FURTHER CONSIDERATIONS FOR THE AUTHENTICATION OF THE PEKING UNIVERSITY CANG JIE PIAN : WITH BRIEF DIGRESSION ON THE SO-CALLED “HAN BOARD” WITNESS

Over the past few years, additional evidence has surfaced that supplements my prior authentication of the Peking University Han strips, and in particular the *Cang Jie pian 蒼頡篇 manuscript from this collection. The present article surveys this fresh evidence, including confirmation of spiraling verso lines on the archaeologically excavated Shuihudi 睡虎地 Han strips, as well as further textual parallels, both with a previously unidentified Cang Jie pian strip from Niya 尼雅, and also with newly published content from the Shuiquanzi 水泉子 *Cang Jie pian. The article also critically examines in greater detail the methodology previously employed to authenticate the Peking University Han strips. Limitations are acknowledged, both in regard to supporting negative appraisals, and in the determination of “novelty” for manuscript features. To demonstrate these issues, the article analyzes a conflict between the Peking University and Shuiquanzi Cang Jie pian manuscripts and introduces the publication of the so-called “Han board” *Cang Jie pian witness largely unknown to the field before. In a previous article, I argued for the authenticity of the *Cang Jie pian 蒼頡篇 bamboo-strip manuscript held by Peking University.1 This assessment was based on the identification of irregular features in the manuscript unattested previously 1. Christopher J. Foster, “Introduction to the Peking University Han Bamboo Strips: On the Authentication and Study of Purchased Manuscripts,” Early China 40 (2017), 167–239. For the publication of the manuscript, see Beijing daxue chutu wenxian yanjiusuo 北京大學出土文獻研究所 ed., Beijing daxue cang Xi-Han zhushu (yi) 北京大學 藏西漢竹書(壹) (Shanghai: Shanghai guji, 2015). The * designates that this title was assigned by the editors of the collection, not written on the manuscript itself. I will only give the * on first mention of a title for a given manuscript. I am very grateful for the aid I have received from many colleagues in preparing this article, and I wish to thank footnote continued on next page © The Society for the Study of Early China and Cambridge University Press 2021. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited. Early China (2021) vol 44 pp 419–464 doi:10.1017/eac.2021.11 *Christopher J. Foster 傅希明, SOAS, University of London; email: cf44@soas.ac.uk. and unanticipated by the state of the field at the time, but then confirmed by archaeologically excavated data first available or fully appreciated only after the Peking University Cang Jie pian was secured. Two such novel features were raised in defense of the Peking University Cang Jie pian: (1) the presence of lines applied extensively across the verso of the manuscript, and (2) textual parallels with content found primarily on the Shuiquanzi 水泉子 *Cang Jie pian manuscript. An analysis of these phenomena indicated with a high degree of confidence that the Peking University Cang Jie pian is indeed a genuine artifact. Over the past few years, study of the Cang Jie pian has benefited from new discoveries and the release of more data. Additional information is now available that supplements my prior authentication of the Peking University Cang Jie pian. In the following, I survey fresh evidence that bolsters this authentication. The evidence includes confirmation of spiraling lines and other unusual features of the verso marks, which were known only from unprovenanced manuscripts before, but are now seen among the archaeologically excavated Shuihudi 睡虎地 Han bamboo strips. Further cases of textual parallels between the Peking University Cang Jie pian and previously unknown Cang Jie pian content are also raised. Examples are drawn from a strip collected at Niya 尼雅 during Aurel Stein’s fourth expedition, and from the full publication of the Shuiquanzi Cang Jie pian manuscript. In light of recent developments, we are also afforded an opportunity to examine in greater detail the methodological assumptions underpinning this assessment. When is a feature both unattested and unanticipated by the state of the field? How do we determine when archaeological data that confirms said features are first known or fully appreciated? What happens if novel features on a unprovenanced manuscript are instead later found to be in conflict with archaeologically excavated artifacts? In order to explicate these points, I examine a conflict between the Peking University Cang Jie pian and recently released content from the Shuiquanzi Cang Jie pian, discussing various possible explanations for the discrepancy. Even more significant, however, has been the announcement that there exists yet another unprovenanced *Cang Jie pian manuscript, held in a private collection and previously unknown to the public. I briefly describe this so-called “Han board” witness and in particular the anonymous reviewers of both this and the prior article for their thoughtful comments, as well as Bai Junpeng 白軍鵬, Cai Dan 蔡丹, Yang Bo 楊博, and Zhang Chuanguan 張傳官 for helping me access publications and data. This research is affiliated with a project led by Yang Bo, entitled “Comprehensive Research on the Writing and Texts of the Peking University Collection of Qin and Han Bamboo-Strip Manuscripts” 北京大學藏秦、漢簡牘文字、文本綜合研究 (YWZ-J020), funded by the PRC Ministry of Education and State Language Commission. This article was completed during a British Academy postdoctoral fellowship. CHRISTOPHER J. FOSTER 420

Over the past few years, study of the Cang Jie pian has benefited from new discoveries and the release of more data. Additional information is now available that supplements my prior authentication of the Peking University Cang Jie pian. In the following, I survey fresh evidence that bolsters this authentication. The evidence includes confirmation of spiraling lines and other unusual features of the verso marks, which were known only from unprovenanced manuscripts before, but are now seen among the archaeologically excavated Shuihudi 睡虎地 Han bamboo strips. Further cases of textual parallels between the Peking University Cang Jie pian and previously unknown Cang Jie pian content are also raised. Examples are drawn from a strip collected at Niya 尼雅 during Aurel Stein's fourth expedition, and from the full publication of the Shuiquanzi Cang Jie pian manuscript.
In light of recent developments, we are also afforded an opportunity to examine in greater detail the methodological assumptions underpinning this assessment. When is a feature both unattested and unanticipated by the state of the field? How do we determine when archaeological data that confirms said features are first known or fully appreciated? What happens if novel features on a unprovenanced manuscript are instead later found to be in conflict with archaeologically excavated artifacts? In order to explicate these points, I examine a conflict between the Peking University Cang Jie pian and recently released content from the Shuiquanzi Cang Jie pian, discussing various possible explanations for the discrepancy. Even more significant, however, has been the announcement that there exists yet another unprovenanced *Cang Jie pian manuscript, held in a private collection and previously unknown to the public. I briefly describe this so-called "Han board" witness and in particular the anonymous reviewers of both this and the prior article for their thoughtful comments, as well as Bai Junpeng 白軍鵬, Cai Dan 蔡丹, Yang Bo 楊博, and Zhang Chuanguan 張傳官 for helping me access publications and data. This research is affiliated with a project led by Yang Bo, entitled "Comprehensive Research on the Writing and Texts of the Peking University Collection of Qin and Han Bamboo-Strip Manuscripts" 北京大學藏秦、漢簡牘文字、文本綜合研究 (YWZ-J020), funded by the PRC Ministry of Education and State Language Commission. This article was completed during a British Academy postdoctoral fellowship. offer initial thoughts on how the existence of this manuscript impacts authentication of the Peking University Cang Jie pian.

Spiraling Lines and Other Verso Features on the Shuihudi Han Strips
The existence of verso lines in and of themselves helps to prove the authenticity not only of the Peking University Han manuscripts, with the Cang Jie pian among them, but also that of other collections of unprovenanced manuscripts, including those held by the Shanghai Museum, Tsinghua University, and Yuelu Academy. 2 Before elaborating, however, please note that I am no longer referring to these collections as "purchased manuscripts." The reason for calling them such in my prior article was to avoid assumptions about the manuscripts' origins and authenticity, as is the case when using the term "looted manuscripts." 3 I am grateful for the critique of this practice by one of the anonymous reviewers, who rightfully points out that using "purchased manuscripts" risks treating these objects foremost as economic commodities and thereby unduly emphasizes their financial value. This in turn may appear to legitimate trade in illicit antiquities, while also downplaying the manuscripts' complex histories and ethical entanglements. This is not my intent, and to avoid this I now adopt the term "unprovenanced manuscripts," following other recent articles that discuss artifacts of this nature. 4 To summarize the prior discussion of the verso lines: Explicit mention is made of verso lines in the 1991 report on the Baoshan 包山 Chu strips, and photographs of various strips' versos with these marks were published infrequently before Peking University's acquisition of the Han strips in January 2009. 5 Yet the extensive presence of verso lines was only fully appreciated late in 2010, following the publication of complete verso photographs for a number of Tsinghua University and Yuelu Academy manuscripts, and the research conducted by Sun Peiyang 孫沛陽 and others on both unprovenanced and archaeologically excavated manuscripts alike. 6 The verso lines therefore may be taken as a novel feature for the Peking University Han strips and for collections acquired before 2010. 7 That verso lines now are widely documented on archaeologically excavated manuscripts, whether newly identified on old caches or newly unveiled on the latest finds, likewise supports the authenticity of the Peking University Han strips and these other unprovenanced collections. 8 This argument hinges in large part on the claim that verso lines were only "fully appreciated" late in 2010. It is an assertion that speaks to both the novelty of the feature for the Peking University Han strips, and also to the viability of raising older caches of archaeologically excavated manuscripts as evidence for authentication. The question ultimately becomes, "who knew what, when?" And the answer to this question must take into account both the accessibility of information and its understanding. While a threshold of 2010 is, in my opinion, uncontroversial-and indeed substantiated once again by the editing of the Shuihudi Han strips, as will be shown below-the fact that limited mention of the phenomenon and partial data had been published earlier may invite skepticism. The verso marks are a challenging case, in that the raw data under discussion has been present on manuscripts unearthed long before the Peking University acquisition (a matter of accessibility to information), but largely neglected until only recently (a matter of its understanding).
