1. The terms “character” and “graph” are used interchangeably herein but, where a distinction is made, characters as they appear on excavated materials are referred to as “graphs.”

2. Zhu Dexi 朱德熙 and Qiu Xigui 裘錫圭, “Zhanguo wenzi yanjiu (liu zhong)” 戰國文字研究(六穜), Zhu Dexi guwenzi lunji 朱德熙古文字論集 (Beijing: Zhonghua, 1995), 31–53, see 31–32 (originally published in Kaogu xuebao 1972.2). Zhu and Qiu's analysis was a revision of an earlier analysis by Chen Mengjia 陳夢家: Chen Mengjia 陳夢家, “Dong Zhou mengshi yu chutu zaishu” 東周盟誓與出土載書, Kaogu 1966.5, 271–81, see 275–76. See below for a detailed presentation of this analysis.

3. Crispin Williams, “Early References to Collective Punishment in an Excavated Chinese Text: Analysis and Discussion of an Imprecation from the Wenxian Covenants,” Bulletin of the School of Oriental and African Studies 74.3 (2011), 437–67.

4. For a brief introduction to the genre of oath and covenant, see Crispin Williams, “Interpreting the Wenxian Covenant Texts: Methodological Procedure and Selected Analysis,” Ph.D. dissertation (University of London, 2005), 76–89.

5. See, for example, Mark E. Lewis, Sanctioned Violence in Early China (Albany: State University of New York Press, 1990), 43–52 and passim.

6. Liu Boji 劉伯驥, Chunqiu huimeng zhengzhi 春秋會盟政治 (Taibei: Zhonghua congshu bianshen weiyuanhui 中華叢書編審委員會, 1962), 1, 216, n.2. Given that the Chun qiu 春秋, the Zuo zhuan 左傳 and other historical texts are by no means comprehensive in their coverage of events, we can be certain that the total number of covenants far exceeded this figure. There is, for example, no mention in the historical records of any of the covenants excavated at Houma and Wenxian.

7. Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu 侯馬盟書 (Beijing: Wenwu, 1976).

8. Henan sheng wenwu yanjiusuo, “Henan Wenxian Dong Zhou mengshi yizhi yi-hao kan fajue jianbao” 河南溫縣東周盟誓遺址一號坎發掘簡報, Wenwu 1983.3, 78–89, but see also 77. The full excavation report is in preparation: Henan Wenxian Dong Zhou mengshi yizhi 河南溫縣東周盟誓遺址 (Beijing, Wenwu, forthcoming).

9. See Crispin Williams, “Dating the Houma Covenant Texts: The significance of recent findings from the Wenxian Covenant Texts,” Early China 35 (2012–13), 247–75; Wei Kebin 魏克彬 (Crispin Williams), “Wenxian mengshu T4K5, T4K6, T4K11 mengci shidu” 溫縣盟書 T4K5、T4K6、T4K11 盟辭釋讀, Chutu wenxian yu guwenzi yanjiu 出土文獻與古文字研究 5 (Shanghai: Shanghai guji, 2013), 280–363, see 293–96.

10. For an overview of the site, see Lothar von Falkenhausen, “The Waning of the Bronze Age: Material Culture and Social Developments, 770–481 B.C.” in The Cambridge History of Ancient China, ed. Michael Loewe and Edward L. Shaughnessy (Cambridge: Cambridge University Press, 1999), 450–544, see 457–59.

11. Henan sheng wenwu yanjiusuo, “Henan Wenxian Dong Zhou mengshi yizhi yi-hao kan fajue jianbao,” 89.

12. For a table giving details for each pit, see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 401–20.

13. Henan sheng wenwu yanjiusuo, “Henan Wenxian Dong Zhou mengshi yizhi yi-hao kan fajue jianbao,” 78.

14. For the Houma covenants, see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 401–20. Final figures for the Wenxian site are not yet published, but an indication of the numbers can be gained from Williams, “Interpreting the Wenxian Covenant Texts,” 47.

15. I identify the named sanctioning spirit in the excavated covenants as Lord Yue 岳公, a mountain deity. In a number of covenant types the spirit is referred to simply as “My superior” wujun 吾君, which may refer to this spirit, or perhaps to the former lords of Jin. See Wei Kebin 魏克彬 (Crispin Williams), “Houma yu Wenxian mengshu zhong de ‘Yue gong’” 侯馬與溫縣盟書中的‘岳公,’ Wenwu 2010.10, 76–83, 98. For an English summary of this identification, see Williams, “Dating the Houma Covenant Texts.” For a discussion of the significance of burying the tablets in pits, see Susan Roosevelt Weld, “Covenant in Jin's Walled Cities: The Discoveries at Houma and Wenxian,” Ph.D. dissertation (Harvard University, 1990), 332–38.

16. The key point being that it demonstrates the assumption among elites that individuals could act independently when it came to participating in political groupings, regardless of their lineage affiliation. See Crispin Williams, “Ten Thousand Names: Rank and Lineage Affiliation in the Wenxian Covenant Texts,” Asiatische Studien LXIII.4 (2009), 959–89; Williams, “Early References to Collective Punishment in an Excavated Chinese Text.” It also reflects the wish of elites to extend political control to individuals and individual households, see Mark Edward Lewis's discussion of the “administrative individual” in his Writing and Authority in Early China (Albany: State University of New York Press, 1999), 20 and passim.

17. Weld: “Covenant in Jin's Walled Cities: The Discoveries at Houma and Wenxian,” 353–54.

18. Each covenant type found at Wenxian generally includes just one set of these four-clauses, with two different stipulations. At Houma the so called “Lineage Covenant Texts” (zongmeng lei 宗盟類) and “Pledge Texts” (weizhilei 委質類) are longer and repeat sets of clauses more than once. For examples, see Susan Roosevelt Weld, “The Covenant Texts from Houma and Wenxian,” in New Sources of Chinese History, ed. Edward L. Shaughnessy (Berkeley: The Society for the Study of Early China and The Institute of East Asian Studies, University of California, Berkeley, 1997), 125–60; Williams, “Interpreting the Wenxian Covenant Texts,” 40–41. As Weld notes, this formulaic structure for oaths is not particular to early China, but is found in oaths of other early cultures, e.g. those of the ancient Near East and Greece. See Weld, “Covenant in Jin's Walled Cities: The Discoveries at Houma and Wenxian,” 46, n. 1; Williams “Interpreting the Wenxian Covenant Texts,” 83–86.

19. I use “character variation” here broadly, to refer to variation at both the component- and calligraphic-level (also referred to as structural- and stylistic-variation). For discussion of these terms, see Crispin Williams, “A Methodological Procedure for the Analysis of the Wenxian Covenant Texts,” Asiatische Studien LIX.1 (2005), 61–114, see 70–72; Matthias L. Richter, “The Fickle Brush: Chinese Orthography in the Age of Manuscripts: A Review of Imre Galambos's Orthography of Early Chinese Writing: Evidence from Newly Excavated Manuscripts,” Early China 31 (2007), 171–92, see 181. Chapter 6 of the work reviewed by Richter, Galambos's Orthography of Early Chinese Writing, provides a good example of the use of the Houma covenant materials in a study of character variation, see Imre Galambos, Orthography of Early Chinese Writing: Evidence from Newly Excavated Manuscripts (Budapest: Department of East Asian Studies, Eötvös Loránd University, 2006), 127–42.

