1. A still more challenging way of textual preservation includes inlaid inscriptions. There is some evidence for inlaying short bronze inscriptions with turquoise in the Anyang period; see for example the gongxingqi 弓形器 (M54:393) excavated from the tomb M45 at Huayuanzhuang, Locus East, published in shehui kexueyuan kaogu yanjiusuo, Zhongguo, Anyang Yinxu Huayuanzhuang Dongdi Shangdai muzang 安陽殷墟花園莊東地商代墓葬 (Beijing: Kexue, 2007), 159Google Scholar, Figure 121:3 and Plate 49:2. More common are inscriptions inlaid with gold from the Eastern Zhou period; however, to the best of my knowledge, no instance of inlaid inscription dating to the Western Zhou period has been reported. All dates for the Western Zhou follow Shaughnessy, Edward L., Sources of Western Zhou History: Inscribed Bronze Vessels (Berkeley: University of California Press, 1991), 217–87Google Scholar. The Western Zhou period is conventionally divided into three subperiods: Early Western Zhou (1045–957 b.c.e.), Middle Western Zhou (956–858 b.c.e.), and Late Western Zhou (857–771 b.c.e.).

2. An incisive and up-to-date overview of issues related to inscription production techniques can be found in Changping, Zhang 張昌平, “Shang Zhou qingtongqi mingwen de ruogan zhizuo fangshi: yi Zeng guo qingtongqi cailiao wei jichu” 商周青銅器銘文的若干製作方式—以曾國青銅器材料為基礎, Wenwu 2010.8, 61–70Google Scholar. For an English translation, see Zhang, Changping, “Some Considerations of the Bronze Inscriptions Techniques Used during the Shang and Zhou Dynasties,” trans. Lu, Ling-en, in Original Intentions: Essays on Production, Reproduction, and Interpretation in the Arts of China, ed. Pearce, Nicholas and Steuber, Jason (Gainesville: University Press of Florida, 2012), 265–81Google Scholar. For an earlier reconstruction of the various modes of bronze inscription production, see Barnard, Noel and Chia-pao, Wan, “The Casting of Inscriptions in Chinese Bronzes—with Particular Reference to Those with Rilievo Guide-Lines,” Soochow University Journal of Chinese Art History 6 (1976), 43–134Google Scholar. See also the discussion towards the end of this article, where the most recent literature is reviewed.

3. Shaughnessy, Edward L., “The Writing of a Late Western Zhou Bronze Inscription,” Asiatische Studien/Études asiatiques 61.3 (2007), 845–77Google Scholar; von Falkenhausen, Lothar, “The Royal Audience and Its Reflection in Western Zhou Bronze Inscriptions,” in Writing and Literacy in Early China: Studies from the Columbia Early China Seminar, ed. Feng, Li and Prager, David Branner (Seattle: University of Washington Press, 2011), 239–70Google Scholar.

4. Nickel, Lukas, “Imperfect Symmetry: Rethinking Bronze Casting Technology,” Artibus Asiae 66.1 (2006), 5–39Google Scholar; Bagley, Robert, “Anyang Mold-Making and the Decorated Model,” Artibus Asiae 69.1 (2009), 39–90Google Scholar.

5. For several instances of fission in Western and Eastern Zhou bronze inscriptions, see Zhichu, Sun 孫稚雛, “Jinwen shidu zhong yixie wenti de shangtao” 金文釋讀中一些問題的商討, Zhongshan daxue xuebao 1979.3, 57Google Scholar; Zhichu, Sun, “Jinwen shidu zhong yixie wenti de tantao (xu)” 金文釋讀中一些問題的探討(續), Guwenzi yanjiu 9 (1984), 409–10Google Scholar.

6. Li ji zheng yi 禮記正義 (Shisanjing zhushu 十三經注疏 ed., 1815; rpt. Beijing: Zhonghua, 1980), 49.379. For translation and discussion of this passage, see Schwermann, Christian, “Composite Authorship in Western Zhōu Bronze Inscriptions: The Case of Tiānwáng guǐ 天亡簋 Inscription,” in That Wonderful Composite Called Author: Authorship in East Asian Literatures from the Beginnings to the Seventeenth Century, ed. Schwermann, Christian and Steineck, Raji C. (Leiden: Brill, 2014), 41–44CrossRefGoogle Scholar.

7. Among the most important works, see for example Matsumaru Michio 松丸道雄, “Sei Shū seidōki seisaku no haikei: Shū kinbun kenkyū—joshō” 西周青銅器製作の背景: 周金文研究·序章, Tōyō bunka kenkyūsho kiyō 72 (1977), 1–128. Matsumaru however acknowledges that aristocrats in regional states could also cast their own inscribed bronzes. See also the insightful treatment of the problem of authorship in Western Zhou bronze inscriptions in Schwermann, “Composite Authorship in Western Zhōu Bronze Inscriptions,” 30–57.

8. Feng, Li, “Literacy Crossing Cultural Borders: Evidence from the Bronze Inscriptions of the Western Zhou Period (1045–771 b.c.),” Bulletin of the Museum of Far Eastern Antiquities 74 (2002), 210–42Google Scholar; Feng, Li, “Western Zhou Bronzes: Archaism as a Divergent Tradition,” in Dialogue with the Ancients: 100 Bronzes of the Shang, Zhou, and Han Dynasties—The Shen Zhai Collection, ed. Kwok, Patrick K. M. (Singapore: Select Books, 2018), 73–83Google Scholar.

9. See my “You ‘tong shi yi ming,’ ‘yi ren tong ming’ he ‘yi jia zhu ming’ san zhong xianxiang kan Xi-Zhou tongqi mingwen de ji zhong bianzuan moshi” 由“同事異銘”、“異人同銘”和“一家諸銘”三種現象看西周銅器銘文的幾種編纂模式, paper presented at the conference Inscribed in Bronze: New Directions in the Study of Ancient Chinese Bronze Vessels and Their Inscriptions, Chicago, May 14–15, 2016. The complexities of vessel-making within aristocratic lineages were insightfully explored by Fenghan, Zhu 朱鳳瀚 in his articles “Jinwen suo jian Xi-Zhou guizu jiazu zuoqi zhidu” 金文所見西周貴族家族作器制度, Qingtongqi yu jinwen 1 (2017), 24–45Google Scholar and “Zongren zhu qi kao: jian ji zai lun Xi-Zhou guizu jiazu zuoqi zhidu” 宗人諸器考: 兼及再論西周貴族家族作器制度, Qingtongqi yu jinwen 2 (2018), 16–28Google Scholar.

10. For various uses of writing in aristocratic lineages as attested to in bronze inscriptions, see Feng, Li, “Literacy and the Social Contexts of Writing in the Western Zhou,” in Writing and Literacy in Early China, ed. Li, and Branner, , 271–301Google Scholar. The point that higher aristocracy had their own scribes who could compose inscriptions for them is commonly raised in scholarship; see, for example, Shaughnessy, “The Writing of a Late Western Zhou Bronze Inscription,” 876–77.

11. For a case study on the compositional process of bronze inscriptions, see Shaughnessy, “The Writing of a Late Western Zhou Bronze Inscription,” 845–77. In addition to the official documents issued by the royal court, a compiler might have consulted transcripts of oral exchanges at the court; see von Falkenhausen, “The Royal Audience and Its Reflection in Western Zhou Bronze Inscriptions,” 267–70.

12. Susini, Giancarlo, The Roman Stonecutter: An Introduction to Latin Epigraphy, trans. M., A. Dabrowski (Oxford: Basil Blackwell, 1973), 46–48Google Scholar. For the manuals, see Cagnat, René, “Sur les manuels professionels des graveurs d’inscriptions romaines,” Revue de Philologie, de Littérature et d’Histoire Anciennes 13.1 (1889), 51–65Google Scholar.

13. Lan-ying, Tseng 曾藍瑩, “Zuofang, getao yu diyu zichuantong: Cong Shandong Anqiu Dongjiazhuang Han mu de zhizuo yiji tanqi” 作坊、格套與地域子傳統: 從山東安丘董家莊漢墓的製作遺跡談起, Guoli Taiwan daxue meishushi yanjiu jikan 8 (2000), 45Google Scholar. For one instance of what Tseng believes was a repertoire catalogue, see Hong, Wu, “Beyond the ‘Great Boundary’: Funerary Narrative in the Cangshan Tomb,” in Boundaries in China, ed. Han, John (London: Reaktion Books, 1994), 81–104Google Scholar.

