THE SANXINGDUI SITE: ART AND ARCHAEOLOGY VOL. 1 (TEXTS) A DISSERTATION PRESENTED TO THE FACULTY OF PRINCETON UNIVERSITY IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY RECOMMENDED FOR ACCEPTANCE BY THE DEPARTMENT OF ART AND ARCHAEOLOGY

jay xu

This article compares the remarkable restorations of two relief sculptures from the Binyang Central Cave at the Longmen Grottoes, Henan Province, during the 1940s. The contrasting restorations of the Offering Procession of the Empress as Donor with her Court and Figure of Wei Mo Chi from the Longmen Grottoes reveal how the objects became embedded physically with changing notions of pedagogical and museum practices during a dynamic moment in the growth of the field of Asian art. In turn, an understanding of the complex transformations of these heavily restored objects will provide vital support for new digital conservation initiatives at the Longmen Grottoes that hope to identify remaining fragments from the site and return them to their original locations.

The authors of this contribution are medieval art historians, married to each other, who participated from in a “Connecting Art Histories” initiative funded by the J. Paul Getty Foundation. Titled “Global and Postglobal Perspectives on Medieval Art and Art History,” it involved exchange teaching and funded eld trips with graduate students and faculty from the University of Toronto and the Guangzhou Academy of Fine Arts (GAFA), in southern China. In the fall of , Linda Safran taught two courses at GAFA: a large undergraduate survey of medieval art and architecture and a small MA- level seminar on medieval Sicily, in anticipation of a trip to Sicily in February . In the fall of that year, Adam Cohen offered a graduate seminar on medieval manuscript illumination to many of the same students at GAFA. The authors were able to augment memories of their seminar meetings with audio recordings made by the students.

THE SANXINGDUI SITE: ART AND ARCHAEOLOGY VOL. 1 (TEXTS) JAY JIE XU A DISSERTATION PRESENTED TO THE FACULTY OF PRINCETON UNIVERSITY IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY RECOMMENDED FOR ACCEPTANCE BY THE DEPARTMENT OF ART AND ARCHAEOLOGY Advisor: Robert Bagley November 2008 © Copyright by Jay Jie Xu, 2008. All rights reserved To J. C. Abstract The Sanxingdui site is a Neolithic and early Bronze Age site in the Chengdu Plain of western Sichuan in southwest China. In 1986, two pits were found at the site containing an astonishing amount of wealth in the form of hundreds of bronzes alongside stone and jade implements, gold objects, and elephant tusks. The most shocking revelation was the bronze sculptures, which account for the majority of the bronzes. Ranging from miniature to monumental in size, they include human-like figures, fantastic creatures, dragons, birds, and trees the likes of which had never been seen before in Sichuan or anywhere else. This sculptural tradition and iconography contrast sharply with other ancient traditions in China. Around the same time, traces of an ancient wall were recognized, leading eventually to the discovery of a large walled settlement. These finds at Sanxingdui prove beyond doubt that the site was home to a major civilization previously unknown. The present dissertation is a comprehensive survey of art and archaeology of the Sanxingdui site. It attempts to sort out the basics of the site: its archaeological history, the components of its archaeological record, the spatial and temporal dimensions of the site, and the elite material culture evidenced in the contents of the two pits as well as other related finds. Focusing particularly on the two pits, the dissertation addresses the nature of the pits; the typology of artifacts and their origins, functions, and cultural associations; the iconography and original appearances of the bronze sculptures; and the bronze fabrication technology and its cultural implications. iv Acknowledgement I have to overcome some measure of emotion to set words down on these few pages of paper. Although I have written a few acknowledgements for other occasions, this one is distinctly different, because I have been waiting to write it for fifteen years. In some ways, writing it is harder than writing the dissertation itself, for how can I acknowledge in so few words all the debts I have accumulated and express all my gratitude? Indeed, both debts and gratitude reach back much further than fifteen years, since the journey leading to this acknowledgement began long before I arrived at Princeton. What follows, then, acknowledges merely a tiny fraction of my indebtedness. For anyone going through life and accomplishing something along the way, big or small, there are people who nourish, inspire, sustain, and lend a helping hand. I have my set of people, extraordinarily so. Let me begin with my parents. Education was always the center of their values in bringing up their children, and we were always encouraged to achieve the best of what we can, even if the best of what we can achieve is modest. This spirit has sustained my life ever since I became self-conscious. A dissertation that has been pursued for fifteen years side by side with a full-time career is bound to exert significant stress on family life, but throughout the process, my wife Jennifer Chen (a.k.a. Chen Jie) has persevered with me, sustained me, and inspired me with encouragement and love. Her unshakable confidence in me and her sense of responsibility that I ought to finish what I signed up for have awakened a fortitude inside me that I did not know I possessed. As a fellow art historian, she has always been the first reader and critic of my manuscript. As the mother of our daughter, Antonia Erdong Xu v (a.k.a. Toni), she has relieved me of much of the responsibility of a parent so that I could work more than full time (as curators usually do) and still find time for the dissertation. Toni is another inspiration for me, for although she cannot comment on her father’s writing (she is often puzzled to understand what I have been writing about, except that it is for a Ph.D., a “pretty hard degree”), she gives Jennifer and me the purpose and joy of our life. In her own way, Toni has made her tangible contribution to this dissertation: in 2003, she made me a picture of a depressed lion who is nevertheless determined to trudge on. It perhaps captures the mentality of not a few fellow students, Princeton or otherwise, on the same journey. Pasted inside the door of the cabinet for the printer, this picture has charmed me and dared me to take it down, for I am not permitted to do so until the dissertation has been completed. My life with Jennifer, and with Toni since 1997, has been richly rewarded with journeys and explorations: from Shanghai to Princeton, to New York, to Seattle, to Chicago, and now to San Francisco. I am most fortunate that, at every station of my life, there have always been people who believed in me and gave me a chance. Mr. Ma Chengyuan, a leading scholar of ancient Chinese bronzes and former director of the Shanghai Museum, for whom I served as secretary in the beginning of my career, inspired in me a strong interest in Chinese art in general and ancient bronzes in particular. Mr. Wang Qingzheng, former deputy director of the Shanghai Museum and an eminent scholar in Chinese ceramics, rubbings, and numismatics, taught me the importance of broadly-based training. Both of them were my true teachers, although I was not their formal student. vi If there is ever a new beginning in one’s life, that beginning for me was on the day of June 9, 1990, when I set foot on the campus of Princeton. Professor Robert Bagley of the Department of Art and Archaeology was responsible for it. He somehow saw a scholar’s potential in me, and invited me to be his student. His decision changed my life in ways that go far beyond this dissertation. Words fail me in expressing my gratitude and affection. My curatorial career in the United States began in 1996 when I was appointed a curator by Mimi Gates, director of the Seattle Art Museum/Seattle Asian Art Museum. Throughout my seven-year service, Mimi encouraged me to finish the dissertation but never subjected me to pressure. Once I saw in her office a banner from her devotees at Yale, “For God, for Country, for Yale, for Mimi.” I feel the same except that I would change Yale to Princeton! Many other people have mentored me in my life and career. Here I can only mention a few. Professor Wen Fong taught me both in scholarship and in museum work. James Wood, formerly director and president of the Art Institute of Chicago, allowed me time to work on the dissertation at the beginning of my tenure there as head of the Department of Asian Art. Without the support and comradeship of fellow students such as Cary Liu and Dora Ching, I do not think I would have had as splendid a time at Princeton. Last but not least, I am profoundly indebted to the archaeologists in Sichuan who excavated Sanxingdui and other sites in the Chengdu Plain and whose work made possible the research that is reported in this dissertation. They are simply too many to mention by name. The publications I have cited must serve as an index to those vii colleagues. The institutions that kindly granted me unprecedented access to archaeological materials include the Sichuan Provincial Institute of Archaeology, the Sanxingdui Museum, the Sichuan University Museum, and the Chengdu Municipal Institute of Archaeology. To all of them, I extend my gratitude. I would also like to thank the members of my dissertation committee, Professors Jerome Silbergeld and Andrew Watsky of the Department of Art and Archaeology of Princeton University and Professor Robert Murowchick of Boston University, for their generous help. Professor Robert Bagley, my principal advisor, demonstrated unfailing patience over the course of fifteen years during which he continued to guide me as his “student who does not graduate.” In essence, I will indeed not graduate, for I continue to look up to him for guidance. And as a student, one usually has more leeway for making mistakes. Undoubtedly, this dissertation has its fair share of mistakes and errors, for which I am solely responsible. viii Table of Contents Abstract iv Acknowledgement v Table of Contents ix List of Tables xiii List of Figures xiv Chapter 1. Introduction Chapter 2. The Archaeological Record of the Sanxingdui Site: A Critical Review 2.1 History of discoveries and fieldwork at the Sanxingdui site 2.1.1 2.4 10 Culture 14 2.1.3 Discovery in 1986: K1 and K2 18 2.1.4 Fieldwork from 1984 to present: recognition of the Sanxingdui 2.1.5 2.3 10 Fieldwork from 1951 to 1986: recognition of the Sanxingdui city 2.2 8 Initial chance discovery at Yueliangwan in 1927 and subsequent excavation in 1934: recognition of the “Hanchow culture” 2.1.2 1 20 The amount of space excavated and publication of the fieldwork 22 2.1.6 Fieldwork at related sites 24 Stratigraphy, pottery typology, periodization, and absolute date 28 2.2.1 Stratigraphy, pottery typology, and periodization 29 2.2.2 36 Absolute date Size of the city, city walls, pits of objects, and other features 38 2.3.1 Settlement during Phase I 38 2.3.2 Settlement of Phases II through IV 40 Defining the Sanxingdui Culture 52 Material Culture of the Sanxingdui Elite: K1 and K2 61 3.1 Research approaches 61 3.2. Burial content and condition 64 Chapter 3. ix 3.3 Nature of the pits 69 Bronzes at Sanxingdui (I): Sporadic Finds and Vessels 79 4.1 Bronzes not from K1 and K2 79 4.2 Vessels from K1 and K2 83 4.2.1 Vessels from K1 84 4.2.2 Vessels from K2 88 4.2.2.1. Vessels of Style III 90 4.2.2.2. Vessels of Styles IV and V(a) 92 4.2.2.3. The Central Plain connection 96 Chapter 4. 4.2.2.4. The middle Changjiang connection and related issues 100 4.2.3 Chapter 5. 5.1 5.2 Function of the vessels 109 Bronzes at Sanxingdui (II): Images and Related Implements 113 Iconography and reconstruction: heads, figures, and masks 114 5.1.1 Facial features 115 5.1.2 Hairstyle and headdress 116 5.1.3 Coloring of facial features 119 5.1.4 The life-sized figure 122 5.1.5 Installation of the heads and their likely appearances 126 5.1.6 Masks 131 5.1.7 133 Stylistic differences between images from K1 and K2 Iconography and reconstruction: trees and associated images and implements 135 5.3 Original ensemble of the images 141 5.4 Other images from K1 and K2 146 5.5 Sources of Sanxingdui imagery 151 5.6 The function of the Sanxingdui images and the reason for their Chapter 6. sacrifice 153 Bronzes at Sanxingdui (III): Casting Technology 156 x 6.1 6.2 Casting technology in the Central Plain 156 6.1.1 Basic techniques 156 6.1.2 The “one-pour mentality” in foundry practice 159 6.1.3 Nature of the clay section-mold casting 161 Foundry practice at Sanxingdui 162 6.2.1 Objects cast in one pour of metal 163 6.2.2 Objects cast in separate pours of metal 165 6.2.2.1. Precasting, casting on, running on, and soldering 166 6.2.2.2. Concurrent use of joining techniques 168 6.2.2.3. The life-sized figure 170 6.2.2.4. The monumental tree 172 6.2.2.5. Devices of mechanical interlock 175 Origin and characteristics of the Sanxingdui foundry practice 177 6.2.3.1. The Central Plain origin 177 6.2.3.2. Characteristics of the foundry practice 179 6.2.3 Chapter 7. 7.1 Lithic and Gold Artifacts at Sanxingdui 184 Lithic artifacts 184 7.1.1 Types, typological sources, and functions 184 7.1.1.1. Sets of stone bi of graduated sizes 188 7.1.1.2. Collared types, and other circular types 190 7.1.1.3. Cong tubes 194 7.1.1.4. Trapezoidal and parallelogram-shaped implements 195 7.1.1.5. Forked blades 197 7.1.1.6. Ge blades, and hybrids of forked blade and ge 204 7.1.1.7. Other blades and implements 209 History and material of local production 211 7.1.2 7.2 Chapter 8. 8.1 Gold artifacts 214 The Sanxingdui site in the context of ancient China 219 Cultural contacts in Phase I 220 xi 8.2 Cultural contacts in Phases II–IV 221 8.2.1 Central Plain contacts 221 8.2.2 The middle Changjiang contacts 224 8.2.3 The northern region contacts 226 8.3 Large network of exchange, limited participation 228 8.4 Qualified isolation: geography and the formation of the Sanxingdui Chapter 9. Culture 230 Conclusions 233 Chapter 10. Epilogue: Expanding Horizon and Legacy of the Sanxingdui Culture 240 10.1 The Shi’erqiao and related sites 240 10.2. The Jinsha site 243 Appendix. Elemental Analysis and Lead-Isotope Ratios at Sanxingdui and other Early Bronze Age Sites 254 Chinese Glossary 257 Works Cited 264 Tables 298 Figures 311–418 xii List of Tables (pp. 298–310) Table 1a Stratigraphy at the Sanxingdui Site from 1980–1986. Adapted from Chen Xiandan 1989a, p. 217 table. Table 1b Periodization Schemes and Cultural Identifications of the Sanxingdui Site. Table 2 Radiocarbon Dates from the Sanxingdui Site. After Xu 2003, table 2. Table 3 Inventories of K1 and K2. After Falkenhausen 2002, pp. 90–2 table 1. Table 4 Elemental Composition of Bronzes from Early Bronze Age Sites. Adapted from Xu 2001, p. 69, table 1. xiii List of Figures (pp. 311–418) Figure 1.1 Topographical map of China. After Ebrey 1996, p. 11. Figure 1.2 Map of Sichuan province (including the Chongqing municipality). After Hong Kong 2007, p. 23 fig. 1. Figure 1.3 Bronze mask with protruding pupils, SXD, K2(2):148, h. 66 cm, w. 138 cm, depth 73 cm. After Bagley 2001, p. 109 no. 22. Figure 2.1 Location of the Sanxingdui site. After Paris 2003, p. 28. Figure 2.2 Extent of the Sanxingdui site. After Beijing 1999a, p. 11 fig. 2. Figure 2.3 The walled city, major finds and fieldwork of the Sanxingdui site. After Xu 2003, p. 150, fig. 1. Figure 2.4 Objects unearthed in 1927 at Yueliangwan. After Dye 1931. Figure 2.5 Excavation loci at Sanxingdui from 1980 to 1986. After Xu 2003 p. 154, fig. 2. Figure 2.6 K1 under excavation. After Beijing 1994a, p. 106. Figure 2.7 K2 under excavation. After Bagley 1990a, p. 55 fig. 9. Figure 2.8 The west wall, looking south. After Tokyo 1998, p. 42 fig. 3. Figure 2.9 Layout of the Rensheng cemetery. After Kaogu 2004.10, p. 15 fig. 2. Figure 2.10 Locations of walled settlements of the Baodun Culture, and other Neolithic sites in the Sichuan Basin. After Tokyo 2000, p. 101 fig. 66. Figure 2.11 F5: the largest house foundation at the Gucheng site. After Wenwu 2001.3, p. 58 fig. 12. Figure 2.12 Pottery types at the Sanxingdui site. After Beijing 1999a, p. 425 figs. 225– 6, p. 426 figs. 227–8. a: Phase I; b: Phase II; c: Phase III; d: Phase IV. Figure 2.13 Jiandi zhan from K1. After Beijing 1999a, p. 146 fig. 76. Figure 2.14 A reconstructed layout of the Sanxingdui city. After Sun Hua 2000, p. 164 fig. 6.9. xiv Figure 2.15 Building foundations at Sanxingdui Locus III. After Kaogu xuebao 1987.2, p. 233 fig. 6. Figure 2.16 Pottery dou with openwork ring foot. a: M10:8, Rensheng cemetery, overall h. 24 cm. After Kaogu 2004.10, p. 19 fig. 11:5. b: AT2806(3):4, ring foot, Xiaojiawuji site of the Shijiahe Culture, existing h. 16.5 cm. After Beijing 1999e, p. 191 fig. 148:5. Figure 2.17 Jade awl-shaped implements. a: M5:6, Rensheng cemetery, l. 29.2 cm. After Kaogu 2004.10, p. 20 fig. 12:22. b: AT1215(2):2, Xiaojiawuji, l. 13.5 cm. After Beijing 1999e, p. 329 fig. 260:2. c. M9:4, Fuquanshan site of the Liangzhu Culture, l. 27 cm. After Huang Xuanpei 2000, p. 84 fig. 63.12. Figure 2.18 Cross-section of the Yueliangwan wall excavated in 1999. Photograph by author, February 27, 2000. Figure 2.19 Adobe brick atop the east wall. Photograph courtesy of Chen De’an. Figure 2.20 Pottery types from Shi’erqiao. After Sun Hua 1996, pp. 125–6 figs. 2–3. Figure 3.1 K1, line drawing. After Beijing 1999a, p. 20 fig. 8. Figure 3.2 K2, line drawing. After Beijing 1999a, p. 159 fig. 81. Figure 3.3 K2, middle layer. After Taibei 1999, p. 196. Figure 3.4 K2, top layer. The trench at the top of the photograph is not a ramp but an intrusive ditch dug in a later period. After Taibei 1999, p. 194. Figure 4.1 Bronze tiger-shaped plaques. a: 1984 Rensheng village near the Yazihe River, l. 38 cm. After Beijing 1994a, pl. 66. b: Rensheng village, l. 43.4 cm, h. 13.2 cm. After Bagley 2001, no. 40. c: Jinsha, l. 26.5 cm. After Chengdu 2005, p. 48 no. 45. Figure 4.2 Rectangular bronze plaques from the 1987 Cangbaobao pit. After Beijing 1998a, p. 81 fig. 3. a: 87GSZJ:16, l. 14 cm, w. 4.9–5.3 cm. b: 87GSZJ:36, l. 13.8 cm, w. 5.2–5.6 cm. c: 87GSZJ:17, l. 13.8 cm, w. 5.2–5.8 cm. xv Figure 4.3 Rectangular bronze plaques from other sites. a: Gaopian, l. 12.3 cm, w. 4.3–5 cm. After Ao & Wang 1980, p. 76 fig. 2.4. b: Erlitou, 87M57:4, l. 15.9 cm, w. 7.5–8.9 cm, ca. 16th century BC. After Kaogu 1992.4, p. 296 fig. 2.1. Figure 4.4 Lithic spearheads. a: Gaopian, jade, h. 23 cm. After Ao & Wang 1980, p. 76 fig. 2.3. b: SXD, K1:141, stone, h. 34.2 cm. After Beijing 1999a, p. 128 fig. 67.1. c: SXD, K1:137, stone, h. 23 cm. After Beijing 1999a, p. 128 fig. 67.2. d: Jinsha, 2001CQJC:18, jade, h. 24.51 cm. After Beijing 2002c, p, 153. Figure 4.5 Jades with notched edges. a: axe, Gaopian, h. 18 cm. After Ao & Wang 1980, p. 76 fig. 2.1. b: hatchet-shaped implement, SXD, K1:235, h. 12.4 cm. After Beijing 1999a, p. 84 fig. 44.3. c: axe, Erlitou, h. 21 cm, ca. 16th century BC. After Kaogu 1984.1, p. 38 fig. 5.2. Figure 4.6 Bronze pan, K1:53, h. 10.4 cm. Reconstruction drawing after Beijing 1999a, p. 43 fig. 26.1. Figure 4.7 Bronze lid, K1:135, h. 11.2 cm. Reconstruction drawing after Beijing 1999a, p. 43 fig. 26.2. Figure 4.8 Bronze zun, K1:163/K1:59. After Beijing 1999a, p. 41 fig. 24. Figure 4.9 Bronze pou, K1:130. After Beijing 1999a, p. 45 rubbing no. 4.1. Figure 4.10 Bronze zun, K1:158/258, h. 43.3 cm. After Beijing 1999a, p. 35 fig. 23. Figure 4.11 Bronze zun from Anhui Funan, h. 50.5 cm, 13th century BC. After Beijing 1996b, pp. 117–8. a: side view centered on flange. b: detail of tiger and human. Figure 4.12 Bronze zun, K2(2):112, h. 31.5 cm. After Beijing 1999a, p. 241 fig. 137. Figure 4.13 Bronze zun, K2(2):135. After Beijing 1999a, p. 239 fig. 135. Figure 4.14 Bronze zun, K2(2):109. Reconstruction drawing after Beijing 1999a, p. 240 fig. 136. Figure 4.15 Bronze zun, K2(2):79, h. 44.2 cm. a: after Beijing 1999a, p. 242 fig. 138. b: after ibid., p. 246 rubbing no. 18. xvi Figure 4.16 Bronze lei, K2(2):70, h. 33.4 cm. After Beijing 1999a, p. 263 fig. 145. Figure 4.17 Bronze lei, K2(2):88, h. 35.4 cm. After Beijing 1999a, p. 264 fig. 146. Figure 4.18 Bronze fang lei, K2(3):205/K2(3):205-1, h. 35.6 cm. After Beijing 1999a, p. 279 rubbing no. 29. Figure 4.19 Bronze lid, K2(2):32, h. 8.1 cm. After Beijing 1999a, p. 275 fig. 149. Figure 4.20 Bronze zun, K2(2):127, h. 41.6 cm. After Beijing 1999a, p. 253 fig. 140. Figure 4.21 Bronze zun, K2(2):129, h. 45.5 cm. a: after Beijing 1999a, p. 254 fig. 141. b: after ibid., p. 257 rubbing no. 20. Figure 4.22 Bronze zun, K2(2):146, h. 52.6 cm. After Beijing 1999a, p. 252 fig. 139. Figure 4.23 Bronze zun, K2(2):151, h. 56.5 cm. After Beijing 1999a, p. 255 fig. 142. Figure 4.24 Bronze lei, K2(2):159, h. 54 cm. After Beijing 1999a, p. 265 fig. 147. Figure 4.25 Bronze lei, K2(2):39/K2(2):39-1. After Beijing 1999a, p. 274 fig. 148.2–3. Figure 4.26 Bronze lei, K2(2):103/K2(2):103-1/K2(2):103-2. After Beijing 1999a, p. 278 rubbing no. 27. Figure 4.27 Bronze bird for a vessel lid, K2(3):193-1, h. 34 cm. After Beijing 1999a, p. 334 fig. 184. Figure 4.28 Bronze fragment, probably of a vessel lid, K2(3):23, h. of bird, 13.3 cm. After Beijing 1999a, p. 266 rubbing no. 23. Figure 4.29 Bronze lid (MA1612) said to come from Hunan Changsha, h. 28 cm, late 13th century BC, Musée Guimet, Paris. After Bagley 2001, p. 150 fig. 50.1. Figure 4.30 Bronze zun from Anyang, M18:13, h. 53 cm, early 12th century BC. After Beijing 1985b, fig. 53. Figure 4.31 Bronze kneeling figure bearing a zun, K2(3):48, h. 15.6 cm. After Beijing 1999a, p. 170 fig. 86. Figure 5.1 Bronze head covered with gold foil, K2(2):45. Before restoration. Photograph courtesy of Yang Xiaowu. See Figure 5.16 for the restored head. xvii Figure 5.2 Fragments of bronze tree before restoration (the photograph shows the trees in Figures 5.30 and 5.40, and two smaller bases). After Bagley 2001, p. 116 fig. 27.1. Figure 5.3 Bronze head, K1:2, h. 29 cm, greatest width 20.6 cm, weight 4.48 kg. After Beijing 1999a, p. 24 fig. 9. Figure 5.4 Bronze head, K1:6, h. 25 cm, greatest width 20.4 cm, weight 3.36 kg. After Beijing 1999a, p. 24 fig. 10. Figure 5.5 Bronze head, K1:5, h. 45.6 cm, greatest width 22 cm, weight 4.54 kg. After Beijing 1999a, p. 29 fig. 17. Figure 5.6 Bronze head covered with gold foil, K2(2):214, h. 48.1 cm, greatest width 22 cm. After Beijing 1999a, p. 184 fig. 101. Figure 5.7 Bronze head, K2(2):58, h. 51.6 cm, greatest width 23.8 cm, weight 5.8 kg. After Beijing 1999a, p. 178 fig. 96. Figure 5.8 Bronze head, K2(2):90, h. 34.8 cm, greatest width 17.2 cm, weight 2.081 kg. After Beijing 1999a, p. 177 fig. 95. Figure 5.9 Bronze head, K2(2):83, h. 13.6 cm, greatest width 10.8 cm, weight 0.71 kg. After Beijing 1999a, p. 172 fig. 88. Figure 5.10 Bronze head, K1:7, h. 27 cm, greatest width 22.8 cm, weight 7.66 kg. After Beijing 1999a, p. 27 fig. 12. Figure 5.11 Bronze head, K1:72, h. 30.8 cm, greatest width 17.8 cm, weight 2.01 kg. After Beijing 1999a, p. 27 fig. 14. Figure 5.12 Bronze head, K1:10, h. 27.8 cm, greatest width 19.4 cm, weight 4.36 kg. After Beijing 1999a, p. 28 fig. 15.2. Figure 5.13 Bronze head, K2(2):17, h. 40 cm, greatest width 18.2 cm, weight 4 kg. After Beijing 1999a, p. 176 fig. 93. Figure 5.14 Bronze head, K2(2):34, h. 36.8 cm, greatest width 17.4 cm, weight 2.4 kg. After Rawson 1996, p. 65 fig. 24.1. Figure 5.15 Bronze head, K2(2):154, h. 17.6 cm, greatest width 10.8 cm, weight 0.691 kg. After Beijing 1999a, p. 175 fig. 89. Figure 5.16 Bronze head covered with gold foil, K2(2):45, h. 42.5 cm, greatest width 19.6 cm, weight 2.55 kg. After Beijing 1999a, p. 183 fig. 98. xviii Figure 5.17 Bronze figure on pedestal, K2(2):149, 150, overall h. 260.8 cm, h. of figure 172 cm. a-e: line drawings from various angles. After Beijing 1999a, foldout following p. 162. f: design of headdress. After Beijing 1999a, p. 164 rubbing 11. g: outer garment. After Wang & Wang 1993, p. 62 fig. 2. h: middle garment. After ibid. i: inner garment. After ibid. j: dragons on the mantle. After Bagley 2001, p. 74 fig. 2.4. k: detail of decorated pedestal. Photograph by author. Figure 5.18 Bronze figure with an animal headdress, K2(3):264, h. 40.2 cm. After Beijing 1999a, p. 167 fig. 84. Figure 5.19 Bronze plaque in the form of a kneeling figure, K2(3):04, h. 13.3 cm. After Beijing 1999a, p. 170 fig. 85.3. Figure 5.20 Bronze figure, K2(3):296-1, h. of figure 10.8 cm. After Beijing 1999a, p. 234 fig. 130.1. Figure 5.21 Bronze figure, K2(3):292-2, h. 8.3 cm. After Beijing 1999a, p. 167 fig. 83. Figure 5.22 Bronze hybrid figure standing on birds, K2(3):327, overall h. 81.4 cm, h. of figure 30 cm. After Beijing 1999a, p. 171 fig. 87. Figure 5.23 Bronze pedestal or miniature building with kneeling figure, K2(2):143-1, fragment, h. 31 cm. After Beijing 1999a, p. 234 fig. 131. Figure 5.24 Bronze kneeling figure holding a forked blade, K2(3):325, h. 4.7 cm. After Beijing 1999a, p. 235 fig. 133. Figure 5.25 Bronze diamond-shaped appliqués. After Beijing 1999a, p. 208 fig. 115. a: integral appliqué, K2(3):202, l. 57.2 cm, w. 23.6 cm. b: two halves, K2(3):197, l. 54.8 cm, w. 12.7 cm; K2(3):8, l. 54.8 cm, w. 12.8 cm. c: four quarters, K2(3):101, l. 29 cm, w. 12.4 cm; K2(3):106, l. 28 cm, w. 12.4 cm; K2(3):8-1, l. 28.6 cm, w. 12.4 cm; K2(3):99, l. 27.8 cm, w. 13.2 cm. Figure 5.26 Bronze mask, K2(2):293, h. 25.5 cm, w. 37.8 cm. After Beijing 1999a, p. 189 fig. 104.2. Figure 5.27 Bronze mask, K2(2):153, h. 40.3 cm, w. 60.5 cm. After Beijing 1999a, p. 189 fig. 103.2. xix Figure 5.28 Bronze mask with protruding pupils, K2(2):148, h. 66 cm, w. 138 cm, depth 73 cm. After Beijing 1999a, p. 197 fig. 110. Figure 5.29 Bronze mask with protruding pupils and a trunk, K2(2):142, overall h. 82.5 cm, h. of mask 31.5 cm, w. 77 cm. After Beijing 1999a, p. 197 fig. 108. Figure 5.30 Bronze tree, K2(2):94, overall h. 3.96 m, h. of trunk 3.59 m, d. of base 0.93 m. After Beijing 1999a, foldout page facing p. 218. Figure 5.31 Bronze bird from a tree, K2(2):213, h. 8 cm. After Beijing 1999a, p. 227 fig. 126.1. Figure 5.32 Bronze bird with human head, K2(2):154, h. 12 cm. Reconstruction drawing after Taibei 1999, p. 99. Figure 5.33 Bronze bells. After Beijing 1999a, p. 292 fig. 161.2, p. 293 figs. 162.1– 162.2, p. 299 figs. 163.1–163.2, 164.1–164.2. a: K2(3):103-31, h. 10.2 cm. b: K2(3):103-28, h. 7.35 cm. c: K2(3):274, h. 8.3 cm. d: K2(3):149, h. 14.3 cm. e: K2(2):103-8, h. 14 cm. f: K2(3):70-7, h. 7.6 cm. g: K2(3):78, h. 12.2 cm. Figure 5.34 Bronze shells. After Beijing 1999a, p. 300 figs. 165.1, 165.4, p. 301 fig. 166.5, p. 310 fig. 169.1. a: K2(2):79-6, d. 9.1 cm. b: K2(3):103-22, d. 6.9 cm. c: K2(3):6, h. 12.7 cm. d: K2(3):265-1, h. 9.5 cm. Figure 5.35 Bronze hanging apparatus, K2(3):124, h. 6.2 cm. After Beijing 1999a, p. 314 fig. 171.1. Figure 5.36 Bronze foils. After Beijing 1999a, p. 320 figs. 173.3, 174.2, 175.2. a: fish-shaped foil, K2(3):194-6, l. 9.8 cm. b: forked-blade-shaped foil, K2(3):194-2, l. 6.5 cm. c: foil with vine pattern, K2(3):194-13, size unknown. Figure 5.37 Fragment of a bronze tree with a bronze collared disk attached to the trunk, K2(3):204, 261, overall h. 59.7 cm, diameter of disk 8.8 cm. After Beijing 1999a, p. 222 fig. 123.1. xx Figure 5.38 Fragment of a bronze tree with a jade collared disk attached to the calyx of a flower, K2(3):20, overall height 50 cm. After Beijing 1999a, p. 225 fig. 124. Figure 5.39 Bronze collared disks. a: SXD K2(3):134, d. 11.2 cm. After Beijing 1999a, p. 282 fig. 154.2. b: SXD K2(2):99, l. 8.9 cm. After Beijing 1999a, p. 285 fig. 157.5. Figure 5.40 Bronze tree (fragment), K2(2):194, overall h. 193.6 cm. a: after Tokyo 1998, p. 99. b: line drawing after Beijing 1999a, p. 220 fig. 121. Figure 5.41 Bronze tree, K2(3):272, overall h. 50 cm. Reconstruction drawing after Beijing 1999a, p. 226 fig. 125. Figure 5.42 Jade implement with incised figures, K2(3):201-4, l. 54.2 cm. After Beijing 1999a, p. 361 fig. 197.1. Figure 5.43 Bronze altar?, K2(3):296, h. 54 cm. Reconstruction drawing based on surviving fragments, after Beijing 1999a, p. 233 fig. 129. Figure 5.44 Bronze bird’s head, K2(2):141, h. 40.3 cm. After Beijing 1999a, p. 333 fig. 183.1. Figure 5.45 Bronze bird on a post, K2(3):301-3, h. 27.8 cm. After Beijing 1999a, p. 335 fig. 185.3. Figure 5.46 Bronze rooster, K2(3):107, h. 14.2 cm. After Beijing 1999a, p. 333 fig. 183.2. Figure 5.47 Bronze tube with dragon, K1:36, h. 41 cm. After Beijing 1999a, p. 34 fig. 20. Figure 5.48 Bronze snake. After Beijing 1999a, p. 326 fig. 178.1, p. 327 figs. 179.1, 179.3. a: head, K2(3):87, l. 54.8 cm. b: middle section, K2(3):56, l. 35.6 cm. c: tail, K2(3): 44, l. 21.2 cm. Figure 5.49 Bronze tiger-like creature, K1:62, l. 11.4 cm, h. 10.8 cm. After Beijing 1999a, p. 35 fig. 22. Figure 5.50 Bronze seated figure, K1:293, h. 14.6 cm. After Beijing 1999a, p. 29 fig. 18. xxi Figure 5.51 Bronze plaques of taotie faces. After Beijing 1999a, p. 198 figs. 111.2 and 112.1, p. 203 fig. 113.3. a: K2(3):221, h. 21.2 cm, w. 35 cm. b: K2(3):231, h. 20.8 cm, w. 26.4 cm. c: K2(3):231-1, h. 12.3 cm, w. 27.8 cm. Figure 5.52 Bronze circular appliqué, K2(3):1, d. 84 cm. After Beijing 1999a, p. 239 fig. 134.1. Figure 5.53 Bronze fragment of a miniature roof, K2(2):143, h. 15.8 cm. After Beijing 1999a, p. 235 fig. 132. Figure 5.54 Comparisons of Shijiahe and Sanxingdui heads. After Falkenhausen 2003, p. 194 fig. 2. Figure 6.1 a: bronze ding, PLZM2:36, Hubei Huangpi Panlongcheng, h. 30.1 cm, 15th–14th century BC. After Beijing 1996b, pl. 32. b: diagram showing the relationship between the Panlongcheng ding and the mold used to cast it. After Fong 1980, p. 72 fig. 16. Figure 6.2 Bronze gui, Freer Gallery of Art (31.10). Detail of precast flange. After Gettens 1969, p. 93 fig. 94. Figure 6.3 a: bronze fang ding, XDM:8, Jiangxi Xi’gan Dayangzhou, h. 97 cm, 14th century BC. After Beijing 1997d, p. 33 fig. 20(A). b: detail of cast-on tiger. After Hong Kong 1994, pl. 35-3. Figure 6.4 Bronze fang ding from Shanxi Pinglu Qianzhuang, h. 82 cm, 14th century BC. After Beijing 1996b, pl. 37. Figure 6.5 Bronze jia, Freer Gallery of Art (35.12). Detail of capped post with run-on join. After Gettens 1969, p. 94 fig. 98. Figure 6.6 Bronze guang, Freer Gallery of Art (39.53). Detail of brazed handle. After Gettens 1969, p. 87 fig. 83. Figure 6.7 Bronze ding from Henan Zhengzhou, h. 77.3 cm, 14th century BC. After Tokyo 1986, p. 62 no. 32. Figure 6.8 a: bronze fang yi, h. 29.8 cm, ca. 12th century BC, Harvard University Art Museums (1943.52.109). After Bagley 1990b, p. 7 fig. 1B. b: diagram showing the relationship between the Harvard fang yi and the mold used to cast it. After Bagley 1990b, p. 9 fig. 5. Figure 6.9 a: bronze zun, h. 36.8 cm, ca. 12th century BC. Freer Gallery of Art (51.19). After Pope et al. 1967, p. 99 pl. 16. xxii b: detail of animal head. After Pope et al. 1967, p. 101 fig. 11. Figure 6.10 Bronze ge with serrated edges a: Sanxingdui, K1:247-2, l. 20.5 cm. After Beijing 1999a, p. 56 fig. 32.5. b: Jinsha, 2001CQJC:169, l. 21.5 cm. After Beijing 2002c, p. 56. Figure 6.11 Bronze mask, K2(2):331, h. 15.4 cm, w. 18.2 cm. After Tokyo 1998, p. 79 (below). Figure 6.12 Bronze mask, K2(3):14 (fragment), h. 25.6 cm. Left side. Photograph by author. Figure 6.13 Bronze head, K1:2, in Figure 5.3. Photographs by author. a: detail of area above right ear. b: casting-on repairs viewed from interior. Figure 6.14 Bronze head, K2(2):82, h. 39.6 cm. Right side. Photograph by author. Figure 6.15 Bronze head, K2(2):154, in Figure 5.15. Detail of top. After Tokyo 1998, p. 57 (left). Figure 6.16 Bronze head, K1:7, in Figure 5.10. Detail of top. Photograph by author. Figure 6.17 Bronze head covered with gold foil, K2(2):214, in Figure 5.6. a: detail of top. Photograph by author. b: right side. After Yang 1999, p. 211. Figure 6.18 Bronze head, K1:5, in Figure 5.5. a: detail of top. After Tokyo 1998, p. 154 (lower). b: left side. Photograph by Paul Macapia. Figure 6.19 Bronze head, K2(2):51, h. 40.4 cm. Detail of top. Photograph by author. Figure 6.20 Bronze mask with protruding pupils, K2(2):148, in Figures 1.3/5.28. a: detail of left eye. Photograph by author. b: detail of nostrils. Photograph by author. c: rear view. Photograph by Paul Macapia. Figure 6.21 Bronze head, K2(2):83, in Figure 5.9. Left side. After Tokyo 1998, p. 117 (lower). Figure 6.22 Bronze head, K2(2):107, h. 36.6 cm. Detail of top. Photograph by author. Figure 6.23 Bronze head, K2(2):90, in Figure 5.8. a: detail of top. Photograph by author. b: rear view. After Tokyo 1998, p. 60 (right). xxiii Figure 6.24 Bronze head, K2(2):14, h. 39.5 cm. Detail of top. Photograph by author. Figure 6.25 Bronze head covered with gold foil, K2(2):115, h. 41 cm. Detail of top. Photograph by author. Figure 6.26 Bronze tree (fragment), K2(2):194, in Figure 5.40, a–e, photographs by author; f: after Beijing 1994a, pl. 41. a: detail of soldered or run-on rim. b. detail of a kneeling figure. c. detail of a hole on the rim. d: detail of damaged side. e: cast rivet seen from inside. f: view of another side. Figure 6.27 Bronze mask, K2(2):128, h. 25.4 cm. Photographs by author. a: left ear, back side. b: detail of right inside. Figure 6.28 Bronze hybrid figure standing on birds, K2(3):327, in Figure 5.22. Detail of soldered join between the figure’s foot and the bird. After Bagley 2001, p. 128 fig. 36.3. Figure 6.29 Bronze mask with protruding pupils and a trunk, K2(2):142, in Figure 5.29. a: rear view. After Rawson 1996, p. 67 fig. 25.2. b: right side. After Taibei 1999, p. 80. c: detail of left ear, back side. Photograph by author. Figure 6.30 Bronze figure with an animal headdress, K2(3):264, in Figure 5.18. a: rear view. After Tokyo 1998, p. 118 (right). b: detail of top. Photograph by author. c: detail of left projection with solder reinforcement. Photograph by author. d: detail of neck, right side. Photograph by author. Figure 6.31 Bronze figure on pedestal, K2(2):149, 150, in Figure 5.17. a: detail of the decorated part of the pedestal. After Yang 1999, p. 210 (right). b: detail of back. After Tokyo 1998, p. 48 (left). c: detail of lower part. After Tokyo 1998, p. 48 (right). d: detail of lower right side. Photograph by author. e: before restoration. After Beijing 1999a, p. 165 pl. 57. f: detail of left side. After Bagley 2001, p. 74 Figure 2. g: detail of upper part. After Tokyo 1998, p. 49. Figure 6.32 Bronze tree, K2(2):94, in Figure 5.30. a: detail of base. After Tokyo 1998, p. 97 (below). xxiv b: detail of a branch join. Photograph by author. c: detail of join of branches to trunk. After Tokyo 1998, p. 97 (above). d: detail showing the dragon. After Tokyo 1998, p. 95. e: detail of one of the four struts joining the dragon to the trunk. Photograph by author. f: detail of a soldered join on the dragon. Photograph by author. g: detail of a bird. After Tokyo 1998, p. 96. h: detail of a calyx. Photograph by author. i: cross section of a broken branch. Photograph by author. Figure 6.33 Bronze circular appliqué, K2(3):1, in Figure 5.52. Detail of join at rim. After Bagley 2001, p. 135 fig. 41.2. Figure 6.34 Bronze mask, K2(2):114, h. 26.6 