20. Examples other than those mentioned in the main text include: Matthias Richter, “Towards a Profile of Graphic Variation,” Asiatische Studien LIX.1 (2005), 169–207; Enno Giele, “Signatures of ‘Scribes’ in Early Imperial China,” Asiatische Studien LIX.1 (2005), 353–87; Matthias Richter, “Tentative Criteria for Discerning Individual Hands in the Guodian Manuscripts,” in Rethinking Confucianism: Selected Papers from the Third International Conference on Excavated Chinese Manuscripts, Mount Holyoke College, April 2004, ed. Xing Wen, International Research on Bamboo and Silk Documents: Newsletter 5.2 (San Antonio, Trinity University, 2006), 132–47; Adam Smith, “Writing at Anyang,” Ph.D. dissertation (University of California, 2008), 247–53, 303–62; Matthias Richter, “Faithful Transmission or Creative Change: Tracing Modes of Manuscript Production from the Material Evidence,” Asiatische Studien LXIII.4 (2009), 889–908; Olivier Venture, “Looking for Chu People's Writing Habits,” Asiatische Studien LXIII.4 (2009), 943–57; Daniel Morgan, “A Positive Case for the Visuality of Text in Warring States Manuscript Culture,” paper for “The Creel-Luce Paleography Forum University of Chicago, 24–25 April 2010,” accessed online: http://cccp.uchicago.edu/archive/2010Creel-LucePaleographyWorkshop/ (Last accessed June 4, 2012); Matthias Richter, The Embodied Text (Leiden, Brill, 2013). See also the essays in “Part II: Scribal Training and Practice” of the book Writing and Literacy in Early China, ed. Li Feng and David Prager Branner (Seattle: University of Washington Press, 2011), 141–236. Those essays are: Ken-ichi Takashima, “Literacy to the South and East of Anyang in Shang China: Zhengshou and Daxinzhuang”; Adam Smith, “The Evidence for Scribal Training at Anyang”; and Matthias Richter, “Textual Identity and the Role of Literacy in the Transmission of Early Chinese Literature.” In the same volume, see also: Anthony J. Barbieri-Low, “Craftsman's Literacy: Uses of Writing by Male and Female Artisans in Qin and Han China,” 370–99. For an overview and discussion of the identification of different scribal hands in the Guodian bamboo-slip texts see Scott Cook, The Bamboo Texts of Guodian: A Study and Complete Translation (Ithaca: Cornell East Asia Series 164–65, 2012), 47–54. For a survey of the literature in Chinese see Li Songru 李松儒, “Zhanguo jianbo ziji yanjiu” 戰國簡帛字跡研究, Ph.D. dissertation (Jilin University, 2012), 7–20.

21. Li Feng, “Ancient Reproductions and Calligraphic Variations: Studies of Western Zhou Bronzes with ‘Identical’ Inscriptions,” Early China 22 (1997), 1–41. As Li points out, Matsumura Michio 松丸道雄 had already considered such questions in a 1977 article: “Sei-shū seidōki seisaku no haikei” 西周青銅器製作の背景, Tōkyō daigaku tōyō bunka kenkyūjo kiyō 東京大學東洋文化研究所紀要 72 (1977), 1–128.

22. As Richter points out, the distinction of scribal hands “demands the observation of a large number of recurrent graphic elements” and, for Chinese texts, that means a focus on commonly occurring characters and character components (Richter, “Tentative Criteria for Discerning Individual Hands in the Guodian Manuscripts,” 137–38). Repeated texts are the ideal materials for such analysis.

23. Both her MA and Ph.D. theses are on this topic: Li Songru 李松儒, “Guodian Chumu zhujian ziji yanjiu” 郭店楚墓字跡研究, MA thesis (Jilin University, 2006); Li Songru, “Zhanguo jianbo ziji yanjiu.” An example of a published work in which she applies this methodology is “You ‘Junrenzhe he bi an zai’ jia yi ben ziji kan xianqin wenxian de chuanchao” 由《君人者何必安哉》甲乙本字迹看先秦文獻的傳抄, Chutu wenxian yu guwenzi yanjiu 出土文獻與古文字研究 4 (Shanghai: Shanghai guji, 2011), 259–69.

24. Li Songru, “Guodian Chumu zhujian ziji yanjiu,” 9–16. For a more recent exposition of this methodology, see Li's Ph.D. thesis: Li Songru, “Zhanguo jianbo ziji yanjiu,” 74–108.

25. Richter also discusses these and related issues, see Richter, “Tentative Criteria for Discerning Individual Hands in the Guodian Manuscripts,” 134–36, 142.

26. These principles, and the terms “features” and “normal variation” are taken from Ron N. Morris, Forensic Handwriting Identification: Fundamental Concepts and Principles (San Diego: Academic Press, 2000), 63–65, 131–35. “Normal variation” refers to minor variations due to factors such as whether the writing is done carefully or carelessly, whether the writer is focused or tired, the nature of the writing implement, etc. It is also referred to as “natural variation,” see Roy A. Huber and A.M. Headrick, Handwriting Identification: Facts and Fundamentals (Boca Raton: CRC Press, 1999), 132–34. For a discussion of such use of the methodology of forensic handwriting identification in the palaeographic identification of scribal hands, see Tom Davis, “The Practice of Handwriting Identification,” The Library 8.3 (2007), 251–76.

27. In an interpretative transcription the standard characters are given for the words I believe are denoted by the original graphs. A formal transcription is one in which components of the ancient graph are represented using the corresponding components of the kaishu 楷書 script (with composite components transcribed using the corresponding composite component, not their separate base components). For this terminology, see Williams, “A Methodological Procedure for the Analysis of the Wenxian Covenant Texts,” 73–83.

28. For a copy and image of this tablet, see Henan sheng wenwu yanjiusuo, “Henan Wenxian Dong Zhou mengshi yizhi yi-hao kan fajue jianbao,” 85 and plate 7. Each individual tablet from Wenxian is identified by its test-square number (prefixed by the letters “WT”), its pit number (prefixed by the letter “K”), and its individual number. The letters are sometimes omitted, e.g. WT1K1-1 becomes 1-1-1. Tablets from Houma are prefixed with “HM,” followed by pit number, colon, tablet number, e.g. HM 1:1.

29. For this identification see He Linyi 何琳儀 and Wu Hongsong 吴紅松, “Shengsheng shi xun” 繩繩釋訓, Zhongyuan wenwu 2006.1, 62–64. For further evidence supporting this analysis, see Wei Kebin (Crispin Williams), “Wenxian mengshu T4K5, T4K6, T4K11 mengci shidu,” 341–42.

30. I adopt an identification of the word here as ji 極 that was suggested by Chen Jian (Personal communication, February 22, 2009).

31. A single text can mix the pronouns that refer to the covenantor: in this text qi 其 “his, her” is used as well as ru 汝 “you”. This arbitrary use of singular personal pronouns by the scribes who prepared the tablets may reflect an oral dimension to the covenant ceremony. It suggests different parts of the covenant were spoken by different people, or sections read by an official to be repeated by the covenantor and the pronoun adjusted accordingly.

32. Williams, “Early References to Collective Punishment in an Excavated Chinese Text.”

33. The words commonly denoted by these characters are: ma 麻 “hemp,” , if taken to be a variant for yi 夷 means “to make level,” fei 非 “to be not,” “be wrong,” shi 是 “this,” “be right.”

34. The analyses include (full references are given below): “[May the covenantor be] without peace and not happy” mi yi fei ti 靡夷匪禔 (Guo Moruo 郭沫若, 1966); “Wipe out my lineage” mo yi wo shi 摩夷我氏 (Chen Mengjia 陳夢家, 1966); “Wuyi [name of the river spirit He Bo 河伯] will punish violation of the covenant” Wu Yi fei shi 無夷非是 (Qi Guiyan 戚桂宴, 1979); “Destroy [and seize your] land and smash [your] lineage” mie di po shi 滅地破氏 (Peng Jingzhong 彭靜中, 1979); “Kill [the covenantor] and exterminate [his] lineage” ma yi [zhi] fei shi 麻夷[之]非氏 (Li Yumin 李裕民, 1983); “with no [degree of] destruction being inappropriate” or “it would not be right if destruction did not [befall him]” mi yi fei shi 靡夷非是 (Imre Galambos, 2005). For the above analyses, see: Guo Moruo, “Houma mengshu shitan” 侯馬盟書試探, Wenwu 1966.2, 4–6; Chen Mengjia, “Dong Zhou mengshi yu chutu zaishu,” 275–76, Qi Guiyan, “‘Ma yi fei shi’ jie” ‘麻非是’解, Kaogu 1979.3, 272, 230; Peng Jingzhong, “Guwenzi kaoshi er ze” 古文字考釋二則, Sichuan daxue xuebao (zhexue shehui kexue ban) 四川大學學報 (哲學社會科學版), 1979.2, 102–4; Li Yumin “Gu zi xin kao” 古字新考, Guwenzi yanjiu 古文字研究 10 (1983), 117–21; Imre Galambos, “A Corpus-Based Approach to Palaeography: The Case of the Houma Covenant Texts,” Asiatische Studien LIX.1 (2005), 115–30. A further suggestion is given by Du Zhengsheng 杜正勝 in his Bian hu qi min 編戶齊民 (Taibei: Lian jing 聯經, 1990), 442–48. Zhu Dexi and Qiu Xigui's analysis is found here: Zhu Dexi and Qiu Xigui, “Zhanguo wenzi yanjiu (liu zhong),” 31–32. For an article supporting Zhu and Qiu's reading and refuting that of Qi Guiyan, see Tang Yuming 唐鈺明, “Chonglun ‘ma yi fei shi’” 重論 ‘麻夷非是,’ in Zhuming zhongnian yuyan xuejia zixuanji: Tang Yuming juan 著名中年語言學家自選集:唐鈺明卷 (Hefei: Anhui jiaoyu, 2002), 101–10 (originally published in: Guangzhou shiyuan xuebao 廣州師院學報, 1989.2). For summaries and discussions of several of the suggested readings for this phrase, see Tsang Chi-hung 曾志雄, “A Study of Alliance Pacts Unearthed at Houma,” Houma mengshu yanjiu 侯馬盟書研究, Ph.D. dissertation (University of Hong Kong, 1993), 111–12 and n. 43; and also Galambos, “A Corpus-Based Approach to Palaeography: The Case of the Houma Covenant Texts,” 121–23.