14. Consider for example the sets of gui-tureens commissioned by two members of the same lineage, the senior Bo Si 伯㺇 and his younger brother Wei 衛. So far 10 gui-tureens commissioned by Bo Si are known, which can be divided into three subsets based on shape and décor. The earlier set (A) has a shorter inscription containing a prayer for blessings. See Zhenfeng, Wu 吳鎮烽, Shang Zhou qingtongqi mingwen ji tuxiang jicheng 商周青銅器銘文暨圖像集成 (Shanghai: Shanghai guji, 2012)Google Scholar, no. 05275, hereafter abbreviated as “Mingtu,” and Zhenfeng, Wu, Shang Zhou qingtongqi mingwen ji tuxiang jicheng xu bian 商周青銅器銘文暨圖像集成續編 (Shanghai: Shanghai guji, 2016)Google Scholar, no. 30460, hereafter abbreviated as “Mingxu”; two other vessels are reported but not published. Two later sets (B [Mingtu 05315–18] and C [Mingxu 30457–58]) differ in shape and décor, but both contain identical inscription commemorating the reception of gifts from the king. The gui tureens commissioned by Wei (Mingtu 05368–69) are completely identical in shape and décor with Bo Si’s set B, and thus can be assumed to have been cast together. However, while the structure and wording of the first half of the Wei gui inscription is nearly identical to that of Bo Si’s sets B and C, the second half takes over the formulas solely seen in Bo Si’s earlier set A. This implies that should the inscriptions be drafted in the workshop, the master copies of previously cast inscriptions would need to be stored here, which seems quite unlikely. The case of Bo Si’s and Wei’s vessels thus seems to confirm that at least within a lineage, the inscriptions for various lineage members were drafted using internal scribal resources, with reference to previously drafted inscriptions, and that the drafts received by the bronze workshops were elaborate and complete. On the relation between Bo Si’s and Wei’s vessels, see Fenghan, Zhu 朱鳳瀚, “Wei gui yu Bo Si zhu qi” 衛簋與伯㺇諸器, Nankai xuebao 2008.6, 1–7Google Scholar.

15. It is unclear to what extent the donor participated in the selection of a vessel’s shape and décor, and once he decided to have an inscription cast into the vessel, whether there were limitations other than material ones regarding the length of the inscription. Notably, there are scores of large vessels bearing very brief inscriptions and small vessels on which the inscriptions run well beyond the common inscriptional area, indicating that it was possible to accommodate longer texts, even on very small vessels, when needed. It would then appear that in the majority of cases, the inscriptions had to fit the vessels and not vice versa. However, from the perspective of production techniques, the inscriptions were not mere appendages of the vessels, as their inclusion necessarily invoked a much more complex production procedure requiring different treatments of piece-molds, as well as specialized personnel who were possibly not available in every workshop.

16. This is generally true for inscriptions with less than 100 characters, i.e., the majority of Western Zhou inscriptions. Longer inscriptions tend to expand horizontally (i.e., increasing the number of columns) rather than vertically (increasing the length of columns). However, some long inscriptions, like that of Larger Yu ding 大盂鼎 or Shi Qiang pan 史墻盤, are divided into two separate blocks, meaning that these smaller inscriptional fields again have the shape of a vertical rectangle. The layout aesthetics of bronze inscriptions certainly deserves further inquiry.

17. These observations are based on the shehui kexueyuan kaogu yanjiusuo, Zhongguo, ed., Yin Zhou jinwen jicheng (xiuding zengbuben) 殷周金文集成 (修訂增補本), 8 vols. (Beijing: Zhonghua, 2007)Google Scholar. Only fully legible rubbings were considered. Gui tureens and ding cauldrons were chosen specifically as the most common vessels to have been inscribed throughout the Western Zhou period.

18. A “character-space” is a virtual graphic cell, mostly in the shape of a vertically oriented rectangle. One character of regular size, or more characters of smaller size could be written into one character-space. It does not belong to the domain of writing a text but to that of formatting a text, being an “organizing principle,” see Steinke, Kyle, “Script Change in Bronze Age China,” in The Shape of Script: How and Why Writing Systems Change, ed. Houston, Stephen D. (Santa Fe: School for Advanced Research Press, 2012), 154–55Google Scholar.

19. “Jicheng” is used as an abbreviation for Yin Zhou jinwen jicheng. Further abbreviations used in the present article are “Xinshou” for Bor-sheng, Jung 鐘柏生, Chao-jung, Ch’en 陳昭容, Ming-chorng, Hwang 黃銘崇, and Kwok Wa, Yuen 袁國華 eds., Xin shou Yin Zhou qingtongqi mingwen ji qiying huibian 新收殷周青銅器銘文暨器影彙編 (Taipei: Yiwen, 2006)Google Scholar, and the above-mentioned “Mingtu” and “Mingxu” for Wu Zhenfeng’s Shang Zhou qingtongqi mingwen ji tuxiang jicheng and Shang Zhou qingtongqi mingwen ji tuxiang jicheng xu bian, respectively.

20. For insightful deliberations on how the display function shaped the visual qualities of bronze inscriptions including the layout, see Steinke, “Script Change in Bronze Age China,” 135–58.

21. The stoichedon style is a term used in Greek epigraphy to refer to a pervasive habit in mainly Attic inscriptions between the sixth and third centuries b.c.e. to engrave letters so that they “are in alinement vertically as well as horizontally, and are placed at equal intervals along their respective alinements”; see Austin, R. P., The Stoichedon Style in Greek Inscriptions (Oxford: Oxford University Press, 1938), 1Google Scholar. In Chinese scholarship, the neat arrangement of an inscription is usually referred to as cheng hang cheng lie 成行成列 etc., but I am not aware of a specific term for this type of graphical arrangement; I thus borrow the term “stoichedon style” for the purposes of this discussion.

22. From the perspective of writing habits and their transmission in time, it would be instructive to make a distinction between ligatures in the original sense of a word (i.e., instances of graphical coalescence when strokes of two or more characters join/touch (as in wu shi 五十) or merge (so-called jiebi 借筆, as in wu yue 五月) and characters that share only the character-space (like si yue 四月 or xiao zi 小子). It would appear that the former cases represent a more customary writing habit, whereas the latter cases are more ad hoc devices for space-saving. However, regarded through the emic perspective of Warring States scribes, the two instances appear to have been considered as functionally identical, as they tend both to be marked by a ligature mark. This shows that the basic characteristic of hewen is in fact the sharing of a graphic cell, and further confirms that the ancient scribes worked with the notion of “character-space.” I will thus refer to ligatures and sharing of a character-space jointly as “combined graphs,” or hewen. For traditional views on ligatures in Chinese paleography, see Zhao, Liu 劉釗, “Guwenzi zhong de hewen, jiebi, jiezi” 古文字中的合文、借筆、借字, Guwenzi yanjiu 21 (2001), 397–410Google Scholar. For a more recent reassessment, see Bao Huifang 暴慧芳, “Hanyu guwenzi hewen yanjiu”漢語古文字合文研究, M.A. thesis (Xinan University, 2009). For an English introduction to the problem, see Galambos, Imre, “Scribal Notation in Medieval Chinese Manuscripts: The hewen (Ligature) and the chongwen (Duplication) Marks,” Manuscript Cultures 2 (2009), 5–9Google Scholar; Park, Haeree, The Writing System of Scribe Zhou: Evidence from Late Pre-imperial Chinese Manuscripts and Inscriptions (5th–3rd centuries BCE) (Berlin: De Gruyter, 2016), 94–97CrossRefGoogle Scholar.

23. While a “combined graph” hewen could function as a direct space-saving device providing ad hoc solutions for textual compressions, reduplication marks chongwenhao could function as a space-saving device only indirectly at earlier stages of the compositional process. At the stage when the composer was selecting the wording of an intended inscription, it is conceivable that sometimes he would favor the use of a certain expression or construction that would enable reduplication, and as a consequence, the abbreviation of the written text. The Early Western Zhou Tai Bao gui 大保簋 inscription (Jicheng 04140) is to my knowledge the only clear-cut instance that does not apply reduplication marks for the writing of an “ABAB” sequence (wang jiang zheng ling yu Tai Bao, Tai Bao ke jing 王降征令于大保大保克敬 “the King sent down a campaign command to the Grand Protector. The Grand Protector was capable of being respectful”; translation after Edward L. Shaughnessy, “The Role of Grand Protector Shi in the Consolidation of the Zhou Conquest,” Ars Orientalis 19 [1989], 51). Interestingly, the layout of this inscription is 9/9/9/7, and it is thus conceivable that it was originally designed as 8/8/8/8 (the same as the more or less contemporary Li gui 利簋) with the use of reduplication marks for Tai Bao 大保. The repetition of the characters Tai Bao may thus indicate some distortion in the process of textual transmission from the master copy to the clay slab with inscription. Slightly later Er you 耳卣 inscription (Jicheng 05384) duplicates the character 耳 in the sentence Ning shi xi Er, Er xiu fu gan ju 寧史錫耳耳休弗敢沮 “The secretary of Ning awarded [me,] Er, [I,] Er was perfect and do not dare to cease [in it]”; here, however, it is quite likely that Er and xiu were misplaced, and that the inscription was intended to read Ning shi xi Er xiu, Er fu gan ju 寧史錫耳休耳弗敢沮 “The secretary of Ning awarded [me,] Er beneficence, [I,] Er do not dare to cease [to be diligent in my service].” “Awarding beneficence” xi xiu 錫休 is a commonly used expression in bronze inscriptions. The layout of this inscription is 6/6/5.

