)  r%(42)  f  Two FYSJ forms of su % 'white silk' (GSR: no. 68) — i.e., (37) (S061) and (38) (Sl 16) — are at the center of this dating procedure, an abbreviated form of the process I apply to FYSJ graphs later on. Comparatively speaking, FYSJ (38) more strongly resembles SWJZ Small Seal graph (39) (Gao: 237) than it does FYSJ (37). The two loops in the si & 'silk' (GSR: no. 876) element of (38), located in the lower part of su fji, are horizontal and oval shaped, so that they roughly match the Small Seal form. Further, the bent horizontal lines in the upper portion of FYSJ (38) — representing the "flowering buds" on the stem of a flower (JMJWCD: 270) — also bear an orthographic affinity to the Small Seal form of the character su f£. All this orthographic evidence suggests that FYSJ (38) is a relatively early form of su H?, something which is substantiated by the fact that in the Warring States period the strokes representing flower buds in this graph had already undergone linearization. The QJWZB does 55  not contain an example of su fjl; however, under the entry for biao |g 'exterior garments, the outside (of dress)' (GSR: no. 1153) it does have an example of the this element (QJWZB: 635) — i.e., SHD example (42) clearly shows horizontal strokes that are no longer bent as in the Small Seal. By contrast, FYSJ (37) more strongly resembles Ancient Clerical forms of the graph su H written in the early Han dynasty; the loops in the si & grapheme lack horizontal orientation and the depiction of the budding flower is distinctly linear in shape. These traits make (37) comparable to the MWD graph (40) current during the early years of the Western Han dynasty and (41) dated to a period some forty years later. Naturally, much could also be written concerning the orthographic variation of the MWD, another major resource for the graphological analysis of the FYSJ. However, I will not comment on the MWD in as much detail as the SHD and FYSJ. It has already been noted that the different texts in the MWD manuscript are all written in Ancient Clerical script, but that there is some variation in orthographic form among them (Shaughnessy: 49). The same conclusion is attainable just by looking at the examples of MWD graphs found in the LZB. Below I cite several examples:  (43)  (44) ^ - * F  (45) 7  (46)  I  (47)  f ^ ( 4 8 )  \  The character represented in MWD examples (43) and (44) (LZB: 920) is fang -fj 'square, region, everywhere' (GSR: no. 740). Although there is some disagreement regarding what is depicted by this graph, Xu Zhongshu's interpretation is both frequently cited and corroborated by other paleographers (for other interpretations offang  see GSR, no. 740,  SWJZZ: 8:7b and Wang Yun _E2§ in JGWZGL: no. 3119; for views which confirm this interpretation see KMJS: no. 1057). Xu Zhongshan believes that fang  was originally a  pictograph of a lei ^ 'plough' (GSR: no. 578), depicting a stem and handle bar and the two shears of its plowshare C f r _ & * , Jh*_„S'fl§„  , TM^W^MMA  , ^M'MMMW  MX) (JGWZGL: no. 3119, JMJWCD: 67). Certainly, the shape of a plough seems discernible in the form of OBI graph (45) (JGWB: no. 1057). 56  In MWD examples (46) and (47) the character shown is hu ^ 'final particle of exclamation and interrogation' (GSR: no. 55). The SWJZ states that this character depicts sound raising upwards (IMi-tlifi£.0i!l) (SWJZZ: 5:32a) and, as is the case with fang jj, the OBI form of this character seems to support this view — refer to example (48) (JGWB: no. 599). In both sets of MWD examples the difference in orthography is self-evident. What the above examples do not clearly show, however, is the tendency of characters in MWD script to be more vertically rectangular than is the norm in SHD script. Having commented on the physical nature of the FYSJ and demonstrated the methodology I will use to date particular graphs, all that remains is to explain the tasks conducted in part four of each graphological study. These tasks are relatively straightforward, for the most part involving the same type of dating procedure discussed above, but with the emphasis shifting from individual graphs to a larger context that includes a number of bamboo slips. At the outset of part four, I select one FYSJ graph from those examined in part three. I then conduct a broader and more thorough investigation of this graph and the area of the FYSJ it can be said to represent. In order to ascertain the correct date of any graph in the FYSJ, it must be compared to other graphs located on the same bamboo slip. Only when orthographic or calligraphic consistency among all the graphs on a slip is established can we be certain that the date given those graphs is accurate. The next phase of part four involves establishing a group of slips which belong to the same source manuscript. This endeavor entails the same kind of comparative procedure undertaken when dating the graphs on individual slips, only the context and number of graphs examined increases in size and number. For this part, I will compare the slip dated in the initial phase of part four to the slips preceding and following it in the FYSJ, looking for consistency in orthographic form and calligraphic style. Using these criteria I analyze as many slips as necessary to establish a complete group of one source manuscript.  57  Section 1): A Graphological Study of ma H 'horse' (GSR: no. 40) and FYSJ slips S121 to S124  I begin this study of the character ma H 'horse' with a brief analysis of the components which comprise it. A basic understanding of this character's makeup will facilitate orthographic analysis of it, something which is crucial to the process of dating. In the preceding chapter, the pictographic nature of ma |§ was already discussed; however, at this point I will review several of the points that were previously made. As evident in example (1) (JGWB: no. 1152) below, the OBI form of ma | § exhibits a very high degree of pictographic quality: the horse's head, mane, legs, body and tail are all clearly visible in the layout of the graph. Xu Shen's observations regarding the Small Seal form of ma H (7) (Gao: 192) substantiate this assertion. The SWJZ reads: "[This pictograph] depicts the shape of a horse's head, tail of hair and four legs" ^MMMm^m  (SWJZZ: 10: la).  Although Xu Shen never saw the OBI form of ma H , he did have access to forms of this character which antedated Small Seal (SWJZZ: 10:1a). For example, his discussion of ma H in the SWJZ contains the gu wen  'ancient script' graph seen in example (8) (SWJZZ:  10: la). The anatomical parts of the horse are more clearly discernible in (8) than in (7) — i.e., the depiction of the head and mane do not overlap and the grapheme representing the 'tail' stands out from the body, making it less ambiguous. Below, written in various different scripts, are a number of examples of the character ma H and other characters that contain the horse element:  The orthodox Seal and Small Seal forms of ma JH, seen in graphs SD (4) (SKZWB: 10:1; JSZD: 1464), SD (5) (SKZWB: 10:2; JSZD: 1467), SWJZ (7) and (9) (Gao: 194) exhibit enough similarity to constitute a particular orthographic strain. For example, vertical rectangularity is a salient feature in the orthography of all these graphs. As mentioned in chapter one, vertical orientation became progressively more prevalent in the orthography of Seal script from the fifth-century to the second-century B.C. This particular characteristic of Seal script is highlighted with a comparison between graph (5) from the Spring and Autumn period and SWJZ graph (9) (c. 100 A.D.). Example (5), the orthodox Seal SD form of the charactery/a M 'to yoke, yoked carriage' (GSR: no. 15), lacks vertical orientation in its form; the jiaftpelement, comprised of //' ji representing an 'arm' and kou P 'mouth' (GSR: nos. 15, 928), is positioned at the side of the ma Mt element, making the teknograph as a whole more box-like in shape. However, the descendant SWJZ Small Seal form ofjia %, which reflects the standard Small Seal form of the Qin dynasty, indicates that by the second-century i  B.C. the Small Seal form of this character had already undergone significant verticalization. The jia Jpf element in SWJZ Small Seal (9) has shifted from the side of the character to the top, creating a more vertically rectangular shape. It warrants note that an appreciable discrepancy in the vertical rectangularity of preQin Seal and Small Seal graphs is quite common. Ajmong the examples of pre-Qin Seal and Small Seal graphs examined in chapter one, there are numerous cases where such a discrepancy exists. In chapter one, graphs (18), (19), (21), (23), (25) show the broader-shape of pre-Qin Seal script particularly well. Examples (4), (5), (7) and (9) above also exhibit several other features which are characteristic of Seal script. These features include a high degree of orthographic uniformity, a sense of balance and proportion, and clearly articulated strokes which are curved and rounded at the ends. As previously discussed in chapter one, compared to the corresponding OBI graph, the Small Seal form of ma ^ is clearly less pictographic — e.g., compare OBI example (1) and SWJZ (7). That being said, the likeness of a horse is still visible in the Small Seal form, albeit in a more symbolic sense. To be specific, in the upper portion the horse's head and mane have transformed into a vertical oval shape with three horizontal lines passing through it and extending out to the right. In the lower portion there are two downward-sloping strokes 59  which represent legs, and a fork-shaped configuration representing the horse's tail. The pictographic quality of ma H seen in pre-Qin Seal and Small Seal script also appears quite frequently among scripts outside this Qin orthographic strain. Example (2) (JWB: no. 1286) is from an inscription on a bronze vessel dated to the middle of the Zhou dynasty (Gao: 192); whereas, (10) (Gao: 192) isfromthe Warring States period. Aside from minor calligraphic differences — lines that are slightly straighter, shorter or longer — there is no major orthographic feature distinguishing (2) and (10)fromtheir Small Seal counterparts. To fully appreciate the pictographic quality seen in these graphs one need only view the orthographic forms ma | § takes in the simplified script written during the second half of the Zhou dynasty, what has been referred to as Six States script. BRI graph (3) (JWB: no. 1286) from the Spring and Autumn period and (6) (Gao: 192) from the Warring States period are good examples of this script. Through extreme orthographic abbreviation and linearization both of these simplified forms ma H lack any trace of pictographic quality. Ancient Clerical script is the simplified offspring of the same orthographic lineage that produced Small Seal script; therefore, it is not surprising that these scripts share several features in common (Qiu: 69). For example, graphs written in Ancient Clerical script often retain the vertical rectangularity of Small Seal script, although to varying degrees. Examples of this from the preceding chapter include: SHD (50), (55), (56) and (57) and MWD (58), (59), (64) and (65). (As will be demonstrated repeatedly in the course of this chapter, the earliest form of Ancient Clerical script from the Warring States period is generally less vertical than the form of the same script written during the early Han.) Further, the two scripts show considerable structural similarity; that is, graphemes and other graphic elements are generally positioned in the same place within the structure of any given graph — i.e., compare SWJZ Small Seal (9) with SHD Ancient Clerical (15) (QJWZB: 722). This point also was made in chapter one — e.g., refer to SHD graphs (50), (52), (55), (56), (57). For the most part, Ancient Clerical graphs tend to resemble their Small Seal predecessors more closely than corresponding graphs written in Six States script. However, at the same time, Ancient Clerical script was without doubt subject to the same forces of simplification and other types of orthographic change as Six States script. The tendency for strokes to be converted into dots in Ancient Clerical script is a very salient manifestation of orthographic evolution, one that will be illustrated on numerous occasions in this chapter. 60  Moreover, characters written in Ancient Clerical script are typically more angular in shape and exhibit a greater degree of cursoriness — i.e., a 'slapdash' or 'slipshod' writing style — than their Small Seal counterparts. This cursoriness often translates into a lack of orthographic balance and proportion in terms of how components are positioned within a larger graphic structure. As I utilize the Small Seal and Ancient Clerical scripts for the purpose of dating the FYSJ, these basic differences will no doubt serve a crucial function. The Ancient Clerical form of ma ,H shares a number of similarities with its Small Seal predecessor. Several typical Small Seal characteristics are discernible in SHD graph (12) (QJWZB: 719), such as the oval shape of the horse's head and the medium-length horizontal lines which run through it. Further, just like the Small Seal orthographic forms of the character ma H , the lines representing the legs and tail of a horse also run down the side of SHD (12). To a slightly lesser degree, SHD (13) (QJWZB: 719) also exhibits signs of Small Seal orthography: the horse head depicted is not oval, but longish horizontal lines cross through it and extend to theright.The longish lines portraying the horse's tail and legs in example SHD (11) (QJWZB: 719) also show the influence of Small Seal orthography. Moreover, this graph's vertical rectangularity does the same. In this case, the verticality is facilitated by the manner in which the strokes representing the horse's mane do not extend to the far right like (12) and (13). Based on this orthographic analysis, it seems evident that of the three SHD ma Mi forms (11) represents the oldest variation. SHD graphs (11), (12) and (13) contrast Small Seal in a number of significant ways, as well. For example, the pictorial quality in all three graphs is diminished— i.e., the tail of the horse is indistinct, as no longer is it written in the fork-shaped configuration seen in the Small Seal form. SHD (12) and (13) also manifest a diminished vertical rectangularity. In both cases this diminishment is chiefly brought about by the three long horizontal strokes crossing through the head of the horse. The same orthographic feature is also plainly evident in the SHD forms of the character jia M seen in examples (14) and especially (15). These elongated strokes not only create a horizontalness in the structure of these four SHD graphs. Their length also creates a sense of orthographic imbalance. More SHD ma ,H graphs take the form seen in (13) than any other (QJWZB: 719, 722,723). This SHD graph contains typical Ancient Clerical traits, like an angular upper 61  structure — i.e., the part of the graph representing the head and mane of a horse, a shape that is horizontally rectangular, and detached or abbreviated leg strokes. Graph (13) shows signs of the Ancient Clerical tendency to convert strokes into dots, something which is much more common in later forms of this script. Finally, like all SHD graphs for ma J§ (13) is written in the seemingly careless or slapdash fashion which is the hallmark of Warring States Ancient Clerical script. However, as will be seen later in this study, the slapdash writing style characterizing in this script is more evident still in other characters. I will now shift attention to examples of ma H from the MWD manuscripts. Overall, MWD graphs (16), (17) (LZB: 1473) and MWD (18) (LZB: 1478) show characteristics of a more mature Ancient Clerical orthography than the SHD graphs examined above. Of course, such orthographic maturity is to be expected from a manuscript which was written several decades after the pre-Qin dynasty SHD. For example, MWD (16) and (18) — in which ma ,H is the semantic determiner in the teknogram can f§| 'three-horse team' (GSR: no. 647) — both exhibit the abbreviated dotted strokes typifying more mature forms of Ancient Clerical script — i.e., compare these MWD graphs with SHD (12) and (15). Moreover, unlike the SHD, none of the MWD ma H graphs feature an oval-shaped horse head, use longish lines to depict the mane or have strokes representing a horse's legs which run down the side of the graph. In addition, MWD graphs (16), (17) and (18) typically possess a sharpened angularity, as well. Overall, it is easy to characterize MWD script as forming the transition between preQin dynasty Ancient Clerical and the more mature Clerical script of the Han dynasty. The MWD form of ma Mi is just one example of this. Graphs (19) and (20) (LZB: 1473) — from the Juyan  (hereafter JY) (c. 125-104 B.C.) and Fenghuang shan JUSUJ (hereafter FH)  (c. 163-150 B.C.) inscriptions, respectively — demonstrate how the ma Mt graph evolved in Clerical script after the early Han dynasty (Zheng: 23). Note the continued use of dots instead of lines and an increasingly sharp angularity in these graphs. The following graphs are from the FYSJ:  The FYSJ contains several different forms of ma Mr- In some of the above examples, ma H stands alone as a monosomatic character, or matrograph — e.g., FYSJ (21) (S001) and (25) (122). In other examples the ma | § graph is part of a teknogram that includes additional components. For example, in FYSJ graph (22) (SO 18) ma |§ is part of the character zou §j@ 'one who raises horse, a horse attendant' (GHYZD: 339); in FYSJ (27) (S124) ma Jig is in the character^ g$ 'post horse, relay horse' (GSR: no. 790); and in FYSJ (24) (SI 14) ma H is found in sao ft 'chaotic, unsettled' (GHYZD: 212). In FYSJ (26) (124) ma if is also positioned to the left of another graphic element; however, that element is missing from the inscription due to damage in the bamboo fragment. Of course, in FYSJ (23) (S073) ma Mr is also part of a larger graphic structure; it combines with the phonophoric jia Jp to form the characterjia %. Comparison of graphs (21) to (27) clearly proves that the FYSJ is a patchwork of many different manuscripts. In these graphs there is a wide assortment of orthographic forms representing both early and late Ancient Clerical writing. A comparison of the ma ,f§ component in FYSJ (22) and (24) clearly illustrates this: while the former is more vertical in form and has longer leg strokes, the latter is written with thicker lines, has shorter leg strokes and is more box-like in shape. Because of its closer orthographic affinity to Small Seal script, it may be supposed that FYSJ (22) is the older of the two — i.e., compare FYSJ (22) to SHD (11). I believe that the ma Mr in FYSJ (23) bears a strong resemblance to MWD Ancient Clerical script. Key to this evaluation is the graph's square shape. More specifically, while the corresponding SHD graph (15) has long horizontal lines that give it more of a horizontal orientation, the shorter horizontal lines in FYSJ (23) do not have the same effect. Another feature that is key in the dating of this graph is the length of its leg strokes, which strongly resemble those seen in FYSJ (24). In fact, only minor calligraphic difference and the fact that (24) exhibits a reduced number of strokes separates the two graphs. Another ma Mr graph which shows some signs of relatively late Ancient Clerical orthography is FYSJ (27) — i.e., notes its short and straight leg strokes. At the same time, however, its vertical orientation seems to suggest an earlier date. I believe that the evidence 63  provided by FYSJ (26) is decisive in this matter. FYSJ (26) is from the same slip as (27), but clearly shows an older orthography — i.e., its legs strokes are long and the tail stroke bends further downward. As such, I propose that FYSJ (26) and (27) represent an orthography from a time somewhere between FYSJ (22) and the two graphs (23) and (24). Something similar can be proposed for FYSJ (21), as well. In view of its vertical rectangularity and has longish legs strokes, (21) is to some degree comparable to (22) and (26). Of all the FYSJ graphs for ma | § , only FYSJ (25) from slip S122 has abbreviated dotted strokes which represent a comparatively late form of Ancient Clerical. As mentioned, this feature began to gain prominence in the MWD inscriptions — e.g., refer to (16) and (18). Consequently, I believe this graph to have been written sometime in the early Han. It is one thing to state that a graph represents a relatively late form of orthography, quite another to actually ascribe it to a later date. If a graph manifesting late or mature orthography is situated in the context of graphs written in comparatively early orthography, it becomes unlikely that the former was actually written at a relatively late point in time. Instead, this type of situation probably reflects orthographic overlay, whereby a lingering or preserved older orthographic form is found side by side with young, developing form. Graphs representing different orthographic forms may co-exist and even be found on the same bamboo slip. One example of orthographic overlay can be found in FYSJ graphs (26) and (27). Although both graphs are found on FYSJ slip 124, the orthography of (27) is more mature making it look like a later form than (26). In order to state with a greater degree of certainty that FYSJ (25) was written during the early Han we must perform certain measures of verification. We must compare this graph to other graphs located in the same immediate context or, as in the case of the FYSJ, on the same bamboo slip. If the other characters written on FYSJ slip S122 also appear written in a mature orthographic form, we may on the basis of orthographic consistency confirm a later date for this particular FYSJ ma MJ graph. The graphs found on FYSJ slip S122 correspond to stanza three of Mao Shi no. 127. This slip reads:  #3Hj?5fcW!3i^ (Hu:  15; Wen: 85), and is translated by Bernhard Karlgren as,  "... the four horses are well-trained; the light carriages..." (1944: 211). Unfortunately, the last three graphs on this slip are incomplete due tofragmentationin the bamboo material. This 64  restricts our analysis of FYSJ slip S122 to the graphs si $3 iadle' (GSR: no. 518; it is also a jiajie  loan character for 'four') and ji KJfc 'to complete, finish, exhaust, all, entirely' (GSR: no.  515) which precede and follow ma |§. Furthermore, the QJWZB and LZB both lack the characters/ $3 in Ancient Clerical form. Consequently, we turn our attention towards an analysis of the FYSJ graph ji KI to determine if it shows signs of mature Ancient Clerical orthography. The graph ji KI from FYSJ slip SI 22 is given below in (28). Corresponding graphs from the SHD and MWD texts are also provided:  Before comparing and evaluating the examples of ji Kjfc given above, a few words concerning the components which comprise this compound graph are in order. Karlgren believes that this graph "shows a kneeling person and a food vessel" (GSR: no. 515). Some paleographers, however, have been more explicit than this in their interpretations. For example, one writes, "[The characterji Jljfc] depicts a person facing food that has been cooked, but his mouth is not turned towards the food. This indicates that the person is full and is about to [get up off his lmees and] leave" ^ - f i A ^ ^ K M f i ^ f _ W ^ f ^ I * m r l 7  , IfeA^MAT  (JGWZGL: no. 338; this view is endorsed in the summary of no. 338 found on p. 381). Using the OBI form of// K5£ seen in (29) (JGWB: no. 657) for reference, its two main components are clearly in keeping with the above explanation. The component on the left represents a food vessel with a lid and maybe a stand of some sort (GSR: no. 941); whereas, the component on the right depicts a human form on his knees and with his mouth pointed away from the vessel. The SD Seal form of this graph seen in (30) (SKZWB:5:16; JSZD: 585), as well as 65  Small Seal STI (31) (SKZWB:5:16), both strongly resemble the OBI graph in terms of pictorial quality. This demonstrates how little the orthography of certain characters changed over the roughly five hundred years from the late Shang to the Spring and Autumn period. By contrast, the Warring States SHD form of//  seen in (32) and (33) (QJWZB: 375) are  somewhat more simplified and noticeably less vertically rectangular. While the 'food vessel' is much shorter in height, it contains the same number of strokes seen in the earlier Small Seal form; however, the stroke representing the 'body' in the two SHD graphs has been linearized and reduced in length. Generally speaking, the lines in the SHD 'food vessel' pictograph are rounded, something which may indicate an affinity to Small Seal script. Moreover, on the whole this teknogram lacks a sense o f balance, as the 'food vessel' is small compared to the human form. Although there is a high degree of uniformity in the two SHD ji  graphs, the  structural domination of the human form is especially evident in SHD (33). The strong resemblance marking SHD (32) and SHD (33) can be attributed to the fact that they are from the same SHD text, entitled Wei li zhi dao J ^ £ J I (QJWZB: 1137-42). The box-like shape of MWD graphs (34) and (35) (LZB: 815) resemble the SHD examples. However, in MWD (34) and (35) the 'food vessel' element exhibits a greater sense of balance. Whereas the upper portion of the food vessel in the SHD graphs is larger than the bottom part, in the MWD examples the two parts are even in size. The MWD graphs as a whole are more evenly proportioned than their SHD counterparts; the food vessel component and the human figure are comparatively equal in terms of height and width. There are some other noticeable differences in the orthographic makeup of MWD (34) and (35). For example, the MWD forms are comprised of sharply angular strokes, and the human figure in both MWD examples has been radically transformed: the upper portion of the 'body' looks more like a loop than an open mouth and the rest of the component shows increased signs of symbolification. In the MWD examples, the drawing of a person kneeling that was once depicted with a high degree of pictoriality has become more abstract. Before comparing FYSJ (28) with the other Ancient Clerical forms, it warrants mention that this FYSJ character is clearly a graphic variant of ji Ijfc; that is, while maintaining the same pronunciation and semantic value, its orthographic structure is meaningfully different. In addition to the elements representing a 'food vessel' and 'a kneeling person', FYSJ (28) contains the grapheme zhi  'foot' (GSR: no. 1258). The same graphic 66  component was seen earlier in chapter one when dealing with the character dao M- In the case of ji  however, this variant form must have been exceedingly rare, since none of the  graphological sources used for this study contained another example of it. Of course, the variant FYSJ form ofJj5i is structurally close enough to the historically continuant form seen in the SHD and MWD examples to date it through a process of comparison. To facilitate this process, the additional element in (28) will not be given further consideration. Although FYSJ (28) is slightly damaged and incomplete, from what is visible this graph exhibits the same sense of structural balance characterizing the MWD form. The food vessel to the left is about equal in size to the figure of a kneeling person and the zhi / _ grapheme to the right. The same sense of orthographic balance is evident in the element of the food vessel. Both the stand holding the vessel and the vessel itself are in proportion to one another. This piece of evidence strongly suggests that FYSJ (28) represents a late Ancient Clerical form, and encourages us to assign this graph the same date as FYSJ (25). However, this evaluation is undermined somewhat by the orthography of the human figure which resembles the more pictographic SHD examples. In my attempt to resolve this impasse, I propose the following orthographic principle: the overall orthographic structure of a character conveys more about its form than any single element in its makeup. Since the balance characterizing the overall structure of FYSJ (28) suggests a relatively mature form, I date this graph to a period in time close to the early Han dynasty. Moreover, since the orthographic form of FYSJ (28) appears to be consistent with that of FYSJ (25), we may conclude that FYSJ slip S122 was likely written at a relatively late date. The next step is to compare FYSJ slip S122 with its immediate environment, the bamboo slips that precede and follow it. The object of this exercise is to discover how many slips actually belong to the same manuscript as FYSJ slip S122. FYSJ S121 represents a different ode than S122; namely, stanza one of Mao Shi no. 119. Based on this information and our knowledge of the physical nature of the FYSJ, the likelihood that S121 is part of the same manuscript as SI22 is relatively low. As a whole the FYSJ is so extremely fragmented that we cannot expect two odes as far apart in the Mao Shi as Mao no. 119 (FYSJ S121) and Mao no. 127 (FYSJ S122) to be from the same