35. See previous note.

36. Chen Mengjia, “Dong Zhou mengshi yu chutu zaishu,” 276. For copies of the graph, see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 325.

37. Zhu Dexi and Qiu Xigui, “Zhanguo wenzi yanjiu (liu zhong),” 50, n. 4.

38. Fangyan jiaojian 方言校箋, ed. Zhou Zumo 周祖謨 (Beijing: Zhonghua, 1993), 13.86. Also Yang Xiong Fangyan jiaoshi huizheng 楊雄方言校釋匯證, ed. Hua Xuecheng 華學誠 (Beijing, Zhonghua, 2006), 13.691–92.

39. Other examples are discussed here: Yang Xiong Fangyan jiaoshi huizheng, 13.691–92, n. 1; Mengzi zhengyi 孟子正義, ed. Jiao Xun 焦循 (Beijing: Zhonghua, 1987), 14.954.

40. Dates supplied with quotes from early texts are those most frequently given for compilation or initial authorship of the full text. These are provided to facilitate comparison when considering the usages discussed. It should be borne in mind that the dates are, in many cases, the subject of debate. It should also be noted that, for transmitted texts, the received versions have passed through many stages of copying and editing since they were first written down. When considering individual characters, in cases where a word did not have a dedicated graph commonly associated with it, the likelihood of significant variation in later reproductions is particularly high. This appears to be the situation we have with the word denoted by the ma 麻 graph under discussion here. For dating of transmitted texts, see the individual entries in Early Chinese Texts: A Bibliographical Guide, ed. Michael Loewe (Berkeley: The Society for the Study of Early China and The Institute of East Asian Studies, University of California, 1993). For discussion of the effects of transmission on early texts, see, for example, Richter, The Embodied Text, 1–9; William H. Baxter, “Zhou and Han Phonology in the Shijing,” in Studies in the Historical Phonology of Asian Languages, Current Issues in Linguistic Theory 77, ed. William G. Boltz and Michael C. Shapiro (Amsterdam and Philadelphia: John Benjamins, 1991), 1–34.

41. Guo yu 國語 (Shanghai: Shanghai guji, 1995), 21.641.

42. Mengzi zhengyi, 14.953.

43. Han shu 漢書, Ban Gu 班固 [Han] and Yan Shi 顏師 [Tang] (Beijing: Zhonghua, [1962] 1992), 51.2330.

44. Shuowen jiezi, Xu Shen 許慎 [Han] (Beijing: Zhonghua, 1963), 148 (7a米部22a).

45. Hou Han shu 後漢書, Fan Ye 范曄 [Liu Song] (Beijing, Zhonghua, [1965] 1995), 80.2600.

46. The reconstruction for the Old Chinese pronunciation of mo 摩 is: mo 摩 < ma < *mˤaj, while that for ma 麻 is: ma 麻 < mae < *mˤraj. Thus 麻 ma could be used to write either *mˤaj or *mˤraj, i.e. the syllable with or without the medial *-r-. I use the Old Chinese reconstruction system of William H. Baxter and Laurent Sagart, see their Old Chinese: A New Reconstruction (Oxford University Press, 2014). When giving reconstructions, the transcription for the Middle Chinese pronunciation appears first, followed by the Old Chinese reconstruction. Baxter and Sagart propose that Old Chinese shows the vestiges of an earlier derivational morphology based on affixation. For example, they suggest a derivational usage of an *-r- infix to indicate distributed action in verbs of action, and intensification in stative verbs. Thus, in the case under discussion, we might conjecture that we have a root 摩 *mˤaj, “to rub,” “to grind,” and a derivation with infix *-r- that indicates repeated action, and thus intensification of the action, giving the meaning “to crush, to obliterate, to pulverize,” leading to the gloss “to destroy.” Other examples of derivation from the root mo 摩/磨 include the derived noun mo 磨 “grindstone,” in which the suffix *-s nominalizes the root to give: mo 磨 < maH < *mˤaj-s. The same process may also be responsible for the noun mo 塺 “dust”: mo 塺 < maH < *mˤaj-s.

47. I do wonder whether the word denoted by these various characters is not in fact hui 毀 “to destroy.” Interchange between this character and those just discussed is not seen in early texts, but phonetically the match is close: hui 毀 < xjweX < *[](r)ajʔ and ma 麻 < mae < *mˤraj. Given these reconstructions, ma 麻 (and the characters with ma 麻 as phonetic discussed above) would certainly have been a suitable candidate to represent the word hui 毀. There is no consensus on how to analyze the character hui 毀 but one common explanation would suggest that the word's basic meaning was close to that of mo 摩 “to rub, to grind,” the root from which we have suggested the intensified word meaning “to crush, to obliterate, to pulverize” is derived. The graph hui 毀 is composed of the components jiu 臼 “mortar,” tu 土 “earth,” and shu 殳 “to beat” and Karlgren suggests this represents the action “to 殳 beat to 土 powder (mostly corrupted: 工) in 臼 a mortar” (Bernhard Karlgren, Analytic Dictionary of Chinese and Sino-Japanese (New York: Dover Publications, Inc., [1974] 1991), 65). The Shuowen jiezi glosses hui 毀 as que 缺 “broken, damaged,” which is not its usual usage, and analyzes it as being derived from the character hui 毇 as abbreviated phonetic (sheng sheng 省聲) (Shuowen jiezi, 289 (13b土部12a)). The word hui 毇 means “to pound [rice/grain] (in order to dehusk it)” and the graph could obviously be analyzed along the same lines as hui 毀, i.e. as depicting rice (mi 米) being pounded in a mortar (Shuowen jiezi, 148 (7a毇部23a)). The characters are given identical sound glosses, suggesting they are in fact just variant forms reflecting extended usages of a single word, i.e., “to pound (rice)” (hui 毇) and “to destroy” (hui 毀). Given their similarity in pronunciation and meaning, it seems possible that the graph hui 毀 denotes the same word as the proposed intensified derivative of mo 摩 that means “to crush, to obliterate, to pulverize” (commonly glossed as “to destroy”). The word hui 毀 occurs in early texts in the context of destruction of homes, lineages, and states, for example: Wu Yue chunqiu 吳越春秋 “Fu Chai neizhuan 5 夫差內傳第五:” “而吳伐二國，辱君臣，毀社稷，...” “[If] Wu attacked the two states, humiliated their rulers and ministers, and destroyed their Earth and Grain altars, …” (Wu Yue chunqiu jijiao huikao 吳越春秋輯校彙考, ed. Zhou Shengchun 周生春 (Shanghai: Shanghai guji, 1997), 5.95). Baxter has conjectured that hui 燬, which is a variant form of hui 毀, may be cognate with huo 火, which would imply that at its root hui 毀 means “destroy by fire,” an analysis which would not accord with that suggested above (William H. Baxter, A Handbook of Old Chinese Phonology (Trends in Linguistics. Studies and Monographs 64) (Berlin; New York: Mouton de Gruyter, 1992), 417). However, early usage of hui 毀 is not limited to destruction by fire, and includes, for example, the meaning “slander” (i.e. “destruction by words,” also written with the variant hui 譭). So it may be the case that the character hui 燬 reflects an extended meaning of the word rather than its root meaning. In Axel Schuessler's reconstructions of Old Chinese, ma 麻 (*mâi) is close to hui 墮/隳 (*hmai) “to destroy (by pulling down),” which would suggest another possible candidate for the word denoted by the graphs discussed above (Axel Schuessler, personal communication, February 3, 2013). Baxter, however, argues that hui 墮/隳 must have final *oj which would then make a connection with characters with the ma 麻 phonetic unlikely (Baxter, A Handbook of Old Chinese Phonology, 422 and 857, n. 312).