It was noted before that the Peking University Han strips and other unprovenanced manuscripts, primarily those among the Tsinghua University and Yuelu Academy collections, bear unusual characteristics to their verso marks. 9 One such characteristic is that verso marks can 6. Sun Peiyang 孫沛陽, "Jiance bei hua xian chutan" 簡冊背劃綫初探, Chutu wenxian yu guwenzi yanjiu 2011.4, 449-62.
7. The verso lines cannot be used as evidence to authenticate collections acquired after the 2010 threshold, however, as we must assume that, from this point onward, forgers may have known to replicate this feature. In other words, the feature is no longer novel. This is consistent with my proposed methodology. Each manuscript, of course, needs to be treated individually, assessing what information is novel and the current state of knowledge in our field at the time of its discovery. I clarify this in response to form "sets," likely fashioned from the same bamboo culm, where the carved line spirals, connecting from the last strips of the set back to the initial strips in the same set. 10 Another characteristic is the occasional rearrangement of strip order between when the verso marks were produced and the manuscript was finalized for writing. This results in abnormalities to otherwise continuous verso lines, including gaps, buffer strips, the displacement of line sections, or the "reversedangled steps" phenomenon. 11 Similarly, reorientation of bamboo strips after production of verso marks appears to have occurred as well, for example with strips rotated 180° before writing, vertically "flipping" the presentation of verso marks. 12 None of these phenomena-like the verso lines themselves-were fully appreciated when Peking University first acquired their Han strips. Yet, unlike the mere existence of verso marks, most of these additional characteristics could only be distinguished upon examination of the complete versos of bamboo-strip manuscripts bearing extensive line carvings or drawings. Neither the statement in the Baoshan report, nor prior photographs of strips' versos, reveal enough information to anticipate the presence of verso mark "sets" with spiraling lines or many of the other anomalies listed above. 13 It is possible that the rotation of strips (whether vertically or horizontally) during some stage 11. The reversed-angled steps and other phenomena are discussed in detail by Thies Staack in: "Identifying Codicological Sub-units in Bamboo Manuscripts," see esp.159-60. For an example, see Tsinghua's *Chu ju 楚居 manuscript, strips 3-6. Additional citations may be found in Foster, "Introduction to the Peking University Han Bamboo Strips," 207, n. 115.
12. This might be another explanation for the appearance of reversed-angled steps, see Staack, "Identifying Codicological Sub-units in Bamboo Manuscripts," 160; He Jin 何晉, "Qianyi jiance zhidu zhong de 'xulian'-yi chutu Zhanguo Qin Han jian wei li" 淺議簡冊制度中的"序連"--以出土戰國秦漢簡為例, Jianbo 8 (2013), 463-64. Staack also notes (158, n. 11) that, on the Yuelu Academy Sasi nian zhiri 卅四年質日 manuscript, ink was used for the verso lines, and the traces of that ink at times appear to run "upwards" towards the top of the strip. Staack suggests therefore that when the lines were applied (or shortly after), the strips were upside down compared to their final orientation (with the written text). He gives strips 36 and 37 as examples.
13. In Baoshan Chu jian 包山楚簡 and Baoshan Chu mu 包山楚墓, the editors report that the verso marks "at times may be used to connect adjacent strips, though at other times they are not correlated" (相鄰的簡有的可據此依次相接，有的則互不相關), which implies that not all of the lines are continuous. But this does not hint at the spiraling sets of strips or the other specific anomalies we are seeing. See Hubeisheng Jing Sha footnote continued on next page of a manuscript's chaîne opératoire may be inferred from archaeological specimens, where notches for securing the binding cord were duplicated on both sides of a strip, or carved into otherwise incongruous locations, suggesting differing orientations for the strip in the course of the manuscript's production. 14 Even this, however, is a very tenuous inference, and must further relate the carving of notches relative to that of the verso lines in the manuscript's chaîne opératoire.
The requisite data for appreciating these features was not publicly disseminated before the Peking University acquisition, limiting the accessibility of this information. Here a distinction needs to be drawn between public and potential private dissemination. We cannot rule out that caretakers of excavated manuscripts accessed this sort of data, or privately shared it with other individuals. There is, however, indirect evidence that even the editors of many of these manuscript caches remained unaware of the verso marks and their importance until the 2010 threshold. This is demonstrated clearly in the editing of the Shuihudi Han strips, presented below. Consider, as well, the statement by Chen Songchang 陳松長 that when we were first editing the [Yuelu Academy] Wei li [zhi guan ji qianshou 為吏治官及黔首 manuscript], we had only a rudimentary understanding of the features present on the strips' versos. Although we later took photographs of the strips' versos with an infrared scanner … at that time we still had a limited appreciation of the verso lines and similar data, with our attention falling solely on any writing present on the back of the strips. 15 Owing to the general inaccessibility of data, and the lack of its appreciation by the limited number of caretakers who might have had access, confirming characteristics like the verso sets with spiraling lines on archaeologically excavated artifacts provides even greater confidence in the authenticity of the Peking University, Tsinghua University, and Yuelu Academy manuscripts. 16 The Liu nian zhiri 六年質日 manuscript unearthed from Shuihudi Han tomb M77 offers the first published photographs of an archaeologically excavated artifact exhibiting a variety of the above verso mark characteristics. In November of 2006, during construction work on the Han-Dan railway line, a hole was accidently drilled into a grave, now labelled M77, within the Shuihudi cemetery, in Yunmeng 雲夢 County, Hubei. 17 Inside this grave was a bamboo basket containing at least 2,137 bamboo strips and tablets. 18 Remarkably, the majority of these strips were not significantly displaced from the positions in which they were deposited originally, when bound as scrolls and stacked together. This has aided immensely reconstruction of individual manuscripts. 19 Only those on the east side of the basket suffered damage from the drilling, breaking a small number of strips and dispersing the pieces. Contents of the cache include event calendars, legal statutes, daybook materials, a mathematical treatise, anecdotal narratives featuring various famous historical figures, and accounting records. 20 A comparison of the tomb 16. Foster, "Introduction to the Peking University Han Bamboo Strips," 221. 17. Hubei sheng wenwu kaogu yanjiusuo 湖北省文物考古研究所 and Yunmeng xian bowuguan 雲夢縣博物館, eds., "Hubei Yunmeng Shuihudi M77 fajue jianbao" 湖 北雲夢睡虎地 M77 發掘簡報, Jiang Han kaogu 2008.4, 31-37, images available on plates 11-16. I will refer to this as the "brief report" in the following.
18. The brief report lists at least 128 intact or mostly intact tablets among this count, made from both bamboo and wood. Furthermore, while the 2,137 number cited does include broken pieces, the editors also write that there are several thousand smaller fragments in addition to these strips (Hubei sheng wenwu kaogu yanjiusuo and Yunmeng xian bowuguan, eds., "Hubei Yunmeng Shuihudi M77 fajue jianbao," 35). In an introductory article for the manuscripts, the editors again assert that the number of 2,137 is certainly lower than the eventual final count. structure and its contents to other previously excavated graves suggests that the burial is from the early Western Han period, which is confirmed by the dates on the event calendars and accounting records. The burial must have taken place shortly after the tomb occupant, Yue Ren 越人, passed away in 157 b.c.e. 21 The brief report for Shuihudi M77 does not mention verso marks appearing on any of the strips. 22 In 2010, Xiong Beisheng 熊北生 published an article on the methodology employed by the editors to arrange the manuscripts in this cache. 23 Xiong emphasizes how they relied heavily on the strips' in situ placement, checked against the content of the texts. Once again, no mention is made of the verso marks, or of how they could be used to establish strip order. This, in particular, suggests to me that the editors of the Shuihudi Han strips still were not aware of the verso marks when preparing the article. The timeframe is consistent with a late 2010 threshold for the full appreciation of this feature by the field, on unprovenanced and archaeologically excavated bamboo-strip manuscripts alike. 24  24. That verso marks are not mentioned in prior surveys of the Shuihudi Han strips could be happenstance. For instance, the 2018 introduction to the manuscripts does not discuss this feature (Xiong, Chen, and Cai, "Hubei Yunmeng Shuihudi 77 hao Xi-Han mu chutu jiandu gaishu," 43-53). My point, however, is that if the verso marks were understood and employed to arrange the Shuihudi Han manuscripts, then this feature undoubtedly should have been identified in an article dedicated to the methodology behind editing the cache.
features of a complete verso (sans writing) of an archaeologically excavated manuscript, retrieved through controlled conditions, has been subjected to a thorough analysis. 25 The Liu nian zhiri manuscript is an event calendar for the sixth year following Emperor Wen's 漢文帝 "second inaugural year 後元," which corresponds to 159-158 b.c.e. It joins a series of other event calendars entombed in M77, the entire series amounting to 719 strips and over a thousand smaller fragmented pieces altogether. 26 Each calendar lists the months for the given year, separated into those that are even-numbered first, then those that are odd-numbered. Whether a month is da 大 "large" (30-day) or xiao 小 "small" (29-day) is noted. Sexagenary ganzhi 干支 day counts are enumerated for every month. Important events are recorded next to the day on which they occurred, and include seasonal activities (e.g., the La 臘 and Setting-Out-Seeds 出種 festivals), official business (e.g., when Yue Ren "attends to matters" or shishi 視事), and private affairs (e.g., the death of a parent, or travel conducted by a daughter). 27 25. Cai Dan, "Shuihudi Han jian zhiri jiance jianbei huaxian chutan" 睡虎地漢簡質 日簡冊簡背劃綫初探, Jiang Han kaogu 2018.4, 125-27, 82. For a brief mention and handdrawing of a potentially spiraling verso line among the Tianhui 天回 Laoguanshan 老官山 cache of Han strips, made slightly prior to Cai's discussion, see Zhongguo Zhongyi kexueyuan Zhongguo yishi wenxian yanjiusuo 中國中醫科學院中國醫史文 獻研究所, Chengdu wenwu kaogu yanjiuyuan 成都文物考古研究院, and Jingzhou wenwu baohu zhongxin 荊州文物保護中心, "Sichuan Chengdu Tianhui Han mu yijian zhengli jianbao" 四川成都天回漢墓醫簡整理簡報, Wenwu 2017.12, esp. 50-51. I thank Thies Staack and Chun Fung Tong for bringing these articles to my attention.