original source. In fact, the calligraphic style characterizing FYSJ slip S121 supports this conclusion. Although none of the graphs written 67  on this slip are complete, the parts visible are written in a much rounder form than those in slip S122. If the graphs on FYSJ slip S121 were whole it would be prudent to conduct a scrutinous orthographic analysis of them to obtain still more detailed orthographic data. But in light of their incompleteness, I will forego further analysis and propose that FYSJ slips S121 and S122 are from different manuscripts. Calligraphic evidence alone seems strong enough to reach this conclusion. The calligraphy in FYSJ slip S124 is also a decisive factor in concluding that it is not from the same manuscript as S122. Moreover, the orthography of these slips is also quite different, as FYSJ graphs (25) to (27) discussed earlier plainly indicate. However, in regard to slip S123 the same verdict cannot be so easily passed. FYSJ S123 is unlike S121 in several significant ways. Firstly, the Mao ode represented by S123 — stanza one of Mao Shi no. 128 — immediately follows stanza three of Mao Shi no. 127 represented by S122. This nearness increases the possibility that these two slips belong to the same manuscript. On the other hand, although it is difficult to be certain, the calligraphic style characterizing FYSJ slip S123 appears slightly different from that found in S122 — i.e., in S123 the strokes are thicker and there is less space between characters. However, since FYSJ slips S122 and S123 only contain a few graphs a piece, any conclusion based on the marginal calligraphic evidence available would be tenuous at best. FYSJ slip S123 contains three clear and intact graphs, and I believe that a comparison based on orthographic form is therefore warranted. The following graphs will be used for this purpose:  (36)  (41)  68  (52)  (53)  (54)  (55)  (56)  Example (36) (S123) is the FYSJ form of the character wen 3 t 'drawn lines, design' (GSR: no. 475). FYSJ (36) is damaged and parts of it are missing; however, it is still sufficiently intact to use for comparison and analysis. According to Karlgren, this graph "shows a man with tattooing on the breast." This view accords with the SWJZ (^_i_ScX5:) (SWJZZ: 9:21a) and the interpretations of numerous paleographers. One such expert writes, "[The OBI graphs of] wen all show a human body with tattooed drawings ... [OBI forms of wen] are probably pictographs of patterns on a body " ^ W ^ A  ••• _3t#^.feff5  (JGWZGL: no. 3236; for similar interpretations see KMJS: no. 1085). If this is so, SWJZ Small Seal graph (37) (QJWZB: 676) may be described as the figure of a man with a head, legs planted apart and arms stretched out to either side revealing a chest and torso which tapers down to a narrow waist. In both calligraphic and orthographic terms, FYSJ (36) resembles MWD (40) (LZB: 918): they are both written in bold, thick strokes; the stubby stroke representing the figure's head is missing; and the arms slant down on similar angles. By contrast, MWD (39) (LZB: 918) and SHD (38) (QJWZB: 676) both appear to be written in an earlier orthographic form. Overall, the shape of the human figure in these two graphs more closely resembles the Small Seal form seen in SWJZ (37). While the strength of this evidence is weakened somewhat by the paucity of wen ^ graphs available in the QJWZB and LZB, it nevertheless points to the conclusion that FYSJ graph (36) represents a relatively late orthographic form. Consequently, I propose that FYSJ (36) indicates an affinity between FYSJ slips S123 and S122 in terms of date. Example (41), also from slip S123, is a FYSJ form of the character yin Tgj 'mat, bed' (GSR: no. 370). It is comprised of the elements cao Pf (GSR: no. 1052) andyin g, which in Classical Chinese means 'rest upon, rely on' (GSR: no. 370). In turn,;///? @ is comprised of the following orthographic components. (A) the pictographic grapheme wei Q which normally represents an enclosure, but in this case may simply represent the square shape of a 69  mat (e.g. GSR: nos. 425,1094; KMJS: no. 783); (B) the grapheme da fc which usually depicts an adult human figure, but here may instead be a corrupted form of what originally were lines symbolizing a pattern in a mat (GSR: no. 317; SWJZZ: 10: 18b; KMJS: no. 783). The QJWZB and the LZB both lack the graph yin _ ; therefore, in lieu of using yin ® for comparison, I instead rely on graphemes wei • and da fc for this purpose. I shall begin by comparing.y/M „ in FYSJ (41) (S123) with SHD graphs (43) and (44) (QJWZB: 459). Since there does not appear to be any significant difference in height or width between yin 0 in (41) and its SHD counterparts — the FYSJ graph is slightly narrower and more vertical — it would probably be more productive focusing attention on other aspects of orthography. Mirroring STI Small Seal form (42) (SKZWB: 6:17) (ca. 219 B.C.), the extended parts of the da fc grapheme in the SHD graphs stretch out to the sides and bottom corners of the surrounding enclosure. This is especially true in the case of SHD (43), which is found in the SHD text entitled Yu shu g§  The orthographic form of yin H seen in the FYSJ clearly does  not resemble the SHD, instead looking much more like the early Han form manifest in MWD (45) and (46) (LZB: 428). Like the MWD graphs, the inner grapheme in the jwz H teknogram found in FYSJ (41) has been simplified, no longer resembling the appearance of a human figure. As part of this simplification, the bottom two strokes of this element do not contact the bottom corners of the box surrounding it. FYSJ (41) is especially close to MWD (45) in appearance; in both cases the uppermost point of the da fc element does not contact the top of the outer box. In light of its close orthographic affinity with MWD (45) and (46), the characteristics of the yin 0 grapheme in FYSJ (41) support the assertion that slip S123 was written sometime close to the early Han dynasty. Of course, in our comparative analysis of FYSJ (41), dating yin 0 is only half of the exercise. To strengthen the conclusions arrived at so far, the other part of the graph yin must also be examined; namely, the cao Jnj) component. The pre-Qin and Small Seal forms of cao n can be seen in SD (47) (SKZWB: 1:22; also see JSZD: 1087) and SWJZ (48) (Gao: 300), which show the character cao ^ 'grass, plants, herbs' (GSR: no. 1049). The SWJZ 70  form of cao |_, with its single pair of cao  Jnfi  graphemes at the top, clearly demonstrates the  influence that orthographic abbreviation had on complex Seal script. However, in the present study, the similarity between these two graphs is more important than their differences. In both (47) and (48), cao Pf still preserves a high degree of pictographic quality, something which would diminish in the early SHD Ancient Clerical script. In many cases, the cao Pi element in the SHD continued to be fairly pictographic, reflecting the influence that Small Seal script had on the Ancient Clerical orthography of the time. For example, in SHD (49) (QJWZB: 37) — yuan ?g 'trees with rich foliage' (GSR: no. 260) — the cao Pf component is still quite clearly articulated and in the QJWZB there are numerous other examples of this (see QJWZB: 39, 38,41). SHD (49) represents an early orthographic form of cao Pf; however, in the SHD there are more mature orthographic forms, as well. For example, in the SHD a large number of the semantic determiner cao jfnf are written in a more slapdash fashion — e.g., SHD (50) je M 'leaf (GSR: no. 633). This later form of cao Pf is continued in the MWD. But in an equal number of cases, the cao Pf signifier in MWD graphs resembles that seen in FYSJ (41). That is, they are somewhat symmetrical in appearance and embody a sense of balance — see examples (51) (LZB: 326) ku „ 'bitter, suffer' (GSR: no. 49), (52) (LZB: 327) ruo g 'agree, approve, compliant' (GSR: no. 777) and (53) (LZB: 336) mo M. 'evening, late' (GSR: no. 802). Even a cursory glance at FYSJ (41) is enough to determine that the cao fnf signifier in this graph is also an example of a more mature orthographic form. Until this point, all the orthographic evidence on FYSJ slip SI 23 suggests a late date. The FYSJ form of wen  and yin _f all suggest an orthographic affinity with early Han  Ancient Clerical script. As such, in my view slip S123 should be considered part of the same original manuscript as slip S122. The orthographic proof supporting this assertion far exceeds the question of calligraphic style mentioned earlier. The third and final graph that I will examine from slip SI23 does not refute this assertion, although it cannot be used as positive evidence either. The FYSJ form of xiang $JL 'elephant, ivory, [cut] figure' (GSR: no. 728) seen in (54) (S123) is different from the SHD version found in (56) (QJWZB: 718) in several significant ways. Graph (54) is written in a much less slapdash manner than (56) and the 71  three downward slanted lines at the bottom of the FYSJ graph are comparatively much longer. Unfortunately, in the QJWZB and LZB there is a paucity of xiang fe graphs and we do not have a MWD form with which to draw comparisons. Due to the orthographic evidence supplied by other graphs on FYSJ slip S123, further analysis of xiang fe need not be pursued. In sum, in this graphological study I have attempted to show that FYSJ slips S122 and S123 likely belong to the same source manuscript. Based on a study of the orthographic features characterizing the various graphs on these two slips, I believe that they were written sometime around the early Western Han dynasty. Citing both calligraphic and orthographic evidence, I decided that FYSJ slip S124 is not part of the same source manuscript as S122 and S123. Despite the high degree of damage to FYSJ S121, there is sufficient calligraphic evidence on this slip to conclude that it also does not belong to the same original manuscript.  72  Section 2): A Graphological Study oiyang # 'South side' (GSR: no. 720) and FYSJ slips S102 to S106  According to Xu Shen, yang H is a teknogram comprised of three distinct graphemes (SWJZZ: 9:34b): (A) the grapheme ri 0 'sun, day' (GSR: no. 404) at the very top of the character; (B) a straight horizontal line in the middle; and (C) wuty]at the bottom. While Xu Shen's structural analysis of yang J| is without fault, graphemes (A) and (B) are probably easier to discuss if taken together in the form of the character dan J_ The Small Seal of this character can be seen in example (1) (Gao: 490) below. According to the GSR, in pre-Han Chinese this tomosomatic character means 'dawn, morning, bright'. Consequently, it seems fitting that Karlgren should comment, "The graph is said to depict the sun rising" (GSR: no. 149). Karlgren's interpretation may be based on the SWJZ, which states that dan __ means 'bright' or 'light' and is a pictograph depicting the sun rising above the horizontal plane of the earth  0 H — _ h — ( S W J Z Z : 7:14a; for a similar interpretation also see  JMJWCD: 84). It certainly warrants mention, however, that in scripts other than Small Seal dan H. was not always written pictographically. In fact, in OBI and BRI scripts antedating Small Seal, this graph clearly lacks the same degree of pictographic quality. Example (2) (JGWB: no. 823) provides an OBI form of dan _B; whereas, example (3) (JWB: no. 888) offers a BRI form from a cauldron dated to the early Zhou dynasty. In the former, the horizontal line which represents the earth in the Small Seal graph takes the form of a misshapen rectangle horizontally written. Although the consensus among paleographers is that in OBI materials dan .0 is used to express the meaning 'dawn' or 'sunrise' (?f  H; 0 tB^B#), it is difficult to  associate this meaning with the structure of the pictograph itself (KMJS: no. 823; JGWZGL: no. 1140). In BRI graph (3) the relatedness of this character's semantic value and its pictographic representation is somewhat clearer. The darkened stem joining the earth at the bottom and the ri jEJ grapheme above can be interpreted as a sign that "the sun has just risen and not yet separated from the ground" (#L0 TO^fH^iift) (JMJWCD: 84). There are Warring States forms of this character which do not match the Seal form either, such as (4) 73  (Gao: 490). Among paleographers there are several different interpretations concerning the pictographic quality implicit in wuty}.The Small Seal form of this character appeared to Xu Shen in the Han dynasty as SWJZ (5) (SWJZZ: 9:34b). He commented on it as follows: "[This graph] represents the banner erected by zhou li. [The graph] depicts its pole and there are three streamers. They are made of assorted silks" MMi^^i^ffiWH_N£ri5 (SWJZZ: 9:34b). (In the same entry of the SWJZZ, Duan Yucai mentions that the term zhou li M i l refers to 'high officials and scholars' A^c± da fu shi.). But Xu Shen's explanation of this graph is questionable, and the opinion of paleographers working with OBI forms of wuty]— e.g., example (6) (JGWB: no. 1134) — differ from the SWJZ. Paleographers propose the following interpretations for this character: (A) It is a pictograph of a hu  a type of hand held tablet used by high officials when in audience with  the emperor. According to Guo Moruo % " T h e ancients would record matters on the hu %rj in order not to forget. This graph accurately depicts its shape" "r^AT^Jilt^ »JiA$t7_ , (KMJS: no. 1134); (B) It is a pictograph of a bird tail (KMJS: no. 1134); (C) It "depicts the wings of a bird in flight"  (JGWZGL: no. 2625); (D) Zhao Cheng M  t$ suggests that (6) is an abbreviated form of two gong ^ graphs side by side, and that in OBI this character represents a device used to adjust a bow ( # t _ ^ , i f c u ^ t l M j E ^ KlXM) (KMJS: no. 1545). The diverse semantic values represented by these various interpretations suggests they are tenuous in nature. To stay on safer ground, I will adhere to the straightforward and conservative opinion of Sun Haibo MMffi., who claims, "This graph's shape is unrecognizable" ^-ff^WW (JGWZGL: no. 2625). I will discuss wu^jm. terms of a curved and faintly S-shaped 'spine' with three slanted strokes running down and to the right. Below are graphs from various different sources that I use in my analysis ofyang #:  74  As is usually the case, the complex Seal and Small Seal form of yang | § — example SD (8) (SKZWB: 6:3; JSZD: 651) from the character yang & 'poplar' (GSR: no. 720); SD (9) (SKZWB: 14:10; JSZD: 1395) from the character yang gg 'light, brightness' (GSR: no. 720); STI Small Seal (12) (SKZWB: 6:3; JSZD: 651); and SWJZ (13) (Gao: 493) — possess a high degree of uniformity. On the whole, the ri 0 grapheme is comprised of a slightly rectangular box containing a short horizontal line. In SD (9) the horizontal line is abbreviated to such an extent that it takes the form of a dot. In three out of four examples, this grapheme is separate from the line representing the earth below. Only in the Small Seal graph (13), which is a form taken from the SWJZ (ca. 100 A.D.), does the curved spine of the wu  element pass  through the horizontal line and contact ri 0. In a slightly variant orthographic strain, which spanned from the early Zhou dynasty to the Warring States period, the ri 0 grapheme in the character yang U is round in shape. Examples of this can be found in (10), (11) and (15) (Gao: 473), all of which show the character tang | § 'hot liquid, reckless' (GSR: no. 720). Graphs (10) and (11) have been dated to BRI from the late Zhou dynasty; whereas, (15) is from the Warring States (Gao: 473). By contrast, the ri 0 grapheme in example (16) (Gao: 473), also from the Warring States, is written with much sharper angles. Further, in this character the ri 0 grapheme's inside horizontal line meets the vertical lines on both sides. As is evident in the GWZLB, the increased/>/>2g zhi hua ^jT/fb, or 'straightening of lines' (see p. 27), seen in (16) typifies the 75  way tang © is written in Six States script of the Warring States period (Gao: 473). It is significant that none of the yang J§ graphs seen in examples (8) to (16) show the ri fEj grapheme connecting with the longish horizontal line underneath. This demonstrates that even as late as the Warring States period, the pictographic quality of dan _B was widely maintained. It seems that only in Ancient Clerical script did dan j=L suffer from a diminishment in pictography. Further elaboration on this point is forthcoming. The Seal form of wu  is also extremely uniform. In SD (8), SD (9), STI (12) and  SWJZ (13) wuty]is characterized by a very salient sense of balance and proportion. In Seal script the three lines slanting down to the left typically do not join the curved spine running from the top to the bottom of wu ^ j . Moreover, in (8), (9), (12) and (13) the shallowness of the bend in this graph's spine contributes to its overall vertical rectangularity, another prominent feature of Seal script. Although not visible in the graphs presently under discussion, vertical rectangularity is an especially dominant trait in Small Seal orthography. Some of these traits can be seen in other scripts from the mid and late Zhou dynasty. For example, BRI (10), (11) and (7) (Gao: 454) have so-called 'spines' with relatively slight bends that do not join the three slanted lines to the left. At the same time, Six States forms of wu  dating to the Warring States period are even more vertically rectangular than their Seal  counterparts — e.g., see examples (15) and (16). Further, in Warring State's graphs (15) and (16) the three slanted lines join the spine. This is a characteristic common to the yang H graph as written in bamboo and stone inscriptions of this period, including the SHD to be examined next. Overall, in the SHD Ancient Clerical script, the ri 0 grapheme is irregularly shaped. It can be slightly angular in shape like (17) (QJWZB: 467) which shows the character si %; more oval-shaped as in (18) (QJWZB: 467) and (23) (QJWZB: 616) which shows the character shang % 'to wound, hurt, damage' (GSR: no. 720); or essentially misshapen like (19) (QJWZB: 467), (24) (QJWZB: 616) and (25) (QJWZB: 989). In each case, the SHD grapheme for ri 0 differs drastically from the horizontally rectangular Small Seal form. Therefore, it is interesting that corresponding MWD graphs written in the Western Han bear a comparatively closer resemblance to Small Seal orthographic form; that is, they exhibit a 76  distinct sense of uniformity, are somewhat less cursory and are square in shape. This is especially true of the ri 0 grapheme in MWD (20) (LZB: 250), (26) and (27) (LZB: 147). But a characteristic common to both SHD and MWD forms ofyang H is the manner in which the ri 0 grapheme joins the line underneath it depicting the edge of the horizon. Sometimes the upper element only partially joins the lower horizontal line; in other cases, the horizontal line actually replaces the bottom line of the ri 0 grapheme. Among the examples of Ancient Clerical graphs contained in the QJWZB and LZB, the only exception to this very salient orthographic trait is (30) (LZB: 250). Not only is this feature present in the SHD and MWD graphs mentioned so far, it is also very prominent in MWD (21) (LZB: 250), in addition to MWD (22) (LZB: 567). I have already mentioned that the merging of ri 0 and the horizontal line beneath it distinguishes Ancient Clerical from other scripts contemporaneous to it. As will be seen shortly, the degree of connectedness between these two components in the Ancient Clerical forms ofyang | § plays an important role in dating the FYSJ. The SHD and MWD forms of wu  to some degree resemble what is seen in Warring  States bamboo sources — e.g., examples (15) and (16). Typically speaking, the three slanted lines in the Ancient Clerical form of this graph also join what I have referred to as the 'spine'. From the examples of MWD graphs given above, (22) is the only exception to this 27  orthographic pattern. Aside from this trait, however, the SHD and MWD forms of wuty]are quite dissimilar. There is a much higher degree of cursoriness in the SHD examples, something which is reflected in an irregularity of form and loss of orthographic balance or proportion. For example, in (19) wu  appears to be positioned on a slant; in (29) the three slanted lines  are comparatively long and the spine more deeply curved; in (18) these same lines are not parallel and have a 'stringy' look; (24) also has a deep bend in the spine and longish strokes. 27  For my examination of the yang J§- graph in the SHD, I refer to the characters si H, shang fl§ and yang fig  found in QJWZB. 467, 616 and 989, respectively. I also looked for examples ofyang J§- in numerous other characters that contain this graph. But the likes of the following characters cannot be found in the SHD — i.e., chang  yang jj§, dang {§, dang H§, yang  tang  chang J^, dang  As for the MWD, I utilize the  characters tang ?|§, shang fj§ and yang p§ found in LZB: 250, 147 and 537 for purposes of examining the yang H graph. I checked the following characters in the LZB for further examples ofyang j|, although to no avail: chang |§, yang  dang Jj§, dang '!#, si §§, yang %, chang i§, dang %  77  Conversely, the MWD form is not really cursory at all, as MWD graphs (20), (21), (26), (28) and (30) are all quite uniform. In all of these examples, the three slanted lines are generally straight, of a similar length and parallel. Further, the ri 0 and wuty]elements are more closely in proportion to each other. Having examined^artg J| in the context of SHD and MWD orthography, I will now move on to an analysis of the FYSJ. As in the case of the SHD and MWD graphs, the examples ofyang J| from the FYSJ are all components in a larger graphic environment. For example, in FYSJ (31) yang J| is part of the character yang p§. Two of the FYSJ graphs I use, (35) (SI 03) and (36) (S106), are incomplete; however, what remains of them nevertheless serves a purpose in my discussion.  As mentioned, the examples ofyang # given here all constitute components of various different teknograms: FYSJ (31) (SOI 1) is part ofyang g§ which also contain the graph fit ^ 'big mound, earthen hill' (GSR: no. 1108); FYSJ (32) (S082) is from the same teknogram; FYSJ (33) (SI02) is from tang jff 'hot liquid' which also contains the semantic determiner shui TR 'water' (GSR: no. 576); FYSJ (34) (S105) is from the teknogram dang % 'carefree, extravagant' (GSR: no. 720) which also contains the grapheme xin I\J (GSR: no. 663). For the most part, I only analyze the elements found along side yang # in these teknograms to substantiate assertions that I make concerning orthography. As such, these accompanying elements should be viewed as secondary forms of orthographic evidence. I start my discussion of FYSJ orthography with graph (31). The same form of orthographic abbreviation manifest in MWD (21), (22), (26), (27) and (28) is also found FYSJ (31); that is, the bottom line of the ri 0 grapheme has merged completely with the horizontal stroke representing the plane of the earth. Although this trait is also manifest in SHD orthography — e.g., SHD (24) and (25), in the Warring States Ancient Clerical form of yang J|, the ri 0 grapheme typically only makes partial contact with the grapheme below. 78  But a characteristic of FYSJ (31) that contradicts evidence suggesting a later orthography is its vertical narrowness. Although not all SHD forms of yang J | are vertically oriented and narrow — see (25) and (29), this feature is still strongly characteristic of pre-Qin dynasty yang # orthography. By contrast, of the MWDj^ong | § forms surveyed, only (27) is actually vertically narrow in form. Overall, I still think the evidence suggesting an earlier form of orthography is out weighed by proof indicating otherwise: (A) Despite the vertical rectangularity of FYSJ (31) as a whole, the shape of its ri 0 grapheme still resembles the MWD form, especially that seen in (21); (B) FYSJ (31) is not written in a slapdash style, and is both balanced and evenly proportioned — e.g., the three downward slanting lines are uniform in length and parallel; (C) The proportion of the fu  element and yang J | in FYSJ (31) more closely resembles that  seen in early Han orthography — e.g., compare FYSJ (31) with MWD (20), (21) and (30). Based on the evidence at hand, I conclude that FYSJ (31) post-dates the Warring States SHD. Compared to FYSJ (31), dating (33) is a more straightforward process. Overall, the broadness of FYSJ (33) more closely resembles the MWD form of yang #. In connection with this trait, the spine in the wu  element has a deeper bend and the three slanted lines are  longish, but straight. MWD (20), (21) and (30) are almost identical in this regard. Further, the shape of the ri 0 grapheme in FYSJ (33) is angular and slightly rectangular in a horizontal sense, which matches MWD graphs (20), (21), (30) among others. Although the dan B. component in (33) is divided, which perhaps violates the general MWD orthographic convention, this anomaly is not without representation in early Han orthography. In MWD (30) we have at least one example of yang # in which the dan 0. element is divided. In addition to its broadness, FYSJ (33) exhibits a sense of balance and equal proportion that is one of the hallmarks of MWD orthography. In fact, the size of the shui 7R grapheme next to yang # in FYSJ (33) is strikingly similar to MWD (22). And FYSJ (33) is not written in the slipshod fashion which typifies SHD Ancient Clerical script. Therefore, I conclude that this graph should be dated closer to the time of the MWD, perhaps even after FYSJ (31). At first glance, FYSJ (32) also seems somewhat perplexing. The roundish ri 0 79  grapheme that only partially connects with the lower horizontal line is in some ways similar to the SHD orthographic form of dan _B. Yet, the symmetrical teardrop-shape of ri 0 in (32) really has no likeness to any of the SHD examples. Further, there are important traits in (32) which resemble FYSJ graphs (31) and (33). For example, on the whole FYSJ (32) is balanced and evenly proportioned like (33) and the numerous other MWD graphs. The proportion of ri 0 and wu #J in relation to one another, as well as the basic orthographic structure of wu ty] itself are chiefly responsible for the sense of balance exhibited in FYSJ (32). The uniqueness of ri 0 in FYSJ (32) is also amply reflected in the adjacent fu which bears no resemblance to the Ancient Clerical form seen in MWD (20), (21), (30) or SHD (25). Neither it is similar to the Small Seal form of/w J£ in example (9). Since FYSJ (32) contains so many features that do not appear related to Ancient Clerical orthography, I will refrain from attempting to ascribe a specific date to it. I believe that this character reflects some influence from script outside the Clerical tradition; namely, Six States script. Interestingly, nowhere in the GWZLB is there an example offu^. which even remotely resembles that seen in FYSJ (32) (Gao: 451-8). Compared to FYSJ graphs (31) and (33), it seems likely that FYSJ (34) is written in an early orthographic form. For example, the oval shape of the ri 0 grapheme and the way it only partially contacts the horizontal line below strongly resembles the typical SHD form of dan j3. But there are other characteristics in FYSJ (34) that indicate an early date, as well: (A) The yang J § element in (34) is vertically narrow just like SHD (17) to (19); (B) The yang J§ in (34) lacks the orthographic balance which is so evident in MWD graphs — the ri 0 grapheme is wider than the wuty]element which tapers down to a point; (C) the downward slanted strokes in wuty]are unusually short, which does not coincide with the MWD form at all. If one looks beyond the yang J | component and examines the entire character dang % represented in FYSJ (34) these three features become even more salient. At the same time, taking the entire character into consideration, we see that it is written in a manner very reminiscent of SHD (18) and (19). In FYSJ (34) and these SHD graphs, yang # floats either above or below the grapheme that it is orthographically affiliated with. Despite calligraphy 80  characterized by finer and more articulate strokes, FYSJ (34) basically matches what is seen in the SHD. I therefore date it to the Warring States period. At this juncture, I will subject one of the FYSJ forms of yang H to a more rigorous dating process involving contextual analysis. Since I focus on a relatively late form of ma H in the first graphological study, here in section two I subject an early form ofyang J | to orthographic scrutiny; namely, FYSJ (34) found on slip S105.1 begin this endeavor by testing the various graphs found in FYSJ slip S105 for orthographic consistency — i.e., are the other graphs in FYSJ slip S105 written in an orthographic form which also suggests an early date? The other graphs on this slip which are available for examination — i.e., not damaged or incomplete — include: H ^ J f ^ . I believe there is sufficient evidence in slip S105 to assert not only that dang % represents a relatively early orthographic form, but also that the other graphs on this bamboo slip do, as well. The following