48. Chen Mengjia, “Dong Zhou mengshi yu chutu zaishu,” 276. For copies of the graph, see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 321. For the Guangya gloss, see Guangya shuzheng 廣雅疏證, ed. Wang Niansun 王念孫 [Qing] (Shanghai: Shanghai guji, 1983), 4b.483.

49. Guo yu, 3.111.

50. This point is also made by Li Yumin, “Gu zi xin kao,” 119–20; and Galambos, “A Corpus-Based Approach to Palaeography: The Case of the Houma Covenant Texts,” 128.

51. Ma Chengyuan 馬承源, ed. Shanghai bowuguan cang zhanguo Chu zhushu, vol. 5 上海博物館藏戰國楚竹書(五) (Shanghai: Shanghai guji, 2005). In this volume, the graph is found on slip 2 of the text Jing Jian nei zhi 競建內之 (168–69). Since that publication, however, it has been convincingly argued that this text should be merged with the text Bao Shuya yu Xi Peng zhi jian, see Chen Jian 陳劍, “Tantan ‘Shangbo (wu)’ de zhujian fen pian, pinhe yu bianlian wenti” 談談《上博(五)》的竹簡分篇，拼合與編聯問題, accessed online: http://www.bsm.org.cn/show_article.php?id=204 (Last accessed April 16, 2014).

52. As will be discussed further below, it may be significant that a graph denoting a word with a dental initial in Middle Chinese (zhi 雉 < drijX) is being used here in place of yi 夷 with its Middle Chinese palatal initial (yi 夷 < yij). The Shuowen jiezi has the graph 鴺, which is structurally equivalent to the Wenxian and Bao Shuya yu Xi Peng zhi jian graph (sharing the same phonetic and semantic components) and is glossed as yihu 鴺胡 “pelican” (now commonly written with the graphs tihu 鵜鶘), with the further comment that the graph is sometimes derived from di 弟 in place of the yi 夷 component (Shuowen jiezi, 81 (4a鳥部23a)), and thus giving another example of contact between yi 夷 and a dental initial.

53. Zhu Dexi and Qiu Xigui, “Zhanguo wenzi yanjiu (liu zhong),” 32. Examples of the graph from Wenxian are given below. An example of the graph is also given in Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 315.

54. Baxter, A Handbook of Old Chinese Phonology, 417.

55. Chen Mengjia, “Dong Zhou mengshi yu chutu zaishu,” 276. For copies of the graph, see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 318.

56. See examples in He Linyi 何琳儀, Zhanguo guwen zidian 戰國古文字典 (Beijing: Zhonghua, 1998), 750–51; Bai Yulan 白於藍, Jiandu boshu tongjiazi zidian 簡牘帛書通假字字典 (Fuzhou: Fujian renmin, 2008), 123–24.

57. Chen had only seen a limited number of excavated covenant texts when he published his article. These included the so-called Qinyang covenant texts (Qinyang zaishu 沁陽載書) in which there is an example that uses shi 氏 in place of shi 是, and he notes this as proof of his analysis. The “Qinyang covenant texts” is the name given to a small group of tablets that originated from non-scientific excavation at the Wenxian site in the mid-twentieth century, see Chen Mengjia, “Dong Zhou mengshi yu chutu zaishu,” 279–81; see also: Williams, “Interpreting the Wenxian Covenant Texts,” 42.

58. Qi Guiyan, “‘Ma yi fei shi’ jie”; Li Yumin, “Gu zi xin kao,” 117–21; Galambos, “A Corpus-Based Approach to Palaeography: The Case of the Houma Covenant Texts,” 123–24.

59. Li Yumin, “Gu zi xin kao,” 120.

60. Ma Chengyuan 馬承源, ed. Shanghai bowuguan cang zhanguo Chu zhushu, vol. 6 上海博物館藏戰國楚竹書(六) (Shanghai: Shanghai guji, 2007), 292. For the identification of fei 非 as bi 彼 see He Youzu 何有祖, “Chu jian san zha liu ze” 楚簡散札六則, accessed online: http://www.bsm.org.cn/show_article.php?id=646 (Last accessed December 3, 2012).

61. He Youzu seems to understand the pronouns as both referring to the person to whom respect is being shown (see previous note), but this does not accord with this pairing of pronouns or the context here which, given the use of the idiom, is clearly talking about two parties. Scott Cook makes a similar point, see Scott Cook (Gu Shikao 顧史考), “Shangbo Chu jian Yongyue zhang jie” 上博楚簡《用曰》章解, conference paper: 2007 nian Zhongguo jianboxue guoji luntan 2007 年中國簡帛學國際論壇 (Taiwan University, November 10, 2007).

62. Another Shanghai-museum text, Cao Mo zhi chen 曹沫之陳 has the phrase 非山非澤，亡有不民, in which the two fei 非 should perhaps also be read as bi 彼 (see Liu Hongtao 劉洪濤, “Shuo ‘fei shan fei ze, wu you bu min’” 說‘非山非澤, 亡有不民,’ accessed online: http://www.bsm.org.cn/show_article.php?id=539 (Last accessed December 3, 2012)). However, the problematic term bu min 不民 means it is not possible to make a conclusive judgment as to the meaning of the two fei 非 here. In the Chu Silk Manuscript (Chu boshu 楚帛書) fei 非 occurs before the term “nine skies” jiu tian 九天. Li Xueqin suggests this fei 非 should be read bi 彼, see Li Xueqin, “Chu boshu zhong de gu shi yu yuzhouguan” 楚帛書中的古史與宇宙觀, in Chu shi luncong 楚史論叢, ed. Zhang Zhengming 張正明 (Hubei renmin, 1984), 145–54. As with much of the Chu Silk Manuscript, this section has various problematic graphs as well as difficulties of interpretation, so the identification of the fei 非 here as bi 彼 is not conclusive.

63. Zhang Shichao (Chō Sei Chou) 張世超 et al., Jinwen xing yi tong jie (Kinbun Keigi Tsūkai) 金文形義通解 (Kyoto: Chūbun Shuppansha 中文出版社, 1996), 709–10.

64. Tang Zhibiao 湯志彪, San Jin wenzibian 三晉文字編, Ph.D. dissertation (Jilin University, 2009), 182–83.

65. Peng Jingzhong, in his analysis of the imprecation (see n. 34 above) suggests fei 非 denotes po 破 “to smash” (Peng Jingzhong, “Guwenzi kaoshi er ze”). The Old Chinese reconstruction for po 破 is po < phaH < *pʰˤaj-s, while that for the bi 彼 is bi 彼< pjeX < *pajʔ, and po 破 and bi 彼 share the same phonetic, so phonetically this is a possible loan. Nevertheless, this is not an attested loangraph usage for fei 非.

66. Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu.

67. The images were taken and processed as part of the collaborative project mentioned above. The project is greatly indebted to Carl Andrews (Laboratory of Computer Science, Massachusetts General Hospital, Boston), who initially oversaw the photography, scanning, image enhancement, and database construction. The survey I conducted for this study was done before a further set of photographs was taken in the summer of 2009 of tablets not originally selected for photography. Texts on the tablets photographed in 2009 are not included in the survey presented here.

68. In the following discussion the character , identified above as a variant form for yi 夷, is written directly as yi 夷, without the additional tu 土 component.

69. See examples in Gao Heng 高亨 (ed. Dong Zhi'an 董治安), Guzi tongjia huidian (Jinan: Qi Lu shushe [1989] 1997), 461; Bai Yulan, Jiandu boshu tongjiazi zidian, 123–24; Bai Yulan, Zhanguo Qinhan jianbo gushu tongjiazi huizuan 戰國秦漢簡帛古書通假字彙纂 (Fujian renmin, 2012), 281–84. The interchange does not appear to occur in Western Zhou bronzes, see Zhang Shichao (Chō Sei Chou) et al., Jinwen xing yi tong jie (Kinbun Keigi Tsūkai), 67–71 and 2918–23.