cm. Photographs by author. a: detail of left ear, back side. b: detail of right ear, back side. Figure 7.1 Stone bi from the 1987 Cangbaobao pit. a: larger ones, diameter of the largest one (87GSZJ:21), 20.3 cm. After Beijing 1998a, p. 84 fig. 7. b: smaller ones, diameter of the smallest one (87GSZJ:9), 3.1–3.5 cm. After Beijing 1998a, p. 86 fig. 9. Figure 7.2 Stone bi with undrilled center, collected in 1976 at Yinchuan in Ningxia Hui Autonomous Region. Qijia Culture, early 2nd millennium BC. After Bagley 2001, p. 159 fig. 8. Figure 7.3 Jade collared types. a: Sanxingdui, ring, K2(2):5-4, overall d. 8.5 cm, d. of perforation 5.5 cm. After Beijing 1999a, p. 371 fig. 204.3. b: Sanxingdui, ring, K2(2):146-3, overall d. 13.7 cm, d. of perforation 6.8 cm. After Beijing 1999a, p. 370 fig. 203.1. c: Sanxingdui, ring, from the 1927 Yueliangwan pit, overall d. 11.2 cm, d. of perforation 7 cm. Sichuan Provincial Museum (110483). After Xiao et al. 2001, p. 15. d: Sanxingdui, disk, K2(2):57-2, overall d. 17.5 cm, d. of perforation 6.7 cm. After Beijing 1999a, p. 369 fig. 202.1. e: Sanxingdui, disk, K2(2):146-2, overall d. 17.8 cm, d. of perforation 6.7 cm. After Beijing 1999a, p. 369 fig. 202.2. f: Sanxingdui, irregularly shaped implement with collaring perforation, K1:204, l. 20.8, d. of perforation 3.3 cm. After Beijing 1999a, p. 84 fig. 43.1. g. Xin’gan, XDM:651, overall d. 16.8 cm, d. of perforation 7.2 cm. After Beijing 1997d, p. 144 fig. 75.1. h. Anyang, 01HDM54:352, overall d. 17.6 cm, d. of perforation 5.7 cm. After Beijing 2005, p. 6. xxv i: Jinsha, 2001CQJC:2, overall d. 16.96 cm, d. of perforation 6.2–6.42 cm. After Beijing 2002c, p. 94. Figure 7.4 Jade circular types. a: Sanxingdui, 1987 Cangbaobao pit, bracelet-like ring, 87GSZJ:33, d. 5.6 cm, h. 2.2 cm. After Beijing 1998a, p. 83 fig. 6.4. b: Sanxingdui, 1987 Cangbaobao pit, ring, 87GSZJ:11, d. 9 cm, d. of perforation 6.4 cm. After Beijing 1998a, p. 82 fig. 5.1. c: Jinsha, bracelet-like ring, 2001CQJC:172, d. 6.92–7.10 cm, h. 3.68 cm. After Beijing 2002c, p. 90. Figure 7.5 Jade cong tubes. a: 1927 Yueliangwan pit, h. 5.5 cm. After Graham 1934, plate following p. 128. b: 1927 Yueliangwan pit, h. 7.2 cm. After Tokyo 1998, p. 176 no. 152. c: 1927 Yueliangwan pit, AK:2.2, 110485, h. 11 cm. Sichuan Provincial Museum. Photograph by author. d: Sanxingdui, K1:11-2, h. 7.6 cm. After Beijing 1999a, p. 81 fig. 42.1. e: Jinsha, 2001CQJC:61, h. 22.26 cm. After Beijing 2002c, p. 84. f: Jinsha, 2001CQJC:1, h. 16.57 cm. After Beijing 2002c, p. 86. Figure 7.6 a: Jade trapezoidal implement, Sanxingdui, K1:81, 97, l. 1.62 m. After Beijing 1999a, p. 64 fig. 35. b: Jade trapezoidal knife, unearthed at Erlitou in 1975, late 16th century BC, l. 60.4–65 cm. After Kaogu 1978.4, p. 270 fig. 1.3. c: Jade rectangular implement, from Hunan Shimen Weigang, M1:07, ca. 1500 BC, l. 48 cm. After Wang & Long 1987, p. 17 fig. 8.1. Figure 7.7 Jade parallelogram-shaped implements. a: Sanxingdui, K2(3):150, l. 36.1 cm. After Beijing 1999a, p. 360 fig. 196.1. b: Sanxingdui, K2(3):194, l. 66.4 cm. After Beijing 1999a, p. 361 fig. 197.2. c: Jinsha, 2001CQJL10:16, l. 18.3 cm. After Beijing 2006b, p. 76. Figure 7.8 Map of the distribution of forked blades. After Deng Cong 1994, end papers. Figure 7.9 Jade forked blades from sites other than Sanxingdui. a: SSY7, from Shaanxi Shenmu Shimao, l. 26.5 cm. Shaanxi Provincial Museum. After Deng Cong 1994, fig. A:6. b: from Shandong Linyi Dafanzhuang, l. 32.5 cm. After Deng Cong 1994, fig. D:7. c: SSY16, from Shaanxi Shenmu Shimao, l. 34.5 cm. Shaanxi Provincial Museum. After Deng Cong 1994, fig. A:14. xxvi d: SSY15, from Shaanxi Shenmu Shimao, l. 30.6 cm. Shaanxi Provincial Museum. After Deng Cong 1994, fig. A:13. e: VM3:4, from Erlitou, l. 54 cm. After Kaogu 1983.3, p. 204 fig. 10.6. f: VM3:5, from Erlitou, l. 48.1 cm. After Kaogu 1983.3, p. 204 fig. 10.5. g: III KM6:8, from Erlitou, l. 49.5 cm. After Beijing 1999c, p. 250 fig. 162.3. h: VII KM7:5, from Erlitou, l. 46–48 cm. After Deng Cong 1994, fig. D:12. i: unearthed in 1958 at Erligang, l. 66 cm. After Deng Cong 1994, fig. B:1. j: SSY17, from Shaanxi Shenmu Shimao, l. 49 cm. Shaanxi Provincial Museum. After Deng Cong 1994, fig. A:7. k: 2001CQJC:955, from Jinsha, l. 42.25 cm. After Beijing 2002c, p. 109. Figure 7.10 Jade forked blades at Sanxingdui. a: surface find in 1984 at Cangbaobao, l. 55.5 cm. After Bagley 2001, no. 52. b: 1927 Yueliangwan pit, no. (3.1)260, l. 39.3 cm. Sichuan University Museum. After Deng Cong 1998, vol. 3, color pl. 277. c: 1927 Yueliangwan pit, no. 313, l. 61 cm. Sichuan Provincial Museum. After Deng Cong 1994, fig. D:8. d: K1:170, l. 48.4 cm. After Beijing 1999a, p. 64 fig. 36.1. e: K1:23, l. 27 cm. After Beijing 1999a, p. 64 fig. 36.2. f: K1:01, l. 25.2 cm. After Beijing 1999a, p. 64 fig. 36.3. g: K1:02, l. 28.2 cm. After Beijing 1999a, p. 64 fig. 36.4. h: K1:275, l. 23.2 cm. After Beijing 1999a, p. 64 fig. 36.5. i: K2(3):167, l. 30.5 After Beijing 1999a, p. 362 fig. 199.2. j: K2(3):322-7, l. 33.6 cm. After Beijing 1999a, p. 362 fig. 198.4. k: K2(3):320, l. 67.8 cm, thickness 0.6 cm. After Beijing 1999a, p. 363 fig. 200.5. Figure 7.11 Jade ge blades from K1 and K2. a: K2(3):322-8, jade, l. 55.4 cm. After Beijing 1999a, p. 385 fig. 209.1. b: K1:97-8, jade, l. 47.1 cm. After Beijing 1999a, p. 94 fig. 48.2. c: K2(3):227-1, jade, l. 33.9 cm. After Beijing 1999a, p. 378 fig. 205.3. d: K1:141-1, 155-2, jade, l. 40 cm. After Beijing 1999a, p. 91 fig. 46.1. e: K2(3):314-6, jade, l. 27 cm. After Beijing 1999a, p. 379 fig. 206.5. f: K1:136, jade, l. 17.8 cm. After Beijing 1999a, p. 97 fig. 49.1. g: K1:246, jade, l. 21.1 cm. After Beijing 1999a, p. 97 fig. 49.2. Figure 7.12 Jade ge blades from other sites. a: ELT Phase III, IIIKM1:2, l. 30.2 cm. After Beijing 1999d, p. 250 fig. 162.5. b: ELT Phase III, bronze, IIIcai:60, l. 27.5 cm. After Beijing 1999d, p. 169 fig. 103. c: PLZM2:14, from Hubei Huangpi Panlongchen, l. 70 cm. After Beijing 2001e, p. 180 fig. 119.1. xxvii d: from a tomb at Henan Zhengzhou Baijiazhuang excavated in 1955, l. 37.9 cm. After Chen & Fang 1993, pl. 19. e: M5:444, from the tomb of Fu Hao, l. 27.4 cm. After Chen & Fang 1993, pl. 32. f. Jinsha, 2001CQJC:60, l. 50 cm. After Beijing 2002c, p. 144. g. Jinsha, 2001CQJC:478, l. 16.2 cm. After Beijing 2002c, p. 142. Figure 7.13 Jade hybrids of forked blade and ge. a: Sanxingdui, K1:151, l. 43.2 cm. After Beijing 1999a, p. 68 fig. 37.1. b: Sanxingdui, K1:75, l. 34.8 cm. After Beijing 1999a, p. 75 fig. 39.1. c: Sanxingdui, K1:235-5, l. 38.2 cm. After Beijing 1999a, p. 81 fig. 41.1. d: Jinsha, 2001CQJC:27, l. 34.6 cm. After Beijing 2006b, p. 85. Figure 7.14 a: Jade ge, K1:23-1, l. 29.5 cm. After Beijing 1999a, p. 94 fig. 48.5. b: Jade hybrid, K1:155-1, l. 39 cm. After Beijing 1999a, p. 97 fig. 49.3. Figure 7.15 Jade axe, K1:266, h. 16.5 cm. After Beijing 1999a, p. 98 fig. 51.3. Figure 7.16 Jade adze, K1:261, h. 18.3 cm. After Beijing 1999a, p. 118 fig. 60.2. Figure 7.17 Jade knives. a: Sanxingdui, K2(2):314-5, l. 27 cm. After Beijing 1999a, p. 390 fig. 210.1. b: Tomb of Fu Hao, l. 33.5 cm. After Beijing 1984a, p. 134 fig. 74 (middle). Figure 7.18 Jade chisels a: Sanxingdui, K1:150, h. 27.1 cm. After Beijing 1999a, p. 107 fig. 55.6. b: Jinsha, 2001CQJC:29, h. 22.61 cm. After Beijing 2002c, p. 138. Figure 7.19 Jade sword-shaped implements. a: Sanxingdui, K1:280, h. 28.2 cm. After Beijing 1999a, p. 97 fig. 50. b: Jinsha, 2001CQJC:583, h. 10.8 cm. After 2006b, p. 98. Figure 7.20 Boulders of jade found in 1974 by the Yazihe River at Suozitian. After Xiao et al. 2001, p. 24. Figure 7.21 Partly-worked and waste stone bi disks from Sanxingdui, exact locale of excavation unknown. After Tokyo 1998, p. 182. Figure 7.22 Gold sheath, K1:1, l. 142 cm, diameter 2.3 cm, weight 463 g (including carbonized wood). a: after Bagley 2001, p. 71 no. 1. b: after ibid., fig. 1.1. c: after Beijing 1999a, p. 61 fig. 34.1. xxviii Figure 7.23 Gold band from Jinsha, 2001CQJC:688, diameter 19.6–19.9 cm, weight 44 g. After Paris 2003, p. 163 fig. 3. Figure 7.24 Gold appliqué in the shape of a tiger, K1:11–1, h. 6.7 cm, l. 11.6 cm, weight 7.27 g. After Beijing 1999a, p. 61 fig. 34.3. Figure 7.25 Gold foils from K2. After Beijing 1999a, p. 354 fig. 195. Figure 8.1 Pottery gu and tripod he from Sanxingdui (left) and Erlitou (right). After Sun Hua 2000, p. 154 fig. 6.5. Figure 8.2 Bronze ge blades. a: Xinfan Shuiguanyin, l. 15–19.8 cm. Photography by author. b: Jinsha, 2001CQJC:646, l. 22 cm. After Chengdu 2005, p. 33. Figure 8.3 Pottery high-stem dou. a: Sanxingdui. After Beijing 1999a, p. 425 fig. 226:7. b: from Xiaojiawuji, H538:15, h. 39.6. After Beijing 1999e, p. 253 fig. 191:17. Figure 8.4 Pottery zun-shaped vessels. a: from Sanxingdui. After Beijing 1999a, p. 426 fig. 227:4. b: from the Dengjiawan site of the Shijiahe Culture, H48:12, h. 14 cm. After Beijing 2003, p. 170 fig. 141:6. Figure 8.5 Pottery hens. a: from Sanxingdui. After Chen Xiandan 1989a, pl. 3:4. b: from Dengjiawan, AT203(2a):1, existing h. 8.1 cm. After Beijing 2003, p. 210 fig. 169:3. Figure 10.1 Shi’erqiao and related sites in Chengdu. After Sun Hua 1996, p. 124 fig. 1. Figure 10.2 Location of the Jinsha site in Chengdu city. After Zhu et al. 2003, p. 250 fig. 2. Figure 10.3 Terraced mound at Yangzishan. Reconstruction drawing. After Kaogu xuebao 1957.4, p. 20 fig. 3. Figure 10.4 House elevated on stakes at Shi’erqiao. Reconstruction drawing. After Wenwu 1987.11, p. 8 fig. 11. Figure 10.5 Excavation loci in the Jinsha site. After Zhu et al. 2003, p. 253 fig. 3. Figure 10.6 Aerial view of exposed wall foundations of the large row buildings excavated at Locus Sanhe Huayuan. After Chengdu 2005, p. 2. xxix Figure 10.7 A pit of elephant tusks at Jinsha, K1. After Chengdu 2005, p. 3. Figure 10.8 a: Archaeologist Zhu Zhangyi at the locus with a large distribution of stone bi, the Jinsha site. Photograph by author, June 27, 2001. b: large stone bi unearthed in 1927 at Yueliangwan, the Sanxingdui site, outer d. 51.4 cm, inner d. 14 cm. Sichuan University Museum (2.3) 29. After Chengdu 2006, p. 13 pl. 8. Figure 10.9 Stone tigers. a: Sanxingdui, 99GSZYT111(17), excavated in 1999 from a ditch next to a wall at Yueliangwan. Photography by author. b: Jinsha, 2001CQJC:684, h. 21.5 cm, l. 28.8 cm. After Beijing 2002c, p. 185. Figure 10.10 Stone coiled snakes. a: Sanxingdui, locus unknown. Photograph by author, July 29, 2002. b: Jinsha, 2001CQJC:719, l. 41.8 cm. After Beijing 2002c, p. 189. Figure 10.11 Stone kneeling human figure from Jinsha, 2001CQJC:716, h. 21.72 cm. After Beijing 2002c, p. 177. Figure 10.12 Bronze standing human figure from Jinsha, 2001CQJC:17, h. 19.6 cm. After Beijing 2002c, p. 43. Figure 10.13 Gold masks. a: Sanxingdui, K1:282, h. 11.3 cm, weight 10.62 g. After Beijing 1999a, p. 61 fig. 34.2. b: Jinsha, 2001CQJC:465, h. 3.74 cm, weight 5 g. After Beijing 2002c, p. 22. Figure 10.14 a: Jinsha, gold circular ornament, 2001CQJC:477, d. 12.5 cm, weight 20 g. After Beijing 2002c, p. 30. b: Jinsha, bronze handled collared disk, 2001CQJC:588, d. 10.24 10.36 cm. After Beijing 2002c, p. 61. Figure 10.15 a: bronze fragment, part of a vessel (zun or lei)’s foot ring, from Jinsha, 2001CQJT8105(7):34, h. 12.3 cm. Style Va, 12th century BC. Photograph by author, July 1, 2002. b: bronze fragment, animal head for the swing handle of a vessel (probably you), from Jinsha, 2001CQJT8406(6):2, h. 4 cm, l. 4 cm. early 10th century BC. Photograph by author, July 1, 2002. xxx Chapter 1. Introduction A quarter of a century ago, scholars of ancient China had little to say about early civilization in Sichuan Basin of southwest China (Fig. 1.1). We know that the basin was conquered and subsequently colonized by the expanding state of Qin in 316 BC, whose army went on to unify China in the third century BC and thus brought Sichuan into the mainstream of Chinese civilization. Before that, written sources make it clear that in the Warring States period (5th–3rd century BC) the basin was occupied by two polities called Ba and Shu with peculiar cultural customs. But for earlier periods, there is hardly any written record from contemporary sources, and the texts composed much later recount myths and legends. Equally wanting were archaeological finds from the field: in the time from around the middle of the second millennium BC, when civilization first emerged in China in the Central Plain, to around the tenth century BC, no city and few bronzes were known in Sichuan, whereas elsewhere Bronze Age civilizations had been prospering. Finds were sporadic even for the following five or six centuries. This dearth gave rise to an impression that Sichuan was a cultural backwater that only belatedly became civilized. This impression was shattered in the summer of 1986 when one of the strangest discoveries in the history of Chinese archaeology was made at a place called Sanxingdui in the Chengdu Plain of western Sichuan (Fig. 1.2). A dozen jades accidentally exposed by brickyard workers led to the discovery of two pits containing an astonishing amount of wealth in the form of hundreds of bronzes alongside stone and jade implements, gold objects, and elephant tusks, altogether weighing more than one metric ton. The two dozen vessels among the bronzes help date the pits to the early Bronze Age, around 1200 BC, 1 corresponding to the Anyang period of the Central Plain. The most shocking revelation came with the bronze sculptures that account for the majority of the bronzes. Ranging from miniature to monumental in size, they include human-like figures, fantastic creatures, dragons, birds, and trees the likes of which had never been seen before in Sichuan or anywhere else (Fig. 1.3). This sculptural tradition and iconography contrast bluntly with other ancient traditions in China, which are primarily known for bronze vessels with little if any sculpture. Suddenly, we are confronted with evidence in staggering abundance for an elite material culture that rivals and differs from any other in Bronze Age China, including that of Anyang. The discovery of the two pits did not come in a complete void, however. Findings of ancient artifacts, almost exclusively in pottery, stone, and jade, actually began in 1927. Two decades after that, intermittent field archaeology resumed at the site. Then, when China started to engage in economic reform in the late 1970s, raging capital construction began, and this has given rise to extensive salvage archaeology across China, including at Sanxingdui. In 1984, traces of an ancient wall were recognized, leading eventually to the discovery of a large walled settlement. Two years later, the two pits were discovered. Today they are still the most dramatic find made at the site. Now, twenty or so years later, after a great many papers and books have been written about Sanxingdui (mostly in Chinese), we have gained considerable insight into the site and its culture. But the initial shock and puzzlement linger, as we still lack a comprehensive examination of the ingredients of the two pits and what they embodied. This gives the impetus and purpose for the present dissertation, which aims at sorting out the basics of the Sanxingdui site: its archaeological history, the components of its 2 archaeological record, the spatial and temporal dimensions of the site, and its elite material culture as embodied in the contents of the two pits as well as other related finds. The two pits are naturally the focus of the dissertation, as they, along with the walls, prove beyond doubt that Sanxingdui was home to a major civilization, and their contents are by far the richest and most enigmatic among all finds. My hope is for an intimate knowledge of the objects buried in the pits: their typology, origin, original appearance, fabrication and so forth. It is also important to set the objects in the context of ancient China so as to gain a measure of understanding of the cultural relationships that linked Sanxingdui with other regions beyond the Sichuan Basin. The basic approach in this research is therefore descriptive rather than interpretative. The research does not begin with or aim at an overarching theory or model. It is through detailed observation and analysis that meaningful interpretations are achieved, and a few will be offered in the dissertation: a re-definition of the archaeological Sanxingdui Culture; an analysis of the nature of the pits; a suggestion as to the function of the bronze images and the reason for their burial; and a discussion of the geographical factor as partial explanation for the uniqueness of Sanxingdui’s cultural development. Since there is neither any trace of writing at Sanxingdui nor any contemporary record about it from elsewhere, the objects from the Sanxingdui site constitute the primary body of evidence. Objects from other regions pertinent to the discussion will be brought in for comparative analysis. Resolutely avoided are textual sources written in later ages. Whatever kernels of historical memory might be retained in those texts, in the absence of any methodology for distinguishing fact from fiction, it is prudent to separate 3 them from artifactual evidence at the current stage of research, when the archaeological record at the site is far from being adequately analyzed and understood on its own right. Perhaps at some future time, it may be possible to combine a well-developed archaeological understanding with textual sources adequately researched independently on their own internal evidence. Nine chapters form the body of the dissertation. Following this introduction, Chapter 2 will review the history of fieldwork and the archaeological record at the Sanxingdui site and its environs, critique existing scholarship, describe Sanxingdui’s rise from a modest settlement to the center of a major civilization, and re-define the Sanxingdui Culture archaeologically. This chapter of critical review will set the stage for the chapters that follow. Chapter 3 commences concentrated research on the two pits excavated in 1986. The chapter will first lay out the research approaches applied, which advocates focus on internal evidence provided by the contents of the pits and comparison with related phenomenon from other cultures. The chapter will then describe the two pits, their contents and burial conditions, review existing interpretations, and identify the nature of the two pits. Chapters 4 to 6 form the core of the dissertation. Extensive firsthand examination of the bronzes from the two pits will be reported in detail. The examination was made possible by the unprecedented access generously granted to me by the archaeologists and institutions in whose custody the objects reside. Chapter 4 will address stray finds from Sanxingdui but will focus primarily on the bronze vessels from the two pits: their 4 typology, stylistic and cultural associations, and function. Chapter 5 will present a comprehensive discussion of the images and related implements, addressing such issues as iconography, reconstruction of the possible original appearance of the Sanxingdui images and their assemblage, sources of the Sanxingdui imagery, the function of the images, and the reason for their burial. Chapter 6 will explore the fabrication technology responsible for the making of the Sanxingdui images through detailed observation of casting features. Comparison with bronze casting in the Central Plain will make it possible to say something about the origin of the technology. This chapter will not only address technical issues, which are of import on their own right, but also cultural implications of the technology employed and the particular mentality evident in the Sanxingdui foundry practice. How the objects were made is as important to understanding the Sanxingdui imagery as how they were conceived stylistically and culturally. Chapter 7 will discuss lithic and gold artifacts at Sanxingdui. Ritual artifacts made of stone and jade have been found widely across the Sanxingdui site and span a long chronological range. The primary aim here is to sort out several basics: types, typological sources, and functions, as well as the history and material of lithic production in the local setting. Throughout the discussion, cultural contacts suggested by these objects will be presented individually as appropriate. The chapter will also examine artifacts made of gold, a material that was rarely used elsewhere in China during the Bronze Age but which constitutes a conspicuous category in the inventories of the two pits. Chapter 8 will bring together relevant material from earlier chapters to discuss cultural contacts between Sanxingdui and places outside the Sichuan Basin. Many traces 5 of contact with other cultures will have been identified and described at various points in the preceding chapters. Here they will be summarized and supplemented with further observations. The regions especially important to Sanxingdui’s external contact will be identified; this will set the Sanxingdui site in the context of ancient China. The particular nature and behavior of Sanxingdui’s participation in a large network of exchange will be commented on and its geographic situation will be proposed as part of the explanation. Chapter 9 will summarize the principal conclusions reached in the preceding chapters, bringing them together in a coherent picture. It can also serve as a quick index to the detailed contents of the dissertation. Chapter 10 will present a brief survey of the settlement at Chengdu about 40 km southwest of Sanxingdui, where another major discovery of elite artifacts was made in 2001 at the Jinsha site. The Chengdu settlement expands the horizon of the Sanxingdui civilization and gives clues to Sanxingdui’s legacy. The description of the Jinsha finds will focus on artifact comparisons and on the relationship between Sanxingdui and Chengdu, but it will also describe all the essential features known presently at Jinsha and other related sites in Chengdu, which begin to fill a frustrating blank in the archaeological record for the six or seven centuries following the time of the Sanxingdui pits. During the fifteen years over which my work on this dissertation has stretched, I have engaged in several related projects, the largest of which was the exhibition titled Treasures from a Lost Civilization: Ancient Chinese Art from Sichuan that toured the North America in 2001–2, for which I was the organizing curator. Parts of an earlier version of the dissertation were published in the catalogue, Ancient Sichuan: Treasures 6 from a Lost Civilization (Bagley 2001), that accompanied the exhibition (my contributions are cited here as Xu 2001). Several years later, I served as the guest editor of a book-length section on the art and archaeology of the Sichuan Basin for the fifth volume of the Journal of East Asian Archaeology. This collection includes a number of papers from different authors originally delivered at the international symposium held at the Seattle Art Museum in conjunction with the exhibition. Part of Chapter 2 was published in that volume (Xu 2003). I have also presented numerous lectures on the contents of the dissertation over the years, and published a few short articles in popular journals. This dissertation supersedes the previous publications and presentations, however, presenting my research in its currently definitive form. 7 Chapter 2. The Archaeological Record of the Sanxingdui Site: A Critical Review Situated in southwestern China and surrounded by mountains and high plateaux, the Sichuan Basin covers about 200,000 sq km. About seven percent of it is flat plains, the rest hills, the terrain ranging from 300 to 700 m above sea level (Fig. 1.1). The climate is mild subtropical with distinct seasons and monsoon rains. Running across the basin is a large network of rivers, primarily the Changjiang (“long river”), which enters the basin at its southwest corner and flows in an eastward direction to exit through the Three Gorges. Along the course, many tributaries feeds into the Changjiang (Fig. 1.2). In ancient times, the long river provided access to high plateaux in the west and in the south, and to regions downriver to the east. In the northern direction, tributaries running southward to the Changjiang enabled human traffic to spread throughout the basin and invited communication with regions to the north. Of the various parts of the basin, the Chengdu Plain at the western edge accounts for the largest tract of flat land, amounting to about 6000 sq km. The plain is a rich alluvial deposit formed by the Minjiang River in the west and the Tuojiang River in the east. It generally descends in elevation from northwest to southeast, from 700 to 400 m above sea level. Despite the menace of periodic flooding, the plain has been densely populated since Neolithic times, and it was the stage on which civilization first emerged in Sichuan. The Sanxingdui site is located in the Chengdu Plain, about 10 km west of the city of Guanghan, which is about 40 km northeast of Chengdu, capital of Sichuan province 8 (Figs. 1.2, 2.1).1 The name of the site—Sanxingdui—literally means “Three Star Mounds,” referring to the three rolling terraces nearby the village of Sanxing, the village obviously being named after the terraces as well.2 On present evidence, the Sanxingdui site covers an area of 10 to 17 sq km as determined by the distribution of similar artifacts.3 The site is roughly 5 to 6 km east to west and 2 to 3 km north to south, spread along the southern bank of the Yazihe River and both sides of the Mamuhe River (Fig. 2.2).4 The center of the site is a walled city that measures about 3.5 sq km (350 hectares) and that includes the whole or parts of five present-day villages, including the village of Sanxing at its south wall (Fig. 2.3).5 Sanxingdui was chosen to name the whole site because of the discovery there in 1986 of K1 and K2, the two pits filled with immense material wealth, which revealed the site to be the center of a previously unknown civilization. The name was officially anointed in 1988 when the Chinese State Council declared the site to be an Important National Cultural Property, the highest ranking in China’s historic preservation system.6 1Beijing 1999a, p. 9. The distance between the Sanxingdui site and Guanghan is variably given in publications as 8 or 9 or 10 km; the difference probably arises from different starting and ending points of the measurement. 2Only half of one terrace survives today, however. 3The site is estimated to cover an area of about 10 sq km in Nanfang minzu kaogu 1992a (p. 308), about 12 sq km in Beijing 1999a (p. 9), about 15 sq km in Chen et al. 1998 (p. 1), and about 17 sq km in Zhao Dianzeng 1996a (p. 232). The figures do not necessarily represent the size of the Sanxingdui settlement at any particular moment in its history since they are obtained, presumably, by estimating the distribution of all cultural remains that broadly fall into the entire time frame of the site. 4Beijing 1999a, p. 15. 5The other four villages are Zhenwu, Huilong, Rensheng, and Dayan (ibid., p. 9). 6Ibid., p. 13. 9 2.1 History of discoveries and fieldwork at the Sanxingdui site7 Archaeology at the Sanxingdui site has a history of about eighty years, but it is a spotty history with times of intensive fieldwork and dramatic discoveries, as well as long periods of inactivity. In this survey, I will describe how the site and its culture came to be recognized. For brevity, only events that define important moments in the site’s history are discussed in detail and pinpointed on Map 5; the degree of detail depends on their state of publication.8 2.1.1 Initial chance discovery at Yueliangwan in 1927 and subsequent excavation in 1934: recognition of the “Hanchow Culture” 7The site and the ancient culture it typifies have been called by various names. In the present dissertation, the names other than the Sanxingdui site and the Sanxingdui Culture will be put in quotation marks to avoid confusion. The definition of the Sanxingdui Culture will be discussed in Section 2.4. 8For chronicles of all the known finds and fieldwork at the Sanxingdui site see Chen Xiandan 2001, and Xiao et al. 2001. 10 The first known discovery at the Sanxingdui site took place in 1927.9 Sometime in that year, a local farmer with the family name of Yan, while deepening an old irrigation ditch at the edge of his residence, uncovered a group of stone and jade artifacts.10 The Yan residence was located on a long curving terrace which the locals had imaginatively viewed as resembling a crescent moon, hence the locale commonly referred to as Yueliangwan (“Moon Bend”). In the layout of the Sanxingdui city known today, Yueliangwan is situated near the north end of a central axis running roughly north and south across the city, with Sanxingdui located towards its south end (Fig. 2.3). In fact, Sanxingdui and Yueliangwan, both being raised above the surrounding fields, formed a long celebrated bit of local topography fondly called Sanxing ban yue (“Three Stars Accompanying the Moon”). 9The discovery at Yueliangwan was first reported in Dye 1931 and Graham 1934. Nowhere in these two reports is it mentioned that the discovery took place in 1927. The lack of explicit mention has since given rise to much confusion about the actual year of the discovery. As far as I knows, the only source from which the date may be deduced is the remark in Dye 1931 (p. 102) that the discovery was made about four years before the spring of 1931. In Lin Minjun 1942 (p. 93), the year 1931 is given for the find, but the author had apparently mistaken the year 1931 when the discovery came to the attention of Rev. Donnithorne, a missionary stationed at Guanghan who initiated the effort to preserve the finds, to be the same year when the discovery was made. As the first Chinese publication on the discovery, Lin Minjun 1942 must have exerted significant influence on later works; the date given there is followed, for example, in Zheng Dekun 1946 (p. 31) and in Beijing 1999a (p. 9), which is the official excavation report of K1 and K2. Another date, the year 1929, is also commonly given in various other publications, the earliest known to me being Feng & Tong 1979, in which the date appears to be based on a misreading of Graham 1934 and Lin Mingjun 1942. I followed this most commonly used date in Bagley 2001 (p. 24). In this dissertation, however, it is decided that the year 1927 may be the most reliable because it is implied in the report by Daniel S. Dye, who first committed the discovery to writing. For a detailed discussion of this issue see Xu Jie 2006. By the same logic, the sequence of events given hereafter follows Dye 1931 and Graham 1934, even though the account in Lin Minjun 1942 sometimes affords remarkably richer details. 10No mineralogical determination was or could be performed on the so-called jades when they were published. Throughout the present dissertation, the term jade/jades should be understood as a conventional label for fine stones not necessarily limited to nephrite. 11 The objects started to disperse soon after the discovery. In the spring of 1931, after a lapse of four years, a number of objects were eventually collected from the Yan family and given to the West China Union University Museum at Chengdu (now the Sichuan University Museum); they became the first group of objects from the site that came into the public record. The publication of Daniel S. Dye, a professor of geology at West China Union University who participated in preserving and investigating the finds, gives an idea of what kinds of object were uncovered.11 They included large disks (bi) of graduated size, some being 70 cm in diameter, small ring(s) with collared opening, forked blades, axes, chisels, cylindrical tubes encased in square prisms (cong), and beads (Fig. 2.4). Apparently the large bi were made of coarse stone and crudely worked, while other items were of harder stones and showed fine workmanship.12 The objects had been buried in a rectangular pit at the bottom of the irrigation ditch. In March 1934, David C. Graham, then curator of the West China Union University Museum, led an excavation at the ditch to find more objects and to secure additional data. He hoped that knowledge of stratification and of the exact locations of objects would make possible more accurate dating and the determination of the culture to which the objects belonged. Besides a few potsherds, a large number of stone and jade objects were unearthed from the original pit.13 Nearby, a cultural stratum was found with 11Dye 1931. 12Dye 1931, pp. 102–4. The size of the largest stone bi is given in Lin Mingjun 1942 (p. 96), and in Feng & Tong 1979 (p. 34). 13Graham 1934 (p. 118) states, “We secured nearly a hundred fragments of stone rings, and stone and jade knives, fifteen turquoise and green stone or jade beads, and over eighty small, flake-like pieces of jade, square or oblong in shape, besides a few pieces of broken pottery.” Lin Minjun, the chief assistant to Graham in the excavation, had a different account, however, stating that the objects excavated from the pit comprised only 12 hundreds of potsherds and numerous fragments of stone and jade. The similarities in artifact type made it clear that the cultural stratum and the pit belong to the same time and culture.14 In all, the excavation retrieved more than six hundred pieces of jade, stone and pottery, mostly potsherds.15 Besides being the first controlled digging at the Sanxingdui site, the 1934 excavation had three important achievements. First, it correlated stone and jade types with pottery types, which would later prove to be the essential artifacts that define the archaeological culture of the site. Second, it clarified the physical context from which the 1927 finds came, even though no major new artifact types were revealed. Third, and perhaps most importantly, it established the antiquity of human settlement of the area and an archaeological culture particular to it, named the “Hanchow Culture” by Graham. Through artifact analysis, Graham dated the “Hanchow Culture” to the beginning of the Zhou dynasty or about 1100 BC,16 a date remarkably close to what would be confirmed half a century later. After the 1934 excavation, fieldwork at the Sanxingdui site apparently came to a halt. In the known record, the next time when the area was investigated by archaeologists was 1951. two fragments of forked blades and a few fragments of stone disks (Lin Minjun 1942, p. 95). 14Graham 1934, pp. 118–9. Graham did not report explicitly on the relationship of the pit and the stratum, but evidently regarded them as belonging to the same level. 15Lin Minjun 1942 (p. 95), and Zheng Dekun 1946 (p. 33). According to the description in Graham 1934, it may be estimated that roughly 56 sq m were excavated (not including several test pits). 16Graham 1934, p. 129. The dating was endorsed by Guo Moruo, a leading epigrapher and scholar of ancient China (see the letter by Guo published in English translation in Graham 1934, p. 130.) 