Paper Menu >>
Journal Menu >>
2013. Vol.2, No.1, 8-24
Published Online February 2013 in SciRes
Copyright © 2013 SciRes.
The Xinshu 新書 Reexamined: An Emphasis on
Usability over Authenticity
Wuhan Irvine Culture and C ommunication, Wuhan, China
Received November 29th, 2012; revised December 30th, 2012; accepted January 5th, 2013
A collection of texts conventionally ascribed to Jia Yi 賈誼 (200-168 BC), the Xinshu 新書 has been
subjected to an ages-long debate regarding its authenticity. The present study disclaims the discovery of
any adequate evidence to prove the text trustworthy; but it finds the arguments for its forgery ill founded.
Rather than present merely an account of this dilemma or attempt to corroborate either position in the de-
bate, this paper argues against the approach in textual criticism that views early texts through a dualistic
prism of authenticity vs. forgery. A case of forgery should be established upon no less concrete evidence
than should one of authenticity. The mere lack of positive evidence can hardly be regarded or used as any
negative evidence to disprove a text. Given the dilemma, the paper suggests treating the Xinshu nonethe-
less as a workable and even currently reliable source for our study of Jia Yi until that very day dawns
upon us with any unequivocal evidence of its forgery detected or, better still, excavated.
Keywords: Jia Yi; Xinshu; Western Han; Authenticity; Forgery; Chinese Textual Criticism
The Xinshu 新書 is a collection of texts traditionally as-
cribed to Jia Yi 賈誼 (200-168 BC). An important figure in
the Chinese intellectual history, Jia Yi played an active role in
initiating a political reform during the early Han. The form of
the extant Xinshu can be traced back to the Song dynasty (960-
1279) if not earlier. But none of the Song editions survives
except by way of Lu Wenchao’s 盧文弨 (1717-1796) edition
of the Xinshu and Lu’s editorial notes therein (Nylan, 1993: p.
162). There are, moreover, texts in the Xinshu that are cor-
rupted. All this resulted in the centuries’ long debate over the
authenticity of the text. There have been scholars who consider
the Xinshu as a ca se where either a “reckless person” (wangren
妄人) wrote a text and deceptively attributed it to Jia Yi or
someone forged a book of Jia Yi by compiling and altering
some quotations of Jia Yi gathered from other sources. In this
study, I will follow Cohen’s suggestion to refer to both kinds of
spurious texts as pseudepigrapha1.
The view of the Xinshu as a pseudepigrapha is explicit in the
Siku quanshu zongmu tiyao 四庫全書總目提要 (Siku tiyao
henceforth). Compilers of the Siku tiyao (Siku compilers hence-
forth) claimed to have identified some other early texts-primar-
ily the Hanshu 漢書—to be the sources based on which the
Xinshu was forged. The unidentified forger, the compilers sus-
pected, forged the Xinshu by fist dividing up some of Jia’s es-
says cited in those other sources, then editing and compiling the
divided passages into 58 chapters, and finally putting each of
the 58 under an imposed title in order to falsely establish it as a
chapter. The Siku compilers opined that the forger’s motive was
to match the total of the forged chapters with the number re-
corded in the bibliographic treatise of the Hanshu (i.e. Hanshu
From late 1950s to early 1960s, Chen Weiliang 陳煒良
(1958) in Hong Kong and a team of four accomplished scholars
in Peking University (i.e. Sun Qinshan et al., 1961) each pro-
vided a comprehensive summary of the debate. Both raised
quite a few original and cogent points. After reviewing the
same set of contending arguments, they each arrived at a con-
clusion in opposition to the other’s. The four scholars in Beijing
conclude their study by arguing for the authenticity of the Xin-
shu. Following the four scholars, Wang Zhouming 王洲明
(1982) conducted a similar collative comparison and reached
the same conclusion. The four scholars’ major points are
incorporated in Wang Zhouming’s study and presented along
with Wang’s substantial original findings. One can thus legiti-
mately treat Wang as a major defender of the Xinshu in modern
China. Without being informed of Chen Weiliang’s insightful
challenge, Wang Zhouming, however, did not have a chance to
deepen his discussion.
Examining a textual issue invariably involves ascertaining
the date of a text. The language of a text has long been an area
where scholars commence their search for evidence. In the past
few years, there have been scholars who attempted linguistic
approach to the received Xinshu, looking for what the text
might reveal linguistically about its date. Wang Zhouming, for
one, included a linguistic discussion of a moderate scope in his
aforementioned research. In Europe and at the turn of the mil-
lennium, Rune Svarverud conducted a much more extensive
research in this respect. His attempts significantly contribute to
the linguistic study of the extant Xinshu. But, his study, as Luo
1As Cohen (2000: p. 195) observes, “[o]ne aspect of the work of the Ch’ing
Dynasty textual scholars […] was the identification of various types o
spuri- ous books. This area of scholarship is called pien-wei 辨偽 ‘distin-
guishing the spurious’. The general term for such books is wei-shu 偽書,
which is often translated as ‘forgery’. However, ‘pseudepigrapha’ (writings
of falsely ascribed authorship) is a more appropriate translation because
‘forgery’, with its implication of intentional fraud, applies to some ty
wei-shu but not to all.”
Shaodan (2002) points out, falls short of providing adequate
proofs2. More work is yet to be done in this field.
Besides language, other textual issues have also been brought
into focus. Some textual features were particularly cited as
evidence of forgery. They include the textual corruption, the
flawed writing style, the lack of reference to the Zuozhuan 左
傳, and the mismatch between some quotes of Jia Yi in the
Hanshu 漢書 and their counterparts in the Xinshu.
The discussion on writing style often touches upon the tex-
tual mismatch between the Xinshu and Jia Yi’s quotes else-
where. Such discussion adequately demonstrates scholars’ dif-
ficulty in maintaining objectivity when they face the current
lack of transmitted or excavated manuscript as a reliable refer-
ence point. The afore-mentioned four scholars in Peking Uni-
versity, for instance, conducted a textual comparison between
the Xinshu and the quotes of Jia Yi in the Hanshu. They con-
sidered the Xinshu authentic because they found the texts in the
Xinshu smoother and more consistent in style than the quotes in
the Hanshu. But, precisely the same kind of comparison left
Chen Weiliang with an opposite impression about the writing
style of the Xinshu. He concluded that the Xinshu was forged by
copying those quotes in the Hanshu. Likewise, Yao Nai 姚鼐
(1732-1815) also considered the Xinshu spurious partly because,
to him, the Xinshu did not seem to present as good a writing
style as did its textual counterparts in the Hanshu. Their con-
flicting views, interestingly, converged on a basic assumption
of Jia Yi’s perfection in writing style. Stemming from the as-
sumption was a conviction that, between the Xinshu and those
quotes, the better written text would necessarily be the original
text. But one may find the common ground questionable at a
time before the perfection of Jia Yi’s writing style can be estab-
lished in the first place.
Chen Weiliang, in addition, noted that the spuriousness of
the Xinshu was evidenced in its lack of reference to the Zuo-
zhuan, because such absence was impossible for someone like
Jia Yi, an official-scholar with acclaimed expertise in Zuozhuan
study in his generation. In the present study, I will point out
that such absence is hardly inconceivable once we take into
consideration the time in which Jia Yi had lived, a time when
neither the Zuozhuan School of historical studies nor the prac-
tice of citing canonical texts was favorably received in political
Since none of the above approaches has yet shown un-
equivocal evidence, the present study will seek hard evidence
primarily in the physical layout of the text. The insufficiency of
relevant excavations may impede further progress in scholar-
ship. This study stuggests that progress can nevertheless be
made on the basis of received texts. The evidence of early text
can be detected in what we may call embedded end title for
passages in the extent Xinshu. Meanwhile, the paper would
disclaim any intention to establish a case of authenticity. What
it opposes is the adherence to the dichotomy of authenticity and
spuriousness where the Xinshu is concerned.
The present study, therefore, would point out that, for all the
lack of substantial evidence to authenticate the Xinshu, it is
unadvisable to label the Xinshu as pseudepigrapha. And this
paper will question the viability of such a practice as treating
the textual corruption of the extant Xinshu to be evidence of its
Following this introduction, the first section of this article
tackles such fundamental questions as what the Xinshu is and
what it was thought to be. I find that, first, no pre-modern edi-
tions of the Xinshu contain Jia Yi’s literary works except in an
appendix in some cases. Second, the evidence I examine does
not support the proposition that the Xinshu was compiled on the
basis of Jia Yi’s quotes in the Hanshu.
The second section begins by summarizing the corrupted
condition of the Xinshu text. Close attention is paid to the pecu-
liar way in which some parts of the text are divided and how
the peculiarity aroused the suspicion of forgery. The discussion
proceeds to relate such peculiarity in the Xinshu to the formats
of silk and bamboo texts in the early times. I will focus on the
way embedded end titles were used in early texts and the
change they underwent. The section cites concrete evidence to
suggest that originally all or some of the chapters in the extant
Xinshu were merely passages rather than freestanding texts.
They were either mistaken for titled essays (i.e. chapters) or
intentionally but erroneously treated as chapters i n later editions.
Hence the peculiar divisions of some of the texts in the Xinshu
are hardly indicative of forgery.
The third section examines the views of Yao Nai and the
Siku tiyao. Both question the authenticity of the Xinshu. Both
have inspired later debates, including Yu Jiaxi’s 余嘉錫
(1883-1955) forceful refute of the Siku tiyao. But unlike the
Siku compilers, who would give credit to the part of the Xinshu
that bore no textual parallel to the Hanshu, Yao Nai considered
the Xinshu entirely untrustworthy. I will argue that Yao’s point
proceeded from a misinterpretation of both the Xinshu and the
The fourth section discussed Chen Weiliang’s textual study.
To my knowledge, there has been hardly any response to
Che n ’ s c h al l e ng e e x c e pt Sv arverud’s boo k. In this section I will ,
besides introducing Svarverud’s points, question the validity of
the evidence Chen cites, which includes the verbosity of the
Xinshu text and the lack of mention of the Zuozhuan in the
In the conclusion, I, on the one hand, acknowledge the insuf-
ficiency of current evidence to authenticate the Xinshu. On the
other hand, I argue that the reliability of the extant Xinshu can
be recognized on the sole basis of the inadequacy of all the
arguments that are hitherto made to prove the Xinshu spurious.
The edition of the Xinshu that I use is primarily the one com-
piled by Yan Zhenyi 閻振益 and Zhong Xia 鐘夏 (2000).
This edition is based upon the Jifu 吉府 (1515 AD) edition,
which is a Ming-dynasty edition and whose source, according
to Yan and Zhong (2000: p. 5), can be traced far back into pre-
modern times because of some identified features of pre-Song
scripts. Other editions that I use include the Siku quanshu edi-
tion and Lu Wenchao’s Baojing Tang抱經堂 edition. Sub-
sumed under this latter edition are a reproduction of the Baojing
Tang edition in 1937 by Shangwu Yinshuguan, a modern edi-
tion by Wu Yun 吳雲 and Li Chuntai 李春臺 (1989), and
2See Luo Shaodan (2002). Svarverud devotes 61 pages to examination ofthe
salient linguistic features in the extant version of the Xinshu and thereby
concludes that the Xinshu is auth entic. Assuming the ignor ance of medi eval
literati—particularly medieval text forgers—about early Chinese, he applies
to the Xinshu the linguis tic criteri a developed by Bernard Karlg ren for id en-
tifying early Chinese texts. Without challenging Svarverud’s view regarding
the authenticity of the Xinshu, Luo cautions that, since the literary language
in the Han dynasty constitutes a transition from early to medieval Chinese,
Karlgrenean method is inadequate in textual studies on documents ascribed
to the Han dynasty. Luo finds out that many salient features that Svarverud
identified in the language of the extant Xinshu can be found in Han Yu’s 韓
愈 (768-824) essays.
Copyright © 2013 SciRes. 9
another modern edition by Wang Zhouming and Xu Chao 徐
超 (1996)3. Fang Xiangdong’s 方向東 (2000) modern edition,
which collates several editions, has also been accorded close
Xinshu or Not Xinshu?—Two Criteria
As observed above, the Xinshu is said to have been forged on
the basis of the citations of Jia Yi in the Hanshu. But in addi-
tion to the Xinshu, there have been other monographs attributed
to Jia Yi. Some are indeed composed of the quotes of Jia Yi in
the Hanshu. We thus need to make certain we are not discuss-
ing one of those other books when we actually mean to exam-
ine the Xinshu. In the face value, the confusion seems unlikely
to occur, for different books are differently titled. However,
title turns out to be an inadequate means of identification, be-
cause “Xinshu” was not an original title of that corpus of essays
attributed to Jia Yi. This title occurred as late as the 6th century
(Wang Zhouming, 1982: p. 17). It is thus necessary to examine
the identity of what came to be the extant Xinshu.
Regarding this issue, Wang Zhouming holds a rather broad
view. Having traced the record of Jia Yi’s works in the biblio-
graphical treatises of dynastic histories as well as such Song-
dynasty bibliographical catalogs as the Chongwen Zongmu 崇
文總目 and Zhongxing Guan’ge Shumu 中興館閣書目, Wang
The above records indicate that the work of Jia Yi was
transmitted all along from the Han Dynasty to the Song
Dynasty. [Its] title varied during the transmission: [It] was
known sometimes as the Jia Yi, sometimes as the Jiazi,
sometimes as the Jia Yi ji, and sometimes as the Jia Yi
xinshu. What [also] varied was the textual division: [It]
could be in 2 juan, 4 juan, 9 juan, 10 juan, or 19 juan.
And there lacked a uniformed way of categorizing it. [The
monograph] was categorically treated sometimes as [a
text of] Confucian School and sometimes as [one of] the
Miscel- laneous School. 上述著錄說明，賈誼的作品，
歸屬亦不同，或歸儒家，或歸雜家 (Wang Zhouming,
1982: p. 18).
Apparently, Wang implies here that the monograph under
discussion could also be known as the Xinshu; because, in his
citation of the Zhongxing guange shumu, “Xinshu” was the
adopted title. Wang (1982: pp. 19-20) later suggested that the
extant Xinshu and the “ancient editions” (guben 古本) of the
book came from the same source. So it is obvious that, along
this long chain of identifications, if any of the monographs is
found to be problematic, the Xinshu will be automatically
subjected to suspicion.
