1. Li Cang, a powerful lord in the area, held the office of Prime Minister of Changsha, a post to which he was appointed in 193 B.C. See Riegel, Jeffrey K., “A Summary of Some Recent Wenwu and Kaogu Articles on Mawangdui Tombs Two and Three,” Early China 1 (1975), 10–16CrossRefGoogle Scholar, particularly 11-12.

2. The title “Zhanguo zonghengjia shu” has been standard since Chinese scholars published the text as a monograph in 1976: Zhanguo zonghengjia shu, ed. xiaozu, Mawangdui Hanmu boshu zhengli (Beijing: Wenwu chubanshe, 1976)Google Scholar. This monograph is a revised transcription of “Mawangdui Hanmu chutu boshu Zhanguo ce shivven” 馬王堆漢墓出土帛書戰國策釋文, Wenwu 1975.4, 14–26Google Scholar. Both of these works are in simplified characters and even the revised transcription is far from an accurate representation of the manuscript. The 1983 edition Mawangdui Hanmu boshu (san) 馬王堆漢墓帛書(參) (ed. xiaozu, Mawangdui Hanmu boshu zhengli [Beijing: Wenwu chubanshe, 1983])Google Scholar is far better, giving the transcription in full-form characters and including for the first time clear photographs of the entire manuscript.

3. See, for example, Kuan, Yang 楊寬, “Mawangdui boshu Zhanguo ce de shiliao jiazhi” 馬王堆帛書戰國策的史料價値, Wenwu 1975.2, 26–28, 68Google Scholar; and Yong, Ma 馬雍, “Boshu bieben Zhanguo ce gepian de niandai he lishi beijing” 帛書別本戰國策各篇的年代和歷史背景, Wenwu 1975.4, 27–40, 26Google Scholar.

Other work on this material published in 1974 and 1975 also includes incorrect information. For example, Han, Xiao 曉菡 claimed in his article “Changsha Mawangdui Hanmu boshu gaishu” 長沙馬王堆漢墓帛書概述 (Wenwu 1974.9, 40–44)Google Scholar, that the text can be divided into twenty-eight accounts, which has been corrected to twenty-seven; or that only one matching account is found in Shiji, the actual number being eight. Lan, Tang 唐蘭 in the article “Zuo tan Changsha Mawangdui Hanmu boshu” 座談長沙馬王堆漢墓帛書 (Wenwu 1974.9, 45–57)Google Scholar, suggested that this text might be the lost work of Suzi蘇子, which is listed in the Hanshu ״Yiwen zhi” 藝文志 ([Beijing: Zhong-hua shuju, 1962], 30.1739)Google Scholar, as being in thirty-one pian. But further studies have not supported this suggestion since the manuscript appears to have been copied out from at least three independent sources (see n. 12 below). Tang Lan did not again mention the connection of this material to the Suzi in his more recent article “Sima Qian suo meiyou jianguode zhengui shiliao” 司馬遷所沒有見過的珍貴史料, in Zhanguo zong-hengjia shu, 123–153Google Scholar.

4. As with the Laozi manuscript found in the same tomb, because this manuscript appears to avoid usage of the character bang 邦, the personal name of the first Han emperor Gaozu 高祖(r. 206–195 B.C.), but does not avoid ying 盈or heng 恒, the personal names of the second emperor, Emperor Hui 惠 (r. 194-188 B.C.), and the third emperor, Emperor Wen 文 (r. 179–157 B.C.), respectively, the manuscript can be dated no earlier than 195 B.C., but no later than 188 B.C.

5. The M version recorded in this paper is a kaishu 槽書 transcription of the manuscript, even though the manuscript often writes older graphs which are close to the Small Seal style. For example, while the manuscript always records for the state name Yan, in the transcription I write the standard graph 燕 without mentioning the older graph.