24. It is a common practice in scholarship on Chinese bronze inscriptions since the Song dynasty that the information about inscriptions’ graphic presentation is fully entrusted to the rubbing itself, and that the annotation is compressed into the number of characters (zishu 字數)—in more recent publications being accompanied by notes on the number of reduplication marks (chongwenhao) and “combined graphs” (hewen). Nearly all epigraphic qualities of the inscriptions are obliterated by such a description, however. For example, for the Hu gui inscription, the Yin Zhou jinwen jicheng description is “122 characters, 1 reduplication mark, 1 ‘combined graph’”; see Zhongguo shehui kexueyuan kaogu yanjiusuo, Yin Zhou jinwen jicheng (vol. 4), 03421. Supplementing such description with the number of “character-spaces” would give much clearer impression. Noel Barnard often used the notion of character-space in his discussions, but viewed it as a modern concept that he often related to the problem of the cataloging of information concerning the length of the inscription; see Noel Barnard 巴納 and Cheung Kwong Yue 張光裕, Zhong Ri Ou Mei Ao Niu suo jian suo ta suo mo jinwen huibian 中日歐美澳紐所見所拓所摹金文彙編 [Rubbings and Hand Copies of Bronze Inscriptions in Chinese, Japanese European, American, and Australasian Collections], Introductory Volume (Taipei: Yee Wen Publishing Company, 1978), 51–52. Lothar von Falkenhausen also uses this term, see his “Ritual Music in Bronze Age China: An Archaeological Perspective,” Ph.D. dissertation (Harvard University, 1988).

25. Most instances of xiao zi 小子 are written as hewen in the bronze inscriptions, including the earliest instances in the Late Shang period; shi you 十又 appears written as hewen in the Middle Western Zhou Yong yu 永盂 (Jicheng 10322) and Fifteenth Year Que Cao ding 趞曹鼎 (Jicheng 02784) inscriptions; zai xia 在下 is written as hewen in the Late Western Zhou Guo Shu Lü zhong 虢叔旅鐘 (Jicheng 00238–44) and Fifth Year Hu zhong 五祀㝬鐘 (Jicheng 00358) inscriptions. Clearly, the conventions for use of hewen were quite stable in time, yet the question is whether all the hewen used in bronze inscriptions stem from general scribal practice, or if there are also some that were devised and used specially for space-saving in bronze inscriptions. A diachronic comparison with oracle bones, covenant texts, and bamboo manuscripts might shed some light on this issue, for the moment, Bao Huifang, “Hanyu guwenzi hewen yanjiu” serves as a handy reference.

26. I am grateful to Ren Xueli 任雪莉 of Baoji Bronzeware Museum 寶鷄青銅器博物館 for providing this photograph and facilitating contact with the Fufeng County Museum 扶風縣博物館 where the vessel is housed; I am equally grateful to the director of the Fufeng County Museum, Wang Yutang 汪玉堂, for the kind permission to reproduce the photograph here.

27. There is no fixed term by which scholars refer to the person who inscribed the clay slabs. If we exclude “craftsman” or “scribe” as inappropriate because they obscure the fact that this was a highly specialized task, the most approximate term seems to be Barnard’s “artisan-scribe.” On the Eastern Han Wu Liang Shrine stela 武梁碑, the stonecutter responsible for carving the texts referred to himself as liang jiang 良匠 (“skilled craftsman”); the carver of another Eastern Han Stela for Sacrificing to the Mountain of Three Dukes 祀三公山碑 (carved 117 c.e.) calls himself simply gong 工 (“artisan”); other Han dynasty terms include zao shi gong 造石工 (“Chief Mason”), bei shi 碑師 (“Stele Master”) and, most commonly, shi shi 石師 (“Master Mason”). Today, these stonecutters are referred to as kegong 刻工, which is a term originating in the Three Kingdoms period, but which still does not distinguish between a stonecutter producing stone reliefs and one carving texts. Regarding the evolution of stonecutters’ self-appellations, see Zhangcan, Cheng 程章燦, Shike kegong yanjiu 石刻刻工研究 (Shanghai: Shanghai guji, 2008), 50–62Google Scholar. For lack of a better term, I borrow the word “ordinator”, which is used in Latin epigraphy to denote the person responsible for transferring the text from the master copy onto an inscriptional surface, which I find is quite a good parallel for the responsibilities in the process of the preparation of the bronze inscription in Western Zhou times. The process of the transfer of the text itself can be then conveniently referred to as “ordination” (from the Latin ordinatio). While using the masculine pronouns in reference to ordinators, I do not exclude the possibility that some of them (if not all) could be females. On female artisans in Early China, see Barbieri-Low, Anthony J., “Craftsman’s Literacy: Uses of Writing by Male and Female Artisans in Qin and Han China,” in Writing and Literacy in Early China, ed. and, Li Branner, 370–99Google Scholar.

28. These include the famous He zun 何尊 (Jicheng 06014), Yi Hou Ze gui 宜侯夨簋 (Jicheng 04320), Larger Yu ding 大盂鼎 (Jicheng 02837) and Zhong fangding 中方鼎 (Jicheng 02785) inscriptions.

29. There are also recent voices arguing that the Larger Yu ding should in fact date to the reign of King Mu; see Shan, Li 李山 and Hui李輝, Li, “Da Xiao Yu ding zhizuo niandai Kang wang shuo zhiyi” 大小盂鼎制作年代康王說質疑, Beijing shifan daxue xuebao 2012.2, 31–36Google Scholar; Khayutina, Maria, “Reflections and Uses of the Distant Past in the Chinese Bronze Inscriptions from the 10th to 5th Centuries BC,” in Historical Consciousness and the Use of the Past in the Ancient World, ed. Baines, John, Henriette van der Blom, Yi Samuel Chen, and Tim Rod (Sheffield: Equinox, 2019), 166–67Google Scholar.

30. Note that in Latin epigraphy, for example, scholars usually agree that the drafts written on ephemeral materials did not always specify the layout and abbreviations; these would follow routine. Quite likely, the drafts were also not made to full size. See Susini, The Roman Stonecutter, 33, 44–47; Grasby, Richard D., “Latin Inscriptions: Studies in Measurements and Making,” Papers of the British School at Rome 7 (2002), 153Google Scholar. Unlike in Chinese or Greek inscriptions, however, the tabular alignment of graphs was not of great importance in the Latin epigraphy which preferred proportional layout of the letters. The size of cast characters in Chinese bronzes was also usually more or less comparable to the size of handwriting. In such situations, the most economical solution would be to exemplify the layout together with the space-saving devices directly in the master copy.

31. Feng, Li, “Ancient Reproductions and Calligraphic Variations: Studies of Western Zhou Bronzes with ‘Identical’ Inscriptions,” Early China 22 (1997), 1–41CrossRefGoogle Scholar. Based on the calligraphic features and the shape and décor of the ding cauldrons, Li Feng tentatively proposed that originally there was a set of 15 ding + 12 gui. This argument presupposes that only one person was responsible for producing inscriptions in each subset consisting of four or five vessels. However, Li Feng’s reasoning was misguided by a photograph of one of the ding vessels published in sheng kaogu yanjiusuo, Shaanxi, wenwu guanli weiyuanhui, Shaanxi sheng, bowuguan, Shaanxi sheng, eds., Shaanxi chutu Shang Zhou qingtongqi 陝西出土商周青銅器, vol. 1 (Beijing: Wenwu, 1979)Google Scholar, Plate 197, which is probably mistaken for another ding from the cache, most likely the Shanfu Lü Bo ding 善夫旅伯鼎 (75QDJ:21). See Wei, Cao 曹偉, Zhouyuan chutu qingtongqi 周原出土青銅器, vol. 3 (Chengdu: Bashu shushe, 2005)Google Scholar, 443. Cao Wei, Zhouyuan chutu qingtongqi, vol. 3, 390–401, provides correct photographs of Ci’s vessels that show that all three ding are nearly identical in shape and décor, and could be thus part of one subset. In fact, it will be shown later that more individuals indeed took part in the production of inscription slabs for the same subset of vessels in a consecutive fashion. The most intriguing textual variation in the set of inscriptions on Ci’s vessels is rendering the ancestor’s name Gui gong 癸公 twice as Zhu gui 朱癸, which is most likely due to an eye-slip to a neighboring line with a gift list; however, the problem is that the same mistake is repeated by two separate hands. This introduces the possibility that the subset inscribed by hand A was reproduced with some temporal distance from the subset by hand B and was not based on a manuscript but directly on the vessel’s inscription. The text of gui 75QDJ:9 inscription would be then copied directly from the gui 75QDJ:10, including the mistake. The other gui inscriptions by hand A would then use a corrected version or were again copied directly from the remaining inscriptions by hand B; however, in the extant sample by hand A, the omission of three characters in gui 75QDJ:13 by hand B is not reproduced. See also n. 88 below.