Wang Zhouming’s chain of identifications noticeably in-
cludes some text corpus categorized under literary sections in
the bibliographical treatises of dynastic histories. Wang (1982:
pp. 17-18), for instance, includes a 4-juan Jia Yi ji, a 2-juan
Qian Han Jia Yi Ji, and a 2-juan Jia Yi Ji in, respectively,
Suishu 35, Jiu Tangshu 47, and Xin Tangshu 60. These mono-
graphs are all in the category of literary works. My study shows
that although the Xinshu circulated under different titles, Jia
Yi’s literary works were not included except as an appendix in
some cases. As Svarverud observes,
[T]he texts attributed to Jia Yi consisted of two main
bodies of texts: The memorials written by Jia Yi himself
as suggestions and grievances on the current situations in
the empire of Western Han times; and the more philoso-
phical and cosmological texts recording the words of Jia
Yi based on his teachings and speeches conducted by
himself, or his disciples and relatives shortly after his
death-most probably both (Svarerud, 1998: p. 8).
Except in the Songshi 宋史, where a 10-juan Jia Yi xinshu is
categorized as a text of Miscellaneous School, the collection
that Svarverud observed is invariably categorized in the section
of rujia 儒家 in both dy nastic histories and other bibliographical
sources. Examples include a 58-pian Jia Yi in the Hanshu, a 10-
juan Jia Zi in the Suishu, a 9-juan Jia Zi in the Jiu Tangshu, a
10-juan Jia Yi Xinshu in the Xin Tangshu, a 19-juan Jiazi in the
Chongwen zongmu, and a 10-juan Xinshu in the Junzhai dushu
In his bibliographical catalog, the Yuhai 玉海, Wang
Yinglin 王應麟 (1223-1296) regarded, firstly, the three Jia Yi
books recorded in those dynastic histories and, secondly, the
Xinshu recorded in the Zhongxing guange shumu as texts of one
and the same tradition. He provided a list of all the essays in the
version of the Xinshu in his time, which overwhelmingly
corresponds to the table of contents of the extant Xinshu. On
that list, there are no works of literature. Moreover, we see on
the list a considerable amount of texts not found in the Hanshu.
A text of this tradition is exemplified by Lu Wenchao’s edition
of the Xinshu, which consists of 58 philosophical and political
essays attributed to Jia Yi. But among the 58 essays, two are
completely missing except for their titles. The 58 pieces are
divided into such categories as shishi 事勢, lianyu 連語, and
zashi 雜事. Some editions, such as Lu Wenchao’s, include Jia
Yi’s Hanshu biography.
If anyone either suggests a different line of textual transmis-
sion or introduces a different monograph, s/he needs to first
define his/her position to the tradition identified by Wang
Yinglin. Otherwise, s/he may risk causing confusion in school-
arship by introducing other books into the tradition without
giving a notice. And that may lead to the error of discussing the
authenticity issues of a wrong book.
Since Wang Yinglin identified the tradition without suggest-
ing any exhaustive list of books, caution should be taken when
one attempts to extend the list by adding books. Although dif-
ference in title is of little importance in this case, it must be
noted that any book we add should meet the following two
criteria reflected in Wang Yinglin’s bibliographical notes and
table of contents. First, the book attributed to Jia Yi should be
categorized in a non-literary section in pre-modern bibli-
ographical sources. Specifically it should fall into the section of
philosophical works—be it “Confucian School” or “Miscella-
neous School. What this criterion entails is the absence of liter-
ary works except, as occasionally is the case, in an appendix.
Sec- ond, the contents of the book, if known, should not be
limited to the memorials of Jia Yi quoted in the Hanshu. It
would be desirable that the book we find meets both criteria,
because the tradition that Wang Yinglin identified would repu-
3Both teams of modern scho lars declare that they h ave largely kept the t ext
of Baojing Tang 抱經堂 edition intact in their editions (see Wu & Li, 1980:
p. 358; Wang & Xu, 1996: p. 4).
Copyright © 2013 SciRes.
diate books that meet either criterion alone. Therefore, once we
add a book that fails to meet both criteria, efforts must be made
to explain why we still consider the book affiliated with the
extant Xinshu. This simple rule can save us from wasting our
time on a book that we never mean to discuss.
Below we will examine how Jia Yi’s literary works are re-
lated to the Xinshu.
Jia Yi’s Literary Works and the Xinshu
Besides the collection of Jia Yi’s works categorized in the
section of philosophy, there are other collections under Jia’s
name but categorized in such literary sections as fu-poetry (e.g.
Hanshu 30), ji 集 (e.g. Suishu 35) and bieji 別集 (as in both
Tangshu). We consider those books outside the tradition of the
Xinshu because they are categorized separately from the books
in the tradition identified by Wang Yinglin.
But, since Jia Yi stands out in Chinese history as both a po-
litical thinker and a literary figure, could it be that, in each of
those dynastic histories, a single text corpus ascribed to Jia Yi
was twice entered with each entry under a different title plus a
different catalog and divided into a different number of juan?
There can be no definite answer to this question. This is a mat-
ter of likelihood.
The practice of multi categorization, to start with, would be
bibliographically confusing. It would seem no more advisable
to us than it did to pre-modern historians. In Hanshu 30, Ban
Gu 班固 (32-92) is found to be meticulous in this regard.
Hanshu 30 was based on the Qilue 七略 (cf. Hanshu, p. 1701;
Suishu, pp. 905-906), a lost bibliography compiled by Liu Xin
劉歆 (fl. 53-23 BC). In Hanshu 30, Ban Gu left some
unequivocal notes where categories in the Qilue were altered4.
Although we are not certain exactly how systematic his
categorization and notes may be, oftentimes what seems to be
double categorization is nonetheless conventionally treated as
an indicator of different bibliographical identities5. It is thus
unjustifiable to treat one monograph in the category of
philosophy as identical with another in the section of literature
just because they are ascribed to the same author, hence our
The first criterion, to recall, also entails the absence of Jia
Yi’s literary works except in an appendix. In Hanshu 30, there
are a 58-pian Jia Yi categorized under rujia and a 7-pian Jia Yi
under the fu-poetry6. None of the fu-poems is included in Lu
Wenchao’s edition of the Xinshu. And, in a note on Jia Yi’s
works in Hanshu 30, Wang Yinglin wrote, “[Under the cate-
gory of] rujia [in] the [bibliographical] treatise [of the Hanshu,
there is a] 58-pian Jia Yi. Besides, there are seven pian of
fu-poems.” 志：儒家：賈誼五十八篇。又賦七篇 (Wang
Yinglin, 1987: Juan 55, p. 3a). Wang Yinglin’s observation was
echoed almost verbatim by Zhao Ximing 趙曦明 (1705-1787)
(1937: p. 86), who put the following note about the records of
Jia Yi’s works in Hanshu 30, “[in] the bibliographical treatise
[of the Hanshu], there are 58 pian [under] rujia. Besides, there
are seven pian of th e fu-poems.” 藝文志：儒家：賈誼五十八
篇。又賦七篇. Their use of the word “besides” 又 is obviously
of great significance; because the word indicates their con-
scious categorical separation of the seven pian fu-poems from
the 58-pian Jia Yi7.
We have mentioned that certain versions of the Xinshu might
include Jia Yi’s literary works in an appendix. But such ver-
sions are normally considered as “other versions” 別本. He
Mengchun 何孟春 (1474-1536) once provided an annotated
table of contents for his edition of the Xinshu in the Ming Dy-
nasty (see Yan & Zhong, 2000: pp. 492-493). The last volume,
according to the table, contains five fu-poems of Jia Yi. It was
He Mengchun himself that referred to this edition as a bieben
(Ibid, p. 492). Qi Yuzhang 祁玉章 (1969: p. 45) further noted
that He Mengchun’s edition differed from other editions in the
5In Hanshu 30, a 29-pian Shangjun 商君 and a 27-pian Gongsun Yang 公
孫鞅 are separately categorized though both were attributed to the pre-Qin
statesman Shang Yang 商鞅 (cf. Hanshu, p. 1735, 1757). The same holds
true with a 2-pian Pang Xuan 龐煖 and a 30-pian Pang Xuan, both o
which are apparently ascribed to the pre-Qin military commander and theo-
rist Pang Xuan (cf. Hanshu, p. 1739, 1757) In these two cases, Ban Gu did
not put any note regarding how each book is related to its suspected coun-
terpart in a different category. It is to be noted that cases like this are rare.
Moreover, what the categorical distribution of Li Kui’s 李悝 works exem-
plifies is a case in which books attributed to the same author, once catego-
rized sep arately, are usually treat ed as differ ent books rat her than simpl y as
different versions of a same book. In Hanshu 30, there are a 7- pian Li Ke
李克, a 32-pian Li Kui, and a 10-pian Lizi 李子 respectively under the
categories of rujia, fajia 法家, and bing quanmou 兵權謀. All these books
are ascrib ed to the pre-Qi n stat esman Li Kui, t hough al l had b een los t by the
time of Suishu 35. Yet, because these books were differently categorized in
anshu 30, they are considered as three different books (see Gu Shi, 1987:
pp. 99, 133, 194; Zhang Shunhui, 1990, p. 238).
6Yao Minghui 姚明煇 (fl. 1914) provided specific titles for five of the
seven fu-poems, such as “Diao Qu Yuan fu” 弔屈原賦, “Funiao fu” 鵩鳥
賦, etc. (see Yao Minghui, 1933: p. 146). There are only two pian that Yao
did not find. But there is a “Diao Xiang fu” 弔湘賦 recorded by Chen
Zhensun 陳振孫 (1983:vol. 674, p. 694a) in the Song Dynasty. It is not
certain whether “Diao Xiang fu” is an alternative title for “Diao Qu Yuan
7Since the mai n-body tex t of th e Xinshu normally begins with Jia Yi’s “Guo
Qin lun” 過秦論, one may wonder whether this exemplifiesa case where
the Xinshu includes a work of literature. This question derivesfrom the fact
that, in pre-modern times, “Guo Qin lun” was occasionally considered as a
work of the fu 賦 poetry (cf. Qian Zhongshu, 1979: p. 888). The fact is,
however, “ Guo Qin lun” i s not exactly a fu-
oem. Rather it is merely a case
where “Mr. Jia [Yi] presents [a piece of] exposition that resembles a fu
[poem].” 賈生作論而似賦 (Ibid. p. 891). I think a parallel case could be
found in Yang Xiong’s 揚雄 (53 BC-18 AD) essay “Jiechao” 解嘲. “As
an essay featuring parallelism,” observed Yang Shuda 楊樹達 (1984: p.
677), “certainly ‘Jiechao’ came to be known as a fu-poem among later gen-
erations [after Yang Xi ong and Ban Gu].” “解嘲”行文偶儷，稱賦自是後世
之事.” Such practice of posterior labeling speaks of an example in which
scholars, from hindsight, “applied the terminology of later-generation liter-
ary stylists to [their] discussion of Ban [Gu’s] book.” 用後世文章家文法說
班書 (Yang S huda, 1984: p. 677).Hence it would be ha rd ly correct to assert
that “Guo Qin lun” and “Jiechao” were meant to be fu-
oems at the time
when they were composed.
4For instance, beneath the category of zhuzi 諸子, there is a note saying
“with a 25-pian Cuju removed [from here].” 出蹴鞠一家，二十五篇
(Hanshu, p. 1745), while beneath the sub category of Jiqiao 技巧,he added
the following note, “with a redundant Mozi omitted to enter a Cuju [in its
place].” 省墨子重，入蹴鞠也 (Ibid. p. 1762). The same kind of mutually
referential notes can also be found between the sub categories of Quanmou
權謀 (Ibid. p. 1757) and Li 禮 (p. 1710) for the book Sima Fa 司馬法.
According to Yan Shigu’s 顏師古 (581-645) note to Hanshu30, words
such as “ removing” 出 and “entering” 入 in Ban Gu’s notes indicate Ban
Gu’s modification of certain bibliographical categories in the Qilue (p.
1706). There are cases where Ban Gu removed certain books from a cate-
gory because he found them double categorized in the Qilue; though both
Qilue categori es made sense. His treatment of “a redundant Mozi” may serve
as a good example. Since the Mozi had already been put under the category
of Mojia 墨家 (see p. 1738), Ban Gu removed its other entry under the sub
category of Jiqiao in the section of Bing 兵, and wrote the note we just
quoted above, though the other category, Bing: Jiqiao, also makes sense
because of those chapters in the Mozi—specifically 11 chapters in Juan14
and Juan 15—that feature detailed di scussions of military techniques.
Copyright © 2013 SciRes. 11
way the texts were divided into volumes. Evidently, even in
this “other edition,” the section of Jia Yi’s literary works is
nonetheless kept outside the main-body text and is reduced to
an appendix8. This tradition seems to be well preserved in mod-
ern times as well. In their edition of the Xinshu, Yan Zhenyi
and Zhong Xia put Jia Yi’s fu-poems in an appendix, to which
the two modern scholars gave such a sub-title as “The fu-poems
and Scattered Texts Not Included in the Xinshu.” 新書未收文
賦及佚文. This title confirms our view that Jia Yi’s literary
works are not included in the main-body text of the Xinshu.
The Xinshu and the Quotes of Jia Yi in the Hanshu:
The Second Criterion
In Xinshu study, scholars unanimously consider Chen Zhen-
sun 陳振孫 (1179-1262) as the first one who claimed to have
found the Xinshu spurious9. Those who question the authen-
ticity of the Xinshu would work hard to substantiate Chen
Zhensun’s view. And those who trust the Xinshu would often
begin their discussion by refuting Chen Zhensun. But a perusal
of Chen Zhensun’s annotation would lead us to discover that
the 11-juan Jiazi that he examined failed to meet at least one of
our criteria. We thus suspect that this Jiazi, regardless how
spurious it might be, had nothing to do with the Xinshu. Let us
quote Chen Zhensun in full and take a close look.
The Jiazi [in] 11 juan. Note: According to the Chongwen
Zongmu, the bibliographical treatises of the Suishu and
the [Jiu] Tangshu both have [Jia Yi’s book] in nine juan.
The treatise of the Xin Tangshu has it in 10-juan. This
edition is [in] 11 juan [and hence] suspicious of error. [Its
ascribed] author is Jia Yi from Luoyang, [who served as]
grand tutor of King of Changsha in the Han Dynasty. [The
compilation consists of] 58 pian [according to] the biblio-
graphical treatise of the Hanshu. This book begins with
“Guo Qin lun” and ends with “Diao Xiang fu.” The rest
[of it] is all composed of excerpts from the Hanshu.