26. The event calendars range from the tenth year following Emperor Wen's "first inaugural year" 前元 (170 b.c.e.) to the seventh year following his "second inaugural year" (157 b.c.e.). Cai, Chen, and Xiong, "Shuihudi Han jian zhong de zhiri jiance," 54-64; Chen Wei, "Shuihudi Han jian Zhiri de shiliao jiazhi" 睡虎地漢簡《質日》 的史料 價值, Zhongguo shehui kexuebao 中國社會科學報, Zhongguo shehui kexuewang 中國社會 科學網, October 8, 2019, www.cssn.cn/zx/bwyc/201910/t20191008_4980224.shtml; and Xu Mingcang 許名瑲, "Shuihudi Han jian zhiri jiance canjuan liri fuyuan shini" 睡虎地 漢簡質日簡冊殘卷曆日復原試擬, Wuhan daxue jianbo yanjiu zhongxin jianbo wenku 武 漢大學簡帛研究中心簡帛文庫, 27. The Liu nian zhiri and Qi nian zhiri 七年質日 uniquely designate seasonal markers like the winter and summer solstices (dongzhi 冬至 and xiazhi 夏至) or the beginning of spring and fall (lichun 立春 and liqiu 立秋). The Liu nian zhiri likewise records when Yue Ren's mother fell ill, died, and was buried. For this, and entries on Yue Ren's daughter, see Xiong, Chen, and Cai, "Hubei Yunmeng Shuihudi 77 hao Xi-Han mu chutu jiandu gaishu," 44. On the relationship between state control and personal identity, and how this complicates designations of such calendars as "public" versus "private" documents, see Chen, "'Event Calendars' in the Early Imperial Era." The Liu nian zhiri is written over sixty-two bamboo strips. This strip number is consistent across the Shuihudi Han event calendars. On each calendar, a single strip is used to list the even-numbered months, followed by thirty strips, one for each sexagenary day count, accommodating for "large" months; another single strip is then used to list the odd-numbered months, again followed by thirty more strips for the sexagenary day counts. Should there be a runyue 閏月 "intercalary month," six more strips are appended (for sixty-eight total), in order to label the month and give its day counts. The Liu nian zhiri strips measure between 26 and 31 cm in length, and were bound by three cords. 28 Faint horizontal lines are carved across the recto, which together with the binding cords, form six rows, where the scribe isolated day counts to their corresponding months. We may be assured that the strip order presented for the Liu nian zhiri is correct, owing both to the fact that the strips were found relatively undisturbed in situ, and to the formulaic structure of its contents-an event calendar with sequential dates. This is a great boon for the study of the Liu nian zhiri verso lines. With the strip order thus secure, we may now investigate the verso marks for a (relatively) complete manuscript in its final form (see Figure 1). 29 There are obvious discontinuities to the verso lines on the Liu nian zhiri. For example, there is a shift in height and gradient to the verso marks when compared between strips 19 and 20. To explain this shift, Cai argues that this is actually a divide between two separate sets of verso lines, with the line on strip 19 "spiraling back" to connect with the marks at the beginning of the manuscript, instead of continuing onto strip 20 (see Figures 2a and 2b). Indeed, should we place strips 28. The physical dimensions of the Liu nian zhiri discussed below are drawn from the general description for the event calendars given in Cai, Chen, and Xiong, "Shuihudi Han jian zhong de zhiri jiance," 54-55.
29. Unfortunately, the photograph of the Liu nian zhiri manuscript verso published in Jiang Han kaogu is of too low a resolution to make out all of the marks clearly. I thank Cai Dan for graciously supplying me with the high resolution version of the same photograph. Figure 1 is after this image, with strip numbers added and the verso marks highlighted. Cai's image includes boxes around the locations of verso marks. On four occasions, I highlight a verso mark which Cai does not "box": on strip 1, at the bottom of strip 18, and in the upper rows of strips 59-60. Cai discusses marks present in all these locations, and I include details of these locations in Figure 2 for the reader's scrutiny. Furthermore, note that Cai's box on strip 52 appears to have been done in white as opposed to black; this gives the impression in Figure 1 that no box is present. Finally, the greyed rectangles are either missing strips or, in the case of 24, a strip on which the verso mark cannot be ascertained (due to obstruction from its encasing for conservation-another sign that verso data was ignored initially in the cache's editing). Readers should take into account that I have not yet been able to personally examine the manuscript, in order to confirm the presence of these marks.      16-19 before strip 1, the verso line is aligned nearly perfectly. I say "nearly" because Cai posits a missing strip before 1, but I do not believe this is necessary. Formal measurements will help clarify this point, though an analysis of the photograph (using pixels to estimate relative distances) suggests that there need not be a missing strip. For the sake of completeness, note also that Cai describes strips 14 and 15 as having two verso marks, upper and lower. These are not boxed in the published image, and their existence is difficult to ascertain with any certainty from the photograph alone. 30 Based on similar observations, Cai divides the manuscript into four sets of strips: set 1 spans strips 1-19; set 2 spans strips 20-38; set 3 spans strips 39-42; and set 4 spans strips 43-62. Of these, set 3 contains four strips without discernable verso marks. The other three sets, Cai believes, originally consisted of around twenty strips, bearing engraved lines that spiraled around the culm tube. Cai offers a reconstruction for the other sets (1, 2, and 4), showing that at certain points after the engraving of the verso line for a given set, but before finalization of the manuscript, strips were either displaced or rotated 180° vertically. 31 An obvious example of the latter is provided by strip 37, which when rotated back 180° vertically, has verso marks that connect smoothly with both the upper and lower lines in its set. For the manuscript's verso according to Cai's reconstructions, see Figure 3. 32 Caution is merited in treating the Shuihudi Han Liu nian zhiri as archaeologically obtained evidence of verso features only known previously on unprovenanced manuscripts. Take for example the 30. It is possible that Cai intended to write "strips 16-19" instead, which would also correct her erroneous count of "four strips" for "strips 14-19" ("Shuihudi Han jian zhong de zhiri jiance jianbei huaxian chutan," 125).
31. Yet these sets of bamboo strips, on the whole, were kept intact through to the binding of the manuscript. This indicates that carving the verso lines and finalizing the manuscript were either localized steps in the chaîne opératoire of the Liu nian zhiri, potentially carried out by a single individual, or that great care was taken to ensure strips from the same verso-line sets remained together during the transportation of blank strips from a given workshop to their scribe end-user. For Cai's hypotheses on how the verso marks functioned in the production of bamboo-strip manuscripts, see "Shuihudi Han jian zhong de zhiri jiance jianbei huaxian chutan," 127.
32. The reconstructions are as follows: For set 1, a strip is missing after 19. I do not include this in Figure 3. For set 2, strip 24 is removed; strip 27 is placed after strip 29; and strip 37 is rotated 180° vertically. Cai suggests that three strips could be missing after 38, with 24 potentially among them. I have not included these in Figure 3. For set 4, strip 43 is moved to after 52; strip 44 is moved before strip 57; strip 62 is rotated 180° vertically and placed after 56. Note that issues seem to persist even after Cai's reconstruction, when consulting the published photograph. The absence of boxes around declared marks is discussed in n. 29 above. Another apparent issue is the height of the verso mark on strip 55, which is incongruous with those to the rest of the line in set 4. In this case, the problem stems from the relative positioning of strips 54-56. Cai has arranged the strips in the photograph so that their bottom ends are flush together. My suspicion is that strip 55 should be shifted upwards, however, as this is suggested by the binding cord remnants on 55, vis-à-vis those on 54 and 56. I thank Stuart Middleton for his immense help in producing the image seen in Figure 3.

Figure 3
Reconstruction of the Liu nian zhiri verso, after Cai, "Shuihudi Han jian zhiri jiance jianbei huaxian chutan." Speculated missing strips not added. spiraling lines. In set 1, the alignment of the marks on strips 19 and 1 is very close. Though perhaps unnecessary, Cai does suggest that one strip is missing between them. For set 2, however, the connection is much more tenuous, with at least three strips now required to link the upper mark on strip 38 back to that of strip 20. In set 4, we have perfect alignment between the marks on strips 61 and 45, but Cai's reconstruction requires various manipulations to arrive at this point, moving multiple strips around to different positions. If allowed to speculate over missing strips or manipulate known pieces at will, without reasoned justifications, then we risk simply inventing whatever imagined scenario we desire to see in the data.
While acknowledging these issues, I agree with Cai that the Shuihudi Liu nian zhiri manuscript provides the first confirmation of sets of spiraling verso lines and these other features on archaeologically excavated artifacts. Although the Shuihudi Han strips were unearthed just over a year before the acquisition of the Yuelu Academy Qin strips, and around two years before the Tsinghua and Peking University acquisitions, the Shuihudi verso data was not published until quite recently. It is therefore stronger evidence for the authenticity of unprovenanced manuscripts from these caches bearing similar verso features (e.g., the spiraling lines), than the presence of verso marks alone. Unfortunately, while this raises our confidence in the Peking University Han strips as a whole, it is unclear if the Cang Jie pian manuscript in specific bears spiraling verso lines, nor does it bear vertically rotated verso marks. 33 Other evidence must be raised for the evaluation of this specific manuscript then, primarily newly identified textual parallels, to which we now will turn.