graphs from various sources are the evidence that I use to support this assertion:  (37) 1*1  (38) r - f (39) / ^ l (40) / I  (41) > T (42) X  (43) /  (44)  (45)  The FYSJ S105 graph to be analyzed next is qi  'uniform, equal, be equal with' (GSR: no.  593) featured in example (37) (S105). According to Karlgren, the original OBI form oiqi 5Hf is a pictograph of "the hair-pins of the married woman [qi] j | " (GSR: no. 593). Both Karlgren's view and the interpretation given this graph in the SWJZ  tt&#M± f-&&Lj&) s  (SWJZZ:  7:33a) are somewhat suspect. Support for the former is not found in the interpretations of 81  present-day paleographers. Moreover, the editor of the JGWZGL writes: "[The character] qi ^ has nothing to do with the shape of an ear of grain or wheat sprouting"  ^^^J^i&MZ-  (no. 2124). But because Xu Zhongshu # t f J 0 endorses it, and for lack of anything more convincing, this botanical interpretation is adequate for the needs of the present graphological study (KMJS: no. 855). As the examples above should verify, orthographic evidence suggests that FYSJ (37) bears a closer affinity to the SHD form seen in (38) (QJWZB: 517), than the MWD form found in (39) (LZB: 1503). The orthographic features possessed by MWD (39) which contrast the FYSJ form include the following: (A) As is generally the case with MWD script, MWD (39) is more vertically rectangular; (B) MWD (39) exhibits a greater inclination towards orthographic balance; (C) The three 'diamond-shaped' graphemes are smaller and not as high. Attempting to further substantiate the date of FYSJ slip S105 using the graph zi 'son, daughter, child' (GSR: no. 964) is a challenge. In both the SHD and MWD, zi =f is written in a number of diverse orthographic forms. The resulting inconsistency makes it difficult to use examples from these sources to establish an orthographic profile for Ancient Clerical forms of this character. Some of the orthographic forms of zi ^ found in the SHD are seen in examples (44), (45) and (46) (QJWZB: 1026). There are several other forms in the SHD not featured in the above examples. Among the MWD forms of zi  are (47), (48),  (49), (50) (LZB: 709). FYSJ (42) (S105) does not really match any of the examples of zi =f found in either the QJWZB or LZB. Duan Yucai writes that the SWJZ Small Seal form of zi f, as seen in (43) (Gao: 49), =  depicts "the shape of a person's head, arms and feet"  feA'S'lt^^M  (SWJZZ: 14:  25b). This interpretation is for the most part correct; however, the majority of paleographers now think that zi ^ was originally a pictograph depicting the form of a child (KMJS: no. 1694; JGWZGL: no. 590; JMJWCD: 36). The 'body' of the zi =f- graph in FYSJ (42) most resembles SHD (44). Although FYSJ (42) is not entirely unlike MWD (47), I believe the flare at the bottom of this stroke in (47) distinguishes the two orthographic forms from one another. In regard to the 'head' portion of 82  zi f-, once again FYSJ (42) seems to resemble SHD graphs the most — e.g., SHD (44) and z  (45). However, here too the FYSJ form bears a marginal resemblance to MWD graphs — e.g., (47) and (50). It would seem that the size of the 'head' is the one feature distinguishing FYSJ (42) from the MWD graphs. In the early Han script, this part of zi =f- is consistently much smaller than in the FYSJ form. The strongest proof for my assertion that FYSJ (42) zi ^ represents an early orthographic form lies in the shape the character's 'arms'. In the Warring States forms of zi f, the child's arms are still slightly bent upward, which reflects an influence from the Small  z  Seal form seen in (43). By the early Han, this stroke had already undergone linearization — e.g., see MWD (47), (49) and (50). In sum, FYSJ (42) and (34) appear to have been written at the same time. Since the character gao  'announce' (GSR: no. 1040) does not exist in the SHD and  MWD I am unable to date it properly. Such an exercise is, of course, dependent upon the comparison of graphs from different Ancient Clerical sources. Fortunately, in comparing the various forms of the graph yang i £ 'sheep, ram' (GSR: no. 732), we obtain valuable evidence for dating FYSJ slip SI05. There can be no doubt whatsoever that the FYSJ form ofyang ^ seen in (51) bears a much stronger resemblance to SHD (52) (QJWZB: 285) than it does to either MWD (53) or (54) (LZB: 1200). A cursory glance at these examples above should suffice as proof of this claim. In JGWZGL, no. 1561, one paleographer writes, "The ancestral OBI form ofyang ^ depicts the shape of horn and four legs" | ^ ^ ; _ | U ^ S ^ 0 J _ ; _ - / ^ - This interpretation is echoed in the SWJZ: "[The graph yang i £ ] depicts the shape of head horns, feet and a tail" fe Mr%B.%2.B (SWJZZ: 4: 33a; also see JMJSCD: 143). This description fits the OBI pictograph seen in (56) (JGWB: no. 510) and SWJZ (55) (Gao: 190) quite well. Using the SWJZ Small Seal as a point of reference, it is plain that while SHD (52) and FYSJ (51) still retain some degree of pictography, the two MWD graphs are more simplified. For example, in MWD (53) and (54) the horns of the ram have either disappeared altogether or become linearized. A l l told, I believe there is enough evidence to propose that FYSJ slip SI05 was written sometime during the Qin dynasty, or even before. 83  FYSJ S105 is one of six slips — S102 to S107 — that corresponds to Mao Shi no. 105 (Hu: 13; Wen: 81-2). Since all of these slips represent the same ode, the possibility that they belong to the same manuscript is very high. If true, this would certainly facilitate the process of identifying different manuscripts in the FYSJ. However, this proves not to be the case here. In the following section, I will attempt to demonstrate that the slips containing verse from Mao Shi no. 105 are not actually from an identical source. For this purpose, I will once more utilize graphic comparison. The orthography of yang # in FYSJ S102 (33) is vastly different from its counterpart FYSJ S105 (34). In the examination conducted earlier, I concluded that (33) represents a relatively mature orthographic form, which contrasts my findings for FYSJ (34). At this juncture, I will examine the dao | f graphs found in FYSJ slips S102 and S104. (For a componential profile of dao M see chapter one.) From the way FYSJ slips S104 and S105 fit together as bamboo fragments (Wen: 82), there is little doubt they are from the same original manuscript. The graphs used for this contextual analysis are as follows:  The FYSJ form of dao j f seen in example (62) from slip S102 is incomplete; consequently, the chuo  element at the left hand side of this graph is largely indistinguishable. However,  from what is visible the chuo *5t, element in (62) does not extend below the shou "gf element, so that it resembles the early Han form of dao M seen in MWD (60) and (61) (LZB: 650). Since the dao M graphs in the MWD all adhere to this orthographic structure, I believe the evidence involving chuo  is sufficiently persuasive to conclude that FYSJ (62) is an early 84  Han orthographic form. As such, FYSJ (62) substantiates the proof already provided by (33) regarding FYSJ slip S102. In the SHD, dao j _ is written in a number of different orthographic forms — e.g., examples (57), (58) and (59) (QJWZB: 108). Example (57) is from the SHD text entitled Fengzhen shi (QJWZB: 1128-37) and is unlike any other dao j _ graph in the SHD. This is because the 'foot' grapheme in (57) does not run into the 'street' grapheme and is positioned beneath shou "|f. However, aside from this anomaly, the SHD forms of dao >_ are typical for Warring States Ancient Clerical script, which are mostly box-like in shape and written in a slapdash or careless manner— e.g., in some cases certain graphemes are misshapen. Interestingly, the shou "if' element in FYSJ (62) resembles its SHD counterparts more than it does MWD (60) and (61). However, this does not dislodge me from my assertion that (62) was written sometime near the early Han. The date that I give the dao >_ graph in S104 — that is, graph (63) above — matches that assigned to the entire S105 slip. The shou ~H element in (63) is particularly wide, something not characteristic of the SHD and MWD forms of dao j _ . In my view, this feature can probably be attributed to calligraphic style. An intriguing piece of orthographic evidence is found in the uppermost horizontal line in the 'head' grapheme in FYSJ (63); that is, the 'head' grapheme as distinctfromthe grapheme which represents either 'hair' or 'horns' (SWJZZ: 9:16-7a; GSR: no. 1102). This stroke is shaped like m^tf  which contrasts the  linearized rendering already current at the time of the SHD in the late Warring States period. Taking into consideration (65) from the complex Seal ZCW inscription, (66)fromQin dynasty Small Seal ceramic inscriptions (these examples are seen in chapter one as 28 and 33) and (67), the SWJZ Small Seal form (Gao: 107), it is easy to see how this stroke developed out of the Small Seal form of dao j _ . It is possible this affinity to Small Seal orthography indicates that FYSJ (63) is written in a form earlier than any seen in the SHD. The orthographic form taken by the chuo ^ element in FYSJ (63) serves to support this assertion. In terms of this element's position in relation to the rest of the graph, in addition to the position of the 'street' and 'foot' graphemes in relation to one another, this FYSJ form of dao j _ closely resembles the Small Seal form seen in both (66) and (67). By contrast, the SHD form of chuo 3fL is already much 85  more abbreviated and cursory. The physical evidence connecting FYSJ S104 to SI 05, coupled with orthographic evidence linking both two slips to the Warring States period, indicates that they are derived from the same original manuscript. Unfortunately, the lu H 'dull, blunt, simple' (GSR: no. 70) graph in FYSJ S104 — e.g., graph (71) — is only partially complete, otherwise even more evidence with which to date slip SI04 would be available. Moving on to an examination of FYSJ SI 05 and SI 06, the first aspect of these slips which attracts interest is their dissimilar calligraphy. The graphs on the former slip are written larger and somewhat closer together. A form of dao ^fi is also found in FYSJ slip SI06 — see example (64) above. By comparing it with the dao Hf in SI05, data regarding the relatedness of these two slips should be attainable. Even at first glance, FYSJ (64) appears to represent an early orthographic form of the graph dao jJt For example, the 'head' and 'hair/horn' graphemes in (64) are strikingly similar to the Small Seal forms seen in (66) and (67): the bottom of the shou i§" element is rounded, the uppermost horizontal line in the same grapheme has not been linearized, and the 'hair/horn' grapheme is trident-shaped. In this regard, (64) resembles Small Seal even more distinctly than (63). I believe this is our strongest orthographic evidence for establishing the date of these graphs in relation to one another. At the same time, the chuo 3_ component in FYSJ (64) seems slightly more abbreviated in form, which might indicate a later date than (63). But it is difficult to know for sure exactly how important this orthographic abbreviation is as a form of evidence. In any case, the dates of FYSJ (63) and (64) in relation to each other is not critical to the present analysis and I am satisfied knowing that these FYSJ graphs both represent early orthographic forms. That the orthographic evidence surveyed suggests (63) and (64) were not written at the same point in time and, therefore, are not from the same source manuscript is the most important consideration here. Further substantiation of this claim is evident in a comparison of different yang J | graphs. More specifically, FYSJ S105 graph (34), and (36) found in FYSJ slip S106. Even though FYSJ (36) is only partially visible from what we can see this graph is clearly different from FYSJ (34). In (36) the ri 0 grapheme is not oval-shaped, the bend in the 'spine' is deeper and the three slanted lines do not join with wuty).Further proof that slip S106 does 86  not belong in the same original manuscript as the two slips preceding it can be attained in a comparison of FYSJ graphs (70) and (71), the former being from slip S106, the latter from SI 04. As seen in example (72) (JGWB: no. 483) from Shang OBI, the character lu H was originally comprised of a 'fish' yufe,and what has been interpreted as a 'mouth' kou • or a basin (GSR: no. 70; JMJWCD: 432). According to the editor of the JGWZGL, in OBI the kou • grapheme was added to this character simply to distinguish it from the character^ fe (see no. 1815). Elsewhere, however, this element is viewed as containing its own semantic value,makinglu%asyssemantic character  ( £ ) _ _ „  ,  _^__W$05; , ^J__f;&)  (JMJSCD: 432). Sometime during the Spring and Autumn period, the kou • grapheme was replaced by bai Q which can be seen in OBI graph (75) (JGWB: no. 984). This matrograph is variously interpreted as a 'phallic symbol' (GSR: no. 782), an abbreviated form of the graph for 'nose' (SWJZZ: 4: 15-6a), or simply a corrupted form of the original grapheme (JMJSCD: 432). Kang Yin jRIg interprets (75) as a pictograph of a white grain of rice or corn. He feels that this graph came to represent the word 'white' after a process of semantic extension (269-70; for the meaning of 'white' also see KMJS: no. 984). For an example of a BRI graph from this period, and a descendant SWJZ Small Seal form see (73) (JWB: no. 475) and (74) (Gao: 215). Returning to the FYSJ examples, the bai Q element is written much broader in (71) than it is in (70). They are written in a different orthography altogether, and clearly do not belong to the same manuscript. Up to this point, I have narrowed down one FYSJ manuscript to definitely include slips S104 and S105.1 have also demonstrated to the best of my ability that neither FYSJ slip S106 or slip S102 belong to this particular manuscript. As such, all that remains is to discover whether or not FYSJ slip S103 belongs to the manuscript represented by S104 and S105. For this purpose, the graphs lu # and yang J§ from slip S103 — FYSJ (69) and (35) -— are available to us. Despite their incompleteness, both of these graphs indicate that FYSJ SI 03 does not belong to the same manuscript as FYSJ slips S104 and S105. For example, from what is 87  visible ofyang |§ in (35), it is clear that this graph is orthographically different from FYSJ (34), belonging to slip S105. Further, the lu _ graph in S103 is substantially narrower than its counterpart in S104 — e.g., compare FYSJ (69) with FYSJ (71). Since neither of the graphs from FYSJ S103 are complete in form, I will not attempt to date them. But before moving on to the third and last graphological study in this chapter, there is one last observation worth mentioning here. The fact that in five bamboo slips from the FYSJ — i.e., S102 to S106 — there are three different forms of dao j _ (see examples 62 to 63) and four different forms of lu _.(see examples 68 to 71) underscores the highly fragmented physical nature of this composite manuscript.  88  Section 3): A Graphological Study of si Jg 'think' (GSR: no. 973) and FYSJ slips S048 to S051 Xu Shen observes that si U is comprised of two graphemes: the upper grapheme I2J; the lower grapheme  (SWJZZ: 10:24b). Experts on etymology both from ancient China and  the modern day unanimously agree that the former grapheme is a pictograph of a 'head' or 'skull.' In regard to this pictograph, Xu Shen writes, "[It depicts] the cap of the skull where the [parts of the] head meet" IS#ISM-fe (SWJZZ: 10:23b). In support of this view, Yao Xiaosui M>#ji§l writes that the OBI form of \% represents a human 'head' and is associated with sacrifice (JGWZGL: no. 1.100). Not a single paleographer listed in KMJS, no. 1115 refutes this interpretation — e.g., Xu Zhongshu writes, "[This graph stands for] the heads attained from enemy states which were used as sacrificial offerings" ffr^rWM.M^-'W^i , MM PPExperts on the subject of Chinese graphology are equally certain that the graph for 'heart' was originally a drawing. Karlgren believed this to be the case (GSR: no. 663) and in the SWJZ Xu Shen states, "[The graph ^ represents a] human heart. It is an organ belonging to the earth element. It [resides] in the body. [This character] is a pictograph" ^ A'L\±JlS^ feS^^&T^  (SWJZZ: 10:24b). (In SWJZZ: 4:22b, Duan Yiicai explains how various human  organs came to represent different elements according to Han cosmology  - r -  e.g.,fei  |if/ iungs'  are associated withjin # 'metal', shen "pf 'kidneys' are affiliated with shut 7K 'water' and gan 3f iiver' is connected with mu fc 'wood'.) Obviously, this pictograph was not immune from the forces of orthographic change that influenced Chinese writing from the Shang to the Han dynasty. Through the ages xin >(jcontinually underwent a process of simplification. Still, when regarding the words of Yu Xingwu T'S^I, "In OBI the 'heart' graph was written Q ; it accurately depicts the shape of the outer profile of the human heart" ^X'L^-^F  ^ JE&.fc'bB#ltitMJ&. (JGWZGL: no.  1934), we should not assume that the Small Seal form is completely dissimilar. There is a high degree of pictography in the Small Seal xin >i? graphs, as well. Below are several examples the character of si,® and the graphemes that are part of it: 89  (4)  ( 9 ) \ K (10)^0(11)  (18)  ^-(19)  \(20)  (5) *V_p(6) ,(5) ^W(6)  V V (7)  (12) S/(13) X(14) Hr (15)  (21)  ^^(22)  V.(23)  %•_.(_) "W(8)  (16) V-(17)  V_  Xl(24)  In the examples above, (9) (JGWB: no. 1115) is taken from Shang OBI and (10) (JWB: no. 1232) from BRI written around the middle of the Zhou dynasty (Gao. 140). The teardropshape characterizing BRI graph (10) resembles that seen in (1) (SKZWB: 10:16) and (3) (Gao: 150), the former a Small Seal STI graph from the Qin dynasty, and the latter a Six States graph from the Warring States period. Clearly, the form of this grapheme remained consistent until its appearance in the Ancient Clerical script of the pre-Qin dynasty. SHD graphs (5) and (6) (QJWZB: 759) show how xin (tj, particularly as a grapheme within a larger compound graph, underwent a transformation. While (6) still maintains a roundness, it has become elongated horizontally, which stands in contrast to graphs such as STI (1) and SWJZ Small Seal (2) (Gao: 150). In (5) the element xin [_] is bpth horizontally rectangular and much squarer in form. MWD graphs (7) and (8) (LZB: 1043) exhibit a continuation of the orthography evident in SHD (5). Both of these MWD graphs are also squarish in shape and exhibit a tendency towards heightened horizontal rectangularity. Compared to what would appear later in Ancient Clerical script, the Small Seal form of xin ifj preserves a considerable degree of pictographic quality. After all, despite some orthographic alternation, the complex Seal and Small Seal forms of xin /f_ seen in STI (1), SWJZ (2) and ZCW (15) (SKZWB: 10:19) — min fg 'to sympathize with, to grieve' (GHYZD: 174) — still resemble their OBI counterpart — e.g., example (11) (Gao: 1.53) showing the character yu ^ 'pleased' (GSR: no. 82). The characteristics distinguishing the Seal form of xin  mainly revolve around orthographic elaboration. For example, graphs (1),  (2) and (15) all contain a stroke which flows down from the bottom of the graph like a tail, 90  and two relatively symmetrical 'bulges' or rounded chambers on either side. By contrast, the 'heart' pictured in OBI graph (11) is simply demarcated with two short lines. The xin  seen in example (12) (Gao: 149), which shows the character nian  'think  of (GSR: no. 670), is the most typical form of this grapheme in early Zhou BRI, and it already shows signs of evolving into the orthographic form manifest in Seal script. This transformation is complete in the BRI of the Warring States, as seen in the Six States graph (13) (Gao: 151), the character xi M 'breathe, sigh, rest' (GSR: no. 925). Example (14) (Gao: 153), a form of the character bei f_% 'grieve, sad' (GSR: no. 579), presents an example from Six States script written during the Warring States period. As is often the case in Six States script, this graph is simplified and has a reduced number of strokes. A high degree of what Pankenier calls symbolification characterizes the way xin  is  written as a grapheme in Ancient Clerical script (34). To prove this point, in SHD graphs (18) through to (21) (QJWZB: 760-1, 770-1), I supply several xin  graphemes which have  been extracted from various different teknograms. For the most part, the cursoriness implicit in these graphs greatly diminishes their pictographic quality and, therefore, their likeness to Small Seal orthography. Example (20), from the teknogram yuan  'resent, resentment'  (GSR: no. 260), still maintains the three U-shaped chambers seen in graphs such as ZCW (15) (ca. third-century B.C.) and Small Seal STI (1). But this remnant of Seal orthography is much less distinct in the other SHD examples given. In example (19), from the teknogram^/ M 'to think' (GSR: no. 957), there is practically no trace of pictographic quality at all. The tendency towards pictographic diminishment continues in examples (22) and (23) (LZB: 971) from MWD and YQS materials, respectively. But there is a second orthographic trait which distinguishes the grapheme xin  in  early Ancient Clerical script. In the majority of examples from Ancient Clerical script given above, the tail-like stroke in xin ;f> is longer than in the Small Seal form. SHD graphs (20) and (21) are particularly good examples of this trend in Warring States orthography. Representing Ancient Clerical orthography from the early Han, MWD (22) also has an elongated tail stroke. Further, YQS (23) (ca. 140 B.C.), a xin /fj. grapheme extracted from kong & 'fear' (GSR: no. 1172), is also a very fitting example of this characteristic. I believe this 91  orthographic feature is part of the trend toward horizontal rectangularity found in Ancient Clerical script. Before proceeding any further, it should be noted that my analysis of xin  has been  based on the assumption that BRI graphs (12) and (13), and Seal graphs (1), (2) and (15) represent a distortion in the pictographic quality of the earlier OBI form. However, another interpretation of the orthographic change which the grapheme xin  undergoes is also  plausible. It is possible that this transformation involves a shift in emphasis rather than orthographic corruption. Instead of stressing the outline of a human heart as is the case in OBI (11), later forms of xin »{> may stress a depiction of arteries passing in and out of the heart. Such a view does not necessarily detract from the argument that is made concerning the orthographic evolution of xin  Even if the BRI and Seal forms of xin >t? depict arteries, the  same linearization and abbreviation can be said to occur. Moreover, a trend towards the diminishment of pictography in xin ii? is still evident. At this juncture, I will examine the two components [tj and >t? as they appear together in the character si Jg. As can be seen in examples (1) and (2), the vertical rectangularity in the Small Seal form of si Jg is very distinct. As such, its shape falls in line with the general trend in Small Seal script. For an example of si Jg written in SHD script we have SHD graph (4) (QJWZB: 758). Actually, despite the blotchy and unclear state of this graph, it is quite valuable to the present study. Not only does SHD (4) provide an example of the overall shape of si Jg written in SHD orthography, it also adds to what we already know concerning the xin  grapheme in early Ancient Clerical script.  Compared to its Small Seal counterparts, SHD (4) exhibits the reduced vertical rectangularity which typifies Warring States Ancient Clerical script. In this way it resembles the si Jg element found in SHD (5), the character Iii J? 'think of, foresee' (GSR: no. 69). By contrast, the si Jg in SHD (6) stands higher, so that it more strongly resembles the comparatively late forms of Ancient Clerical orthography represented by MWD (7) and (8). Interestingly, since SHD graphs (5), (6) and (4) are all from the text fe^^M. Wei li zhi dao (QJWZB: 25:1-18), it is unlikely that they were written at different times. Naturally, it is conceivable that these SHD graphs were written by different scribes. Indeed, while (5) and 92  (4) are located early in the text of the Wei li zhi dao (25:1, 25:7), (6) is from a location closer to the middle of the text (25:16). This orthographic variation in the same Warring States text can be attributed to orthographic overlap as discussed earlier. However, for a more explicit resolution to this case of orthographic contradiction, a more detailed orthographic analysis focusing on the SHD is needed. Having introduced the graphic components of the character si Jg and briefly surveyed its orthographic evolution, I am now free to turn my attention towards analysis of the FYSJ form. Unlike ma H and yang H examined in the previous sections of this chapter, I will consider only two examples of si Jg from the FYSJ. And of these two, most of my attention will be concentrated on FYSJ (26) from slip S050, since its counterpart from FYSJ slip S049 is partially damaged. The two examples of si Jg in the FYSJ are as follows:  Several important features distinguish FYSJ (26) as an Ancient Clerical graph, such as the structural makeup of the r_g and the degree of pictographic diminishment in the xin ifj signific. Whereas the Small Seal form of $ is marked with X , in the Ancient Clerical form the same element is always marked with -f-. However, as scholar Wen Xingfu comments, FYSJ (26) clearly resembles the Small Seal form of si Jg more than it does the Ancient Clerical form (97). For example, compared to SHD (4), (5) and (6), FYSJ (26) is more vertically rectangular and contains a $ element which is both less square and less horizontally oriented. The $ element in SHD (6) is also rounded; however, unlike the Small Seal form it is somewhat misshapen and horizontal. Moreover, the xin I(J signific in FYSJ (26) is still comparatively pictographic and its tail-like stroke does not extend beyond the main body of the graph as is the case in SHD (5), (18), (20) and (21). FYSJ (26) differs from the MWD form for similar reasons. While the MWD forms of 57 Jg exhibit signs of an evolving Ancient Clerical orthography, FYSJ (26) preserves Small 93  Seal orthography. The distinction between MWD and FYSJ orthography can be rendered in terms of specific contrastive features. For example, the f_| element in FYSJ (26) is roundish compared to the square horizontalness characterizing MWD graphs (7) and (8) — e.g., compare FYSJ © with MWD  FYSJ (26) has clearly articulated strokes and is relatively  pictographic compared to the highly linearized rendition of si,§[ seen in MWD (7) and (8). Further, unlike FYSJ (26) the MWD form contains a tail stroke that extends to the right of the graph proper — e.g., compare FYSJ X with MWD ^ w . As such, in terms of overall orthographic structure, the FYSJ form of si , H more strongly resembles Small Seal graphs like STI (1) and SWJZ (2). The orthographic features implicit in the SHD, MWD and FYSJ graphs enables us to date them in relation to one another. Because of the linearization evident in MWD (7) and (8), there is no doubt that FYSJ (26) represents a comparatively older form. At the same time, its vertical rectangularity and structural similarity to Small Seal STI (1) and SWJZ (2) suggests that FYSJ (26) was written prior to any of the SHD graphs. On the basis of what proof is offered here, I hypothesize that the earliest form of Ancient Clerical orthography possessed a strong vertical rectangularity that matched Small Seal script. The FYSJ dang % graph discussed in section two of this chapter (see example 34) may be cited as another example of this early Ancient Clerical orthography. The tentative proposal 1 make here is very intriguing and it may have a significant impact on our understanding of how Ancient Clerical script evolved. But further orthographic analysis is clearly necessary to substantiate this finding. Having dated FYSJ graph (26), I will now examine the other graphs found in FYSJ slip S050. Does the general orthography found on FYSJ slip S050 match that of FYSJ (26)? Since many of the characters on the preceding slip S049 are the same as those on S050, whenever suitable I will analyze them in pairs. By adopting this approach I will simultaneously be able to date slips S050 and S049, as well as discern whether or not they originally belonged to the same original manuscript. The fact that both slips contain verse from Mao Shi no. 44 suggests they originally belonged to the same manuscript before it was broken into fragments and partially integrated into the FYSJ. Although this methodological principle did not stand up to scrutiny in the graphological analysis conducted in section two, I still believe it to be generally valid. 