6. The Zhanguo ce text was edited by Liu Xiang 劉向(77–6 B.C.), but after being partially lost was reconstructed by Zeng Gong 曾繁 (1019–1083) late in the eleventh century. This text, after being repeatedly collated, gave rise in the middle of the twelfth century to two separate versions of Zhanguo ce: Yanchuan Yao shi ben Zhanguo ce 剡川姚氏本戰國策 (for which I use the Sibu beiyao edition), published in 1146 by Yao Hong (fl. 1131-1146); and Bao shi Zhanguo ce shi juan 鮑氏戰國策十卷 (Sibu congkan ed.), published in 1147 by Bao Biao (fl. 1147). Their content is similar but the arrangement of the accounts differs. Both the Yao and the Bao versions are useful for textual comparison and are valued with equal weight in this study. Whenever Bao Biao alters the characters based on his own judgment but notes the original reading (yuan zuo 元作), I will use the original characters (see, e.g., note 23). James Crump gives a detailed discussion of the textual history of the Zhanguo ce; see his “The Chan-kuo Ts'e and its Fiction,” Toung Pao 47 (1960), 305–375, particularly 322-330Google Scholar.

7. Shiji (Beijing: Zhonghua shuju, 1959)Google Scholar.

8. The matching YZ version is from Zhanguo ce “Zhao ce 趙策 4,” 21.9b–11a, and the BZ version is from “Zhao juan: Xiao Cheng wang” 趙卷：孝成王, 6.53a–55a. The S version is from Shiji “Zhao shijia” 趙世良 43.1822. The S version records at thebeginning that the incident in this account took place in the first year of King XiaoCheng, which is 265 B.C. The Great Queen is the late king Hui Wen's 惠文 wife andthe mother of the new king Xiao Cheng.

9. The Chinese editors suggest that the damaged M graph at #15 could be 赦 she < *xjăg ‘to pardon’ which is phonetically and semantically similar to the 恕 shu < *śjag ‘to indulge one's wishes’ that appears in Z and S. If so, the two words are cognate lexical variants. The Old Chinese reconstruction is from Tonghe, Dong 董同解, Shang-gu yinyun biao gao 上古音韻表稿 (1947; rpt. Taibei: Zhongyang yanjiusuo Lishi yuyan yanjiusuo, 1967)Google Scholar.

10. At this point in the manuscript there is the notation wubai liushi jiu 五百六十九 (569), which refers to the number of characters in the account.

11. Nothing is known about Zi Yi except that he is mentioned in the Suoyin 索隱 commentary to the Shiji as “a wise man in Zhao”; see Shiji 43.1824.

12. The M version records the total number of characters used in account 8 as 569. Only accounts 15, 16,17,18 and 19 record such numbers, and a sum total number of the characters used in these five accounts, 2870, is given at the end of account 19. This suggests that these five accounts formed a single unit among the original source material from which the “Zhanguo zonghengjia shu” manuscript was copied, and furthermore indicates that the entire manuscript was copied from at least three independent sources: one that may have contained accounts 1–14, one that contained 15–19, and one that contained 20–27. Two Japanese scholars, Kudō Moto'o 工藤元男 and Fujita Katsuhisa 藤田勝久, have carefully examined the content of the manuscript and concluded that the manuscript is from three separate sources; see Kudō, , “Maōtaishutsudo ‘Sengoku jūoka sho’ to Shiki” , in Chūgoku seishi no kisoteki kenkyū , ed. kenkyūshitsu, Waseda daigaku bungakubu Tōyōshi (Tokyo: Waseda daigaku shuppanbu, 1984), 1–26, esp. 3–4Google Scholar; and Fujita, , “Maōtai hakusho ‘Sengoku jūōka sho’ no kōsei to seikaku” , Ehime daigaku kyōyōbu kiyō 愛媛大學敎養部紀要 19 (1986), 429–458, esp. 439Google Scholar.

13. Crump, James mentioned this similarity in “A Summary of Recent Articles on the Chan-kuo ts'e,” Early China 1 (1975), 15–16CrossRefGoogle Scholar.