70. The development that allowed interchange between shi 氏 and shi 是 was a sound change that the word shi 氏 underwent. Baxter and Sagart note that loanwords and xiesheng contacts indicate a velar preinitial for shi 氏 that subsequently drops leaving the dental (Baxter and Sagart, Old Chinese: A New Reconstruction, section 4.4.4). Only after this development does shi 是 become a suitable graph to write shi 氏. The reconstructions are: shi 氏: *k.deʔ > *g.deʔ > *deʔ > dzyeX > shi, and for shi 是: *deʔ > dzyeX > shi. Once the velar preinitial *g- drops, the two words are pronounced identically (*deʔ) providing the conditions for interchange of the graphs by scribes. Among the Houma and Wenxian materials the two graphs are both used to denote the word shi 氏, although shi 是 is more frequent. The use of a loangraph and orthograph being used contemporaneously in this way is seen in other cases, e.g.: ce 冊 and ce 策, fei 飛 and fei 蜚, mei 眉 and mi 麋 (see Qiu Xigui, Chinese Writing, trans. Gilbert L. Mattos and Jerry Norman (Berkeley: The Society for the Study of Early China and The Institute of East Asian Studies, University of California, 2000), 268). That the shi 是 is generally more frequently used than shi 氏 in the imprecation phrase should not be taken as evidence that the word denoted is in fact shi 是. Comparing frequency of usage in this way is not a reliable indicator of meaning as is clear if one compares the usage of these two graphs in the two large Wenxian pits WT1K1 and WT1K14. These pits each held thousands of examples of the same oath type, but in pit WT1K1 the variation with shi 氏 occurs in 0.4% of those cases in which the last graph is legible, compared to 56.3% of corresponding examples in pit WT1K14. Thus, although the word denoted is clearly the same in both pits, in pit WT1K14 the use of shi 氏 is somewhat more common than that of shi 是, while in pit WT1K1 shi 是 is almost exclusively used. While it is tempting to conjecture that this discrepancy indicates that the pits date to different periods (i.e. WT1K14's more frequent use of shi 氏 reflecting a time at which this usage of shi 是 was not yet as prevalent), it could also simply reflect different habits among different groups of scribes, or different exemplars used during the writing process.

71. For examples of the wang 亡 graph, see Shanxi Sheng Wenwu Gongzuo Weiyuanhui, Houma mengshu, 325 (under the entry for ma 麻). Some of the Wenxian examples with this variation also have the alternative use of shi 氏 for shi 是.

72. Li Yumin, “Gu zi xin kao,” 120.

73. Chunqiu Zuo zhuan zhu (rev.ed), ed. Yang Bojun 楊伯峻 (Beijing: Zhonghua, [1990] 2006), 989–90 (“Xiang” 襄 11.3).

74. Chunqiu Zuo zhuan zhu, 169 (“Zhuang” 莊 6.3). In fact, this example could also be read without taking wang 亡 to mean “to destroy,” i.e.: “The loss of Deng will certainly be due to this man.”

75. Shi ji 史記, Sima Qian 司馬遷 [Han], (Beijing: Zhonghua, [2nd ed. 1982] 1992), 44.1855.

76. Zhanguo ce 戰國策, Liu Xiang 劉向 [Han] (Shanghai: Shanghai guji, [2nd ed. 1985] 1995), 25.922. (“Qin Wang shi ren wei Anling Jun” 秦王使人谓安陵君).

77. Zhanguo zonghengjia shu 戰國縱橫家書, ed. Mawangdui Hanmu boshu zhengli xiaozu 馬王堆漢墓帛書整理小組 (Beijing: Wenwu, 1976), 81 (“Qin keqing Zao wei Rang Hou” 秦客卿造謂穰侯).

78. Zhongguo shehui kexue yuan kaogu yanjiusuo 中國社會科學院考古研究所 ed., Yin Zhou jinwen jicheng 殷周金文集成 (Beijing: Zhonghua, 1984), vol. 8, no. 4341.a–c. The transcription and interpretation follow Li Xueqin 李學勤, “Ban gui xu kao” 班簋續考, Guwenzi yanjiu 13 (1986), 181–88.

79. Zhongguo shehui kexue yuan kaogu yanjiusuo, Yin Zhou jinwen jicheng, vol. 15, no. 9735.1–4 (for this phrase, see 9735.3). It is interesting to see that the Zhongshan bronzes appear to distinguish this verbal use of wang亡 from its use to denote the existential negative wu 無 by adding the “walking” radical chuo 辶.

80. Zhongguo shehui kexue yuan kaogu yanjiusuo, Yin Zhou jinwen jicheng, vol. 9, no. 4615.

81. Zhongguo shehui kexue yuan kaogu yanjiusuo, Yin Zhou jinwen jicheng, vol. 5, no. 2840.

82. A commonly given English example of this phenomenon is hearing “kiss this guy” for “kiss the sky” in the Jimi Hendrix song “Purple Haze.” The popular term “mondegreen” is sometimes used for these and other types of mishearing.

83. For examples in transmitted texts, see Gao Heng, Guzi tongjia huidian, 316–17; for examples in excavated texts, see Bai Yulan, Jiandu boshu tongjiazi zidian, 262–63; Bai Yulan, Zhanguo Qinhan jianbo gushu tongjiazi huizuan, 662–64.

84. Edwin G. Pulleyblank, Outline of Classical Chinese Grammar (Vancouver: University of British Columbia Press, 1995), 109. Schuessler suggests that this wang 亡 *maŋ is derived from a root *ma with suffix *-ŋ, the character 無 *ma reflecting the root. He suggests the negative mi 靡 *m(r)ajʔ (Schuessler: *mai), also used in place of wu 無 in transmitted texts, is also derived from this root. See Axel Schuessler, ABC Etymological Dictionary of Old Chinese (Honolulu: University of Hawai'i Press, 2006), 507, 518, 382. Baxter, on the other hand, suggests the possibility that the full form of the root was *maŋ and the *ma was an unstressed variant (that could only be used when another word followed). William Baxter, personal communication, May 19, 2014.

85. Pulleyblank argues that reading wang 亡 as wu is not supported by Shijing rhymes (Pulleyblank, Outline of Classical Chinese Grammar, 109) but this may be related to the particular nature of these examples (see, for example, Baxter's suggestion, in the previous footnote, that the word could not be read in its unstressed form at the end of a phrase).

86. For discussion and examples of this phenomenon, see: William H. Baxter, “Aspects of Old Chinese Morphology: Reading between the Characters in Early Chinese Texts,” conference paper: 4th International Conference on Classical Chinese Grammar (Vancouver, 2001).

87. Even if the *l- in yi 夷 < yij < *ləj had not yet started to develop to y-, the two are still both coronal consonants so *l- could perhaps have had the same effect. Several scholars, e.g. Edwin Pulleyblank and Zhengzhang Shangfang (鄭張尚芳), do indeed reconstruct the final coda of the ge 哥 rhyme group, to which mi 靡 belongs, as *-l, which would then match the initial of yi 夷.

88. The usual reconstruction for mi 靡 is *m(r)aj, the *(r) indicating that the medial *r is unconfirmed. Confusion of the word with wu 無, as conjectured here, would suggest mi 靡 did not have medial *r, so it is reconstructed in this diagram as *maj. See also n. 46 above.

89. The one example of this variant found in Houma pit 92, belongs to the Lineage Covenant Texts, Type 4 category, see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 115, 226 (tablet 92:5).

90. Houma pit 67 contains the covenant type known as the Confiscation Texts (na shi lei 納室類), see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 39–40, 73–74; and Weld, “The Covenant Texts from Houma and Wenxian,” 148–50. An example of the covenant type from Wenxian pit WT1K2 is given in Williams, “Early References to Collective Punishment in an Excavated Chinese Text,” 440–41, and discussed more fully in Williams, “Interpreting the Wenxian Covenant Texts,” 373–447.

91. In tablet 4 from this pit the scribe uses the wang 亡 variant to write the phrase, but then writes ma 麻 after the phrase. This suggests some sort of confusion due to what appears to have been the option to use either variation. The scribe perhaps looked between two separate exemplars each of which used a different variant. Shanxi Sheng Wenwu Gongzuo Weiyuanhui, Houma mengshu, 150, 277.

92. Shuowen jiezi, 196 (9b 勿部 13b).

93. Zhang Guangyu 張光裕, Guodian Chujian yanjiu I – wenzibian 郭店楚簡研究: 第一卷 文字編 (Taibei: Yiwen, 1999), 88.

94. See Bai Yulan, Zhanguo Qinhan jianbo gushu tongjiazi huizuan, 547–50.

95. Qiu Xigui, “Shi ‘wu’ ‘fa’” 釋‘勿’‘發,’ in Guwenzi lunji 古文字論集 (Beijing: Zhonghua, 1992), 70–84.

96. Xunzi jiji 荀子集解, ed. Wang Xianqian 王先謙 (Beijing: Zhonghua, [1988] 1997), 291 (“Qiang guo 彊國”).

97. Li ji zhengyi 禮記正義 in Shisan jing zhushu 十三經注疏, ed. Ruan Yuan 阮元 [Qing] (Beijing: Zhonghua, [1980] 1991), 1314 (“Tangong xia” 檀弓下 10.86).