13 2.1.2 Fieldwork from 1951 to 1986: recognition of the Sanxingdui Culture Although there was a hiatus of nearly two decades after 1934, the initial discovery and excavation at Yueliangwan had a definite impact on later fieldwork. Since 1951, archaeologists have repeatedly returned to the area in their search of ancient remains. The decade from 1951 to 1963 may be reasonably called an “era of Yueliangwan” in the history of Sanxingdui archaeology. One category of artifact constantly encountered in the Yueliangwan area was stone and jade implements, including finished, semi-finished objects, as well as raw and waste materials. The cultural deposits on the side near the Mamuhe River were found to be as deep as 2 m. By 1961, several field surveys had led archaeologists to pronounce that the Yueliangwan area was the center of a site spreading over an area about 3 km long between the Yazihe and Mamuhe rivers, the cultural homogeneity among various locales being attested by similar potsherds. They named the site the “Zhongxing site” after the local commune. This is the extent of the Sanxingdui site known then.17 From September through December 1963, excavations were conducted in the Yueliangwan area under the directorship of Feng Hanji, a leading archaeologist of the time.18 They were the first excavations at the Sanxingdui site in thirty years. Covering a total of 150 sq m, the excavations afforded fresh understandings. The archaeologists identified two cultural strata and hence two distinguishable developmental stages of the ancient culture. The features excavated included plots of artificially arranged pebbles, dirt surfaces burned hard and reddish, six tombs, and parts of building foundations, which 17Wenwu cankao ziliao 1954.3, Wang & Jiang 1958, and Wenwu 1961.11. 18The excavation was first briefly described in Song Zhimin 1983 (pp. 72–3), and Song Zhimin 1991 (pp. 209–13). 14 began to reveal the complexity of cultural features in the area. The artifacts included about 30,000 potsherds, and a few jade and bone implements. Possible traces of a bronze foundry were also discovered, comprising some bronze fragments, malachite lumps, copper slag and fragments of coarse pottery utensils, perhaps crucibles.19 It must be acknowledged that, to some extent, the prominence of Yueliangwan was due to chance and choice; chance being that it was the locale where the first major discovery took place, and choice being that the later archaeologists, guided by the finds in the 1930s, chose to concentrate their efforts in that area. This choice was surely reasonable and evidently justified by the field results. On the other hand, by preventing the archaeologists from employing their limited resources and manpower in other areas, it shaped the archaeological record in a particular way. We must be acutely aware of such factors when evaluating the archaeology of the Sanxingdui site or indeed any other. Sanxingdui came to notice in 1956, when a field survey revealed dense deposits of potsherds in the area, prompting the archaeologists to call it the “Sanxingdui site” for the first time, as distinct from the cultural remains at Yueliangwan.20 No fieldwork was performed in the area in the following years, however, probably due to the efforts at Yueliangwan. In 1975, when a brickyard was set up at Sanxingdui and began to dig clay on the terraces, a large number of potsherds were exposed, and local archaeologists determined that the terraces contained rich cultural deposits. But nothing could be done to preserve them. By the spring of 1980, one and half terraces had been dug away.21 19Ma Jixian 1992. 20Wang & Jiang 1958, pp. 30–1. 21Xiao et al. 2001, p. 26. 15 After the Cultural Revolution (1966–76), the pace of fieldwork accelerated at the Sanxingdui site as it did elsewhere, as a result of rapid economic development throughout the country, but planned survey and excavation gave way chiefly to salvage operations. Due to the urgency to obtain archaeological information about the Sanxingdui terraces before they were completely dug away by the brickyard, the archaeologists started to concentrate in that area. This prompted a period of intensive excavations in the area and thus unwittingly began what may be called the “era of Sanxingdui” in the history of Sanxingdui archaeology. In 1980–1981, excavations were carried out at a locale later labeled as Sanxingdui Locus III, led by Wang Youpeng, Chen De’an and Chen Xiandan, archaeologists from the Department of Field Archaeology at the Sichuan Provincial Museum; this would later become the Sichuan Provincial Institute of Archaeology. The team first conducted a test excavation over an area of 125 sq m (five 5 x 5 m test squares) in May 1980, and then from November 1980 through May 1981 excavated an area of 1,100 sq m (forty four 5 x 5 m test squares). The total area excavated in 1980–81 exceeds 1,600 sq m and the excavations yielded a great number of potsherds and an array of features.22 Then, from April through December 1982, Chen De’an and Chen Xiandan led the team for another season of excavation nearby, at what was to be labeled Sanxingdui Locus I, exposing an area of 100 sq m (Fig. 2.5).23 22The size of 1,600 sq m includes also more than ten test squares over an area of about 400 sq m, which yielded no cultural remains. 23Kaogu xuebao 1987.2. In Chen Xiandan 1989a (pp. 213, 215), it is mentioned that the area excavated in 1980–81 was later labeled as Sanxingdui Locus III, and the area excavated in 1982 was later labeled as Sanxingdui Locus I; the area covered by the 1982 excavations is mentioned to be 150 sq m, however. 16 The 1980–81 excavation at Sanxingdui Locus III was the most instrumental in defining the archaeological culture at the Sanxingdui site. The archaeologists recognized a series of cultural strata and, based on more than 100,000 potsherds, a core group of pottery types associated with them. Through stratigraphy and pottery typology, they established a periodization documenting three phases of cultural development. To that, the 1982 excavation added one more phase. Typological comparison and radiocarbon dating enabled the archaeologists to estimate the duration of the culture as extending from the late Neolithic period or about 2500 BC, to about 1000 BC, i.e., the time corresponding to the end of the Shang period or the beginning of the Western Zhou period in the Central Plain. And they formally proposed to name the culture the “Sanxingdui Culture”, which has since been commonly accepted.24 A chronological control of the site was thus finally established. In another respect, the large number of artifacts, mostly sherds of pottery vessels, as well as numerous features such as building foundations, a pottery kiln, and tombs, made it possible to describe the content of the culture in a more substantial way than before. The pottery types also served as a primary link to connect Sanxingdui with locales previously excavated or surveyed and with sites in other parts of the Chengdu Plain, thus enabling the archaeologists to begin mapping the size of the site as well as the spatial distribution of the culture. From March to May 1984, the same team excavated at Xiquankan, about 600 m north of Sanxingdui, covering an area of 175 sq m (seven 5 x 5 m test squares). The excavation yielded a large amount of potsherds whose types conform to the core group recognized in the 1980–81 excavations. From October 1984 to January 1985, another 24Kaogu xuebao 1987.2, p. 249. 17 excavation was carried out at Sanxingdui Locus III over an area of 125 sq m. In the spring of 1986 the team, again trying to keep ahead of the brickyard but also in order to provide practical training for students from the Sichuan University, launched the largest excavation yet at the Sanxingdui site, digging at Sanxingdui Loci I and III as well as a new locale later labeled as Sanxingdui Locus II (Fig. 2.5).25 The total space uncovered was 1,325 sq m. The excavation revealed the thickest cultural deposits (about 2.5 m) with the most numerous strata and the largest quantity of features and artifacts, including many small pottery animals.26 The excavation at once confirmed and enriched the results of the earlier excavations at Sanxingdui. The temporal span was once again determined to be the late Neolithic through Western Zhou with a duration of 2,000 years. 2.1.3 Discovery in 1986: K1 and K2 1986 proved to be the most rewarding year yet for the Sanxingdui archaeologists. The “era of Sanxingdui” reached a dramatic height in the summer of 1986 with a discovery that is arguably the strangest in Bronze Age archaeology of China. On a hot summer’s day, July 18, two months after archaeologists and archaeology students had finished their largest excavation, workers from the local brickyard, while digging clay at Sanxingdui Locus II, came upon a dozen jades and reported their find to the archaeologists. The archaeologists, led by Chen De’an and Chen Xiandan, began work on the same day and soon discovered something totally unexpected—a pit containing objects of types never seen before, such as a dozen lifesized heads with sharply defined facial features and bulging eyes (Fig. 2.6). When the 25Chen Xiandan 1989a, pp. 215–7. 26Chen Xiandan 1988, p. 10. 18 excavation was concluded on August 14, they had unearthed from the pit, later designated as K1, more than 400 artifacts of bronze, gold, stone, jade, amber, and pottery, along with thirteen elephant tusks, dozens of cowry shells, and three cubic meters of burnt animal bones mixed with wood and bamboo ash. Then, on the very day they concluded, some brickyard workers, having moved to dig clay about 30 m away, exposed a bronze head, and the second pit, K2, was discovered (Fig. 2.7).27 Excavation of K2 began on August 20 and lasted until September 17. Its content proved to be still richer. It contained 67 elephant tusks, 4600 cowry shells, and hundreds of artifacts (the final excavation report, Beijing 1999a—which numbers fragments individually— reaches a total of 1300), including many jade and stone objects, and an amazing pile of bronze sculptures, some of them very large and very strange. The discovery of K1 and K2 completely changed the understanding of the Sanxingdui site. Whereas the excavations and field surveys up to then had succeeded in defining the Sanxingdui Culture in basic archaeological terms and in establishing the site as a settlement of substantial size, perhaps surrounded by walls (see discussion below), information about the material culture of the local elite was not very rich. The 1927 discovery and the 1934 excavation at Yueliangwan did bring to light important specimens of jade implements, and later fieldwork yielded some more, but they were not artifacts impressive enough to suggest to the archaeologists that they were unearthing the material remains of a civilized society. The two pits, by preserving in staggering abundance the material culture of an elite previously unknown and unsuspected, proved without doubt 27Throughout this dissertation, factual information about K1 and K2 is taken from the final excavation report, Beijing 1999a, unless otherwise noted. The final report was preceded by preliminary reports in Wenwu 1987.10 (K1) and Wenwu 1989.5 (K2). 19 that the site was home to a civilization startlingly different from any other in Bronze Age China. 2.1.4 Fieldwork from 1984 to present: recognition of the Sanxingdui city As a result of the 1980s excavations, the size of the known site was substantially enlarged, now including the Yueliangwan area in the north and the Sanxingdui area in the south. In 1984, an excavation at Xiquankan, a short distance north of Yueliangwan by the Yazihe river, yielded tantalizing clues that the terrace there might have been manmade. A subsequent test digging in September proved that it was indeed a man-made structure, probably a wall.28 An “era of the walls” began in 1985 when the search for them became earnest. During that year field survey and test digging indicated that many other terraces in the site were man-made walls dating from generally the same time as other cultural remains, thus suggesting that the Sanxingdui site was a walled settlement. Those walls included the Sanxingdui terraces, terraces to the west of Sanxingdui that run roughly north-south between the Yazihe and Mamuhe rivers, and terraces to the east of Sanxingdui running roughly parallel to the west wall (Figs. 2.3, 2.8).29 Since then and throughout the 1990s, several excavations were conducted under the directorship of Chen De’an to map out the walls and to understand their structure, method of construction, and date. In 1988, a formal excavation on the last half of the surviving terrace at Sanxingdui finally proved that it was indeed a wall. The same confirmation was obtained through excavations in the following years on the east and west walls. Between November 1994 and January 1995, another wall south of Sanxingdui 28Xiao et al., 2001, p. 31. 29Ibid., Beijing 1999a (p. 12), and Taibei 1999 (p. 198). 20 was detected and then confirmed by test digging. This overturned the previous understanding that the Sanxingdui terraces were the remains of the city’s south wall, and it enlarged significantly the walled area.30 In January 1999, an excavation at Yueliangwan proved that the Yueliangwan terrace was a man-made wall as well. On present information the city measured 1.6–2 km from east to west and about 2 km from north to south, enclosing an area of 3.5 sq km (350 hectares) in a roughly trapezoidal plan with the south wall forming the base of the trapezoid. It is not clear whether there was once a north wall; the site is bounded on the north by the Yazihe river (Fig. 2.3). The city is comparable in size to the largest known city of its time, Zhengzhou in the Central Plain, principal city of the Erligang Culture (c. 1500–1300 BC).31 From 1987 through 1990, several field surveys organized to map out the extent of the Sanxingdui site revealed that the city was at the center of a site that covered more than 10 sq km (Fig. 2.2).32 This is the extent of the Sanxingdui site known to the present. In addition to the fieldwork devoted to the walls, a cemetery at the Rensheng village was discovered in November 1997, again by local brickyard workers. The subsequent excavation in 1998 exposed 29 small tombs (Fig. 2.9), two of which had been completely destroyed by the brickyard workers. The tombs were neatly laid out, with their corners pointing to the cardinal directions. The ten furnished tombs yielded altogether five pottery vessels (including one lid), 22 small jade objects, 37 stone beads, two stone balls, and one large elephant tusk.33 This is the first and so far the only time 30Taibei 1999, p. 199. 31Chen et al. 1998, p. 1. 32Nanfang minzu kaogu 1992a, pp. 295 and 308. See also note 3. 33Kaogu 2004.10. Besides three jade objects illustrated in this report, two other jade objects from the tombs are illustrated in Chen De’an 2000 (p. 123 figs. 146–8). 21 that a cemetery of human burials was found at the Sanxingdui site.34 No complete human skeleton was found, however, all having apparently been rammed intentionally as were the walls and floors of the tombs, a peculiar phenomenon awaiting further clues for understanding. 2.1.5 The amount of space excavated and publication of the fieldwork During the period of eighty years from 1927 till now, the major fieldwork conducted at the Sanxingdui site may be summarized as follows: (1) The 1934 excavation at Yueliangwan, about 56 sq m (2) The 1963 excavations at Yueliangwan, 150 sq m (3) The 1980–81 excavations at Sanxingdui Locus III, over 1,600 sq m (4) The 1982 excavation at Sanxingdui Locus I, 100 sq m (5) The 1984 excavation at Xiquankan, 175 sq m (6) The 1984–85 excavation at Sanxingdui Locus III, 125 sq m (7) The 1986 excavations at Sanxingdui Loci I, II, III, 1,325 sq m (8) The 1986 excavation of K1 and K2 at Sanxingdui Locus II (9) The 1988–89 excavation of the Sanxingdui terrace (10) The 1990 excavation of the east wall (11) The 1991–92 excavation of the west wall (12) The 1994–95 excavation of the south wall (13) The 1997–98 excavation of the Rensheng cemetery, 934 sq m (14) The 1999 excavation of the Yueliangwan terrace 34Three small tombs without any burial furnishings had earlier been excavated at Yueliangwan in 1963 (see Ma Jixian 1992, p. 313.) 22 The amount of space covered by these excavations is limited: a total of somewhat more than 5,000 sq m have been excavated in the whole site by 1994.35 Publication of these excavations and other field surveys is less than adequate. Only the 1980–81 excavation and the 1986 excavation of K1 and K2 have been published in formal reports (Kaogu xuebao 1987.2; Beijing 1999a). The 1934 excavation at Yueliangwan and the 1997–98 excavation of the Rensheng cemetery were published in preliminary reports (Graham 1934; Kaogu 2004.10), the former in much detail. The report of the 1963 excavation had to wait thirty years for publication, the delay no doubt caused by the Cultural Revolution and consequent dispersal of excavation records. When published, the report appeared as a recollection of personal notes taken by one of the archaeologists (Ma Jixian 1992), and is therefore far from being exhaustive. Particularly lamentable is the absence of a report of the 1986 excavation at Sanxingdui Loci I, II and III, the largest in scale and the richest in results in every way. Only a brief account of that excavation has been published, included in a summary of fieldwork between 1980 and 1986 (Chen Xiandan 1989a). Besides these reports, there are a few reports of field surveys, and useful bits of information have occasionally been disclosed in short articles or news releases. This woeful inadequacy, coupled with the limited scope of the excavations, prevents one from giving a detailed account of the settlement and its various components beyond a 35This is a rough estimate calculated from available statistics up to 1994. According to the description in Graham 1934, it may be estimated that roughly about 56 sq m was excavated by him in 1934 (not including several test-pits). According to Ma Jixian 1992 (p. 310), a total of 150 sq m was excavated in 1963. According to Beijing 1999a (p. 9), a total of 5,000 sq m was excavated between 1980 and 1994; according to the statistics given in Taipei 1999 (p. 23), the 5,000 sq m seems to include 1,000 sq m of walls. The number could be significantly larger when the amount of space excavated since 1994 is added, but judging from my personal experience from numerous visits to the site, the total space will still be quite small compared to the size of the city. 23 general overview. Such an overview, which will be presented in Sections 2.2 and 2.3 of this chapter, nevertheless provides a meaningful backdrop against which to understand the art and culture of the social elite of the Sanxingdui site as embodied in the materials of K1 and K2. 2.1.6 Fieldwork at related sites Beyond the Sanxingdui site, several other sites in the Chengdu Plain have been identified as wholly or partly contemporaneous. The earliest of these are walled settlements of the Neolithic Baodun Culture that flourished in the third millennium BC, particularly the second half.36 Named after the first major site where it was observed, the Baodun Culture is characterized by a set of pottery vessels whose types give only limited hint of contact with Neolithic cultures outside the Sichuan Basin.37 Though Baodun sites are scattered over much of the basin, the culture seems to have been centered in the 36The Baodun Culture is alternatively known as the Baoduncun culture. For surveys and studies of the Baodun Culture see Wang et al. 1997, Jiang et al. 1997, Feng Guanghong 1999, Wang et al. 1999 (pp. 3–4), Wang & Sun 1999, Sun Hua 2000 (pp. 309–14), Tokyo 2000 (pp. 99–121), Song Zhimin 2000, Wang & Jiang 2000, Lin Xiang 2001, Jiang et al. 2001a, Jiang et al. 2002 (pp. 2–7), Jiang & Li 2002 (pp. 58–85), and Wang (2003). 37Tantalizing clues of cultural connection in pottery have recently been revealed in the fieldwork at the fourth-millennium BC Yingpanshan site in Mao Xian county in the upper valley of the Minjiang River in the northwestern edge of the Sichuan Basin. Some traits in fabrication technique, vessel type and decorative style of the Baodun pottery find parallels at Yingpanshan, whose primary cultural affiliation was with the Majiayao Culture further north. However, the large discrepancy in time between Baodun and Yingpanshan makes tentative any claim of cultural connection. For discussions on this subject, see Jiang & Chen 2001 (p. 31), Beijing 2002a (p. 76), and Jiang Zhanghua 2005 (p. 16). Closer in time but farther in distance, similarities with Baodun pottery have been observed in the late Neolithic site at Baoshan in the Hanzhong Basin lying immediately north of the Sichuan Basin across the Qinling mountains (Beijing 2002b, pp. 175–6). In another direction, a few pottery types of the Baodun Culture are similar to those of the third-millennium BC Shijiahe Culture of southern Hubei in the middle Changjiang region (Wang 2003, p. 118). 24 Chengdu Plain, where six walled settlements have been discovered since 1995: the Baodun site in Xinjin county, the Yufu site in Wenjiang county, the Gucheng site in Pi Xian county, the Mangcheng site in Dujiangyan municipality, the Shuanghe site in Chongzhou county, and the Zizhu site near the Shuanghe site (Fig. 2.10). The settlements are spaced from southwest to northwest of Chengdu at intervals of 20 to 50 km. They vary in size, the two smallest each covering about 0.1 sq km (10 hectares, Mangcheng and Shuanghe), another one more than 0.2 sq km (20 hectares, Zizhu), the next two a little more than 0.3 sq km (30 hectares, Gucheng and Yufu), and the largest 0.6 sq km (60 hectares, Baodun). Four of the settlements are rectangular (Baodun, Mangcheng, Gucheng, and Shuanghe), one square (Zizhu), and the other irregular in shape (Yufu). Three of them have double walls (Mangcheng, Shuanghe, and Zizhu), one enclosed by another. Walled settlements as large as these testify to an ability to mobilize substantial labor forces; they suggest a stratified society with a developed agricultural base.38 Several features recur at each of the six settlements. All were built on high natural terraces, and their walls were set along the edges of the terraces, making the ground 38Fieldwork at the Baodun sites has been carried out chiefly by the Chengdu Municipal Institute of Archaeology, which is exemplary for its efficiency in publishing excavations results. The excavation of the Baodun site in 1996 has been formally published in Tokyo 2000, which was preceded by two preliminary reports: Kaogu 1997.1, and Kaogu 1998.1. The latest study of the Baodun site is Li Mingbin 2001a. Four other walled settlements have been published in preliminary excavation reports, summaries, or studies: the Yufu site in Wenwu 1998.12, Jiang & Li 1998, Li Mingbin 2001b, and Li & Chen 2001; the Gucheng site in Wenwu 1999.1, Wenwu 2001.3, and Beijing 2001a; the Mangcheng site in Fan Tuoyu 1992, Kaogu 1999.7, Chengdu wenwu 2001.1 (pp. 62–3), Beijing 2001b, and Beijing 2001c; and the Shuanghe site in Tokyo 2000 (p. 102), Wang & Jiang 2000 (pp. 154–7), Jiang et al. 2001a (pp. 703–4), and Kaogu 2002.11. The Zizhu site, discovered in the fall of 1997, has not been excavated but is briefly described in Tokyo 2000 (p. 102), Wang & Jiang 2000 (p. 157), Chengdu wenwu 2001.1 (p. 63), and Jiang et al. 2001a (p. 704). 25 surface higher inside than outside.39 At Baodun, Yufu and Gucheng, archaeologists learned the technique of wall construction by dissecting the cross section. The walls were built of compacted earth. First, a central section was built with layer after layer of earth piled on horizontally and tamped or beaten with timbers. When the section reached a certain height, slopes on both sides were built with slantingly piled and beaten layers. This produces a very thick wall with a flat central part and a gentle slope on either side; the cross section is trapezoidal, with the wall much thicker at the base than at the top.40 Pebbles and sand were used in varying degree for construction, sometimes for reinforcing the walls and often for paving the wall surfaces. The Baodun walls have a further peculiarity that no city gate has so far been found. Since evidence of flooding is often found, it is possible that the walls were meant to function more as dikes for flood control rather than as defensive ramparts. The siting of settlements on high terraces was perhaps itself a defense against floods. The settlements are located along river banks, always in transitional zones where the terrain changes from hill to plain; their elevations range from 500 to 700 m. These locations apparently combined easy access to water with greater 39Jiang & Li 1998 (pp. 26–7), Sun Hua 2000 (pp. 302–23), Tokyo 2000 (pp. 102–3), and Wang & Jiang 2000 (p. 159). 40Tokyo 2000 (p. 103), and Wang & Jiang 2000 (p. 159). In the more sophisticated rammed-earth technique commonly used in the Central Plain, earth was poured in thin layers between wooden forms and pounded to great hardness. To a large degree, I suspect that the difference arose from the different tempers of earth: the earth in the south is much more sticky and tends to cohere, while the earth in the north is fine and powdery and therefore had to be rammed hard to hold the shape. Walls built in the same way as at the Baodun sites are characteristic of the Shijiahe Culture (Wang et al. 1997.2, p. 24; and Jiang & Li 2002, pp. 75–6). Urban construction in that region has a history that can be traced back as far as the beginning of the fourth millennium; the largest Neolithic walled settlement so far known in China, covering about 1 sq km (100 hectares), was built at the Shijiahe site during the first half of the third millennium. It is quite possible that Baodun learned wall construction from Shijiahe. For the Shijiahe city and wall construction see Nanfang minzu kaogu 1992b, pp. 277–80. 26 security from flooding than areas further downstream in the center of the Chengdu Plain.41 Within the settlements, building foundations have been excavated. The largest, a rectangle 50 by 11 m, was found at the Gucheng site (Fig. 2.11). It might have been a communal dwelling or a palace. The construction technique is unknown outside Sichuan: ditches were dug in the ground, wooden pillars were erected with their bases in the ditches and surrounded by pebbles, and dirt was then poured into the ditches to fill up the space; straw or bamboo mixed with mud was packed between the wooden members to form the walls.42 Smaller houses in the settlements were mostly built in a similar way, but a few were built without digging ditches in the ground but only holes for the wooden pillars.43 Besides these walled settlements, smaller sites of the Baodun Culture without walls have also been found in several locations, including Shijiefang, Huachengcun, and Jinsha, all in Chengdu, Qingjiangcun in Pi Xian,44 and as we shall see later, the Sanxingdui site. It appears that at the time of the Baodun Culture, the Chengdu Plain had already hosted a stratified society with settlements of at least two distinct levels. Moreover, certain pottery types indicate cultural connections between the Baodun sites and a number of other Neolithic sites previously known inside the Sichuan Basin, including those in the north of the Chengdu Plain and in the eastern part of the basin 41Wang Cunwu 1998, Jiang & Li 1998 (pp. 26–7), Liu Xingshi 1998, and Zhu Zhangyi 2001 (p. 27). Of these studies, Liu Xingshi 1998 has produced the most forceful argument for the nature of the walls as dikes for flood control. 42Wenwu 2001, 3, pp. 58–9. 43Tokyo 2000 (pp. 104–5), and Wang & Jiang 2000 (p. 159). 44For the Shijiefang site see Zhu Zhangyi 2001. For the Huachengcun site see Liu & Rong 2001. For the Jinsha site see Huaxia kaogu 2002.3 and Beijing 2004. For the Qingjiangcun site see Beijing 2001d and Wenwu 2003.1. 27 reaching as far as the Three Gorges, and a few sites in other parts of Sichuan (Fig. 2.10).45 Bronze Age sites that co-existed partly with Sanxingdui include Shaxi in Ya’an county at the southwest edge of the Sichuan Basin, and in the Chengdu Plain, Guilinxiang and Shuiguanyin in Xinfan county, Qingjiangcun in Pi Xian county, about a dozen small sites in Guanghan and Shifang counties, and Yangzishan, Shi’erqiao, Hetaocun, and Jinsha in Chengdu.46 The Sanxingdui site also had far-reaching connections with regions beyond the Sichuan Basin, which will be brought into discussion at appropriate points later in this dissertation and summarized in Chapter 8. 2.2 Stratigraphy, pottery typology, periodization, and absolute date This section will discuss and review in detail the temporal dimensions of the Sanxingdui site. Archaeologists in China commonly rely on two basic tools to obtain relative dating of settlements of the Neolithic period and Bronze Age: stratigraphy and pottery typology/seriation. In the simplest terms, stratigraphy proceeds by interpreting the layering of physical debris that cultural and natural forces constantly deposit. As debris of a later time is usually deposited atop an earlier one, the layering can therefore serve as an index of temporal order, though various factors sometimes cause stratigraphical reversal 45For discussions of these connections see Wang & Sun 1999 (pp. 69–73), Tokyo 2000 (pp. 116–20) (same as Jiang et al. 2001a, pp. 715–9), and Wang & Jiang 2000 (p. 162). 46For the Shaxi site see Nanfang minzu kaogu 1990, and Li Minbin 1999a. For the Guilinxiang site see Wenwu 1997.3. For the Shuiguanyin site see Kaogu 1959.8. For the Qingjiangcun site see Beijing 2001d, and Wenwu 2003.1. For the Hetaocun site see Wenwu 2003.4. For the sites in the Guanghan and Shifang counties see Nanfang minzu kaogu 1992a. For the Yangzishan, Shi’erqiao, Jinsha and related sites in the city of Chengdu, see Chapter 10. For general survey of those sites in the Chengdu Plain except the Hetaocun and Jinsha sites see Sun Hua 1996, Sun Hua 2000 (pp. 102–5, 146–8), and Jiang et al. 2002 (pp. 11–6). 28 and thus make the work of obtaining the correct order of layers or strata less easy than it sounds. Pottery typology/seriation is possible because of the ubiquitous presence of pottery vessels in habitation sites. Pots break, but the resultant potsherds are virtually indestructible and can often be restored. Archaeologists can, through pottery types, delimit a site’s temporal duration and spatial extent. As time goes on, shape, decoration, clay color and clay type of pottery vessels all tend to change gradually. By gauging the degree of change, archaeologists can estimate the developmental stages or phases of a given site. Sometimes two or more strata may need to be combined to form one phase; conversely, one stratum–phase may need to be subdivided into several subphases. If change happens abruptly, it often, though not always, suggests to archaeologists the presence of outsiders (traders or immigrants or conquerors) who introduce new types or new customs. In short, stratigraphy and pottery typology/seriation work together to define the basic dimensions in time and space of a site. As for absolute dates, carbon-14 dating calibrated by dendrochronology is the basic tool by which to gain absolute chronological control. But sometimes, as in the case of the Sanxingdui site, dates can also be obtained by noting the presence at the site of objects whose age is known from other associations. 2.2.1 Stratigraphy, pottery typology, and periodization For ease of discussion, the cultural strata at the Sanxingdui site revealed in various excavations are summarized as follows, (1) The 1934 excavation at Yueliangwan: one cultural stratum47 (2) The 1963 excavations at Yueliangwan: two cultural strata48 47Graham 1934, pp. 118–9. 29 (3) The 1980–81 excavations at Sanxingdui Locus III: five cultural strata, plus two strata largely free of cultural content interposed between cultural strata49 (4) The 1982 excavation at Sanxingdui Locus I: two cultural strata50 (5) The 1984 excavation at Sanxingdui Locus III: five cultural strata51 (6) The 1984 excavation at Xiquankan: three cultural strata52 (7) The 1986 excavation at Sanxingdui Locus I: five cultural strata The 1986 excavation at Sanxingdui Locus II: five cultural strata The 1986 excavation at Sanxingdui Locus III: thirteen cultural strata53 (13) The 1997–98 excavation of the Rensheng cemetery: one cultural stratum54 The issue of stratigraphy presented itself almost at the very beginning of the fieldwork at the Sanxingdui site. When Graham excavated at Yueliangwan in 1934, he recognized one cultural stratum, and evidently thought, with good reason, that the pit he found and the cultural stratum belonged together, although he did not explicitly say so.55 In 1942, Lin Minjun voiced the opinion that the two were of different dates, with the cultural stratum belonging to the Neolithic period whereas the pit to a Zhou date. Lin 48Ma Jixian 1992, p. 312. 49Kaogu xuebao 1987.2, pp. 228–32. 50Kaogu xuebao 1987.2, p. 249. Chen Xiandan 1989a, however, claims that four cultural strata were discerned in this excavation (p. 215). 51Chen Xiandan 1989a, p. 217 table. This table usefully correlates the strata excavated from 1980 to 1986, but one must exercise caution in using it because the table uses his own distinctly problematic periodization scheme (see discussion below). 52Ibid. 53Ibid. 54Kaogu 2004.10, p. 15. 55See relevant discussion in Section 2.1.1. 30 based his dating of the cultural stratum on the absence of metal, and the perceived primitiveness of the pottery types and stone implements that characterized the stratum. For the date of the pit, he identified its stone and jade objects as types described in the Zhou li and accepted their Zhou date as commonly believed then. He dismissed the presence in the cultural stratum of similar objects as the result of a later intrusion.56 Lin’s conclusions were largely followed by Zheng Dekun in his influential study of ancient cultures of Sichuan published in 1946, and in it he further identified the pit burial as the result of a sacrifice to the local Minshan Mountain during the Eastern Zhou period, an identification guided solely by texts on Zhou offering rituals.57 The overbearing influence of traditional texts is obvious in both studies, which arbitrarily forced an absolute date on the pit as well as the dismissal of the evidence shared by the cultural stratum and the pit. Their interpretations exerted considerable influence at the time. It was not until 1961, when the field survey in the Sanxingdui site proved once again the co-existence of the artifact types, that some archaeologists began to reject their assertions.58 What Graham described as a single cultural stratum must in fact have included more than one stratum, however. The 1963 excavation at the same locale identified two cultural strata distinguished by differently colored soil.59 When the report of this excavation was published in 1992, its author had the benefit of a four-phase periodization 56Lin Minjun 1942, p. 100. He associated some of the sherds with the black pottery of the Longshan Culture at Shandong Chengziya. His dating of jade types followed Laufer 1927. 57Zheng Dekun 1946, pp. 39–42. See also Cheng 1982, pp. 64–5. 58Wenwu 1961.11. Yet, regrettably, the lesson is largely forgotten as many scholars have continued to this day to employ later texts as the primary guidance in their interpretation of the finds at the Sanxingdui site. 59Ma Jixian 1992, p. 312. 