Moreover, in Volume 11, [there is a] “Biography of Jia
Yi” abridged [from the Hanshu]. Those parts that are ab-
sent from the Hanshu are invariably shallow, motley, and
[hence] unworthy of any attention. [Therefore the book]
could be anything but Jia Yi’s original work. 賈子十一
Zhensun, 1983: Vol. 674, p. 694a).
Based on Wang Yinglin’s notes and table of contents, our
second criterion stipulates that in no case can there be an edi-
tion of the Xinshu with its contents limited to the memorials of
Jia Yi in the Hanshu. From Chen Zhensun’s account, we learn
that the book he witnessed featured a main-body text that “be-
gins with ‘Guo Qin lun’ and ends with ‘Diao Xiang fu’.” In
addition, there was an abridged biography in Volume 11 that
was included presumably as an appendix. The rest of the book,
as we are told, “is all [emphasis added] composed of excerpts
from the Hanshu” with only a few occasional exceptions. Then
how can this 11-juan Jiazi be an edition of the Xinshu? How
can this book possibly be affiliated with the tradition that Wang
Yinglin identified? Chen Zhensun was certainly right when he
judged the 11-juan Jiazi to be a pseudepigrapha. It is us who
have misread Chen Zhensun and resultantly formulated an
erroneous impression that by denouncing the 11-juan Jiazi,
Chen Zhensun was questioning the authenticity of the Xinshu.
Textual Division and the Textual
Corruption of the Xinshu
The Xinshu is known to consist of 58 chapters. But none is
mentioned in early bibliographies. Wang Yinglin’s Yuhai is the
first known source that has listed the 58 titles. Since the Xinshu
is a collection of essays, each essay constitutes a titled chapter.
There are two titles whose subordinate texts were lost early.
Fifty-eight is also the number for chapters recorded in the Siku
tiyao. The Siku compilers indicated that only 55 chapters had
survived the long process of transmission from the Song
In the late 1990’s, Rune Svarverud (1998: pp. 8-11) listed
those 58 chapters in his book—including the two titles with
missing texts—along with their distributions among the ten
juan of the extant Xinshu. This number for the chapters matches
the number recorded in Hanshu 30 (cf. Hanshu, p. 1726).
Physical Condition of the Text
Among the received 50 some chapters, some do not look like
chapters. And there are numerous corrupted passages, which
gives rise to quite a few cases of contextual mismatch and gaps
in logic (Chen Weiliang, 1958: pp. 4-5). Yet such an imperfect
version, as we know now, is the remainder of several re-edi-
tions and reproductions over time, especially those in the Tang
and Song times (Svarverud, 1998: p. 33).
As a concrete example, in a chapter entitled “Qin shu wei
luan” 親疏危亂, we encounter a rather abrupt beginning like
There is something that Your Majesty would not do now.
[I,] Your subject [,] would not dare present [anything] less
than the entire truth [regarding the current] situation. [Let
us] suppose that the sub-celestial terrain were the same as
an earlier time when] Marquis of Huaiyin stayed en-
feoffed at Chu, Qing Bu stayed enfeoffed at Huainan,
Peng Yue stayed enfeoffed at Liang, Han Xin stayed en-
feoffed at Han, Zhang Ao stayed enfeoffed at Zhao, Guan
Gao stayed [in his position] as a minister, and Lu Wan
stayed enfeoffed at Yan; [let us further suppose that] Chen
Xi were still in Dai. [In a word, let us] suppose that these
few lords were all well and alive, remaining at their indi-
vidual fiefs. Had Your Majesty ascended to the throne at a
moment like that, could Your Majesty consider Yourself
safe? [I,] Your subject, know as a fact that Your Majesty
could not. 陛下有所不為矣，臣不敢不畢陳事制。假
8According to Chao Gongwu 晁公武 (fl. 1151-1161), there are also “cer-
tain [versions of the Xinshu that] incl ude [Jia] Yi’s biog raphy in Hanshu as
an appendi x.” 或取漢書賈誼傳附於後 (Chao Gongwu, 1983: p. 214b).
9Dates of Chen Zhensun’s life span are provided according to He Guan-
gyan’s 何廣棪 (2001) research.
10Compiled on the basis of Lu Wenchao’s edition, the extant Xinshu com-
rises 56 or 55 chapters, depending on whether the “Guo Qin lun No. 2” and
“Guo Qin lu n No. 3” are combined.
Copyright © 2013 SciRes.
下之不能也 (Jia Yi, 1937: pian 23, p. 32).
Wu Yun and Li Chuntai share Lu Wenchao’s view that the
first sentence makes a rather odd beginning11. The Siku compil-
ers, Chen Weiliang, and Huang Yunmei all take this kind of
textual corruption to be indicative of a case where the Xinshu
was forged by first splitting up the quotes of Jia Yi in the
Hanshu and then artificially imposing a title on each split passage.
To the Siku compilers, such corrupted condition seemed to
suggest that the Xinshu was forged on the basis of a number of
lengthy quotes of Jia Yi in other early texts, especially the
Hanshu and the Da Dai liji 大戴禮記.
This view came to be widely accepted in the 20th century.
Cai Tingji蔡廷吉 (1984: p. 27), for one, observed that, since
the Song times at the latest, it had been a common practice
among pre-modern editors of the Xinshu to make up a total of
58 chapters so as to match this record in Hanshu 30. Huang
Yunmei (1980: p. 263) and, especially, Chen Weililang (1958:
pp. 4-5) are arguably the most prominent advocators of the
view expressed by the Siku compilers. They both suggested that
the Xinshu was artificially created by collecting, altering, and
finally imposing titles on, the various quotes of Jia Yi’s texts in
Without subscribing to this skepticism, Svarverud (1998: p. 8,
2000: p. 13) suggested that certain chapters might have each
been deliberately divided up into 2 or 3 passages in order to
make up the number of 58. He rightly observed that the relation
was “very hard to establish now” between the 55 or 56 chapters
in the extant Xinshu and the unidentified 58 pian mentioned in
Hanshu 30. Such disorder in the Xinshu significantly adds
complexity and would hence easily arouse doubts. As Gardner
[T]he textual division of a Chinese—as of a Western—
book is normally dictated by the development of the sub-
ject. Any alteration in such division must ordinarily re-
flect either omission of some part of the original text, ad-
dition to it, or a complete recasting of the former treat-
ment of the subject … It follows that … alteration in the
number of pien, sections, in an ancient text, or of juan,
now chapters, in a modern one, both of which are subject
divisions, is … a matter of grave concern … If, then, we
detect a variation in the textual division of a Chinese work
in the history of its transmission, it behooves us to inquire
carefully for an explanation (Gardner, 1961: pp. 41-44).
But forgery can hardly be the only explanation. One factor
could be the physical damage the text suffered. Skeptics of the
Xinshu have never convincingly ruled this out. Take the Xinshu
text we just quoted. Its abrupt beginning and short length make
it look more like a miss-located passage than a freestanding text.
The earliest printed editions of the Xinshu were, like many
other pre-modern texts, woodblock editions in the Song dynasty
(Qi Yuzhang, 1969: p. 44; Wang Zhouming, 1982: p. 17; Cai
Tingji, 1984: pp. 53-54). Chinese texts previous to woodblock
printing existed in large amount as scripts on bamboo slips or
silk. In cases where a long text was inscribed on several sepa-
rate pieces of slips or sheets of silk, any accidental derangement
of the chunks and pieces could result in serious textual disorder
(Yu Jiaxi, 1958: pp. 546-547).
Besides, late Professor Zhang Shunhui 張舜徽 (1962: pp.
28-29) pointed out that, after the emergence of wood-engrave
printing, book producers would make a point of ensuring an
economical use of engraving materials at the expense of the
physical layout of ancient editions. Zhu Taiyan 朱太岩 (1989:
p. 47) also noted that, in the age of woodblock printing, book
producers tended to seek to create new, and hence change the
old, text formats for the mere purpose of making their products
look appealing to their customers. The text format we see in
woodblock printing is therefore not always a faithful replica of
an early edition. Chen Weiliang is right in noting that some
chapters look fragmented. But, to consider the fragmentation
indicative of spuriousness is to over stretch the significance of
the evidence12. Skeptics of the Xinshu never adequately expli-
cated to what extent textual fragmentation could be considered
evidence of forgery.
But, if it is merely a passage rather than a chapter and if the
title is not something artificially imposed by a forger, why else
is there a title for this passage?
Embedded Passage Titles and Its Possible Application
in the Xinshu
The answer can be found in what we may call passage titles
or embedded end titles found in bamboo and silk texts up to the
Han times, if not later. As passage titles instead of chapter titles,
they were positioned at the end of each passage. I will argue
that this format and the change it underwent can adequately
account for both the peculiar textual division and the textual
corruption in the received Xinshu.
The Han edition of the Yili 儀禮 suggests a format that may
confuse an inexperienced medieval or modern reader13. Since
this text is a Han edition, one cannot rule out the possibility this
might have been one of the formats adopted in the early collec-
tion of Jia Yi’s works.
Some pre-modern scholars already noted that, in early Chi-
nese texts, there were often subtitles assigned to individual
passages within a chapter (Zhang Shunhui, 1962: pp. 35-37). In
the 1940s, Professor Zhang Shunhui was among the first few
modern scholars who provided comprehensive accounts of this
format, observing in particular that the format could be found in
early texts such as the Xunzi and the Lüshi chunqiu14 (Both
texts are noticeably pre-Qin texts compiled during the Western
Han). Later on, there were more scholars following their steps.
Knoblock (1988: Vol. 1, p. 113), for one, noted that “the Lü
edition (of the Xunzi) divided the text of each book into para-
12Much as Chen Weiliang maintained that the textual problems could not
have been the result of scattered bamboo slips with rotten strings or of a
delinquent script copier, he did not provide any detailed examples to support
his point. Chen has provided two detailed charts to compare between the
inshu and other do cuments, especiall y the Hanshu. But, with those charts,
Chen makes no attempt to examine the fragmentation found in the Xinshu.
(see Chen Weiliang, 1958: pp. 4, 6-21).
13It is a format in which all the chapter numbers and chapter titles were
inscribed on the backside of each chapter’s first [bamboo] slip. By that we
mean on th e reverse side of each ch apter’s first line. Long chap ters of over
dozens or one hundred slips were numbered on the back of the first slip
whereas their titles were scribed on the back of the second slip (Chen Meng-
ia, 1980: pp. 301-302).
14For instance, Professor Zhang Shunhui (1990: p. 184) once observed that
the Lüshi chunqiu 呂氏春秋, another text from the State of Qin, was the
first early Chinese text that was known to have adopted this format. From
Professor Zhang’s preface, we learn that the source where he made this
observation was ori
inated from his manuscri
t in 1946.
11See Wu & Li, 1989: p. 104. Lu Wenchao’s opinion can be found in Lu’s
note in Jia Yi, 1937: p. 32.
Copyright © 2013 SciRes. 13
graphs covering single topics. Some of these paragraphs con-
tain embedded titles.”
Embedded end titles can also be found in excavated texts. In
the Shuihudi Qinmu zhujian 睡虎地秦墓竹簡 (4th?-3rd? BC),
end titles like “chuli lü” 除吏律, “chu dizi lü” 除弟子律,
“Wei hulü” 魏戶律 , and “Wei benming lü” 魏奔命律, etc. all
suggest pre-Han use of such titles than in the Han dynasty15.
Other examples, as Yates noted, can be found in the Huang-Lao
texts excavated at Mawangdui in mid-southern China (Yates,
1997: pp. 197-198). Since the Shuihudi and Mawangdui texts
are respectively Qin and Han scripts, it is safe to consider the
use of such titles as a common practice during the Qin-Han
period at the latest16. In some early texts, titles and subtitles are
found respectively at the end of an essay and the passages or
paragraphs in the essay (Wang Liqi, 1999: pp. 265-267). In
others, there can be an essay title that covers each essay while
there are subtitles placed at the end of each passage (Jiang Bo-
qian, 1990: pp. 1 19-120). It is not known whether the titles were
provided by the authors or imposed by the compilers of the
texts. But one thing is certain: Even if they proved to be poste-
rior impositions, it would not mean the text itself is spurious17.
These titles were, admittedly, not well preserved in later edi-
tions of early texts. And this format could easily mislead those
who reproduced early texts in pre-modern times, resulting in
their erroneous presentation of early texts. Some scholars (e.g.
Gu Yanwu, 1990: pp. 908-909; Zhang Shunhui, 1962: p. 35)
imputed the mishandling of these titles to the medieval and later
editorial confusions that mistook those embedded end titles for
part of the main-body text.
If an editorial note by Lu Wenchao is right, we may say such
editorial mistake can indeed be detected in an essay entitled
“Wumei” 五美 in the Xinshu. In Lu’s edition, this chapter
ends with a sentence that reads, “Your Majesty’s fear of what
has, for a long time, kept Your Majesty from implementing
these five desirables?” 陛下誰憚而久不為此五美? While
leaving the sentence intact, Lu Wenchao put the following note
immediately beneath the sentence: “The last two graphs [i.e.
“五美”, meaning “five desirables”] must be a title of the pre-
ceding passage” 末二字當目上文. In Lu’s judgment, therefore,
the two graphs 五美 at the very end were originally meant to
be a title rather than part of the last sentence of the passage.
Hence according to Lu’s reading, the last sentence should be:
“Your Majesty’s fear of what has, for a long time, kept Your
Majesty from implementing this?” 陛下誰憚而久不為此?
The extant Xinshu, as is indicated in our Introduction, can be
traced back to some Song editions that eventually survived in
Lu Wenchao’s edition and Lu’s notes therein. It can be inferred
from Lu’s note that the last two graphs, “wumei,” had already
been treated as part of the main-body text in the Song editions
that Lu witnessed. Yet the sentence, just by itself, would be
grammatica lly and idiomat ically correct rega rdless whether this
disyllabic word were added or not. Besides, in Lu Wenchao’s
edition, the same two graphs were also used in the beginning of
the passage as a chapter title. Difference in opinion has thus
occurred regarding where this disyllabic word belonged.
In the past, there have been modern annotators who exempli-
fied the difference in their annotated editions of the Xinshu. In
their 1989 edition, Wu Yun and Li Chuntai mixed the disyllabic
word into the main-body text. Wang Zhouming and Xu Chao
(1996: p. 62), in contrast, dropped the disyllabic word and then
added an editorial note, declaring that they had done so in ac-
cordance with two editions in the Ming dynasty as well as the
quote of the text in Hanshu 48 (i.e. biography of Jia Yi). Fang
Xiangdong (2000, p. 89) quoted Lu Wenchao’s note without a
comment. The same year, Yan Zhenyi and Zhong Xia (2000: p.