98. See also n. 151 below where it is noted that Lu Deming 陸德明 (556–627 c.e.) glosses the first character (written mei 昧) of what appears to be a later record of this imprecation in the Gongyang zhuan as having an old pronunciation equivalent to the character wen 刎. This may just be coincidence but one might speculate that it originates from the variant with wu 勿 discussed here. On a separate note, it will be observed that both variants used in place of ma 麻, i.e. wang 亡 and wu 勿, are graphs that scribes would have been familiar with as denoting negatives. The graph ma 麻 is also found in excavated texts denoting the existential negative mi 靡 (read third tone), e.g. in the Zhongshan 中山 texts. In the case of the wang 亡 variant, scribes may, then, have associated the graphs ma 麻 and wang亡 as phonetically very similar negatives, and this may partly explain the interchange between them we see in the imprecation phrase, particularly if there was already some confusion about the word actually denoted here. The negative wu 勿 was not only different in pronunciation to both mi 靡 and wang 亡, but has a different function, being a prohibitive negative rather than an existential negative. This belies the speculation that scribes wrote wu 勿 as an alternative to what they understood to be the existential negative mi 靡. This variant does not, then, constitute evidence that the original word of the phrase itself was a negative function word.

99. Since it is difficult to be sure that these are a valid variation and not just an unusual calligraphic variant, they are not included as a separate category in the appended tables.

100. Examples from Zhang Shichao (Chō Sei Chou) et al., Jinwen xing yi tong jie (Kinbun Keigi Tsūkai), 2980–81; Tang Zhibiao, San Jin wenzibian, 736–37.

101. On the basis of this variant, we might speculate that bi 彼 was used in the written form of the oath, but that during the covenant ceremony the covenantor would have been required to speak this self-imprecation using the pronoun wo 我 “my.” This would correspond with the inconsistent use of pronouns referring to the covenantor elsewhere in the covenant texts (see n. 31 above). The infrequency of first-person pronouns in the written submission and imprecation clauses perhaps reflects an unwillingness on the part of the scribes to be repeatedly writing a self-curse.

102. The covenantor's name is perhaps Qi 奇, but the central strokes of the lower component do not seem to be a perfect match for the kou 口 expected in qi 奇.

103. For example Yu Yue 俞樾 et al., Gushu yiyi juli wu zhong 古書疑義舉例五種 (Beijing: Zhonghua, [1956] 1983), and a further supplement to this (with English translation): Pei Xuehai 裴學海, trans. Achilles Fang, “Fourth Supplement to the Ku-Shu i-i chü-li” 古書疑義舉例四補, Monumenta Serica, vol. 50 (2002), 549–654; also Chen Yuan 陳垣, Jiaokanxue shili 校勘學釋例 (Beijing: Zhonghua, 1959).

104. For example Qiu Xigui 裘錫圭, “Tantan Shangbo jian he Guodian jian zhong de cuobiezi” 談談上博簡和郭店簡中的錯別字, Zhongguo guwenxian shijiang 中國出土古文獻十講 (Shanghai: Fudan daxue, 2004) 308–16 (originally published in Huaxue 華學 6, 2003). And see also the summary in Li Songru, “Zhanguo jianbo ziji yanjiu.” 99–100. For an example where identification of scribal error explains an otherwise perplexing phrase, see Chen Jian 陳劍, “‘Shangbo 6—Kongzi jian Li Huanzi’ chong bian xin shi 《上博（六）•孔子見季桓子》重編新釋, accessed online: http://www.gwz.fudan.edu.cn/SrcShow.asp?Src_ID=383 (Last accessed March 26, 2014). See the discussion of the graph to be read fu 敷, in item 3 of section 3 “Analysis.”

105. See nn. 103 and 104 above.

106. Man-Tak Leung et al., “A Model of Writing Chinese Characters: Data from Acquired Dysgraphia and Writing Development,” in Writing: A Mosaic of New Perspectives, ed. Elena L. Grigorenko et al. (Hoboken: Taylor and Francis, 2012), 357–68, see 361.

107. David Moser, Slips of the Tongue and Pen in Chinese, Sino-Platonic Papers 22 (March, 1991). Adam Smith applies research into the cognitive processes involved in reading to the question of the emergence of literacy in China, see Adam Smith, “Writing at Anyang,” 49–137. Parts of his discussion are applicable to an understanding of variation and error in early Chinese writing, for example 53–57, 80–82, 111–12.

108. Unless otherwise cited, the following is largely based on the article “Cognitive Model of Writing,” Encyclopedia of the Human Brain (Oxford: Elsevier Science & Technology, 2002), accessed online: http://search.credoreference.com/content/entry/esthumanbrain/ii_cognitive_model_of_writing/0 (Last accessed March 14, 2014). I am grateful to my colleague Sanako Mitsugi 三ツ木紗奈子 for comments on an earlier draft of this paragraph.

109. Jocelyn R. Folk and Brenda Rapp, “Interaction of Lexical and Sublexical Information in Spelling: Evidence from Nonword Priming,” Applied Psycholinguistics 25 (2004), 565–85; Katherine K. White et al., “Why Did I Right That? Factors That Influence the Production of Homophone Substitution Errors,” The Quarterly Journal of Experimental Psychology 61, 7 (2008), 977–85.

110. The latter suggestion is based on the understanding that readers and writers of Chinese analyze characters at the component level, see Man-Tak Leung et al., “A Model of Writing Chinese Characters,” 358–59.

111. Man-Tak Leung et al., “A Model of Writing Chinese Characters,” 362. The authors refer to these units as “logographemes.”

112. The figures for frequency assume that lacunae in some examples match the corresponding graph for that particular variation.

113. Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 34.

114. The pronoun object is usually written as nü 女 (ru 汝) “you” or zhi 之 “him, her,” and the very first character of the phrase, usually a graph taken to be denoting di 諦, is sometimes replaced with a lexical variant, e.g. yong 永 “always, forevermore.”

115. Two examples (rows c and d in Table 2) with lacunae that may be this variant cannot confidently be assigned to either scribe so it is possible one or two other scribes may have also used this variant.

116. This is the regular sound change expected for *g-, palatalizing before front vowel *-i. See: Baxter, A Handbook of Old Chinese Phonology, 210–13.

117. Baxter, A Handbook of Old Chinese Phonology, 446–64, particularly 450.

118. Baxter, A Handbook of Old Chinese Phonology, 456–58.

119. This is the process known as *i-fronting: “the fronting of original *ə to i in syllables where both initial and coda were acute” (Baxter, A Handbook of Old Chinese Phonology, 577). Baxter notes that this process may have “applied differently, or at different times, in different Old Chinese dialects.” (Baxter, A Handbook of Old Chinese Phonology, 456).

120. Gao Heng, Guzi tongjia huidian, 530–33.

121. Bai Yulan, Jiandu boshu tongjiazi zidian, 140; Wang Hui 王輝, Guwenzi tongjia shili 古文字通假釋例 (Taibei: Yiwen, 1993), 611.

122. This corresponds to David Moser's use of the concept of “spreading activation” to explain certain speech and writing errors in modern Chinese. For an example of this type of error in modern Chinese writing, see his Slips of the Tongue and Pen in Chinese, 27 (example 75). For comparable cases in received texts, see the examples of identical characters with different meanings in adjacent phrases (shangxiawen tongzi yiyi 上下文同字異義) in Yu Yue et al., Gushu yiyi juli wu zhong, 3–4.

123. Scribe A also wrote the variation ▉夷非氏 (WT1K14-2069). The tablet is broken before the yi 夷 but we can conjecture that the missing graph is the ma 麻 we would expect in this position. The “” marks two lines: . This form is a reasonable match for the character er 二 “two” as it appears in the date found on tablets from pit WT1K1 at Wenxian, e.g.: (1-1-1991). No straightforward explanation would account for the addition of the word er 二 “two” here and I assume that this is a further sign of Scribe A's uncertainty as to how to write this phrase correctly.

124. In this case, the correction could have been given orally, or possibly written to the side of an incorrectly written example of the phrase on an already prepared tablet, which was then used as the model for further copying. For examples of such errors in received texts, see Chen Yuan, Jiaokanxue shili, 30–33.