31 scheme constructed with the results of the 1980–81 and 1982 excavations at Sanxingdui. The two strata at Yueliangwan were accordingly identified as belonging to two phases: the lower one to Yueliangwan Phase I and the upper one to Yueliangwan Phase II, which corresponds respectively to Phase I, and Phases III and IV at Sanxingdui.60 Of the five cultural strata discerned in the 1980–81 excavation, the fifth and fourth—the two deepest layers—contained pottery vessels of similar traits although they were separated by a layer of soil free of any cultural content. Between the fourth and the third strata, which were also separated by a layer largely free of artifacts, a considerable difference in pottery types was discerned. Pottery features were again similar between the third and the second strata. The archaeologists therefore collapsed the five strata into three phases: Phase I: the earliest, the fifth and fourth cultural strata Phase II: the third and second cultural strata Phase III: the first cultural stratum Of these three phases, Phases I and II appear to be separated by a long time span as indicated by distinct differences in pottery traits, whereas Phase III probably followed Phase II immediately.61 Of the two cultural strata identified in the 1982 excavation, the lower one was equivalent to the first stratum in the 1980–81 excavation whereas the upper one showed different traits.62 This upper stratum became recognized as representing Phase IV in 1987; its pottery types are most recently summarized in the final 60Ibid., pp. 322–3. See also Li Mingbin 1999b. 61Kaogu xuebao 1987.2, pp. 228–32. Recent fieldwork at the Yufu site seems to supply a missing link between Phases I and II (see Li Mingbin 2001b). 62Ibid., p. 249. 32 excavation report of K1 and K2.63 In the following discussion, this four-phase scheme will be referred to as the Report Scheme to avoid confusion with other periodization schemes. The excavations from 1980 through 1982 thus for the first time established a comprehensive stratigraphy and periodization for the Sanxingdui site. Because of its fundamental importance, archaeologists through the years have offered much criticism and many attempts to refine the periodization. A direct critique was first published by Huang Jiaxiang, who pointed out various inconsistencies in the factual information provided by the report of the 1980–81 excavation as well as possible mistakes in interpretation.64 His critique, remarkable for its rigor, does not challenge the basic stratigraphy and periodization of the report, however, but injects a welcome caution about many details. Chen Xiandan, a co-author of the 1980–81 excavation report, incorporated further results from excavations through 1986 in his study, and proposed a somewhat different four-phase scheme (Table 1a).65 However, his scheme is marred by obvious mistakes and confusion, as pointed out by Sun Hua,66 and has been largely ignored by other scholars. Sun Hua has produced so far the most influential revision of the periodization.67 He endorsed the basic validity of the Report Scheme, but proposed a three-phase scheme with more finely divided subphases: 63Wenwu 1987.10 (p. 1), Zhao Dianzeng 1987 (p. 19), and Beijing 1999a (pp. 424–7). 64Huang Jiaxiang 1990. 65Chen Xiandan 1989a, pp. 217–23. For the relationship between Chen Xiandan’s periodization and cultural identification and those of the report and of Sun Hua, see Table 1b. 66Sun Hua 1992, pp. 11–3. 67Sun Hua 1992. 33 Sun’s Phase 1 = Report Scheme Phase I Sun’s Phase 2 = Report Scheme Phases II and III Sun’s Phase 3 = the stratum of K1, and the upper stratum of the 1982 excavation.68 Sun further divided the three phases into six subphases illustrating the gradual changes in pottery traits. Phase 1: Subphase 1 Phase 2: Subphases 2–4 Phase 3: Subphases 5–6 Sun’s study is solidly anchored in pottery typology with sensitive attention to changes in both vessel types and traits. Consequently, his subphases more closely mirror the changes of the pottery vessels at the Sanxingdui site than the Report Scheme, and his study has since exerted much influence on other efforts at periodization.69 Yet the two schemes are essentially compatible as can be seen from the equations given above. Sun later modified his periodization by moving his subphase 5, to which he thinks both K1 and K2 belong, to Phase 2.70 His updated periodization, hereafter referred to as the Sun Scheme, now stands as: Phase 1: Subphase 1 = Report Scheme Phase I Phase 2: Subphases 2–5 = Report Scheme Phases II, III, and 68In 1986, when K1 and K2 were excavated, K1 was found to lie beneath a stratum containing potsherds typical of the Report Scheme’s late Phase III, and K2 beneath a deposit that can be dated, also by pottery typology, to early Phase IV (Beijing 1999a, p. 427). 69Recently published schemes such as Li Boqian 1997, Wang & Zhang 1999, and Jiang et al. 2002 (pp. 7–11) are mostly in general agreement with the finely divided subphases of Sun’s periodization. 70Sun Hua 2000, p. 97 and pp. 142–4. 34 early Phase IV Phase 3: Subphase 6 = Report Scheme late Phase IV For the purposes of the present dissertation, the Report Scheme is sufficiently finely divided, and it will therefore be used. Combing the typological studies so far known, we may summarize the pottery types for each phase as follows. Phase I is characterized by only a few types, including ring-footed dou, and pots with sharply folded mouth rim. Their fabric is either grayish and fine in texture, or brownish and sandy, and they often bear cord-marked or incised patterns (Fig. 2.12a). The dou continued in Phase II with changes in traits, and the brownish sandy fabric became the majority. Phase II is distinguished, however, by changes rather than continuity, with the advent of several new types that would persist through Phases III and IV, including guan with tapered profile and small flat bottom, tripod he, and high-stem dou. Also new are ladles in the shape of a bird with a hooked beak. The dominant pattern—cord-marking—became coarser (Fig. 2.12b). Phase III shows close affinity with Phase II in fabric, type and pattern, though with gradual changes in specific morphological traits and addition of a few new types such as gu, zunshaped vessel with broad shoulder but small flat bottom, and tripod pot with a broad collar along the neck. More vessels are plain in surface in this phase (Fig. 2.12c). Most vessels of Phase IV have gray-brownish sandy fabric, but the fraction of fine grayish fabric increased significantly. Continuity shows in several types such as guan with tapered profile and small flat bottom, tripod he, bottle, and vessel lid, but their morphological changes are larger than before and more vessels are now plain in surface. The major new types of this phase were cups and bowls with pointed or nearly pointed 35 bottom (Fig. 2.12d). The change between Phases IV and III is certainly larger than that between Phases III and II. The pottery type that has provoked the most discussion is the bowl with pointed bottom (jiandi zhan), which did not appear in the 1980–81 excavation but showed up in large quantities in the 1982 excavation, and later in other excavations, including K1 (Fig. 2.13). Clearly this vessel type was developed late in the occupation of the Sanxingdui site, but in which phase? In the Report Scheme it is claimed that the type first appeared toward the end of the Phase III.71 In Sun’s first periodization, it did not emerge until Phase 3, but in his later Sun Scheme, its appearance is dated to Subphase 5 of Phase 2.72 The difference may seem a small technicality. However, when different cultural identities are proposed for the phases, as in the Sun Scheme, the difference takes on a weighty significance, as will be discussed in Section 2.4. 2.2.2 Absolute date The absolute dating published in the report of the 1980–81 excavation (including remarks about the 1982 excavation) relied on the following evidence. A calibrated radiocarbon date was obtained from a sample of charcoal from Phase I remains, which gives the radiocarbon date of 4075±100 BP (with 5730 half-life) and the calibrated date of 4500±100 BP.73 The fine leiwen-spirals decorating some of the latest pottery objects 71Beijing 1999a, p. 424. 72Sun Hua 1992 (p. 19), and Sun Hua 2000 (p. 144). 73Prior to the publication of an updated collection of radiocarbon dates in Beijing 1991, radiocarbon dates were calibrated in China using the calibration curve constructed by Paul E. Damon and his collaborators and published in 1972 (see Beijing 1982, p. 439 for reference and explanation). In this method, the calibrated dates are expressed in the same manner as the uncalibrated dates. Kaogu xuebao 1987.2 does not specify the lab no. of 36 from the 1980–81 excavation, i.e. objects of Phase III, are similar to the leiwen on Shang bronzes. Besides, the upper stratum of the 1982 excavation yielded pottery vessels with pointed bottom, the likes of which had appeared in abundance in the Shuiguanyin site, a site that can be dated to the end of the Shang period or the beginning of the Western Zhou period through associations in bronze.74 By now, a series of 28 radiocarbon dates is available, narrowing the error limits of the radiocarbon process and giving a reasonably clear range of dates for the Sanxingdui site (Table 2).75 Excluding two outliers at each end, the range of absolute dates appears to be from the early third millennium to about 1000 cal BC. The demarcation between Phase I and Phase II seems to fall around 2000 BC.76 Artifact associations make it possible to roughly define the time spans of the other three phases: Phase II, first half of the second millennium BC, and Phases III and IV, second half of the second millennium BC, with c. 1200 BC as a convenient though very approximate demarcation between III and IV.77 To a large degree, this dividing line between Phases III and IV depends on stylistic evidence supplied by the bronze vessels from K1 and K2; this has been the sample that gives the dates, but it is easy to see in Table 2 that only the sample of lab no. ZK-0973 has the radiocarbon date of 4075±100. 74Kaogu xuebao 1987.2, p. 249. For dating of the Shuiguanyin site see Kaogu 1959.8 (p. 410), and Jiang Zhanghua 1998a (p. 6). 75Eighteen of the dates, acquired from 1980 through 1989 and originally published in Kaogu and Wenwu, are summarized and updated in Beijing 1991, pp. 224–6. Falkenhausen has turned those dates into a useful radiocarbon profile (Falkenhausen 2002, p. 93 table 2). In addition, Table 2 includes seven calibrated dates published in Kaogu 1992.7 (p. 660) and 1993.7 (pp. 648–9), and three uncalibrated dates published in Wenwu 1996.6 (p. 93). I have re-calibrated all the dates using the latest calibration softwares. For basic principles and problems of the radiocarbon method see Bowman 1990. 76Beijing 1999a, p. 424. 77For detailed comparisons in pottery types for absolute dating see Sun Hua 1992 (pp. 19–21), Li Boqian 1997 (pp. 273–4), Wang & Zhang 1999 (pp. 17–20), and Beijing 1999a (pp. 424–7). 37 addressed by Bagley, Chen De’an, and other scholars.78 The vessels will be reviewed in detail in Section 4.2. 2.3 Size of the city, city walls, pits of objects, and other features Like any habitation site, the Sanxingdui site must have changed in size and features throughout its history of occupation, but fieldwork at the site does not yet afford detailed information about each phase. Only a very rough estimate may be made by putting together relevant pieces of information. Before doing so, a caveat must be voiced about the way the layout of the Sanxingdui city is all too often described: all the features being lumped together without regard to their particular time period. Sun Hua, for example, has produced in this way a reconstruction of the layout of functions in the city that is at least inaccurate and potentially wrong (Fig. 2.14).79 Although the excavations and their publication at present cannot make possible a detailed phase-by-phase account, every effort should be made to identify the phase to which features belong when they are described. 2.3.1 Settlement during Phase I In the days of Phase I, the longest of the four phases, the Sanxingdui site was probably only a modest settlement. Among the areas so far excavated, remains of this 78Bagley 1988, Bagley 1990a, and Beijing 1999a (pp. 428–31). For general discussion of the dating of bronzes of this time see Bagley 1987 (pp. 19–32), and Bagley 1999 (pp. 146–55). 79Sun Hua 2000, pp. 162–5. 38 phase exist at Sanxingdui Locus III, Yueliangwan, Xiquankan, and Rensheng.80 As we recall, field surveys in the Yueliangwan area by 1961 had led archaeologists to identify the site as the “Zhongxing site” covering an area about 3 km long between the Yazihe and Mamuhe rivers with Yueliangwan as its center. Perhaps the archaeologists would not be entirely wrong in pronouncing Yueliangwan as the center of the site if only Phase I were concerned. Sanxingdui Locus III is the northernmost among the three Sanxingdui locales and north of the Sanxingdui wall (Figs. 2.3, 2.5). Significantly, no Phase I remains were excavated in either Locus I or Locus II in the 1982, 1984 and 1986 excavations. It is therefore possible that Locus III might represent the southernmost limit of the Sanxingdui site during Phase I. Only few features can be assigned definitely to Phase I. Three of the eighteen building foundations at Locus III probably date from this phase: two of them, labeled F16 and F18, are circular in floor plan, and the other, labeled F17, is square. F16 and F18 measure about 3 and 4 m in diameter respectively, while F17 is 3.5 m long on each side enclosing a space about 12.25 sq m (Fig. 2.15).81 Several other foundations of this phase were partially exposed in the 1963 excavation at Yueliangwan.82 Burials of young adults or children were found beneath building foundations, perhaps ritual offerings made at the 80Phase I strata were excavated at Sanxingdui Locus III in both 1980–81 and 1986, and one stratum of Phase I was excavated at Yueliangwan in 1963, at Xiquankan in 1984 (see Table 1a), and at the Rensheng village in 1998. 81Kaogu xuebao 1987.2, pp. 233, and 249 table 2. Huang Jiaxiang 1990, however, challenges the accuracy of the report’s assignment of these three house foundations to Phase I (pp. 1031–2). Given the similarity in building construction with Baodun settlements to be mentioned below, the possibility of their belonging to Phase I has become quite strong. 82Ma Jixian 1992, p. 313. 39 time of laying the foundations.83 The Rensheng cemetery, datable to late Phase I and early Phase II (Fig. 2.9),84 constitutes a major feature of the period. Although the cemetery now lies outside the west wall, there was probably no wall at the time. Similarities in features and artifact types indicate that Phase I of the Sanxingdui site belonged to the Baodun Culture. The pottery vessels resemble those from Baodun walled settlements, so do the stone tools (small axes, adzes, and chisels). Usually well made, the stone tools attest a developed agriculture. Building construction was by the socalled “wooden frame–mud wall” technique, the same as was used at Baodun settlements.85 Some of the artifacts from the Rensheng cemetery suggest distant contacts, however: a pottery dou is close in style to the Shijiahe type (Fig. 2.16), whereas three awl-shaped jade implements find parallel not only in the Shijiahe Culture but also in the Liangzhu Culture of the third millennium BC in the costal area of the lower Changjiang region (Fig. 2.17).86 2.3.2 Settlement of Phases II through IV 83Chen De’an 1998, p. 58. 84Kaogu 2004.10, p. 21. 85Wang & Sun 1999 (pp. 63–5), Sun Hua 2000 (pp. 304–6), and Tokyo 2000 (pp. 114–5) (same as Jiang et al. 2001a, pp. 713–4). Prior to the discovery of the Baodun Culture, Wang & Ye 1993 (p. 264) first observed similarities in artifact types with Bianduishan in Mianyang and other similar Neolithic sites in northern Sichuan. Sun Hua 1992 (p. 23) proposed to classify the remains of Phase I as belonging to the “Bianduishan Culture”. The Bianduishan site is now identified as having close cultural affinity with the Baodun Culture. For Bianduishan and related sites in northern Sichuan see Kaogu 1990.4, Wang & Ye 1993, and Ma Xinxin 1993. For their relationship with the Baodun Culture see Wang & Sun 1999, p. 63. 86Kaogu 2004.10 (p. 22), and Jiang Zhanghua 2005 (pp. 17-8). Both mention also similar examples found at the Shijiefang and Jinsha sites in Chengdu. 40 Phases II and III at the Sanxingdui site show close affinity; the changes in pottery that prompt the archaeologists to distinguish them are subtle. A larger distinction exists between Phases III and IV but continuity there also is quite apparent. At the present stage of fieldwork there are not many features known for these phases and the walls evidently existed throughout. These phases will therefore be discussed together, noting changes in the settlement wherever possible. The walls at the Sanxingdui site seem to have been built in Phase II, as the pounded earth that composes them contains only potsherds of Phase I, and in all the three walls on the east, west and south, the earth lies atop a stratum belonging to Phase I while being broken into by strata of Phase II or later. The moat outside the walls contains deposits mainly of Phases III and IV, indicating that the walls were used through the end of the habitation.87 On the other hand, the complex layout of the city does give the impression of a gradually growing settlement (Fig. 2.3). The three Sanxingdui terraces, together measuring about 200 m long but perhaps originally much longer, lie on an east-west axis several hundred meters north of the south wall.88 The Xiquankan and Yueliangwan terraces seem to bisect the city in another way, on a north-south axis. The city was evidently divided into quarters by these and perhaps more walls yet to be discovered. It is possible that the walls were added at intervals during Phase II and later to accommodate a growing population and to deal with the wanderings of the rivers. On present information, the east wall of the city is estimated to have extended about 1.8 km long, the remnants visible today measuring about 1 km. The south wall 87Chen De’an 1998, p. 62. 88Chen et al. 1998, p. 1. 41 measures about 0.6 km, the west wall about 0.8 km. These walls average 40 m thick at the base and 20 m thick at the top. The highest surviving parts, belonging to the west wall, are 6 to 10 m high. The Sanxingdui terraces are about 6 m high. The city measured 1.6 to 2 km from east to west and about 2 km from north to south, enclosing an area of 3.5 sq km (350 hectares) in a roughly trapezoidal plan, the south wall forming the base of the trapezoid.89 In some respects the walls point to continuity with Baodun: they have the same gently sloping sides as Baodun walls, and they too may have been built for flood control. The cultural strata revealed in the 1980–81 excavation are interrupted twice by a layer of earth largely free of artifacts and at least one layer of blackish silt exists in the strata discerned during the 1986 excavation,90 suggesting that they might be alluvial deposits. At an elevation of 500 m, Sanxingdui is in the flood plain. The Mamuhe River, now a tiny stream often dry of water, was a broad river in ancient times, and both it and the Yazihe River flooded periodically. Sanxingdui archaeologists have not been able to determine the ancient course of the Mamuhe River, but Liu Xingshi, an environment specialist, has raised the possibility that a shift to a course cutting through the ancient city had something to do with the city’s abandonment.91 That is a serious possibility to reckon with: the map of the Sanxingdui site shows that the Mamuhe River makes an 89Chen et al. 1998 (p. 1), and Chen De’an 1998 (pp. 61–2). 90For the 1980–81 layers see Kaogu xuebao 1987.2, p. 231. That one such layer exists also in the 1986 strata can be inferred from a remark in Lin Xiang 1987 (p. 81) where it is mentioned that on top of the eighth strata is a layer of blackish silt largely free of artifacts that might be an alluvial deposit. It is also mentioned that the eighth stratum belongs to the fourth phase. Checking Table 1a makes it clear that the only eighth stratum belonging to the fourth phase is that of the 1986 excavation. However, it cannot be ascertained to which particular locale this eighth stratum belongs, for all of the three locales excavated in that year, Loci I, II and III, have an eighth stratum. 91Liu Xingshi 1998, p. 37. 42 abrupt change of course where it meets the west wall, and the middle section of the surviving wall makes a turn as well as if to follow the course of the river. Also, as Liu observes, although the ancient city is now bounded in the north by the Yazihe River, there are terraces immediately next to the river which are possibly remnants of a north wall.92 Actually we may recall that the first trace of man-made wall was discovered at Xiquankan, a terrace along the river. If indeed there was a north wall, much of it must have been washed away when the Yazihe River changed course. Moreover, there seems to be evidence arguing against the walls being defensive ramparts: having gentle slopes, the walls would surely have been easy for an enemy to scale, particularly in places where the outer face slopes more gently than the inner face (Fig. 2.8). Significantly, of the three layers of likely alluvial deposits, one of the 1980–81 layers belongs to Phase I while the other separates Phase I from Phase II, and the 1986 layer belongs to Phase IV. Floods may have been a perennial problem at the site. The wall building technology at Sanxingdui was undoubtedly acquired from the Baodun Culture. As dissection at the east wall showed, in cross section the wall consists of three distinct parts, a middle part in which the pounded earth layers are horizontal, and inner and outer parts in which the layers slope. The excavation at the Yueliangwan wall in 1999 revealed the same feature there (Fig. 2.18).93 The earth used to build the walls was dug immediately outside them, leaving a moat 20–30 m wide,94 and making the ground level higher inside than outside in a way reminiscent of the walled Baodun 92Ibid. 93The information about the east wall was obtained in personal communication with Chen De’an in October 1999 during a visit at the site. During the same visit, I also witnessed the excavation at the Yueliangwan wall. Structures of the other walls are unknown at present. 94Chen De’an 1998, p. 62. 43 settlements (Fig. 2.8). But the Sanxingdui walls are somewhat more sophisticated. A layer of adobe brick—bricks shaped in forms and sun-dried but not fired—has been found on the top surface of a section of the east wall (Fig. 2.19). This is the earliest such use known in the Chengdu Plain, but we do not know exactly when this layer of adobe brick was added atop the wall. In the most conspicuous manner, the walls embodied the rise of Sanxingdui from a modest settlement of secondary importance in Baodun times to the metropolis of a new epoch, its size overshadowing by far the earlier Baodun settlements. This must have been a dramatic moment. As the walls changed the landscape, other human structures must have been changing the area within them. We may imagine that, as the walls rose, palatial buildings, temples, public monuments and so forth were being erected as well, though few traces of them have survived the ravages of time.95 Inside the city, remains of Phase II were excavated at Xiquankan in 1984 and at Sanxingdui Locus III in 1980–81 and 1986. Remains of Phase III were uncovered in those excavations, and in addition, at Locus III in 1984. Remains of Phase IV were excavated at Locus I in 1984, Locus III in 1984, and Loci I, II and III in 1986 (Table 1a). Features included mainly building foundations and pits. A count of more than 50 building foundations has been given, but without breakdown into phases.96 On present information, 15 of them at Sanxingdui are known to belong to Phase II (numbered F1– 95The contents of K2 actually seem to encapsulate in sculptural form some of these structures such as the bronze pedestal or building in Figure 5.23, the bronze altar(?) in Figure 5.43, and the bronze roof-like structure in Figure 5.53. 96Chen et al. 1998, p. 2 44 15), while the numbers for Phases III and IV are not clear.97 Those fifteen foundations are rectangular and, in a few cases, square in floor plan. The circular plan of Phase I seems to have gone out of fashion. The buildings were built in the same “wooden frame– mud wall” method as before. The sizes of the buildings seem to be on average somewhat larger than in Phase I, varying from 9 to 76 sq m, with the majority between 11 and 19 sq m.98 The largest is a rectangle 8.7 m deep and 23 m wide, occupying 200 sq m and comprising multiple rooms.99 Much larger than the average, it must have been a structure of considerable importance. Yet it is still modest by comparison with the 550-sq-m building in the Gucheng settlement of the Baodun Culture. It would not be surprising if the Sanxingdui city had buildings much larger than those that have been found. At present, foundations appear to congregate in the Sanxingdui area, but this may only be because excavations of large scale have so far not taken place anywhere else. The picture may well change after more fieldwork. Rectangular pits containing ritual objects seem to be a regular feature of the Sanxingdui site. Nine pits have so far be excavated in the city. 97The information about the fifteen building foundations is from Kaogu xuebao 1987.2, pp. 234–7. The 1986 excavation exposed dozens of building foundations, but their periodization has not been published. 98According to Kaogu xuebao 1987.2, p. 253 table 2. Different ranges of sizes for building foundations are given in several publications. Kaogu xuebao 1987.2 (p. 233) gives the average of 10 to 25 sq m, which is for both Phase I and Phase II foundations, but the sizes given in the table at the end of the report seem to differ from this range. Chen Xiandan 1989a (p. 227) gives the average of 14 to 35 sq m, while Chen De’an 1998 (p. 61) and Chen et al. 1998 (p. 2) give the average of 15 to 30 sq m. None of these three publications gives information about the dates of the buildings or about the excavations from which the figures derive. 99Chen De’an 1998 (p. 61), and Chen et al. 1998 (p. 2). The information that this building dates from Phase II was provided by Chen De’an in a personal communication in 1999. 45 (1) One pit at Yueliangwan found in 1927 contained many stone and jade objects, including 20 or so stone bi of graduated sizes. Some of the objects retained traces of a red pigment on the surface.100 There are two accounts about the layout of the disks. According to one, they were found piled up in the shape of a cone with smaller ones atop larger ones.101 According to the other, the top of the pit was covered with disks gradually diminishing in size, lying in a flat or horizontal position. The sides were lined with disks in a vertical position, also gradually decreasing in size. Under the horizontal layer were buried other objects of finer stones.102 The pit was over 2 m long, nearly 1 m wide, and about 1 m deep, with the longer sides oriented almost east and west.103 (2) One pit at Yueliangwan was found in 1964, about fifty or sixty meters away from the 1927 pit. It contained about 300 hundred pieces of stone 100Dye 1931 (p. 102), Graham 1934 (p. 116). The inventory of objects and their quantity uncovered from the pit were never known. Sometimes it is estimated that as many as three or four hundred objects were found; see, e.g., Feng & Tong 1979, p. 31, the source of which is not given. None of the earliest accounts, Dye 1931, Graham 1934 and Lin Minjun 1942, gives a specific count, however, and their wording seems to make it less than probable that the number would be in the hundreds; see Dye 1931 (p. 102), Graham 1934 (p. 114), and Lin Minjun 1942 (pp. 93–4). 101This account was originally given by photographer Jin of the West China Union University Museum, who joined Rev. Donnithorne and Dye in a visit to the Yan family in June 1931 to record the locale and finds. See Lin Minjun 1942 (p. 93), Zheng Dekun 1946 (p. 33), and Cheng 1949 (p. 56). 102Graham 1934, p. 116. This account was apparently given by Rev. Donnithorne, who draw the layout published in Graham 1934. 103Graham 1934, p. 116. 46 and jade materials including finished, partly worked objects and raw materials.104 (3) One pit at Suozitian in the Yueliangwan area was found in 1974. It contained several dozens of stones with polished surfaces—possibly whetstones.105 (4 & 5) Two pits found at Sanxingdui in the spring of 1986 contained a small number of bronzes, stone and jades, burned bones, and ashes.106 (6 & 7) Two major pits, K1 and K2, were found at Sanxingdui in the summer of 1986. K1 was 4.5–4.64 m long and 3.3–3.48 m wide at mouth, 4.01 m long and 2.8 m wide at bottom, and 1.46–1.64 m deep. K2 was 5.3 m long and 2.2–2.3 m wide at mouth, 5 m long and 2–2.1 m wide at bottom, and 1.4–1.68 m deep.107 (8) One pit at Cangbaobao, a terrace about 400 m east of the 1927 pit, was found in 1987. It contained stone and jade objects as well as bronzes. The former category includes eight jade rings, one jade bracelet-like ring, three stone axes, one small stone cong, and 21 stone bi of graduated sizes allegedly stacked up in the shape of a cone. Other types of stone and jade objects whose quantities have not been published include chisel–shaped implement(s), cores drilled from perforating the stone bi, and stone balls for slingshot. The bronzes are three plaques of a type comparable to those 104Feng & Tong 1979 (p. 31), Ao & Liu 1991 (p. 332), and Beijing 1999a (p. 15 note 12). 105Ao & Liu 1991 (p. 332), and Xiao et al. 2001 (p. 24). 106Beijing 1999a, p. 12. 107Beijing 1999a, pp. 19 and 157. 47 unearthed at Erlitou, c. 1500 BC, in the Central Plain. The pit is about 2 m long, 1 m wide, and 0.4 m deep, and in addition to artifacts, it was sprayed with a red pigment and contained burnt bits of bone and ash.108 (9) One pit at Shizinao at the south end of what remains of the east wall, found in 1987, contained stone and jade objects.109 Related are two more pits outside the Sanxingdui site. (10) A pit found in 1976 at Gaopian, 10 km to the northwest, contained one bronze plaque similar to those from Cangbaobao, one stone axe, one jade axe-shaped blade, and one jade spearhead.110 (11) A pit found in 1987 at Mayang in Yanting county of north Sichuan contained 10 stone bi of graduated sizes laid out horizontally at the bottom of the pit. The pit measures 3.2 m long, 1.2–1.6 m wide, and 1–1.3 m deep.111 These eleven pits are published in varying amounts of detail, ranging from booklength formal report (K1 and K2) to detailed preliminary report (1927 Yueliangwan and 1987 Cangbaobao), brief report (1976 Gaopian and 1987 Mayang), or just brief remarks (1964 Yueliangwan, 1974 Suozitian, 1986 Sanxingdui, and 1987 Shizinao). The dates of K1 and K2 are the most securely established: late Phase III and early Phase IV respectively. The presence of Erlitou-type bronze plaques in the 1976 Gaopian 108Beijing 1998a. 109Zhao Dianzeng 1996b, p. 93. 110Ao & Wang 1980. That the objects was found in a pit is actually not mentioned in this brief report. However, both Beijing 1998a (p. 88) and Zhao Dianzeng 1996b (p. 94) identify it as a pit. 111Zhao Zike 1991, and Zhao Dianzeng 1991. 48 pit and the 1987 Cangbaobao pit suggests a possible date equivalent to Phase II.112 The dates for the other pits are less certain, but the 1963 excavations at Yueliangwan and the 1980–81 excavations at Sanxingdui provide some important reference points. In both excavations fragments of stone bi were unearthed, at Yueliangwan from the upper cultural stratum, which is datable to Phases III and IV, and at Sanxingdui from strata datable to Phases II and III. Significantly, none was found in Phase I strata at either location.113 Moreover, no such bi have appeared in any of the Baodun settlements. This stone artifact type almost certainly appeared in the Chengdu plain for the first time during Phase II, and it seems to have lasted through Phase IV. The 1927 Yueliangwan pit and the 1987 Mayang pit may thus be broadly dated to Phases II to IV, but the absence from these pits of stone or jade dagger-axes (ge), abundantly represented in K1 and K2, perhaps suggests that Phase II is a more likely possibility.114 The other five pits are barely published, but they appear to date broadly from Phases II to IV as well. These pits all contained stone and jade artifacts, often similar types, except for the 1964 Yueliangwan pit and the 1974 Suozitian pit whose inventories exhibit remarkable differences.115 None seems to have contained functional tools and weapons or traces of burial furniture and human remains, but sometimes burnt bone and ash (K1 and the 1987 Cangbaobao pit). Objects were often smeared with a red pigment (the 1927 Yueliangwan pit, K1, K2, and the 1987 Cangbaobao pit), and were often burned (K1, K2, the other two 1986 Sanxingdui pits, and the Cangbaobao pit). Their layout in the pit often shows a 112Beijing 1998a, pp. 88–9. However, the possibility that the plaques might be later imitations also exists; see discussion in Section 4.1. 113Ma Jixian 1992 (pp. 321–2), and Kaogu xuebao 1987.2 (pp. 230–1, 248–9). 114Zhao Dianzeng 1996b (p. 94) dates the two pits earlier than K1 and K2 to about 1500– 1400 BC without giving specific reason. 115The artifact types contained in the 1987 Shizinao pit is not known at all. 49 deliberate arrangement (the 1927 Yueliangwan pit, K1, K2, the 1987 Cangbaobao pit, and the 1987 Mayang pit).116 Given these shared traits, it is reasonable to propose that these pits were the result of ritual offerings made at different times.117 Among them, K1 and K2 of course stand out by their enormously richer content, particularly in the category of metal, and also by their much larger size. They seem to represent offerings of the highest order in the local community, but they may also reflect a dramatic increase of wealth of the local elite. The content and nature of the two pits will be discussed in detail in Chapter 3. The 1964 Yueliangwan pit and the 1974 Suozitian pit may have been storage pits, as the former contained fragments, half-finished objects and raw materials in addition to finished objects, while the latter contained whetstones. Similar debris and raw material have been found elsewhere in this general neighborhood, including at Xiquankan, suggesting that workshops for stone and jade were located here.118 Workshops may have existed in other areas also, for a large number of stones, some of which had been sliced or half sliced, were found at Sanxingdui Locus III in 1986.119 The 1963 excavations in the Yueliangwan area also yielded possible traces of bronze foundry,120 but the evidence is far from conclusive. Clay cores for bronze casting were found among cultural remains in 116Curiously the two versions for the layout of bi in the 1927 Yueliangwan pit correspond to the allegedly layout of the disks in the Cangbaobao pit and the Mayang pit respectively. 