70) remarked in their annotated edition of the Xinshu that since
there were no other occurrences of end titles in the extant Xin-
shu, they judged these two graphs to be redundant graphs rather
than an embedded end title.
In this particular case, I suggest basing our judgment primar-
ily on the text itself instead of on either the Hanshu or any post-
Song editions of the Xinshu. Although neither the presence nor
the absence of the two graphs would result in a grammatically
incorrect sentence, a scrutiny of the text will lead us to the con-
clusion that only the absence of those graphs is contextually ap-
This passage was composed to advise the emperor to institute
a system of dizhi 地制 (enfeoffment). Instituting the system
would, as the author argued, make the emperor’s virtue known
to the “sub-celestial world” in five aspects. Thus the “five de-
sirables” refer to the world’s five kinds of recognition of the
emperor’s virtue. Apparently, one can securely infer from this
argument that all the emperor was capable of was having the
five desirables “done” to his majesty himself. In other words,
the five desirables were not something that the emperor was
able to initiate directly. They were, rather, the five rewards for
the emperor to earn, the five things that the “sub-celestial
world” would do to his majesty as a response to and reward for
his successful establishment of that system. The object of the
verb “為”, therefore, has to be the enfeoffment system, not the
“five desirables”. And the pronoun “此”—meaning “this”
—should be taken as a singular demonstrative pronoun that
denotes the system of enfeoffment. Lu Wenchao is, therefore,
correct in considering the last two graphs external to the last
15In addition to these titles, there are also two end titles for individual trea-
tises which modern archaeologists were able to identify only after they had
soaked the bamboo texts in water for a long time (see Shuihudi Qinmu zhu-
ian, 1978: p. 2 , footnote 1) .
16Yet, sometimes some basic training in archaeology and Chinese textual
criticism is required for one to be able to identify the titles. Otherwise, one
could be under an erroneous impression that the use of titles in the currently
available Qin and Han scripts was very rare and exceptional. In the case o
Shuihudi texts, for example, there were two end titles which modern archae-
ologists were able to discern only after they had soaked the bamboo texts in
water for some time (Shuihudi Qinmu Zhujian, 1978: p. 2) It sh ouldbe noted
that the quantity of identified end titles in the Shuihudi scripts is signifi-
cantly above the level of being rare and exceptional. The occurrences of end
titles in the Mawangdui texts are also evident. There is, for instance, a title
“at the end of each essay in the [Mawangdui silk] manuscripts” that Yates
translates in his book. I concur with Yates that the use of such end titles “is
not the practice of early and middle Warring States texts” but one that “starts
to appear in the third century B.C.E.” (Yates, 1997: p. 197) Besides their
occurrences in excavated scripts, residues of end titles can also be found in
some post-Han editions of Han compilations of early texts, such as the Xunzi
and the Lüshi chunqiu (see also Yates, pp. 197-198; Zhang Shunhui, 1990:p
17According to Zhang Xuecheng 章學誠 (1738-1801) (1985: p. 395), chap-
ters of ancient texts usually remained untitled. Professor Zhang Shunhui
(1990: p. 184) also pointed out that the chapter titles of most pre-Qin texts
were posterior impositions. One may thus deduce that the same holds true
for passage titles.
As for whether the two graphs merely remain as a redundant
word or function as a title, there can be no definitive answer.
But there is no mutual cancelling between the two possibilities.
It was likely that the two graphs, “wumei”, remain in the text as
Copyright © 2013 SciRes.
both a redundant word and an embedded end title.
There were, to recall, pre-modern scholars and book compil-
ers who mistook embedded end titles for part of the main-body
text. There were also editors who would, by mistake or on pur-
pose, change former print-formats (Gu Yanwu, 1990: pp. 908-
909; Zhang Shunhui, 1962: p. 35 ; Zhu Taiyan, 1989: p. 47). The
Song editions of the Xinshu, though extant no more, had re-
mained to be “the sole sources for all later editions” in the form
of Lu Wenchao’s commentaries on them (Svarverud, 1998: p.
24). Cherniack observes that the Song dynasty was a time when
Chinese book culture favored change. Consequently, “the sup-
ports that had earlier served to stabilize the texts were weak-
ened, and canonical texts, like other texts, became open to tex-
tual innovation” (Cherniack, 1994: p. 21). Her comment is in
agreement with Zhu Taiyan’s (1989: p. 47) description of the
fashion during the Song dynasty among reproducers of ancient
Because the use of embedded end titles was popular in the
Han era and the modification of early formats was a common
practice during the Song times, it stands to reason to suspect
that there were originally other end titles in the Xinshu. But a
thorough removal of them would require both knowledge about
early formats and the ability to recognize such titles. In the
Xinshu, other end titles could be easily detected because it was
grammatically or idiomatically impossible to merge the titles
into the sentences that they individually followed. But the
presence of the embedded title “wumei” does not make its pre-
ceding sentence any less idiomatic than does its absence. The
title hence survived deliberate removals and remained as re-
dundant graphs in the text. Chen Weiliang discusses textual
corruption from the primary perspective of falsification. But
passage title is an important factor to consider in this matter.
The Siku Tiyao and Yao Nai
To discredit a historical document normally means to find it
anachronistic. As late as the Qing Dynasty, compilers of the
Siku Tiyao and—especially—Yao Nai both claimed to have
found concrete evidence of forgery in the Xinshu.
The Siku Tiyao
In their discussion of the Xinshu, the Siku compilers re-
marked that “the book is not entirely authentic, nor is it entirely
forged” 其書不全真，亦不全譌 (Siku quanshu zongmu tiyao,
juan 49). The evidence of its not being “entirely authentic,”
they declared, lied in the corrupted condition of the text and the
titles that seemed to have been artificially imposed on passages
(Ibid). These are the topics that we discussed in the previous
section. The reason they considered the Xinshu not “entirely
forged” was because they found that, among the parts of the
Xinshu with no textual parallels in the Hanshu, there were texts
rich in early teachings with their sources traceable in ancient
classics. The compilers thus differ from many modern skeptics
of the Xinshu in that they were not unwilling to give credit to
the parts of the Xinshu that lacked textual parallels in the
Hanshu, though the compilers’ judgment in this regard is rather
Yao Nai’s skepticism, by contrast, was directed to the entire
Xinshu. He considered the Xinshu questionable in textual and
historical aspects. In the textual aspect, the grounds of his skep-
ticism do not differ significantly from those of other skeptics
discussed in our previous sections. But his criticism in histori-
cal aspect deserves our particular attention.
In the “Dengqi” 等齊 Chapter of the Xinshu, the author said
he would “sigh over” the practice of using gold seals among the
ministers of enfeoffed lords. Yao Nai considered the “sighing”
as an anachronistic react to what had already come to be nor-
mal by Jia Yi’s time, when lords began to lose interest in hon-
oring codes of protocols appropriate to their individual statuses
and ranks. What is more, in the chapter of “Dengqi,” enfeoffed
lords are addressed as bixia 陛下. And the term fei 妃 (concu-
bine) is found to be an alternative title for hou 后 (queen). Yao
Nai argued that, firstly, bixia was a title that no lords in the Han
era could possibly adopt because the titled had been allocated
for the emperor alone in the Qin Dynasty. Secondly, to refer to
a queen as a fei was a practice that did not occur in history until
as late as the Wei-Jin period. So, its occurrence in the “Dengqi”
Chapter seemed to suggest that the Xinshu was forged in or
after the Wei-Jin times.
As regards the use of the title bixia, Wang Zhouming points
out that the purpose of the chapter was to denounce the practice
of addressing a lord as bixia (Wang Zhouming, 1982: p. 12).
The chapter was, in other words, taking issue with the lords’
disregard of protocol codes. As for referring to a queen as fei, it
should be noted that the original meaning of fei is “spouse”, not
“concubine” (Zhang Shunhui, 1983: juan 24, p. 5b). And ac-
cording to Wang Zhouming (1982: p. 21), the character 妃 in
this sense is pronounced pei instead of fei. Therefore, the sen-
tence “Tianzi zhi pei hao yue hou” 天子之妃號曰后 in
“Dengqi” simply means “the spouse of the Son of Heaven is
known as a Queen”.
Moreover, Yao Nai’s criticism of the chapter regarding the
“sighing over” the use of gold seal may instantly remind us of
what Wang Guowei 王國維 (1877-1927) found about the
“Dengqi” Chapter. The chapter mentions some official titles in
the courts of enffeoffed states. Wang Guowei had once judged
those titles to be particular only to the governance structure of
the central court. For this reason, he had long considered this
chapter untrustworthy until one day he witnessed a number of
clay seals that had been in use during the Han Dynasty in the
court of an enffeoffed state. Right on the seals, Wang noticed
precisely those titles that he had found in the “Dengqi” chapter.
Wang said he then realized that what was said in “Dengqi” was
true: It was true that enffeoffed lords during the Western Han
had lost interest in protocol codes and begun emulating the
emperor in the system of officialdom (Wang Guowei, 1973:
juan 18, p. 921).
But if this was a common practice among the lords, does it
mean Yao Nai was right in pointing out the “sighing” in the
“Dengqi” Chapter to be an anachronistic reaction to it? The key
issue here, I think, is not what practice was common and nor-
mal in that period, but rather what practice would look intoler-
able to the author of the chapter no matter how common, nor-
mal, or even fashionable the practice might have been in his
time. It was all a subjective matter on the part of the author. A
historical and objective perspective would thus be totally ir-
Chen Weiliang’s Criticism Reviewed
As a scholar in modern time, Chen Weiliang (1958) epito-
Copyright © 2013 SciRes. 15
mizes all skepticism of the Xinshu and contributes to signifi-
cantly deepening our study. His criticism—which examines the
Xinshu from literary, historical, and philosophical perspectives
(Nylan, 1992: p. 169)—strikes as more comprehensive and so-
phisticated than that of all his predecessors. Critics of the Xin-
shu today can hardly afford to ignore his study. Nylan (1992)
and, especially, Svarverud (1998) have both presented in-depth
reviews that did justice to Chen’s contributions. My review
below will discuss Chen’s criticism on the minben 民本thought
and what one may call senary-composite cosmology in the
Xinshu. In addition, I will discuss his arguments on the stylistic
issues in the Xinshu and how the Xinshu was related to other
texts, especially to the Shuoyuan 說苑 and Zuozhuan 左傳.
Five or Six
In the Xinshu, both the “Liushu” 六術 and “Daode shuo”
道德說 chapters argue that everything in the universe is
six-fold. According to Hanshu 48, however, Jia Yi suggested—
over 20 years after the founding of the Han Dynasty—that the
Han Empire establish yellow as the most important color,
change the calendar of his time, and replace the quinary cus-
tomary and ritual system with a senary system, in which, for
example, physical dimensions of utensils were to be set at six
lengths (i.e. chi 尺 and/or cun 寸) and the emperor would use
a six-horse chariot. Chen Weiliang (1958: p. 27) considered
such discrepancy between those chapters and Hanshu 48 sug-
gestive of a case of textual forgery on the part of the Xinshu.
The discrepancy is admittedly obvious. But I do not think it
substantiates an assumption of Jia Yi’s consistent insistence
upon the implementation of a quinary system. Viewed from a
perspective of Five-Phase 五行 theory in pre-modern China18,
succession of dynastic regimes would follow a cycle of destruc-
tion 相剋 in the sequence of Wood-Soil-Water-Fire-Metal-
Wood. The founding father of the Han Dynasty, Emperor
Gaozu (r. 206-195 BC), initially considered himself an agent of
Fire. Before long, he was convinced that his regime represented
the element of Water, whose corresponding numbers were, to
note, both One and Six. Later, during the reign of Emperor Wen
(r. 179-157 BC), Jia Yi and Gongsun Chen 公孫臣 (fl. 165
BC) successively petitioned to have Soil recognized as the
dynastic element and to honor both its color yellow and its
corresponding number Five (Hanshu, p. 1270)19. But it was not
until about 60 years later that the Han Empire adopted the
institutions they had suggested (Ibid, pp. 199-200).
Thus we know, the early period of the Western Han era un-
derwent a transition from upholding Water to upholding Soil.
The transition overlapped with Jia Yi’s lifetime and coincided
with the discrepancy that Chen Weiliang discovered.
In response to Chen Weiliang’s argument, Svarverud (1998:
pp. 129-134) has made a cogent point from the perspective of
Chinese intellectual history. He finds that numbers Five and Six
had been, by Jia Yi’s time, well integrated in early Chinese
philosophical and political thoughts, with the former number
“corresponding to the Five Elements, or Phase, in cosmos” and
the latter “as the Heavenly number describing the cosmic origin
of things” (Svarverud, 1998: p. 132). And he further contends
that Chen Weiliang’s “interpretation of an opposition between
the cosmological significance of five and six as evidence for the
non-reliability of these chapters in Xinshu is superficially ar-
There is ample evidence for a cosmology interpreting
Heaven as containing the essential six qi [氣] which in
turn give birth to different qualities in the world of things,
not [Svarverud’s italic] excluding but rather forming a
comprehensive cosmology incorporating the historical
cyclicity of the Five Elements in early Han cosmology.
Based on pre-Han and early Han cosmology the scheme
of all things origination in a six-fold division presented in
the chapters Liushu and Daodeshuo seems, on the con-
trary, to conform to expectations with regard to the writ-
ings of the Han eclectic Jia Yi (Svarverud, 1998: p. 133).
Svarverud’s point is argued through ample citations of early
sources. To the list of his examples, I would add the Guoyu 國
語, where the following cosmological theorization is found:
The Heaven is six-fold whereas the Earth five-fold. Such
are the numeric constants [in the cosmos]. The heavenly
[sextuple] function as longitudes while the earthly [quin-
tuple] as latitudes. The longitudes and latitudes [inter-
weave] with no error, hence the [formation of cosmic]
經緯不爽，文之象也 (Guoyu, 1978: juan 3, p. 98).
Wei Zhao’s 韋昭 (204-273) annotation elaborated on this
notion of cosmic pattern and substantiated the concepts of sex-
tuple and quintuple with the following remark.
In the heaven there are six qi, namely yin, yang, wind,
rain, darkness, and brightness. On earth there are five
elements. [They are] Metal, Wood, Water, Fire, and Soil.