32. Meaning that each column has the same number of character-spaces but these are not in line with each other. I am indebted to Yegor Grebnev for suggesting this term to me (personal communication, April 9, 2017).

33. One gui inscription (75QDJ:13) omits three characters xiang 享, yong 用, and qi 其, but still pertains to the 10x11 layout (eight characters in the ultimate column). See the related discussion below where based on this feature, I suggest that hand B perused the same exemplar text as hand C.

34. Since there was enough space to accommodate columns of 11 character-spaces on these two ding cauldrons, one may speculate that the 10x11 layout was motivated not by more space but by the general habit of having an inscription vertically “longer” than wider.

35. The 9x11 layout again seems to reflect the general habit of writing longer rather than wider. There is also a possibility that the original master copy had only le and that you was added later, but I am inclined to rule out this possibility, since we do not commonly see le appearing in gift lists on its own. Interestingly, it is the modifier you “bronze-plated” that is omitted from the modifier-head noun phrase “bronze-plated bridle,” so the inscription remained grammatical despite this omission. While this may be a coincidence, it may also hint that the person who decided to make this omission was literate to some degree.

36. For the various types of layout on bronze bells in Shang and Zhou periods, see von Falkenhausen, “Ritual Music in Bronze Age China,” 626–44.

37. Note also that each of the two pairs of gui vessel and lid inscriptions are performed by a different hand, see Xiaoxia, Liu 劉曉霞, “Xiaochen Lai gui xin lun” 小臣𬣆簋新論, Kaogu 2016.4, 109Google Scholar. For further implications of this observation, see the discussion below.

38. For an instructive discussion on the causes of violations of the stoichedon order in Greek inscriptions, see Osborne, M. J., “The Stoichedon Style in Theory and Practice,” Zeitschrift für Papyrologie und Epigraphik 10 (1973), 258–70Google Scholar.

39. The line Hu bai qi shou, dui yang wang xiu 虎拜稽首, 對揚王休 “Hu with folded hands bows prostrate and extolls the King’s beneficence,” in the Shi jing 詩經 hymn “Jiang Han” 江漢, also employs the trisyllabic variant, arguably also for purposes of space-saving, though here the goal is to preserve the tetrasyllabic verse. Note that bai shou qi shou was also used as a polite (often speech-opening) formula; such use is attested to not only in bronze inscriptions, but also in the “Shao gao” 召誥 or “Li zheng” 立政 chapters from Shang shu 尚書 or the Warring States manuscripts from Xincai Geling 新蔡葛嶺. For a more detailed discussion, see my “You tongqi mingwen de bianzuan jiaodu kan Xi-Zhou jinwen zhong ‘bai shou qi shou’ de xingzhi” 由銅器銘文的編纂角度看西周金文中“拜手稽首”的性質, Qingtongqi yu jinwen 1 (2017), 541–59Google Scholar, where I argue that bai shou qi shou in Western Zhou bronze inscriptions is used predominantly as a polite formula expressing gratitude.

40. The highly phonetic writing of the character yang 揚, in which only the phonophoric component 昜 is written instead of the usual complex, is also noteworthy.

41. This phenomenon can be sporadically observed in longer inscriptions, see for example Larger Ke ding 大克鼎 (Jicheng 02836) or Hu ding 㫚鼎 (Jicheng 02838) inscriptions. For a brief discussion about these inscriptions, see n. 90 below.

42. The issue of casting of bronze inscriptions will be discussed in detail below. For a brief introduction to the piece-mold casting procedure, see Shaughnessy, “Sources of Western Zhou History,” 37–43; for a more elaborate treatment, see Li, Yung-ti, “Co-Craft and Multicraft: Section-Mold Casting and the Organization of Craft Production at the Shang Capital of Anyang,” in Craft Production in Complex Societies: Multicraft and Producer Perspectives, ed. Shimada, Izumi (Salt Lake City: University of Utah Press, 2007), 184–94Google Scholar.

43. An alternative to the term “blueprint” tentatively used in this article would be the bibliographic term “setting copy.” Apart from its conciseness, I find the technical undertone of the term “blueprint” quite fitting in the context of inscription-making.

44. I am indebted to Jeffrey R. Tharsen for drawing my attention to this case (personal communication, March 2015). For the general complexities involved in the transposition of text from an exemplar manuscript onto an epigraphic artifact, see the examples from Latin epigraphy discussed by Mallon, Jean, “Paléographie des papyrus d’Egypte et des inscriptions du monde romain,” Museum Helveticum 10 (1953), 141–60Google Scholar; see also Grasby, “Latin Inscriptions,” 151–56.

45. Noel Barnard, in association with Kwong-yue, Cheung, The Shan-fu Liang Ch’i Kuei and Associated Inscribed Vessels (Taipei: SMC, 1996), 37–71Google Scholar.

46. For the reading of this character, see Haoliang, Yu 于豪亮, “Shaanxi sheng Fufeng xian Qiangjiacun chutu Guo Ji jiazu tongqi mingwen kaoshi” 陝西省扶風縣強家村出土虢季家族銅器銘文考釋, Guwenzi yanjiu 9 (1984), 259Google Scholar.

47. Barnard and Cheung, The Shan-fu Liang Ch’i Kuei and Associated Inscribed Vessels, 68–71. However, eight or sixteen bells usually formed a chime from the Middle Western Zhou period onward.

48. Shimin, Wang 王世民, “Xi-Zhou ji Chunqiu Zhanguo shidai bianzhong mingwen de pailie xingshi” 西周暨春秋戰國時代編鐘銘文的排列形式, in Zhongguo kaoguxue yanjiu: Xia Nai xiansheng kaogu wushi nian jinian lunwenji (er ji) 中國考古學研究—– 夏鼐先生考古五十年紀念論文集 (二集), ed. kaoguxue, Zhongguo yanjiu bianweihui (Beijing: Kexue, 1986)Google Scholar, reprinted in Shimin, Wang 王世民, Kaoguxue shi yu Shang Zhou tongqi yanjiu 考古學史與商周銅器研究 (Beijing: Shehui kexue wenxian, 2017), 431–32Google Scholar. The same observation was made recently also by Jeffrey R. Tharsen; see his “Chinese Euphonics: Phonetic Patterns, Phonorhetoric and Literary Artistry in Early Chinese Narrative Texts.” Ph.D. dissertation (University of Chicago, 2015), 147 n. 8. Note, however, that Tharsen’s reconstruction does not consider the dimensions of the two bells, and it appears that he considers bell F to have been originally preceding bell E, and that only the characters Liangqi 梁其 (bell E) and the final six characters on bell F were misplaced. However, it is clear from the dimensions of the two rubbings that bell E indeed preceded bell F, and it will be shown that in fact, the last two characters Liangqi on bell E are the only two characters from the present inscription originally projected to appear on this bell.

49. The best example of this is one of the Forty-third Year Qiu ding 卌三年逑鼎 cauldrons unearthed from the famous cache in Yangjiacun 楊家村, Meixian 眉縣, Shaanxi in 2003 (2003MYJ:7); for rubbing, see Xinshou 748. This inscription was originally prepared on three separate clay slabs that were embedded into the inner core mold; however, in the case of 2003MYJ:7 ding, the first and the third slab were embedded in the wrong order, resulting in sequence 3–2–1 instead of 1–2–3 (thus, the columns of the inscription run as follows: original columns XX–XXIX, then columns X–XIX, and then I–IX). For an Early Warring States period example of misplaced slabs, see Huaqiang, Song 宋華強, “Aomen Chongyuan xin jian Chu qingtongqi chu yi” 澳門崇源新見楚青銅器芻議, Zhongwen xueshu qianyan 3 (2011), 193–96Google Scholar.

50. The inscription on bell D in fact omits the character Liang 梁 and writes zai shang 才上 as hewen; thus, it only contains 61 character-spaces. The visualization in Figure 8 correspondingly omits position 124 of the original character Liang.

51. This would suggest that at least in this case, the casters were concerned with facilitating the reading of the inscription. The question is who made the decision to alter the layout of the inscription, which obviously demanded several editorial actions in the blueprint.

52. Since the bells were graded in size, we can expect that bell G would accommodate more characters than bell H; thus, the larger blueprint for bell D can be expected to split between characters 115 and 116, possibly even between characters 120 and 121 (see Figure 9). Further possible scenarios for the textual discrepancy might be: 1) 36 character-spaces were conceived originally for bell E. The person responsible for the task, possibly the ordinator, would then take the manuscript with 74 characters (which was used to inscribe bell C) and mark the place between character 36 and 37 to split the text. The manuscript would then be split into two parts, one containing characters 1–36, and another with characters 37–74; 2) 38 characters were planned for the inscription, but the person responsible counted not from the beginning but from the end (see also later discussion for this point); 3) the larger blueprint was just mechanically split into two halves; in this case, we could conjecture that the blueprint originally had nine character-spaces in each column, and thus the split would appear exactly between characters 36 and 37, resulting in two smaller blueprints, one with characters 1–36 and another with characters 37–74. Except for the last one, however, these scenarios do not satisfactorily explain why exactly 36 or 38 characters would be planned for the bell E inscription.