Textual Parallels: A Strip from Niya Collected During Aurel Stein's Fourth Expedition
In 王, helping to confirm the Niya site as the Jingjue Kingdom. 36 Stein was not permitted to take the artifacts he collected out of China, so he deposited them in the British consulate in Kashgar. The strips he collected were later presented to local authorities, then unfortunately lost. Before leaving, however, Stein took photographs of the strips (with the aid of the British Consul George Sherriff), with the negatives and prints deposited in various institutions; soon these too were forgotten by scholarship. Through the efforts of Wang Jiqing 王冀青, sets of Stein's fourth expedition photographs have been rediscovered in the British Library, and currently stand as our only evidence for an important, if small, collection of Han period manuscripts.
Unable to consult the actual artifacts, we must depend upon the photographs that remain. The clarity of the photographs is remarkable, in light of the circumstances under which they were produced, yet they are far from ideal. Stein himself, in fact, felt dissatisfied with the images, and hired Thomason College of Civil Engineering at Roorkee, India, to make "improved" versions of the glass negatives. As Wang Jiqing has discussed, this improvement entailed someone, likely ignorant of Chinese, mechanically retouching the characters' strokes, and ultimately such efforts proved to be more of a hindrance than an aid to the strips' decipherment. The British Library appears to hold not only the "improved" glass negatives, but also a complete duplicate set of negatives (made via exposure on direct duplicating film) and gelatin silver prints that were based on both the nitrate originals and glass negatives. The strip that concerns us here is N.XIV.20, but the precise source of the published photographs of N.XIV.20 awaits confirmation. 37 It is no surprise, therefore, that scholars have disagreed over the transcriptions of these manuscripts. Furthermore, N.XIV.20 is damaged, with most characters missing portions of their right components, which uniquely frustrates transcription of its text. In his initial study of Stein's fourth expedition Han strips,  40 Despite the damage to the right side of N.XIV.20, partial strokes often remain, hinting at the orthography of the missing right components. Characters 2, 3, 4, 5 and 7 on N.XIV.20 all match up well with the components found on PKU 40 for the corresponding characters. Character 4 on N.XIV.20 may bear an additional horizontal stroke at its top, and the clarity of 5 is less than ideal, but the parallels for both are discernable. In the case of character 1 on N.XIV.20, the upper portion of the character is indistinct, yet the two angled strokes, running in opposite directions, are suggestive of the top component to ju 䡞 on PKU 40. For 6 on N.XIV.20, it is the left side that is blurry, but the right component matches dian 點 on PKU 40. This string of text is non-grammatical, yet it fits the typical format we see for Cang Jie pian text. It is not found in any received or excavated works other than the Peking University Cang Jie pian. 41 Furthermore, there is precedent for the existence of Cang Jie pian materials at Niya, as one other Cang Jie pian strip was identified at this  This identification further supports the authenticity of the Peking University Cang Jie pian. Although Stein collected this piece close to a century ago, the artifacts have since disappeared, with the photographs likewise long neglected in foreign institutions. Wang Jiqing first published these photographs in 1998, but the initial transcriptions of N.XIV.20 have been misleading; for this reason, no scholar has since raised the possibility that this is Cang Jie pian content. 43 We may thus treat the text here as an unattested and unanticipated feature of the Peking 42. Wang Yue 王樾, "Lue shuo Niya faxian de Cang Jie pian Han jian" 略說尼雅發現 的蒼頡篇漢簡, Xiyu yanjiu 1998.4, 55-58. Wang actually introduces two strips (labelled N14:1 and N14:2), but only the first was clear enough for Wang to read. This strip, N14:1, was broken, but can be rejoined together. Its recto reads: 谿谷阪險丘陵故舊長緩 肆延涣囗, which parallels content found on PKU 4 and other witnesses. On N14:1's verso are three characters, which Wang transcribes as: 人仝人. The strip was picked up by a worker during a Sino-Japanese expedition in 1993.  University Cang Jie pian now confirmed on an archaeological specimen, only fully appreciated after the Peking University cache was secured. 44 The strength of this evidence is similar to that of the presence of verso marks: we have public dissemination of partial data before acquisition of the Peking University Han strips, yet this information was not yet understood by the specialist discipline. 45 The discovery that N.XIV.20 belongs to the Cang Jie pian has broader significance beyond just the authentication of the Peking University manuscript as well. Studied together with the one other Cang Jie pian strip previously recovered from Niya, it is now possible not only to confirm details about these strips' prior deposition on site, but also show that multiple chapters of the Cang Jie pian circulated at Niya. The presence of extensive content from an important scribal primer in the ostensibly foreign kingdom of Jingjue 精絕 bears witness to Han diplomacy and warrants closer investigation. 46 When studying unprovenanced manuscripts, it is important to acknowledge the destructive impact of looting, and to highlight the consequential loss of archaeological context. N.XIV.20 offers a particularly clear example in this regard: although the strip is interesting enough as Cang Jie pian content in isolation, it becomes vastly more significant when placed within its archaeological context of Han period Niya.

Textual Parallels: Newly Published Shuiquanzi Cang Jie pian Content
Although the identification of parallel content between N.XIV.20 and PKU 40 strengthens the authentication of the Peking University Cang Jie pian, it is burdened by the public dissemination of data prior 44. Stein kept meticulous dairies and attempted to record his surveys with the detail appropriate to an archaeological expedition of his time. Yet the propriety of Stein's methods, especially in this instance, may be questioned. N.XIV.20 was picked up by a worker, most likely Abdul Ghafar, whom Stein had sent to the site surreptitiously. Wang, "Sitanyin disici Zhong-Ya kaocha suohuo Han wen wenshu," 273, 275.
45. We are left, therefore, with the unlikely possibility that (1) the physical strip still exists somewhere and was available to a potential forger, or that they noticed photographs of this broken strip sometime after 1998; and (2) that said forger then tailored the eventual PKU Cang Jie pian manuscript, so that the text on PKU 40 both accords with the fragmentary writing preserved on N.XIV.20, and does so in a way in which the content might make sense in a scribal primer like the Cang Jie pian, defying the then current expert transcriptions given for N.XIV.20.
46. Wang, "Lue shuo Niya faxian de Cang Jie pian Han jian," 58; Fu, "Sitanyin disici Zhong-Ya kaocha suohuo Niya yizhi Cang Jie pian Han jian yi mei." to the Peking University manuscript's acquisition. The Shuiquanzi Cang Jie pian, on the other hand, constitutes archaeologically excavated data that was not available to a potential forger before the Peking University acquisition, as it was unearthed just before Peking University secured its Han strips, and was published-still partially-long afterward. 47 For this reason, the Shuiquanzi Cang Jie pian remains our strongest body of evidence against which we may compare the content of the Peking University Cang Jie pian, at least until new manuscripts are unearthed. 48 In 49. Zhang Cunliang "Shuiquanzi Han jian Cang Jie pian zhengli yu yanjiu." These overlaps do not take into account the Han board witness, which will be discussed below.
50. One major contribution of Zhang's dissertation is that it conveniently consolidates labels for the Shuiquanzi Cang Jie pian cache, which varied across prior publications. Citations to individual Shuiquanzi strips follow the labels and transcriptions found on Zhang, "Shuiquanzi Han jian Cang Jie pian zhengli yu yanjiu," 55-59. The Peking University Cang Jie pian transcriptions follow those in Beijing daxue cang Xi-Han zhushu (yi). Here I use the interpretative character forms, though slight differences may arise when considering the strict transcriptions given in both works, including added, missing, or moved components. I also drop paratextual information, such as chapter titles. Recall that the Shuiquanzi Cang Jie pian adds three-character "extensions of meaning" after the four-character base lines of Cang Jie pian text; this footnote continued on next page Other unique parallels may be found between PKU 46 and 47 vs SQZ C024; PKU 52 vs SQZ C066; PKU 63 vs SQZ C082; 53 and PKU 64 vs SQZ C108. 54 To reiterate, the Shuiquanzi Cang Jie pian was recovered during fieldwork that took place from August to October 2008. According to Zhang Cunliang, they completed an initial set of transcriptions for the Shuiquanzi strips on the eve of the National Day 國慶節 holiday that fall. 55  54. SQZ C108 can be re-joined to SQZ C092, which provides further overlapping content with PKU 64. This parallel was noted in Foster, "Introduction to the Peking University Han Bamboo Strips," 227.
55. Zhang and Wu Hong 吳葒 had also taken photographs of the strips. There is no mention that this data was shared with others, until the preliminary report was released in 2009. See Zhang, "Shuiquanzi Han jian qiyanben Cang Jie pian lice," 61. data, as the preliminary report for the Shuiquanzi cache was published at the end of 2009. Private dissemination of this data was possible, but only for a very limited window of time (October 2008 to January 2009), in which representatives of Peking University probably had some knowledge of the Han strips for sale. For these reasons, confirmation that novel content in the Peking University Cang Jie pian is also found on the archaeologically excavated Shuiquanzi Cang Jie pian serves as strong proof of the former's authenticity.