With the six heavenly qi functioning as longitudes while
the five earthly elements as latitudes, [the cosmos] is
以地之五行為緯，而成之也 (Guoyu, 1978: juan 3, p.
Evidently, this passage in the Guoyu confirms Svarverud’s
point that numbers Six and Five were integrated in early Chi-
The Minben 民本 Thought
In such Xinshu chapters as “Dazheng No. 1” 大政上 and
“Dazheng No. 2” 大政下, Chen Weiliang identified “the po-
litical thought of treating-people-as-basis” 民本政治思想.
After comparing the themes of these two chapters with that in
Jia Yi’s quotes in the Hanshu, he judged the two chapters to be
artificial early writings. Central to this observation is apparently
his assumption that things absent in authorized dynastic histo-
ries are unreliable. In terms of the minben thought, however,
the present study does not find the two Xinshu chapters incon-
sistent with the “Guo Qin lun” quoted in the Shiji 史記, which
18Readers who do not read Chinese please consult Needham, et al. (1954-
1999: Vol . 5, part 3) for a det ailed acco unt of the Five- Phase th eory in early
19Without documenting his point, Chen Zhi 陳直 (1901-1980) (1979: p.
288) maintains that the quinary system that Jia Yi proposed applied only to
the number of characters in the official titles inscribed on the official seals.
By that, he means that Jia Yi suggested using five-character official seals to
replace their contemporary four-character seals. Chen Zhi also points out
that the five-character seals that Jia Yi suggested were not in use until Em-
peror Wu’s reign. I have found evidence indicating Emperor Wu’s adoption
of five-character seals in the Shiji (p. 1402) and Hanshu (p. 1245).
Copyright © 2013 SciRes.
was no less established as standard history than was the Hanshu.
Besides, I do not think that there was only one kind of minben
thought in early China.
The term minben is arguably a short form for the phrase “min
wei bang ben” 民惟邦本 (common populace is the foundation
of the state) in Chapter “Wuzi zhi ge” 五子之歌 in the old-
text Shangshu 尚書. According to the chronology suggested by
Yan Ruoqu 閻若璩 (1636-1704) (1796), the old-text Shang-
shu may not have predated Jia Yi’s time. But the “minben
thought” is nevertheless a convenient posterior label for the
kind of early persuasions that accorded importance to the
common populace. Mencius (7B: 14) is widely quoted to have
said “(in a state) the people are the most important; the spirits
of the land and grain (guardians of territory are the next; the
ruler is of slight importance” (Chan, 1963: p. 81). One would
agree with Chen Weiliang on the marked difference between
this type of minben thought and that conveyed in those two
Chen Weiliang’s (1958, p. 27) observation on the “spirit of
Legalism” (fajia jingshen 法家精神) in Jia Yi’s thought is well
founded. With very few exceptions (e.g. Yang Shuda, 1984:
Vol. 7, p. 479), scholars generally agree with Ban Gu that Jia
Yi “was well versed in (the thoughts of) Shen (Buhai) and Han
(Fei)” 明申韓 (Hanshu, p. 2723). Scholars in the Song Dy-
nas ty pai d particular attention to this aspect of Jia Yi. Ye Shi 葉
適 (1150-1223) once remarked that Jia Yi’s Warring-State type
of strategies was adorned in his use of Confucian principles
(see Qian Zhongshu, 1979: p. 893). Chen Liang 陳亮 (1143-
1194) (1987: pp. 126-128) regarded the “heterodox learning”
異端之學 of Legalism as an important complement to the
Confucianism in Jia Yi’s thought. In Northern Song times, Su
Shi 蘇軾 (1986: Vol. 1, p. 315) took for granted Chao Cuo’s
晁錯 (200-154 BC) “heretical ideas” of Legalism but felt sur-
prised at the same ideas manifested in Jia Yi’s teaching. To
Wang Fuzhi 王夫之 (1619-1692) (1996: Vol. 10, p. 104) in
the Ming-Qing period, Jia Yi “sounded scarcely different from
Li Si” 去李斯之言也無幾.
Given all these attestations, one would still expect Chen
Wailiang to explain why he had considered the minben thought
and the “spirit of Legalism” mutually exclusive. In Chapter 23
of the Guangzi 管子, it is said “[a] hegemonic or kingly domi-
nance would start from treating the common populace as a
foundation. The state will be stable [so long as] the foundation
is in order. [Should] the foundation become chaotic, the state
would be in jeopardy.” 夫霸王之所始也以民為本，本治則國
固，本亂則國危 20. Further reading of the chapter will take us to
a focused discussion on how to accomplish dominance and
ensure the triumph of one’s state in wars of annexation. Such a
topic is of course more pertinent to the will of a ruler than it
would be for the interest of the common populace. This obvi-
ously speaks of a certain convergence between Mencian min-
ben thought and Legalism.
Early Legalist thinkers lacked the chance to examine enough
of the formindable force of nationwide mass riots, such as the
untouchable’s general uprisings by the untouchables towards
the end of the Qin Dynasty21. They therefore variously empha-
sized what Mark E. Lewis calls “sanctioned violence” as their
answer to all socio-political situations22. Around the fall of the
Qin Empire, early-Han literati found in sanctioned violence
something that might arouse vehement reaction from the
masses. Such awareness was amply reflected in the political
discourse of the early Han period, including the speeches of Jia
Yi. In this regard, the minben thought detected by Chen Weili-
ang in the Xinshu does not seem to significantly deviate from
Jia Yi’s Legalist inclinations reflected in the Hanshu. I choose
to refer this kind of minben thought as practical minben
It is said in Chapter “Dazheng No. 1” that a ruler “cannot af-
ford not to fear the people” 民不可不畏也. This presumably
comes from the “Jiugao” 酒誥 chapter of the Shangshu, where
King Cheng of Zhou said that “previously, the sage king of the
Yin held [both] the Heavenly Sovereignty [and] petty com-
moners in awe” 在昔殷先哲王迪畏天顯小民. Thus, insofar as
they were the object of the sage king’s fear, “petty commoners”
were considered second only to the “Heavenly Sovereignty.”
The Xinshu chapter elaborates on this idea of fearing the people,
cautioning that a ruler, “therefore, cannot slight even the most
base, nor deceive even the most foolish, commoner among
general populace,” 故夫民者，至賤不可簡也；至愚不可欺也
because, “from ancient times to now, [it has always been the
case that] he who sets himself against the people will be sooner
or later overcome by the people”. 故自古至於今，與民為讎
者，有遲有速，而民必勝之. In the same chapter, a ruler’s
regime is said to be doomed once people loathed what the ruler
wanted them to do. Likewise, in the Shiji, we catch Jia Yi ar-
guing eloquently in his “Guo Qin lun” that, upon the loss of
people’s favor, the Qin Empire—a formerly invincible con-
queror of six well armed states—could not even withstand the
strike from a crowd of untouchables who were armed with only
wood clubs and hoes. This kind of reasoning found in an or-
thodox history is by no means incompatible or inconsistent with
the practical minben thought conveyed in the two Xinshu chap-
As is summarized in the Introduction, in late 1950s and early
1960s respectively, Chen Weiliang in Hong Kong and a team of
four scholars in Beijing compared the texts of the Xinshu and
that of Jia Yi’s quotes in the Hanshu. After comparing almost
21But, even i n that ear ly peri od, Xun zi 荀子 (4th-3rd cent. BC), the alleged
mentor of such major Legalist theorists as Han Fei 韓非 and Li Si 李斯,
was already a ware that the masses wer e to the r uler what wat er was to bo at.
By that he meant that, just as water was capable of both carrying and over-
turning a boat, so were the masses capable of upholding and overthrowing
their ruler (see Xunzi, 1936: Vol. 20, p. 100). In the Liji 禮記, incidentally,
it is also argued that a gentlemen cannot afford to overlook the general
populace an y more than one can af ford to overlook water, which is capab le
of taking human life if one is careless about it (Liji, 1983: Vol. 116, 412b).
Like the thought of Xunzi, the Laozi 老子 was also a rich source of influ-
ence to Legalism. Argument similar to the forgoing, for instance can also be
found in Chapter 39 of the Laozi.
22According to Lewis, sanctioned violence was “modes of inflicting harm o
taking life which men accept, approve, and even prescribe.” By that he
specifically refers to the imposition of the will of the ruling upon the ruled,
or that of one state’s will upon another, through institutionalized use of force
(Lewis, 1990: p. 1).
20In the recei ved edit ion of t he Guanzi, the first sen ten ce read s “ 夫霸王之所
始也以人為本.” According to Yan Changyao’s 顏昌嶢 (1868-1944)
(1996: p. 214, p. 219) annotation, the extant edition of the Guanzidates back
to the Tang Dynasty, when the graph “人” was supposed to be used in place
of “民” in order to avoid the taboo name of Li Shimin 李世民, Emperor
Taizong of the Tang Dynasty. I therefore restored the graph “民” in my
citation of the Guanzi here.
Copyright © 2013 SciRes. 17
the same set of texts, Chen Weiliang and the team of four
scholars has each reached an opposite conclusion to the other’s
argument. Whereas Chen Weiliang contended that the quotes of
Jia Yi in the Hanshu presented a smoother reading than did the
Xinshu23, the four scholars judged the latter to be smoother in
reading than the former. Obviously, different scholars’ personal
tastes for writing styles were at play in their textual criticism.
If Zhang Xuecheng’s 章學誠 (1738-1801) observation can
be accepted, we must take into account that there were times
when pre-modern historians edited and revised the historical
documents they quoted24 (Zhang Xuecheng, 1956: p. 65).
And one would question the assumption Chen Weiliang and
the four scholars shared despite their difference. It is an as-
sumption that idealizes Jia Yi’s writing style. Based on this
shared assumption, Chen and the four scholars in fact agree that,
between the Xinshu and Jia Yi’s quotes in the Hanshu, a bet-
ter-written text must be the original text. Chen Weiliang par-
ticularly shows an inclination to idealize Jia Yi’s writing. He,
for example, considered the “Daoshu” 道術 chapter unauthen-
tic because he found part of the essay “long-winded, meaning-
less, continuously garrulous, [and hence] unlikely to be what an
ambitious and patriotic Jia Yi would have cared to say.” 冗長
言也 (Chen Weiliang, 1958: p. 26). As a similar example, his
comment on two other Xinshu chapters is: “As a writer with a
natural and flowing style, Jia Yi could never have possibly
written such self-repetitive and awkward sentences. Such is the
shallowness and vulgarity of the forger.” 文章疏宕如賈誼者
亦決不會為此重複笨拙之句，此乃偽者之淺陋 (Ibid, p. 28).
In both cases, Chen Weiliang (1958: pp. 4-5, 26-27) considered
the textual and stylistic problems suggestive of a forger whose
writing skill was markedly inferior to that of Jia Yi.
Svarverud (1998: p. 81) rightly imputed some of the repeated
passages to the physical damage of the bamboo texts. Regard-
ing those problems that are more stylistic than physical in na-
ture, one would wonder whether a stylistically imperfect text is
necessarily a forged text. I choose to answer the question
through a discussion on critics’ reception of the first part of Jia
Yi’s “Guo Qin lun” (i.e. “Guo Qin lun No. 1”).
Although textual scholars disagree about the authorship of
the Xinshu, they all regard “Guo Qin lun No. 1” as an authentic.
Chen Weiliang largely relied on “Guo Qin lun No. 1” in his
discussion of Jia Yi’s thought. As a text with a clear line of
transmission, “Guo Qin lun No. 1” is quoted in full in the Shiji.
From there, according to Svarverud (1998: p. 48), Ban Gu
quoted it in his Hanshu. Later on, it was included in numerous
prestigious anthologies, of which Xiao Tong’s 蕭統 (501-531)
Wenxuan 文選, Yao Nai’s Guwen Cilei Zuan 古文辭類纂,
and Yan Kejun’s 嚴可均 (1762-1843) Quan Shanggu Sandai
Qin Han Sanguo Liuchao Wen 全上古三代秦漢三國六朝文
are only a few examples.
Yet “Guo Qin lun No. 1” has not always been favorably re-
ceived among critics. Scholars normally appreciate its vigorous
style. But they notice that it is a text that features sentences “too
[lengthy] to recite or read aloud” 難於諷誦 (Huang Kan, 1962:
pp. 145-146), redundant expressions, “unsymmetrical and un-
stable parallelism” 對偶偏枯杌隉, “piled-up phrases” 堆疊成
句, and “lavishness in words in excess of substantiality in
meaning” 詞肥義瘠 (Qian Zhongshu, 1979: p. 891). Could it
possibly be coincident that Chen Weiliang found in those Xin-
shu chapters exactly the same kind of stylistic problems as
those other modern scholars found in “Guo Qin lun No. 1?”
In fact, there were also pre-modern scholars under the same
impression about Jia Yi’s writing. Zhu Xi 朱熹 (1130-1200),
for one, caught Jia Yi “scribbling haphazardly all along on the
wings of [his] talent”. 只是乘才快，胡亂寫去 (see Qian
Zhongshu, 1979: p. 888). His remark echoes an earlier remark
by Su Shi, who found Jia Yi “opulent in talent while inadequate
in insights”. 才有餘而識不足 (Su Shi, 1986: Vol. 2, p. 777).
Chen Weiliang has certainly demonstrated incisive discern-
ment in identifying Jia Yi’s stylistic flaws. But let us not forget
that such flaws are shared by the texts that Chen considers un-
reliable and those that he trusts with no reservations.
The Xinshu vs. Other Early Texts
Chen Weiliang has also taken note of some textual similari-
ties between the Xinshu and three other texts25. He found the
texts in the Xinshu shorter than their textual counterparts in
those three books, and thus suspected the Xinshu of being partly
forged on the basis of those books because, had those three
books quoted the Xinshu, their texts would have been either of
same length as, or shorter than, the Xinshu texts26.
However, among the three text ual parallels between “Xiuzhe ng
yu No. 1” 修政語上 (Xinshu) and “Jundao” 君道 (Shuoyuan),
I find only one case in which the Xinshu text is shorter.
Svarverud must have noticed this as well. He observes that
Chen Wailiang has based his “argument on a limited number of
passages among all the parallel passages in these texts”
(Svarverud, 1998: p. 62).
As a matter of fact, to say that, between the two sources, the
longer text is the original text, we have to eliminate one more
possibility suggested by the four scholars in Peking University.
It is the possibility that both the Xinshu and non-Xinshu texts
were quoting an unidentified third source instead of each other
(see Sun Qinshan et al., 1961: p. 65). This is a possibility that
one can never rule out. But since the third source is still hardly
identifiable, and considering the large quantities of early texts
lost to time, let us base the discussion solely on what is known.