94  FYSJ S049 contains the following verse from Shijing no. 44, stanza one: Z l ^ ^ H _ _ 7 ? TfeMWS^P- Some of the characters in the FYSJ version differ from those found in the corresponding section of the Mao Shi — i.e., ~^^i^ Hi}lK9M"a :  J S ^ "Two young  gentlemen embark in the boat, floating on, it goes far away; longingly I think of you ..." (Karlgren, 1944: 185). Of course, Karlgren's translation of this passage must be reconsidered before it can be deemed applicable — whether in part or in whole — to the FYSJ version; however, this is not the objective of the graphological study conducted here. In FYSJ slip S050, the transmitted variant jing M 'bright' (GSR: no. 755) is replaced with guang % 'bright' (GHYZD: 90), but the rest of it is identical to slip S049. (In OC both of these characters is pronounced with a [k-] initial and, according to Chen p. 279, belong to the yang H§ rime-group. Based on their similar OC pronunciation and semantic value Hu Pingsheng, p. 57, asserts that jing §: and guang 3fe are tongjia zi  or 'allographs'.)  Below are the graphs which undergo analysis in my comparison of FYSJ slips S050 and S049, as well as the ensuing S051:  Our first look at FYSJ slips S050 and S049 involves graphs (28) (S050) and (29) (S049), both of which are forms of the character cheng  'to mount, ascend, ride, drive' (GSR: no. 895).  According to Karlgren, the OBI form of this character, seen in graph (27) (JGWB: no. 719) above, depicts "a man [da  mounted on a tree [mu  (GSR: no. 895). Lin Yiguang # H  3 £ asserts a similar view, adding only that in Zhou BRI two symmetrical graphemes shaped like DC were grafted to the main body to represent the feet of the person in the tree (JGWZGL: no. 239). Graphological evidence given in Gao Ming's GWZLB confirms the appearance of the 'foot' graphemes as early as the middle of the Zhou dynasty — e.g., see 95  example (31) (Gao: 274; also see JSZD: 26). These two graphemes can be clearly discerned in the SWJZ Small Seal form of this character, seen in the above example (30) (Gao: 274). Unfortunately, among the stone inscriptions compiled in the SKZWB and the JSZD there are no SD, ZCW or STI Small Seal forms of cheng  to further substantiate this observation.  In my opinion, there is no doubt that FYSJ (28) and (29) are from the same original manuscript. While they exhibit some minor calligraphic variation, in terms of orthographic form these two graphs are immistakingly the same. As such, the cheng ^ graphs in FYSJ slips S050 and S049 function as strong evidence for the assertion that these slips represent the same source manuscript. But what does the orthography comprising (28) and (29) tell us about when they were written? Might their orthography refute the date given the si Jg graph in FYSJ example (26)? To answer these queries, graphs from the SHD and MWD must be taken into consideration. As is the case with the SWJZ Small Seal form of cheng ffc, in SHD graphs (32) and (33) (QJWZB: 410) the arms of the human figure in the tree no longer form part of the orthographic structure. In the GWZLB there is evidence showing this orthographic development as early as the mid-Zhou— see BRI graph (31) (JWB: no. 738). However, while the two symmetrical graphemes representing feet are written separately in SWJZ (30), in SHD (32) and (33) they have joined together and been orthographically altered. Consequently, the da fc and mu fc matrographs in these SHD graphs appear as three horizontally linearized strokes one above the other. But there is another major orthographic development manifest in the SHD form of cheng  In SWJZ (30) the mu fc matrograph appears as a vertical line passing through two  U-shaped elements, one pointed upwards and the other downwards. By the late Warring States, the top U-shaped element had already become a straight horizontal stroke as a result of linearization. And the lower U-shaped element had transformed into two shortened slanted strokes. The different mu fc form and more linearized da fc element seen in SHD graphs (32) and (33) are crucial features for dating cheng ^ in Ancient Clerical script. Another important characteristic in the SHD form of cheng f|t is its box-like shape. The MWD form of cheng ^ can be seen in example (34) (LZB: 1075). Compared to 96  SHD (32) and (33) what is most distinctive about (34) is its heightened linearity. While the simplified da fc element in this MWD graph is still not entirely straight, the horizontal lines symbolically representing 'feet' and 'tree branches' have no slope or angle in them whatsoever. As is evident in YQS graph (35) (LZB: 1075), the tendency towards increased linearity appears to have been a continuous process in the Han dynasty. Naturally, as discussed previously, the vertical rectangularity characterizing MWD (34) is typical for the orthography of Ancient Clerical MWD script. In exhibiting this verticality the direction taken by Ancient Clerical orthography in the early Han dynasty seems to contradict developments manifest earlier in SHD script. That is, rather than follow the SHD form and maintain less of a vertical orientation, MWD orthography reverts back to a more verticalized form resembling Small Seal script. This phenomenon has been discussed earlier. One final observation to make concerning MWD (34) involves the two downward slanting strokes at the bottom of the graph. According to the SWJZ, these strokes represent the roots of a tree (SWJZZ: 6:1a). In MWD (34) they vary in length, with the right stroke extending well beyond the main body of the cheng graph. In terms of orthography, FYSJ graphs (28) and (29) are both quite narrow and tall. As such, there appears to be a resemblance between cheng f|t in FYSJ (28) and (29) and its counterpart in MWD (34). However, while the verticality of the FYSJ graphs rules out the possibility they belong to the same generation of Ancient Clerical script as the SHD, it does not necessarily mean that they approach the MWD in date. I submit that there is a possibility that FYSJ (28) and (29) are written in an orthography antedating the SHD. My rationale for this proposal is chiefly contingent on three main factors: (A) The symbolified 'man' and 'feet' components in (28) and (29) are less linear than in the SHD form, and thereby more strongly resemble the heightened pictorial quality implicit in Small Seal; (B) Neither the strokes representing the roots of a tree or the decreased linearity in (28) and (29) resemble what is seen in the MWD form; (C) The vertical rectangularity characterizing FYSJ (28) and (29) also suggests an affinity to Small Seal orthography. In this way, FYSJ (28) and (29) resemble the FYSJ form of si Jg. Another interesting aspect of the FYSJ form of cheng ^ is the long vertical line 97  running down through the center. None of the other Ancient Clerical forms of cheng  share  this orthographic feature, and it is tempting to use it for dating FYSJ (28) and (29). But since the Small Seal form of this character also lacks what may be described as a long vertical 'trunk', it is difficult to know exactly how to use this uniquely FYSJ characteristic for such a purpose. I believe the evidence provided by FYSJ (25), (26), (28), (29) strongly suggest that FYSJ slips S050 and S049 were written at a time predating the SHD. If the assertions " concerning the aforementioned graphs are deemed valid, then these slips were likely written sometime around the middle of the third-century B.C. At the same time, these two sets of FYSJ graphs also show that slips S050 and S049 are part of the same manuscript. Further proof of this last point is evident in the way other graphs from these two slips match both calligraphically and orthographically. For example, the two zi ^ 'son, daughter, child' (GSR: no. 964) graphs in these slips — FYSJ (36) from S049 and (37) from S050 — are nearly identical to one another. The same can be said for FYSJ (38) (S049) and (39) (S050), which are both graphs of the character bao _ 'shrubbery, bushy' (GSR: no. 1113). The vertical rectangularity evident in FYSJ graphs (36) through to (39) clearly resembles that seen in FYSJ (28) and (29). The significance of this similarity is two-fold: not only does it show that all these graphs were likely written by the same scribe sometime during the late Warring States period, it also indicates that FYSJ slip S051 does not belong to the same manuscript as slips S049 and S050. Compared to the manner in which the graphs on FYSJ slips S049 and S050 are written, the orthography characterizing FYSJ slip S051 seen in (40) and (41) is quite contrastive. Both FYSJ (40), an example of the character guo H 'state, country' (GSR: no. 929), and the graph you fc 'assist' (SWJZZ: 2:21a) seen in FYSJ (41), are very broad and box-like in shape. On the basis of this very salient orthographic dissimilarity, I believe the manuscript represented by S049 and S050 does not extend to S051. But what of FYSJ slip S048 and those which precede it? How many other fragments belong to the manuscript represented by FYSJ S049 and S050? I attempt to address this question in the following section. In order to do so effectively, it is necessary to consider the 98  following graphs from both the FYSJ manuscript and various other sources:  ^  ^  (49) 7(50)  TV  (56)  L(57)  (62)  (69)  I  (63)  3£ (51) •  t^t (52)^1  (53)  *H(58)"7%(59)  I (64)  (70) 1 (71) 1  (66) ' /  I (65)  (72)  2* (54)  (55)  V(60)  (67)  (73) ^ )  V(61)0  _ (74)  (68)  f r  (75)  None of the characters on FYSJ slip S048 are found among the twenty-one characters written on FYSJ slips S049 and S050. This presents a challenge for the graphologist. As should be obvious from the analysis of slips S049 and S050, direct comparison between two graphs representing the same character usually yields the clearest results. The calligraphy on FYSJ slip S048 looks very similar to that seen in the two subsequent slips: graph size, the space between graphs, and the width of the strokes all appear the same. Consequently, a cursory analysis of slip S048 in relation to S049 and S050 would likely lead to the conclusion that all three FYSJ slips were written by the same hand. However, this interpretation is challenged by orthographic evidence to the contrary. Hoping to discover if FYSJ slip S048 is actually affiliated with S049 and S050,1 will begin my analysis with the FYSJ S048 character guan % 'flute' (GSR: no. 257). The guan % graph from FYSJ slip S048 is seen in S049 and S050. This character is comprised of three different components: on the bottom, yuan % is a drawing of a human figure with its head marked by two horizontal lines (GSR: no. 257; JGWZGL: no. 23); above 99  this is the grapheme representing a "roof (GSR: no. 146; SWJZZ: 7:6b); at the very top, zhu t-T is a pictograph o f bamboo' (GSR: no. 1019; JGWZGL: no. 3097). The FYSJ form of guan ^ consists of several salient characteristics: (A) it shows distinct vertical rectangularity; (B) one of the legs stemming from the 'head' grapheme is long and hooked at the end; (C) the ends of the 'roof element do not go lower than the 'head' grapheme; and (D) the 'bamboo' element is articulately written. More specifically, in this pictographic form the two small leaves stemming from vertical stalks of bamboo are clearly defined (see Ye Yusen 3g3_$£ in JGWZGL, no. 3097). Unfortunately, none of the graphological sources at my disposal — i.e., GWZLB, SKZWB, JSZD, QJWZB, LZB — contain either Small Seal or Ancient Clerical forms of the character guan  I will, therefore,  approach guan ;PG by analyzing it in two separate parts. First, I will treat wan ^5 as one unit — i.e., the 'roof mdyuan ju 'man' components. Following this, I will conduct an analysis of zhu  ti-  The SWJZ Small Seal form of wan % seen in (43) (QJWZB: 552) differs considerably from the FYSJ form of this character found on slip S048. The ends of the 'roof grapheme in the Small Seal form droop down very low to the bottom of the graph, and the legs in its yuan j_  matrograph contain rounded curves. Since the si ,© and cheng M graphs in FYSJ slips  S049 and S050 tend to resemble Small Seal form, this first piece of orthographic evidence from FYSJ S048 seems to suggest that this slip is not related to the same manuscript. Actually, the SHD forms of wan % seen in examples (44) and (45) (QJWZB: 552) are more closely related to Small Seal form than the FYSJ form. Although SHD (44) and (45) are much less vertically oriented than SWJZ (43) — which is to be expected in the early form of Ancient Clerical orthography, their roof and yuan jt components are relatively similar to the Small Seal form. The implications that this orthographic affinity has in terms of dating SHD (44) and (45) should be obvious. By contrast, MWD (46) (LZB: 599) looks remarkably similar to SHD (44) and (45), which I think suggests this graph represents a relatively early orthographic form despite being written in the early Han dynasty. But among the examples given above there is another MWD form of wan % which contrasts (46). In MWD (47) (LZB: 599) the roof grapheme 100  does not droop low at the ends and the grapheme representing a man's legs are no longer bent. Moreover, MWD (47) also exhibits a more salient vertical orientation. My view that (47) manifests more orthographic maturity than the SHD examples of wan % is substantiated by JY graph (48) (LZB: 599). This graph was written later than MWD (47), and it still retains similar orthographic features. In sum, the orthographic evidence presented in regard to wan % convincingly indicates that FYSJ (42) was written sometime closer to the beginning of the Western Han than the end of the Warring States. There is no disputing that FYSJ (42) exhibits a much closer orthographic affinity to MWD (47) and subsequent JY graph (48), than it does to either SHD (44) or (45). The 'roof and 'head' elements in the FYSJ and MWD graphs match almost exactly. Further, in both cases the right leg in the yuan J Q component extends down considerably farther than the left leg. Perhaps the comparatively extreme length of this particular stroke in FYSJ (42), as well as its fine lines, can be attributed to calligraphic style or spatial considerations. The one trait setting MWD (47) apart from FYSJ (42) is the degree to which the right 'leg' stroke inyuan ju extends rightward. In MWD (47) this stroke extends beyond the 'roof element, while in FYSJ (42) it does not. This same orthographic feature is seen in the MWD form of cheng ffc and it may represent a minor orthographic transformation that occurred sometime after the writing of FYSJ slip S048. If this is the case, we may assign a more precise date to FYSJ (42), placing it in terms of orthographic maturity before the MWD. Orthographic evidence derived from the zhu  element for the purpose of dating  FYSJ (42) is also fairly persuasive. When used as a semantic determiner, the Small Seal form of zhu |-j is generally written in one of two ways. The first is evident in (49) (Gao: 309), which is the SWJZ Small Seal form of the character^ =p 'a wind instrument [made of bamboo] with thirty-six tubes' ( _ _ _ - | - 7 N S _) (SWJZZ: 5:17a). In this form the short strokes stemming from the vertical stalks of bamboo are straight. Based on graphological evidence provided in the GWZLB, this is probably the oldest form of zhu t-j, going as far back as Shang OBI — see example (55) (Gao: 309). The second form can be found as the semantic determiner in SWJZ Small Seal (50) 101  (Gao: 310) ji % 'winnowing basket' (GSR: no. 952). Unlike the zhu fs element in (49), in this case the short strokes stemming from the central vertical line are slightly arched, or rounded. In bronze and stone inscriptions written before the Qin dynasty, this form of zhu fS, in one variation or another, is the most common (JSZD: 953-72). According to my survey of the QJWZB (329-35), the SHD form of zhu fs is usually highly abbreviated when used as a classifier. As such, its pictographic quality is sharply diminished. For substantiation of this claim refer to SHD (51) (QJWZB: 332) and (52) (QJWZB: 335), which are examples of the characters deng If 'step of a stair' (GSR: no. 961) and zhen % 'needle' (GSR: no. 671), respectively. These graphs show how the classifier zhu fs is often written in the SHD as a longish horizontal line intersected by two straight vertical lines. By contrast, a survey of the LZB shows that in MWD orthography zhu fs reverts back to a form more similar to Small Seal. The MWD form of zhu fs is typically written as a vertical line accompanied by two straight strokes slanting downward. For illustrations of this tendency refer to graphs (53) (LZB: 1173) and (54) (LZB: 1176), the first of which is another form of the character deng If, while the latter is jie gjj 'knots or joints of bamboo or other plants' (GSR: no. 399). An examination of zhu fs in FYSJ (42) indicates a strong orthographic affinity to both the Small Seal, as well as the MWD Ancient Clerical form. Since the rest of FYSJ (42) exhibits features more closely related to the MWD orthographic form, I will interpret the evidence provided by zhu fs in (42) as indicating the same thing. As a result, its seems probable that the zhu fs classifier in FYSJ (42) represents a later orthographic form. Of course, this view falls in line with the conclusion arrived at concerning wan  giving us even  more reason not to associate FYSJ slip S048 with the same manuscript as FYSJ slips S049 and S050. Assigning a date for an entire FYSJ slip based on the analysis of a single graph lacks persuasiveness. More orthographic evidence must be gathered and compared to enhance the integrity of the dating procedure. To my mind, most of the other graphs on FYSJ slip S048 are to some degree suitable for a concentrated comparative analysis. However, the character 102  wei H 'avoid, taboo' (GSR: no. 571), which is one of the graphs represented on slip S048, is an exception. Not enough primary materials exist for wei U to analyze it with the same degree of certainty found elsewhere. For while numerous forms of wei !$ exist in Zhou BRI and stone inscriptions (JSZD: 1201), only one is found in excavated materials from the Qin and Han (LZB: 1355). Moreover, the QJWZB and LZB both lack examples of the graph wei  which  might serve as a substitute. Since there is an abundance of orthographic evidence to work with on FYSJ slips S048,1 will by-pass discussion of wei ft here. The next character from FYSJ S048 to undergo examination is you ^ 'have, there is' (GSR: no. 995) — e.g., FYSJ (69). In the Small Seal form ofyou M seen in (70) &you X element representing a hand with three fingers (SWJZZ: 3: 17b) is set above another element that stands for 'meat' (#J-M¥#p*l) (see Yang Shuda #jfM in JMJWCD: 123). But the combination of these two elements to form the character you ^ did not begin until early Zhou BRI. Prior to this the primary OBI form of this character consisted of the hand only (Gao: 61) — see OBI graph (75) (JGWB: no. 350). Even at a glance, it is easy to tell that FYSJ (69) more strongly resembles early Han MWD graphs (73) and (74) (LZB: 64), than the Warring States SHD graphs (71) and (72) (QJWZB: 196). The most striking characteristic setting SHD (71) and (72) apart from their counterparts is the shape of you X In the SHD this element is typically more angular and positioned on more of an angle. Additionally, whereas the bottom ends of the 'meat' element in SHD (71) and (72) are more even, in the FYSJ and MWD orthographic forms one end extends lower than the other. But FYSJ (69) is not exactly identical to the MWD forms either. For example, in MWD (73) and (74) you X is almost completely horizontal and the short strokes written inside the graph representing the meaning of 'meat' are abbreviated. I interpret this evidence as indicating that FYSJ (69) was written later than the SHD, but not quite as late as the early Han. Thus, orthographically speaking, FYSJ (69) supports the date assigned to slip S048 given earlier. FYSJ slip S048 also contains the character shuo 103  'speak' (GSR: no. 324), which is  seen in example (56). In the right-hand side of this teknogram is the element dui j& 'glad, open' (GSR: no. 324). In regard to dui j& Karlgren writes, "The graph has 'man' [ren Jl] and 'mouth' [kou P], meaning of the top strokes obscure" (GSR: no. 324). Xu Xuan  argues  that the upper part of dui j& is composed of kou P 'mouth' and ba /\ and that together the idea of air being dispersed is conveyed ( „ £ £ P $ A  ,fe&£.#S0(JGWZGL: no. 43). This  interpretation is also found elsewhere and will be adequate for the present needs of this study (KMJS: no. 1062). Karlgren writes that the element yan if 'big flute, speak, talk' is "a drawing of a kind of flute" (GSR: no. 251). Xu Shen agrees thatj^m "gf conveys the meaning of'talk' or 'speech' ( J l i ^ M W ^ i f X but he does not commit himself to an explanation of the character's structure (SWJZZ: 10:7a). Both these interpretations are deficient: Karlgren's view in this case is simply incorrect, and the interpretation given in the SWJZ is too vague. For the paleographical view, I turn to the summary in JGWZGL, no. 722 which reads, "The primary form of yan if is derived from she i§- [the graph for tongue]. The addition of — [horizontal lines] above signifies speech being emitted from the tongue, making it a hypodigmatic character" (  „  , MS$±  , m m ft^S" , U t t * * ) . (The term  'hypodigmatic' may be defined as follows: "graphically indicative or suggestive in a nonarbitrary way of a word with an intangible and therefore not concretely depictable meaning" — see Boltz, 1994: 180.) This description becomes clearer if the OBI form of this character is referred to — as such, see example (61) (JGWB: no. 277). In Ancient Clerical script, the she  element became completely linearized. Since the  QJWZB does not have an example of the character shuo  instead I use the SHD form of  tuo SH 'to peel off (GSR: no. 324) seen in examples (57) and (58) (QJWZB: 835). SHD (57) and (58) exhibit orthographic forms of dui j& that differ considerably. The kou P grapheme in both SHD graphs is quite square in shape, but the ren Jl grapheme in (58) is comparatively of much shorter. Further, in SHD (58) the strokes of the ren JL grapheme originate at the same point on the underside of the kou P grapheme. By contrast, in SHD (57) these downward strokes originate from opposite corners of kou P. 104  Sinceyan g was not discussed in the context of SHD orthography, I will leave it out of my discussion of MWD graphs (59) and (60) (LZB: 1340), as well. Overall, the orthography of dui j& in MWD does not match that of the two SHD examples. To begin with, the MWD form of dui j& is characterized by a kou P grapheme which is less angular than those in SHD (57) and (58). In addition, to an extent unprecedented in earlier Ancient Clerical orthography, the two strokes in the MWD form of ren Ji differ drastically in length. These strokes do not have a bend in them like (57) and they are not short like (58). Moreover, MWD (59) and (60) are more vertically rectangular than SHD (58) and (59). Without a doubt, the dui j& component in FYSJ (56) clearly resembles its MWD counterparts. The orthographic features described in connection with the MWD are practically identical in FYSJ (56). Consequently, I can confidently assign a date to FYSJ (56) which approximately matches that given FYSJ (42) and (69). Example (62) from FYSJ slip S048 is a form of the character^/ W 'to be pleased' (GSR: no. 790). However, (62) is incomplete and lacks the xin ; 0 grapheme which usually serves as the semantic determiner for this character. Both the QJWZB and LZB lack examples ofyi W, but fortunately they contain the character ze fl| 'choose' (GSR: no. 790) which is sufficiently similar to compare with the FYSJ graph. With the material available, I will be able to compare the various forms of the character^/ ij§ 'to spy' (GSR: no. 790) found in the FYSJ, SHD and MWD. But before doing so, a few words concerning the makeup of this character are in order. According to the GWZLBjy? 1§ did not exist before the Warring States period (Gao: 159,133,476). However, reference materials specializing in OBI graphs provide some indication that this may not actually be the case (KMJS: no. 4695). In any case, in both Six States and Small Seal scripts, yi lg has a relatively consistent orthographic form (Gao: 133), as is evident in example (63), from the Warring States period, and the SWJZ Small Seal graph (64) (Gao: 133). Karlgren does not have an explanation for the structural composite of yi Ip (GSR: no. 790). But according to the SWJZ this character is composed of three main components (SWJZZ: 10:13b): (A) at the top is a mu g 'eye' grapheme (GSR: no. 1036 ); (B) in the 105  middle is a da A 'great, greatly' grapheme (GSR: no. 317); (C) at the bottom is a form of gan 'shield' (GSR: no. 139; Gao: 364; JMJWCD: 12). In the Small Seal form of>" W, which bears the characteristic vertical rectangularity, all of these graphic components are clearly visible. The SHD form of ze % can be seen in (65) and (66) (QJWZB: 829). while the corresponding MWD graphs are found in (67) and (68) (LZB: 392). In comparison to SWJZ (64), the main differences in the Ancient Clerical form ofyi Ip is its upright mu @ grapheme. The shape of this grapheme in the various Ancient Clerical scripts becomes progressively squarer with the passing of time. The Small Seal form ofyi ip and its Ancient Clerical counterpart also differ in terms of pictoriality. There is a much greater degree of linearity in SHD (65) and (66), and still more in MWD (67) and (68). Linearization is a crucial orthographic feature distinguishing the Ancient Clerical forms of yi Ipl. Another feature which serves the same purpose is the reduction of strokes. Since SHD (65) has the most strokes, 1 believe it likely represents the oldest Ancient Clerical orthographic form ofyi lj§. The fact it is obviously the least linearized supports this evaluation. By the same token, because it is clearly the most simplified, MWD (67) is probably the latest and most mature form of yi 1|. Placing FYSJ (62) into this orthographic evolution is somewhat complicated, as it contains features characteristic of both the SHD and MWD forms ofyi 1§ — e.g., although the mu @ grapheme in it is roundish, taken as a whole it is nevertheless highly linearized, as well as simplified. Consequently, I think it should be dated sometime between (66) and (68), the latest SHD form and the earliest MWD form. As the result of this analysis, the rough date allotted FYSJ (62) is in alignment with the conclusions reached so far regarding FYSJ slip S048. In sum, in the above graphological study I attempt to show that FYSJ slips S049 and S050 are both representative of the same manuscript. There is an abundance of evidence which supports this view. At the same time, I assert that this manuscript is probably written in an orthographic form antedating the SHD bamboo manuscripts. FYSJ graphs (25), (26), (28) and (29) are all crucial for the formation of this argument. Although not much effort was 106  spent analyzing FYSJ S051,1 believe that the orthographic evidence reviewed in connection with this slip plainly shows that it was not written by the same hand as S049 and S050. The graphs on S051 are typically broader and more box-like in shape than those seen on slips S049 and S050. By contrast, considerable effort was spent proving that FYSJ slip S048 also does not belong to the same manuscript as FYSJ S049 and S050. Rather than being written in a very early form of Ancient Clerical script, FYSJ S048 appears to be written in a form of orthography datable to the Western Han dynasty.  107  Conclusion: In the introduction, I outlined my primary objectives for this chapter. They are: (A) to discover the true physical nature of the FYSJ; (B) to learn when this manuscript was written. In regard to (A), I proved that the FYSJ, as a composite text made up of numerous different manuscripts from very different origins, is much more fragmented than previously suggested by scholars working in the field of FYSJ research. In fact, my research has shown that in many cases as few as two bamboo slips represent one individual original manuscript. This conclusion contrasts Wen Xingfu's somewhat unsubstantiated assertion that groups of up to seven or eight bamboo slips can comprise a single source manuscript in the FYSJ (978). It is not an aim of this study to deal with Wen's observations point by point; however, the limited scope and lack of rigorous methodology characterizing his brief study in themselves make his conclusions suspect. To the detriment of his argument, when evaluating the calligraphy of the FYSJ, Wen fails to apply a methodological approach which takes into consideration the systematic use of analytical criteria on a correlative basis. Further, in that Wen's overall examination of FYSJ script does not operate on a level of graphs and their features, the view he puts forth concerning FYSJ calligraphy tends to reflect impressions, not actual data. Since Wen's orthographic analysis is even more limited — he only takes two graphs into account, it really lacks any significance as a scholarly exercise. Therefore, I see no reason to further criticize the particular failings of his orthographic examination. The manner in which I conducted the orthographic analysis in my own study should suffice to expose the shortcomings implicit in Wen's cursory approach. Finally, for the purpose of discerning graphic dissimilarity in the FYSJ, I do not see the advantage of dividing analysis between calligraphy and orthography as is the case in Wen's analysis. However, rather than discount Wen's graphological study entirely, 1 will suggest that considered together our respective studies might offer a more balanced perspective of the nature of the FYSJ. Certainly, further study along the same lines as what 1 have conducted in chapter two is needed. There are several important concerns which I have not addressed — e.g., Are there any cases of a particular source manuscript appearing in more than one location within the composite manuscript of the FYSJ? Having spent considerable effort formulating an effective approach to orthographic analysis that is specific to the FYSJ, I believe this study may function suitably as a launchpad to further research on this manuscript. 