Complications to the PKU 65 and SQZ C058 Parallel and their Methodological Significance
In my first survey of the Peking University and Shuiquanzi Cang Jie pian witnesses, I raised an overlap between PKU 65 and SQZ C058 as one example of parallel novel content. 56 There are two complications with this parallel that belatedly have come to my attention. Addressing these issues can help clarify the limitations of my approach, and provides an opportunity to discuss the methodological assumptions underpinning it. The text of PKU 65 and SQZ C058 is compared in Figure 7. Certain conditions must be met in order for this parallel to serve as evidence for the authenticity of the Peking University Cang Jie pian.
To raise the Shuiquanzi manuscript as archaeological confirmation of the content, a timeline must be established showing that this data was accessible only after the Peking University acquisition, as outlined above. For the Peking University data, it is necessary to show that the content is unattested previously and unanticipated by the state of the 56. Foster, "Introduction to the Peking University Han Bamboo Strips," 227, where SQZ C058 is identified as SQZ2_T:63 (SQZ2_P:45) 57. Zhu Fenghan's strict transcription for yang 卬 is ; while jue 蟨 is from 广 as the upper component. For the overlapping characters, Zhang Cunliang's most recent analysis of the Shuiquanzi piece gives ⿰彳⿺辶 as the strict transcription for ying 迎; he also gives ⿸广欮 for jue 厥. There are additional updates to Zhang's readings at points beyond these three parallel characters, but they do not require detailed comment for our purposes. field before the manuscript was secured. In this instance, that novel feature is the content found on PKU 65, and the final characters in particular: 堯舜禹湯顡卬[趮] [蟨] (which reads "Yao, Shun, Yu, Tang, resolute, look towards, [restless], [hurried]"). 58 The first complication concerns the novelty of this content.
How can we demonstrate that this feature is both unattested and unanticipated? Demonstrating novelty (in both regards) demands, in a certain sense, the notoriously problematic task of proving an absence. 59 For the present case study, my approach was (1) to survey all published wood and bamboo-strip manuscripts unearthed prior to the Peking University acquisition for potential parallel content; (2) to check my findings against other scholarly compilations of Cang Jie pian manuscripts; and finally (3) to review scholarship in the field more broadly for discussions that may be relevant to the parallel in question. 60 The massive quantity of materials to survey, both among primary sources and secondary scholarship, together with the possibility of unpublished data or data otherwise difficult to access, raises the probability of overlooking evidence.
Such was the case in my evaluation of the PKU 65 and SQZ C058 parallel. Among the richest caches of manuscripts with Cang Jie pian content are the hundreds of shavings collected by Aurel Stein during his second expedition to China, now held in the British Library (strip labels: YT). 61 As noted before, when Édouard Chavannes first organized the Han strips collected by Stein, he prioritized publishing only the most legible specimens in his Les Documents Chinois, leaving out these shavings. This omission was rectified by the Yingguo guojia tushuguan cang Sitanyin suohuo weikan Han wen jiandu 英國國家圖書 58. My translation of PKU 65 here is based on Zhu Fenghan's annotations. Zhu reads wai 顡 as yi 毅 ("resolute") and jue 蟨 as 蹶 ("rushed"). The reason that the 禹湯 顡 parallel is not treated as a novel match in the previous survey is because this content also appears on YT 2780 and YT 3251, which Hu Pingsheng identified as potential Cang Jie pian materials already in the British Library volume (see "Yingguo goujia tushuguan cang Sitanyin suohuo jiandu zhong de Cang Jie pian canpian yanjiu" 英國國家圖書館藏 斯坦因所獲簡牘中的《蒼頡篇》殘片研究, 73). This identification was published before the Peking University acquisition.
59. To be more specific, it entails proving an absence before the existence of the manuscript bearing the feature in question. With unprovenanced manuscripts, like the Peking University Cang Jie pian, this timeline inevitably will be imprecise. The most conservative approach is to use the date the manuscripts arrived at Peking University and may be regarded as secured. 館藏斯坦因所獲未刊漢文簡牘 (aka Yingtu) volume published in 2007, but only in part. 62 It turns out that, once again, over a hundred shavings inadvertently were left out of the Yingtu volume as well. The International Dunhuang Project website included photographs for the missing shavings in its database, but more formal dissemination of the data only occurred with articles released during the summer of 2016. 63 Among them is an important piece, YT 1852, with parallel content related to PKU 65 and SQZ C058, not mentioned in my prior discussion: 禹湯㥟卬廣厥賓分 笵□ . 64 My previous neglect of this strip highlights a limitation in my approach: since demonstrating novelty demands proving an absence, all data must be accounted for and ruled out, which is unrealistic or often impossible. As a result, our evaluations must be constantly revisited and Before elaborating upon the significance of YT 1852, let me note that this limitation to demonstrating novelty applies to both the attestation and anticipation of the feature in question, though in slightly different ways. While the example above pertains mostly to prior documentation of the actual feature (i.e., finding the same content written on other Cang Jie pian manuscripts already unearthed), it is also important to ensure that a forger could not have reasonably anticipated the feature's existence (on genuine artifacts, perhaps yet to be found) based on the state of knowledge in the field at that time. For example, imagine that the line on PKU 65 instead read: 堯舜禹湯文武成康 (listing out the proper names "Yao, Shun, Yu, Tang, Wen, Wu, Cheng and Kang"). Even if archaeologists discovered Cang Jie pian manuscripts with this same content in the future, it offers only weak evidence for the authenticity of our re-imagined Peking University manuscript. This is because a forger may anticipate pairing the sages Yao, Shun, Yu and Tang, with the names of the first Zhou kings, Wen, Wu, Cheng and Kang. Indeed, other early received texts such as the Mozi 墨子, often list Wen and Wu after Yao, Shun, Yu and Tang; similarly, other works, like the Shang shu 尚書 and Zuo zhuan 左傳, discuss Wen, Wu, Cheng and Kang together as a unit. 66 The reason that YT 1852 is noteworthy for the PKU 65 and SQZ C058 overlap, however, is not because it greatly impacts our evaluation of the content's novelty. While the SQZ corpus is the best archaeological check for authenticating the Peking University Cang Jie pian to date, certain content among the British Library corpus was not fully appreciated before Peking University acquired their Han strips. This was my argument for the YT 3559 and PKU 3 overlap in the prior article, echoing my analysis of the Niya strip above, and it equally applies to YT 1852 here. 67  second complication: that the Peking University Cang Jie pian conflicts in various ways with newly excavated or appreciated archaeological data for these lines.
Our three witnesses read as shown in Figure 8 (removing the [] from the PKU 65 transcription for 趮蟨).
There are certain character variants present on YT 1852. Most significantly, on PKU 65 the final two characters, while fragmentary, clearly have 目 as a left component. But on YT 1852, the characters in the parallel positions instead read bin fen 賓分, neither of which being written from 目 or a graphically similar left component. There are, moreover, other pieces among the British Library cache that write out bin fen fan 賓分笵, such as the gu prism YT 1791, suggesting that this is not simply an isolated occurrence on YT 1852. 68 A less obvious but perhaps more troublesome conflict involves the structure of the Cang Jie pian text and, specifically, where the Peking University and SQZ manuscripts suggest line divisions. Although PKU 65 is fragmentary, ample blank space remains on the strip below ⿰目□⿰目□, while the bottom end of the piece is also level. A notch is also recorded at the top of the fragment, with partial writing remaining just above it as well, which means it is most likely the middle notch. 69 This indicates that the fragment was once the bottom of a strip, with ⿰目□⿰目□ concluding the text on this specific writing support. The Peking University Cang Jie pian is strictly formatted: every strip bears twenty characters of base text, amounting to five complete lines from the Cang Jie pian. In other words, the structural logic of the manuscript necessitates that the final characters on every strip must also conclude a line of Cang Jie pian base text. 70  On SQZ C058, however, the line division must be different. This again is signaled by the structural logic of the Shuiquanzi Cang Jie pian, which adds three-character "extensions of meaning" (I will refer to this as commentary for convenience) to the four-character base lines. Through comparison against PKU 65 and YT 1852, the distinction between base text and commentary on the Shuiquanzi witness is clear: … 禹湯 + 稱不絕 // 䫉迎 趮厥 + 怒佛甘 (which then translates as "… Yu and Tang; praised eternally // Pulchritudinous, arresting, restless, hurried; enmity is unsavory." 72 Because of the commentary's placement, we are assured that the line break for the base text must follow "Yu and Tang" on the Shuiquanzi manuscript. Note that this appears to be the case for the British Library edition of the Cang Jie pian as well, since bin fen 賓分 begins one side of the YT 1791 gu prism. On other gu prisms with primer content, an individual chapter is written in its entirety on a single piece, with each side of the prism bearing the same number of characters (for the Cang Jie pian, this would be twenty characters per side, for a sixty-character chapter). 73 Following this formatting, the first characters on the top of the prism begin their own lines. The triangular punctuation mark on YT 1791A secures this fragment as the top of the gu prism, and therefore indirectly establishes that bin fen 賓分 starts a new line. 74 Novel features on unprovenanced manuscripts offer a unique opportunity to test against newly excavated or appreciated archaeological finds for a positive authentication. But what if the data 71. Beijing daxue cang Xi-Han zhushu (yi), 132. 72. My translation of SQZ C058 is based on Zhang Cunliang's annotations. Zhang reads mao 䫉 as miao , and takes 迎 to be a miswriting of 抑, arguing that they mean "beautiful in appearance." He interprets jue 厥 as a loan for jue 蹶, which pairs with zao 趮 (alt. zao 躁) in the sense of "rapid movement" or "restlessness. from new archaeological finds present conflicts instead, as with the PKU 65, SQZ C058, and YT 1852 parallel? Does this then implicate the unprovenanced manuscript, here the Peking University Cang Jie pian, marking it as a forgery? While confirming that even a single novel feature on a unprovenanced manuscript exists among newly excavated archaeological finds would, theoretically, prove the authenticity of the artifact in question, the opposite does not hold true. Later conflicts with the archaeological data might raise suspicions about the unprovenanced manuscript, especially should they consistently appear in an ever more robust corpus of new finds. Yet it remains possible that the unprovenanced manuscript is both authentic and also exceptionally unique. This reveals another limitation to my approach: that it is designed to enable positive authentication, but does not provide similar accommodation for negative appraisals.