In the following, I will compare two texts between which the
Xinshu and non-Xinshu texts are equally readable but of sig-
nificantly different lengths. The point I wish to make is that,
even if we, for the sake of argument, rule out the possibility of
there being a third source, forgery on the part of the Xinshu is
still not the best explanation for a text parallel. Now let us ex-
amine the example below, where the Shuoyuan passage is sig-
ificantly longer than its counterpart in the Xinshu.
23Likewise, Yao Nai in pre-modern China considered the Xinshu spurious
partly because, to him, the Xinshu texts did not read as smoothly as their
textual counterparts in the Hanshu.
24In the Hanshu, we see a long quote of Jia Yi’s petition that is capped by
Ban Gu’s introduction to it. The introduction declares that this long quote
only “roughly” [dalue 大略] presents Jia Yi’s petition. As a pre-modern
annotator of the Hanshu, Yan Shigu 颜师古 noted that this was “probably
[a case where] the historian simply put [in the history] what was important
and relevant”. “And,” said Yan Shigu, “[that is why Ban Gu] later said in the
final comment that [he] ‘selectively’ included in [Jia Yi’s biography]those
[essays] that were pertinent to the affairs of that time.” 蓋史家直取其要切
耳。故下贊云掇其切於世事者著於傳 (Hanshu, p. 2260).
25They are Dong Zhongshu’s 董仲舒 (198-106 BC) Chunqiu Fanlu 春秋
繁露 and Xi u Xinag’s Shuoyuan 说苑 and Xinxu 新序.
26The examples he cites are, s pecifically, “Ron gjing” 容經 (in the Xinshu)
vs. “Yubei ” 玉杯 (in the Chunqiu Fanlu), “Chunqiu” 春秋 (in the Xinshu)
vs. “Zaishi No. 4” 雜事四 (in the Xinxu), and “Xiuzheng yu No. 1” 修政
語上 (in the Xinshu) vs. “Jundao” 君道 (in the Shuoyuan).
Copyright © 2013 SciRes.
Copyright © 2013 SciRes. 19
Source “Xiuzheng yu No. 1 (Xinshu) “Jundao” (Shuoyuan)
取道者不以手而以耳，… Tang said, “medicines and foods are
tasted by the lo wly before [they are] submitted to the honorable.
Words of admonishme n t are prese nted to the honorable before being
heard by the lo wl y.” Hence it i s [ a matter of] edification to submit
medicines a nd foods to the honorable afte r having the lowly taste
them. [And] it is [pertaining to] the Dao to present words of
admonishment to the honorable before [let t ing the words b e] heard
by the lowly. Thus [one can savor much of [good] taste when eating
[one’s foods] after having the foods tasted by others. [But] on e will
get very little [out of] word s i f [one] hears [the words] after having
them savored by others. So, because of this, a wise ruler’s [way of
dealing with] words will always be hearing them personally,
selecting from them personally, collecting them personally, and
store them up personally. Therefore, it is sagacious to take the Dao
quickly27, prominently [rewarding] to im p le ment [the Dao] quickly,
[and] nice to quickly apply [the Dao] to the masses. For this reas o n,
[he] who seeks the Dao will do i t with [his] heart instead of [his]
eyes. [And he ] who takes the Dao will do it with his ears instead of
“medicines and foods are first tasted by the lowly before [they are]
submitted t o t he honorable. Words of admonishment are first
presented to the honorable before being heard by the low l y.” Hence
it is [a mat ter of] edification to submit m edicines to the honorable
after havin g t h e lowly taste them. [A n d ] i t i s [ p ertaining to] the Dao
to present word s of admonishment to the hono rable before [letting
the words be ] heard by the lowly. Thus [one] can savor much of
[good] taste when eating [one’s foods] after having the foods tasted
by others. [But] one will get very little [out of] words i f [one] hears
[the words] after having them savored by ot hers. Because of this, a
wise ruler’s words must be [obtained by] l is t ening to others, hearing
others, selecting from others, taking from others, collecting from
others, storing them u p through ot h ers, [and] implementing them
through others. Therefore, it is sagacious to take the Dao quickly,
prominently [rewar d ing] to im pl ement [the Dao] quickl y, [and] n i ce
to quickly apply the Dao to the masses28. For this reason, [he] who
seeks the Dao will do it with [his] heart inst ead of his eyes. [And he]
who takes the Dao will do it with his ears instead of his hands.
Between these passages, the Shuoyuan passage obviously
owes its more extensive length to its longer sequence of “必自
X Verb之” structure. To me, the “也” in the former’s “必自也
Verb 之”makes better sense than the latter’s “他”, because what
the entire passage says is that since one gets “very little [out of]
words if [one] hears [them] after having them savored by oth-
ers”, a “wise ruler” should “personally” deal with and work on
what is advised to him rather than do it through others. So, the
“自” in the structure is more likely to be an adverb [meaning
“by o ne se lf” ] t ha n a pre p osi ti on [meaning “from” or “through” ].
Thus, in spite of the longer text on the part of a non-Xinshu
source, the comparison above does not support the view that the
Xinshu was quoting the Shuoyuan. If one had to explain their
different lengths in terms of one source’s quoting another, then
we should say this looked rather like a case in which the author
of the Shuoyuan had been elaborating on a passage that he saw
in the Xinshu.
Finally, as indicated above, there have been large quantities
of early texts lost to time. Of the 29 quotes of Confucius in the
Mencius, only eight have had their sources identified. But few
of the eight quotes present perfect textual parallel to their
sources (Gu Yanwu, 1990: p. 339). Considering all this, it is
hardly feasible or constructive to compare lengths in search of
an authentic text.
The Xinshu and the Zuozhuan
From Hanshu 88, we learn that Jia Yi was the leading
scholar of the Zuozhuan study in his generation. In his time, he
studied, taught, and wrote about, the Zuozhuan. His expertise in
the study of the Zuozhuan is confirmed in Wu Chengshi’s 吳承
仕 (1885-1939) Jingdian shiwen xulu shuzheng 經典釋文序
錄疏證 (1984: p. 123). However, Wang Zhong 汪中 (1744-
1794) (1869: neipian 3, p. 5) noticed that, in those parts where
history of the Spring-Autumn period was cited and discussed,
the Xinshu did not present sufficient textual parallels to the
Zuoahuan. In other words, the author of the Xinshu did not
demonstrate enough of his familiarity with the Zuozhuan.
Inter-textual relation in early literature is a complicated mat-
ter. In pre-modern times, there was not a definitive demarcation
drawn between quoting and paraphrasing. Just as Wang Mao
王楙 (1151-1213) observes, “[there have been cases where] the
ancients did not present verbatim quotes of classics and early
philosophies. [They] often added and omitted [words] at will.”
古人引用經子語，不純用其言，往往隨意增減 (Wang Mao,
1939: p. 117)29.
Liu Fenglu 劉逢 祿 (1776-1829) (1995: p. 254a) also con-
sidered the connection between the Xinshu and the Zuozhuan
weak. From there, he proceeded to question the conventionally
alleged authorship of the Zuozhuan. From the late 19th to the
beginning of the 20th centuries, however, Liu’s judgment en-
countered vigorous challenge. Zhang Binglin 章炳麟 (1868-
1936) refuted Liu Fenglu’s view on several occasions. In his
Chunqiu Zuozhuan xulu 春秋左傳敘錄, Zhang Binglin exhib-
ited numerous examples to argue that the Chunqiu (i.e.
Spring-Autumn Period, 770-467 BC) lore in the Xinshu did not
contradict but rather complement the Zuozhuan30 (e.g. Zhang
Binglin, 1982: Vol. 2, pp. 841-843). Similarly, Yu Jiaxi also
Besides, by the Han Dynasty, although manual transcription
had long been an important means of preserving texts, the tradi-
tion of oral transmission did not extinct (Zhang Shunhui, 1990:
p. 5) Both oral and manual transmissions were liable to gener-
ate different versions of a text. Gardner once enumerated sev-
eral cases in which rival versions might occur (Gardner, 1961:
pp. 48-52). Thus nothing can guarantee an extant text to be the
only version there has been in history.
27Read 數 as 速.
28Read 藏 as 臧.
29Zhu Guozhen 朱國楨 (?-1632) must have also noticed this. See his Yon
Zhuang xiaopin 湧幢小品, juan 18.
30In a speech h e delivered in his old age, Zhang Bing l in no t ed i n passing that
the Xinshu “abounded in the citation of Master Zuo (Qiuming)” (Zhang
Binglin, 1995: p. 121). See also his Liu Zizheng Zuo Shi Shuo 劉子政左氏
remarked that, in the Xinshu, Chapter “Baofu” 保傅 was
partly based upon the Zuozhuan (Yu Jiaxi, 1963: pp. 266-267).
But the view of Wang Zhong and Liu Fenglu seems never-
theless valid. Usually, without annotators’ aid, (e.g. Yang
Shuda, 1984: Vol. 10, p. 751; Yang Bojun, 1981: pp. 719, 788-
789, 1195) the parallels between the Zuozhuan and the Xinshu
would easily pass unnoticed, for they are mostly imperfect par-
allels. Among all the 12 citations of the Zuozhuan recorded in
The Xinshu with parallel passages from other pre-Han and Han
texts, wording parallels are sporadic and rare31. More surpris-
ingly, the title of Zuozhuan is simply not mentioned in the Xin-
shu. What first-class expert would demonstrate so little knowl-
edge in his acclaimed area of expertise?
There has been, to my knowledge, no one in China—at least
no one in the Chinese mainland—that has attempted to answer
this question. In Europe, Svarverud has suggested two answers,
between which, the first one is forthright. The lack of reference
to the Zuozhuan is, as he suggests,
possibly an indication that Jia Yi was not [Svarverud’s
emphasis] well versed in the Zuozhuan, and may not even
have composed the Zuozhuan commentary entitled Zuozhuan
xungu 左傳訓故, as is recorded in Hanshu, Rulinzhuan
儒林傳 (Svarverud, 1998: p. 126).
Then why there is such a record in the Hanshu and Liu
Xiang’s 劉向 (77?-6? BC) Bielu 別錄 will become an unset-
tled issue. Compared with this first answer, Svarverud’s second
answer is more sophisticated. “Another more plausible expla-
nation,” says he,
seems to be that these historical events [of Spring-Au-
tumn period] did not serve the intention of being historical
records but rather moralistic anecdotes substantiating the
lessons to be drawn from history. Jia Yi was thus less
concerned with their historical accuracy and more con-
cerned with their pedagogical effect on teaching in the
state of Liang. As suggested by Wang Zhong these events
may possibly also belong to a lost corpus of early orally
transmitted texts not recorded in Zuozhuan (Svarverud,
1998: p. 126).
This is indeed an interesting point. Speaking of the records in
the Zuozhuan, however, perhaps little matters how accurate
they might be historically, many of them had already been ren-
dered moralistic enough by their author to meet various peda-
gogic purposes. So, if Jia Yi did know them well, how did he
find them unusable in his teachings? Why did he resort to an-
ecdotes instead of the Zuozhuan? Here, Chen Weiliang’s chal-
lenge is still not something we can afford to ignore. But if we
review the transmission of the Zuozhuan, we will find that for-
gery is anything but the sole, or even the best, explanation for
the lack of mention of the Zuozhuan in the Xinshu.
As Svarverud points out, Wang Zhong suspected that the
events cited in the Xinshu were transmitted from texts other
than the Zuozhuan. This suspicion might have been justified
long ago in the 20th century. In the 1970s, a silk text of
Spring-Au- tumn history entitled Chunqiu shiyu 春秋事語
was excavated in Changsha, Hunan Province32. Scholars agreed
that the text was composed in the Warring States period and
scribed around the founding of the Han Dynasty, between 210
and 190 BC (Zhang Zhenglang, 1977: pp. 36, 38). According to
Zhang Zhenglang 張政烺, this is a text composed for peda-
gogic pur- poses, intended to prepare youths for further study of
the Spring-Autumn annals (SAA henceforth). Without address-
ing an opposite opinion held by Tang Lan’s 唐蘭 33, Li Xueqin
李學勤 (1989: pp. 2-4, 6) substantiated Zhang Zhenglang’s
view by arguing that the Chunqiu shiyu was largely adapted
from the Zuozhuan. While taking note of some differences be-
tween the two texts, Li Xueqin concluded that the Chunqiu
shiyu was “indeed a categorical early work of Zuozhuan stud-
ies” 實為早期《左傳》學的正宗作品, particularly a work of
Xunzi’s tradition within the Zuozhuan school of historical
studies. In- terestingly, while Wang Zhong noticed the signifi-
cant discrep- ancy between a Spring-Autumn anecdote in the
“Shenwei” 審微 chapter of the Xinshu and its counterpart in
the Zuozhuan, Li Xueqin found the same anecdote in the Xin-
shu impressively parallel to its corresponding record in the
Chunqiu shiyu. What this seems to suggest is that it is one thing
not to directly cite the Zuozhuan, but it is quite another to devi-
ate from the Zuo- zhuan school of SAA study.
The affinity between the Chunqiu shiyu and Zuozhuan is also
noted by Wu Rongzeng 吳榮曾 (1998) and Wang Li 王莉
(2003) from different perspectives. And after comparing the
style, syntax, narrative format, viewpoints, and chronology of
events in this text and those in the Zuozhuan, Li Xueqin went
on to tackle the significance of the title “Shenwei”. He cited the
following heritage line of the Zuozhuan school recorded in Lu
Deming’s 陸德明 (556-627) Jingdian shiwen 經典釋文 and
Liu Xiang’s Bielu34: Zuo Qiuming 左丘明—Zeng Shen 曾申
—Wu Qi 吳起—Wu Qi 吳期—Duo Jiao 鐸椒—Yu Qing 虞
卿—Xunzi—Zhang Cang 張蒼—Jia Yi35. Li Xueqin observed
that, in the Zuozhuan school, there were four works that con-
tained the character wei 微 in their titles36. And from Li
Xueqin (1989: p. 5), we learn that Yan Shigu 顏師古 (581-
645) glossed the meaning of the wei to be “to explain the subtle
33Tang Lan’s view was expressed on August 28, 1974, at a forum sponsored
by the journal of Wenwu. A minute of the forum entitled “Zuotan Changsha
Mawangdui Hanmu boshu” 座談長沙馬王堆漢墓帛書 is published in
34The heritage chain presented by Lu Deming 陸德明 ends with Jia Yi
whereas that by Liu Xiang 劉向 ends with Zhang Cang 張蒼.