117This point has been made earlier in Zhao Dianzeng 1996b, pp. 94–5. 118Chen Xiandan 1988 (p. 10), and Xiao et al. 2001 (p. 24). 119Chen Xiandan 1988, p. 10. 120See description in Section 2.1.2. 50 the Sanxingdui area also, including in the earth fill of K1,121 but again such traces are not sufficient to establish the presence of a bronze foundry in the area. Beside walls, building foundations and pits, the features included also a pottery kiln of Phase IV discovered in 1982 at Sanxingdui Locus I,122 and a few small tombs poorly furnished or with no burial goods at all.123 The preceding descriptions make clear that much more fieldwork is needed to reveal and map out the features inside the city. At this point, we have little idea about the layout of the city and the daily lives of its inhabitants. The reconstructed layout shown in Figure 2.14 claims that the Yueliangwan area was the location of royal palaces because the high terrace there would be suitable for such a function and because there were many pits of stone and jade objects.124 Such a claim is premature. The same may be said about the proposal that the temples of gods were located in the Sanxingdui area simply because K1 and K2 are located there.125 Outside the city, little fieldwork has been performed apart from a few surveys to estimate the extent of the site. While they afforded the important understanding that the site covers an area of 10 to 17 sq km, the surveys did not reveal at which time during its occupation the Sanxingdui settlement reached maximum size, nor whether the size of the site is equivalent to the size of the settlement at its largest. For the first question, if we have to guess, I would suggest the time around 1200 BC, when the city seems to have 121Beijing 1999a, p. 22. 122Chen Xiandan 1989a, p. 215. 123Kaogu xuebao 1987.2 (p. 238), and Ma Jixian 1992 (pp. 312–3). 124Sun Hua 2000, p. 163. 125Ibid. 51 possessed enormous wealth as attested by the content of K1 and K2. As for the second question, there is no clue at all. Whether inside the city or outside, one conspicuous absence in the present archaeological record is major tombs. So large and rich a city is likely to have buried the dead of its ruling class ostentatiously. For that, we have to wait for future fieldwork to find out. 2.4 Defining the Sanxingdui Culture For archaeologists in China as elsewhere, “archaeological culture” commonly means a distribution of material culture traits that can be observed consistently over a restricted area and within a given period.126 It is delimited in both time and space. Routinely, the archaeological culture is defined mainly by pottery types because of their ubiquitous presence in human habitations from the Neolithic period on. Although artifacts of other materials like stone and features such as tombs and building foundations often figure in the definition of a given culture, they are clearly of secondary importance in actual practice. The distinction between an archaeological culture and the culture of a site must be made clearly. An archaeological culture is routinely shared by multiple sites, but a site may have multiple cultural affiliations, as it may have remains belonging to more than one culture, whether diachronically through time or synchronically across space. In practice, particularly in the early stages of fieldwork at a site, it is often difficult to see the cultural distinctions clearly. 126Jiang Chunfang 1986, p. 253. 52 As discussed before, the Sanxingdui site is divided into four phases in the Report Scheme. This scheme treats all the phases as belonging to the same culture, namely the Sanxingdui Culture.127 In contrast, Sun Hua’s three-phase scheme originally published in 1992 treats each phase as a different culture, which he named as the “Bianduishan Culture,” the “Sanxingdui Culture,” and the “Shi’erqiao Culture.”128 To estimate the merits of the two schemes, let us briefly summarize the differences in pottery typology between the Report Scheme’s four phases. A major typological difference exists between Phases I and II: the latter is distinguished from the former by the advent of a whole new set of vessel types, and continuities between the two are limited. In Phase III, the vessel types of Phase II continued with little change, but new types also appeared. Particularly noteworthy is the emergence of jiandi zhan, the bowl with pointed bottom, towards the end of this phase, when it appeared for the first time in K1. In Phase IV, jiandi zhan and other vessels sharing the pointed bottom became a major presence. On the other hand, several vessel types persisted from Phase II through this phase, including but not limited to guan with tapered profile and small flat bottom, tripod he, and bottle, though their morphological changes are more noticeable (Figs. 2.12b–d). By comparison, the difference between Phases III and IV is larger than that between Phases II and III. Since the differences between Phases I and II in pottery typology appear indeed large and qualitative, it seems warranted to recognize the two phases as representing two different cultures in the pottery-based definition customary in Chinese archaeology. Lending critical support to this distinction is the rise of the city with magnificent walls 127Kaogu xuebao 1987.2, p. 250. 128Sun Hua 1992, p. 23. 53 during Phase II, which signifies a basic change in the nature of the settlement.129 In this light, continuities between the two phases must be considered as secondary to the changes, though they indicate a direct relationship between them. The pottery typology recently revealed at Baodun settlements has confirmed Sun’s distinction and Phase I is now re-named as that of the Baodun Culture.130 Since Phases II and III are particularly close, it is also not without reason that Sun Hua first combined them, minus K1 of late Phase III, into one phase, his Phase 2. The reason for excluding K1 from Phase 2 is obviously the presence of jiandi zhan, which heralded the popularity of vessels with pointed bottom in Phase IV. Sun subsequently grouped K1 and Phase IV into one phase, his Phase 3. However, does the difference between Phases 2 and 3 warrant defining a new culture, or given the considerable continuity seen in other vessel types, would it be more appropriate to consider it as reflecting a large gap with some missing link between two developmental stages within the same culture? Phrased in another way, how much cultural significance should one give to jiandi zhan? For Sun, the advent of jiandi zhan in K1 signals a new culture because the vessel and related types would not only become a distinctive characteristic of Phase IV at Sanxingdui but would also appear widely distributed at a cluster of sites at Chengdu, with Shi’erqiao as their type site. Moreover, Phase 3 represents for him a stage of decline at 129Although walled settlements had appeared earlier at Baodun settlements in the Chengdu plain, the Sanxingdui city is much larger in size, undoubtedly the product of a very different society. 130For a summary of Phase I of Sanxingdui in the context of the Baodun Culture see Li Mingbin 2001c. 54 the Sanxingdui site.131 That is why he classified Phase 3 as representing the Shi’erqiao Culture. Sun’s three-culture theory, like his three-phase scheme, is a pioneering contribution that has remained deeply influential. Most Sichuan archaeologists have come to accept his three-cultural identification.132 However, while the identification of Phase 1 as belonging to the Baodun Culture seem justified, his distinction between the Sanxingdui Culture and the Shi’erqiao Culture is problematic in some regards. First of all, the Shi’erqiao Culture identification causes the confusion and incongruity of labeling the major archaeological finds of the phase, i.e. K1 and K2, with a name taken from another far more modest site. Sun Hua’s cultural identification was based solely on the evidence of pottery without regard to material culture of the elite. Two dozen jiandi zhan in K1 outweighed nearly one metric ton of bronze plus many other valuables from the two pits.133 This not only reflects an unbalanced assessment of material culture, but also exposes an innate problem of pottery-based cultural definition, a topic on general methodology that cannot be covered by this dissertation. Judging from the material wealth from K1 and K2 as well as the city walls, it is unquestionable that labeling the Sanxingdui site of that phase by the name Shi’erqiao is not appropriate. 131Sun Hua 1992, pp. 17–8, 23. 132Those archaeologists include Chen De’an, one of the leading excavators at the Sanxingdui site (see Zhao & Chen 2001, pp. 461–65). Chen Xiandan, another leading excavator, however, maintains the position that there was only one culture at the Sanxingdui site throughout its duration; he even proposes that the Baodun Culture should be re-identified as the “Baodun phase of the Sanxingdui Culture” (see Chen & Liu 2002). Chen Xiandan’s position in effect regards the Neolithic stage of the Sanxingdui settlement as the same as its Bronze Age stage, a position clearly untenable. 133The estimate of one metric ton is given in Guangming ribao, December 10, 1986. 55 Sun himself must have been bothered by the inconvenience that K1 and K2 impose. In the updated study he published in 2000, he modified his periodization by moving K1 and K2 to his Phase 2, allowing the two pits to be identified as belonging to the Sanxingdui Culture.134 This new three-phase scheme is what was referred to in the preceding discussion on stratigraphy and periodization as the Sun Scheme. Yet, the modification seems to create a new challenge to his distinction between the Sanxingdui Culture and the Shi’erqiao Culture. Now that jiandi zhan originates in the Sanxingdui Culture, this vessel type loses the position Sun had claimed for it as a signifier of a new culture, instead favoring the idea that Phases 2 and 3 were two developmental stages of a single culture. Another challenge to Sun’s identification of Phase 3 as belonging to the Shi’erqiao Culture exists whether or not K1 and K2 are included in that phase. The challenge can be presented from several angles. First, there seems to be no conclusive evidence that the Sanxingdui site was in decline during Phase 3 as Sun claims. In Tables 1 and 2 we can see that remains of Phase IV, which is basically equivalent to Sun’s Phase 3, have actually been excavated at more locales than those of any other phase.135 The explanatory remarks published by Chen Xiandan for Table 1a state that the remains of Phase IV can be further divided into two subphases, which seems to suggest that their distribution and quantity are not small.136 As there has been no quantitative analysis 134Sun Hua 2000, p. 142–4. 135The eighth stratum in the 1986 excavation at Sanxingdui, which Chen Xiandan dates to his fourth phase, is classified in Sun’s periodization as belonging to Phase 2, but the removal of that stratum does not affect the statement made here . 136Sun Hua 2000 (p. 8) states that Phase 3 remains at the Sanxingdui site have so far only been excavated at a few small areas including the Xiquankan locus. Jiang & Li 2002 (p. 110) similarly states that the cultural remains of the period corresponding to Phase 3 are 56 undertaken or published for any phase, and since fieldwork has only touched a very small portion of the site, it seems premature to state that Phase IV (or Phase 3) was a time of decline. Indeed, the Sanxingdui city was abandoned at the end of this phase, but that need not be the result of a drawn-out process of decline. Second, there seems to be no conclusive evidence that vessels with pointed bottom were more widely distributed at Chengdu than at Sanxingdui, an assertion that is the sole basis for Sun’s cultural identification of Phase 3. Of the thirteen strata at the Shi’erqiao site, it is generally agreed that the tenth through the thirteenth represented an early stage of culture at the site, which is variably called the Shi’erqiao Culture or an early stage of it.137 The thirteenth stratum, the earliest, contained vessels with pointed bottom, guan with tapered profile and small flat bottom, tripod he, high-stem dou, and ladle in the shape of a bird with a hooked beak. Most of these types continued in the twelfth stratum, while the eleventh and tenth strata seem to contain only vessels with pointed bottom (Fig. 2.20).138 This artifact list instantly makes clear that these strata or at least some of them are contemporary with the Sanxingdui site. In Sun Hua’s estimation, the thirteenth through the eleventh strata correspond to Sanxingdui Phase 3.139 Among the cluster of settlements in Chengdu, however, few can be dated to this phase. A stratum at the Fuqinxiaoqu site was considered by Sun as dating from an interim time between thin and therefore the city was clearly in decline, or perhaps abandoned already. It is unclear what information these statements are based on. 137Wenwu 1987.12, Sun Hua 1996 (pp. 123–9, and 141–2), Jiang Zhanghua 1998b (pp. 146–56), and Wang et al. 1999 (pp. 4–5). 138Wenwu 1987.12, pp. 4 and 14. 139Sun Hua 1996, pp. 125 fig. 2, and p. 129. Jiang Zhanghua 1998b (p. 162) equates the thirteenth and twelfth as contemporaneous with Phase 3. 57 the eleventh and tenth strata, thus immediately following Phase 3.140 Otherwise, no other site was specified as containing remains of this phase. And we know that the Shi’erqiao site covers a maximum area of 3 hectares,141 much smaller than even the Baodun sites. It is hard to imagine how it is possible for jiandi zhan and other vessels with pointed bottom to have a wider distribution at Chengdu than at Sanxingdui. Many other sites at Chengdu indeed had these vessels, but they date from later times than Phase 3. The continued popularity of vessel types with pointed bottom in those sites surely has no bearing on the discussion. On the other hand, there seems to be no scarcity of vessels with pointed bottom at Sanxingdui. In the report of the 1980–91 excavation, it is stated that the upper stratum of the 1982 excavation, which corresponds to Phase IV (or Phase 3), yielded a large number of such vessels; the number becomes more impressive when we realize that the total space excavated was only 100 sq m.142 It would be interesting to compare the distributional ratios per square meter of these vessels at similar locales in the two sites, but no quantitative analysis of this kind has been done, and appropriate data has not been published to enable such analyses. In addition to the challenges to Sun’s identification discussed above, it is important to point out that the pottery assemblage in Shi’erqiao’s thirteenth and twelfth strata exhibits a remarkable continuity with Sanxingdui in such vessel types as guan with tapered profile and small flat bottom, tripod he, high-stem dou, and ladle in the shape of a 140Sun Hua 1996, p. 132. For pottery types excavated at Fuqinxiaoqu see Wang Yi 1991, pp. 298–302. 141Jiang Zhanghua 1998b, p. 146. In Sun Hua 1996 (p. 123), the size is given as 1.5 hectares only. 142Kaogu xuebao 1987. 2, p. 249. 58 bird with a hooked beak. These types form the core of the assemblage of the Sanxingdui Culture. As mentioned before, the same vessels were present at Sanxingdui at Phase 3. In view of this continuity, the importance of jiandi zhan as a harbinger of a new culture was perhaps over-emphasized. Vessels with pointed bottom probably did not become the defining character of a culture until later. Lastly, as Sun himself pointed out, the Shi’erqiao Culture represented a transitional phase,143 which clearly indicates that this culture lacks a defining character unmistakably identifiable. For all these reasons, I propose to drop the appellation—the Shi’erqiao Culture, and propose to define the temporal dimension of the Sanxingdui Culture as starting from the Report Scheme’s Phase II and ending with Shi’erqiao’s twelfth stratum. The Sanxingdui Culture so delimited is based on a core pottery assemblage, but also takes into consideration of elite material culture such as represented in K1 and K2, and the city walls. Moreover, this re-definition expresses a hierarchical relationship in the settlement pattern of the Chengdu Plain, with the Sanxingdui site as the primary center and the Shi’erqiao site as a satellite of Sanxingdui.144 Other satellite sites include Guilinxiang, Shuiguanyin, Qingjiangcun, Yangzishan, Jinsha, and as far as Shaxi at the southwest edge of the Sichuan Basin, all previously mentioned in Section 2.1.6. However, as the settlement pattern of the Sanxingdui Culture in the Chengdu Plain is yet to be studied, it 143Sun Hua 2001, p. 483. 144Li Boqian 1997 also includes Shi’erqiao and related sites in his definition of the Sanxingdui Culture, but the temporal dimension of his definition is broadly extended to include remains in Chengdu as late as the early Spring and Autumn period. His scheme thus does not take into consideration the crucial change in the pottery inventory between the 12th and the 11th strata at the Shi’erqiao site as well as the marked changes in bronzes during the long span of time. 59 is impossible to comment on what exact relationship the Sanxingdui site had with other sites and how they maintained such relationship. The artifactual content of the Sanxingdui Culture has been defined largely in terms of pottery typology as discussed in Section 2.2.1 and in this section. What has been missing in the debate on the definition is the material culture of the elite class—preserved for us in enormous richness in the two pits K1 and K2. Describing and understanding this elite culture are the primary objective of this dissertation, to which the following chapters will now turn. 60 Chapter 3. Material Culture of the Sanxingdui Elite: K1 and K2 The two pits, K1 and K2, are extraordinary in every respect. They are much larger than any other pit whose size is known,145 and their contents make the Sanxingdui Culture one of the richest bronze cultures of ancient China. A large portion of the contents, particularly the bronze images, is unprecedented and puzzling. No similar find has been made anywhere else. The uniqueness of these two pits presents a significant challenge for interpretation. In the quest for understanding and explanation, scholars have engaged in heated debate ever since the discovery in 1986. By now, an enormous volume of published opinion is available on the subject, in quality varying from carefully constructed argument to fanciful story telling. 3.1 Research approaches The approaches applied in the present dissertation are two. One is close examination of the finds themselves, comparing the two pits and their contents. Reliance on internal evidence is particularly important given the uniqueness of the Sanxingdui finds. The other approach is comparison or contrast with other cultures. This is necessary not only for illuminating cultural connections but also for bringing out salient features that might otherwise remain less than obvious if looked at only internally. Comparison needs to be done with caution, however, and its usefulness must be judged case by case. As Bagley points out when commenting on comparison with Shang sacrifices at Anyang, the objects from the Sanxingdui pits, above all the bronze images, are so different from 145See description in Section 2.3.2. 61 anything found at Anyang that information about Anyang sacrifices seems almost irrelevant for explaining the sacrifice at Sanxingdui.146 Nevertheless, as he also points out, “Anyang ritual does have points of contact with Sanxingdui finds and it can at least suggest possibilities for interpreting the Sanxingdui finds. More importantly, perhaps, comparison with Anyang brings into sharper focus the contrast between the Shang civilization of metropolitan north China and that of the Chengdu plain.”147 A judicious balance of analogy and distinction is clearly called for. A textual approach that figures prominently in the Anyang archaeology is not available for the study of Sanxingdui because no inscriptions except for half a dozen marks on pottery vessels have been found at the site to shed light on the finds or on the culture expressed by them. The pottery marks have not been deciphered and are probably not writing.148 The only known writing native to Sichuan takes the form of short, as yet undeciphered inscriptions on a small number of bronze dagger-axes (ge) dating from the end of the fifth century BC or the beginning of the fourth. Contemporary with these inscriptions are about a hundred pictographic emblems found on bronze seals, weapons, and tools.149 On present evidence, the Chinese writing system was not adopted in Sichuan until after the Qin conquest late in the fourth century BC. Secondary textual sources offer no help either. Those sources may be divided into two groups: sources more or less contemporary with the two pits at Sanxingdui and sources compiled much later. The first group consists of a few oracle bone inscriptions at 146Bagley 1988 (p. 83), and Bagley 1990a (p. 60). 147Bagley 1990a, p. 60. 148Lin Xiang 1989 (p. 24), and Duan Yu 1991 (p. 22a). The assertions by Lin Xiang and Duan Yu that those marks represent writings are unfounded, and Duan Yuan’s attempt at deciphering them is guesswork without convincing corroborative evidence. 149Duan Yu 1991. 62 Anyang and at Zhouyuan, the heartland of the Zhou empire, which mention the name Shu.150 Shu was the name of the polity based in the Chengdu Plain during the Warring States period (5th–3rd century BC). The appearance of the same name in oracle texts of the late second millennium BC does raise the possibility that the region was so called at that time as well. However, those laconic texts do not specify the location of the Shu under discussion, and attempts to link the Shu in oracle texts with the Sanxingdui site are fraught with problems, a review of which has led Sage, a scholar who favors the link, to conclude, “these reservations about shu (or sou) for the present effectively must restrain any fully confident use of the oracle inscriptions as historical raw material.”151 In a study by Gu Jiegang, attempts to attach the name to anything in Sichuan earlier than about the eighth century have been convincingly refuted.152 In any case, currently available archaeological evidence does not justify connecting Sanxingdui with the Shu of later periods, as members of a single cultural continuum, which the labeling of the Sanxingdui site as a Shu polity would imply. 150For survey of the oracle inscriptions that mention Shu see Dong Qixiang 1983 (pp. 1– 7), Lin Xiang 1985, and Lin Xiang 1989. For a perceptive discussion of those inscriptions, their study by other scholars, and problems associated with their use see Sage 1992, pp. 31–4. 151Sage 1992, p. 34. However, the problems do not prevent Sage from adopting the stance, “Convenience and a fair measure of confidence dictate that this power [i.e. the entity at the Sanxingdui site] should properly be known as Shu.” (p. 34). 152Gu Jiegang 1981, pp. 1–71. This critical study is not cited in the section on the oracle bone inscriptions in Sage 1992. No study arguing for the attachment rivals Gu’s study in rigor and exhaustive analysis. References in later texts—the Zuo zhuan, whose composition cannot be earlier than the second half of the 5th century B.C., probably later, and the “Mu shi” chapter of Shang shu, of uncertain date—that have been widely accepted as justifying the application of the name to early archaeological cultures are discussed in Xu Zhongshu 1982, pp. 7–26. For dates of these two texts see Loewe 1993, pp. 69–71, and 377–80. 63 The other group of textual sources comprises myths and legends written down a thousand years or more later.153 The huge disparity in time and the nature of their contents make them inadmissible as evidence for historical events that might be connected with the Sanxingdui finds. At the present time, when the evidence from the field has yet to be digested thoroughly on its own terms, it is prudent not to confound that effort with textual sources of any kind, let alone sources as unreliable as these. Plentiful examples of such confusion already exist in the story telling rampant nowadays. Interpretations in this and following chapters are therefore to be arrived at through the two approaches mentioned above. This chapter will concentrate on the nature of the two pits, while chapters 4–7 will discuss in detail individual categories of artifacts. Relevant studies by other scholars will be considered and reviewed in the course of discussion. 3.2. Burial content and condition The two pits are located about 50–60 m south of the Sanxingdui terraces at Locus II (Figs. 2.3, 2.5). K1, the first pit to be discovered, is a rectangular shaft approached by shallow entrance ramps (Figs. 2.6, 3.1). The pit is 1.46–1.64 m deep, and measures 4.5– 4.64 m by 3.3–3.48 m at the top, and 4.01 m by 2.8 m at the bottom. Its corners point to the cardinal directions. Ramps converge on the pit from three sides. As the pounded earth fill in the ramps is the same as in the pit, it seems clear that the ramps and the pit were constructed at the same time. The ramps are now incomplete because parts of them were destroyed by later features and parts dug away by the brickyard workers at the time of the 153For a thorough and critical review of the texts see Gu Jiegang 1981, which finds them far from reliable historical documents. 64 discovery, but their structure can be clearly understood judging from what has survived. One central ramp enters the pit perpendicularly at the middle of its southeast wall; it measures 0.34–0.4 m deep, 3.85 m long, 1 m wide at the top and 0.94 m wide at the bottom. At the farthest end away from the pit, this ramp is connected with another ramp, which runs parallel with the pit’s southeast wall. This second ramp, much of it destroyed, is about the same width and depth as the first. It must have originally extended longer than the southeast wall, and at both ends perhaps connected with a ramp that ran parallel with the central ramp. Those two ramps were almost completely destroyed, but as they approached the pit, they clearly turned at a 90° angle toward it and enter it at the south corner of the southwest wall and the east corner of the northeast wall. In this way, three ramps enter the pit at three points, which basically concentrate on the southeast wall. K1 contained more than 400 artifacts of bronze, gold, stone, jade, amber, and pottery, along with 13 elephant tusks, 62 more or less intact cowry shells, and three cubic meters of crushed and burnt animal bones mixed with a small amount of wood and bamboo ash. The bones were in small bits, making measurement impossible; identification has consequently proven difficult. Examination of a small sampling of bones by a zoological expert indicated that medium-sized herbivores like pig, sheep, and goat account for the majority, with a few large-sized herbivores like cattle and buffalo. Though not included in the samples examined, elephant molars were found as well and were determined to belong to Asian elephants. Human bones such as skull fragments and limbs are suspected, but the fragments are too small to ascertain, and no toes and teeth, the most durable of human bones, could be found. At present, the presence of human remains appears a slim possibility. 65 The 178 bronzes include four vessels plus one lid (Figs. 4.6–4.10), 13 more or less human heads (Figs. 5.3–5.5, 5.10–5.12), 107 rings, and 44 triangular ge blades with serrated edges (Fig. 6.10).154 The objects of stone and jade, numbering about 200, include more than 60 tool-like implements and more than 70 blades, mostly forked blades, ge blades, and a curious hybrid of the two unknown anywhere else.155 There were four gold items, most notably a tube of sheet gold 142 cm long, originally the sheathing of a wooden staff. And there were nearly 40 pottery vessels of four types, the majority being 22 jiandi zhan (Fig. 2.13). The contents of the pit seem to have been deposited sequentially: small objects of stone, jade, and gold first; next the bronze heads and vessels; then the elephant tusks mixed with burnt bones, ash and some small bronze and pottery objects; and finally the pottery vessels and large blades of jade and stone. The objects were not carefully deposited, however; most of them lay randomly in the south corner and along the southeastern and southwestern walls of the pit. The bones and ash showed the similar concentration, forming a slope with the highest level in the south corner and along the southeast wall tapering gradually toward the north corner and the northwest wall (Fig. 3.1). Obviously the contents were pushed into the pit from the south corner and along the southeast wall, probably by way of the ramps, though the ramps may have served some other purpose as well given their rather intricate structure. After deposition, the pit and 154Judging from their flimsiness, the triangular ge blades must not have been functional weapons, but were probably used as some sort of ornament. 155The statistics given here for lithic artifacts in K1 and later for those in K2 lump “jades” and “stones” together, as there has not been comprehensive scientific identification of them. For a detailed comment on the lithic classification in the excavation report (Beijing 1999a) see note 374. 66 ramps were then filled with earth pounded hard; mixed with the earth are some clay cores from bronzes. Before they were deposited in the pit, the bones and most of the objects in K1 had been subjected to high temperatures. Some of the bronzes show traces of melting (e.g., Fig. 5.10), others had melted into unrecognizable lumps. Since the walls of the pit show no sign of fire or smoke, the burning must have occurred before the bones and objects were dumped into the pit. Some items, particularly those of jade and stone, had been broken as well as burned. It is clear that the breakage occurred before burial, because different parts of a single object were sometimes found scattered. The bones had also been broken, indeed pounded to small fragments with sharp edges. K2 is 30 m southeast of K1 (Figs. 2.3, 2.5, 2.7, 3.2–3.4). It lacks the ramps of K1, and is a narrower rectangle: 5.3 m long and 2.2–2.3 m wide at the top, 5 m long and 2– 2.1 m wide at the bottom. But it has the same depth (1.4–1.68 m) and the same orientation to the cardinal points. Far richer, the pit contained 67 elephant tusks, 4600 cowry shells, and hundreds of artifacts (the final excavation report—which numbers fragments individually—reaches a total of 1300). The 735 bronzes include a wealth of figurative items, ranging in size from miniature to monumental and in type from heads to whole figures to strange bronze trees alive with birds and other creatures. A human figure standing on a pedestal is life-sized, figure and pedestal together measuring 2.6 m high (Fig. 5.17a-e). The largest of the bronze trees is 4 m high (Fig. 5.30). At the other extreme of size, a kneeling figure holding a forked blade is only 5 cm high (Fig. 5.24). At least six of the 44 bronze human heads in the pit were originally covered in part with gold foil (Figs. 5.6, 5.16). There are also 23 bronze masks of varying sizes, 20 of which bear 67 the same facial features as the human heads while three others are animal masks with fantastic features (Figs. 5.26–5.29, 6.11). Taotie-like faces appear in nine flat plaques (Fig. 5.51). Aside from the monumental tree, several other much smaller trees survived in fragments (Figs. 5.40, 5.41), and there are a good many small bronze ornaments, some of which may originally have hung from the branches of the bronze trees (Figs. 5.33–5.39). For many of these objects K1 has nothing comparable, but among the less spectacular items there is considerable overlap between the two pits (Table 3). Both yielded bronze rings, vessels, and thin triangular ge blades, and among the 121 jade and stone artifacts (not including beads), most of the lithic types had been present in K1. Still, K2 had 500 small jade, stone, and ivory beads and tubes, types not represented in K1; and K1 had hybrids of forked blade and ge blade unknown in K2. K2 also did not contain pottery and bones. The deposit, despite breakage and burning, comprised three distinct layers, resulting from sequential filling of the pit. The bottom layer consisted of a wide variety of tree fragments, small bronze ornaments, and jade and stone items. A small amount of grass and wood ash as well as burned cowry shells were also found in this level. The middle layer was made up entirely of the larger bronzes, including the life-sized standing figure, human heads and masks, animal masks, large tree fragments, large appliqués, and vessels (Fig. 3.3). Strewn on top of these were the elephant tusks, forming the third layer (Figs. 2.7, 3.4). Within each layer the objects had been scattered at random, and many were found upside down or lying on the side at the time of excavation, including all the vessels. Small jade and stone objects and cowry shells originally contained in the vessels had thus spilled out. Objects were apparently thrown in from all sides of the pit, as they 68 filled up the pit in a rather even pattern. As with K1, the earth fill of the pit was pounded hard. If the fill contained any core material from bronze casting, the excavation reports do not mention it. Many of the small objects and cowry shells first dumped into the pit were found mixed with ash and show clear traces of scorching or burning. Some large bronzes like human heads, masks and vessels were burned as well. Other objects scorched or burned included elephant tusks and jade and stone objects with obvious cracks resulting from exposure to high temperature. Overall, a smaller fraction of the contents were subjected to this treatment than those in K1, but they show much evidence of deformation or breakage before burial. To take only one example, the life-sized figure was broken diagonally across the robe, and the two pieces were found in the pit lying in opposite directions: in Figure 3.3 the upper half is near the center of the pit, almost right side up, the head pointing toward the upper left corner; the lower half (the bottom of the robe, the feet, and the pedestal) is upside down, near the right wall of the pit, with the base of one of the bronze trees lying across it. There is no question, therefore, that the statue was broken up intentionally before the fragments were deposited in the pit. The majority of objects in K2 must have been similarly damaged before burial, though breakage in some cases may have resulted from the pounding of the earth fill. 3.3 Nature of the pits The content and burial condition of K1 and K2 provide essential clues as regards their nature. The preliminary excavation report of K1 identifies it as a burial of sacrificed offerings for good reasons: the fact that objects were burned and the presence of burned 69 bones and ash certainly suggest a burnt sacrifice; also, the fact that some bronze bronzes contained cowrie shells seems to indicate their use as ritual offerings.156 The excavators later drew further support from other factors such as the fact that many objects were deliberately broken.157 In the preliminary excavation report of K2, additional support is sought from a picture incised on a jade implement that apparently shows a ritual in progress (Fig. 5.42), and from the phenomenon that the objects were deposited in an orderly sequence by category. The report also emphasizes that the bronze heads, masks, 156Wenwu 1987.10, pp. 4, 13–4. The report also attempts to bolster this line of thinking by comparison with known sacrificial activities at Anyang, the Shang capital, in the Central Plain. For example, drawing comparison with sacrifices at Anyang where actual human beings were slaughtered, the report identifies the bronze heads as substitutes for actual human victim (ibid.). It further identifies the V-shaped neckline of the heads as symbolizing their execution (p. 14). This latter conclusion is untenable, however; as will be discussed in Section 5.1.5, the neckline is evidently related with the dress for human figures. As pointed out in Section 3.1, comparison with Anyang practices needs to be done with caution. For a summary comparison of Anyang and Sanxingdui sacrifices, which brings out major differences between the two cultures, see Bagley 1990a, pp. 61– 2. This comparison is a foundation on which some of the interpretations to be presented in Section 5.6 are based. For another comparison of sacrificial pits at Sanxingdui and at Anyang see Xu Ziqiang 1993. 