125. The usual reconstruction for this would be *tˤek-s, with the post-coda *-s corresponding to the falling tone of the Middle Chinese reading. The conjecture of a segmentation error here would require that in the dialect of the scribes this *-s had been dropped. There is evidence for such dialectical differences, for example Lu Fayan 陸法言 (born c. 562 c.e.)'s preface to the Qieyun specifically mentions areas in which the falling and entering tones are not distinguished (see translation and discussion in Göran Malmqvist, “Chou Tsu-Mo on the Ch'ieh-yün,” Bulletin of the Museum of Far Eastern Antiquities 40 (1968), 33–78, see 35–37).

126. Baxter and Sagart currently reconstruct this as *[g](r)ək. The square brackets indicate that the initial could be more complex, but the simple initial accords with the suggested segmentation error. The “(r)” is included because a medial *-r- cannot be ruled out, but the conjectured error here would suggest it was not present.

127. Errors such as this that persist in received texts are those that produced a graphically similar attested character, for examples see Chen Yuan, Jiaokanxue shili, 20–23. In the case discussed here, the scribe writes the right-side of the graph with a combination of components giving a graphically similar form, but does not produce a different attested character. Such reanalysis of components was common during the development of the script, see, for example, Liu Zhao 劉釗, Guwenzi gouxing xue 古文字構形學 (Fuzhou: Fujian renmin, 2006), Chapter 6 and passim.

128. The shi 視 of the submission-clause phrase (諦極視汝) in tablet WT1K14-3763 is hard to make out, but may be the variant graph.

129. We could conjecture that the error occurred during direct copying when the scribe's eye returned to a previously written tablet being used as a model, and unintentionally picked up the last character from the previous phrase (perhaps due to its being conspicuous at the end of a column of characters). Or, as seems likely, both phrases may have been memorized, but confusion as to the meaning of the second phrase contributed to errors in its reproduction. If we conjecture that the scribe, at some level, felt this variation was legitimate, the fei 非 would have to be taken as a verb, and this would accord with Li Yumin's analysis of the phrase, as discussed above.

130. Scribe O actually writes the last character not as shi 氏 but as the graphically similar di 氐. This is discussed in the next paragraph.

131. This suggests an unintentional muddling of the phrases as they were held in working memory, and the phonetic similarity between shi 視 and yi 夷 discussed above may have contributed to the error. That is to say, the scribe omitted the shi ru 視汝, but after writing ma 麻 the phonetic similarity of the following word yi 夷 to shi 視 triggered the writing of the omitted phrase, after which the original phrase was completed with the yi fei shi 夷非氏.

132. The loss of the coda *-j from fei 非 gives the pronunciation of bu 不, and so this example can be analyzed as error due in part to “phonetic decay” during mental processing of the word, see Moser, Slips of the Tongue and Pen in Chinese, 24–27.

133. For similar examples in received texts, see Yu Yue et al., Gushu yiyi juli wu zhong, 119–25. The phonetic similarity of yi 夷 and shi 視 perhaps contributed to the insertion of the nü 女 (ru 汝) in this particular position.

134. For copies of these tablets, see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 171–72 and for photos of HM 1:40 and HM 1:41, see 89.

135. Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 172.

136. The 麻夷女非氏 (WT1K14-1229) variant just discussed is structurally the same (taking nü 女 to be denoting the pronoun ru 汝), but that example was convincingly shown to be an error by a careless scribe.

137. The graph which I believe is denoting pou 剖 is derived from fu 付, which I take to be the phonetic component: fu 付 < pjuH < *p(r)o-s and pou 剖 < phuwX < *pʰˤ(r)oʔ. See Wei Kebin (Crispin Williams), “Wenxian mengshu T4K5, T4K6, T4K11 mengci shidu,” 288–89.

138. For a discussion of the meaning and identification of this graph, see Williams, “Dating the Houma Covenant Texts.”

139. Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 328. 33 variant forms are given.

140. See: Li Jiahao 李家浩, “Shi ‘bian’” 釋“弁,” Guwenzi yanjiu 古文字研究 I (1979), 391–95.

141. Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 328.

142. See: Galambos, Orthography of Early Chinese Writing: Evidence from Newly Excavated Manuscripts, 127–42 (particularly 141), 145.

143. See n. 65 above for the possibility that the scribe was thinking of the verb po 破 “to smash.”

144. In the Houma mengshu a hand copy is given but not the original image, see Shanxi sheng wenwu gongzuo weiyuanhui, Houma mengshu, 264.

145. A number of other apparent variations in the imprecation phrase are difficult to analyze because of problems with legibility and are not discussed. These include the following ( indicates an additional or possibly variant graph): 麻夷非是 (1-14-5087), 麻夷非是 (1-14-2040), 麻夷非 (4-9-4), 非氏 (1-2-95 and 1-2-167), 麻夷 (1-1-60), 是 (5-14-30). Tablet WT1K17-43 writes麻夷□是, the additional graph is . I had wondered whether this was shu 庶 “children of secondary wives,” which would have fitted the proposed analysis well (combining with the shi 氏, the male offspring of the primary wife). However, the match with shu 庶 is not fully convincing, and it turns out that this scribe is unpredictable. This tablet is one among three from this pit (tablets 43, 110 and 112) written by the same scribe, all of which do not use the standard covenant type for this pit but instead a very basic text just requiring loyalty, without any specific stipulations and, in each copy, the scribe varies the wording of the loyalty phrase. Furthermore, in tablet 43, the scribe appears to have left out the covenantor's name in the name clause and so it may be simpler to conjecture that the additional graph is the covenantor's name, which the scribe has appended at the end.

146. See n. 16 above.

147. Wei Kebin (Crispin Williams), “Houma yu Wenxian mengshu zhong de ‘Yue Gong’”; also Williams, “Dating the Houma Covenant Texts.” In this case, a single scribe used the graph yu 獄 in place of the commonly found graph denoting this spirit's name. The graph yu 獄 could be analyzed as denoting yue 嶽, which is also commonly written with the character yue 岳. This led to the realization that the commonly found graphs used for the name could be analyzed as early forms of the character yue 岳.

148. Weld, “The Covenant Texts from Houma and Wenxian,” 132; Galambos, Orthography of Early Chinese Writing: Evidence from Newly Excavated Manuscripts, 127–42.

149. Zhu Dexi and Qiu Xigui, “Zhanguo wenzi yanjiu (liu zhong),” 31–32.

150. Chun qiu Gongyang zhuan zhushu 春秋公羊傳注疏 in Shisan jing zhushu 十三經注疏, ed. Ruan Yuan 阮元 [Qing] (Beijing: Zhonghua, [1980] 1991), 2312 (“Xiang” 襄 27, 2).