The 77. Note that both bin 賓 (PKU 2 and 8) and potentially fen 分 (FY C054) appear elsewhere in early editions of the Cang Jie pian. While repetition of characters occurs in the Peking University manuscript (e.g., hu 胡 on PKU 8 and PKU 11), this is rare (and indeed, PKU 8 is a known site of editorial manipulation). As such, we may speculate that YT 1852 is a simplification or miswriting, making variation of the sort just discussed more plausible. content; and it remains possible that, in the future, other manuscripts will be unearthed which prove this relationship. 78 Less attention has been paid to the conflict in line divisions between PKU 65, SQZ C058, and, potentially, YT 1791. Qin Hualin 秦樺林 early on noted that the Shuiquanzi manuscript appends commentary after "Yu and Tang" (禹湯), which necessitates a different line break, and argues that grouping "Yu and Tang" together with "Yao and Shun" (堯 舜) makes more sense thematically. 79 Most scholars now apply this line division to the Peking University manuscript without further comment, with only Hu Pingsheng 胡平生, to my knowledge, acknowledging the materiality of PKU 65 and the problematic placement of the partial characters . 80 Yet even Hu disregards the strict formatting of the Peking University manuscript, prioritizing the evidence given in the Shuiquanzi Cang Jie pian.
In my opinion, there are three potential resolutions to this conflict. The fragmentary nature of PKU 65 invites debate over whether or not it was once the bottom of a strip. Damage might have erased or destroyed ink after , and the notch above yao 堯 could be an incidental tear. Examination of the physical artifact is necessary to evaluate this appropriately, but this explanation does not seem promising based on the photographs alone. Assuming then that PKU 65 is the bottom of the strip, we may follow Hu Pingsheng and suppose that the Peking University Cang Jie pian forgoes its otherwise strict formatting in this section of text (either purposefully or as a scribal error). There are no other examples elsewhere in the Peking University Cang Jie pian where it diverges from the established format, however, and I find it unlikely that this is a sole exception. 81  unique to the manuscript in question that, while feasible, cannot be tested against new archaeological finds.
Alternatively, the conflict could also be due to edition-level variation between the Peking University Cang Jie pian and that of the Shuiquanzi (and possibily British Library) Cang Jie pian, with content added or deleted. The Han shu 漢書 Yiwen zhi 藝文志 gives a detailed textual history for the Cang Jie pian, in which it posits that lüli shushi 閭里書 師 or "village teachers" combined three prior texts (the Cang Jie 蒼頡, Yuanli 爰歷, and Boxue 博學) into a single work, with fifty-five chapters, each sixty characters in length. The Peking University Cang Jie pian chapters are all over a hundred characters and must be different from this Village Teachers edition; the Shuiquanzi and British Library Cang Jie pian witnesses, on the other hand, likely derive from the Village Teachers edition, with chapters sixty characters in length. Yet beyond the parsing of chapters, comparison of these manuscripts reveals very little manipulation in the content of their base texts, with most variants at the level of individual characters or words, making the edition-level variation intimated by PKU 65 vs SQZ C058 (and perhaps YT 1791) rather surprising. 82 Regardless, edition-level variation is both feasible and liable to archaeological confirmation.
To summarize the discussion above, there are limitations to my methodology for authenticating the Peking University Cang Jie pian, which are amply demonstrated by the PKU 65 and SQZ C058 parallel. On the one hand, ensuring novelty demands an analysis of all previous data. Because of the sheer amount of data to survey, and the possibility of unpublished or otherwise inaccessible data, claims for novelty must be constantly revisited and updated. In this case, YT 1852 was found to contain content related to the PKU 65 and SQZ C058, though this does not present a serious challenge to the novelty of PKU 65. On the other hand, this methodology is oriented towards positive authentication, but cannot prove a negative appraisal. Conflicts with newly excavated or appreciated archaeological finds might raise suspicions, especially if they consistently occur when compared against an ever more robust archaeological corpus, but do not theoretically prove forgery. PKU 65 82. On the textual history of the Cang Jie pian, see Han shu, 30.1721; Foster, "Study of the Cang Jie pian: Past and Present," 3-18. On the stability of content between Pre-Village Teachers and Village Teachers editions, see Foster, "The Shape of the Text: Gu Prisms and Han Primers." The only significant edition-level variant extant to date is between the Peking University and Fuyang manuscripts, but the Fuyang Cang Jie pian is also a pre-Village Teacher edition. See PKU 8-9 and FY C002. More tenuous editionlevel variation could be suggested by SQZ C072, but no photograph is available yet for this strip to allow us to check Zhang's transcriptions and analysis. My suspicion is that this content is misidentified by Zhang. differs with YT 1852 in its content ( vs bin fen 賓分, respectively), and conflicts in its implied line divisions compared to SQZ C058 and YT 1791 (breaking after shun 舜 vs after tang 湯, respectively). Yet feasible explanations can be offered to resolve these conflicts, with various hypotheses which both can and cannot be tested against new archaeological finds. 84. Since Chinese scholarship now largely adopts "Han board" as an appellation for this cache, I will follow this convention only to ease cross-referencing; this is not intended to condone any judgments over the antiquity of the manuscripts (i.e., as being of the Han period) that may implied by this convention. It is difficult to assess what footnote continued on next page As we did with the Peking University Han strips, we once again encounter a collection of manuscripts that potentially possess immense scholarly value, but whose promise is restricted by the illicit means in which it was obtained. Lacking proper provenance, the Han board manuscripts are either looted artifacts or forgeries. While this article focuses on matters of authentication, it must be emphasized, on the former possibility, that the Han board collection presents additional anxieties over professional ethics. In deciding how to responsibly handle these materials, I still believe we must "weigh between, on the one hand, the material and intellectual losses that may be suffered in the future by further incentivizing looting and, on the other hand, the material and intellectual losses we will suffer imminently by neglecting looted artifacts already on the market, as well as the future loss of neglecting those that may surface later." 85 Yet unlike the Peking University strips, the Han board manuscripts are held by a private collector, not a public institution. With artifacts in private collections, our evaluation is impacted by extra concerns. Access to items in private collections is more restricted. This withdraws artifacts from the realm of shared cultural heritage. It also potentially biases research on them, including their authentication. Scholarship on these objects can lead to the direct enrichment of an individual collector, who stands to benefit from positive appraisals of their collection's worth. Government oversight of public institutions helps to ensure responsible stewardship, whereas no such oversight is in place for private collections, thereby threatening the preservation of the artifacts. These are among the serious concerns that need to be weighed when treating objects in private collections. 86 My interest in the Han board Cang Jie pian manuscript, in the context of the present article, is mainly theoretical in scope and limited to the impact that the existence of such an artifact may have on our authentication of the Peking University Cang Jie pian manuscript. To this end, I briefly describe the Han board collection based on the information provided in Xinjian Han du Cang Jie pian Shi pian jiaoshi, and I survey, with minor elaboration, arguments raised by other scholars on the Han board Cang Jie pian's authenticity, which have been disseminated information circulated privately beforehand about the Han board Cang Jie pian. Liu mentions approaching Li Xueqin 李學勤 and Wang Hui 王輝 for appraisals. Liu, Xinjian Han du, "Houji" 後記, 273.
85. Foster, "Introduction to the Peking University Han Bamboo Strips," 233. 86. This is a topic that I will elaborate upon in further detail as part of a forthcoming Cambridge Elements piece discussing the use of unprovenanced artifacts in early China studies. publicly elsewhere. This is not intended as my own authentication of the Han board manuscripts, which requires both further research and greater dialogue about the concerns raised above.
The collection contains 119 wood-board pieces, of which fifty-seven belong to the Cang Jie pian, fifteen to Shi pian yi, forty-six to Shi pian er, and one to the Fengyu shi. 87 The intact boards are approximately 47 cm long, 5.4-6.1 cm wide, and 0.6-0.7 cm thick. 88 The top and bottom ends are level. At about 1.7 cm from the top of each board there is a hole, 0.5 cm in diameter, seemingly to accommodate a binding cord, allowing the boards to be strung together into a complete text. The top of each board is painted red, running ∼2.8 cm down. The writing is in black ink, in a mature Han clerical script (approaching bafen 八分), which Liu dates to sometime after the mid-Western Han. 89 On the Cang Jie pian and Shi pian boards, there are numerical labels (e.g. di yi 第一 "first") written above the holes. The main text, however, is written in the space underneath the red coloring, in three columns. The Cang Jie pian and Shi pian yi have twenty characters per column, for sixty characters total per board; Shi pian er and Fengyu shi squeeze an additional four characters into the end of third column on their boards, for sixty-four characters 87. The count of fifty-seven pieces for the Cang Jie pian includes a board initially identified by Liu as HB SP1 2. On its reassignment to the Cang Jie pian, see below.
Although there are fifty-seven pieces affiliated with the Cang Jie pian in the Xinjian Han du volume, a number of these are small fragments. It is possible multiple fragments could belong to what was a single intact board.