157Chen & Chen 1987, p. 27. Not all the excavator’s arguments for identifying the pits as depositories of sacrifices will be listed here, for some of them are hard to verify and some turn out to be wrong. For example, they think that the bronzes were made on the spot specifically for inclusion in the pits; see ibid, Chen Xiandan 1989b (p. 36), and Chen Xiandan 1989c (pp. 13–4). They believe that they found bronze dross, fragments of clay cores, and molds for casting in the pits, which indicate to them that objects were made nearby the pits. The evidence is, however, not as strong as it sounds. There seems to be discrepancy in accounts regarding those remains. Bronze dross is mentioned in the preliminary report of K1 but not in that of K2; the final excavation report, Beijing 1999a, does not mention it at all for either pit. It is quite possible that the so-called bronze dross is actually melted lumps of bronze. No mold fragments are mentioned in any of the reports. The clay cores, mentioned also only for K1, do not necessarily indicate the presence of bronze foundry nearby, for bronzes were not necessarily cleared of clay cores after casting even if there would be no problem to do so. Objects like heads in particular might not have needed all the core material to be removed, for it would probably not interfere with their mounting. However, although the excavators’ reasoning on this point is questionable, it is possible that objects like bronze ge blades, too flimsy for practical use, were made specifically for sacrifice. 70 trees as well as the jades were objects of ritual offering.158 Chen Xiandan later also pointed out that many objects were smeared with red or blackish pigments.159 To these arguments it may be added that breaking and burning were conceivably ways of “killing” artifacts so that they could make the passage from this world to some supernatural realm; in other words, they were symbolic actions effecting metaphysical transformation. For these reasons, the identification of the pits as burials of sacrificed offerings seems at least a solid tentative conclusion. At the same time, it must be pointed out that the exact purpose(s) or recipient(s) of such sacrifices cannot be identified with certainty on present evidence. Internal comparison between K1 and K2 seems to suggest that they represent two performances of a same type of sacrificial action, in which offerings were first burned and broken and then buried. The two pits share essential similarities. They are close together, comparable in size, and oriented alike, which indicate a prescription for their construction. Given that the two pits are of different times, it seems that the area was also well maintained over time.160 Although the objects in the pits appeared chaotic at a first glance, they actually had a certain order in sequence of deposition. The existence of sequence betokens purposeful procedure rather than wanton destruction or discard, as noted in the preliminary report of K2. The contents of the two pits, though vastly 158Wenwu 1989.5, p. 19. 159Chen Xiandan 1989c, p. 12. 160Not all scholars agree that the two pits differ in time. The dating of K1 and K2 will be further addressed in Section 4.2. In addition to K1 and K2, another two pits have been found in the Sanxingdui area, as mentioned in Section 2.3.2. Located just a short distance to the southeast at Locus II (Fig. 2.5), they contained a few bronzes, jade and stone artifacts, and burned bones and ash. Though far more modest than K1 and K2, the pits seem to demonstrate an essential compatibility in burial content and condition, and were probably sacrificial pits as well. Their presence reinforces the impression that the Sanxingdui area was a ritual ground where sacrifices were performed. 71 different in richness, are generally comparable in the range of bronzes and jades, and in other valuables such as cowry shells and elephant tusks. Most of the objects had been burned and/or battered before deposition. And both pits were sealed with a filling of pounded earth. The two pits nevertheless differ in several ways, some of which have been previously noted in Section 3.2. To sum up, K2 was vastly richer than K1, particularly in bronze images: heads, masks, trees, birds, even models of what might be temples or altars (Fig. 5.43). By contrast, K1 had more stone implements, and contained pottery whereas K2 did not. The objects were more often burnt in K1, more often broken in K2. K1 had a large quantity of bones and ash but K2 had only a small quantity of ash. K1 had ramps, K2 did not.161 The disparity in material between the two pits might be due to the lapse of a few decades between the two pits, a lapse that might have seen much change in the city’s wealth and in its craft industries. Falkenhausen proposes another explanation, believing that the higher level of material consumption in K2 possibly reflected patronage by higher-ranking elements in the society.162 Unless the two pits are deemed as dating from the same time, however, a possibility that contradicts the evidence from bronze vessels that will be discussed in Section 4.2, the disparity in material wealth in a diachronic comparison does not necessarily indicate difference in social hierarchy. Wealth is famously unpredictable and can easily change over time. Besides, the content of K1 161Falkenhausen 2002 (pp. 68–74) also enumerates in detail similarities as well as differences between the two pits and their contents. 162Falkenhausen 2002, p. 70. 72 included a tube of sheet gold 1.42 m long, originally the sheathing of a wooden staff, which undoubtedly belonged to the regalia of a person of the highest status.163 Furthermore, in Chapter 5 it will be argued that the bronze trees in K2 replaced actual trees consumed in K1; and that many other bronze images likes heads and masks were copies of wooden images that perished in the burning. In other words, the content of K1 may have very well represented only a portion of a much larger assemblage that was destroyed and burned, and the differences between the two pits may have been smaller than they seem today. The two pits may have had the similar types of offerings, but fabricated in different materials, resulting in a large disparity in archaeological record. In studying the contents of these two pits, we must continually be alert to what may have been lost. As regards the other differences between the two pits, they seem less than essential, too. That the contents were more often burned in K1 and more often broken in K2 is a difference in degree, which does not alter the essential fact that they were burned and battered; for symbolic purposes the treatment may have been the same. The disappearance of ramps in K2 was arguably the most important change, but while it resulted in a different distribution of the contents in the pits, it does not seem to have altered other features of the sacrifice as enumerated before. 163Falkenhausen also states that bronze heads covered with gold foil are seen only in K2 (ibid., p. 69). This statement is incorrect, however, as K1 did yield a gold foil of a face (K1:282), which had surely been applied to a head though it is not clear whether the head was of bronze or of another material (see Beijing 1999a, p. 61 fig. 34.2, p. 62 pl. 15.2; Tokyo 1998, no. 113). 73 The differences between the two pits, therefore, do not necessarily contradict the conclusion drawn from their similarities, that they were two performances of a same type of sacrificial action, albeit with changes over time.164 Not all the scholars agree that K1 and K2 were sacrificial pits, however. Xu Chaolong, for example, raises a series of challenges to the identification,165 but all of them may be answered with solid reasons. (1) Xu questions whether the production force at Sanxingdui was capable of sustaining such consumptive sacrifices. This question was raised on the wrong premise that such sacrifices were conducted on a regular basis, which needed not to be the case. The two pits undoubtedly represent extraordinary, perhaps catastrophic, occasions that called for sacrifices on such large scales. There is no reason why the wealth could not be devoted to religious activities of the highest importance. (2) Xu feels that it would be ridiculous to destroy and bury precious ritual offerings whose production likely cost huge resources over a long period of time. Whereas Xu viewed breaking and burning as negative act of destruction, probably out of modern sensibility, 164In view of the differences between the two pits, Falkenhausen feels that “it does not seem altogether certain that they constitute two instances of the same type of deposit, or manifestations of exactly the same kind of ritual activity” (Falkenhausen 2002, p. 68). On the other hand, in view of the similarities, he feels that “they may have been related phenomena—variation on the same theme” (Falkenhausen 2002, p. 70). He appears undecided on the issue. It is important to point out that the same ritual activity can have different manifestations, and different manifestation does not necessarily betoken difference in the nature of the ritual activity. To draw an analogy, funerals can have a myriad of different manifestations, or performances. The matter really boils down to what constitutes the essential qualities of a given ritual activity. Differences must be carefully analyzed and, again, the possibility that the sacrifice resulting in the deposit of K1 involved objects of perishable material needs to be considered. 165Xu Chaolong 1992a, pp. 33–5. Xu’s challenges were aimed at refuting the studies in Chen Xiandan 1989c. While I agree to Chen’s general identification of the pits as sacrificial deposits, many of his detailed explanations, which Xu rightfully questions, are indeed fraught with problems rising from uncritical use of textual sources or unrestrained speculation. Uncritical use of textual sources is a problem with Xu as well, however, and actually with many scholars working on the subject. 74 the ancient people at Sanxingdui needed not to feel the same if the purpose of such acts was to symbolically transform the objects for acceptance by a higher supernatural authority. (3) Xu cites the absence of similar practice at Anyang and the lack of textual record of such sacrifice as evidence negating the nature of the pits as burials of sacrificial offerings. Here he seems to have neglected basic difference between the Sanxingdui Culture and that of the Central Plain. His negative evidence has no bearing on the Sanxingdui ritual practice. (4) and (5) Xu questions the reason why the human heads and so forth, apparently images of the ruling class, would be sacrificed and used as surrogates of real human sacrifices. A possible answer to this question will be presented in Section 5.6. (6) Judging from the probability that many of the bronzes were used for a long time before burial, Xu questions the assertion that the bronzes were made especially for sacrifices. This is indeed a valid challenge; the weakness of that line of reasoning has been discussed in note 157 as well. However, for whatever purpose were the bronzes made and no matter how long had they been used prior to the event, they surely could be used as ritual offerings. Zhang Xiaoma raises some of the same challenges as Xu Chaolong, but also an additional objection on the ground that no related architectures such as temples and altars or arena for sacrifices have been found in the vicinity of the two pits.166 It is obviously possible that traces of such structures may have disappeared beyond recognition over the period of more than two thousand years or they are yet to be discovered in future fieldwork.167 Furthermore, we should not presume that the acts resulting in the two burials necessarily took place in architectural settings, nor should we presume that the 166Zhang Xiaoma 1996. 167Moreover, argument from absence of evidence is generally weak. 75 acts of burning and breaking must have taken place in the same locale as the burials. As pointed out in Section 3.2, the burning clearly did not occur inside the pits. The discussion in Chapter 5 on bronze images will make it clear that many components of objects consumed in the acts were not buried in the two pits, further raising the possibility that burning and breaking may have occurred elsewhere from the burials.168 Quite a number of alternative interpretations have been proposed, but they are problematic for various reasons.169 Lin Xiang suggests that the deposits represent shamanistic acts of abandoning ritual implements that no longer worked magic,170 but why the inclusion of bones and valuables such as elephants tusks and cowry shells, which do not fall into his category of ritual implements? Some scholars identify the pits as burials of tomb goods that accompany tombs not yet located, but careful search of the vicinity has not produced any.171 Zhang Minghua proposes that the pits are cremation burials in which, naturally enough, little trace of the deceased was found.172 Nevertheless, although the small size of the bone fragments in K1 makes it difficult to rule out the presence of human remains, not even the most durable of human bones—toes and teeth 168Sun Hua also argues against identifying the pits as sacrificial deposits by citing later texts on sacrifices and Anyang practices; see Sun Hua 1993a (pp. 3–4), and Sun Hua 1993c (pp. 73–4). Those evidences, however, obviously have no direct bearing on the Sanxingdui practice. 169Some of the following hypotheses have been refuted in Wenwu 1989.5 (p. 19), Bagley 1990a (p. 60), and Chen Xiandan 1997 (pp. 10–2). The reasons are not always the same as those of mine. 170Lin Xiang 1987, p. 81. 171This identification is cited in Wenwu 1989.5, p. 19. 172Zhang Minghua 1989. 76 —were found, nor was any found in K2. Barnard regards the pits as hoards buried with the intention of recovery at a later date,173 but why the breakage and burning? Like Barnard, Qian Yuzhi views the pits as hoards of ritual paraphernalia and treasures hurriedly buried for safekeeping by the defenders of the Sanxingdui city at the imminent sacking of the city by invaders, but to account for their breaking and burning, he imagines that they were salvaged from ravages of battles and then buried.174 Xu Chaolong argues that the deposits indeed resulted from a violent change of dynasties, with the new political power destroying and discarding objects of the former dynasty.175 Sun Hua proposes two possibilities along the similar line, one of which is, if the two pits date from the same time, the deposits were either loot discarded by invaders or ritual objects destroyed and abandoned by the vanquished.176 A problem common to these proposals is obvious: they all depend on the premise that the two pits date from the same time, which is highly unlikely. And why would so many animals have to be killed, burned, and laboriously pounded into bits? For the invader or conqueror theories in particular, why was the trash so neatly buried? Why were valuables such as ivory and recyclable bronze discarded? Sun Hua himself expresses a similar doubt by wondering why the invaders would not carry off the war loot. Sun Hua’s second theory is that, if the pits date from different times, they represent discards of temple furnishings occasioned by the change of reigns in which the new ruler deemed inauspicious the possessions of the former ruler.177 The questions 173Barnard 1990. 174Qian Yuzhi 1992, p. 53. 175Xu Chaolong 1992b, p. 46. 176Sun Hua 1993b (pp. 4–5), and Sun Hua 1993c (pp. 75–6). 177Ibid. 77 raised above regarding the slaughtering and pounding of animals and the neat burial of objects are equally applicable to this theory. Moreover, we may remind ourselves that religious activity could not be carried out in an economic vacuum, but is always intimately tied to economic resources, particularly for periodic activity. If we agree that K1 and K2 are about a few decades apart in time, the periodic frequency would be remarkably high, and economic consideration would presumably figure quite importantly. As mentioned earlier, Xu Chaolong raises one of his objections against the identification of sacrificial pits by questioning the sustainability of such activities on a regular basis. His question would be relevant if such destructions were indeed performed regularly. Falkenhausen has most recently proposed a related possibility, that the local inhabitants had a custom stipulating periodic ritual destruction and rebuilding of temples.178 His interpretation is in essence the same as identifying the pits as sacrificial burials by emphasizing on their nature as ritual deposits, but he attempts to imagine a particular ritual. For this particular suggestion, the same question regarding the economic sustainability may be raised, and both his and Sun’s interpretations would require more similar pits to be found in future fieldwork.179 The preceding discussion has summarized the other identifications so far known to have been proposed for the two pits, none of which seems more plausible than the vague identification of K1 and K2 as sacrificial pits. 178Falkenhausen 2002, p. 85. 179In that regard, perhaps some of the other pits could be argued for as depositories of similar disposals, but the huge difference in wealth as well as in size would pose a problem. 78 Chapter 4. Bronzes at Sanxingdui (I): Sporadic Finds and Vessels Finds of bronze at the Sanxingdui site show a dramatic imbalance between the riches of K1 and K2 and the comparative poverty in the rest of the site. Probably due in part to insufficient fieldwork, this imbalance nevertheless highlights the extraordinary nature of the two sacrifices. On the other hand, bronzes other than those from K1 and K2 are important, despite their scarcity in quantity, for shedding light on the cultural connections of the site and on the origin of its bronze industry. This chapter will therefore begin with them, and then move on to discuss the bronze vessels from the two pits. Chapter 5 will concentrate on the images from the two pits, and Chapter 6 will explore the fabrication technology responsible for the making of the images. 4.1 Bronzes not from K1 and K2 In the known record, the first bronze object found at the Sanxingdui site appears to be a tiger-shaped plaque unearthed in 1984 at the Rensheng village by the bank of the Yazihe River (Fig. 4.1a).180 Another similar but larger plaque was found in the same area (Fig. 4.1b).181 Both show in silhouette a stealthily walking predator with muscular legs, powerful claws and gaping jaws, but the creature of the smaller plaque has more elaborate teeth, snout, and ears or horns. Both are inlaid with small chips of turquoise. 180Beijing 1994a (pl. 66), Chen De’an 2000 (no. 57). 181The date of this find is not clear. Xiao et al. 2001 (p. 24) puts it generally among the finds made since 1981. 79 The back of the larger plaque is bare of decoration but carries four small loops meant for attaching it to something else.182 Besides these two animal-shaped plaques, three rectangular plaques were unearthed in 1987 from a pit at Cangbaobao, and another similarly shaped plaque was found in a pit at Gaopian, 10 km to the northwest of the Sanxingdui site (Figs. 4.2, 4.3a).183 All four are similar in size, but one of the Cangbaobao plaques is a solid plain piece (Fig. 4.2c), while the others have geometric patterns either in openwork (Fig. 4.2a) or in turquoise inlay (Figs. 4.2b, 4.3a). The patterned ones also have four small loops for attachment. The plaque in Figure 4.2b shows impressions of finely woven fabric on the front and of woven bamboo on the back. The tiger-shaped and the rectangular plaques are obviously related by their similar attachment device and more importantly by their turquoise inlay. These traits further connect them, particularly the rectangular ones, to plaques excavated at Erlitou in the Central Plain (Fig. 4.3b).184 The rectangular plaques from the two sites are essentially of the same type in shape and technology. It is quite possible that this type of inlaid plaque 182I have not had a chance to examine the back of the smaller plaque because the plaque apparently can no longer be found. A third tiger-shaped plaque is now known at the Jinsha site in Chengdu (Fig. 4.1c), the recessed grooves on whose body were clearly meant for inlays. 183Beijing 1998a, and Ao & Wang 1980. The pits are listed in Section 2.3.2. 184A total of three have so far been found at Erlitou; all are illustrated in Beijing 1996b, pls. 20–22, but upside down. The excavation reports are Kaogu 1984.1 (a photograph showing this plaque in situ appears in Beijing 1993, p. 121), Kaogu 1986.4, and Kaogu 1992.4. Another plaque, very similar to the Erlitou one published in Kaogu 1986.4, was unearthed at Tianshui, Gansu province (see Zhang Tian’en 2002). Moreover, there are quite a few plaques in museum and private collections similar to those excavated ones (see Li Xueqin 1991 for a survey of them). Zhao Dianzeng was first to point out similarity between the rectangular plaques at Sanxingdui and Erlitou (Zhao Dianzeng 1993, p. 82). Li Xueqin 1997b provides some details of the connection and difference in style. 80 originated at Erlitou since a bronze foundry capable of manufacturing such items had come into existence there no later than 1500 BC, while at Sanxingdui we have no evidence apart from the plaques themselves of bronze casting at this early date.185 However, whether the Sanxingdui plaques were actual imports from Erlitou or later imitations is harder to decide.186 On the one hand, it is easy to imagine these attractive and conveniently small objects circulating in networks of trade or exchange and adopted for different uses in different cultural contexts.187 On the other hand, the Sanxingdui plaques display some marked differences: they are all a little slenderer in proportions and their surface designs are quite different from the Erlitou ones, which evoke some kind of animal face. The plaque in Figure 4.2b has two rows of double circles, perhaps vestiges of animal faces.188 Of course it is impossible to know if Erlitou only produced plaques decorated with animal faces but never ones like those at Sanxingdui. But significantly, unprovenanced plaques in museum and private collections as well as the one excavated in Gansu Tianshui seem to bear closer affinity to the Erlitou plaques. Against this backdrop, 185For Erlitou metallurgy see Su et al. 1995 (pp. 95–9), and Bagley 1999 (pp. 141–2). Jenny So suggests that the plaques originated at Sanxingdui, and that the Erlitou examples were imported from Sanxingdui (Bagley 2001, p. 159 note 25). On present evidence, at least, it seems easier to believe the opposite. 186Perhaps it is not a pure coincidence that both tiger-shaped plaques were found in the Rensheng village, where Phase I tombs were later exposed. 187The plaques found at Erlitou, by contrast with those at Sanxingdui, came from tombs. They were found at the chest of the deceased on two occasions, and by the side of the body on the third, which suggest that they were used as ornaments attached on the chest. For surveys of the Erlitou plaques see Wang Jinqiu 2001, and Ye & Li 2001. The impressions on the Sanxingdui plaque in Figure 4.2b suggest that it was wrapped in cloth and placed with a bamboo artifact, but its exact function is not known. The context of the Gansu Tianshui plaque is not known. 188Du Jinpeng was the first to point out that the Sanxingdui plaques are later derivatives of Erlitou originals by emphasizing such relationships (Du Jinpeng 1995). 81 the Sanxingdui ones are quite distinctive, which seems to strengthen the possibility of their being local products. In either case, the connection with Erlitou is clear. Like the rectangular plaques, the tiger plaques at Sanxingdui might be imports or local products, though their uniqueness seems to favor the latter possibility.189 Their inlaid patterns are rather unusual, not obviously related to the markings of a real tiger as seen on a gold appliqué of a tiger from K1 (Fig. 7.24), nor to the markings commonly given to tigers on bronze vessels, e.g. a zun from K1 (Fig. 4.10). The date of all these plaques is problematic, thanks to the uncertainty as to their place of fabrication and to the lack of clear archaeological context. Like K1 and K2, the Cangbaobao pit was found by brickyard workers digging clay. By the time archaeologists arrived on the scene the stratigraphy of the area had been too badly disturbed to allow relating the pit to other parts of the Sanxingdui site. The Gaopian pit apparently did not have clear stratigraphical relationship to the site either, and the tiger plaques were stray finds.190 Nevertheless, the jades of the Gaopian pit suggest a date of burial no later than K1: its jade spearhead compares closely with stone examples from K1 (Figs. 4.4a–c), and its jade axe has notched edges similar to those of a jade hatchet-shaped implement in K1 as well as of Erlitou examples (Fig. 4.5).191 Neither the spearheads nor the notched edges occurred in 189So far the only known example outside the Sichuan Basin comparable to these plaques is a fragment of a tiger-shaped foil excavated from tomb M41 at Xi’an Laoniupo, which however does not have turquoise inlay (see Liu Shi’e 2001, p. 297, p. 298 fig. 258, pl. 155.1). Three-dimensional tigers with turquoise inlay were excavated from Fu Hao’s tomb at Anyang (see Beijing 1984a, p. 111, pl. 76). 190The tiger plaque from Jinsha was among the finds bulldozed out by construction workers and thus lost its stratigraphic context. See Section 10.2 for the Jinsha discovery. 191For a color illustration of the Erlitou axe in Figure 4.5c see Yang 1999, p. 149 no. 40. The point about the similarity in notched edges with Erlitou jades is made by Beijing 1998a (p. 89), which also argues that the Sanxingdui bronze plaques betray a primitiveness in casting that is not seen among the bronzes of K1 and K2. This second 82 K2. Given the possibility that the rectangular plaques might be imports from Erlitou, it seems prudent to assign all the plaques to the time range from 1500 to 1200 BC, in other words, to Phase III. At any rate, these plaques seem to be the oldest bronzes yet found at Sanxingdui. Apart from the plaques, few early bronzes have yet turned up in the neighborhood of Sanxingdui.192 Then, however, we suddenly have overwhelming evidence from K1 and K2. The majority of them, the images, are surely the products of a local industry, for nothing like them has ever been seen anywhere else. But the vessels find close comparisons in other regions. They not only point to cultural connections, but also serve as the most important criterion for dating the pits. The following section will explore these vessels in detail. 4.2 Vessels from K1 and K2 K1 and K2 contained altogether about 25 bronze vessels, but all were damaged and only 10 were in a condition to be restored.193 Even some of the restored vessels were so badly broken or melted that restoration required the addition of large missing portions. The zun in Figure 4.10, for instance, is two-thirds modern material.194 The shapes of the damaged vessels are in most cases recognizable, however; in striking contrast to the point is questionable, for the serrated ge blades in both pits could not have been more difficult to cast than the plaques, and they are quite poor castings. 192Elsewhere in the Chengdu Plain, a burial at the Xinfan Shuiguanyin site, about 25 km northwest of Chengdu, contained three bronze ge blades said to resemble Erligang types (Kaogu 1959.8, p. 408) (Fig. 8.2a). The identification, if correct, would imply a date c. 1500–1300 BC. More recently, an Erligang-type bronze ge blade was excavated at the Jinsha site (Fig. 8.2b), see discussion in Section 10.2. 193The ten restored vessels comprise seven zun and three lei. See the lists of K1 and K2 vessels below. 194Chen De’an, personal communication, October 1999. 83 bronze images, the vessels all belong to types long familiar elsewhere and their decorative styles show no trace of local peculiarity. This grants a large measure of confidence for dating the vessels by typological and stylistic comparison against a wellestablished stylistic sequence and with particular counterparts in other regions.195 The dates of the bronzes in turn help to date the pits. 4.2.1 Vessels from K1 K1 yielded four vessels falling into three types, in addition to a fragmentary lid. They display styles widely distributed during the Upper Erligang period and the transition period (14th to 13th century BC), and may therefore be dated accordingly.196 The vessels are listed below in the order of their likely dates. (1) Pan, K1:53, two fragments, 14th century BC (Fig. 4.6) (2) Lid, K1:135, fragmentary, 14th–13th century BC (Fig. 4.7) (3) Zun, K1:163/K1:59, two fragments, early 13th century BC (Fig. 4.8) (4) Pou, K1:130, two fragments, late 13th century BC (Fig. 4.9) (5) Zun, K1:158/258, restored, late 13th century BC (Fig. 4.10) 195For stylistic sequence of bronze vessels from Style I through Style V see Bagley 1999, pp. 146–55. Detailed analysis of individual vessel types and decorative styles may be found in Bagley 1987. On general discussion of the dates and stylistic affinity of Sanxingdui bronzes see Bagley 1988 and Bagley 1990a. Beijing 1999a (pp. 430–1), Jiang & Li 2002 (pp. 131–40), and Nanba Junko 2002 (pp. 142–6) also lists vessels from elsewhere showing stylistic similarities with the Sanxingdui vessels, but some of the conclusions drawn from such comparisons are not necessarily acceptable. 196The bronze styles of the Erligang and transition periods have been studied in depth by Bagley, who argues that their wide distribution was probably caused by an expansion of the Erligang Culture, and that regional bronze cultures eventually sprang up on the Erligang legacy; see Bagley 1977, Fong 1980 (pp. 111–7), Bagley 1999 (pp. 168–80). For other studies of the transition period and/or its bronzes see Nanba Junko 1990, Chen Fangmei 1991, and Tang Jigen 1999. 84 The pan K1:53 is probably the earliest in date among K1 bronzes (Fig. 4.6). As a vessel type, the pan seems to have first appeared around the middle of the Erligang period (1500–1300 BC). Among the few early examples, the closest in shape and decoration to the present one is a pan from Chenggu (1981CHLTT:6) in Shaanxi province, both featuring a double row of circular dots.197 Vessels similarly decorated include two lei (PYWH6:15, and PYWM11:34) from Panlongcheng in Hubei province, and another from Gaocheng in Hebei province.198 Lids are not often encountered in finds of bronzes. The lid K1:135 is decorated in low relief with eyes set among diagonally extended filaments and curls often seen on bronzes of the Erligang period and the transition period (13th century) (Fig. 4.7).199 The earliest version of this design among excavated bronzes seems to appear on another pan from Chenggu (1980CHLTT:6) and on a yan at Panlongcheng (PLZM2:45).200 The design on the K1 lid is more elaborate. The fragments K1:163 and K1:59 in Figure 4.8, probably of a same zun, are quite similar to a zun from Chenggu (1974CHBSTT:2), both rather crudely fabricated and covered with a mesh of spiraling patterns in thread relief.201 With a large flaring mouth, short middle section, and high foot, the Chenggu zun appears typologically more advanced than Erligang zun vessels. 197The Chenggu pan was reported in Wang Shouzhi 1988.6 (p. 7, pl. 1.3), and Zhao Congcang 2006 (pp. 32–3, p. 33 fig. 34.2, pl. 97). 198For the two Panlongcheng lei vessel see Beijing 2001e, p. 286 fig. 209, p. 287 fig. 210. For the Gaocheng lei see Beijing 1985a (p. 131 fig. 78.2), and Beijing 1996b (pl. 134). A lei in the Palace Museum, Beijing displays a Style II taotie amidst rows of dots; for a color illustration see Beijing 1999b, no. 5. 199A handled cup in the Shanghai Museum bears both this design and the double row of circles like that on the pan (see Beijing 1996b, pl. 162). 200For the Chenggu pan see Zhao Congcang 2006 (p. 3, p. 10 fig. 12, pl. 96). For the Panlongcheng yan see Beijing 2001e, pp. 173–4, p. 174 fig. 115, color pl. 16, pl. 49.2. 201The Chenggu zun was first reported in Tang et al. 1980, p. 215, pl. 2.2. See also Beijing 1979 (p. 96 pl. 102), and Zhao Congcang 2006 (p. 77, p. 78 fig. 76, pl. 29). 85 Pou K1:130 in Figure 4.9 carries decoration similar to that of a pou from Gaocheng, which shows an early effort at a Style IV image–ground distinction.202 However, one of the two fragments, a piece from the vessel’s shoulder and middle section, appears also to have high relief: the taotie’s horn rises slightly from the background. The Sanxingdui vessel therefore seems to be somewhat more advanced than the Gaocheng pou. Though covered in familiar texturing patterns of Style III, zun K1:128/258 is exceptional for its motifs (Fig. 4.10). The middle section is divided into three compartments by vertical crenellated flanges, on top of which hover horned dragons whose heads are fully sculptural, their prominence serving to accentuate the presence of the flanges. The dragons’ bodies swell gently out of the background pattern on the vessel shoulder into high relief and appear in a zigzagging fashion, clearly suggesting a sense of motion. The middle section contains, in each compartment, a symmetrical design in which a two-bodied tiger apparently holds in its mouth the head of a human figure whose arms and legs are spread out on both sides. Like the heads of the dragons, the tiger heads project powerfully into three dimensions. Their twin bodies, spread horizontally across the compartment, imitate the symmetry of a taotie. Distinctly different from contemporary taotie decoration, this design is known from only one other vessel, a larger zun found at Funan in Anhui province, more than a thousand kilometers to the east (Fig. 4.11). On the Funan zun the human face beneath the tiger’s jaws can be clearly seen (Fig. 4.11b). Curiously, while the two vessels are very close in design, they are poles apart in 202The Gaocheng pou was reported in Kaogu 1973.1, p. 28 fig. 8.4, pl. 9.3. It is among the evidence used by Bagley to argue that Style IV decoration first appeared during the transition period (Bagley 1987, p. 316). 