151. The Baxter/Sagart reconstruction for mei 昧 is: mei 昧 < mwojH < *mˤ[u][t]-s, but *-e- is also an option for this vowel: compare their reconstruction for mei 妹, which shares the same phonetic component: mei 妹 < mwojH < *C.mˤə[t]-s. The final *-t simplifies to *-j through “final cluster simplification” and we can conjecture this had already happened when this graph was selected for use for this syllable. On this basis, the reconstruction becomes *mˤəj-s. As discussed, the rhyme categories *-aj and *-əj merge in some dialects and the analysis of fei 非 < pjəj < *pəj denoting bi 彼 < pjeX < *pajʔ in the covenant texts requires this to have been the case in the Jin dialect of the covenants. The words mei 昧 *mˤəj-s and ma 麻*mˤraj would have been affected in the same way, making them phonetically very similar. Another approach for this character would be to consider its sound gloss in the Jingdian shiwen 經典釋文 which says: “The old pronunciation is [that of the character] wen 刎” (Lu Deming 陸德明 [Tang], Jingdian shiwen 經典釋文 (Shanghai: Shanghai guji, 1985), 1251). We suggested that the variant in which ma 麻 is replaced by wu 勿 is denoting wen 刎 “to cut apart, to cleave.” Nevertheless, it seems unlikely that this relatively unusual variant is the model for the phrase as used in the Gong yang zhuan. Moving on to the second character, it was noted above that the Wenxian texts include a variant graph in this position, which is also found in the Shanghai slip text Bao Shuya yu Xi Peng zhi jian denoting the word zhi 雉. The use of the graph zhi 雉 in the Gongyang zhuan in this phrase is thus the same variation with the use of a different variant graph. As discussed, the rhyme category of yi 夷 can be reconstructed as *-ij, giving *lij, a very close match to zhi 雉 *lrijʔ. The third graph is bi 彼, the word that the adopted analysis argues is denoted by the fei 非 of the excavated covenants. If we accept that the Gongyang zhuan phrase is indeed the same imprecation, then its use of bi 彼 provides further support for this analysis of fei 非. The final graph is somewhat more problematic. The reconstruction for shi 視 is shi 視 < dzyijH < *gijʔ-s, while that for shi 氏 is shi 氏 < dzyeX < *k.deʔ. As noted, in the covenant texts shi 視 takes the phonetic di 氐 from which we can infer the initial had palatalized to give *dʑijʔ, resulting in a close match with the root of shi 氏, the *d- (and the development to Middle Chinese dzy- may already have been underway). Zhu and Qiu note that there are examples in early texts of interchange between shi 是, the graph commonly found in the covenant texts to denote shi 氏, and shi 示, the phonetic component in shi 視, for example in the Zhou li the word qi 祇 “earth spirit” < gjie < *ge (<*k.de) is consistently written with the character 示 shì < zyijH < *s-gij-s (Zhu Dexi and Qiu Xigui, “Zhanguo wenzi yanjiu (liu zhong)” 32). One can also not help noticing that, in bronze inscriptions, graphs that can be confidently taken to be early variant forms of shi 視 frequently have shi 氏 “lineage,” rather than di氐 (or indeed shi 示) as an apparent phonetic. For example: (Yuan ding 員鼎); , , (Zhongshan Wang Cuo zhaoyu tu 中山王兆域圖); (Ping Yin ding 平陰鼎); (Xin'an jun ding 信安君鼎); (He zun 何尊). This of course would suggest shi 視was an ideal graph with which to denote shi 氏. There may, then, be more to be said about the reconstruction of shi 視's Old Chinese pronunciation. Nevertheless, the vowels *-i- and *-e- are already close, and the *-j coda in *-ij may have brought the *-i- closer to the *-e-. So, overall the use of shi 視 for shi 氏 seems plausible. The fact that those speaking and recording this phrase may have been uncertain as to the precise identity of the individual words of which it is composed may also have led to some confusion as to the correct pronunciation of its component syllables.

152. For a summary and discussion of recent research on modes of textual transmission in early China, and particularly the debate about the degree of an oral component in text reproduction, see Scott Cook, The Bamboo Texts of Guodian, 76–82. See also Li Songru, “Zhanguo jianbo ziji yanjiu,” 118–20.

153. For the use of writing in the projection of authority in early China see Lewis, Writing and Authority in Early China, particularly the Introduction and Chapter 1.

154. See, for example, Constance Cook, “Education and the Way of the Former Kings,” in Writing and Literacy in Early China, ed. Li and Branner, 302–36, see 333–35; Robin Yates, “Soldiers, Scribes and Women: Literacy among the Lower Orders in Early China,” in Writing and Literacy in Early China, ed. Li and Branner, 339–69, see 340–45.

155. The oaths of the excavated covenant texts were almost certainly read out during the covenant ceremonies, the covenantors themselves probably also being required to speak the oath (as is hinted at by the mixed use of pronouns within single texts). Mark Lewis discusses collective oaths sworn before hunts and battles, as well as the reading of covenants during the covenant ceremony, see Mark Edward Lewis, Sanctioned Violence in Early China (Albany: State University of New York Press, 1990), 18, 24–25, 46, 67–70.

156. Although primarily concerned with “texts with a transmission history,” Martin Kern's “Methodological Reflections on the Analysis of Textual Variants and the Modes of Manuscript Production in Early China” (Journal of East Asian Archaeology 4.1–4 (2002), 143–81) includes discussion of different modes of “textual reproduction” and “manuscript production” that is relevant to this issue.

157. Chunqiu Zuo zhuan zhu, 969 (“Xiang” 襄 9.5).

158. Chunqiu Zuo zhuan zhu, 435 (“Xi” 僖 25.3).

159. Sun Yirang 孫詒讓, Zhouli zhengyi 周禮正義 (Beijing: Zhonghua, [1987] 2000), 2852–57. For a translation of this section see Williams, “Interpreting the Wenxian Covenant Texts,” 87.

160. Chunqiu Zuo zhuan zhu, 440 (“Xi” 僖 26.3).

161. In general (there are exceptions) the tablets do not appear to have been written in advance leaving a gap where a name would be added later.

162. Both the expectation to attend, and verification of attendance is evident from the occasional tablet which adds a note at the end recording that the covenantor for whom the tablet was prepared was elsewhere (e.g. “X [the covenantor's name] has gone to Y [place name]”). This is clear evidence that when attendance was being verified the covenantor in question had departed, and might suggest that the tablet was also originally prepared in absentia.

163. The suggestion of pre-existing lists is pertinent to the question of the early development of individual and household registration. To date, the earliest excavated examples of such registers are from the Qin period (see Yates, “Soldiers, Scribes and Women,” 359). However, various types of register are mentioned in received texts as early as the sixth century b.c.e. in Chu and the fourth century b.c.e. in Qin (see Lewis, Writing and Authority in Early China, 26–27). Whether or not lists of the names of the covenantors existed prior to the covenants, they surely existed after the covenants. Each pit of tablets itself could be characterized as a registry of names prepared for the sanctioning spirit, and it is reasonable to assume that a list of those covenanting was recorded to be preserved above ground as a record, and possibly also for future administrative purposes.

164. For the requirement in late third century b.c.e. Qin statutes for village leaders to supply lists of adult males in their communities, see Yates, “Soldiers, Scribes and Women,” 360.

165. Crispin Williams, “Ten Thousand Names,” 977–78.

166. It could, of course, equally well reflect a group of relatives lined up in person in front of a single scribe, to whom they then individually gave their names as the scribe wrote the tablets. It seems unlikely that a written copy of the oath was sent out to a multitude of different groups (e.g. kin groups) who were expected to procure and prepare the tablets themselves and bring them along to the covenant ceremony for burial.

167. Li Songru, “Zhanguo jianbo ziji yanjiu,” 71–73.

168. The possibility that two people were involved in the copying process is suggested in part by a Western Jin (281–316 c.e.) ceramic paired-figurine model that Li Songru discusses. The object represents two officials kneeling and facing each other, noses almost touching, one holding a square-shaped tablet of some kind in one hand and writing on it with the other, the other figure holding something (the catalogue description suggests they are bamboo slips) in both hands (Li Songru “Zhanguo jianbo ziji yanjiu,” 73; the object is held by the Hunan Provincial Museum, for an online catalogue entry see: http://www.hnmuseum.com/hnmuseum/collection-info/collection-info!frontCollectionDetail.action?id=15cfcc9c295444f8a81f87541f29e845 (Last accessed April 3, 2014)). The catalogue entry suggests the object represents the act of collation (jiaochou 校讎) of a text, and quotes Liu Xiang 劉向 (79–78 b.c.e.), who described the process as involving two people. Scott Cook also refers to this Han source, suggesting the practice may have developed from an earlier method of textual copying involving two people, one reading the text to the other (Scott Cook, The Bamboo Texts of Guodian, 77).

169. This point is made by James Royse in his study of early Greek scribal habits. It seems that early Greek scribes wrote in the same way, holding the papyrus in one hand and pen in the other. See James R. Royse, Scribal Habits in Early Greek New Testament Papyri (Leiden: Brill, 2008), 98–100. Many of Royse's general points about scribal habits can be applied to the discussion of scribes and scribal error in early China.

170. Royse, Scribal Habits in Early Greek New Testament Papyri, 100, n. 116.

171. As for the identity of the scribes, and their status and training, the tablets do not provide relevant evidence. For a discussion of the training and career of scribes in the Qin and early Han periods, see Yates “Soldiers, Scribes and Women,” 345–60. For related discussion and an analysis of references to scribes in the Zhou li, see Martin Kern, “Offices of Writing and Reading in the Rituals of Zhou,” ed. Benjamin Elman and Martin Kern, Statecraft and Classical Learning: The Rituals of Zhou in East Asian History (Leiden: Brill, 2009), 65–93. For scribal training in the Shang period, see Adam Smith, “Writing at Anyang,” 303–84.

172. See the sections above on the variant with 亡 and those with .

173. William H. Baxter, “Aspects of Old Chinese Morphology,” 6. Quoted with permission of the author (personal communication, May 19, 2014).

174. The covenant texts prepared for burial may, then, be an exception to the observation that excavated texts with a specific administrative or practical function generally have fewer errors than literary texts (see, for example, Li Songru, “Zhanguo jianbo ziji yanjiu,” 99).