108  Concerning objective (B), for my purposes I feel that detail outweighs scope in importance. In chapter two I analyzed almost two hundred and fifty graphs from roughly a dozen different slips to date the multitude of different orthographic forms used in the FYSJ. After all, to make my contribution to this area of FYSJ research meaningful, attaining a high degree of accuracy is paramount. I believe my conclusions offer much more insight into the date of the FYSJ than what was previously offered by Wen Xingfu (97). Analyzing the FYSJ in terms of orthographic feature, rather than on a level of script, I placed FYSJ graphs within the early evolution of Chinese script. The understanding I reached concerning the common orthographic traits implicit in Small Seal and Ancient Clerical script, enabled me to assign approximate dates to FYSJ graphs using solid orthographic evidence. Clearly, discerning the degree to which elements of Small Seal and Ancient Clerical orthography are present in FYSJ graphs was a painstakingly meticulous process which involved frequent comparative analysis. Considering calligraphy as another factor in my methodology complicated analysis even further, but it contributed to the results obtained. I showed that FYSJ slips S122 and S123 were written in the early Western Han, that FYSJ S104 and S105 are a product of the Warring States period and represent a form of script preceding that found in the SHD inscriptions ca. 240 - 217 B.C., and that FYSJ S049 and S050 are again earlier in origin than the SHD. Arriving at conclusions such as this marks very clear progress in our understanding of the very complex FYSJ manuscript. In light of the evidence provided in this chapter, I question the assumption that the FYSJ can be attributed to one school of Shijing scholarship. To my mind the tendency for Chinese scholars to assert this view despite the orthographic diversity characterizing the FYSJ can only be regarded as overly speculative (Hu, 1988: 30-1, Wen: 98). There is a strong possibility that numerous different versions of the Shijing are represented to some degree in this composite manuscript.  109  Chapter Three: A Selective Lexical Study of the Fuyang Shi jing Introduction:  This study opened with a discussion of the basic methodology needed to research the orthography and lexical variation in the FYSJ. In chapter one, I outlined a methodology for graphological analysis that is based on a knowledge of orthographic development from Shang OBI to the Western Han dynasty. From this broad understanding of orthographic change in early Chinese writing, I derived specific criteria for the analysis of Small Seal and Ancient Clerical scripts. Furthermore, placing the FYSJ within this framework, I conducted a study of this manuscript which maintained two important approaches to graphological criticism: orthographic feature, and contextual consistency. The results of this orthographic evaluation were summarized in the conclusion to chapter two, and need not be reiterated here. The overview of methodology in chapter one also includes a summary of the pre-Han writing system. Furthermore, in this chapter I discuss several approaches to identifying and analyzing graphs and lexicons in the language of this period. For example, I demonstrate that phonetic compatibility is an important criterion for associating different words to a polysemous graph. I also acknowledge phonetic compatibility to be an essential characteristic of word families, as well as the semantic basis upon which phonetic borrowing occurs. But in order to utilize this understanding of pre-Han Chinese in a form of methodology, the phonetic constraints dictating compatibility must be more clearly stated. As discussed in the first chapter, the principles governing phonetic compatibility in pre-Han Chinese are as follows: (A) For finals to be considered phonetically compatible, their main vowel should be identical and their final vowel homorganic — i.e., articulated at the same place in the vocal tract; (B) For initials to be considered phonetically compatible, they too must be homorganic. For my research on the Shijing, I rely primarily on Karlgren's Old Chinese reconstruction and the Shijing rime-groups to conduct tests for homorganicity. The rimegroups derived from the Shi jing are generally considered the most reliable means with which to judge homorganicity in finals. Of course, my overview of Old Chinese phonology and its applications is aimed at serving a specific purpose. In the third chapter, I utilize the above methodology to study the lexical variation discovered after comparing the FYSJ and the various received versions of the Shijing. My chief objectives are: 110  (A) to provide a new reading of certain odes in the Shijing; (B) to provide an academic survey of the types of lexical variation associated with the FYSJ; (C) to determine, in all of the cases investigated, which form of lexical variation represents the proximate original. Chapter three is divided into six separate studies. Lexical Variation Study no. 1 deals with FYSJ slip SO 18 and a form of lexical variation involving syntax. In this study I discuss the absence of the particle hu ^ in the FYSJ and draw a comparison with the Mao and Lu Shi. I also outline two other occurrences in the FYSJ where a change in syntax is implied — i.e., FYSJ S017 and S026. The second study features a case of lexical variation involving synonymity; i.e., hu •£§ in FYSJ SI 15/Lu Shi and he  jfff  in Mao  572/  no. 115. Although these  two studies involve lexical variation, they do not result in a significantly different readings of the Shijing text. This is because the changes in syntax and synonymity discussed in Lexical Variation Studies nos. 1 and 2 have no actual impact on wording. Consequently, Lexical Variation Studies nos. 1 and 2 can be viewed as addressing objective (B). In Lexical Variation Study no. 3,1 shift my attention to objective (A). For example, the character ju  found in FYSJ S045 offers a new reading of Shi jing no. 41, one which  challenges that seen in the Mao version. This study is followed by two similar studies that also produce different readings — i.e., in Lexical Variation Study no. 4, long |S[ from FYSJ S048 is offered in place of long ffr from the Mao/Lu and Qi Shi; in Lexical Variation Study no. 5,1 consider xiang fe from FYSJ S123 in connection with chang ^§ found in three transmitted versions of the Shijing. The sixth and final study in chapter three involves scribal error and is an example of the complex problems that face research on lexical variation. Therefore, Lexical Variation no. 6 should also be considered primarily in relation to objective (B). To my mind, the task of finding the proximate original is a logical extension of the other endeavors engaged in chapter three. Once a significant new reading of the Shijing has been discovered, academicrigordemand that the more legitimate or authentic version be identified. For the study of ancient Chinese manuscripts, the question "Which textual variation represents the original?" forms the motivation behind all textual criticism. In 111  addressing objective (C), I adopt Yumiko Blanford's interpretation of the proximate original, as defined in the following quotation: "The proximate original is the one variant that has a greater probability of representing the original word than does any other variant, while recognizing that the determination of the original cannot always be achieved with full confidence" (188). There are several recognized methods for determining the proximate original. For example, one is the lineage method, which entails finding the variant with the highest frequency of representation (Boltz, 1997: 266-7, 1984: 193-6: Blanford: 202). That is, if there are three or more texts — or lines of textual transmission — which represent lexical variant Y, and only one which represents lexical variant Z, according to the lineage method Y would be the proximate original. On the reliability of this method, Blanford writes, "When this rule is applied, it determines the proximate original almost automatically and greatly reduces the number of variation cases for which we need to use other rules to determine the proximate original" (Blanford: 199-201). A second principle used to discern the proximate original among several lines of textual transmission is called lectio difficiliorpotior, or "the priority of the more difficult reading" (Boltz, 1997: 268; Blanford: 202). Boltz outlines this principle as follows: "When one of two variants is rare, obscure, or in some other way more difficult to understand than is the other one, the former is more likely to represent the original than the latter... The rationale is that it is more likely for an editor or copyist to change, either consciously or inadvertently, a word or phrase that is difficult to understand (the lectio difficilior) into one that is easier to understand (the lectiofacilior), than the other way around" (1997: 268). Finally, a third method that I use for determining the proximate original relies on an examination of context. Sometimes this involves choosing the lexical variant which is better suited to an ode's narrative and theme. Examples of this can be found in Lexical Variation Studies nos. 3 and 4. Other times a study of context might involve semantic parallelism and involve choosing the lexical variant which more closely matches a corresponding word in another line or stanza — e.g., Lexical Variation Study no. 5.  112  LexicalVariation Study, no. 1 FYSJ, slip S018: Mao, no. 25; Lu: hu ^ This study of lexical variation differs somewhat from the others dealt with in chapter three. Instead of comparing two different lexicons against one another, Lexical Variation Study no. 1 represents a case where one textual variant contains a character not found in another. The situation described here still concerns lexical variation, since it involves the study of a particular lexicon and the significance that its presence or absence has in regard to the given context. Yumiko Blanford categorizes cases such as this separately, referring to them as "vacant variation" which she defines as follows: "[cases] in which no word is written in one or more versions while one or more words are written at matching places in another version(s)" (189). Making a distinction between vacant and lexical may sometimes be useful for purposes of description; however, since the procedure followed to evaluate vacant variation does not differ from what is termed lexical variation, I do not distinguish them to the degree Blanford does. (It is worth noting that William Boltz does not recognize vacant variation as something separate from lexical analysis. See Boltz, 1984,1997.) FYSJ slip SOI 8 contains a total of five graphs, and reads:  (Hu, 1988: 3;  Wen: 64). (There is no punctuation in the FYSJ and I do not add any in my citations of it.) The last four graphs on this bamboo slip are undamaged and clearly written; however, the uppermost graph is incomplete due to the damaged state of the material it is written on. But since the shi ^ 'pig, swine' (GSR: no. 1218) signific is still discernible in this graph, we may safely identify it as zong $£, which is variously defined by Han glossarists as either a 'oneyear-old' or 'three-year-old' pig (see the Mao zhuan and Zheng Xuan's jian in Wang Xianqian: 517; also see SWJZZ: 9:36b). The corresponding passage in both the Lu Shi and Mao no. 25 reads: J S § | £ #  , TSI^I1SJ3I  "By one discharge five young pigs! Lo, you young  grooms and gamesters! "(ICS: 10:11; Karlgren, 1944:179; Wang Xianqian: 120). Before entering into a discourse focusing on the character hu  ^ *g D (GSR: h  no. 55), I will provide a  general overview of the above verse-line from the Mao Shi. Because the vacant variation in Lexical Variation Study no. 1 has an impact on syntactic structure, it is necessary to examine 113  the entire verse-line containing hu first. This approach will make it easier for the reader to understand the discrepancy between the variations examined in this study. The character zong marks the end of line two in the third stanza of Shijing no. 25. Since its position is situated outside the verse-line containing the particle hu  it has no  bearing on the lexical study underway here. In the following third line, there are two characters which I will endeavor to explain before discussing the particles preceding them. These characters are zou iff and yu JJL According to the Mao zhuan, zou Jg and_yw Jjt combine to stand for a kind of tiger that is white with black stripes. This tiger displays a quality of benevolence by refusing to eat raw meat (IgjJt , iiiR-tiJ °fii^M3<C,  ^^^M)  (Ma: 104; also see Karlgren, 1942: 106). On the other hand, the Han school's gloss states that zouyu HjJt was the official title of those responsible for overlooking the animals in the imperial menagerie (MWkfc^M-WMM) (Ma: 104; Wang Xianqian: 120; Karlgren, 1942: 106). To complicate matters even more, the Lu gloss extracted from the Xinshu 0fW (The New Text) — a text written during the Han dynasty — takes zou H to refer to the park of Emperor Wen (SC^-^M), tmdyu j£ as its games keeper (B^WJSO (Ma: 104: GSR: no. 59, 132). In his "Glosses on the Kuo Feng [Guofeng] Odes," Karlgren outlines his preference for the interpretation attributed to the Han school: (A) The interpretation put forth in the Mao zhuan does not fit the overall context of the ode; (B) The definition offered in the Lu gloss lacks substantiation from other preclassical or classical texts (1942: 106). (According to Pulleyblank, p. 3, the term "preclassical" refers to the period roughly antedating the mid fifth-century B.C., while "classical" postdates this.) I agree with (A) since the ode would not make any sense as a hunting song or anything else if the meaning of zouyu HJH was changed to 'zou-yu beast' or 'white tiger' (refer to the translation in Karlgren, 1942: 179). Qing philologist Mao Ruichen may have been correct when he wrote that zouyu HHt originally referred to a kind of animal, but was later borrowed to represent an official function or capacity (106). In support of the Han school's interpretation, there are several preclassical and classical texts which refer to zou fg as a "keeper of horses and caretaker of carriages"  ^ffiltflifiA)  — e.g., the Zuo zhuan _-ff (The Chronicles of Zuo) 114  and the Liji  ijgfS (The Record of Ritual) (ICS: 11:225, ICS, Liji: 6; HYDCD. 12:868). Further, according  to Han glossarists commenting on the preclassical Shang shu j^jfl (The Book of Documents), the character^ Jjf referred to an official who overlooked the wilderness and wild game ("£f^ #WllJ#JI|#^'B') (ICS: 9:3). The c h a r a c t e r £ is similarly used in texts written during the classical period (HYDCD: 8:846). Having dealt with the content words (shici  |H|) in the verse-line presently under  scrutiny, I will now focus on the particles (xuzi _ ^ ) y w T and jie lg. To begin with, there is evidence that>^ ^ already functioned as a particle in Shang OBI. Moreover, }w  continued  to be common as a particle throughout the preclassical period and until the late Warring States period when it was replaced in general usage by the particle^/ ]fc *o (He: 729: GSR: no. 61). Line three in the third stanza of FYSJ/Mao Shi no. 25 opens with yu  *gjwo (GSR: no. 97),  which has several functions in preclassical Chinese — e.g., it indicates the passive use of a verb and represents the words 'to, at, in, from' (Dobson: 859-60; in Classical Chinese yu replaced>>w ^ to express passive sense — see He: 737). For example, in Mao no. 261 there is a line which reads: A l E T i E (ICS: 10:138). Dobson translates this sentence in the passive sense: "[He] entered and was received in audience by the king" (860). Further, a line from Mao no. 127 reads: S t & T ^ f "Follow the duke to the hunt" (ICS: 10:55; Dobson: 859; also see He: 730). However, none of these meanings apply to yu Instead, I believe that the yu  in FYSJ SO 18 and Mao no. 25.  found here is either an abbreviated graphic variant, or  erroneous form of xu of *5(jwo (GSR: no. 97), which is nearly homophonous to yu ^ in Old Chinese. According to Karlgren's reconstruction,}^/  and xu nf are homorganic in terms of  initials; that is, both are produced from the soft palatal region at the back of the mouth. (O'Grady, p. 24, defines the velum as the "soft area towards the roof of the mouth"; whereas, the uvular is the "smallfleshyflap of tissue ... [which] hangs down from the velum.") Actually, in his "Loan Characters in Pre-Han China," Karlgren himself states that the initials [*g-] and [*%-] are phonetically compatible (1963: 10). As far as finals are concerned, both yu  and xu tff are identical in Karlgren's Old Chinese, and belong to the yu 115  ft rime-group (Chen: 172-3). The homorganicity implicit in the Old Chinese pronunciation of these two characters makes them freely interchangeable in preclassical texts such as the Shi jing. However, if yu  is just another form of the particle xu uf, what does it mean? From the  early Zhou to the Han dynasty, xuffffunctioned as a self-standing interjection to either express surprise and apprehension, or mark a sigh or summons (Dobson: 628; He: 638; GSR: no. 97). The Lu version of Shijing no. 25 uses xu Iff and, as will be seen shortly, an interjection really suits the context of the verse-line (Wang Xianqian: 120). Elsewhere, xu uf is used in a similar manner in Han Shi no. 11, as well (Wang Xianqian: 61). In the same verse-line of Shijing no. 25, another particle that frequently "stands alone" in early Chinese texts is situated immediately after yu ^Ixu of. I refer to a character written t_ in FYSJ S018 and I}i in the Mao/Lu Shi which has several functions in texts written prior to the Han dynasty. These include: (A) It is a metrical particle — e.g., Mao no. 191: If  "Even so, no one takes warning" (ICS: 10:88; Dobson: 402); (B) It is a  demonstrative pronoun — e.g., Mao no. 69: {SflJi&^l "What did this lead to?" (ICS: 10:32; Dobson: 403); (C) It is an interjection — e.g., see the line _ ^ ' S A "Ah! my loved one!" from Mao no. 3 (ICS: 10:2; Dobson: 403; also see He: 286-7, GSR: no. 5). In the context of the verse-line studied here, the meaning conveyed in (C) is the most suitable. There is no doubt that jie !_ and jie IJ_ are graphic variations representing the same word with an identical meaning. Firstly, this kind of variation is hinted at by certain graphological evidence. In the unattested HI form seen in FYSJ SO 18, the conventional kou P 'mouth' signific has been replaced with a yan H" speech' signific. Phonological consideration of this relationship results in the very same conclusion: kou P and yan "gf are not the core phonophorics in their respective graphs; subsequently, their exchange has no bearing on pronunciation — which Karlgren reconstructs as *tsja in Old Chinese (GSR: no. 5). Additional proof that jie M and jie IU are variants of the same graph and mean the same thing can be found in other excavated manuscripts from the Han dynasty. In the MWD inscriptions written during the early Han, _ is also written as in the form of l i (Hu, 1988: 26,44). The phrase yujie ^p[]§ / ^f!_ found in FYSJ S018 and Mao no. 25 functions as an 116  interjection in archaic and ancient Chinese. Although Han dynasty glossarists do not discuss this phrase in regard to Shijing no. 25, they do comment on it elsewhere. In Mao no. 11 there is a line which reads: Tig^^ "Oh, the lin [a mythic creature with the body of a deer, the tail of an ox and a single horn]" (ICS: 10:5-6; Karlgren, 1944: 175; Yuan Ke: 451). The gloss attributed to the Han school comments on the yujie ^ I3H seen here as follows: "It is a word [expressing] a sigh"  _. (The Han gloss uses the attested graph xu nf instead ofyw =f —  see Wang Xianqian: 61.) Further, the zhuan for Mao no. 11 takes yujie ^flU to mean exactly the same thing (Wang Xianqian. 62). Contemporary scholars are in agreement with the glosses for yujie  written during the Han dynasty. For example, Dobson demonstrates that in  preclassical texts yujie  H functions as a self-standing interjection which expresses "a groan  or a sigh" (860). Similarly, in Gudai Hanyu xuci tong shi "r^ftitllt_ii$lff (A Complete Interpretation of Ancient Chinese Particles), He Leshi -fqf^i also demonstrates that this phrase means 'to sigh' (638-9). Dobson supports this interpretation with an example from Mao Shi no. 31, which reads:  "Alas, [we] are parted!" (ICS: 10:14; Dobson: 861).  In Mao/Lu Shi no. 25, the particle hu ^ is affiliated with the interjection yujie T As a particle, or function word, hu  has several different uses in preclassical and classical  Chinese (He: 226-30). But its function as a marker of "vocatives, imprecations, expletives, etc." is of chief importance here (Dobson: 330). (Vocatives are words used for direct address; an imprecation refers to an offensive utterance or invocation of evil; and expletives are. " A swear word, particularly one inserted into the middle of a phrase as a meaningless emotional intensifier" — see Trask: 98,136,299.) Dobson cites ? H ^ as an example demonstrating 1  the hu ^ particle's role as a marker. He indicates that this example is drawn from the Shijing and translates it simply as "Alas!" Since only Mao Shi no. 25 contains ^fH^F-, it is certain that Dobson's example was taken from this ode (see ICS: 10:419). As such, taking Dobson's interpretation into account, we may read T  in Mao no. 25 as conveying a vocative  sense; that is, as addressing the "young grooms and gamesters" mentioned in Karlgren's translation. A similar example of hu ^ from Mao no. 256 reads: Jt^-hi "Alas! my child" 21  (ICS: 10:133; Dobson: 330). In this case, the particle hu ^ marks yu 117  another preclassical  form of vocative interjection used either singly or in collocation with other characters (Dobson: 864). For our purposes, it is important to stress that in both of these examples the interjection is not dependent on hu ^p-; that is, the interjections yujie : p _ and yu ]fc can be found in the Shijing unaccompanied by the marker hu  And it is precisely the independent  nature of yujie  in FYSJ SOI 8. The absence of the  that allows for the segment  ^nflffjjf  marker hu ^ in this slip does not have a significant impact on the meaning of the verse-line. Clearly, with or without hu  the line should be translated the same way. Both Karlgren's  "Lo!" and Dobson's "Alas!" are permissible because they imply that somebody is being addressed. The shortened verse-line in FYSJ SO 18 provides valuable insight into how the various versions of the Shijing took form. As Hu Pingsheng comments, FYSJ SO 18 shows that the use of particles in the Shijing originally depended a lot on the conventions of the different regions and dialects in the early Zhou dynasty  WS.BHWSiS)-  As well, the singers, scribes  and compilers linked to the Shijing in their various capacities probably all contributed to the use of particles and variable syntactic structure found in the text ( j t l l f r f  mmmmmmm ,  , mmmm^m  (1988:  0  ,  fti^fRjUPJ  , t#  27). In the FYSJ there is further  evidence that vacant variation occurred among the different versions of the Shijing in ancient China. For example, both FYSJ SO 17 and S026 also appear to represent a version(s) of the Shijing which is characterized by shortened verse-lines. Of course, the close proximity of SO 17 and SO 18 suggests they are possibly from the same original manuscript. And the same may also be true of FYSJ slip S026. But an orthographic analysis of these slips, conducted in the manner seen in chapter two, would be necessary to prove such a conclusion. Due to the effort required for such an endeavor, I will not attempt to connect these FYSJ slips here. I feel that combining orthographic and textual aspects is the next step in the research of the FYSJ. Certainly, the work done in this thesis should facilitate further research along these lines. In FYSJ slip SOI7, the text reads: "... 'Fallow-deer': forty-four characters" ^ W © ^ (Hu, 1988: 3; Wen: 64). Both Hu Pingsheng and Wen Xingfu believe the contents of this slip refer to the number of characters used to write an alternate, previously unknown version of 118  Shijing no. 23. They base their conclusion on the character jun | f "fallow-deer" (GSR: no. 485). This FYSJ character is a graphic variant ofjun  as seen in the contents of Mao Shi  no. 23, as well as this ode's traditional title (ICS. 10:11). Qing scholar Duan Yucai concluded that the jun g component in Jiif  —  which depicts an enclosure surrounding a stalk of grain  — likely took the form of he ^ as part of the orthographic simplification characteristic of Small Seal script ( M ^ i t l t H ^ - f i ) (SWJZZ: 10:22; GSR: no. 485; Hu, 1988: 44). Karlgren agreed that these two graphs are variants; he believed J f f to be the attested form and jun H the variant (GSR: no. 485). A count of all the characters used in Mao no. 23 shows that there are forty-seven, three more than in the Shijing version referred to in FYSJ SO 17. Since stanzas one and two of Mao no. 23 contain only tetrasyllabic verse-lines, the extra characters are unlikely to be found here. But the conditions differ in this ode's completely pentasyllable final stanza, which reads: #?7OT£^ , M&Wfc^  , M&MWt "Slowly! Gently! Do not  move my kerchief; do not make the dog bark!" (ICS: 10:11; Karlgren, 1944: 178). Hu Pingsheng believes this stanza contains several particles that could be deleted. He recommends removing the particle xi ^ from lines one and two, and ye i+i from the third line (1988: 26, 44). By doing so, Mao Shi no. 23 would be written entirely in tetrasyllabic verse, and at no loss whatsoever to its meaning. FYSJ slip S026 refers to yet another unknown Shijing text which contains fewer characters than its transmitted counterparts — or at least the Mao Shi. Slip S026 reads 0 ft A + T N ^ , which may be translated as, "'Oh sun, oh moon!': Ninety-six characters" (Wen: 1  66; Hu, 1988: 4; for my translation of the first two characters in S026,1 refer to Karlgren, 1944: 180). Whereas the first part of this slip relates the traditional title of Shijing no. 29 — e.g., 0 ft "Oh sun, oh moon!"; the remaining portion states the number of characters this ode contains. By contrast, the Mao version of this ode contains a total of ninety-nine characters, which means it is three characters longer than the one referred to in FYSJ S026. Although the FYSJ contains two other slips which pertain to Shijing no. 29 — e.g., S024 and S025, nothing in them helps explain which three characters are missing in FYSJ S026 (Wen: 65-6: Hu, 1988: 46-7). Therefore, as in the treatment of SO 17,1 look to the Mao Shi for assistance. 