86 quality. The zun from Funan is executed with extraordinary finesse, all the lines sharply cut and the creatures appearing alive. It is among the finest castings of its time or any other. The Sanxingdui zun by contrast is quite crude; its quality does not approach the draftsmanship of the Funan vessel. Nor does its design include the taotie faces that on the Funan zun are centered on the flanges (Fig. 4.11a).203 Judging from the simplified and crude quality of the Sanxingdui tiger-human zun, some scholars feel that the vessel should be somewhat later than the Funan zun, perhaps a copy of the latter or derived from a same origin. An associated assumption is that the Sanxingdui zun was locally produced.204 On present evidence, however, no solid conclusion can be drawn on either point. Difference of time is not the only possible explanation for difference in casting quality. But the vessels do suggest an extensive network of exchange and communication, in which Sanxingdui must have participated. The other vessels from K1 further strengthen this impression. Against this backdrop of widely shared bronze styles, it is hard to decide if any of the Sanxingdui bronze vessels was fabricated locally.205 Their limited quantity and variety in contrast with the wealth of bronze images seem to suggest imports, and additional clues will be seen in the following analysis of the bronze vessels from K2. 203The zun from Funan is reported in Ge Jieping 1959, front cover and its inside page. For discussion see Shi Zhilian 1972, Bagley 1987 (pp. 23, 34–5). For comparison with the Sanxingdui zun see Li Xueqin 1989 (pp. 229–30), Bagley 1990a (pp. 64–5), Li Xueqin 1993 (p. 77), and Shi Jinsong 1998a (pp. 56–7). The theme of tiger–human combination is also expressed in other forms; for surveys see Xu Lianggao 1991 and Shi Jinsong 1998a. 204Li Xueqin 1989, Li Xueqin 1993 (p. 77), Shi Jinsong 1998a (p.58), Shi Jinsong 1998b (p. 51). 205Bagley was the first to suggest that the vessels from K1 may not be local castings in view of their styles (Bagley 1988, p. 80). 87 At any rate, the Sanxingdui zun must be close in time with the Funan zun, and their Style III high relief indicates that both likely date from the latter half of the transition period.206 This date for the Sanxingdui zun is supported by other vessels in K1, which date from a generally compatible range of time. This in turn helps date the pit to the same time, i.e. toward the end of the 13th century BC. 4.2.2 Vessels from K2 K2 contained several times as many bronze vessels as K1, but they are all zun or lei.207 In addition, there are a lid (Fig. 4.19), a lid knob, and a bird that can be confidently identified as belonging to a vessel lid (Fig. 4.27). Again these probably belonged to lei.208 As a vessel type, the zun originated in the Erligang period. Its basic structure had not changed at the time of K1 and K2: flaring mouth, sloping shoulder, tapering middle section, and ring foot. The bottom of the vessel is at the height of the bottom of the middle section. The middle section and foot are always divided into three compartments, each compartment of the middle section accommodating a taotie. The lei is typologically related to the zun, and it too originated in the Erligang period, differing only in having truncated mouth. In the Central Plain, however, this type of lei hardly survived beyond 206For development of high-relief Style III see Bagley 1987, pp. 22–3. 207A fragment, K2(2):117, apparently part of a vessel’s shoulder, is tentatively identified as belonging to a hu in Beijing 1999a, p. 265, p. 277 fig. 152, p. 279 rubbing no. 30, and p. 281 pl. 101.4. However, it could very well be the shoulder of a lei. 208For illustrations of the lid knob see Beijing 1999a, p. 281 pl. 101. The bird is listed in Beijing 1999a under “miscellaneous ornaments”. 88 the transition period.209 In the Changjiang region, it was carried on, but with significant changes. For ease of discussion, all the vessels from K2 are listed below except for some small fragments. (1) Zun, K2(2):112, restored, Style III, early 13th century BC (Fig. 4.12) (2) Zun, K2(2):135, fragmentary, Style III, early 13th century BC (Fig. 4.13) (3) Zun, K2(2):109, fragmentary, high-relief Style III, late 13th century BC (Fig. 4.14) (4) Zun, K2(2):79, restored, Style IV, late 13th century BC (Fig. 4.15) (5) Lei, K2(2):70, restored, Style IV, late 13th century BC (Fig. 4.16)210 (6) Lei, K2(2):88, restored, Style IV, late 13th century BC (Fig. 4.17) (7) Fang lei, K2(3):205/K2(3):205-1, Style IV, two fragments, late 13th century BC (Fig. 4.18) (8) Lid, K2(2):32, fragmentary, Style IV, late 13th century BC (Fig. 4.19) (9) Zun, K2(2):127, restored, Style V(a), early 12th century BC (Fig. 4.20) (10) Zun, K2(2):129, restored, Style V(a), early 12th century BC (Fig. 4.21) (11) Zun, K2(2):146, restored, Style V(a), early 12th century BC (Fig. 4.22) (12) Zun, K2(2):151, restored, Style V(a), early 12th century BC (Fig. 4.23) (13) Lei, K2(2):159, restored, Style V(a), early 12th century BC (Fig. 4.24) 209Two Style III examples unearthed at Anyang are among the latest known lei of this type in the Central Plain; for color illustrations see Beijing 1997b, pls. 77 and 78. For a discussion of the lei form see Bagley 1987, no. 1, pp. 145–6. 210Beijing 1999a (p. 270 pl. 97) claims to publish the same vessel, but it is a wrong photo showing another view of lei K2(2):88 published in pl. 98. 89 (14) Lei, K2(2):39/K2(2):39-1, fragments, Style V(a), early 12th century BC (Fig. 4.25) (15) Lei, K2(2):103/K2(2):103-1/K2(2):103-2, fragments, Style V(a), early 12th century BC (Fig. 4.26) (16) Bird for vessel lid, K2(3):193-1, fragmentary, 13th–12th century BC (Fig. 4.27) (17) Fragment, probably of a vessel lid, K2(3):23, 13th–12th century BC (Fig. 4.28) These vessels may further be grouped into two categories. The first category comprises three early zun, (1)–(3), which have Style III decoration of a kind widespread during the transition period. The rest of the list forms the second group. They are decorated in Styles IV and V(a) and display traits that have been identified as characteristics of the middle Changjiang region, in other words, of southern Hubei and northern Hunan.211 4.2.2.1 Vessels of Style III Among the three zun of this category, zun K2(2):112 in Figure 4.12 is probably the earliest. Its shape is short, its mouth flares modestly, and it does not yet have flanges, thus giving a rather restrained appearance. The surface decoration consists of texturing patterns of densely packed quills, volutes, and spirals amid which three pairs of hooked eyes stand out to indicate the presence of three taotie encircling the vessel surface. The 211On the rise of bronze cultures in the middle Changjiang region and characteristics of local bronzes see Bagley 1987 (pp. 32–6, 267–75, 539–51), Bagley 1992, and Bagley 1999 (pp. 208–19). See also Xu 1998, and Nanba Junko 2002 which basically follows Bagley’s studies. For the development of Style IV see Bagley 1987, pp. 23–4. For the development of Style V(a) see ibid., p. 55 note 104, no. 36 (pp. 249–51), no. 43 (pp. 267– 75), no. 78 (pp. 437–9). 90 positions of the taotie are emphasized by slightly raised median ridges between the eyes, and directly above them, bovine heads on the shoulder that intrude into the top register of the middle section. The taotie features a mouth that extends continuously nearly the full width of the face (the mouth is more clearly visible on the foot), a type commonly seen among bronzes of the transition period.212 The design seen on zun K2(2):135 in Figure 4.13 embodies an early experiment at Style IV, with the contours of horns and bodies delineated in thicker lines than the texturing patterns. Similar attempts can be seen on the Gaocheng pou mentioned previously in discussion of the pou from K1, on a zun from Hui Xian, and on a you from Zhengzhou (XSH1:11).213 The decoration should probably be still classified as Style III, however. Although the vessel has lost much of its mouth and foot, the closest parallel in both decoration and shape is undoubtedly a zun from Lingbao in Henan province; the similarity in the heavy scalloped flanges is striking.214 Zun K2(2):109 is monumental; in its fragmentary state it still measures an impressive 69 cm high, taller than any other vessel from K2 (Fig. 4.14). The high-relief Style III taotie features reclining C-shaped horns and a mouth that splits into double jaws. 212For similar taotie designs see Beijing 1984b (p. 5 fig. 9, p. 6 fig. 10, p. 12 fig. 27, p. 28 figs. 65–66, p. 33 figs. 83–84, p. 43 fig. 115, p. 64 fig. 178), Bagley 1987 (no. 51), and Beijing 1997b (pl. 91). A feature not seen here but commonly appearing in similar examples is curly hooks between the lips, suggestive of teeth. For examples of zun vessels of similar shapes of this period see Beijing 1996b (pls. 112, 114), and Beijing 1994b (p. 274 fig. 144[yi]1, 3). The Sanxingdui zun’s wall is rather straight and thus somewhat different from the curve commonly displayed by the other zun vessels referred to here. 213For the Gaocheng pou see note 202. For a color illustration of the Hui Xian zun see Beijing 1996b, pl. 120. The Zhengzhou you was first reported in Wenwu 1983.3, p. 54 figs. 14–15, p. 55, and color plate; see also Beijing 1996b (pl. 136), Beijing 2001f (p. 821, p. 822 fig. 553, color pl. 37, pl. 230). 214The Lingbao zun was first reported in Kaogu 1979.1, p. 20. For a color illustration see Beijing 1996b, pl. 113. 91 The bodies are detached from the face and appear in a S shape rising up toward the edge of the register. Below the bodies are a pair of dragons flanking the taotie. A zun in the Shanghai Museum thought to be from Anhui province bears a close resemblance to this vessel in decoration, including the taotie design, the prominent bovine heads and the jagged flanges. The vessel shapes were probably similar, too, although the mouth of the Sanxingdui zun is incomplete. Similar high-relief Style III taotie can be seen on a pou from Gaocheng (M112:4) and a lei from Xin’gan in Jiangxi province (XDM:44).215 Like vessels from K1, these three vessels are hard to connect with any specific place of production as they feature widespread styles of the transition period. 4.2.2.2 Vessels of Styles IV and V(a) The vessels of this category demonstrate remarkable consistency in shape and in design, though the designs are executed differently in Style IV or Style V(a). The shapes commonly exhibit a upward expansion with widely flaring mouths, which extend far beyond the middle section. The foot is high, exceptionally so in some cases. At the same time the middle section tends to become proportionally compressed, setting off the other two parts and accentuating the changes of direction in the vessels’ profile. The visual effect of these zun vessels is immediately striking if we compare zun K2(2):112 in Figure 4.12 and zun K2(2):151 in Figure 4.23: the latter is imposing, almost flamboyant, with an irrepressible soaring quality. The lei vessels show a striving for a imposing effect also, as 215For an illustration of the Shanghai Museum zun see Bagley 1987, p. 86 fig. 73. For the Gaocheng pou see Beijing 1985a, p. 129, p. 131 fig. 78.1, and color pl. 3; see also Fong 1980, no. 13; this pou is different from the Style IV one also from Gaocheng discussed before. For the Xin’gan lei see Beijing 1997e (p. 73, pp. 76–8 fig. 41 [A–C], color pl. 19.1, pl. 20.1), and Beijing 2006a (pp. 71–5). 92 comparison between lei K2(2):70 or lei K2(2):88 on the one hand and lei K2(2):159 on the other will amply show (Figs. 4.16 or 4.17, and 4.24). Otherwise, these three vessels are basically identical in structure, all having a very deep middle section with an almost straight wall that creates a weighty appearance. In decoration all the zun have the following features. Three raised bowstrings encircle the neck. Three animal heads sit on the shoulder and intrude into the middle section to hover over the central axes; the heads feature flat sheep or bovine horns and often carry a small bird or curved flange on the head. All the vessels have three jagged flanges on the middle section and the foot, emphasizing the tripartite division of the circumference into compartments. The short flanges on the shoulder are replaced by thin flat birds. Each compartment is occupied by a taotie design. The foot has in addition a plain register, interrupted by three rectangular holes aligned with the flanges because the mold was divided there and the holes served for core extension. One or two raised bowstrings encircle this plain register. The decoration on the lei vessels is basically the same, but the animal heads actually hang from the top edge of the middle section, though they look as if they project from the shoulder. The upright horns, sometimes very prominent, as seen on lei K2(2):159 in Figure 4.24, tend to dominate attention. The most peculiar trait is the presence of four sets of vertical flanges, hence four animal heads and four identical units of decoration in each register. This quadripartite division of a round shape had its origin in the transition period as seen, for example, on the Xin’gan lei (XDM:44) mentioned earlier. 93 The taotie, the dominant motif on these vessels, deserves discussion in more detail. The designs seen on these vessels may be broadly grouped into three types distinguished primarily by the shape of the taotie’s mouth and bodies. (A) Taotie with continuous mouth that extends nearly the full width of the face. This type of taotie appears in Style IV on vessels listed above as (4)–(8) (Figs. 4.15–4.19), and in Style V(a) on vessels (9), (10) (Figs. 4.20, 4.21).216 Its narrow bodies, when shown, are issued directly from the sides of the hooked eye sockets (Figs. 4.15–4.17, 4.19). The Style V(a) designs of this type seems to be based on the Style IV version like that in Figure 4.18: compare the shape of the horns and the ears; and both taotie are bodiless. For another interesting detail, compare Figures 4.15b and 4.21b: the Ioniccapital-like volutes next to the mouth, to which the claws are attached, appear the same. The designs also have differences, however. The nose in Style V(a) designs has nostrils in the form of whorls, clearly based on an earlier high-relief Style III version as seen in Figure 4.14 and it also has two new hooks projecting outward at the level of the eyebrows. The eyes are greatly diminished in size to the point that they are barely noticeable, the space now shared by a pair of eyebrows borrowed from classic Style V taotie.217 The mouth is filled with volutes (the same volutes curiously cover the eyebrows in Figure 4.21). The visual effect of this kind of design is confusing: the taotie does not try very hard to catch attention. (B) Taotie with double jaws and detached bodies, which is rendered in Style V(a) on vessels (11)–(14) (Figs. 4.22–4.25). It features a mouth that splits into double jaws. 216This type of taotie is the same kind as the one discussed in Bagley 1987, pp. 341–2, pp. 541–3. 217For an example of classic Style-V taotie see Bagley 1999, p. 155 fig. 3.9. 94 The bodies, always shown, are detached from the face and appear in a S shape next to the eyes and horns. Clearly, this taotie was developed straight out of its precursor in highrelief Style III as seen in Figure 4.14,218 and it shows remarkable constancy with little variation in detail except that the dragons flanking the taotie on both sides appear in different degrees of vagueness depending on space available to them; in one case the motif is abbreviated to a simple C shape (Fig. 4.24). (C) Taotie with double jaws and connected bodies Type C appears on the foot of vessels of Style V(a), (9)–(13) and (15) (Figs. 4.20– 4.24, 4.26). It seems to be related with both the previous two types: the nose is basically the same; it has double jaws, but its narrow bodies are connected directly to the eyes in a way reminiscent of Type A. Nevertheless, its horns are unmistakably unique and remain the most conspicuous feature of this taotie. Shaped like the letter U turned upside down and free of filler pattern, the horns stand out rather sharply against the background.219 This type of taotie remains consistent whether appearing on vessels whose middle sections bear Type A taotie or Type B taotie. Besides these three types of taotie, the fragmentary lei in Figure 4.26 presents a strange combination of Types A and B on its middle section. The taotie actually has two mouths: one derived from Type A, and beneath it the second one with double jaws. It also has two sets of horns: one set similar to those seen in Figure 4.21, covered with scales, 218In high-relief Style III, taotie with detached bodies sometimes has a continuous mouth as seen, for example, on the Gaocheng pou (M112:4) cited before. See note 215 for reference. 219On a bronze hu unearthed at Shilou in Shanxi province, the taotie with the same horns appears in a conspicuous, elaborate form, albeit without bodies (see Fong 1980, no. 21). This taotie is among the evidence that prompts me to suspect a Changjiang connection among Shanxi bronzes, a subject that has been briefly discussed in Xu 1998 and will be dealt with in a future study. 95 and below them the other set featuring C-shaped horns of Type B. The taotie has eyebrows, the way by which they are wrapped by volutes is similar to that in Figure 4.21, but its bodies are S-shaped and detached. The draftsman had obviously mixed two taotie designs together. The foot of this lei, on the other hand, carries a taotie typical of Type C. The preceding discussion of the taotie designs seen on K2 vessels reveals one interesting phenomenon that needs some further comments. Type A designs appear in both Style IV and Style V(a). As Loehr and Bagley have shown, Style V(a) was developed out of high-relief Style III, which had grown out of Style III, while Style IV also originated in Style III but gave rise to Style V.220 In other words, Style V(a) was not based on Style IV. The reasoning for this conclusion is clear: the image-ground distinction achieved by Style IV does not exist in Style V(a) in which both the taotie and background patterns are covered with spirals of the similar density, making the taotie less than distinct; the effects of these two styles are thus diametrically opposed to each other. As pointed out above, visual confusion with Style V(a) is amply evident in Figures 4.20 and 4.21. Therefore, the relationship between Type A taotie designs of Style IV and Style V(a) must be understood as a borrowing of motifs rather than stylistic evolution. At the same time, the phenomenon illustrates the staying power of Style III tradition and a stronger interest in patterning than in image making. The lei in Figure 4.26 discussed above, which mixes two taotie in one design, demonstrates the same preference. 4.2.2.3 The Central Plain connection 220For relevant references see notes 195, 206, 211. For Style V see Bagley 1987, pp. 28– 30. 96 The typological and decorative traits of the Styles IV and V(a) vessels discussed above connect them unequivocally to the middle Changjiang region. Before citing examples from that region, however, it is necessary to examine the issue of cultural connections with the Central Plain because many scholars hold the opinion that the Sanxingdui vessels were local products resulting directly or indirectly from influences of the Shang culture in that region (the Shang culture being referred to covers the entire time span from the Erligang to Anyang period.)221 Certainly the vessels from K1 as well as the three earlier vessels from K2 were products of the widespread Erligang culture and its successors of the transition period. The situation changed late in the transition period and during the Anyang period, however, when distinctive regional cultures sprang up from a common Erligang base. Thereafter we are dealing less with influence radiating outward from the Central Plain than with cultural equals in a relationship of mutual influence.222 In theory and practice alike, it is not appropriate to think in terms of a single unvarying 221Publications expressing this view include Huo Wei 1989, Li Xueqin 1989, Li Xueqin 1993, Zheng Zhenxiang 1993, Li Xiandeng 1994, Zheng 1996, Yu Weichao 1996, Xu Lianggao 1998, and Song Zhimin 1998. Li Xueqin, aware of stylistic similarities between the Sanxingdui vessels and those from the middle Changjiang region and the Chenggu region as well as their differences from Central Plain bronzes, argues for indirect influence. Zheng Zhenxiang entertains both possibilities, but concludes that “it is likely that peoples of the Chengdu plain… were in contact with the central Shang state in Henan province but that they had closer links with the southerly local cultures of Hunan and Hubei.” (Zheng 1996, p. 246). Yu Weichao generally identifies the vessels as copies of the Shang originals (p. 61). Huo Wei (p. 38) and Xu Lianggao (p. 230) separately assert that the vessels copied the Shang prototypes with local characteristics. Li Xiandeng 1994 and Song Zhimin 1998 (pp. 112–5) represent an extreme case, both arguing that the Sanxingdui vessels directly copied Anyang bronzes. In addition, Ge & Linduff 1990 holds that the vessels (and some jades) are modeled after a variety of Shang prototypes (p. 508), but seems to doubt their local production by noting an absence of evidence suggesting that they were made locally (p. 512). 222For studies of the contacts and influences between Anyang and surrounding regions, particularly the middle Changjiang region see Rawson 1992 and Rawson 1993. 97 relationship between the Sanxingdui vessels and those of the Central Plain. Vessels of different periods must be analyzed separately. To elucidate the relationship between the vessels from K2 and those of the Central Plain, let us contrast them with an example from Anyang. A zun from M18 at Anyang Xiaotun will suffice for the purpose; it is as close a parallel to the Sanxingdui zun as can be found at Anyang (Fig. 4.30).223 Some important differences in shaped may be noted: though the Anyang zun has a taller neck and mouth, it flares less; its middle section is deeper and its foot shorter. It is thus more earthbound than the K2 vessels. Zun K2(2):127 in Figure 4.20 is arguably the least daring in shape among the K2 vessels, yet the difference is already clear. Examples like K2(2):151 in Figure 4.23 drive the point home. In several ways, the Anyang vessel shows new inventions in decoration whereas the K2 vessels retain old conventions. The most noticeable is the triangular designs on the Anyang zun’s neck, an area that on earlier vessels and on the K2 examples carries no decoration except bowstrings. The flanges on the Anyang vessel have become notched and uniform in size, and the center of the taotie is highlighted by the addition of another flange. The K2 vessels continue with the Erligang tradition in the use of crenellated flanges that gradually diminish in size from top to bottom, and the taotie’s nose retains the slightly raised ridge. The animal heads on the K2 vessels sit astride the shoulder and the middle section, again an old convention from the Erligang period, while those on the Anyang zun truly sit on the shoulder with their muzzles projecting in the air. However, 223The zun was reported in Kaogu xuebao 1981.4, p. 497. For a color illustration see Beijing 1997b, pl. 97. Its decoration is rendered in Style IV. For a very similar zun decorated in Style V see Beijing 1997b, pl. 95. 98 the vessels from K2 display one new trait unparalleled by earlier bronzes or contemporary vessels from Anyang, that is, the flat bird sitting on the shoulder replacing the short flange. Similar birds or creatures hard to identify often stand on the animal heads. These uses of birds are characteristic of middle Changjiang bronzes.224 In addition, the fragment K2(3):23 in Figure 4.28, probably from the lid of a bronze lei (see below), features a high-crested bird in the form of a flat plaque, like a much elaborated flange or a spectacularly feathered version of the little birds on the vessel shoulder. Similar birds also replace the flanking dragons in the Style IV design seen in Figure 4.16. In taotie designs, the Anyang and K2 vessels show clear similarities as well as differences. The techniques of Style IV and Style V(a) are shared, and some taotie designs are similar enough to indicate that they are different variations of the same images (Style IV designs, and Style V[a] in Figures 4.22–4.25).225 Yet the Style V(a) renditions of taotie as seen in Figures 4.20 and 4.21 (both the middle section and the foot) have no parallel in the Central Plain, nor does the whimsical mixture seen in Figure 4.26. Likewise, Style V designs and a type of exploded Style V(a) taotie typical of Anyang are not seen among the K2 vessels.226 Finally, another distinctive trait appears on vessels of Style V(a) from K2: the inside wall is depressed in correspondence to the relief on the outside, thus keeping the wall uniform in thickness. This was done perhaps to save metal or to prevent casting defects that could be caused by uneven cooling of the metal if the wall were uneven. The 224Bird motifs and crenellated flanges on middle Changjiang bronzes are discussed in detail in Bagley 1987, pp. 544–8. 225Bagley suggests that the Style IV taotie seen on middle Changjiang region bronzes is copied from Anyang designs (Bagley 1987, p. 543). 226For an example of exploded Style V(a) taotie see Bagley 1987, no. 78. 99 technique was invented in the transition period, the zun from Anhui Funan being one example (Fig. 4.11).227 In the Central Plain it ceased to be used in the Anyang period, at least in part due to advanced casting prowess. In the south, a general conservatism seen in some aspects of the vessels discussed above might have been a contributing factor for its continued use. As part of the technical peculiarity of these vessels, the holes on the foot are always large rectangles, whereas on the Central Plain vessels the holes are smaller, usually cross-shaped, and sometimes there are no holes at all. In summary, the zun vessels from K2 are distinctly different from their counterparts at the Central Plain. Their connection, seen mainly through taotie designs, is limited. As regards the lei vessels, so far no comparable shape is known from Anyang or elsewhere in the Central Plain at the stages of Style IV and V, i.e., the late 13th century BC and later. The vessels from K2 thus embody a development parallel to that in the Central Plain rather than one under its influence. In the following part, when the provenance of the K2 vessels is examined, we will come to see that even that limited connection does not apply to the Sanxingdui site because the vessels are most likely imports to Sanxingdui from the middle Changjiang region. 4.2.2.4 The middle Changjiang connection and related issues All the differences from the Anyang counterpart enumerated above are applicable to zun and lei vessels from the middle Changjiang region, because they have the same typological and decorative traits as those from K2. Similar objects from the middle 227The inside depressions on this type of vessels were pointed out and discussed in Bagley 1987, pp. 267, 272. 100 Changjiang region are listed below. No detailed comments seem necessary; brief remarks on the taotie designs and on other features will be made as appropriate.228 (1) Lei from Yueyang, Hunan province, h. 50 cm.229 Style IV, Type A taotie without body on middle section and foot. The border at the top of the middle section, made up of alternating whorls and flowerets, is the same as that on three K2 lei vessels in Figures 4.24–4.26.230 Stylistically, this lei seems to be intermediate between lei K2(2):70 or K2(2):88 and lei K2(2):159 (Figs. 4.16 or 4.17, and 4.24), as it has the Style IV design of the former but a higher foot and neck that bring it closer to the shape of the latter. Moreover, it has the same thin flat birds on the shoulder as the latter vessel. (2) Vessel lid said to come from Changsha, Hunan province, h. 28, Musée Guimet, Paris (MA1612) (Fig. 4.29). The Style IV Type A taotie on the lid is similar to that on lid K2(2):32 in Figure 4.19, which has also the same quadripartite division of the surface by hooked flanges. As such a division is characteristic of lei vessels, these two lids most likely belonged to lei. Similarity is also obvious between the high-crested bird on the Guimet lid and the bird K2(3):193-1 in Figure 4.27, which suggests that 228The list is not exhaustive. Zun and lei vessels in museum or private collections without known provenance are not listed even if their styles leave no doubt of their Changjiang origin. Vessels of other types carrying similar designs are not listed here either. For surveys of zun and lei vessels unearthed in the Changjiang region see Shi Jinsong 1998b, and Zhang Changping 2004. 229The vessel was first reported in Hunan kaogu jikan 1984. For color illustrations see Beijing 1998d (pls. 93–94), Bagley 2001 (p. 146 fig. 48.1). 230This pattern, popular in the middle Changjiang region, is discussed in detail in Bagley 1987, pp. 543–4. 101 the K2 bird probably belonged to a lid, perhaps of a lei also. Another fragment mentioned earlier, K2(3):23 in Figure 4.28, might also be identified as belonging to a lid: its taotie faces toward the center like those on the other two lids. But the similar high-crested bird it carries becomes a flange.231 (3) Zun from Jiangling, Hubei province, h. 63.5 cm.232 Style V(a), Type B taotie on middle section, Type C taotie on foot. (4) Zun from Jiangling, Hubei province, h. 46.2 cm. From the same find, style similar as above. (5) Zun from Zaoyang, Hubei province, h. 53 cm.233 Similar to (4). (6) Zun from Yueyang, Hunan province, h. 56.5 cm.234 Similar to (4) and (5). (7) Zun from Huarong, Hunan province, h. 72.3 cm.235 Style V(a), Type B taotie on middle section, Type C taotie on foot, shape among the most flamboyant. The massive animal heads are of the type usually seen on lei as in Figure 4.24. 231This fragment is mistakenly identified as from a vessel’s shoulder in Beijing 1999a, p. 252. 232This and the next vessel were first published in Wang Congli 1993. In addition, there is a third zun of the same style apparently found in 1994 in the same area. All the three zun were on view at the Jingzhou City Museum when I visited on October 17, 2000. 233The vessel was first reported in Xu Zhengguo 1988 and Xu Zhengguo 1990. 234The vessel was first reported in Xiong Chuanxin 1981. For a color illustration see Beijing 1983, pl. 38. 235The vessel was first reported in Wenwu 1972.1. For illustrations see Beijing 1972 (p. 30), and Beijing 1983 (pl. 17). 102 (8) Lei from Shashi, Hubei province, h. 51.5 cm.236 Style V(a), middle section combining traits seen on several vessels from K2: lei K2(2):103 in Figure 4.26 (borders of alternating whorls and flowerets at both the top and bottom edges), zun K2(2):129 in Figure 4.21 (mouth and eyebrows), and horns and detached bodies of Type B taotie; Type C taotie on foot. (9) Lei or pou said to come from Changsha, Hunan province, h. 38 cm, Museum of Far Eastern Antiquities, Stockholm.237 Style V(a), Type B taotie on middle section, Type C taotie on foot. A border at the top of the middle section is made up of alternating whorls and flowerets. (10) Sheep-horned animal head from Liuyang, Hunan province, h. 13 cm.238 Similar to the animal heads on the Huarong zun (7) and lei K2(2):159 in Figure 4.24. The vessels from K2 are identical in style to those from the middle Changjiang region.239 As previous studies by Bagley have established the probability of the style being a development that took place in the middle Changjiang region,240 the question of the place of production of the K2 vessels arises: were they locally made at Sanxingdui after the middle Changjiang models or were they imported? Several scholars have 236The vessel was reported in Peng Jinhua 1987. 237Bagley 1987, p. 274 fig. 43.6. 238The head was reported in Hunan kaogu jikan 1986, p. 28. 239Although there is no square vessel like the fang lei in Figure 4.18 among those known to have been excavated in the middle Changjiang region, a fang zun in the collection of the National Palace Museum, Taibei, resembles it closely in the taotie design, while its tell-tale small birds folded around the vessel corners and animal heads with wing-like crests on the shoulder leave no doubt its southern provenance. The museum also has a fang lei similar in decoration though much more squat in shape. For discussion of the Taibei fang zun see Bagley 1987, p. 36. For color illustrations of the two vessels see Chen Fangmei 1998, nos. 52 and 69. 240See note 211 for relevant references. 103 commented on the issue. Bagley was the first to observe the middle Changjiang connection, and based on the stylistic affinity, to raise the possibility that the vessels might be imports.241 Thorp, Falkenhausen, Nanba Junko, and Zhang Changping advocate this possibility also.242 Li Xueqin, on the other hand, argues for their being copies, and Rawson and Shi Jinsong seem to share this opinion.243 Several factors argue for the possibility that the Sanxingdui vessels were indeed imports from the middle Changjiang region. None of them may be decisive by itself, yet when taken together, they are persuasive. These factors are as follows. (1) Quite a few vessels from K2 have small circular holes drilled in their foot rims, some directly through the decoration: zun K2(2):112, zun K2(2):109, lei K2(2):88, fang lei K2(3):205/205-1, zun K2(2):127, zun K2(2):129, lei K2(2):159 (Figs. 4.12, 4.14, 4.17, 4.18, 4.20, 4.21, 4.24).244 These appear to be modifications carried out at Sanxingdui, 241Bagley 1992, pp. 225–6. Earlier in Bagley 1988 (p. 84) he pointed out the stylistic similarity, and in Bagley 1990a (p. 65) he left open the possibility that such vessels might have been made in both Sichuan and the middle Changjiang region. 242Yang 1999 (p. 225), Falkenhausen 2002 (p. 66), Nanba Junko 2002 (pp. 146, 149), and Zhang Changping 2004 (pp. 125–6). Okamura Hidenori 1998 (p. 206), however, asserts that the vessels in the middle Changjiang region were imports from Sanxingdui, an assertion that is hard to justify on present evidence. 243Rawson 1996 (p. 72), Shi Jinsong 1998b, Li Xueqin 1999 (pp. 287–8). Rawson also has another opinion. Commenting on lei K2(2):159 (Fig. 4.24), she suggests, “It seems probable that the peoples from Sichuan and Hunan copied their bronze vessels from a common source.” (ibid., p. 70). The common source implied is undoubtedly that of the Central Plain. There is no such a source to be “copied from” in the Central Plain, however. This lei is very different from any counterpart in that region, embodying changes and inventions unseen there. 