35Wang Zhouming (1982: p. 25) suggests that Jia Yi had another mento
called Master Wu 吳公. The heritage line Wang Zhouming presents is:
Xunzi—Li Si 李斯—Master Wu—Jia Yi. I would assume that this is based
upon the following account in Hanshu 48, “Jia Yi was a native of Luoyang.
A [youth] of 18, he was well reputed in his prefecture for [his ability]to
recite [and] interpret classics [and for his] literary talents. Master Wu, Gov-
ernor of Henan heard of his outstanding endowments. [He, thereupon,]
summoned [ Jia Yi and] placed [Jia] at his menxia. [Master Wu] treated [Ji
Yi] with exceeding favor [emphasis added] … As formerly a townsman o
Li Si, [Master Wu] once studied under [Li Si’s] mentorship.” 賈誼,雒陽人
下，甚幸愛。……[吳公]故與李斯同邑，而嘗學事焉。It seems that Wang
Zhouming associated the menxia 門下 with mensheng 門生 (disciple).
But Chen Zhi 陳直 (1979: p. 288) pointed out that the term menxia here
meant menfu zhi xia 門府之下 (official patronage). I choose to accept
Chen Zhi’ s inter pretation because I judge th e phras e xing’ai 幸愛 to be not
as an appropriate expression to describe mentorship as it is for describing
36Among the four, authors of the Zuozhi Wei 左氏微 and Zhangshi Wei 張
氏微 are unidentified but suspected to be, respectively, Zuo Qiuming and
Zhang Cang (Li Xueqin, 1989: p. 5).
31Cf. Ho Che Wah, et al., 2007, p. 41, p. 137, p. 147, p. 149, p. 152, p. 154,
p. 163, p. 20 1, p. 220, p. 222, p. 223, p. 228.
32The original title of this text is unknown. The title Chunqiu shiyuwas
given by modern scholars afte r i t s excavation.
Copyright © 2013 SciRes.
meaning” 釋其微旨. Li Xueqin held that wei was a common
genre in the Zuozhuan school of SAA study, which featured
short discussions of the morals and lessons to be drawn from
the Zuozhuan. He observed that works in this genre had some-
times used materials not included in the Zuozhuan. And, as a
spin-off of his research, Li discovered that the style and format
of Chapter “Shenwei” in the Xinshu were typical of this genre
(Li Xueqin, 1989, pp. 5-6).
Moreover, Li found the excavation spot of the Chunqiu shiyu
significant. As is generally known, Changsha was where Jia Yi
once served as Grand Tutor for the King of Changsha. Consid-
ering his official post in Changsha and the fact that the Chunqiu
shiyu was intended for pedagogical purposes, one may not feel
very surprised at the similarity that Li Xueqin identified be-
tween the Chunqiu shiyu and “Shenwei”.
Yet, we still have good reason to wonder why Jia Yi demon-
strated so little of direct knowledge of the Zuozhuan. A remark
by Wang Mao (1939: pp. 133-134) regarding the use of yijing
逸經 during the Western Han may deepen our perplexity.
The yi 逸 in the term yijing means “to be left out” (Yang
Bojun, 1981: p. L, 1984: p. 197) A yijing thus refers to a jing
(經 classic text) that was not adopted in the imperially author-
ized curriculum of classic studies, which in turn means that the
study of a yijing was not a practice under imperial patronage.
As Yang Bojun (1981: p. L) suggests, the Zuozhuan had long
been considered a yi chunqiu 逸春秋 (i.e. an extracurricular
text of SAA studies) in the Han Dynasty. It was, according to
Yang Bojun, considered a yi chunqiu even in the time of Wang
Chong 王充 (27-97) in the Eastern Han Dynasty. This means
the Zuozhuan had remained extracurricular throughout the en-
tire Weste r n Han era.
This, however, still does not adequately explain the lack of
its reference in the Xinshu. Although the study of extracurricu-
lar classic texts (i.e. yijing) did not enjoy governmental patron-
age, it was not banned, either37. In fact, Wang Mao noticed that
there had been a number of scholar-officials during the Western
Han who actually took delight in citing such texts in their writ-
ings and speeches. The people Wang Mao listed were mostly
scholar-officials after Jia Yi’s time. But, on the list there was
also Xiao He 蕭何, who significantly predated Jia Yi. Wang
Mao claimed to have found numerous cases of yijing-citing in
the Han Dynasty. In that case, although the Zuozhuan could not
possibly be used as an orthodox canon, why was it not even
honored in the Xinshu in the capacity of a yijing? And, as a
logical question to follow, could the author of the Xinshu still
be someone specialized in Zuozhuan study, living in a time
when yijing-citing was at least perfectly acceptable if not nec-
My answer to these questions is invariably affirmative. We
need, firstly, examine the transmission of the Zuozhuan during
the Western Han. In spite of the long transmission that the
Zuozhuan school of SAA studies had undergone, the Zuozhuan
never became a popular classic in the imperial court until the
reign of Emperor Ping of the dynasty (r. 1 BC-6 AD), which
was over 160 years after Jia Yi’s death. Previously, during
Emperor Ai’s (r. 6-1 BC) time, the Zuozhuan school once at-
tempted to establish an official institute for the Zuozhuan study.
But they did so only to meet with oppositions from the Imperial
Erudites (Hanshu, 1962: p. 1976). Hence Pi Xirui 皮錫瑞
(1850-1908) (1936: p. 79) agreed with Liu Chang’s 劉敞
(1019-1068) that the Zuozhuan was unwelcome among West-
ern-Han literati. In fact, this problem was so impressive among
later generations of scholars that even a record of the Liu 劉
collateral group in the Zuozhuan would be enough to arouse
suspicion38. During the Western Han, the popular and official
schools of SAA studies in the imperial court had first been the
Gongyang 公羊 tradition and then Guliang 穀梁 tradition.
Yet even the prevalence of these two traditions also predated
Secondly, one must take into consideration the political at-
mosphere in the imperial court at that time. It is generally
agreed that the Western Han was a time when the imperial
court largely discouraged citation of classics. As a most widely
cited example, Lu Jia 陸賈, a scholar-official previous to Jia
Yi, once got scolded by Emperor Gaozu for citing classics from
time to time. What the emperor wanted him to do instead was
to summarize the political lessons that could be drawn from the
succession of the Qin Dynasty by the Han Empire (Hanshu,
1962: p. 2113).
Things did not change much in Jia Yi’s generation. Emperor
Wen was the ruler that Jia Yi served. From the biography of
Zhang Shizhi 張釋之, we learn that this emperor would still
expect an official to “lower” his arguments down to a practical
level instead of “issuing very high[-sounding] theories.” At
such an imperial request, Zhang Shizhi, too, had to confine his
topic to the lessons of the Qin-Han period (Hanshu, 1962: p.
The Xinshu are divided into such sections as shishi, lianyu,
and zashi. According to Yu Jiaxi (1958: p. 541, 544), the texts
in the shishi section were written to the emperor whereas those
38Liu was the surname of the royal family in the Han Dynasty. In the Zuoz-
huan, Lord Wen year 13, there is a sentence reading “those who remained
[came to] be [known as] the Liu collateral branch” 其處者為劉氏. Kong
Yinda 孔穎達 (547 -648) suspected the sentence to be a deliber ate interpo-
lation by the Zuozhuan scholars of the Han times. Those scholars—such is
Kong Yingda’s suspicion—mixed this artificial sentence into the Zuozhuan
in hope of enhancing the popularity of the Zuozhuan in their time. “Explor-
ing [its] context,” says Kong, “[one would find] the sentence odd. [I] in-
tensely suspect that this sentence was not what [the original author]had
intended [to say]. Basically, [I] think, [it was because] ancient learning was
abandoned at the beginning of the Han Dynasty [and hence]Master Zuo
[Qiuming’s work]was not eminent in the world then. For lack of means to
fulfill themselves, [those]early scholars of classics … thus interpolated
these words with a view to winning the favor of [their] generation. 討尋上
不顯於世，先儒無以自申，……插注此辭，將以媚於世 (Du and Kong,
1983: Vol. 143, p. 424a).Kong’s suspicion is now proven to be ill founded
(see Yang Bojun, 1981: pp. 596-597). However, such suspicion itself best
exemplifies the deep impression left among later generations of Chinese
scholars regarding the unpopularity of the Zuozhuan during the Western Ha n
37As scholars during the reign of Emperor Wu (r. 141-87 BC), for instance,
Sima Tan 司馬談 and Sima Qian admittedly used much of the Zuozhuan in
the Shiji. But the Shiji was a private undertaking rather than an officially
commissioned compilation. As Wang Guowei 王國維 (1973: Vol. 11,
shilin 3, p. 513) pointed out, the Shiji was “a work co mp o s ed i n privacy” 私
家著述, which its author did not dedicate to the imperial court. This being
the case, t he use of th e Zuozhuan in the Shiji neither disproves the status o
the Zuozhuan as a yijing nor ind i cat es t h e acceptance of the Zuozhuan by the
imperial academia. The unofficial nature of the Shiji is not only noted in the
scholarship on the historical literature of pre-modern China (e.g. Loewe,
1966: p. 281), but also reflected in Sima Qian’s own postscript to the Shiji.
(i.e. “Taishigong zixu” 太史公自序) In the postscript, Si ma Qian stated that
he had composed the Shiji to convey “on e s cholar’s view” 一家之言, which
he would save for the “sages and worthies of later ge nerations” 後世聖人君
子 rather than present it to his contemporaries. Sima Qian, in addition,
expressed the same idea in h i s l etter to Ren An 任安.
Copyright © 2013 SciRes. 21
in the lianyu and zashi sections were largely Jia Yi’s unofficial
speeches recorded by his disciples or followers. If we accept
Yu Jiaxi’s view, then it seems interesting that one encounters
more Spring-Autumn stories in lianyu and zashi than in shishi39.
As is generally known, although “high-sounding” theorization
was not welcome in court, history had always been an impor-
tant subject of academic discipline. By the Han Dynasty, it had
been a long tradition among scholars of various persuasions to
resort to history in their theoretical undertakings (Luo Jun,
1995: pp. 96-97). Although this tradition could have been af-
fected by the political fashions of the time, it was not likely to
be terminated by the fashions. What was dispensed with in the
imperial court was not necessarily as dispensable in education.
In Chapter “Xianxing” 先醒 in the lianyu section of the Xin-
shu, one can find an example where Jia Yi, as a royal tutor,
taught a prince how to interpret three Spring-Autumn anecdotes.
However, in education, the contents of teaching would inevita-
bly be circumscribed by the outcome of academic and political
contentions. That may account for the lack of direct citation and
mention of the Zuozhuan even in those Xinshu lectures.
It was not until several decades after Jia Yi’s time did the
knowledge of the Zuozhuan come to be manifested in the
speeches of those scholar-officials with expertise in Zuozhuan
study. Liu Xin, for instance, once quoted a short passage from
Zuozhuan Lord Cheng year 13 in his memorial during Emperor
Cheng’s reign (32-7 BC) (see Hanshu, 1962: p. 979, 980, foot-
note 1). Note that Liu’s memorial with a quote of Zuozhuan
historically heralded the Zuozhuan school’s large-scaled but
abortive campaign to institutionalize the Zuozhuan study in 6
BC, the very year immediately after Emperor Cheng’s reign,
which means that such direct use of the Zuozhuan occurred
over 130 years after Jia Yi’s death. Another example is Du Ye
杜鄴, grandson of Zhang Chang 張敞, who was an offi-
cial-scholar very knowledgeable about the Zuozhuan. In his
early years, Du Ye had studied under Zhang Chang’s son—i.e.
Du’s own maternal uncle—Zhang Ji 張吉 and inhered thereof
the academic heritage of the Zhang family. In the year of 2 BC,
Du Ye wrote a memorial to his emperor, which contained the
In the past, the Earl of Zheng yielded to Madam Jiang’s
wish, hence the eventual cataclysmic usurpation of power
by [his younger brother] Duan. King Xiang of Zhou do-
mestically suffered from the trouble caused by Queen
[Mother] Hui, [which resulted in] his hazardous exile in
王內迫惠后之難，而遭居鄭之危 (Hanshu, 1962: p.
This is a passage featuring forthright citations of Zuozhuan
Lord Yin year 1 and Lord Xi year 24 (Yang Shuda, 1984: Vol.
9, p. 658)40. As an official message to the court, perhaps it was
no coincidence that this occurred squarely in the thriving period
of the Zuozhuan school in the Western Han—though the Zuo-
zhuan study was still not institutionalized at that time. By the
time this message was written, over 160 years had elapsed since
Jia Yi’s death. And it is noteworthy that the memorials of Liu
Xin and Du Ye both emerged at a time when the Western Han
Empire was approaching its collapse. The official establishment
of the Zuozhuan as an orthodox canon took place in the Eastern
Han Dynasty (25-220), which was historically a separate era
from the Western Han.
Political contentions during the early Western Han are known
to have been complicated by the competitions among schools of
thoughts. One glaring example is found in the vying between
Confucianism and Huang-Lao Daoism. By overtly citing and
emphasizing the Zuozhuan in political discourse, a scholar-
official might risk inviting otherwise avoidable frictions and
hostilities in court. The frictions might come from both the
throne and the official-scholar’s colleagues. Considering both
this background and the fragmented condition of the Xinshu,
one can hardly exclude the possibility that certain bamboo texts
about the Zuozhuan were simply ripped off and destroyed41.
Forgery, therefore, can hardly be the best—still less the sole—
explanation of the lack of references to the Zuozhuan in the
The textual corruption of the Xinshu is a plain fact from
which one can deduce neither way regarding the issue of its
authenticity. However, the use of an embedded end title in its
textual layout is found to be consistent with Western Han for-
mat of texts. A forger aware of this early format would have
tried to forge more of such titles and make each of them show-
ier or more ostentatious than the only end title “wumei”. Con-
versely, a forger who was ignorant about this early format
would not have thought to end an already complete sentence
with a redundant disyllabic word in the first place. Either way,
we start from hypothesizing the forgery of the text but end in
disproving the hypothesis.