119  The question which must be satisfactorily answered is: are there particles in Mao no. 29 which may be removed without effecting its meaning? The answer to this query, it turns out, is rather conspicuous. Mao no. 29 is entirely tetrasyllabic, except for the third line of each stanza which is five characters in length (Hu, 1988: 27, 47). In each case, this third line is identical and reads: TbPRZ.A'Q " • O such a man is he!" (ICS: 10:13; Dobson: 605; also see Karlgren, 1944: 180). Hu Pingsheng writes, "If the three*/  characters [from Mao no. 29]  are removed, we get precisely ninety-six characters and the whole ode becomes tetrasyllabic" , f&#7L-f-7W , . a ^ f f - t M H ^  (1988: 47). Because xi ^ is just a metrical  particle, translated by Dobson simply as "O!", it is easily the most expendable character in stanza three of Mao no. 29 (Dobson: 605; also see He: 606). The insight into the Shi offered by FYSJ slips S018, S017 and S026 leads to an important question: is it possible to distinguish which variation represents the proximate original? On one hand, there is the FYSJ text (or texts, depending on the relationship of the three FYSJ slips mentioned above) which consistently omits unnecessary particle words; on the other, the Mao Shi appears to permit a higher degree of inconsistency in terms of meter. I will not discuss this issue in connection with FYSJ S017 and S026, focusing instead on FYSJ slip SO 18 which is featured in this particular study. One way to view the problem is to assume that at some point in time principles of uniformity and consistency were asserted upon the Shijing text. Since the phrase yujie  fjij 'Alas!' does not necessarily require any  sort of marker, it is possible that sometime before the Former Han, hu ^ was edited out of the original Shijing. According to this hypothesis, Mao and Lu Shi no. 25 are more likely than FYSJ S018 to represent the original, unedited version of the Shijing text. It is not impossible that this revision occurred as part of an effort to standardize the text of the Shijing; that is, reduce by one character the pentasyllabic verse-lines in certain odes, so that they conform to this classic's predominantly quadrisyllable meter. We have seen how grammatical markers do not always contribute to the meaning of the text; therefore, it is reasonable to assume they were the target of this standardization. But to make this argument more persuasive, the issue of motivation must be addressed. Why would an early school oi Shijing scholarship deem it necessary, or preferable, to render the meter of this classic more uniform? Does this proposed standardization reflect the influence of a certain 120  ideology or was it the result of development in the Chinese language? It may be possible that at some stage in preclassical or classical Chinese, the marker hu ^ simply became optional or obsolete. At the same time, usage of the marker hu ^ in different versions of the Shijing may involve the influence of different pre-Han dialects. These questions and conjectures involve subjects such as ancient dialectology, the evolution of pre-Han Chinese and a comparison of oral textual form versus written textual form, all which lie well outside the scope of this thesis. Consequently, I will not delve further into this line of speculation, leaving it for a future research project. Another way to determine which text best represents the original Shijing lies in a principle of consensus, or what has been referred to as the "lineage method" (Blanford: 2002) — i.e., which lexical variant, if any, is found in the majority of primary texts. In the case presently under inspection, there are two known lineages that utilize the marker hu ^ ; both the Mao and the Lu 5/2/contain the verse-line: T ^ ^ H ^ ^ C S : 10:11; Wang Xianqian: 120). By contrast, the FYSJ contains a version of this line which lacks hu  This discrepancy  constitutes strong evidence that the Shijing referred to in FYSJ SOI 8 is a later form of this classic, and not the proximate original. As such, the results of the lineage method accord with those derived from the application of a uniformity principle. LexicalVariation Study, no. 2 F Y S J , slip S115; L U : hu ffi Mao, no. 115:  he j q f  This study of lexical variation centers around synonymity. Because lexical variants of this sort do not involve a significant change in meaning, in this study I will conduct only one study of synonymous characters. After all, a new reading of the Shijing based on an analysis of lexical variation in the various manuscript and transmitted texts is the primary concern of this chapter. However, since this chapter deals with the theme of lexical variation, 1 would be remiss to bypass synonymous variation completely. Therefore, in the following pages I will present one more brief study of this. For a discussion of the synonyms not treated in this 121  thesis, the reader is referred to Hu Pingsheng's "Fuyang Han jian Shi jing jian lun" ^-WMWi WfMffim ("A Brief Discussion of the Han Manuscript Version of The Book of Odes Found at Fuyang") (1988: 26). Moreover, Hu's "Fuyang Han jian Shijing yi wen chu tan"  ^^^SS  ff#M^$L$3# ("The First Exploration of Variant Characters in the Han Manuscript Version of The Book of Odes Found at Fuyang") may also be consulted (1988: 36 88). T  The present case of synonymous lexical variation concerns the particles hu $J and he •fnj. The former form is contained in FYSJ slip Sl 15, which reads: ^^MikM^  0 MM (Hu,  1988: 14; Wen: 84). It is also found in the Lu Shi (Wen: 84; Wang Xianqian. 418). The first and second lines of stanza three in Mao no. 115 are identical to FYSJ Sl 15 in every regard, except that he j&J is replaced with hu £§: r p ^ M jfe , joj^ g f£|_- "... you have your wine and food, why do you not daily play your lutes..." (ICS: 10:51; Karlgren, 1944: 207). Wang Li believes hu ^ff and he jsj to be cognates (435); however, I believe phonetic evidence suggests a different conclusion. While hu j$ is pronounced *g o (with a [+round] main vowel) in b  Karlgren's reconstruction of Old Chinese and belongs to the rime-group ge  he JPJ is *g a: h  (with a [+unround] main vowel) and falls into the>^ M, rime-group (GSR: nos. 1,49; Chen: 172, 188). However, there is no doubt that they possess essentially the same meaning. For example, the GSR states that he jBJ originally meant 'carry', but that it could also be a loan for 'which, what, how, why' (no. 1). At the same time, the GSR assigns the meaning 'how, why, what' to hu  (no. 49). Dobson defines hu  as an interrogative substitute meaning  'how', 'what' or 'why' (341-2), and further states that in archaic Chinese this particle is often used in collocation with other characters, such as the negative bu^fi — e.g., hu bu "why not... ?" (342). He similarly views he jBJ as an interrogative substitute in preclassical and classical texts, one which can also be combined with bu  to mean "why not..." (316).  Referring exclusively to classical texts, Pulleyblank defines hu t$ and he jBJ in a way that corroborate Dobson's views. Pulleyblank writes that in ancient Chinese he jBJ means "what; why, etc" and "is the ordinary interrogative pronoun for things"; for hu $\ he comments that it is "mostly confined to abverbial usage meaning 'why, how'" (1995: 93-,5). 122  Elsewhere, Qing scholar Chen Qiaocong ltflf$£ comments that hu  and he ffi can be  used interchangeably (tongyongzi JffiM^X something which suggests synonymity (Wang Xianqian: 418). Chen cites the line, SfgWS " can there be settlement..." from Mao no. 29, then refers to the corresponding Mao zhuan which comments that hu  is equivalent  to he ffi in terms of meaning (^gMft) (ICS: 10:13; Karlgren, 1944: 180; Wang Xianqian: 418). The Mao zhuan offers the same gloss for hu $ J elsewhere; namely, when commenting on Mao no. 90: ^ g ^ H " should I not be glad!" (ICS: 10:40; Karlgren, 1944: 200; Wang Xianqian: 363; Wang Li: 435). Zheng Xuan's interpretation of hu  concurs with the Mao  zhuan. In regard to a verse-line from Mao no. 264,ffiWMrM"Why do [they] do these evil  things?", Zheng also uses he -fBj as a gloss for hu £g (ICS: 10:140; Dobson: 342; Wang Xianqian: 992). But the use of hu  as an interrogative substitute is not confined to the Shi  jing alone; this particle performs the same function in other preclassical texts. For example, in the Shang shu there is a passage which reads: %MMWk ,  $PMW>$U  1  "If [one] does not think  [about it], how [will they] attain [it], i f [one] does not act [on it], how [will they] complete [it]" (ICS: 9:17; Wang Li: 435). Han glossarist Kong Anguo ?L^ffl comments that the particle hu £5 in this passage is semantically equivalent to he -for (Wang Li: 435). In each of the above examples, hu $J is consistently used in the primary preclassical source, while he jBJ is always part of the corresponding gloss from the Han dynasty. As might be expected, this pattern of usage is not merely coincidental. Pulleyblank points out that the interrogative substitute hu ~$ was common in the preclassical period, but not in the subsequent classical period (1995: 95). Actually, this assertion is substantiated by the frequency of the phrase hubu  'why not' in the Mao Shi compared to the semantically  equivalent hebu jnpfx Whereas the former occurs on a total of ten occasions in the Mao Shi, the latter form takes place only once (ICS: 10:227, 231). This information is relevant to the lexical variation found in FYSJ slip SI 15; it strongly suggests that the hufiftparticle in SI 15 represents the original Shijing, as recorded during the preclassical period. Presumably, sometime in the early Han, the Mao school of Shijing exegesis replaced the original hu 123  particle with the synonym he {&]", which was more common at the time and, therefore, more familiar. Of course, the appearance of hu  in the Lu Shi also implies that this particle  represents the proximate original. When the majority of textual variants utilize a particular form, the lineage method requires that this form be taken as the proximate original (Boltz, 1997: 266-7; Blanford: 201-2). In this case, both FYSJ Sl 15 and the Lu Shi use hu ffl; therefore, we may confidently assert that the particle he jBJ found in Mao no. 115 should be emended. LexicalVariation Study, no.3 FYSJ, slip S045: ju Jj§ Mao, no. 41: che  J|L  FYSJ S045 corresponds to a verse-line in stanza three of Mao no. 41. Whereas, FYSJ S045 reads, M  M f ^ M ^ n M  (Hu, 1988: 6; Wen: 70), Mao Shi no. 41 reads: ^Wff^ , t l  ^ f H l ^ "... if you are affectionate and love me, I will hold your hand and go with you in your carriage..." (ICS: 10:19; Karlgren, 1944: 184). There are several instances of variation in these two sources. One case of variation involves the particles ran  and er nf, which are likely  either cognates or different graphic representations of an identical word (Wang Li: 156; Wang Xianqian: 202). Hu Pingsheng also views these two characters as being completely interchangeable, and we may be certain that lexical variation is not an issue here (1988: 55). In the present study, I will analyze the other discrepancy which exists between these parallel texts; that is, the lexical variation involving the FYSJ characterju j_ and its counterpart ju $ from the transmitted text of the Mao Shi. My analysis of this lexical variation will result in an interesting alternative reading of Shijing no. 41. In fact, My conclusions will show that the characterju ]j£ in Mao no. 41 likely represents a misreading of the text which should be emended. The Shijing rime-groups and Karlgren's reconstruction of Old Chinese both indicate that ju j_- and ju ^ were phonetically compatible in the preclassical period. Both characters 124  belong to the yu ft rime-group and the GSR assigns them completely homophonous Old Chinese pronunciations — i.e., *kjo or * k j w 3 (Chen: 169, 170; GSR: nos. 49, 74). Obviously, complete homophony is not standard between two lexical variants; in fact, when dealing with entirely different words the opposite is generally expected. However, there are conditions which sometimes require homophony among lexical variants, such as when there is a particular rhyme sequence which must be followed. Since both ju jg and ju ^ fit the rhyme sequence governing the first three lines of this ode, they pass the first test required for this particular case of lexical variation — i.e., they are are phonetically suitable for the position they occupy. (The other two characters, hu M 'fox' *g wo and wu 'crow, raven' *o, also fall h  into the yu  rime-group — GSR: nos. 41,61; Chen: 170,173) The next consideration for ju  Jg andju J|t is related to their semantic value. Whether or not they suit the meaning of the verse-line in which they are situated must be determined. This is the question I will turn to next. The graph ju $ has represented the Chinese wordfor 'carriage, chariot' since very early on (GSR: no. 74; HYDCD: 9:1186). The definition given for ju M in the SWJZ is as follows: "The general name for [any kind of framed] body with wheels" &1fm&M$i-& (SWJZZ: 14:37a). But this character is found in inscriptions produced long before the Han dynasty. Actually, the ancestral OBI form ofju Zhongshu writes, "[The OBI form ofju  is a pictograph of a carriage or chariot. Xu  depicts the shapes of a chariot's two wheels and  body, as well as its shaft, yoke, railing and so on. Wheels are the most important symbol of a chariot; consequently, sometimes [the graph] is abbreviated in the shape of two wheels"  ZMtiiBM , m , Is ,ffimm, mMMZm^Wmm, ^ ^ - ^ ^ ( K M J S : no. 1629; also see JGWZGL: no. 3145). The two OBI forms ofju  mentioned by Xu Zhongshu — i.e.,  the complete and abbreviated forms — are given below (JGWB: no. 1629):  There can be no doubt concerning the general meaning of the characterju ^ . Its meaning and 125  orthographic development are relatively straightforward compared to a lot of other characters. However, its usage in Mao Shi no. 41 is a slightly different matter. The line from Mao no. 41 containingyzv $L parallels a line in stanza one which reads, Hrf-fBHf "• • • I will hold your hand and go with you...," and a line in stanza two that is  similar: $tr£|l] jlf "... I will hold your hand and go home with you..." (Karlgren, 1944: 1845). According to Karlgren's translation, the final character in these verse-lines conveys motion, and a sense of togetherness between the lady narrating the ode and her lover. Both of these elements seems appropriate for the theme of this ode, which Karlgren claimed to be about a lady conveying "her wish to go with a gentleman to his home as his wife" (1944: 184). But there is at least one scholar who views this reading ofju $L as problematic. Yao Jiheng  believes that tongju [WJi|f "go with you in your carriage" overlaps with the  preceding long gui [Wljff "go home with you" in terms of meaning (wenyi chongfu J t - f i J i l S ) (66; see Hu, 1988: 55). The point Yao raises is valid, since in traditional China gui | § refers to a bride being brought to her husband's home in some form of appropriate conveyance (HYDCD: 5:367; GSR: no. 570). In the Mao Shi there are numerous instances of gui jtf being used in this sense (see ICS: 10:222-3). One such example can be found in the lines from Mao no. 6 which read: ^ i i ^ J i i , 'K^'MM "• • • this young lady goes to her new home, she will order well her chamber and house..." (ICS: 10:3; Karlgren, 1944: 174). However, the shortcoming proposed by Yao Jiheng does not necessarily disallow the line fjg^-|l)^.. Actually, if not for the discovery of the FYSJ and slip S045, there is nothing in Yao's criticism concerning Mao no. 41 which suggests a definite need for emendation. From the perspective of grammar the use oiju ^ is also problematic. Because this character is a noun, it does not match the verbs xing fj and gui $f found in corresponding positions in the previous two stanzas of Shi jing no. 41; that is, while the verb gui f§ is parallel in grammatical function with the verb xing f j , the noun ju  violates the pattern set  by these two words. This is strong evidence that ju ^ may not originally have been the intended character for stanza three of this particular ode. But without any concrete evidence showing which character was actually intended for the position occupied by ju 126  there  would be no point in challenging the use ofju  Once again, however, new insight into this  matter is offered by the FYSJ. The primary purpose of the present Lexical Variation Study is, of course, to determine if the line fjjg^fi]:^ can be read differently or actually improved by replacing^ M with ju J_ from FYSJ S045. At this juncture, I will attempt to explore this matter. The characterju JU in FYSJ slip S045 most likely means "dwell, reside in" (GSR: no. 49). A quick perusal of some of the other words represented by this graph in pre-Han Chinese — e.g., 'sit down', 'occupy', tranquil' — is enough to dismiss them from serious consideration. Inserting ju J_ into the verse of Shijing no. 41, stanza three, actually improves the meaning in several respects. For example, usingyw JU in this context removes the semantic overlapping caused when tongju [WJ^: is paralleled with tong gui |Ijj?f. Instead of conveying the act of traveling in a carriage, FYSJ slip S045 in translation reads: "I will hold your hand and live with you." At the same time, the characterju jg, being a verb itself, does not violate the grammatical pattern established by the verbs xingft and gui J£f in the first two stanzas of the ode. Finally, I believe that usingju j g to fill this lexical position in Shijing no. 41 also contributes to a sense of narrative climax — i.e., in the first stanza, representing the early stages of their relationship, the bride and her lover may only engage in walking together; the second stanza shows the married couple following tradition and going to the husband's home in a carriage; in the closing stanza, after all the preliminaries, the couple is finally living together. A strong sense offinalityor conclusion is manifest in the ode when two stanzas emphasizing action — xing f j and gui f§ — are followed by a third stanza signaling a halt in motion. There is also a sense of thematic completeness afforded by this interpretation, as it encompasses the full courting procedure. In sum, I feel that FYSJ S045 is probably the correct version of Shijing no. 41; whereas, the corresponding Mao no. 41 is likely the product of phonetic borrowing. This conclusion corresponds to the explanation William Baxter offers for why such an error might have occurred in Mao Shi no. 41: "[the scribe] simply chose a character to match the sound of a recited or memorized text." Baxter elucidates this point by referring to the lexical variation found in FYSJ S045 and Mao no. 41, concluding: "Whichever is the better reading, a scribe at 127  some point substituted one character pronounced *k(r)ja [*kp I *kjwo] for another, because they were homonyms" (Baxter: 360). Although the contextual evidence suggesting that ju jf? is the proximate original is quite persuasive, it does not entirely discount the credibility of the Mao version. Unfortunately, neither the lineage method or the principle of lectio difficilior potior can be applied to this case of lexical variation. LexicalVariation Study, no. 4 FYSJ, slip S048: tong | i j M A O no. 42; L U , QI: tong %  In its entirety, FYSJ slip S048 reads: ® M = M =  (Hu, 1988: 6; Wen: 70).  The symbol "=" is a notation device used by the scribe of S048 to represent duplication. These short horizontal markings are an abbreviated way of indicating that the graphs tong | f and guan ^ are repeated in the context given. Before moving on to an analysis of FYSJ S048, I will examine the lines in Mao no. 48 which correspond to it, and contain the character tong &. It reads:  ,  fi^f^  , s f t f e l K I C S : 10:22) and has been translated by Karlgren  as, "The good girl is pretty, she gave me a red pipe [i.e., some tube-like plant]; the red pipe is bright; I delight in the beauty of the girl" (Karlgren, 1944: 185). Both the Lu and Qi Shi also use the character tong % in this particular ode (Wang Xianqian: 207, Wen: 70). In my study of the received version of Shijing no. 42,1 will begin with an examination of the character guan If. As Lexical Variation Study no. 4 unfolds, the important role this character has in the analysis will be made clear. Scholars such as Lu Deming  (556-627  A.D.) in the Tang dynasty, interpreted the character guan If in this ode to mean 'writing brush' (Lu Deming: 1.230; HYDCD: 8:1198). But this interpretation originates in the Han dynasty, when the Mao zhuan, the erudite Zheng Xuan and the commentaries of the Qi and Lu schools all related this character to the inner court office called 'Female Scribe of the Red Brush'  (Wang Xianqian: 207-8). The female scribes belonging to this office were  primarily responsible for recording events which occurred in the quarters of the empress and 128  imperial concubines, documenting both the accomplishments and faults of these inner court residents. The Mao zhuan reads,"... if the [female] scribe does not record [their] faults, she is committing a crime and will be executed" j£^iBM ,  (Wang Xianqian: 207). In  connection with this, the Qi commentary explains the significance of the character tong ff  y  'red' as follows: "As for the color red [associated with guan *gf], it simply refers to vermilion lacquer. The [female] scribe officials recorded events [in the inner palace]; consequently, they used a red brush to [imply] recording events with a 'vermilion heart'" , ^^XW^fc'btSM  , ^ f O f ° ^MM,  (Wang Xianqian: 207). The term chixin jfc>(j 'vermilion heart' refers  to a sincere and focused attitude (DKJ: 10:815). Zheng Xuan viewed tong ffr in the same way, writing, "Regarding the 'red' tube, it is the vermilion shaft of a brush" ffifg , H^1?"& (Wang Xianqian: 207). For the above definition of the character guan If, taking tong f& to mean 'red' is obviously quite suitable; however, it also fits another interpretation of guan U outlined below. The majority of contemporary scholars and translators have eschewed the view that guan U refers to a 'writing brush' in Mao no. 42. This view is clearly colored with a socioethical agenda, the result of approaching the Shijing as a form of exemplary reading in the Han dynasty (Saussy: 106-50). But a consensus has not been reached regarding how this segment of Mao no. 42 should be read. Karlgren's view has been given in his translation given above. In the explanatory notes attached to this translation, Karlgren adds: "Kuan [guan *_f] 'pipe' also means 'tube' generally, and there has been much speculation as to the nature and use of the tube here. But the parallel with stanza three shows that the girl had merely in her promenades picked some simple growing things, a pipe (reed or slender bamboo, for a flute), a young shoot (edible, a delicacy?), and charmingly given them to her boy as tokens of love" (Karlgren, 1944: 185). In order to properly present Karlgren's view, the relevant portion of the third stanza is hereby cited: "From the pasture-grounds she presented me with a young shoot; it is truly beautiful and remarkable" i ^ J f H , $Q0zMM (Karlgren, 1944: 185; ICS: 10:20). Proof that guan i f can refer to a 'pipe-shaped' or 'tube-shaped' object in pre-Han Chinese — e.g., Karlgren's "reed" or "slender bamboo" — is attainable from a number of sources. For example, a passage from the Zhuangzi reads, "This is simply like peering at 129  heaven through a tube or pointing at the earth with an awl — they're too small for the purpose" M l f l ^  ,  M$mi&& , W ' J W ( M a i r :  163; HISS: 20:45; HYDCD: 8:1198).  Further, the Lun heng (The Balanced Inquiries) (ca. first-century A.D.) reads, "Those pipes of the jcwo-pipes and sheng-pipes resemble the throat [in a] person's mouth" A f f l l ^ l ? , mAZ-Hm-fc (HYDCD: 8:1198). Several scholars agree with Karlgren that guan _5 refers to a 'tube-like' or 'pipeshaped' plant in Shijing no. 42. For example, Jeffrey Riegel offers the following translation: "The Chaste Maiden, so clever,/has given us the Vermilion Stalk" (Riegel: 154-5; for another example see Yu Guanying: 33). Moreover, there is thematic evidence that also supports this reading. For the motif involving the use of plants or flowers as gifts of courtship is quite frequent in the Shijing (Mao nos. 6, 14, 67, 188; see ICS: 10:3, 7, 31, 86). However, one problem with this view is its broad interpretation of guan  As alluded to by Karlgren, in  pre-Han Chinese this character refers to tubular objects in general, and not specifically botanical items. Similarly, if we take tong f& to mean 'red,' it can be used to modify a wide variety of different objects. But there is no disputing that Karlgren's reading of the lines fq® is still permissible.  WM , WMMf$  Arthur Waley also interpreted guan H? as a tubular object; however, he believed it to mean 'flute': "Give me a red flute./The flush of that red flute/Is pleasure at the girl's beauty" (33). Certainly, tong ffr 'red' can be used with 'flute'. Moreover, there is a considerable amount of evidence substantiating this interpretation of guan  From the Han dynasty, the  SWJZ defines guan _r as follows: "It is like a [seven-holed bamboo] c/tt-flute [but] with six holes" ttttBfcll (SWJZZ: 5:18a). In the Erya ff ft (On Approaching Elegance), written during the Qin and early Han, guan _? is described as a type of woodwind instrument that is divided into various different sizes (ICS: 16:72; DKJ: 8:804). Furthermore, in a chapter from the Xunzi Irj-f entitled, " A Discussion of Music" yue lun J^M, the character guan _F has a 1  similar meaning: "... as the decorations in the peace dance he uses feather ornaments and yak tails; and he sets the rhythm with sounding stones and woodwinds" WPAMW '•  $tIM@W  (Gao Heng: 60; Watson: 116). In the Zhouli jtj _ (Institutes of Zhou) there are numerous 130  references to guan If as a woodwind used in various forms of ceremony (ICS: 4:42-3; DKJ: 8:804). Finally, guan *_f is used in a similar manner in the Yili  (The Book of Etiquette  and Ceremony) (ICS: 7:39; HYDCD: 8:1198; also see Wang Xianqian: 1098). More importantly, however, guan If clearly refers to a musical instrument in the Shijing itself: in Mao no. 280 there is a line which reads, "Pan-pipes and flutes are ready and begin" H l f fi!j§ (ICS: 10:146; Waley: 218; also see Karlgren, 1945: 90). Zheng Xuan's corresponding commentary to this ode reads, "A [guan *_f] woodwind resembles the fife, [you] close [your lips] and blow into it" ^£nt£#fW2fc2. (Wang Xianqian: 1027). Further, in Mao no. 301 we find the verse-line, " A shrill music of flutes"  (ICS: 10:155; Waley: 225; also see  Karlgren, 1945:97). The FYSJ manuscript's use of the graph guan  in place of guan  substantiates  Waley's translation further. Wang Li maintains that guan % and guan If are variant characters for the same word— i.e., "woodwind" (551). Certainly, the Old Chinese pronunciation of these two characters exhibits the necessary phonetic interchangeability to be allographs: Karlgren assigns the Old Chinese pronunciation *kwa:n to both guan ^ and guan If (GSR: nos. 257, 157), and both characters belong to the yuan % rime-group (Chen: 297). Earlier evidence that suggests phonetic compatibility can be found in Lu Deming's comments for Mao no. 301  (395; also see DKJ: 8:788). But for our purposes, the strongest  evidence for the relationship between guan % and guan | f comes from the actual text of the Shijing. Mao no. 274 has a verse-line which reads: "... musical stones and flutes resound..." ©^jrtrJKF (ICS: 10:144; Karlgren, 1945: 89). Further proof that the graphs guan ft and guan If refer to the same word can be found in various other texts from the classical period (DKJ: 8:788; HYDCD: 8:1172; Wang Li: 551; SWJZZ: 5:6-7a). Like the first interpretation of tong guan fft^ given above — i.e., 'red [tube-like] plant,' Waley's interpretation that a woodwind is referred to in Shijing no. 42 is entirely acceptable. In fact, since tong ffr 'red' is no more closely affiliated in semantic terms to 'flute' than it is to '[tube-like] plant', one would be hard pressed to choose one over the other. Context may provide some clue as to which of these readings is more suitable, and a study of this is forthcoming. 