88. Details on the physical constitution of these boards may be found on Liu, "Qianyan" 前言, Xinjian Han du, 1. 89. Besides "Qianyan" 1, Liu repeats this assessment in "Han du Cang Jie pian de chubu yanjiu" 漢牘蒼頡篇的初步研究, Xinjian Han du, 220. There he refines the date to the end of Emperor Wu's reign or after. Zhang Chuanguan and Bai Junpeng both argue that the calligraphy is already bafen script, and therefore dates closer to the early Eastern Han. Zhang  (2) it has chapter divisions that match those previously reconstructed for the Village Teachers edition (e.g., as seen on the gu prism JY 9.1, with the placement of the Shuiquanzi character counts, or with the placement of triangular punctuation in the British Library shavings); (3) the labels on the Han boards with Cang Jie pian content, as read by Liu, do not exceed "fifty-five" (the chapter count for the Village Teachers edition); and (4) the opening lines of Shi pian yi, found together with the Han board Cang Jie pian, refer to "fifty-five chapters that copy out the Cang Jie" (蒼頡之寫五 十五章, see HB SP1 1.a). Since the boards are numbered, we can begin to piece together chapter order. 91 The manuscript bears over 2,160 characters, which amounts to roughly two-thirds of the Village Teachers Cang Jie pian's text. 92 Known or suspected Cang Jie pian content seen previously 90. Liu notes that HB SP2 20 is an exception, as he counts seventy-two characters altogether on this board. This is, however, difficult to evaluate from the photographs. See also n. 96 below. 91. Liu in Xijian Han du occasionally posits duplicate labels (using 甲 and 乙, e.g., HB 18甲 vs HB 18乙), arguing that this is an "enlarged" version to the Village Teachers edition of the Cang Jie pian ("Qianyan," 1-4; "Han du Cang Jie pian de chubu yanjiu" 漢 牘蒼頡篇的初步研究, 215-242). Most of these duplicate labels have since been corrected by subsequent scholarship, though clearer photographs are necessary before any resolution is possible. For now, I adopt the following changes. VT stands for Village Teacher edition chapter #, and is compared to the HB labels given by Liu: VT 2 = HB SP1 2; VT 10 = HB 54; VT 11 = HB 11乙; VT 20 = HB 10; VT 21 = HB 11甲; VT 17 = HB ?4; VT 18 = HB 18乙; VT23 = HB 43乙; VT 23 = HB ?1; VT 25 = HB 8; VT 45 = HB 35 乙; VT 54 = HB 24; and VT 55 = HB 53乙. We cannot definitively place HB content in any of the following individual VT chapters: VT 8,9,22,27,28,32,35,40,41,42,43  among other manuscript finds appears on the Han board witness. This includes the "opening chapter" frequently encountered among the Dunhuang and Juyan strips, as well as most of the content found in the Fuyang, Shuiquanzi, and Peking University Cang Jie pian manuscripts, including some unique parallels. 93 There is also novel content. 94 Notably absent is any content found in the "proper names" chapters, seen among the British Library, Yumen huahai 玉門花海 and Majuanwan 馬圈灣 strips, whose affiliation to the Cang Jie pian has been debated. 95 Besides the Cang Jie pian, this collection contains two other primers, which Liu titles Shi pian yi 史篇一 and Shi pian er 史篇二. Like the Cang Jie pian, both Shi pian texts rhyme every other four-character line. 96 They read in a more narrative style, akin to the so-called opening chapter of the Cang Jie pian ("Cang Jie created writing, and taught it to later generations" 蒼頡作書以教後嗣). In fact, Zhang Chuanguan 張傳 官 argues that Shi pian yi board 2 is misidentified by Liu and instead constitutes the second chapter of the Cang Jie pian Village Teachers edition, serving as a continuation of the Cang Jie pian's opening. 97 Regardless, Shi pian yi itself establishes a close relationship to the Cang Jie pian, when in its first lines it makes reference to the fifty-five 2,360 total. The Han shu describes the Village Teachers edition of the Cang Jie pian as fifty-five chapters, sixty characters per chapter, which means that edition was 3,300 characters long.
93 96. HB SP1 18 has three seven-character long sentences towards the end of the board, and the rhyme scheme appears to shift to accommodate them as well. Note that the sixty-character count for the chapter is preserved by what Liu presumes is the use of a repetition punctuation mark after yi 益 at the end of HB SP1 18.b. The mark is not visible on the photograph. HB SP2 18 and 19 seemingly both have a single threecharacter line. In each case, the content is similar and the first character of the sentence in question is nian 廿. This, however, could stand for the two characters, ershi 二十 ("twenty"), which would then restore the four-character sentence structure. 97. Zhang, "Tantan xinjian mudu Cang Jie pian de xueshu jiazhi," 334-38. Zhang chapters of the Village Teachers Cang Jie pian: "When endeavoring to study writing, the boards (suited) to instruct youths, are the fifty-five chapters transcribed by Cang Jie" (寧來學書告子之方蒼頡之寫五十五 章, SP1 1.a). 98 The postface to the Shuowen jiezi 說文解字 bemoans how various scholars mistakenly believed that Qin clerical script had ancient origins, and that the Cang Jie pian likewise "was the creation of ancient emperors" (古帝之所作) with "phrasing that has a mystical art about it" (辭有神僊). 99 The opening lines of the Shi pian yi may reflect such a belief. Of course, it is possible to take the final two lines here as "the fifty-five chapters that copy out the Cang Jie" instead, giving a more explicit citation of the text title. 100 Frequent mention (and veneration) of "scribes" (shi 史) and of the importance of "writing" (shu 書) likewise demonstrate the pedagogical objectives of Shi pian yi. 101 Whereas Shi pian yi is relatively short (fifteen board pieces), Shi pian er bears more extensive content (forty-six board pieces). 102 Beyond an exhortation to study, which again opens the primer, Shi pian er surveys a broad variety of topics, ranging from the origins of the cosmos and natural cycles, to family relations and expected behaviors, important life events for men and women, and the duties of rulers and officials. Finally, one board among this collection writes out a poem, referred to by Liu as the Fengyu shi. This title is not Liu's invention, but rather that of Zhang Feng 張鳳, who adopted it for a nearly identical poem written 98. The transcriptions for HB SP1 1.a follow Liu in Xinjian Han du, 151. On the association of fang 方 ("boards") with gu 觚 ("prisms")-multi-faced writing supports commonly employed with primers during the Han dynasty; see Foster, "The Shape of the Text: Gu Prisms and Han Primers," forthcoming. Another option is to take fang 方 as "method." 99. Shuowen jiezi zhu 說文解字注 (Shanghai: Shanghai guji, 1988), 15.763. 100. Liu argues that another reference is made to the Cang Jie pian textual system in Shi pian er, when it appears to mention the Fanjiang 凡將 arrangement of Sima Xiangru 司馬相如; see Xinjian Han du, HB SP2 52.a, 198). This is a more dubious proposition, and has been contested in: Zhang Chuanguan, "Xinjian Han du mengshu sanzhong jiaodu biji (sishisi ze)" 新見漢牘蒙書三種校讀筆記(四十四則), Fudan daxue chutu wenxian yu guwenzi yanjiu zhongxin xuezhe wenku 復旦大學出土文獻與古文字研究 中心學者文庫, January 6, 2020, www.gwz.fudan.edu.cn/Web/Show/4521, published in Chutu wenxian yu guwenzi yanjiu 9 (2020), 353-58.
102. If Liu's transcription of the label "Fifty-seven" on SP1 57 is correct, then this suggests significant textual loss for Shi pian yi, should it prove to be authentic. This is not the typical "authentication conference" seen with other institutional collections, such as Peking University, nor are the reported comments by Li or Wang affirmations of the boards' authenticity, but to the contrary, rather striking in their measured avoidance of any such affirmation. 106 Following the announcement of Xinjian Han du Cang Jie pian Shi pian jiaoshi's publication and the initial release of data, skepticism arose over the authenticity of these manuscripts. 107 This skepticism was not only because the manuscripts lacked a secure provenance, or because they had yet to be subjected to scientific testing such as radiocarbon dating. Other conspicuous features on the manuscripts have invited doubt. Zhang Chuanguan mentions, for example, the peculiar writing support, as well as certain character forms, seeming "unusual" (不同尋常) and "out of the ordinary" (與眾 不同) at first glance. 108 Despite these concerns, Zhang Chuanguan and other scholars recently have expressed confidence in the authenticity of the Han board Cang Jie pian. Zhang believes that: "It would have been extremely difficult to make the wood-board Cang Jie pian by stringing together previously known lines from the Cang Jie pian alone; no modern [forger] could have anticipated the new insights provided by its contents," as these insights "surpass the present state of knowledge for research into Cang Jie pian" (筆者認爲新見木牘《蒼頡篇》是很難僅僅根據以往的《蒼頡篇》文句 連綴而成的；其內容所提供的新知，絕非現代人所能臆測, and, prior to this, 這些新知恐怕已超出了現有的蒼頡篇研究水平). 109 An example would be Zhang's identification of HB SP1 2 as the previously missing second chapter of the Cang Jie pian, mentioned before, a novel feature among the Han board manuscripts that Zhang claims is verified by data not fully appreciated or anticipated before in the field (e.g., the identity of YT 1844, YT 2667, or YT 3222). Zhang's arguments are akin to those that I raise above concerning the Shuihudi verso sets and the Niya Cang Jie pian parallel in relation to the Peking University manuscript.
Irrespective of the Han board Cang Jie pian's authenticity, this manuscript does weave together known content from other Cang Jie pian finds in subtle ways, suggesting new interpretations for the older materials. This includes, for example, obviating "hidden" chapter divisions in the Village Teacher edition of the text, previously obscure (but knowable) from the Shuiquanzi Cang Jie pian, through changes in rhyming between SQZ C037, C038+C039, and C040 vs SQZ C114; as well as SQZ C044 vs SQZ C045. 110 Another curious implication of the Han board chapter order is that it suggests the Shuiquanzi manuscript only entails the first twenty chapters of the Village Teachers edition, with the phrase yuanli 爰歷 appearing at the end of the twentieth chapter.