But does this signify the authenticity of the text? We must
reiterate our earlier concession about the inadequacy of such a
single—albeit valid—piece of evidence. To textual critics who
accept nothing less than definitive authenticity, I would suggest
looking also for definitive evidence of forgery, because a case
of forgery must be established on no less concrete evidence
than should a case of authenticity. As shown in the present
40According to the edition of the Hanshu kuiguan 漢書窺管 in my posses-
sion, the latter citation is that of “Zuozhuan Lord Xi year 25”.(cf. Yang
Shuda, 1984: p. 1029). I suspect that its number “25” is a typographic error.
This citation by Du Ye is a citation of Zuozhuan Lord Xi year 24, not 25.
41By the Han Dynasty, there had already been a long history of censorship in
China. It is generally known that a large quantity of ancient documents was
destroyed in the Qin Dynasty. But that may not have been the first case o
censorship. Ban Gu noted that with the decline of the hegemonic culture o
the Zhou Dynasty, various lords began to violate former norms; and, as they
did so, they removed from early documents some texts about former pro-
prieties (Hanshu, 1962: p. 1029). Accor ding to Yang Sh uda (1984: Vol. 3, p.
130), Ban Gu’s observation was most likely to be based on Mencius 5B: 2.
Speaking of Jia Yi, Liu Xiang was the first known compiler of Jia’s writings
Considering the fact tha Jia Yi and Liu Xiang were about a century apart, we
cannot securely assume that Jia’s works were handed down to Liu Xiang’s
generation complete and intact.
39The Spring-Autumn stories in the shishi section are both few and brief.
The only exception in this regard is Chapter “Shenwei.” But this chapter, i
we recall Li Xueqin’s 李學勤 (1989: p. 5) judgment, is written in the genre
of wei 微. Unlike o t h er articles in this sect i o n , “ S henwei” bears no evi d ence
to indicate that it was a petition addressed to the court. For one thing, an
author of a petition would refer to himself as chen 臣and the emperor as
bixia 陛下. But neither of these is found in “Shenwei.” In fact “Shenwei” is
simply not composed in the form of a petition or correspondence, where the
author directly addressed another person or a certain sector of the govern-
ment. Rather, “Shenwei” squarely meets Li Xueqin’s description of the
genre wei insofar as it is a moralistic discussion. As such, it contains elabo-
rate presentations of four Spring-Autumn anecdotes; though, as scholars
have pointed out, the stories significantly differ from their counterparts in
Copyright © 2013 SciRes.
study, no skeptics of the Xinshu have provided any adequate
evidence to prove their point. The most solid basis for their
skepticism is still the mere lack of any adequate proof of au-
thenticity, which is anything but concrete evidence of forgery.
Once the invalidity of their basis is ascertained, what we see
is the limitation of the dichotomy of authenticity and spurious-
ness itself. Individual scholars’ common sense is still required
in their effort to gauge the extent of a text’s usability in the
scholarship on its conventionally ascribed era and author. I
would suggest always treating the Xinshu as a usable datum for
our study of Jia Yi until the very future day dawns upon us with
any concrete and unequivocal evidence of forgery discovered in
textual or, better still, archaeological research.
Cai Tingji 蔡廷吉 (1984). Jia Yi yanjiu賈誼研究. Taibei: Wenshizhe
Chan, W.-T. (1963). A source book in Chinese philosophy. Princeton,
NJ: Princeton University.
Chao Gongwu 晁公武 (fl. 1151-1161) (1983). Junzhai dushu zhi 郡
齋讀書志. Yin g yi n wenyu ange Siku quansh edition, Vol. 674 . Taipei:
Chen Liang 陳亮 (1143-1194) (1987). Longchuan ji 龍川集. In Deng
Guangming 鄧廣銘 (Ed.), Chen Liang ji 陳亮集. Beijing:
Chen Mengjia 陳夢家 (1980). Hanjian zhui shu 漢簡綴述. Beijing:
Chen Weiliang 陳煒良, Chen Bingliang 陳炳良, & Jiang Runxun 江
潤勳 (1958). Jia Yi yanjiu 賈誼研究. Hong Kong: Qiujing Yinwu,
Chen Zhensun 陳振孫 (1179-1262) (1983). Zhizai shulu tijie 直齋書
錄題解. Yingyin wenyuange Siku quanshu edition. Taipei: Sangwu.
Chen Zhi 陳直 (1901-1980) (1979). Hanshu xinzheng 漢書新證.
Tianjin: Tianjin Renmin.
Cherniack, S. (1994). Book culture and textual transmission in Sung
China. Harvard Journal of Asiat i c S tudies, 54, 5-125.
Cohen, A. (2000). Introduction to research in Chinese source materials.
New Haven, CT: Far Eastern Publications, Yale University.
Du Yu 杜預 (222-284), & Kong Yingda 孔穎達 (547-648) (1983).
Chunqiu Zuozhuan zhu shu 春秋左傳注疏. Yingyin wenyuange
Siku quanshu edition. Taipei: Shangwu.
Fang Xiangdong 方向東 (2000). Jia Yi ji huijiao jijie 賈誼集匯校集
解. Nanjing: Hehai Daxue Chubanshe.
Gardner, C. S. (1961). Chinese traditional historiography. Cambridge:
Gu Shi 顧實 (1876-1956) (1987). Hanshu yiwenzhi jiangshu 漢書藝
文志講疏. Shanghai: Shanghai Guji.
Gu Yanwu 顧炎武 (1613-1682) (1990). Ri zhi lu 日知錄. In Huang
Rucheng 黃汝成 (1799-1837) (Ed.), Ri zhi lu jishi 日知錄集釋.
Shijiazhuang: Huashan Wenyi Chubanshe.
Guoyu 國語 (4th cent.? BC) (1978). Taipei: Jiusi Chuban Youxian
Hanshu 漢書 (1st cent.) (1962). Beijing: Zhonghua.
He Guangyan 何廣棪 (2001). Chen Zhensun sheng zu nian xin kao
陳振孫生卒年新考. Wenxian, 87, 159-161.
Ho Che Wah 何志華, Chu Kwok Fan 朱國藩, & Fan Sin Piu 樊善標
(2007). The Xinshu with parallel passages from other Pre-Han and
Han texts《新書》與先秦兩漢典籍重見資料彙編. Ho ng K ong : The
Chinese University of H o ng Kong.
Huang Kan 黃侃 (1962). Wenxin diaolong zhaji 文心雕龍札記.
Huang Yunmei 黃雲眉 (1980). Gujin weishu kao buzheng古今偽書
考補證. Jinan: Qilu Shushe, 1980).
Jia Yi 賈誼 (200-168 BC) (1937). Xinshu 新書. Congshu jicheng
chubian edition. Shanghai: Shangwu.
Jiang Boqian 蔣伯潛 (1990). Jiaochou muluxue zuanyao 校讎目錄學
纂要. Beijing: Beijing Daxue.
Knoblock, J. (1988). Xunzi: A translation and study of the complete
works. Stanford: Stanford University Press.
Lewis, M. E. (1990). Sanctioned violence in early China. New York:
State University of New York.
Li Xueqin 李學勤 (1989). Boshu Chunqiu shiyu yu Zuozhuan de
chuanliu 帛書《春秋事語》與《左傳》的傳流. Guji zhengli yanjiu
xuekan, 20, 1-6.
Liji 禮記 (1st cent. BC) (1983). Yingyin wenyuange Siku quanshu
edition, Vols. 115-116. Taipei: Shangwu.
Liu Fengliu 劉逢祿 (1776-1829) (1995). Zuoshi Chunqiu kaozheng
左氏春秋考證. Xuxiu Siku quanshu edition. Shanghai: Shanghai
Luo Jun 羅焌 (1874-1932) (1995). Z huzi xueshu 諸子學述. Changsha:
Luo, S. (2002). Inadequacy of Karlgren’s linguistic method as seen in
Rune Svarverud’s study of the Xinshu. Journal of Chinese Linguis-
tics, 31.2, 270-299.
Needham, J. et al. (1954-1999). Science and civilization in China.
London: Cam bridge University.
Nylan, M. (1993). Hsin shu. In M. Loewe (Ed.), Early Chinese texts: A
bibliographical guide. Berkeley: Institute of East Asian Studies,
University of California.
Qi Yuzhang 祁玉章 (1969). Jiazi tanwei 賈子探微. Taipei: Sanmin
Qian Zhongshu 錢鍾書 (1979). Guan zhui bian 管錐編. Beijing:
Shiji 史記 (1st cent. BC) (1963). Beijing: Zhonghua.
Shuihudi qinmu zhujian 睡虎地秦墓竹簡 (4th?-3rd? BC) (1978).
Siku quanshu zongmu tiyao 四庫全書總目提要 (1724-1805) (1933).
Su Shi 蘇軾 (1037-1101) (1986). Wenzhong Su Shi wenji 文忠蘇軾
文集 (prefaced 1173). In Su Dongpo quanji 蘇東坡全集. Beijing:
Beijing Shi Zhongguo Shudian .
Suishu 隋書 (7th cent.) (1973). Beijing: Zhonghua.
Sun Qinshan 孙欽善, Wei Jiangong 魏建功, Wu jingcun 吳競存, &
Yin Falu 陰法魯 (1961). Guanyu Jia Yi Xinshu zhenwei wenti de
tansuo 關於賈誼新書真偽問題的探索. Beijing Daxue xuebao, 5.
Svarverud, R. (1998). Methods of the way: Early Chinese ethical
thought. Leiden & Boston: Brill.
Svarverud, R. (2000). Jia Yi Xinshu zhi chuanshi bianjie賈誼《新書》
之傳世辨解. Wenxian, 84, 12-31.
Wang Fuzhi 王夫之 (1619-1692) (1996). Du Tongjian lun 讀通鑒論.
In Chuanshan Quanshu Bianji Weiyuanhui (Ed.), Chuanshan
quanshu 船山全書 (Vol. 1 0). Changsha: Yuelu Shushe.
Wang Guowei 王國維 (1877-1927) (1973). Guantang ji lin 觀堂集林.
Hong Kong: Zhonghua.
Wang Li 王莉 (2003). Chunqiu shiyu yanjiu erti 《春秋事語》研究二
題. Guji zhengli yanjiu xuekan, 105, 42-45.
Wang Liqi 王利器 (1999). Wang Liqi xueshu 王利器學述. Hangzhou:
Zhejiang Renmin Chubanshe.
Wang Mao 王楙 (1151-1213) (1939). Yeke congshu 野客論叢.
Wang Yinglin 王應麟 (1223-1296) (1987). Yuhai 玉海. Beijing:
Wang Zhouming王洲明 (1982). Xinshu fei weishu kao 新書非偽書
考. Wenxue Yichan, 17-28.
Wang Zhong 汪中 (1744-1794) (1869). Shu xue 述學. Yangzhou:
Wang Z. M., & Xu Chao 徐超 (1996). Jia Yi ji jiaozhu 賈誼集校注.
Beijing: Renmin Wenx u e .
Wu Chengshi 吳承仕 (1885-1939) (1984). Jiangdian shiwen xulu
Copyright © 2013 SciRes. 23
Copyright © 2013 SciRes.
shuzheng 經典釋文序錄疏證. Beijing: Zhonghua.
Wu Rongzeng 吳榮曾 (1998). Du boben Chunqiu shiyu 讀帛本《春
秋事語》. Wenwu, 501.
Wu Yun 吳雲, & Li Chuntai 李春臺 (1989). Jia Yi ji jiaozhu 賈誼集
校注. Zhengzhou: Zhengzhou Guji Chubanshe.
Xunzi 荀子 (1st cent . BC) (1936). In X. Q. Wang王先謙 (1842-1918)
(Ed.), Xunzi jijie 荀子集解. Shanghai: Shangwu.
Yan Changyao 顏昌嶢 (1868-1944) (1996). Guangzi jiao shi管子校
釋. Changsha: Yuelu Shushe.
Yan Ruoqu 閻若璩 (1636-1704) (1796). Shangshu guwen shuzheng
尚書古文疏證. Tianjin: Wushi Kanben.
Yan Zhenyi 閻振益, & Zhong Xia 鐘夏 (2000). Xinshu jianzhu 新書
校注. Beijing: Zhonghua.
Yang Bojun 楊伯峻 (1981). Chunqiu Zuozhuan zhu 春秋左傳注.
Yang Bojun 楊伯峻 (1984). Luyu yizhu 論語譯註. Hong Kong:
Yang Shuda 楊樹達 (1885-1956) (1984). Hanshu kuiguan 漢書窺管.
Shanghai: Shanghai Guji.
Yao Minghui 姚明煇 (fl. 1914) (1933). Hanshu yiwenzhi zhujie 漢書
藝文志注解. Shanghai: Dazhong Shuju.
Yao Nai 姚鼐 (1936). Xibaoxuan wenji 惜抱軒文集. Shanghai:
Yates, R. D. S. (1997). Five lost classics: Tao, HuangLao, and Yin-
Yang in Han China. New York: Ballantine Books.
Yu Jiaxi 余嘉錫 (1958). Siku tiyao bianzheng 四庫提要辯證. Beijing:
Yu Jiaxi 余嘉錫 (1963). Yu Jiaxi lun xue za zhu 余嘉錫論學雜著.
Zhang Binglin 章炳麟 (1868-1936) (1982). Zhang Taiyan quanji 章
太炎全集. Shanghai: Shanghai Renmin.
Zhang Binglin 章炳麟 (1995). Guoxue jiangyan lu 國學講演錄.
Shanghai: Huadong Shifan Daxu e C hubanshe.
Zhang Shunhui 張舜徽 (1962). Zhongguo gudai shiji jiaodu fa 中國
古代史籍校讀法. Shanghai: Zhonghua.
Zhang Shunhui 張舜徽 (1990). Hanshu yiwenzhi tongshi漢書藝文志
通釋. Wuhan: Hubei Jiaoyu Chu b a nshe.
Zhang Xuecheng 章學誠 (1738-1801) (1985). Wenshi tongyi 文史通
義 Jiaochou tongyi 校讎通義. In Ye Ying 葉瑛 (Ed.), Wenshi
tongyi jiao zhu 文史通義校注. Beijing: Zhonghua.
Zhang Zhenglang 張政烺 (1977). Chunqiu shiyu jieti 春秋事語解題.
Wenwu, 248, 36-39.
Zhao Ximing 趙曦明 (1705-1787) (1937). Yanshi jiaxun (zhu) 顏氏
家訓(注). Shanghai: Shangwu.
Zhu Taiyan 朱太岩 (1989). Mantan xieke gong—Diaoke gongyi de
liubian (1) 漫談寫刻工——雕刻工藝的流變(一). Guji Zhengli
Yanjiu Xuekan, 19, 46-47.