53. Moving from larger to smaller vessels seems to have been a common practice in the production of the graded vessels of a set. Typically, the layout and calligraphy of larger vessels in a set was performed carefully and with high quality, but moving towards smaller pieces, mistakes would occur. For a good example, see the set of Forty-third Year Qiu ding cauldrons unearthed in 2004 from a cache in Yangjiacun, Meixian: the two largest cauldrons (YJ18 [Mingtu 02503] and YJ13 [Mingtu 02504]) are cast without textual flaws; in the third cauldron (YJ3 [Mingtu 02505]), the slabs with inscriptions were misplaced (see n. 49 above); the fourth, fifth, and seventh cauldrons (YJ6 [Mingtu 02506], YJ12 [Mingtu 02507] and YJ16 [Mingtu 02509]) are also without mistakes; the inscription on the sixth cauldron (YJ5 [Mingtu 02508]) misses characters zai 才 [II.2] and wo 我 [XVI.6], and the inscription on the seventh (YJ2 [Mingtu 02510]) misses the character zu 且 [VII.10]. The inscription divided onto the two smallest cauldrons (YJ8 [Mingtu 02511] and YJ4 [Mingtu 02512]) omits a whole portion of 15 character-spaces, but this omission was probably intentional; see the discussion below.

54. Barnard and Cheung, The Shan-fu Liang Ch’i Kuei and Associated Inscribed Vessels, 42, 43 Figure 17.

55. Hayashi Minao 林巳奈夫, “In Shū seidōki meibun chūzōhō ni kansuru jakkan no mondai” 殷周靑銅器銘文鑄造法に關する若干の問題, Tōhō gakuhō 51 (1979), 38–40; Li Feng further suggested that this may be connected to the sharing of responsibilities in the production of subsets of a final set or a result of later reproduction; see his “Ancient Reproductions and Calligraphic Variations,” 13–15, 24–26, 37, 40.

56. Such a copying scenario was already anticipated by Edward L. Shaughnessy, “The Writing of a Late Western Zhou Bronze Inscription,” 874 n. 29. Notably, a certain period of time was needed for inscribed slabs to dry before they were fired, which made them available for consultation during the ordination of subsequent slabs.

57. One can only speculate about the reasons for the mismatch of the two blueprints. While negligence is a fairly acceptable explanation, I wonder whether the fact that the first character on the second blueprint was crossed or blackened might have been a contributing factor, as it might have been interpreted as an index sign for the beginning of the text. Compare the later practice of marking the beginning of text by a blackened upper margin in the Mawangdui manuscripts Jingfa 經法, Shiliu jing 十六經, Cheng 稱 and Daoyuan 道原, or the use of black squares to mark the beginning of smaller textual units within the abovementioned Jingfa and Shiliu jing manuscripts. For the Mawangdui marks, see Xigui, Qiu 裘錫圭, ed., Changsha Mawangdui Han mu jianbo jicheng (qi) 長沙馬王堆漢墓簡帛集成 (柒) (Beijing: Zhonghua, 2014), 60–76Google Scholar (photographs). We could further speculate that the blueprint was two-sided, and in his search for the beginning of the text, the ordinator opted for the wrong side, being misguided by the blackened character.

58. Shaughnessy, Sources of Western Zhou History, 267–71.

59. Numerically, this process can be rendered as follows: 10/10/10/10/10/10/10/10/10/7 (vessel inscriptions) →10/10/10/9/10/10/10/10/10/7 (lid 04280.1) →10/10/10/10/10/10/10/10/10/6 (lids 04279.1 and 04282.1). Another possibility is that the omission was marked prior to the ordination of the 4280.1 lid, but the character was mistakenly included. Since only a local correction was made without shifting the position of remaining characters, it is clear that the mistake was detected only after the entire inscription was laid out during the proofreading process. Subsequently, the character Ke was deleted and the four characters were rewritten again. On proofreading, see the discussion below. The observation that the omission of Ke was deliberate is clear not only from this reconstruction, but also from the fact that should this omission be unintended, it should be easy to identify it with the help of the master copy during proofreading. Note that should the blueprint have five character-spaces per column, the omitted character Ke would be positioned as the first character of the eighth column of the blueprint, i.e., possibly a visually salient position.

60. Xiaoneng, Yang, “The Shi Ke Xu: Reconsideration of an Inscribed Late Western Zhou Vessel,” Artibus Asiae 52.3/4 (1992), 192–93CrossRefGoogle Scholar. Based on the measurements, Yang observes that only the Jicheng 04468 lid inscribed by hand C would fit on the vessels Jicheng 04467.2 or Xinshou 1907, i.e. vessels inscribed by hand A.

61. These were common ligatures at the time. All extant Shi Ke xu inscriptions write yi you (or sometimes chang yi) into one character-space, but none of them ligate si pi, even though these characters are written into one character-space in the Jicheng 04467.1 inscription and are extremely close to each other in the Xinshou 1907 lid inscription, i.e. in the two inscriptions by hand B.

62. Yang Xiaoneng, “The Shi Ke Xu,” 204–5, notes this textual corruption and proposes rightly, in my view, that a pan basin was cast together with the set of xu containers.

63. In this case, both suggested positions overlap; see Figure 12. The position of the alternative vessel name at the end of the blueprint is suggested by the inscription on a Late Chunqiu Yin yi 匜 (Jicheng 10284). The inscription identifies the vessel as hui pan (沬) 盤 “pan-basin for face-washing,” but the character yi 匜 “ewer” is added after the closing phrase of the inscription (zi zi sun sun, yong bao yong zhi 子子孫孫，永寶用之 “for generations of descendants, eternally use it as treasure”). This indicates that the blueprint drafted for a pan inscription was also to be employed for a yi inscription, and that the alternative vessel name yi “ewer” was noted at the end of the blueprint. In this case, however, the ordinator had not replaced the character pan as intended, and simply included the annotated character yi at the end of the inscription. Similar traces of the sharing of a blueprint by different vessel types can also be observed in the Late Western Zhou Tai Shi Shi Jiang yi 太師氏姜匜 (Mingtu 14999) inscription. Here, the water-pouring vessel yi denominates itself as bao pan 寶盤 “precious pan-basin.” As the pan basins and yi pouring vessels were usually cast and used in a set, it is quite likely that in this case a blueprint prepared for a pan was copied verbatim on the yi inscription without changing the vessel name. The same situation occurs on the Middle Western Zhou Qiu Wei he 裘衛盉 inscription (Jicheng 09456), which denominates itself also as bao pan. Any research on vessels’ self-appellation should therefore take into consideration the phenomenon of the sharing of blueprints in the bronze vessel production process.

64. While the omission of you “indeed” from the phrase ze you wei 則唯 “then indeed it was” is certainly acceptable, the omission of mi “canopy” from the phrase hu mi xun li 虎冟 (鼏-冪) 熏 (纁) 裏 “tiger-skin canopy with light-red lining” obviously leads to undesired mangling “tiger [skin] with light-red lining.” It is therefore understandable that the omission of mi was reconsidered and replaced by the omission of you. For further elaboration on you, see Pei, Shen 沈培, “Xi-Zhou jinwen zhong de ‘you’ he Shang shu zhong de ‘di’” 西周金文中的“”和《尚書》中的“迪”, Guwenzi yanjiu 25 (2004), 218–24Google Scholar.

65. This also applies to the case of the Liangqi zhong bells above. But compare Susini, The Roman Stonecutter, 42: “We must envisage a fairly complex relationship, full of uncertainties and misunderstandings, among the different participants in the production of an epigraphic monument, and we must include the possibility that they were the same person.”

66. For this division of the procedure of inscription-making, see Grasby, “Latin Inscriptions,” 151–56. For an analogy from ancient Egypt, see the set of ostraca that served as handy blueprints during the process of inscribing the burial chamber in the tomb of Nakhtmin (late 14th century b.c.e.); see Barbara Lüscher, Die Vorlagen-Ostraka aus dem Grab des Nachtmin (TT 87) (Basel: Orientverlag, 2013). I am grateful to Christelle Alvarez for this and other references concerning the ancient Egyptian pyramid texts.

67. I am grateful to Thies Staack for pointing out that the use of such an intermediary manuscript would also increase the likelihood of incurring mistakes in the inscription (personal communication, August 17, 2018). This is certainly true, and it is exactly due to such mistakes that we can retrieve more information about inscription-making in this period.

68. See for example the set of Forty-third Year Qiu ding cauldrons mentioned in the above n. 53, where it seems that only the larger cauldrons were proofread. In the set of the Ci gui tureens, omission of three characters in the 75QDJ:13 inscription (Jicheng 04310) and misspelling of the ancestor’s name in 75QDJ:9–10 inscriptions (Jicheng 04306–07) indicates that they were probably also not proofread.