131  There is other strong evidence in FYSJ S048 indicating that guan _f refers to a woodwind instrument. To uncover what this evidence is, however, the relationship between the FYSJ character tong | f and its counterpart tong f& must first be clarified. Further, early Chinese texts must be examined in order to illuminate the meaning of tong If. According to Karlgren, the Old Chinese phonetic properties tongflfand tong f& might suggest a relationship based simply on phonetic borrowing — something which would make them graphic variants. Karlgren assigns the Old Chinese pronunciation *d u:g to tongflfand *d o:g h  h  to tong f&, and his work on phonetic compatibility shows that these two characters are interchangeable (GSR. nos. 1176,1008; 1963: 1-18). Moreover, although the phonetic principles adopted for this study permit only finals with identical main vowels to be interchangeable, the similarity of the main vowels in Karlgren's phonetic reconstruction of tong M and tong f& suggests that applying this criteria too rigorously here might be a mistake. After all, the vowels [u:] and [o:] found in the Old Chinese pronunciation for these graphs are both [+round] and [+back]. In addition, whereas [u:] is [+high], the vowel [o:] is a high mid-range sound. However, there is good reason for not simply regarding tong | f and tong ffc as graphic variants. Typically speaking, characters that can be switched in a process of phonetic borrowing are entirely disparate semantically. For example, as discussed in chapter one, a character such as gong  *ku:g "prince, public" might be used to replace gong  $j *ku:g "merit" (Karlgren, 1963: 3). This semantic disparity normally results with the meaning of the character borrowed not suiting the context it is placed in. I contend that such is not really the case here, since tongflfand tong ffr both bear a viable semantic association with the object — i.e., guan ^Iguan % 'flute'. The above paragraph provides the evidence needed to show that tongflgfis not simply a loan character for tong  This hints that the tongflfcharacter's relationship with guan  warrants further examination. Whereas 'red' is not object-specific and can refer to practically anything to which a property of color can be ascribed, as a modifier tong | f bears a much more specific semantic association to guan ^Iguan % 'flute'. The GSR no. 1176 states that tong | f refers to "Eloeococca and kindred trees." However, such brief mention in Karlgren's 132  work contradicts the large amount of commentary found throughout the Chinese philological tradition that focuses on identifying tong |SJ (Wang Xianqian: 238-9; Ma Ruichen: 181-2). To cite one example, the SWJZ associates tong | f with the quality of being rong ^ 'flowered' (SWJZZ: 6: 18-9a; GSR: no. 843). The complete philological perspective on the character tong | f need not be related here. Ma Ruichen's Mao shi zhuan jian tong shi %^MMSM (A Complete Interpretation of the Mao Shi, Commentary and Notes) offers a sufficiently comprehensive summary of this research, stating that tong #§} was one of several types of trees used in early China to make various string and woodwind instruments (182). The locus classicus for this understanding of tong  is found in Mao no. 50. This ode directly relates  the tree represented by the tong |5J graph to the production of different musical instruments. Karlgren's translation of the verse-line in Mao no. 50 which contains the character tong flf reads: "[the prince planted the palace] with yi tree, t 'ung tree [tong H], Catalpa, lacquer tree, so that they could (cut:) make guitars and lutes" f t M # ^ ,  (ICS: 10:23; Karlgren,  1944: 187). The type of tree represented by tong $f was such a common material for making instruments, that in later dynasties it actually became synonymous with qin 3^ "guitar" (HYDCD: 4: 973-5). Based on the understanding of tong | f arrived at above, 1 believe FYSJ S048 may be translated as: "she gave me a flute made oftong-treewood; the flute of long-tree wood is bright." This is a very legitimate reading of the verse-line, and one which challenges the readings which are based on modifying guan *_f with the character tong ffr. Certainly, as a modifier of the character guan If 'flute', tong |S| 'to«g-tree' is more semantically specific than the alternative, something which suggests it is the most suitable character for the context. However, from another perspective, a consideration of this ode's overall context also indicates that interpreting guan i f to mean 'flute' may be incorrect. Since stanza two makes reference to yi M "young shoot", a level of imagistic consistency is achieved in Shijing no. 42 if guan If is taken to mean "[tube-like] plant." On the basis of this argument, Karlgren's interpretation of this ode is preferable over the other two. On the other hand, the courtship theme of this ode makes it possible to interpret both the yi ft "young shoot" and guan If 133  'flute' as examples of phallic imagery. Once more, a consistency of imagery is implicit in the relationship of these two objects. While contextual evidence is sometimes quite useful for finding the proximate original among a pair of variant texts — i.e., the preceding Lexical Variation Study no. 3 — more success is attained in this case if the lineage method is applied. Since a total of three lines of transmission utilize the character tong f&, it seems almost certain that the Shijing text represented by FYSJ S048 is not the original version. Rather, it is a variant version, but one which also offers a perfectly legitimate reading of the ode. LexicalVariation Study, no. 5 FYSJ, slip S123: xiang f$L  M A O , no. 128; Han; Qi: chang $J§ (> chang J | ) FYSJ slip S123 reads: XM^M  (Hu, 1988: 15; Wen: 85). Towards the bottom of this  bamboo slip there is some damage, and parts of the third and fourth graph are incomplete. However, both context and the parts of these graphs that are visually discernible enable us to identify them as xiangfeand gu igt A careful perusal of the copying {moben  found in  Hu Pingsheng and Han Ziqiang's Fuyang Han jian Shi jingyanjiu ^WsWfav&$$ffi% (Research on the Bamboo Slips from the Han Dynasty Found at Fuyang) supports this interpretation (see "Fuyang Han jian Shijing moben," no. 4). Mao no. 128 and Karlgren's translation of it read: jZM&ktk "• • • there are striped floor-mats and protruding wheelnaves..." (ICS: 10:128: Karlgren, 1944: 211). The version of this line found in the Han andQi 572/ reads the same as Mao no. 128 (Wang Xianqian: 443: Wen: 85). Comparing FYSJ S123 with the corresponding line in the Mao/Han/Qi Shi, there is only one case of variation, that involving the FYSJ character xiang fe and chang %. Lexical Variation Study, no. 5 will engage in an analysis of these two characters. In terms of Old Chinese pronunciation, the characters xiang fe and chang % are phonetically compatible. For example, evidence from the SWJZ shows that xiangfeand yang # are interchangeable asphonophorics (SWJZZ: 5: l i b ; 11: 2a; Hu, 1988: 78). When dealing with Old Chinese pronunciation, the SWJZ should generally be considered an 134  anachronistic source and used with caution. The pronunciations given in the SWJZ reflect the way characters were read in the Han dynasty, and do not always reflect Old Chinese phonology. But in this case, Karlgren's Old Chinese reconstruction and the Shijinghmegroups support the conclusions drawn from the SWJZ. Karlgren assigns the Old Chinese pronunciation *dzjarj to xiang fe and *t^jarj to chang g§ (GSR: nos. 720, 728), and both graphs belong to the yang p§ rime-group (Chen: 275, 277). This evidence indicates that in terms of initials these two characters are homorganic — e.g., both the fricative [*dz-] and the aspirated [*t-] are plosive alveolar sounds. Moreover, the main vowels and final consonants of these two graphs are phonetically identical. Exactly what this phonetic compatibility signifies will be known shortly. Before discussing the FYSJ lexical variant xiang fe, the meaning of its counterpart chang | § must be identified. In connection with the matter, there is a question of graphic variation which must also be resolved. In the modern day, it is generally believed that chang 4§ represents the word 'long' in preclassical Chinese (GSR: no. 720; HYDCD: 10:817). Interestingly, in contradiction to this view, chang $|§ is rarely associated with this meaning in texts from this period. In fact, it appears that Mao Shi no. 128 — featured in the present study — is the only extant preclassical text where chang $j§ is actually used in this sense (HYDCD: 5: 816; DKJ: 5:922). I believe this association between the word 'long' and the graph chang $g becomes more common in texts written during the classical era, although some effort is required to prove this. In the Confucian text attributed to Mengzi ^f-  (372-289  B.C.?), there is a line which reads: |l/fcf§3<j (ICS: 15:29). Precisely the same line is also found in later texts, such as the Shiji ^.gg (The Record of History), attributed to Si Maqian i H H S (145-90? B.C.), and the Lun heng  by Wang Chong  (27-97? A.D.) (Wang Li:  537-8; HYDCD: 10:816). As this line indicates, chang fj§ is usually understood to mean 'abundant, luxuriant' (sheng ___) when used in connection with trees or plants (HYDCD: 10:816; GHYZD: 27). However, I contend that this definition is inaccurate, and does not take into account all the available evidence. This definition seems to be imbued with a particular set of qualities associated with botanical items that have grown either long or tall; that is, qualities such as 'abundance', 'luxuriance', etc. As such, 1 think it constitutes a case in which 135  a character is defined solely on the basis of context, which can be misleading. 1 would instead interpret chang % in the line ^ 7 f c t t £ S to mean 'long', translating the entire line as, "The grass and trees are long and flourishing." My reasons for this interpretation of chang $g are as follows. There is proof in the SWJZ that chang  represents the word 'long' both in classical  Chinese and, more specifically, in the line ^ 7 f c i | § £ § ; . In this work, Xu Shen uses chang  the  attested character for 'long', in assigning semantic value to the graphing J | (SWJZZ: 9: 34b; GSR: no. 721). Ma Ruichen suggests that the interchangeability ofyang f| and chang i§ means 'long' can also be assigned to the graph chang %. This interchangeability is based on the understanding that yang J§, pronounced *djarj in Old Chinese and belonging to the yang pg rime-group, is the phonophoric in chang % (Ma: 374: GSR: no. 720; Chen: 275). Since the SWJZ is roughly contemporaneous to the Lun heng, Xu Shen's definition ofyang j | (> chang |§) may be used to interpret chang % in the line: IM^SIrS;- By doing so, the  translation I proposed earlier is possible — i.e., "The grass and trees are long and flourishing." At the same time, the zhuan corresponding to Mao no. 128 also indicates that chang $§ referred to the word 'long' during the Han. This gloss explicitly equates the semantic value of the word represented by the character chang % with chang jg; ($§iR , MM -&) (Wang Xianqiang: 443). Subsequently, I believe the Mao zhuan provides valuable insight into the meaning of chang % as used in ^ / ( ^ ^ ^ from the Shiji. Since both the Mao zhuan and the Shiji are relatively contemporaneous to one another, there is no chance of this association being prejudiced by anachronistic analysis. Since both the SWJZ and the Mao zhuan associate chang ij# with chang J | , the relationship between these two characters warrants more attention. Evidence strongly suggests that chang % is not the attested graph for the word 'long'; instead, it is likely a phonetic loan character or variant character sometimes used in place of chang JH. On the other hand, chang Jit. matches all the criteria necessary to make it the attested character for 'long': it appears to be the ancestral graph for the word 'long' and it is widely used in this 136  sense in early Chinese texts and inscriptions. At this juncture, I will refer to graphological evidence concerning chang ft from early oracle-bone and bronze inscriptions. There sources show undeniably that chang ft is the ancestral graph for 'long', as well as other semantically affiliated words. For example, Yu Yongliang ^ a c ^ thinks the OBI form of chang ft is a pictograph depicting the appearance of a human figure with long hair. Yu believes that the semantic value of chang ft was eventually extended to encompass the meaning 'a long time' ( ft*feASft^  ° ^IWMMX^m) (JMJSCD: 190). Xu Zhongshu agrees with this view,  writing that while chang ft originally depicted long hair in Shang OBI, it later came to also mean 'long' — i.e., the antonym of short (glA^ftH , M f e ^ f t S ^ f t ) (KMJS: no. 1133). Below, I given two examples of the graph chang ft: the graph on the left is from a Shang OBI source (Gao Ming:  51), while the  one on the right is taken from BRI (JSZD:  1371):  In Warring States BRI, the character chang ft refers to several words that are related to 'long'. For example, in various different BRI sources the character chang ft means 'long' ( ft),  'length' (changdu  ftg)  and 'a long time' (changjiu  ft^)  (JWCYZD:  883-4;  JMJSCD:  190). Similar meanings for chang ft are found in received texts from throughout the Zhou dynasty. In the Shijing itself this graph also represents the meaning 'long'. This is evident in a verse-line from Mao no. 129 which reads: MPiUlft "• • • the road is difficult and long..." (ICS: W  10:56; Karlgren, 1944: 211).  ,  14:24;  ft—A'W¥ Ames, 1998:  "They are certain to have a nightcoat half the body in length" (ICS: 136). Another somewhat  'tall' — in Mao no. 106: 3f 10:46-7; Karlgren, 1944: 204).  from the Zhuangzi  Elsewhere, chang ft means 'length' in the Lun yu mm$- #  ft— i.e.,  different sense is implied by chang  , JSWft^ "Lo! How splendid, how grand and tall" (ICS: The same meaning is associated with chang ft in a chapter  (ca. mid to late Warring States): MMAR^ 137  , ffi i W^fe , W&SStfJ  "Your person is eight feet two inches tall, your face is radiant, your lips are like shining cinnabar..." (HYDCD: 11:577; Mair: 301). Both the temporal and spatial meaning of 'long' are expressed in preclassical and classical sources (HYDCD: 11:577). For the former, 1 cite Mao Shi no. 299: JUMftM , Mifcf^ll "• • • he has followed that long road, and subdued all this multitude..." (ICS: 10:153-4; Karlgren, 1945: 95). In light of this evidence regarding orthographic makeup, overall usage and semantic value, it seems clear that chang ft is the ancestral graphic form for the word 'long', as well as its attested form in preclassical and classical texts. Therefore, to my mind, chang ft is the legitimate graph for this verse in Mao. 128, and chang % simply a variant character. At this juncture, we may proceed to examine the FYSJ lexical variant xiang fe. In GSR no. 728, Karlgren writes that xiang fe stands for a number of different words — e.g., "elephant," "ivory," "to imitate," "pictured," etc. Because the latter definition best suits the context of Shijing no. 128, it is of chief interest to the present study. In his own study of FYSJ SO 123, Hu Pingsheng assigns a semantic value to xiang fe which is comparable to Karlgren's; he associates xiang fe with the meaning 'designed' or 'patterned' (1988: 78). However, these definitions must be scrutinized before we accept them. If we are able to prove that xiang fe definitely means 'pictured' — or something similar— in preclassical Chinese, FYSJ slip S123 could shed light on our understanding of Shijing no. 128. Karlgren also interprets xiang fe as 'pictured' in his translation of a passage from Mao no. 47, which reads: ^^fgfg , $rj[l| P-U?K ,felSSSl"• • • she is gracefully compliant, (beautiful) like mountain and river, (suitable for =) worthy of the pictured robe..." (ICS: 10:21-2; Karlgren, 1944: 186; also see Wang Xianqian: 223). Moreover, the zhuan for Mao no. 47 seems to agree with Karlgren regarding the meaning of xiang fe. This zhuan reads  fej@  , ffiiilfeW, and may be translated, "As pictured garments are honorable, she took them as [her] attire"fejg##,ffrlZkfeW(Ma: 171; Wang Xianqian: 223). However, a clearer idea of what is referred to by key to Karlgren's view of the character xiang fe. Wang Xianqian writes that xiangfufejgrefers to garments usually worn by rulers in ancient times to symbolize their merit and praiseworthiness. In the context of Mao no. 47, the xiang 138  fitfeJIHworn by the female protagonist indicates that she possesses similar qualities. Further, xiangfufe$galso represents a type of ritual attire worn by kings and their queens which is  referred to in classical sources as huiyi ft3£ (Wang Xianqian: 224; also see Ma: 171; SWJZZ: 8:53a). Han glossarists refer to huiyi ft^c as being patterned or decorated, and in SWJZ entry for the character hui ft reads: ffl&ft=EM2MW fcMM& "The Zhou li states, 'The :  ritual dress [hui ft] is the attire of kings and queens. It is [also] called the patterned [= pictured, decorated] robes'" (Wang Xianqian: 224; Ma: 171; SWJZZ: 8:53a; for the original passage in the Zhou li, see ICS: 4:15). All of this evidence from primary sources and commentary buttresses Karlgren's definition of xiang fe. Hu Pingsheng observes a certain relatedness between xiang fe and the character xiang fc, as mentioned in the SWJZ (1988: 78; SWJZZ: 8:6la). The HYDCD substantiates this claim, assigning xiang H a definition which is similar to 'pictured' or 'patterned' ( M i t  fe  (9: 135). It is likely that xiang fe and xiang fc are graphic variations of one another, their only difference being that the latter form also contains semantic classifier no. 145 yi & 'clothing'. Wang L i agrees with this, claiming that xiang fe is the primary graphic form for xiang fc (fc^^fFfe) (367). Further, Wang Xianqian writes that in ancient Chinese these two characters were interchangeable (fefc"^^jfifi=) (224). As far as specific phonetic information for xiang fc is concerned, we are faced with unexpected limitations; Karlgren does not provide an Old Chinese pronunciation for this character in his GSR (nos. 550, 728). However, xiang fc is a yang |§ rime-group character, and to my mind this sufficiently indicates that it is phonetically compatible with xiang fe (Chen: 277). As such, the meaning and use of xiang fc can also be applied to the character xiang fe as found in FYSJ S123 (see Hu, 1988: 78). Hu Pingsheng utilizes evidence regarding the use of xiang fc in classical texts to bolster his view of xiang fe in the FYSJ S123. He cites a passage from the Jijiu plan ^ifc  H (The Speedy Mastery of Characters) (ca. 40 B.C.), which reads:  fcf?fii_f  "... pictured and ornamented, carved and painted beyond compare..." (Hu, 1988: 78; HYDCD: 9: 135; Wang Xianqian: 224). Hu Pingsheng interprets xiang fc in this text as 139  referring to 'patterned ornamentation' wenshi $tM (1988: 78). To this point, I have concluded that chang % used in Mao/Hu Shi no. 128 is a phonetic loan character for chang H which means 'long'. Moreover, based on the evidence just given, the character xiang fe in FYSJ SI 23 apparently means 'pictured' or 'patterned'. The next step is to determine which of the two lexical variants is most suitable for the ode in question. Is it possible to discern which of the variant constitutes the proximate original? As Hu Pingsheng comments, xiang fe quite adequately suits the context of stanza one in Shijing no. 128 (1988: 79). Revising Karlgren's translation, we may read the FYSJ version of this ode as follows: "... there are striped floor-mats and pictured [= patterned] wheel-naves...." This reading implies a semantic parallelism in the verse-line; the phrases wenyin 3tH "striped floor-mats" and xiang gufejgfc"pictured wheel-naves" are both comprised of adjectives which describe some sort of ornamented surface. In this sense, the FYSJ of version no. 128 is preferable to the Mao/Han Shi. Before subscribing to Hu Pingsheng's argument completely, I will look for additional proof that wen  bears a semantic value which lends itself to parallelism with xiang fe.  Karlgren provides the following pre-Han definitions for wen 3r: "drawn lines," "design," "striped," and "ornamented" (GSR. no. 475). The definitions for wen x given in the HYDCD — drawn from preclassical and classical sources — basically agree with Karlgren. They are: (A) "A pattern of disorderly intersecting colors" ^fe££tfftllfl5; (B) "Lines, stripes; a decorative design, pattern" $£Jt, ?E$t; (C) "Tattooed writing or patterns" $lJM3t^ WtffcffiC (HYDCD: 6:1512). Moreover, there is evidence showing that wen  referred to a  drawing or patterned design as far back in the Chinese language as Shang OBI. In the second chapter, I mentioned that in OBI the graph wen j£ likely depicts "patterns on a body," or a tattooed drawing (JGWZGL: no. 3236: KMJS: no. 676; also see SWJZZ: 9:21a). While the grapheme which represents patterning on a human chest eventually disappears from wen 3t, it is still manifest in Zhou BRI. To illustrate the OBI and BRI forms of wen j£, I provide the following graphs (JGWB: no. 1085; JSZD: 569):  140  In the context of Shijing no. 128, there is no doubt that the gist of the above definitions for wen 3£ semantically parallels the meaning of xiang fe. In conclusion, I believe that Hu Pingsheng's arguments regarding the characters xiang fe and wen  are very persuasive. In  terms of context, the lexical variant xiang fe in FYSJ S123 actually seems preferable to the alternative; that is, chang  'chang J | : In addition to matching the preceding character wen St  in terms of meaning, as a modifier xiang fe is also much more context-specific than chang U§. Taken to mean 'long' chang % can be used to modify almost any animate or inanimate object; whereas, the range of objects associable with xiang fe 'pictured' is more restricted. Based on this evidence, xiang fe appears to represent the original version of Shijing no. 128, while the use of chang J | in the transmitted versions of this ode seems to be the result of phonetic borrowing or some error. If only one line of textual transmission was represented by chang %, there would be legitimate grounds to replace it with, xiang fe. But since more than one transmitted version of the Shijing contains chang  the lineage method used earlier in  this chapter points towards adopting this character as the proximate original. Unfortunately, we lack evidence to explain the apparent contradiction between the results of context analysis and the lineage method. But for the time being, I prefer not to take Mao/Han/Qi Shi no. 128 as the undisputed original. Instead, in light of the legitimate reading offered by the FYSJ, I regard the transmitted version as contestable and the matter of the proximate original unresolved. Variation Study, no. 6 FYSJ, slip  S124:j/ ,||  M A O , no. 128; H A N : //'  H  While Lexical Variant Study no. 5 deals with stanza one of Mao no. 128, this study 141  examines lexical variants in stanza two of the same ode. According to Wen Xingfu, FYSJ slip S124 reads: ^^I^S^Il t £ l | | § (86). By comparison, the corresponding segment from the Mao Shi reads: TNII^^ , # I t f i ,ffiJIIIIJ"... the six reins are well in hand; the blackmottled grays and the black-maned bays are in the center (in the yoke), the black-nosed yellows and the blacks go as outside horses..." (ICS: 10:56; Karlgren, 1944: 211). Damage to the bottom of bamboo fragment S124 actually makes the final graph impossible to distinguish (see "Fuyang Hanjian Shi jing mo ben," no. 4 inHu, 1988: 79). Consequently, HuPingsheng is not as bold as Wen in his treatment of this particular graph, and does not attempt a reading of it — i.e., in Fuyang han jian Shijingyanjiu, Hu's reading of slip S124 is identical to that of Wen Xingfu, except that it does not contain the character gua f| (1988: 15, 79). Wen Xingfu does not give reasons for placing gua ft in his reading of S124; however, there are some good reasons for making this addition: (A) It makes sense that the partially visible graph is gua ft, because this graph is found in same immediate context in the Mao Shi. In reconstructing FYSJ SI24, it makes better sense to utilize related textual data from a parallel version of the Shijing, than to draw from other sources. This approach reduces the possibility of drawing conclusions which are arbitrary in nature. (B) The meaning of gua ff is semantically consistent with the context of this ode as a whole. According to the zhuan for Mao no. 128, gua f| refers to a kind of horse that is yellow with a black mouth (f^HHtf 0 S ) (Wang Xianqian: 444; GSR: no. 18). The categorization of horses on the basis of color and appearance, as is done in the case of gua ft, is the norm in Mao no. 128. For example, according to Karlgren qi IS refers to a "blackmottled gray horse" (Karlgren, 1944: 211; also see HYDCD: 12:851). As for the character zhu H , also mentioned in Mao no. 128, the zhuan states that this is a horse with a "white left foot" ( £ J 1 0 W) (Wang Xianqian: 443; Karlgren, 1944: 211: HYDCD: 12:796). Regarding the same character, the SWJZ states: "[it is] a horse whose back left foot is white" Wi^Jx.^. QUI (SWJZZ. 10: 6a). Four stallions mentioned at the start of second stanza in Mao no. 128 are described as "very big" (iZUtfc JL-^) (for the definition of kong JL see Wang Xianqian: 445; 142  HYDCD: 4:177; the GSR, no. 1108 defines fit as 'big and fat'). Further, the Mao zhuan states that liu  represents a horse with a reddish body and black mane ( ^ ^ M f f i H ® ^ )  (Wang Xianqian: 444; also see HYDCD: 12:868). Ma Ruichen takes ®F to be the complex form of the graph liu fg, and cites the SWJZ to substantiate the meaning given in the zhuan: Ws, ^ H M S M l i a "[As for liu IS], it is a reddish horse with a black furry tail" (374; SWJZZ: 10: 3a; also see Wang Xianqian: 444). In regard to li £ , the SWJZ gives the following definition: H ^ S f e "A horse which is deep black" (SWJZZ: 10:3a). And the zhuan corresponding to Mao no. 297 concurs: #fcMBll "[A horse which is] pure black is called// H " (ICS: 10:152; Wang Xianqian: 1064). The emphasis on color and size which characterizes the descriptions mentioned in this ode is an important element in the narrative of Mao no. 128. (C) If Wen is correct and the graph in question is indeed gua Jd, the fact it does not occupy the exact same place in the verse-line as its Mao counterpart is interesting, but not especially unusual. Actually, reversing the order of two characters standing next to one another in a pair is not uncommon in the Shijing. To demonstrate this, below I will provide four examples of reversible pairs from this classic (Xiang, 1980: 281). But before doing so, it warrants noting that the examples of reversal given all appear to be comprised of words which are to a large degree semantically similar — e.g., both words will refer to related phenomena like some kind of light, a modifier which expresses a similar meaning, or a type of dwelling, etc. In this way, examples one through to four below indirectly suggest that the pair made up of huo li H H , which is found in the Mao and Han Shi, might possibly be written as // huo H i t , as well. After all, both characters in this pair refer to horses which are distinguished by a particular appearance. This hints at the direction that my argument will eventually take, only I must first address the question of the FYSJ character >>/ H . More on this shortly. The four examples of reversible pairs from the Shijing are as follows: Example!: Mao no. 197 reads: ^ f f t j i (ICS: 10:94) Mao no. 258 reads: f&feBjftr (ICS: 10:135) 143  According to the GSR, both gong H and jing f & can mean "respectful, reverent" (nos. 1182, 813). And the HYDCD agrees with this assessment of the two characters (7:508, 5:486). Examples demonstrating the semantic similarity of gong | l andjing |ft can be found in numerous classical and preclassical sources. A passage in the Lun yu reads: "Since exemplary persons are respectful and impeccable in their conduct, are deferential to others and observe ritual propriety, everyone in the world is their brother" ^^WMM.^.,  ^ A I ^ W ^ H , 0#S  ^F*3 , WJlJfr'fe (Ames: 154: ICS: 14:31, italics mine). Moreover, the following passages may be compared: the Lun yu reads: "And his other disciples ceased to treat Zilu with respecf A ^ t H ^ (Ames: 145; ICS: 14:27, italics mine); furthermore, the Shang shu reads: "... the one in a position of younger brother does not think of Heaven's (clearness=) clear laws and so he cannot respect his elder brother" T l ^ ^ A M , Tblfc&faWSl (Karlgren, 1950: 42; ICS: 9:33, italics mine). Example 2: Mao no. 256 reads: WW A S (ICS: 10:133) Mao no. 257 reads: S A M (ICS: 10:134) In this example, ren A and min S can both refer to 'the masses' or 'common people' (bai xing Utt), as opposed to government officials or rulers (HYDCD: 1:1932, 6:1420). Some past scholars, such as the Song dynasty's Zhu X i  and the Tang's Kong Yingda J  have suggested that these two words convey an identical meaning (HYDCD: 6: 1420). However, for the purposes of this study, it is not important to discern the precise degree of semantic similarity in the words examined; we do not need to know, for example, if ren A and min J | are exactly synonymous, or if they are semantically separated by some slight nuance in meaning. Consequently, both in regard to example 2 and the following examples 1 will not pursue this matter in any detail. Example 3: Mao no. 6 reads:  (ICS: 10:3)  Mao no. 237 reads:  (ICS: 10:119)  The words jia ^ and shi _ seen in the above pair can both mean 'house, home' (HYDCD: 144  3:1422, 3:1459; GSR: no. 32). Below, both zhao $B and ming Eg can mean "bright" in preHan Chinese (HYDCD. 5: 594, 5: 684; GSR: nos. 760,1131). Example 4: Mao no. 247 reads: BS^Wg!fe(ICS: 10:126) Mao no. 273 reads: WQMM (ICS: 10:144) This chapter is not the place to provide an in-depth explanation of the various linguistic factors involved in the phenomenon illustrated here. The fact that reversible ordering is characteristic of the Shijing is enough to substantiate Wen Xingfu's claim regarding the questionable graph in FYSJ S124. At this juncture, it should suffice to mention that what is seen in the above examples constitutes a particular linguistic unit in Old Chinese which is characterized by pairs of words related in terms of grammatical function; namely, the words comprising the pairs are both nouns, verbs or adjectives. Further, it has been observed that the characters contained in these pairs can all stand on their own (Xiang, 1980: 281). The alternative to what Wen Xingfu proposes for this particular verse-line seems more probable — at least at first. This alternative constitutes a more obvious and straightforward approach to identifying the seventh graph in FYSJ S124. In short, this graph could be //' H . Because the semantic value of //' J§ embodies a description of appearance and color — in the same manner as gua ft, this character ideally suits the context of Shijing no. 128. Moreover, the alternative involving //' H has one distinct advantage over that which Wen proposes: it follows the word order implicit in the Mao version. According to this interpretation, both in the case of Mao Shi no. 128, as well as the corresponding FYSJ, the second character in the third verse-line of stanza three is //' Hf. By virtue of its relative simplicity — e.g., it requires less explanation and does not involve a reversal of word order — the second alternative adheres to Occam's razor. This well-known principle of logic is expressed as, "Entities [that is, assumptions used to explain phenomenon] should not be multiplied beyond what is needed" (Benet's: 706; Boltz refers to Occam's razor in his analysis of the Laozi, see 1997: 267). Further, from a slightly different perspective, the second alternative is also supported by the lineage method as discussed in Lexical Variant Study no. 1 and elsewhere. For the Han version of this ode contains the character li H just 145  like the Mao Shi. On these grounds, it seems relatively straightforward to conclude that the graph Wen thinks is gua if should instead be identified as // Ji. However, an additional factor must be considered before deciding which of the two alternatives is actually most preferable. That is, which is more compatible with the FYSJ graphs IU, gua ft or // ,§!? If we take the indistinguishable graph on slip S124 to be gua ft, and assume that a reversal of word order occurred in this line of the FYSJ, the counterpart to yi m should be // J§. But if we adopt the more straightforward alternative, one which does not require reverse ordering, the study of lexical variation should concentrate on the characters yi p and gua ft. From a strictly semantic perspective, yi If does not suit the overall context of the ode, and does not match up well with either // Jf or gua $|. Unlike these two characters,}'/ fjf does not refer to the appearance of a horse; instead,}'/ H refers to a horse that possesses a certain function or capacity. Following Xu Shen's interpretation in the SWJZ, the GSR defines yi B as: "post horses, relay horses" (no. 790; SWJZZ. 