In intriguing ways, this information corresponds to and conflicts with the textual history given in the Hanshu Yiwenzhi, where Li Si's Cang Jie precedes the Yuanli; but the former is listed in seven chapters (zhang 章) and the entirety of Cang Jie, Yuanli, and Boxue are in twenty chapters. 111 This brief introduction aside, the concern of the present article is on whether or not the existence of the Han board Cang Jie pian impacts our evaluation of the Peking University Cang Jie pian manuscript's authenticity, and not on the authenticity of the Han board Cang Jie pian. Since the Han board Cang Jie pian does not constitute archaeologically excavated data, it cannot be used positively to confirm features once novel to the Peking University Cang Jie pian, unless and only until the Han board Cang Jie pian itself is first properly authenticated. 112 Even then the evidence would by necessity only be indirect and thus not ideal. 113 It is crucial, moreover, to establish a timeline for the acquisition, study and publication of the Han board cache, to understand how this data relates to the timing of that for the Peking University bamboo strips. Only then 110. For more on the rhyming in the Shuiquanzi manuscript, see Zhou Fei, "Cang Jie pian zonghe yanjiu" 蒼頡篇綜合研究, Ph.D. dissertation (Tsinghua University, 2017), 153-60.
111. Zhang Chuanguan suggests SQZ C129 might be affiliated with HB 46, but the connection is tenuous. See Zhang, "Tantan xinjian mudu Cang Jie pian de xueshu jiazhi," 346-47. Unlike the "hidden" chapter divisions in the Shuiquanzi manuscript, the validity of this observation depends solely on the authenticity of the Han board Cang Jie pian and therefore awaits further research for confirmation.
112. Such a feature might entail content from the Peking University Cang Jie pian not attested before, but now found on the Han board Cang Jie pian, e.g., HB ?1 + HB 8 vs PKU 24-28.
113. This is because the Han board Cang Jie pian's authenticity would itself rely upon other archaeologically excavated data. may we judge the relative novelty of features on these manuscripts and the potential for access to this information by theoretical forgers.
Public announcement and dissemination of Xinjian Han du began in the fall of 2019, which marks our most conservative terminus ante quem for the existence of the Han board data. The date of publication listed on the volume itself is June of 2019, and Zhonghua shuju inevitably required a period of time beforehand for editing and other preparations. 114 Whether or not the Han board data existed prior to this point, and if so for how long beforehand, is uncertain. Liu reports that: "In the autumn of 2009, I was fortunate enough to inspect a collection of wooden boards at a friend's residence in Beijing, and obtain photographs of these artifacts" (二〇〇九年秋，在北京一友人處，我有幸獲觀一批木牘並得 到實物的圖片資料). 115 Liu does not discuss anything more about who the collector may be or the date of the boards' acquisition by them. If this report is to be trusted, a fall 2009 terminus ante quem comes before the public release of substantial data on the Peking University Cang Jie pian manuscript. Without knowing how much time passed between when the Han board manuscripts appeared on the illicit antiquities market and their inspection by Liu, it is also possible that the Han board Cang Jie pian circulated before the appearance of the Peking University manuscripts.
Liu describes his work on the Han board collection in more detail within the postscript of Xinjian Han du, where he states that he completed an initial draft of his annotated transcriptions by the spring of 2013, and then briefly updated them following publication of the Peking University Cang Jie pian. The postscript itself is dated August 2018. We must take Liu at his word for this timeline, and certain internal evidence from the Xinjian Han du coincides well with these dates, including: (1) The latest works cited in the volume are from 2016; 116 (2) Liu does not make use of Zhang Cunliang's dissertation and the updated information available therein for 114. The citation information listed in the front matter reads: 新見漢牘《蒼頡篇》 《史篇》校釋/劉桓編著. -北京：中華書局，2019.6 ISBN 978-7-101-11048-7. The named editor for Zhonghua shuju is Chen Qiao 陳喬, who is also thanked by Liu in the postscript; see Liu, "Houji," Xinjian Han du, 274.
While the timeline given above remains uncertain, if we take it at face value we must confront another problematic scenario: that a forger used the (in this case theoretically genuine) Han board manuscript to create the Peking University Cang Jie pian, at some point before either were attested (pre-2009). We may exclude this possibility since, despite the great overlap in content between these two manuscripts, there are novel features on the Peking University Cang Jie pian that are not seen on the Han board Cang Jie pian, confirmed by newly excavated or appreciated archaeological data. The textual parallel between PKU 64 and SQZ C088, C108 and C092 remains unique to the Peking University and Shuiquanzi manuscripts. This is true for part of the PKU 40 vs N.XIV.20 parallel as well. There is also content on the Peking University Cang Jie pian that is still unattested elsewhere, awaiting further discoveries for confirmation.
The surprising publication of the so-called Han board Cang Jie pian manuscript held in a private collection does, however, highlight how tenuous claims for novelty can be when used for authentication. A possibility we cannot exclude, for instance, is that there is a third manuscript-looted, but genuine-also in a private collection but unknown to the field, acquired prior to the Peking University Cang Jie pian, based on which a hypothetical forger could have drawn inspiration for replicating textual content, material features, or both. Although I find it infeasible, an even more troubling proposition is that such knowledge 117. See Liu, "Fanli" 凡例, Xinjian Han du, 1-2, as well as the labelling used for citing Shuiquanzi pieces throughout Liu's annotations. Recall that Zhang's dissertation was embargoed until 2019, see n. 48 above.
118. Liu, "Qianyan," Xinjian Han du, 2. 119. Liu generally lists parallels between the Han board Cang Jie pian and the other main witnesses in the following order: FY → SQZ → YT → PKU (see e.g., HB 3, Xinjian Han du, 10-13). He reports engaging with the British Library Yingtu volume in 2012, after working on the Fuyang and Shuiquanzi caches. Similarly, Peking University parallels are appended at the end, often with much briefer comments than the other witnesses, even when the overlap is much more extensive. When Peking University data is mentioned near the beginning of a "Shuoming" section (e.g., for HB 5 and HB 6), it seems to concentrate on content released early on, prior to the publication of Beijing daxue cang Xi Han zhushu (yi) in 2015.
could then have been used to manipulate looted ancient bamboo strips, which were prepared or bound already, but left blank without any writing. 120 The reason I find this infeasible is that it presumes writing on waterlogged strips, kept in an orderly fashion, which preserves information such as the original verso line relationships. 121 Yet if we do allow for this, we may confront a situation where an act of modern forgery combines both physical materials and textual content of genuinely ancient origins. It is a scenario that might challenge our very understanding of "authenticity" itself.

Conclusion
Recent developments in the study of early Chinese manuscripts and the Cang Jie pian have brought to light new data and understandings which bolster the positive authentication of the Peking University Cang Jie pian. The archaeologically excavated Liu nian zhiri manuscript from the Shuihudi Han cache confirms the existence of verso "sets," a novel feature found on several unprovenanced manuscripts from the Peking University, Tsinghua University, and Yuelu Academy collections, raising our confidence in these collections more generally. For the Peking University Cang Jie pian in particular, novel overlaps have been found Keywords: Peking University manuscripts, authenticity, Cang Jie pian, Niya, Han board witness, 北大漢簡, 真實性, 《蒼頡篇》, 尼雅, 漢牘本, between a newly identified strip from Niya (N.XIV.20) and recently released data on the Shuiquanzi Cang Jie pian. These observations provide additional evidence that the Peking University Cang Jie pian is indeed genuine.
Yet the conversation above is also cautionary. My methodology for appraising the Peking University Cang Jie pian is based on showing that novel features on the unprovenanced manuscript, unattested and unanticipated before its acquisition, are later confirmed through new archaeological discoveries. This approach is concerned with positive authentication, which limits its usefulness for negative appraisals. More importantly, it depends upon judgments about the "novelty" of features and archaeological controls. It is difficult, if not impossible, to establish novelty definitively. The belated recognition of YT 1852 as an additional witness to the overlap between PKU 65 and SQZ C058, and the consequent conflicts in wording and line divisions, demonstrate both of these points. The publication of the Han board Cang Jie pian, supposedly long in the possession of a private collector but unknown to the field at large, is an even starker example. No matter how confident our appraisal, assessments about authenticity must be continuously revisited and are liable to change as additional information comes to light. 提要 過去幾年，一系列新材料補充了此前我對北京大學藏漢簡及其中《蒼 頡篇》的真實性的判斷，使其得到了進一步的証實。本文提出了這些新 的證據，包括通過科學考古發掘的睡虎地漢簡簡背的"螺旋狀的劃線"痕 跡，也包括北大《蒼頡篇》與至今尚未被確認的一枚尼雅漢簡《蒼頡 篇》和與新公布的水泉子漢簡《蒼頡篇》的可相匹配的內容。本文也 更嚴格地檢討了我以前證實北大漢簡真實性時所採用的鑒定方法，承 認在否認真實性和確定簡牘的"新奇性"特徵這兩個過程上都面臨一程度 的困難。為了說明這些問題，本文還分析了北大漢簡和水泉子漢簡《蒼 頡篇》的一處差異，也介紹了目前還未被廣泛認識的所謂"漢牘"《蒼頡 篇》的相關基本情況。 北京大學漢簡《蒼頡篇》鑒定補考--兼論所謂"漢牘" 《蒼頡篇》的相關問題 傅希明