69. On Lu gui, see Edward L. Shaughnessy, “Newest Sources of Western Zhou History: Inscribed Bronze Vessels, 2000–2010,” in Imprints of Kinship: Studies of Recently Discovered Bronze Inscriptions from Ancient China, ed. Edward L. Shaughnessy (Hong Kong: The Chinese University Press, 2017), 151–54. Based on the maturity of this inscription, some scholars date the vessel to the following reign of King Gong (917–900 b.c.e.); see for example Wei, Han, “Lu gui niandai ji xiangguan wenti” 䚄簋年代及相關問題, in Xin chu jinwen yu Xi-Zhou lishi 新出金文與西周歷史, ed. Fenghan, Zhu 朱鳳瀚 (Shanghai: Shanghai guji, 2011), 57–70Google Scholar.

70. The same kind of mistake also appears on the inscriptions on the lid of Shi Ju gui 師遽簋 and on the Yong yu 永盂. In the Shi Ju gui lid inscription (Jicheng 04214), dated by some scholars to the third year of King Mu’s reign (ca. 954 b.c.e.), either the character zai 才 (II.5) or Zhou 周 (II.6) were omitted during the ordination and then supplemented upon proofreading, but as a result, the ultimate column has one character-space fewer than planned. In the Yong yu inscription (Jicheng 10322), commonly dated to the twelfth year of King Gong’s reign (ca. 906 b.c.e.), the fifth column omitted either character jue 氒 (V.1) or ming 命 (V.2) during the ordination; again, the missing character was supplemented after proofreading, but one character-space is missing in the last column as a consequence.

71. This phrase is miswritten also in the Jicheng 04026 inscription; however, there are two factors that deem the Jicheng 04026 inscription (present location unknown) rather suspicious: 1.Its size, layout, calligraphic features, and even spacing between characters are almost completely identical with the Jicheng 04024.2; 2. The shape and décor of the gui vessel bearing this inscription differs significantly from the Jicheng 04024 and Jicheng 04025 vessels. Since the Jicheng 04024.2 inscription was first published in the Qing imperial catalogue Xi-Qing gu jian 西清古鑑 in 1755 and thus was long available for potential forgers, the authenticity of the Jicheng 04026 inscription, first published in 1937, remains questionable.

72. Guo Moruo 郭沫若, Liang Zhou jinwen ci da xi kaoshi 兩周金文辭大系考釋, vol. 3 (Tokyo: Bunkyūdō, 1935), 181a. The vessel inscription Jicheng 04025.2, also by the same hand, reads correctly shi you yi yue 十又一月, but has in the last column sun sun sun sun 孫二孫二 “grandsons and grandsons” instead of zi zi sun sun 子二孫二 “sons and grandsons”; this easily correctable mistake (by simply effacing the “silk” element in the first sun 孫) was left unmarked.

73. Sun Zhichu, “Jinwen shidu zhong yixie wenti de shangtao,” 56–57. Note the inferior quality of this inscription’s calligraphy. The sequence zuo hui qi yi wan nian wu jiang sun xiang 乍 (作) (沬) 其也 (匜) 萬年無疆孫畗 (亯-享) should thus read zuo hui yi, qi wan nian wu jiang sun xiang 乍 (作) (沬) 也 (匜)，其萬年無疆孫畗 (亯-享) (“makes yi for face-washing, he will for ten thousand years without limit for generations perform (ritual) service”).

74. In case of Zheng Guo Zhong gui vessels, should the Jicheng 04026 inscription be authentic, the repetition of the transposition sign in two inscriptions by the same hand would suggest that the obelism was copied from the blueprint. Thies Staack further proposes that the editorial mark was originally an independent sign in the blueprint, but that it was wrongly interpreted by the ordinator as an additional stroke of the character you 又 (personal communication, August 17, 2018). It is worth noticing that the same ordinator performed the text in the correct sequence in Jicheng 04025.2, but had not corrected the passage in Jicheng 04024.2. Given the spurious nature of Jicheng 04026, I am inclined to leave this question open. The overall scarcity of these marks in the inscriptions seems to suggest that in the two cases under scrutiny, they were probably copied from the blueprint.

75. The locus classicus for reconstruction of this procedure is Barnard, Noel, Bronze Casting and Bronze Alloys in Ancient China (Tokyo: Monumenta Serica Monograph, 1961), 157–61Google Scholar. See also Barnard and Wan, “The Casting of Inscriptions in Chinese Bronzes,” 54–55.

76. Chusheng, Chen 陳初生, “Yin Zhou qingtongqi mingwen zhizuo fangfa pingyi” 殷周青銅器銘文製作方法評議, Jinan xuebao 1998.1, 117–21Google Scholar; Nickel, , “Imperfect Symmetry,” 5–39; Dong Yawei 董亞巍, Fan zhu qingtong 範鑄青銅 (Beijing: Beijing yishu yu kexue dianzi, 2006), 110–13Google Scholar; Zhang Changping, “Shang Zhou qingtongqi mingwen de ruogan zhizuo fangshi,” 61–70; Guan Shuqiang 管樹強, “Shang Zhou qingtongqi yinxian mingwen zhuzao fangfa tantao” 商周青銅器陰線銘文鑄造方法探討, (paper presented at the Peking University Graduate Student Conference “Bronzes and Bronze Inscriptions,” Beijing, December 25, 2016, unpublished conference handbook, pp. 178–92; a digest of this paper is published as Shuqiang, Guan 管樹強, “You qingtongqi mingwen zhuzao fangfa tan guwenzi shidu de jige wenti” 由青銅器銘文鑄造方法談古文字釋讀的幾個問題, Zhongguo wenzi xuebao 8 (2017), 69–76)Google Scholar. Ceramic molds with traces of the tube-lining technique have been unearthed from the Zhougongmiao bronze workshop site in early 2000s and are analyzed in Chen Yang 陳陽, “Zhouyuan zhutong yizhi suo chu taofan de chubu yanjiu” 周原鑄銅遺址所出陶範的初步研究, M.A. thesis (Peking University, 2005). For a critique of tube-lining, see Bagley, “Anyang Mold-making and the Decorated Model,” 39–90. It should be noted that the existence of tube-lining, in my view, does not exclude the existence of carving on the clay model; it is quite likely that the two techniques could have been used complementarily. In light of the present evidence, the question is no longer whether tube-lining was used, but since when, on what scale, and how exactly was it performed. Nickel, “Imperfect Symmetry” and Zhang Changping, “Shang Zhou qingtongqi mingwen de ruogan zhizuo fangshi,” observe the use of tube-lining on piece-molds with décor already in the Yinxu period, while Dong Yawei, Fan zhu qingtong, 110–13, asserts that most of the Shang bronze inscriptions were produced by tube-lining.

77. Nickel, “Imperfect Symmetry,” 16.

78. Zhang Changping, “Shang Zhou qingtongqi mingwen de ruogan zhizuo fangshi,” 61–70. As for the earlier vessels, Zhang shows evidence that the Late Western Zhou bronze bell Chu Gong Jia zhong 楚公𧱌鐘 (Xinshou 659) was also inscribed with the use of tube-lining; see ibid., 67.

79. I inspected the Guanzhuang foundry excavation site on October 29, 2017 and studied the unearthed inscription slabs, on which intaglio grooves of few characters are preserved; in certain places, the rilievo residues of character strokes that have been built up from clay can be observed. The same technique was employed on the molds with decorative motives unearthed from this site. I am indebted to Gao Xiangping 郜向平, Hui Xiping 惠夕平, and to the organizers of the Academic Conference on Shang and Zhou Bronzes and Bronze Inscriptions 商周青銅器與金文學術研討會, Henan Provincial Institute of Cultural Heritage and Archaeology, Zhengzhou (October 28–29, 2017) for providing the opportunity to visit the site. The discovery of the inscription slabs and ceramic molds was reported at this conference by Gao Xiangping, “Xingyang Guanzhuang yizhi faxian de zhutong yicun” 滎陽官莊遺址發現的鑄銅遺存 and Hui Xiping, “Guanzhuang yizhi H1573 mingwenfan keng fajue ji xiangguan wenti de chubu renshi” 官莊遺址H1573銘文範坑發掘及相關問題的初步認識. Some of the introductory remarks presented at the conference were recently published in Gao Xiangping 郜向平, Zhao Hao 趙昊, and Ding Sicong 丁思聰, “Henan Xingyang Guanzhuang yizhi faxian Liang Zhou ji Han dai shougongye zuofang yicun” 河南滎陽官莊遺址發現兩周及漢代手工業作坊遺存, Zhongguo wenwu bao, Feb. 22, 2019, 8. Several Western Zhou period inscription slabs were unearthed from the bronze foundry site at Luoyang Beiyao 洛陽北窯, but only line-drawings that do not allow a reliable assessment of the production technique were published in shi wenwu gongzuodui, Luoyang, “1975–1979 nian Luoyang Beiyao Xi-Zhou zhutong yizhi de fajue” 1975–1979 年洛陽北窯西周鑄銅遺址的發掘, Kaogu 1983.5, 439Google Scholar Figure 12.