10:17a; also see HYDCD: 12:907). Moreover, since Shijing no. 128 is about a lady longing for a husband who is somewhere on a "war expedition," mention of a 'post horse' is really out of place (Karlgren, 1944: 211). But is there phonetic evidence indicating that one of these matches is more likely to occur in the context of Shijing no. 12 8? In the case of matching}// ,p to the character // Jl, there exists at least some chance of phonetic compatibility. This might suggest a relationship which is based on phonetic borrowing, or perhaps even cognate variation, although only if semantic evidence permits it. In Karlgren's Old Chinese reconstruction, the character}//1? is assigned the pronunciation *djak, and // H the pronunciation *ljeg (GSR: nos. 790, 878). While at first glance these pronunciations seem phonetically incompatible with one another — e.g., they contain different main vowels and final consonants, an analysis of phonetic feature indicates that these violations of phonetic criteria may be overlooked in this case. After all, in that the main vowels [*a] and [*e] are both [+unrounded] and [+front], and the consonants [*-k] and [*-g] are both [+velar] and [+plosive], the finals Karlgren assigns to yi H and // H are homorganic. (I refer to Pullum, p. 295, for this phonetic equation.) Further, the initials [*1-] and [*d-] are both [+voiced] sounds that are pronounced from the alveolar region in the front of the mouth, 146  which makes them completely homorganic. The most compelling reason not to violate the phonetic criteria adopted for this study lies in the evidence provided by the STz/y/Mg rimegroups, which indicate that yi H belongs to the duo  rime-group, and li H to ge  group  (Chen: 221,183). Their affiliation to different rime-groups means these characters are not necessarily anything but lexical variants; however, I will not rule out the possibility of a relationship between yi H and // H that is based on some kind of phonetic borrowing. Karlgren's Old Chinese pronunciation ofyi H does not match gua ft as well as it does li H — i.e., the *djak he assigns to yi H differs from gua ft *kwa (GSR: no. 18) in a number of ways. For example, the initial [*k-] is produced from the velum positioned just above the throat region at the back of the mouth. Thus, it is not homorganic with the initial [*d-] sound assigned to the character;// HL As explained in chapter one, although some evidence suggests that no phonetic constraints should be placed on initials, in this thesis 1 have adopted the very clear rule that phonetic compatibility for word initials must be based on homorganicity. Although both yi H and gua ft fall into the ge  rime-group, which makes them  phonetically compatible in terms of finals, this has no bearing on the incompatibility of the initials. Therefore, I do not think there is any chance that these two characters are phonetically interchangeable in Old Chinese (Chen: 186). Taking the phonological evidence into consideration, I side with Wen Xingfu's conclusion that a reversal of word order results in a matchup of the lexical variants y/ H in FYSJ SI24 and // H in the Mao/Han Shi. This conclusion contrasts the semantic evaluation arrived at earlier, something which attests the complexity of lexical studies in general, and the value of phonological evidence. However, despite my assertion of phonetic compatibility between;// H and // H, I do not think that either of these characters is an actual phonetic loan. Instead, I attribute this particular case of lexical variation to scribal error. It seems quite feasible that the scribe responsible for writing FYSJ slip S124 mixed up two characters which sounded alike to him, and wrote one which is incorrect. For this type of mistake to have occurred phonetic compatibility in the strictest sense would not be required; the characters need only have been close enough in terms of pronunciation to cause confusion. Such a mistake is especially understandable if the erroneous character seemed to fit the context, 147  which is probably the case here. It is my opinion that the character li _ found in both Mao no. 128 and the Han version of the Shijing is both most suitable for the context and the proximate original; whereas the FYSJ variant is the product of human error which took place sometime after the original Shijing was compiled. As mentioned earlier, in this particular ode all of the horses mentioned are characterized by their appearance. In keeping with this indisputable narrative tendency, the character // J§ refers to kind of horse that is "pure black." By contrast, the character yi P refers to a kind of horse which performed a certain function in early China. Conclusion: Since the results of the orthographic study conducted in chapter two have already been discussed in detail elsewhere, in this section I will focus only on the accomplishments of the third chapter. Lexical Variation Study no. 1 primarily dealt with FYSJ SO 18 and the corresponding segments from Mao/Lu Shi no. 25. The comparison of these texts involved syntactic change brought about by the absence of hu ^ in the FYSJ version. This particle seems to be an optional syntactic element, and in written form its deletion does not seem to lead to a significant change in meaning. At the same time, however, taken in a larger context of dialectology and linguistic evolution, the presence of hu ¥• in some versions of the Shijing and not others has interesting implications. I reviewed several possible implications suggested by Hu Pinsheng, but the scope of this chapter did not permit extensive elaboration on subjects outside the identification of lexical variants and pursuit of the proximate original. In this chapter, several methods were applied to find the proximate original between two lexical variants, the most widely recognized and convincing of which is the lineage method. It indicates that the FYSJ variant likely does not represent the proximate original in this case; instead, it may signify a later effort to standardize the Shijing text, convert the pentasyllable meter into one which is tetrasyllable. In order to show that syntactic variation is not only seen in FYSJ SO 18, in chapter three I also outlined the same phenomenon in two other FYSJ slips: S017andS026. The second study in this chapter also took into consideration a particular type of lexical variation which involved synonymity. The examination of the synonyms hu 148  and he  jBJ from FYSJ Sl 15/Lu Shi and Mao Shi no. 115, respectively, eventually lead to the uncovering of a proximate original. Not only is hu  an older form of the interrogative  substitute, but there are also more lines of transmission using this character. Although Lexical Variation Study no. 6 does not directly follow the two studies summarized above, it belongs in the same category; rather than actually result in a new reading of one of the poems in the Shijing, this study contributes to a general overview of the kind of lexical variation possible when comparing the FYSJ with its transmitted counterparts. As such, Lexical Variation Studies nos. 1,2 and 6 all address thesis objective (B), as stated in the introduction to this chapter. In no. 6,1 showed that the FYSJ character^/ | f is not a graphic variant of //' JU seen in the Mao/Han Shi. Instead, I proposed that yi H is a lexical variant which was selected on the basis of a similarity of sound The fact that the lineage method indicates // H is the legitimate character for this particular context substantiates this claim. The other three studies in chapter three all presented new readings of the Shijing text; consequently, they all dealt with thesis objective (A). As a result of the analysis conducted in Lexical Variation Study no. 3, the FYSJ S045 characterju J_ proves to be a preferable alternative to ju $L found in Mao no. 41. More specifically,yu j_ seems more suited to the narrative structure and theme of Shijing no. 41. However, since the lineage method and the principle of lectio difficiliorpotior do not apply to this case of lexical variation, I could not prove conclusively that FYSJ S045 represents the proximate original. The lineage method does suggest tong % is the proximate original in Lexical Variation Study no. 4. While tong f& is supported by a total of three lines of transmission, the alternative — tongflffrom FYSJ S048 — is represented by only one. But FYSJ S048 still offers a completely legitimate reading of Shijing no. 42. In fact, if we take guan _5 in lines two and three of the second stanza to mean 'flute', the modifier tong ff> 'red' is not nearly as object-specific as tong |f, which refers to a kind of wood that is especially suitable for making woodwind instruments. Similar disappointment is encountered in Lexical Variation Study no. 5. Here the context again suggests that FYSJ S123 xiang ;§?. is preferable to its alternative chang % in Mao/Han/Qi Shi no. 128. But the lineage method indicates that chang % is the proximate original.  149  Thesis conclusion: In the course of this thesis, I have approached a study of the FYSJ from several different perspectives. After providing a detailed overview of the methodology used in this thesis in chapter one, I proceeded to employ this methodology in the orthographic analysis undertaken in chapter two. The main objectives in chapter two are: (A) to date the FYSJ by studying the script it is written in; (B) to determine the .extent to which the composite FYSJ manuscript is comprised of different source manuscripts. While this orthographic analysis did not take the entire FYSJ into account, something which would have required a study all its own, I nonetheless believe that it did lead to some very telling results. For example, it showed that some of the FYSJ is written in an orthography which probably dates back to the middle of the third century B.C. At the same time, other parts of this manuscript are written in a form of script datable to the early Western Han dynasty. Moreover, the orthographic analysis conducted in the second chapter also demonstrated the degree to which the FYSJ manuscript can be divided into parts from other preexisting manuscripts. It appears that sometimes as few as two bamboo slips from the FYSJ represent a separate original manuscript. Previous attempts to date the FYSJ and evaluate its physical nature were very narrow in scope and cursory in terms of methodology. As such, this thesis represents the first time that this important manuscript has received the rigorous orthographic analysis it really warrants. For reaching a clear understanding of the makeup of the FYSJ is crucial if we are to avoid certain fundamental mistakes, such as treating the FYSJ as if it were a cohesive manuscript written by the same hand. In this sense, the FYSJ is a much different manuscript than either the SHD or MWD. Moreover, for graphologists or scholars working in other respectivefieldsto begin to consider the FYSJ in earnest, dates must be attached to it. Dates are necessary for the FYSJ to be placed within the lineage of Shijing scholarship and studied as such — e.g., we know now that at least some of the FYSJ actually antedates the Mao Shi ca. 150 B.C. Further, dates are needed to place the FYSJ within the evolution of early Chinese script. At this point we know that the FYSJ can shed light on the transition period during which Small Seal script yielded in terms of usage to the burgeoning Ancient Clerical script. And the FYSJ has much to offer the study of individual Chinese graphs; as the orthographic study undertaken in this thesis suggests, the FYSJ doubtlessly contains 150  numerous examples of graphs previously unseen in Small Seal or early Ancient Clerical form. As such, these graphological findings should be used to update resource materials such as the LZB. This thesis was designed to function as a launchpad for further exploration of the FYSJ manuscript, and, indeed, to do justice to the areas of interest just mentioned individual studies are required. The methodology discussed in chapter one also proved indispensable in the lexical analysis conducted in chapter three. To reiterate once more, the objectives 1 set for the third chapter of this thesis are as follows: (A) to reread several works in the Shijing using evidence of lexical variation from the FYSJ; (B) to provide examples of different kinds of lexical variation represented in the FYSJ; and (C) whenever possible to uncover the proximate original from the various lexical variations analyzed. While objective (C) has never been attempted by either Chinese or Western scholars, points (A) and (B) have been given some attention. However, as in the case of the orthographic analysis of the FYSJ, these previous attempts to study lexical variation in this manuscript are marred by a lack of methodological consideration. I believe that this thesis contributes to our understanding of lexical variation and the FYSJ just by advancing a systematic and reliable form of methodology. Like the study of orthography found in chapter two of this thesis, the six lexical studies conducted in chapter three offer insight into the nature of the FYSJ. In this case, however, it is the textual makeup of the FYSJ which is highlighted. For example, one lexical study demonstrates that the version of Shijing nos. 23, 25 and 29 represented by FYSJ slips SO 17, SOI 8 and S026 is comprised of a different syntactic structure than the corresponding transmitted Shijing. Findings such as this contribute a great deal to our understanding of how the Shijing took form. Further research from the perspective of historical linguistics is needed for us to fully understand the implications of these findings. But of all the lexical studies conducted in chapter three, those which suggest new interpretations of certain odes are probably the most significant both for specialists and the general readership. For one thing, because the Shijing is a text which is read and referred to by scholars with interests as broad as literature, history, religion and culture in general, any time a new translation of one of its poems can be proposed, there is no doubt it will receive widespread recognition.  151  Bibliography Ames, Roger T. The Analects of Confucius: A Philosophical Translation. Toronto: Ballantine, 1998. Sun-tzu, The Art of War: The First English Translation Incorporating the Recently Discovered Yin-ch 'ueh-shan Texts. New York: Ballantine, 1993. Baxter, William H. A Handbook of Old Chinese. Vol. 64 of Trends in Linguistics: Studies and  Monographs. New York: Mouton de Gruyter, 1992. Blanford, Yumiko F. "A Textual Approach to 'Zhanguo zhongzhengjia shu': Methods of Determining the Proximate Original Word Among Variants." Early China 16 (1991): 187-207. Boltz, William. "Manuscripts with Transmitted Counterparts." In Edward L. Shaughnessy, ed. New Sources ofEarly Chinese History, edited by Edward L. Shaughnessy, 251-83. Berkeley: The Institute of East Asian Studies; University of California, 1997. The Origin and Early Development ofthe Chinese Writing System. No. 78 of the  American Oriental Series. New Haven: American Oriental Society, 1994. "Textual Criticism and the Ma Wang tui Lao tzu." Harvard Journal of Asiatic Studies AAA (1984): 185-224. Chen, Chusheng W-%B*$L, comp. Jin wen changyongzidian M^Jft [A Dictionary of Frequently Used Bronze Characters]. Xi'an: Shaanxi renmin, 1987. Chen, Fuhua ¥>M'1MM and He Jiuying jqfTLjgt Guyun tongshao l^MMM [A General Outline of Ancient Rhyme]. Beijing: Zhongguo shehui kexue, 1987. Chou, Fa-kao, chief ed. Jin wengulin #;s;g&# [An Etymological Dictionary of Archaic Chinese Bronze Inscriptions]. 16 Vols. Hong Kong: The Chinese University of Hong Kong, 1974. Dobson, W.A.C.H. A Dictionary of Chinese Particles. Toronto: University of Toronto Press, 1974. Downer, G.R. "Derivation by Tone-change in Classical Chinese." Bulletin ofthe School of Oriental and African Studies 22 (1959): 258-90.  152  Duan, Yucai WSM, annot. Shuo wen jie zi zhu MXM^fe [An Annotated Version of the Explanation of Compounds and Analysis of Graphs]. 8th ed. Shanghai: Shanghai guji chuban she, 1997. Fu, Juyou 11M|I# and Chen Songchang WJffeM, eds. Mawangdui Han mu wen wu MEEtf£WM JL^S [The Cultural Relics Unearthed from the Han Tombs at Mawangdui]. Changsha: Hunan chuban she, 1992. Gao, Heng jSpf. Shi jing jin zhu WM^fe [A Modern Interpretation of The Book of Odes]. Shanghai: Shanghai guji chuban she, 1980. Gao, Ming j|f B§, comp. Gu wenzi leibian " f ^ I S l ^ M i i [A Compilation of Archaic and Ancient Graphs]. Beijing: Zhonghuashuju, 1980 Harper, Donald. " A Chinese Demonography of the Third Century B.C." Harvard Journal of Asiatic Studies 45 (1985): 459-98. He, Lezhi ^ U l i et al. Gudai hanyu xuci tongshi ^JXlMMMMMW  [A Complete  Interpretation of Ancient Chinese Particles]. Beijing: Beijing chuban she, 1989. Henricks, Robert G. Lao-tzu Te-tao Ching: A New Translation Based on the Recently  DiscoveredMa-wang-tui Texts. New York: Ballantine, 1989. " A Complete List of the Character Variants in the Mawangdui Texts of Lao Zi." Journal of Chinese Linguistics 10 (1982): 207-30. Hightower, James Robert. Han Shih Wai Chuan: Han ling's Illustrations of the Didactic  Application of the "Classic ojSongs''.Vol. 11 of Harvard- Yanching Institute Monograph Series. Cambridge: Harvard University Press, 1952. Hong, Diaotao ^MM, ed. Lizi bian f f t ^ J l [A Compilation of Graphs in Clerical Script]. Beijing: Wenwu chuban she, 1991. Hu, Pingsheng $ F F : £ - "Some Notes on the Organization of the Han Dynasty Bamboo 'Annals' Found at Fuyang." Early China 14 (1989): 1-26. . and Han, Ziqiang |g g 5£, eds. Fuyang han jian Shijingyanjiu  ^MMWWMfflft  [Research on Bamboo Slips of the Han Dynasty Found in Fuyang], Shanghai: Shanghai guji chuban she, 1988. 753  Hung, William, gen. ed. A Concordance to Chuang Tzu. Supplement No. 20 in HarvardYenching Institute Sinological Index Series. Cambridge: Harvard University Press, 1956. Kang, Yin MW- Wen ziyuanliu qian shuo S^W^MWi  [A Simplified Explanation of the  Origins of Chinese Compounds and Graphs]. Beijing: Renming chuban she, 1979. Karlgren, Bernhard. "Loan Characters in Pre-Han Texts." The Bulletin for the Museum of Far Eastern Antiquities 35 (1963): 1-18. "Grammataca SericaRecensa." The Bulletin for the Museum ofFar Eastern  Antiquities 29 (1957): 1 -332. "Cognate Words in the Chinese Phonetic Series." The Bulletin for the Museum of Far Eastern Antiquities 29 (1956): 1-18. "The Book of Documents." The Bulletin for the Museum of Far Eastern Antiquities  22(1950): 1-81. "The Book of Odes, Ta ya and Sung." The Bulletin for the Museum of Far Eastern Antiquities 17 (1945): 65-99. "The Book of Odes, Kuo feng and Siao ya." The Bulletin for the Museum of Far Eastern Antiquities 16(1944): 171 -256.  "Glosses on the Kuo feng Odes." The Bulletin for the Museum of Far Eastern Antiquities 14 (1942): 71-247. Keightley, David. "The Origins of Writing in China: Scripts and Cultural Contexts." In The Origins of Writing, edited by Wayne M. Senner, 171-201. Lincoln: University of Nebraska Press, 1989. Lau, D.C. and Chen Fong Ching, series eds. Concordances to the Erya, Xiao jing. Vol. 16 and 17 in The ICS Ancient Chinese Texts Concordance Series. Hong Kong: The Commercial Press, 1995. , series eds. A Concordance to the Lunyu. Vol. 14 in The ICS Ancient Chinese Texts Concordance Series. Hong Kong: The Commercial Press, 1995. , series ed. A Concordance to the Chunqiu Zuo zhuan. Vol. 11 in The ICS Ancient  Chinese Texts Concordance Series. Hong Kong: The Commercial Press, 1995. 154  _ series ed. A Concordance to the Mao shi. Vol. 10 in The ICS Ancient Chinese Texts Concordance Series. Hong Kong: The Commercial Press, 1995. , series ed. A Concordance to the Yi li. Vol. 7 in The ICS Ancient Chinese Texts Concordance Series. Hong Kong: The Commercial Press, 1994. , series ed. A Concordance to the Shang shu da zhuan. Vol. 5 in The ICS Ancient Chinese Texts Concordance Series. Hong Kong: The Commercial Press, 1994. , series ed. A Concordance to the Zhou li. Vol. 4 in The ICS Ancient Chinese Texts Concordance Series. Hong Kong: The Commercial Press, 1994. , series ed. A Concordance to the Li ji. In The ICS Ancient Chinese Texts Concordance Series. Hong Kong: The Commercial Press, 1992. Li, Fang-kuei ^jj^.. Shang guy in yanjiu ±^^W% [Research on the Phonology of Old Chinese]. Beijing: Shangwu yinshu guan chuban, 1980. Li, Ling ^ 9 and Keith McMahon. "The Contents and Terminology of the Mawangdui Texts on the Arts of the Bedchamber." Early China 17 (1992): 145-85. Li, Meili ^$£11. Mawangdui Han mu yanjiu mulu Mr^-^MMW^^M [A Catalogue of the Research on the Han Tombs at Mawangdui]. Changsha: Hunan sheng bowuguan, 1992. Loewe, Michael. "China's Sense of Unity as Seen in the Early Empires." T'oungPao 80 (1994): 6-25. Lu, Deming itlgBjE}. Jingdian shi wen fgJ&L^|j£ [An Interpretation of the Text in the Classics and Documents]. 3 Vols. Shanghai: Guji chuban she, 1985. Luo, Zhufeng H'tt®, chief ed. Hanyu da cidian tJiMAff Jft [The Great Chinese Dictionary]. 12 Vols. Hong Kong: Joint Publishing, 1988. Mair, Victor. Wandering on the Way. Toronto: Bantam, 1994. Morohashi, Tetsuji t f f l f f g ^ . Dai KanwaJiten Aift^tfl?ft [The Great Chinese-Japanese Dictionary]. 12 Vols. Tokyo: Taishukan, 1994. Ma, Ruichen H3inf||. Mao shi zhuan jian tongshi 155  [A Complete Interpretation  of the Mao Odes, Commentary and Notes]. Beijing: Zhonghua shuju, 1989. Matsumaru, Michio f&tllM-W- and Takashima, Kenichi ft H i ^ — . Kokotsu moji jishaku soran ^WSC^-^W^W [Synthetic Index for Interpretation of Oracle-Bone Characters]. Tokyo: Tokyo University Press, 1994. Mattos, G.L. The Stone Drums of Ch 'in. Nettetal: Steyler, 1988. ., trans. "Studies on Archaic Chinese." Monument a Serica 33 (1974-75): 219-87. Norman, Jerry. Chinese. New York: Cambridge University Press, 1988. O'Grady, William and Michael Dobrovolsky. Contemporary Linguistic Analysis. 3d ed. Toronto: Copp Clark Ltd, 1996. Pankenier, David, ed. On Script and Writing in Ancient China. Stockholm: Skrifter utgivna av F5reningen for Orientaliska Studier, 1974. Pulleyblank, Edwin G. Outline of Classical Chinese Grammar. Vancouver: University of British Columbia Press, 1995. "How do we Reconstruct Old Chinese?" Journal of American Oriental Society 112.3(1992): 365-82. "Some New Hypotheses Concerning Word Families in Chinese." Journal of Chinese Linguistics. 1. 1 (1973): 111-24. Pullum, Geoffrey K. and William A. Ladusaw. Phonetic Symbol Guide. 2d ed. Chicago: The University of Chicago Press, 1996. Qiu, Xigui HHJIE. Wenzixue gaiyao X^f^flfciS [An Outline of the Study of Writing]. Beijing: Shangwu yinshu guan, 1990. Riegal, Jeffrey. "Classic of Odes." In The Columbia Anthology of Traditional Chinese Literature, edited by Victor Mair, 156. New York: Columbia University Press, 1994. Saussy, Haun. The Problem ofa Chinese Aesthetic. Stanford: Stanford University Press, 1993. Shang, Chengzuo j^t/fc#. Shi ke zhuan wen bian ^EMiilzJZM [A Compilation of Stone Inscriptions in Seal Script]. Hong Kong: Zhonghua shuju, 1976. 156  Shaughnessy, Edward L. / Ching, The Classic of Changes: The First English Translation of the Newly Discovered Second-Century B.C Mawangdui Texts. New York: Ballantine,  1996. " A First Reading of the Mawangdui Yijing Manuscript." Early China 19 (1994): 4773. Silbergeld, Jerome. "Mawangdui, Excavated Materials, and Transmitted Texts: A Cautionary Note." Early China 8 (1982-3): 79-92. Tang, Chengyuan Mf&Gz- Jin shi zidian ^B^-^ [A Graphological Dictionary of Bronze and Stone Graphs], Beijing: Zhongguo shu dian, 1995. Trask, R.L. A Dictionary of Grammatical Terms in Linguistics. New York: Routledge, 1993. Van Zoeren, Steven. Poetry and Personality: Reading, Exegesis, and Hermenuetics in  Traditional China. Stanford: Stanford University Press, 1991. Waley, Arthur. The Book of Songs: The Ancient Chinese Classic of Poetry. New York: Grove  Weidenfeld, 1987. Wang, Fang-yii. Introduction to Chinese Cursive Script. No. 68 of Mirror Series A. New Haven: Far Eastern Publications, Yale University, 1958. Wang, Jingzhi j £ f f ^ i . Shi jing tong shi f?#Mft8 [A Complete Interpretation of The Book of Odes]. Taizhong: Furen Daxue, 1991. Wang, Li 3 : il. Tongyuan zidian jl] M^-M [A Dictionary of Cognate Words]. Beijing: Shangwu yinshu guan, 1982. Wang, Wenhui EEStM, comp. Jianmingjin wen cidian ^BilfeJtflljSt [A Concise Dictionary for Bronze Inscriptions]. Shanghai: Shanghai cishu chuban she, 1998. Wang, Xianqian EE£M- Shi san jiayiji shu W^-^K^MM [A Compendium of Annotations on the Mao and Three Schools Odes]. Beijing: Zhonghua shu ju, 1987. Wang, Yinzhi Jing zhuan shi ci M&MM [An Interpretive Dictionary the Classics and Commentaries]. Edited by Wang Yunwu I i S and Zhu Jingnong ^ M J I Beijing: Shangwu yinshu guan, 1932. 157  Watson, Burton. Hsun Tzu: Basic Writings. New York: Columbia Press, 1963. Wen, Xingfu Shijing Mao shi zhuan Zheng qian bianyi W$£=&WWM- PUIS [An Analysis of the Book of Odes, the Mao Commentaries and Zheng's Notes]. Taibei: Wenshizhe chuban she, 1989. Wu, Hung MM- "Art in a Ritual Context: Rethinking Mawangdui." Early China 17 (1992): 111-44. Wu, Jiulong ^fiM et. al. eds. Sunzijiaoshi M-f$M [A Revised Interpretation of Sun zi, The Art of War]. Beijing: Junshi kexue chuban she, 1990. Xiang, X i [ R J | | . Shijing cidian % [A Dictionary of The Book of Odes]. Chengdu: Sichuan renming chuban she, 1986. . "Shijing li de fuyinci" g ^ M M f t l ["Reduplicative Words in The Book of Odes"]. Yuyanxue luncong ^m^%m9. [A Compendium of Expositions on Linguistics] 6 (1980): 27-54. Xu, Fuchang #811- Shuihudi qinjianyanjiu fti^iftfl'^ifFfg [Research on the Qin Bamboo Inscriptions from Shuihudi]. Taibei: Wenshizhe chuban she, 1993. Xu, Peigen $3&%k and Wei Rulin MlkM- Sun Bin bingfa zhu shi ^ K ^ & f t S [A Revised Version of Sun Bin's Art of War]. Taibei: Liming wenhua shiye gongsi, 1976. Xu, Zhongshu ftt^f?- Jia gu wen zidian ^ # ^ 1 ^ ^ [A Dictionary of Oracle Bone Script]. Chengdu: Sichuan shuju chuban she, 1988. Yao, Jiheng Shijing tong lun flMMtt [A General Discussion of The Book of Odes]. Beijing: Zhonghua shuju, 1958. Yates, Robin D.S. "The Yin-Yang Texts from Yinqueshan: An Introduction and Partial Reconstruction, With Notes on their Significance in Relation to Huang-Lao Daoism." Early China 19(1994).  Yip, Po-ching and Don Rimmington. Chinese: An Essential Grammar. New York: Routledge, 1997. Yu, Guanying ^M^.  Shi jingxuan zhu jitMitrJt [A Selective Annotation of The Book of  158  Odes]. Hong Kong: Daguang chuban she, 1966. Yu,Changhong ^UlifT. and Han Juelin fj|i§[#. Changyongwenyan xuci shouce #M3t"iFJul ff THH} [A Manual of Commonly Used Classical Chinese Particles]. Hebei: Hebei renmin chuban she, 1993. Yu, Xingwu T ^ H , chief ed. Jia gu wenzi gulin ^ j^^^gfcfif [A Treasury of Commentaries for Oracle Bone Graphs]. 4 Vols. Beijing: Zhonghua shu ju, 1996. Yuan, Ke $0$$. Zhonggua shenhua chuanshuo cidian ^llf^tSfllil&ifM [A Dictionary of Chinese Myths and Legends]. Shanghai: Shangwu yinshu guan, 1986. Yuan, Zhongyi H#—. Qin dai tao wen #ft|^3t [Pottery Inscriptions from the Qin Dynasty]. Xi'an: San Qin chuban she, 1987. Zhang, Shichao jjjltilg and Zhang Yuchun jjg3E#, comp. Qin jian wenzi bian ^ ^ 5 C ^ H [A Compilation of Graphs from Qin Bamboo Inscriptions]. Beijing: Zhongwen chuban she, 1990. Zheng, Huimei WMM.- Hanjian wenxue de shufa yanjiu tJtfi^PKltftWfu [Research on the Calligraphy in Han Bamboo Manuscripts]. Taibei: Guoli gugong bowu yuan, 1984.  159  


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