14. The relation between versions differs from account to account. For example, in account 21, the M and Ζ versions are close, but are remote from the S version. Therefore, M and Z must be derived from a “prototype of ‘M and Z’,” while S derives from another prototype. An interesting illustration of different textual branches occurs in acccount 5, for which Z records two matching stories, which we might denote as Z1 and Z2. The M version is similar to Z2, suggesting that they are derived from one prototype; by contrast the S version is similar to Z1, so that both of them must derive from another prototype. In accounts 15, 16, 20 and 24, each version is equally similar and dissimilar to each other, so that we cannot establish a meaningful tree of lineage; therefore, every variation has to be worked upon independently. What all this means is that different rules have to be developed and applied for each account.

15. For the definitions of lectio difficilior and lectio facilior, see Maas, Paul, Textual Criticism, tr. Flower, Barbara (Oxford: Oxford University Press, 1958), 13Google Scholar; Reynolds, L.D. and Wilson, N.G., Scribes and Scholars (1968; rpt Oxford: Oxford University Press, 1974), 199Google Scholar; and Boltz, William G., “Textual Criticism and the Ma Wang tui Lao tzu,” Harvard Journal of Asiatic Studies 44.1 (1984), 201CrossRefGoogle Scholar.

16. For example, the graph is interchangeable with ji 欲 ‘weary’, 綵 with xi 裕 ‘coarse hemp cloth’, with xi 都 ‘gap’, etc.

17. It was taboo to record an emperor's personal name once the emperor had died; see, for example, Beck, B.J. Mansvelt, “First Emperor's Taboo Character and the Three Day Reign of King Xiaowen,” Toung Pao 73.1–3 (1987), 68–85, esp. 75–76CrossRefGoogle Scholar.

18. Index de Fang yan, avec texte critique, Index Series (Tongjian congkan), 2 vols. (Beijing: Bali daxue Beijing Hanxue yanjiusuo, 1951), 3/25/52Google Scholar.

19. The dialectal word zhi ‘to recover’ must be related to the word zhi ‘to recover’ written with 治 in the standardized writing system. This suggests that in the southern Chu dialect the word zhi ‘to recover’, generally reconstructed as *d'əjg, and the word zhi ‘to know’, generally reconstructed as *tjeg, were phonetically close enough to be represented by the graph 知 or 智.

20. Wang Niansun 王念孫 also suggested that yi is a mistake for xu; see Dushu zazhi 讀書雜志 (rpt. Taibei: Taiwan shangwu yinshuguan, 1974)Google Scholar, “Zhanguo ce,” 1.99–100. In this case, the lineage method also determines the proximate original to be the M and S variant.

21. Because of this failure of S to suit the context, the Zhonghua shuju edition suggests, without reference, that the last word, yi 已, is an error for hu 乎, an interrogative marker, so that the sentence would mean “Won't our state be harmed?” (Shiji 44.1859). Realizing, as will be shown below, that wu 无 was originally xian 先, it can be seen that this emendation is unnecessary.

22. The graph for xian in the photograph of M is actually very similar to the graph for wu. The Chinese editors are to be credited for transcribing it as xian.

23. The YZ version is sometimes considered to be superior to the BZ version; see, for example, Liangshu, Zheng 鄭良樹, “Lun Yao Hong jiaozhu ben Zhanguo ce de youdian ji qi liuchuan” 論姚宏校注本戰國策的優點及其流傳, in his Zhujian boshu lunwenji 竹簡帛書論文集 (Beijing: Zhonghua shuju, 1982), 228–246Google Scholar. The reasons for such a claim are: that YZ mentions the source material while BZ hardly acknowledges it; that YZ retains the division of thirty-three pian that Zeng Gong reconstructed following Liu Xiang, whereas BZ is divided into ten juan based first on geography, and then according to the names of the kings of the states; and that BZ adds commentaries that are not at all reliable, etc. However, this does not necessarily mean that the text that Bao used in his collation was unreliable. In fact, the characters recorded in Bao's notes as “yuan zuo X,” “originally written as X,” sometimes coincide with the M version, and thus retain the proximate original. This suggests that BZ may preserve the original text better than YZ.