80. Guan Shuqiang, “Shang Zhou qingtongqi yinxian mingwen zhuzao fangfa tantao,” 184–85; Dong Yawei, Fan zhu qingtong, 112, proposed that tube-lined strokes could be sometimes carefully trimmed, especially in earlier (and shorter) inscriptions, which could explain the difference in calligraphy between earlier and later inscriptions. Sakikawa Takashi 崎川隆 concurs and points out that an inscription slab from the former Luo Zhenyu’s 羅振玉 collection (now in possession of the Lüshun Museum in Dalian, Liaoning Province, dating between the Late Shang and Early Western Zhou period) was also produced by tube-lining or a similar method (personal communication, October 29, 2017). One may expect the technique to have evolved over time. For the Xiaomintun clay slab, see kexueyuan kaogu yanjiusuo Anyang gongzuodui, Zhongguo shehui, “2000–2001 nian Anyang Xiaomintun Dongnandi Yindai zhutong yizhi fajue baogao” 2000–2001 年安陽孝民屯東南地殷代鑄銅遺址發掘報告, Kaogu xuebao 3 (2006), 376Google Scholar Figure 21:3 and Plate 15:2. For the Lüshun Museum slab, see bowuguan, Lüshun, ed., Lüshun bowuguan guancang wenwu xuancui: taoci juan 旅順博物館館藏文物選粹: 陶瓷卷 (Beijing: Wenwu, 2009), 28Google Scholar.

81. The left-to-right direction is corroborated by the appearance of corrections in the initial columns of longer inscriptions, as discussed above. This is quite an interesting observation because the ordinators might have also proceeded with the ordination from right to left, which was a general habit at the time. The fact that they did not shows, in my view, that they preferred the consecutiveness (and thus intelligibility) of the text they were writing to habitual writing directionality, which would suggest that they might have relied on reading the text when copying it, and that they were at least to some extent literate. The left-to-right direction might be also more suitable for right-handed ordinators, although this is only speculation since we do not know exactly how tube-lining was performed. Finally, when vertically written texts are copied from their end, omissions of letters or characters result in garbling of the text in a particular fashion, as is sometimes the case in the ancient Egyptian retrograde inscriptions. I am not aware of any instance of such garbling in Zhou dynasty bronzes. For an insightful case study of mistakes in the transmission of a text from a perishable to durable medium in ancient Egypt, see Alvarez, Christelle, “An Epigraphical Journey in the Pyramid of Ibi: Between Textual Transmission and Mistakes,” in Current Research in Egyptology 2015: Proceedings of the Sixteenth Annual Symposium, University of Oxford, United Kingdom, 15–19 April 2015, ed. Alvarez, Christelle, Belekdanian, Arto, Gill, Ann-Katrin, and Klein, Solène (Oxford: Oxbow, 2016), 20–33CrossRefGoogle Scholar.

82. Lukas Nickel was quick to note that the ordinators using tube-lining would have to write in mirror writing. He also connects this, correctly in my view, to the emergence of the phenomenon of “reversed writing” (fanshu 反書), prominent especially in the Chunqiu period bronzes. See his “Imperfect Symmetry,” 37. It is worth noticing that, for the purposes of the study of calligraphy, the inscriptions in fanshu are of the utmost importance, since in case the inscriptions were produced by the tube-lining technique, they in fact represent the only instances of normal writing (zhengshu 正書) from this period.

83. One more common feature of Western Zhou cast inscriptions is that the lower part of their grooves is usually wider than their mouth, a phenomenon that Rutherford Gettens called “undercutting”; see Gettens, Rutherford John, The Freer Chinese Bronzes, Volume II: Technical Studies (Washington, DC: Smithsonian Institution, 1969), 141–47Google Scholar. Lukas Nickel pointed out that tube-lining can explain the undercutting as well as the absence of mechanically produced inscriptions; see his “Imperfect Symmetry,” 36–38. Zhang Changping has further explained the presence of rilievo grid commonly seen in inscriptions; see Zhang Changping, “Shang Zhou qingtongqi mingwen de ruogan zhizuo fangshi,” 65. Guan Shuqiang adds another specific feature of tube-lined inscriptions, i.e., the thickened intersections of two or more strokes (the so-called tuan dian 團點) created by redundant clay in places where one clay cord was piped over another; see Guan Shuqiang, “Shang Zhou qingtongqi yinxian mingwen zhuzao fangfa tantao,” 185; Guan Shuqiang, “You qingtongqi mingwen zhuzao fangfa tan guwenzi shidu de jige wenti,” 72. Mechanically reproduced inscriptions can be seen on several Chunqiu period bronzes; see Takashi, Sakikawa 崎川隆, “Chunqiu shiqi qingtongqi mingwen zhuzao gongyi zhong jixie fuzhi jishu de chuxian yu fazhan” 春秋時期青銅器銘文鑄造工藝中機械複製技術的出現與發展, in Chutu wenxian yu wuzhi wenhua 出土文獻與物質文化, ed. C., Adam Schwartz 史亞當 (Hong Kong: Zhonghua, 2017), 301–22Google Scholar.

84. Barnard and Cheung, The Shan-fu Liang Ch’i Kuei and Associated Inscribed Vessels, 261–62. Barnard called these residual characters “skeleton sketch-lines.” I agree with Edward L. Shaughnessy (personal communication, December 15, 2017) that it might be more fitting to call them “ghost” characters.

85. Feng, Li 李峰, “Xi-Zhou qingtongqi mingwen zhizuo fangfa shiyi” 西周青銅器銘文製作方法釋疑, Kaogu 2015.9, 78–91Google Scholar. Li Feng correctly identifies the sketch-lines under the character jing 巠 as xiao 孝 and under the character fang 方 (mistakenly rendered as zai 才 in the article) as zai 才. He also identifies the sketch-lines under character sheng 聖 as wang 王; however, from the textual sequence, we can be certain that this is rather the character jing 巠 (note that the bottoms of characters wang 王 and jing 巠 are identical). It seems from Li’s identification that he has not realized the relationship between the “ghost” characters and the characters registered in the final inscription.

86. Ya, Zhou 周亞, “Guanyu Da Ke ding de jige wenti” 關於大克鼎的幾個問題, Qingtongqi yu jinwen 1 (2017), 306–22Google Scholar.

87. I am greatly indebted to Zhou Ya, who has provided me with the opportunity to study a clear photograph of the Larger Ke ding inscription, and to the Shanghai Museum for the kind permission to reproduce the relevant close-ups here. Close scrutiny, however, did not reveal any further “ghost” characters. It is obvious that the inscription was polished upon the vessel’s completion, proceeding from the left end of the inscription towards the right side, but the polishing procedure was not completed, and as a result, the rilievo grid lines and “ghost” characters were preserved at the right-hand part of the inscription. The reason for leaving the remainder of the inscription unpolished was to ensure that the text will remain legible; see discussion below. Figure 15 provides my own line-drawings, which are not as nice as Barnard’s, but are based on clearer rubbings and photographs and thus reflect the actual appearance of “ghost” characters more accurately. A fairly clear photograph of the Larger Ke ding inscription has been published in Zhang Tian’en, ed., Shaanxi jinwen jicheng, vol. 4, Baoji juan: Fufeng, 32–33.

88. In light of the above discussion, the confusion of Zhu gui 朱癸 for Gui gong 癸公 on 75QDJ:10 gui inscription seems somewhat more understandable. All three characters involved, zhu 朱 (Old Chinese in Baxter-Sagart reconstruction *to), gui 癸 (*kʷijʔ), and gong 公 (*C.qˤoŋ) were placed either at the beginning or at the end of their columns, i.e., in visually salient positions, which may have been a contributing factor. The expected addition of the character gui 簋 (*kʷruʔ), possibly next to the character ding or next to the upper or lower margin of the column could also add to the confusion.

89. This shows that the person who performed tube-lining was able to navigate successfully the snarl of grooves, of which many were not to be tube-lined.

90. Such a solution was in fact suggested by Li Feng and Zhou Ya; see Li Feng, “Xi-Zhou qingtongqi mingwen zhizuo fangfa shiyi,” 85–89; Zhou Ya, “Guanyu Da Ke ding de jige wenti,” 318–22. Barnard himself related the phenomenon of “ghost” characters to the technique of inter-stroke space excavation, but asserted that such a production technique was very unlikely to have been employed for inscriptions as long as that of the Larger Ke ding; consequently, he left the question open; see his The Shan-fu Liang Ch’i Kuei and Associated Inscribed Vessels, 238–39, 261–67. One more peculiarity of the Larger Ke ding inscription is an empty space between columns XI and XII. While it may appear at first glance that this might be another remnant of the correction of the engraved sketch, from the content of the inscription, it is clear that the empty column is located between two thematical units—Ke’s eulogy of his ancestors and of the ruling king on one hand (exactly 110 character-spaces) and the rendering of the investiture ceremony on another (exactly 170 character-spaces)—and serves basically as a section break between two paragraphs. A similar phenomenon can be observed on the Middle Western Zhou Hu ding (Jicheng 02838) inscription, in which three different events are recorded, each of them beginning in a new column.

91. Zhou Ya, “Guanyu Da Ke ding de jige wenti,” 318.

92. In his article, Li Feng had already observed this phenomenon in the inscription on the bottom of the Guo Ji gui 虢季簋 unearthed from the Sanmenxia cemetery (M2001:86, Xinshou 18) and suggested that it was a result of the additional corrective carving performed on the finalized clay slabs after the strokes were damaged during the insertion of slabs into the piece-mold assembly. The fact that such hypothetical corrective efforts were nearly always thwarted by polishing, however, seems to deem such a scenario invalid. See Li Feng, “Xi-Zhou qingtongqi mingwen zhizuo fangfa shiyi,” 89.

93. The intaglio sketch grooves of character strokes or décor lines in places from which the clay cord dropped or broke off when the mold was disassembled after casting can be observed on the clay slabs with inscriptions and décor unearthed from the Guanzhuang site and also on several remains of Late Western Zhou ceramic décor molds unearthed from the bronze foundry site at Lijia 李家, Fufeng County, Shaanxi. The Lijia case, for example on the mold H66:46, was described already by Chen Yang, “Zhouyuan zhutong yizhi suo chu de taofan chubu yanjiu,” 28, Figure 18:1. Unused sketch-lines of decorative motifs are also reported from the Lijia site; see for example the ceramic mold H66:41 in Chen Yang, “Zhouyuan zhutong yizhi suo chu de taofan chubu yanjiu,” 28, Figure 18:2. The ash pit with the two specimens is dated to the beginning of Late Western Zhou period; see ibid., 32, 34. See also ibid., 35–36 for the difference between the molds from the Early Western Zhou Zhougongmiao 周公廟 foundry site and the later Lijia site.

94. The Larger Ke ding inscription was prepared divided on two separate inscription blocks. That the left-hand side of the inscription was also polished is indicated by several instances of “missing” strokes, for example in characters wang 王 (XV.4), yu 余 (XVI.8), yu 于 (XIX.8), or zhen 朕 (XXV.4). The comparison of calligraphic features seems to suggest that the two blocks were inscribed by two different ordinators, which is a phenomenon highly unusual in early Chinese epigraphy. This would explain the uneven quality of the two parts of the inscription: while the right-hand side is full of casting flaws, the left-hand side is noticeably more successful, and so is its calligraphy. The left-hand ordinator was clearly more skilled or dutiful than his right-hand colleague.

95. For another case showing casters’ concern with the legibility of an inscription, see Edward L. Shaughnessy, “On the Casting of the Art Institute of Chicago’s Shi Wang Ding: With Remarks on the Important Position of Writing in the Consciousness of Ancient China” (paper presented at the conference “The Age of Transition: Bronzes and Molds Found in Daijiawan, Shigushan, Baoji County, Shaanxi,” Beijing, November 30, 2015).

96. This means that, should more “ghost” characters be discovered in inscriptions, it will necessarily be in those that remained unpolished. Original vessels or high-quality photographs of their inscriptions will need to be consulted, since rubbings can easily fail to register the “ghost” characters. One more “ghost” character can be identified in the inscription on the lid of the Middle Western Zhou Qisheng Lu fangyi 齊生魯方彝 (Jicheng 09896), where the character shi 十 (I.4) is written over the “ghost” character you 又 in the phrase wei ba nian shi you er yue 隹八年十又二月 “It was the eighth year, twelfth month.” Clearly, the character shi was omitted during the initial ordination and character you was sketched in this position (I.4); before tube-lining, the mistake was discovered, and the correct character shi was tube-lined in this position, but as the sketched grooves of the character you in this position were not levelled up, upon casting, they registered as rilievo “ghost” strokes around the intaglio character shi. For a clear photograph of this inscription, see Zhang Tian’en, Shaanxi jinwen jicheng, vol. 2, Baoji juan: Qishan 岐山, Fufeng, 35.

97. See for example the arguments based on phonetic patterns in inscriptions as raised in Jeffrey R. Tharsen, “Chinese Euphonics.” For another point exemplified by reassessment of the function of the phrase bai shou qi shou 拜手稽首, see my “You tongqi mingwen de bianzuan jiaodu kan Xi-Zhou jinwen zhong ‘bai shou qi shou’ de xingzhi,” 541–59.

98. Eastmond, Antony, “Introduction: Viewing Inscriptions,” in Viewing Inscriptions in the Late Antique and Medieval World, ed. Eastmond, Antony (Cambridge: Cambridge University Press, 2015), 2–3CrossRefGoogle Scholar.

99. For a discussion of visibility and legibility of the Western Zhou bronze inscriptions as well as their audience, see Olivier Venture, “Étude d’un emploi ritual de l’écrit dans la Chine archaïque (XIIIe–VIIIe siècle avant notre ère),” Ph.D. dissertation (Université Paris 7–Denis Diderot, 2002), 277–91.

100. As an example of this, although without the necessary discussion on textual production, see Xiaohua, Qin 秦曉華, “Xi-Zhou jinwen ‘ze you’ shixi,” 西周金文“則”試析, Guwenzi luntan 2 (2016), 189–92Google Scholar.

101. Note that the use of a single manuscript for an entire set of inscriptions did not necessarily hinder the manufacturing output of the workshop because the other ordinators could be working in parallel on other inscriptions. The reuse of a single manuscript would thus in fact be economical, both time-wise and resource-wise: it would expedite the production process because no additional time would need to be invested in the production of extra blueprints, and it would spare the material for the production of writing supports. It would also prevent further copyists’ errors.

102. To my knowledge, there are only few instances of textual leaps similar to a “missing strip.” The inscription on Xing ding 𤼈鼎 (Jicheng 02742) seems to lack one column (seven characters) between columns III and IV (the sequence runs … xi ju liang, bai qi / yong zuo huang zu wen kao yu ding 錫駒兩拜稽/用作皇祖文考盂鼎 [“was awarded two foals. With folded hands bowing / and thus make [this] yu-type cauldron for my august ancestors and cultured father”]). From the comparison with the Xing hu 𤼈壺 inscription (Jicheng 09726–27), which records events from the same year, we can assess that seven missing characters are shou, gan dui yang Tian zi xiu 首敢對揚天子休 “prostrate, [I] take the liberty to extoll the beneficence of the Son of Heaven.” However, unlike other Xing’s vessels, the ding-cauldron was already unearthed during the Song dynasty and its inscription is only preserved in line-drawing; it is possible that the “missing” column was present in the original inscription but was omitted during the copying process in Song times. This is probably the case also for another inscription recorded by Song scholars, that of “Shi Qin Gong” ding 師秦宮鼎 (Jicheng 02747), which also drops at least one column of seven characters between columns III and IV. Interestingly, the authors of the imperial catalogue Chongxiu Xuanhe Bogutu 重修宣和博古圖 (1107–1123) were confident enough to mark the missing column (as well as other missing characters) in the “Shi Qin Gong” ding inscription with small circles (Siku quanshu 四庫全書 ed., 3.35b), but no catalogue marks omission for the Xing ding inscription (recorded in Wang Qiu’s 王俅 Xiaotang jigulu 嘯堂集古錄 (1176) and Xue Shanggong’s 薛尚功 Lidai zhongding yiqi kuanzhi fatie 歷代鐘鼎彝器款識法帖 (1144); while Wang Qiu follows Bogutu in marking the omissions in inscriptions with small circles in his catalogue, Xue Shanggong does not observe this practice). The first scholar to note the omission in the Xing ding inscription (and moreover to reconstruct it correctly and link it with the “Shi Qin Gong” ding inscription) is thus probably Mengjia, Chen 陳夢家 in 1958; see his Xi-Zhou tongqi duandai 西周銅器斷代 (Beijing: Zhonghua, 2004), 336–37Google Scholar.

103. On the reconstruction of the habitat of bronze foundry craftsmen in the Western Zhou period, see Xingshan, Lei 雷興山, “Lun Zhouyuan yizhi Xi-Zhou shiqi shougongyezhe de ju yu zang: jian tan teshu qiwu zai juluo jiegou yanjiu zhong de zuoyong” 論周原遺址西周時期手工業者的居與葬—– 兼談特殊器物在聚落結構研究中的作用, Huaxia kaogu 2009.4, 95–101Google Scholar.

104. For examples of these inscriptions, see Edmondson, Jonathan, “Inscribing Roman Texts: Officinae, Layout, and Carving Techniques,” in The Oxford Handbook of Roman Epigraphy, ed. Bruun, Christer and Edmondson, Jonathan (Oxford: Oxford University Press, 2014), 112Google Scholar. See also Sandys, John Edwin, Latin Epigraphy: An Introduction to the Study of Latin Inscriptions (Cambridge: Cambridge University Press, 1927), 57Google Scholar.