244In cases where information about the holes provided below conflicts with that provided in the final excavation report, Beijing 1999a, the information was obtained from my own examination. Zun K2(2):112, 5 holes, drilled from outside at the undecorated bottom edge of the foot. Zun K2(2):109, 3 holes, drilled from outside directly on the bottom border of circular dots. 104 perhaps to adapt the vessels to a local placement different from the one their middle Changjiang casters intended them for. (2) The birds that appear on the vessels are completely different from the birds on other Sanxingdui bronzes. The birds on the vessels feature a straight spiky beak, circular eyes and eye sockets, a scale pattern as part of surface decoration, and sometimes a high, florid crest. The birds on other Sanxingdui bronzes typically have a hooked beak and oval eyes in pointed eye sockets (Fig. 5.30). (3) There is also a clear distinction in surface decoration between the vessels and the images. While the vessels are densely decorated, most of the bronze images have no surface decoration at all. Those that do, most notably the statue of a standing figure (Fig. 5.17), bear intaglio patterns that were comparatively easy to cast. Insofar as these patterns betray an influence from vessels, they depend on transition-period designs rather than on the intricate Style IV or Style V(a) high relief of the zun and lei. The bulging eyes on the bronze heads and masks are rendered all the more strange by their failure to make even the slightest allusion to the ubiquitous taotie.245 And conversely, influence from bronze images is also absent from the decoration of the vessels. Lei K2(2):88, 6 holes, drilled directly through decoration and placed next to the flanges; two flanges each having two holes symmetrically placed, and the other two flanges each having one hole. Fang lei K2(3):205/205-1, 4 holes, drilled at the center of each side on the undecorated bottom edge. Zun K2(2):127, 3 holes, drilled on the undecorated bottom edge. Zun K2(2):129, 2 holes, drilled directly on the decoration, but one might be a casting defect. Lei K2(2):159, 5 holes, some drilled directly through the decoration. 245One type of flat plaques among Sanxingdui images presents taotie-like faces (Fig. 5.51), but again, their features are related to transition-period taotie like those seen on zun K2(2):88 in Figure 4.17. 105 (4) The techniques by which the Sanxingdui bronze images were cast seem to betray a distinctly different mentality from that of the casters of vessels, as will be discussed in detail in Chapter 6. Moreover, as it has already been observed, K1 had fewer vessels but more types, whereas K2 had more vessels but fewer types. The limited repertoire of shapes in K2 might partly reflect the limited selection that was available to Sanxingdui patrons from their trading partners down the river. Or perhaps Sanxingdui patrons did exercise some choice because zun and lei were the only types useful for the local ritual, the other bronze types most characteristic of the middle Changjiang region, namely animal-shaped vessels and large bells, are not represented at Sanxingdui.246 The change in vessel types between the two pits may represent a standardization in ritual during the meantime. The discussion of the bronze images in Section 5.1.7 will suggest that the latter possibility is strong. Now let us briefly look at the routes by which the imports might have entered Sichuan and the networks of trade or exchange that might have brought them. The middle Changjiang region is most obviously connected with the Sichuan Basin by the Changjiang River, and the presence in the Three Gorges region of a Style V(a) zun, listed below in (4), argues for the use of this route. Besides Sichuan, the middle Changjiang had connections with other regions also, as attested by finds of similar vessels. Interestingly, all these regions so far known lie outside the Central Plain where the Shang power was situated. The vessels are listed as follows. 246Noted in Bagley 1999, p. 213. 106 (1) Zun (1963CHBSTT:1) from Chenggu, Shaanxi province, h. 46 cm.247 Style IV, Type A taotie on middle section and foot. (2) Zun from Qingjian, Shaanxi province, h. 17 cm.248 Style IV, Type A taotie on middle section and foot. The flanking dragons on the middle section are replaced by high-crested birds as on zun K2(2):70 in Figure 4.16. (3) Lei from Jiangning, Jiangsu province, h. 28.7 cm.249 Style IV, Type A taotie on middle section. (4) Zun from Wushan, Sichuan province, h. 42 cm.250 Style V(a), Type A taotie on middle section, Type C taotie on foot. Particularly similar to zun K2(2):129 in Figure 4.21. (5) Zun from Liuan, Anhui province, h. 70 cm.251 Style similar to four vessels in the previous list: two zun from Hubei Jiangling, (3)–(4); one from Hubei Zaoyang, (5); and one from Hunan Yueyang, (6). These vessels, together with comparisons drawn from the K1 vessels and the three early vessels from K2, suggest that two additional regions were important in this network of exchange as far as Sanxingdui is concerned: the Hanzhong Basin, where Chenggu is 247The vessel was first reported in Zhu et al. 1966, then in Tang et al. 1980 (p. 214 fig. 4.1, 4.2, p. 215, pl. 2.4), now formally in Zhao Congcang 2006 (p. 70, p. 71 fig. 69, pl. 27). See also Beijing 1979, no. 111. 248The vessel was first reported in Dai Yingxin 1980, p. 95, pl. 12. For illustrations see also Beijing 1979, no. 61. Qingjian is located in northern Shaanxi on the west bank of the Yellow River, across the river from Shanxi Shilou where the bronze hu referred to in note 219 was found 249For a color illustration see Beijing 1998d, pl. 96. 250The vessel was first published in Beijing 1997f, pl. 43, and first reported in Beijing 1998e, p. 9. 251The vessel was first reported in Zhongguo wenwu bao (August 11, 1999), and in Wenwu 2000.12. 107 located, and the western Anhui region in the Huaihe River drainage where both Funan and Liuan are located (Liuan is south of Funan). Geographically, the Hanzhong Basin is connected with the middle Changjiang region by the Hanshui River and the Huaihe River drainage is close to the Changjiang region. All three regions were part of a widespread cultural sphere in the Erligang and transition periods, and they probably maintained a close relationship during the Anyang period.252 The Hanzhong Basin is also immediately north of the Sichuan Basin across the Qinling Mountains, and culturally close to the Sichuan Basin in several respects.253 It is possible that some of the middle Changjiang imports might have been acquired via the Hanzhong Basin rather than by the more direct route up the Changjiang.254 Sanxingdui’s connection with Chenggu may be pushed back to earlier times. Although the source for the early vessels from K1 and K2 is hard to pin down, as noted earlier, we might now speculate that some of the early bronzes at Sanxingdui, like the pan K1:53 in Figure 4.7 and the zun K1:163/K1:59 in Figure 4.8, which find their closest parallels at Chenggu, were fabricated there and exported to Sanxingdui, or at least were transmitted by way of Chenggu. 252In the Chenggu and Yang Xian areas of the Hanzhong Basin a large number of bronze vessels, ranging in date from Erligang through the Anyang period, have been unearthed, and even peculiar faces recalling an example from distant Xin’gan. For the Chenggu and Xin’gan faces, see Bagley 1999, p. 180 fig. 3.19. On bronzes in the Hanzhong Basin see Zhao Congcang 1996, Bagley 1999 (pp. 178–80), and Zhao Congcang 2006. 253See discussion in Sections 8.1 and 8.2.3. 254Both routes were noted in Li Xueqin 1989. His thesis, however, is the transmission of the Shang culture from the Central Plain by the two routes. Shi Jinsong, on the other hand, believes that the Chenggu zun vessels copy the Sanxingdui vessels, which in turn copy vessels from the middle Changjiang region. He therefore argues for a transmission of the vessel type from Sanxingdui to the Hanzhong Basin (see Shi Jinsong 1998b). Basically Shi is arguing that the vessels were cast in each of the places where they have been found, a hypothesis that seems unnecessary (and that does still assume the movement of actual vessels from one region to another). 108 Separately, we may notice that some other early vessels at Sanxingdui find close parallels in the Anhui region, i.e., the zun K1:158/258 featuring the tiger-human theme in Figure 4.10, and the high-relief Style III zun K2(2):109 in Figure 4.14. We might also see a direct connection between the Hanzhong Basin and western Anhui in the comparison of two high-relief Style III vessels, a lei from Chenggu and a zun from Funan.255 Again, the possibility exists that the Hanzhong Basin was the route of transmission. That the zun K2(2):109 was an import is suggested by the holes on the foot. Another early vessel, zun K2(2):112 in Figure 4.12, was modified in the same way. 4.2.3 Function of the vessels The function of the vessels from K1 and K2 has already been touched on in the preceding discussion. The list below further enumerates particular conditions in which some of the vessels were found at the time of the excavation. Their function emerges clearly. (1) Several vessels have cast defects, that is, holes in the bottom and/or wall: zun K2(2):112, zun K2(2):129, zun K2(2):151, lei K2(2):103/K2(2):103-1 (Figs. 4.12, 4.21, 4.23, 4.26). (2) Many of the vessels are smeared with a red pigment on the outer surface: zun K2(2):112, zun K2(2):79, lei K2(2):70, lei K2(2):88, zun K2(2):127, zun K2(2):146, zun K2(2):151, lei K2(2):159 (Figs. 4.12, 4.15–4.17, 4.20, 4.22–4.24).256 255For color illustrations of the Chenggu lei and the Funan zun see Beijing 1996b, pls. 129–30, 115–6. 256Beijing 1999a is not thorough and consistent in providing information about the conditions of the objects. The information provided here is partly based on my examination and does not tally exactly with the report. 109 (3) Many vessels contained other objects.257 Zun K1:158/258 (Fig. 4.10), fragments of burned jade and stone implements, including one jade axe, one jade chisel, 62 cowry shells, partially melted ornaments of thin sheet bronze, one bronze square collared disk, and unidentified charred objects. Zun K2(2):135 (Fig. 4.13), one piece of jade raw material. Zun K2(2):109 (Fig. 4.14), one bronze human mask, two bronze diamond-shaped appliqués. Zun K2(2):79 (Fig. 4.15), two jade fragments of miniature forked blades, 70 ivory beads, two fragments of decorated ivory objects, five bronze jingles, four bronze circular pendants, four bronze turtle-shell-shaped pendants, 10 bronze scallop-shaped pendants, and one bronze eye. Lei K2(2):70 (Fig. 4.16), one jade chisel, one whetstone, one bronze bird-shaped ornament, one bronze serrated ge blade, three bronze circular pendants, two bronze 257The actual contents of the vessels have to be retrieved from dispersed information in Wenwu 1989.5 and Beijing 1999a. My method of retrieval is as follows. The description of lei K2(2):70 in Beijing 1999a (p. 253) does not mention that it contained any objects, yet earlier in Wenwu 1989.5 (p. 9) it is mentioned that the vessel contained a bronze birdshaped ornament, K2(2):70-9. The accession number assigned to this bird ornament reveals a system that assigns numbers to objects according to what they were contained in. Beijing 1999a lists many objects as K2(2):70-. This suspicion is verified by the information about lei K2(2):88 published in Beijing 1999a. The description of this lei on pp. 253–4 does not mention that the vessel contained anything, but on p. 404 the description of jade beads and tubes mention that they were contained in the lei, and their accession numbers appear as K2(2):88-. The information provided here is therefore derived from a search for objects that are identified by the accession numbers of the vessels. On a separate note, caution needs to be exercised in judging the contents in vessels; certain objects may have ended up in vessels by accident during the burial process. Unfortunately there is no way to make certain. The pan K1:53 is excluded from the present list because it is very possible that objects found with it at the time of the excavation present one such accidental association: the pan is shallow and badly damaged. On the other hand, many objects must have been poured out of vessels while they were broken, burned and dumped into the pits. All the vessels in K2 were found lying on their sides. 110 scallop-shaped pendants, one bronze bird-shaped foil, one bronze fragment of a bird’s tail, and one bronze fish-shaped pendant. Lei K2(2):88 (Fig. 4.17), 284 jade beads of various kinds, 42 jade tubes of various kinds, about 40 jade chisels, one jade collared ring, and one gold forked-blade-shaped foil. Zun K2(2):127 (Fig. 4.20), a large number of cowry shells, and one jade collared ring. Zun K2(2):129 (Fig. 4.21), 935 cowry shells, and one gold forked-blade-shaped foil. Zun K2(2):146 (Fig. 4.22), two jade collared disks, 602 cowry shells, of which 375 are complete, and one gold fish-shaped foil. Lei K2(2):103/K2(2):103-1 (Fig. 4.26), 46 ivory beads, 1,488 cowry shells, of which 1,233 are complete, three tiger molars, three bronze eyeballs, six bronze jingles, one bronze circular pendant, five bronze turtle-shell shaped pendants, and five bronze scallop-shaped pendants. The list reveals, first of all, that none of the vessels should be called wine containers, though that seems to have been the function of their counterparts in the Central Plain. They were obviously not intended to hold liquid; casting defects were never repaired because they did not need to be. The vessels were used as containers for solid objects offered in some kind of ritual,258 and they were themselves consumed as 258In the middle Changjiang region, bronze vessels are occasionally found full of jades or small bronzes. For examples see Fong 1980 (pp. 129–31 no. 25), Bagley 1987 (p. 337 fig. 57.4, pp. 373–4, p. 380 note 4), Zheng & Tang 2000, and Yu Yanjiao 2002. The jade collared rings and small beads and tubes contained in the Sanxingdui vessels appear similar to those found in the middle Changjiang vessels, but the other contents are different. It is possible that the Sanxingdui users not only imported the vessels but also 111 offerings. In Figure 4.31, a figure, perhaps a woman judging from the prominently depicted nipples, is seen kneeling at the summit of an openwork base and bearing on her head a zun vessel with a high lid. The figure has the same facial features as bronze heads and mask, leaving it without doubt she was part of the local imagery. It is tempting to believe that she is bringing an offering for deposit in the pit. some of their contents. Perhaps the function of the vessels copied from the middle Changjiang practice as well. 112 Chapter 5. Bronzes at Sanxingdui (II): Images and Related Implements Images and related implements account for the vast majority of bronzes from K1 and K2. They are unprecedented and unique. The combination of their quantity and quality leaves no doubt that they were products of a local bronze industry. However, the city where they were made and the people who made and used them are known from archaeology only sketchily, from inscriptions and texts not at all; and the evidence from the objects themselves is far from complete. A glance through the objects from the two pits makes it clear that, rich and breathtaking as the bronze images are, they are much reduced from their original condition—not only because of more than three thousand years of burial but also because of the particular processes to which they were subjected at the time of sacrifice. Almost all of the objects were intentionally battered and/or burned before burial, leaving them damaged or in some cases completely destroyed (Figs. 2.6, 3.3, 5.1, 5.2, 6.31e). Some objects may have been put into the pits unfinished, and some were evidently deposited incomplete (that is, some fragments of an object made it into the pits while others did not). Besides, in their original settings many images were apparently components of large assemblages; the other parts of the assemblages, made from other materials, perished or never found their way into the pits. All these factors put severe limits on what we can hope to understand about the objects and the ritual they served. As argued in Section 3.1, we must rely on internal relationships, or in other words, combine the evidence provided by the archaeological context with the evidence supplied by the objects, and on that evidence venture our reconstruction and interpretation. 113 We may begin by noticing that the bulk of the bronzes can be sorted into two loose categories in terms of iconography. The first consists of heads, figures, and masks, most of which share a striking physiognomy characterized by sharply cut features and enormous eyes. These faces are distributed along a scale that ranges from comparatively human (Fig. 5.3) to more animal-like or simply weird (Figs. 5.28, 5.29). The second category consists of trees, ornaments for trees, and creatures associated with trees, birds especially. Though neither exhaustive nor mutually exclusive (human figures appear for instance on the tree base in Figure 5.40), the two categories are a convenient aid to discussion. A small number of images, not obviously related to either of the two categories, might be tentatively put in a third. The discussion in this chapter will describe the iconography of the images, attempt a reconstruction of their original appearance, and briefly comment on the likely purposes behind their making and their destruction at the sacrifices. 5.1 Iconography and reconstruction: human heads, figures, and masks This first category comprises 13 heads and two masks from K1 that bear human- like features (Figs. 5.3–5.5, 5.10–5.12)259, and many more from K2, including the monumental statue of a life-sized figure standing on a pedestal (Fig. 5.17). Others from K2 include 44 heads (Figs. 5.6–5.9, 5.13–5.16), one figure wearing an animal-shaped headdress (Fig. 5.18), one kneeling figure bearing a zun on the head (Fig. 4.31), three plaques in the form of a kneeling figure (Fig. 5.19), figures appearing on fragments of a 259For comprehensive inventories of the heads and masks from K1 see Beijing 1999a, pp. 23–33, and p. 449 table 1. 114 structure, possibly model of an altar (Fig. 5.43), and 20 masks (Figs. 5.26, 5.27).260 In addition, figures are depicted on a jade implement (Fig. 5.42). All of these human images share a set of standard facial features except for the head in Figure 5.3, which displays some marked differences to be discussed later in Section 5.1.7.261 5.1.1 Facial features The standard facial features of the bronze heads, figures, and masks may be summarized as follows, though individual renderings vary somewhat. Massive eyebrows stand in relief. Oversize slanting eyes take an almond shape, the upper edge of the eye socket relatively straight while the lower edge droops in a deep curve. Bulging eyeballs are creased from inner to outer canthus, resulting in two facets. The shape of the eye sockets and the creased eyeballs make the faces appear to look downward. Cheekbones are prominently marked by two sweeping curves issuing from the nose. The nose is long and prominent. The mouth is a thin horizontal groove with downward hooks at the corners, the jaw a sort of horizontal molding carried all the way around to the hinge behind the ears. Ears, like the eyes, are disproportionately large, and they stand nearly at the right angle to the face. Ear hollows are represented by an upward curling hook and below it a horizontal shield with a pointed tip. Ear lobes are always pierced. At the temples, sideburns are often indicated by raised rectangles. The emphatic sharp-edged features and downward gaze give the faces similar expressions: always solemn, sometimes even severe or supercilious. The angular features 260For comprehensive inventories of the heads, figures and masks from K2 see Beijing 1999a, pp. 162–95, pp. 231–5, pp. 466–7 table 13, and pp. 468 table 14. 261In addition, figures illustrated in Figures 5.20–5.24 all lack heads or upper bodies, but they are related with the images listed here in clothing, pose, and hand gesture. 115 also seem to indicate that the faces wear masks. The masks vary in some details, however. The head in Figure 5.5 appears to wear the most complicated one: it has two fins projecting backward from the front of the head and then descending behind the ears to join the molding of the jaw, giving the effect of a sort of border running around the face. The head in Figure 5.6 shows another form of mask whose sharply triangular sides are indicated by deep grooves in the bronze. This head as well as three others, including the one in Figure 5.16, have their masks highlighted in gold foil.262 5.1.2 Hairstyle and headdress While the heads share similar facial features, there are clear differences among them in hairstyle and headdress.263 The differences are expressed not so much from one individual head to another as at the level of the group. Two groups may be recognized on present information. One features a hairstyle with clipped hair across the back of the head (Figs. 5.3–5.5, 5.8, 5.9, 5.17e, 5.18, 5.40, 5.43),264 the other a tapering pigtail of braided hair tied near the top by a horizontal band (Figs. 5.10–5.16).265 Heads in the clipped 262The other two heads covered in gold are K2(2):137 and K2(2):115. For K2(2):137 see Beijing 1999a (p. 184 fig. 100, p. 186 pl. 64.4), and Bagley 2001 (p. 95 fig. 14.4). For K2(2):115 see Beijing 1999a (p. 183 fig. 99, p. 186 pl. 64.3), and Bagley 2001 (no. 13). 263For useful summaries of hairstyles and headdresses as well as garments and other personal adornments seen on the heads and figures see Cai Ge 1995, Huang Jianhua 2001a, and Huang Jianhua 2002 (pp. 240–64). 264On the head in Figure 5.9, the hairline is indicated all the way around the head— clipped short above the forehead, trimmed to modest sideburns at the temples, and descending in back as far as the bottom of the ear. 265Rawson holds the opinion that it is not clear whether the pigtail is of cloth or hair, or a combination of both (Rawson 1996, pp. 65–7). 116 hairstyle commonly if not always appear with domed cranium (Figs. 5.5–5.7, 5.9),266 and they usually wear various elaborate headdresses. The two domed heads in Figures 5.6 and 5.7 each have a curved projection on the back that might be the edge of a tight-fitting cap or helmet; or it could be a fringe of hair since the projection does not continue around the head. The head in Figure 5.9 wears a braided crown or garland, set around a stepped-back edge on the top of the head. Similar braids are apparently depicted on the kneeling figures on the altar in Figure 5.43. The heads in Figures 5.3 and 5.4 both have a stepped-back edge at the top, too, the latter also having four holes. These edges were evidently meant for securing some kind of headdress. The head in Figure 5.8 wears a headband decorated with rounded narrow rectangles in sunken line. This headband is the same as worn by the life-sized figure in Figure 5.17. The figure’s headdress further includes on top of the headband a circular plaque, now partly damaged, featuring a pair of monstrous eyes (Fig. 5.17f), which is an enlarged version of the eyes seen on the four animals forming the decorated part of the pedestal (Fig. 5.17k).267 The two large lozenge-shaped pupils are clearly visible on the sides, and the inner canthi of the eye sockets meet at the front center, touching the top edge of the headband. The eyebrows are partly missing, as are the upper parts of the eye sockets. Perhaps the head in Figure 5.8 had a similar headdress, but with its upper part made in a perishable material now lost. Furthermore, at the back of the standing figure’s head are two rectangular holes set on a diagonal, obviously meant for attaching another 266Other published heads with domed cranium are K2(2):137 and K2(2):63. For K2(2):137 see note 262 for references. For K2(2):63 see Beijing 1999a, p. 182 fig. 97, p. 186 pl. 64.1. 267The identification of the plaque as featuring eyes was made by Alain Thote, who kindly shared it with me in a personal communication on April 11, 2003. 117 ornament in the figure’s hair. The head in Figure 5.7 provides just one such example: the diagonally placed ornament takes the form of a curved tube with flaring ends and a constriction around the middle, probably imitating a piece of cloth that tied it to the hair or hat. The tube has two open ends, which suggests that it held some further ornament. Similar hole(s) for the purpose of mounting ornaments are seen on the heads in Figures 5.5 and 5.6 and on the gesturing figure in Figure 5.17.268 This gesturing figure also has the most fantastic headdress so far known, one shaped like an animal’s head. Interpreting this bizarre hat is made easier if we look at the winged animal that supports the small altar in Figure 5.43. What the present figure wears as a hat is clearly equivalent to the head of the animal that supports the altar: we see the same large eye on the side of the head, the same pattern on the throat, three freestanding projections on the top, a star inside a circle near the mouth, and at the front end a large open mouth. Of the three projections, the one at the front of the head looks like an elephant’s curling trunk while the two on the side resemble floppy ears. Knowing that elephants were important ritual offerings at Sanxingdui and that they probably appear on the pedestal of the life-sized figure (see discussion in Section 5.1.4), we might begin to see the animal here as a stylized elephant. Yet another type of headdress can be seen on the miniature figures kneeling on the base of the fragmentary tree in Figure 5.40: the figures have a clipped hairstyle and their headdresses are distinguished by hooked projections rising from the back. The same headdress is worn by the kneeling figure in Figure 5.19, and a somewhat more elaborate version appears on a strange hybrid in Figure 5.32. 268The remnant of an ornament appears at the back of the head K2(2):137. See note 262 for references. 118 In the altar shown in Figure 5.43, the standing figures with clipped hair sport a prominent headdress that appears like an open book, its lower edge decorated with squares formed by volutes. Finally, the standing and kneeling figures depicted on the jade implement in Figure 5.42 show still more examples of headdresses, and the design chased on the gold sheath in Figure 7.22 displays a type of pronged headdress. In these two pictures, however, the hairstyles cannot be known. As mentioned earlier, the other hairstyle features a tapering pigtail. In two examples the pigtail appears to be part of the headdress, for it continues from the cap or headband indicated by the slightly raised edge on the smooth upper part of the head (Fig. 5.10).269 All the heads of this type are flat on top, and they appear to have no additional headdress, although we cannot exclude the possibility that headdresses made of a perishable material were originally mounted on the flat top. In any event, this type of head appears to be more common than the heads with clipped hair and elaborate headdresses. Of the 13 heads from K1, four are of the type with clipped hair while nine have the flat top and pigtails. Of the 44 heads from K2, a mere six are of the first type while 38 are of the second. Besides, the two gesturing figures in Figures 5.17 and 5.18 have clipped hair and the most elaborate headdresses. Given the similar facial features and expressions, it is quite possible that the hairstyles and the headdresses were a primary means of indicating status, the heads with clipped hair being generally higher in rank. 5.1.3 Coloring of facial features 269The other head is K1:11 (see Beijing 1999a, p. 27 fig. 13, p. 30 pls. 3.3 and 3.4, p. 524 color pl. 3). 119 In addition to modeling in shape, there is clear evidence that all the facial features were originally highlighted in colors. The eyebrows, eye sockets, pupils, and sideburns were painted in black; the nostrils, lips and ear hollows were painted in red; and the pigtail was painted in either color. The kneeling figure in Figure 5.19 retains the strongest, most substantial traces of black paint. Similar traces survive in various degrees on numerous other images listed below; the traces are often very faint, and sometime they are completely lost but known to have existed at the time of the excavation.270 (1) Eyebrows: heads, K2(2):2, K2(2):12, K2(2):47, K2(2):48, K2(2):53, K2(2):55, K2(2):58 (Fig. 5.7), K2(2):72, K2(2):78, K2(2):118, K2(2):214 (Fig. 5.6); figure, K2(3):04 (Fig. 5.19); masks, K2(3):57, K2(2):60, K2(2):102, K2(3):119, K2(2):153 (Fig. 5.27), K2(2):331. (2) Eye sockets or pupils: heads, K2(2):12, K2(2):47, K2(2):53, K2(2):78; masks, K2(3):57.271 (3) Eye sockets: heads, K2(2):51, K2(2):58, K2(2):118; figure, K2(3):04; masks, K2(2):60, K2(2):102, K2(2):331. (4) Pupils: head, K2(2):48; figure, K2(3):04; masks, K2(2):153, K2(2):331. 270The final excavation report—Beijing 1999a states that of the forty four heads from K2 all the relatively intact ones retain traces of cinnabar on the ear hollows, nostrils and lips, and traces of black paint on the eye sockets, eyebrows and pigtail (p. 169); and that the less corroded images among the masks and the plaques bearing taotie-like faces can be seen to retain traces of black paint on the eye sockets, pupils and eyebrows (p. 188). However, the report is inconsistent in describing individual artifact conditions at the time of the excavation. The summary provided here combines information from the report and from my own examinations. Besides facial features, it is mentioned in Beijing 1999a (p. 169) that the fragment of a hybrid figure standing on birds K2(3):327 retains traces of cinnabar on the figure’s skirt as well as on the birds beneath the figure’s feet, and traces of black paint in the decoration on the shanks; this fragment is illustrated here in Figure 5.22. 271The information about these images from Beijing 1999a (pp. 466, 190) does not specify whether the traces of black paint survive on the eye sockets, or pupils, or both. 120 (5) Nostrils: heads, K2(2):48, K2(2):51, K2(2):58. (6) Ear hollows: heads, K2(2):34 (Fig. 5.14), K2(2):53, K2(2):58.272 (7) Lips: heads, K2(2):34, K2(2):51, K2(2):58, K2(2):118, K2(2):214; masks, K2(3):57, K2(2):102, K2(2):153. (8) Sideburns: heads, K2(2):48, K2(2):55; figure, K2(3):04; mask, K2(2):102. (9) Pigtail (black): heads, K2(2):34, K2(2):51, K2(2):118; (red): heads, K1:10 (Fig. 5.12), K2(2):53, K2(2):72. These colorings would surely have formed a vivid contrast with the golden color of the uncorroded bronze. Perhaps the colors were meant to give sensory powers to the images, or to heighten the impression of these powers already emphasized by oversize eyes, noses, ears, and mouths. A related phenomenon is the masks of gold foil that cover some of the heads (Figs. 5.6, 5.16). At least seven heads may have once had such masks, the reason for them uncertain.273 If the gold mask were applied when the head was newly cast, its color would probably not look too different from the color of the bronze, which depending on its composition might be a paler yellow, or a trifle reddish. More likely, the gold mask was applied when the head had been in use for some time and the golden color of the bronze had dulled or disappeared due to corrosion. Application of the gold mask, by once again giving the head a glowing sheen, might have served to bring it new life. This speculation seems to find some corroborating evidence in the gold head in Figure 272Curiously the head K2(2):12 seems to show traces of black paint in the ears (see the description at Beijing 1999a, p. 466). 273See note 262 for the other two heads from K2 covered with gold foil. In addition, there are two gold masks from K2, K2(2):62-1 and K2(3):147, and one from K1, K1:282. Those gold masks probably belonged to heads as well. For K2(2):62-1 see Beijing 1999a, p. 355 pl. 135.3. For K2(3):147 see Beijing 1999a, p. 352 fig. 194, p. 355 pl. 135.2. For references to K1:282 see note 163. 121 5.6. The head appears to retain some trace of red paint in the groove of its mouth now exposed by the partly damaged gold foil. The paint must have been applied before application of the gold foil was envisaged, for otherwise it seems unlikely that colors would have been painted directly on bronze in areas that would remain hidden under the gold mask. In other words, the gold mask was probably a later addition to an object that had been in use for some time without it.274 5.1.4 The life-sized statue The most prominent individual among the heads and figures is naturally the lifesized gesturing figure standing on a high pedestal (Fig. 5.17). As the only surviving example of a complete sculpture, it provides a wealth of iconographic details and an essential key for reconstructing the likely appearance of the human heads before they were sacrificed. The figure, probably a man, is immediately striking for its oversize hands forming two large circles obviously meant for holding something. What might he have held? We know that he was buried among sixty-seven elephant tusks, and noticing that the two hands do not form a straight line between them but rather a curve, it has been suggested by various authors that he held an elephant tusk in a gesture of offering or to signify his power.275 The man’s head has the same blocky shape and sharp-cut features as the other heads; his sumptuous headdress or crown has been introduced in the preceding discussion. 274The left ear of the head in Figure 5.16 retains traces of a discolored dark substance tracing the contours of the ear hollow. The coloring must have been intended for the same purpose as that applied directly on bronze. 275See e.g. Bagley 1990 (p. 62), Barnard 1990 (p. 256), Rawson 1996 (p. 61). A few other suggestions, less plausible, are summarized in Rawson 1996 (p. 63 note 2), and Huang Jianhua 2002 (pp. 97–8). 122 The now sightless eyes must originally have been animated with touches of paint, and other features would surely have been similarly highlighted. The man is bare-footed yet so lavishly dressed that we might wonder whether the bare feet had some ritual significance. The other figures, whether standing or kneeling, all seem to wear shoes (Figs. 5.19, 5.20, 5.23, 5.24, 5.40, 5.42). Perhaps this man stood on a piece of sacred ground that required him to remove his shoes for ritual reasons. On close examination, his costume can be seen to consist at least of three layers of clothing. The outermost layer is a short garment that ends below the waist, its lower hem indicated by a raised horizontal line crossing the surface of the statue. Sewn with a heavy braid or cord along its top edge, the garment is wrapped around the left side of the body under the left arm, over the right shoulder, and knotted by the braid at the back. The small hole in the back between the two knots was perhaps meant for a fastening device. The two ends of the piece of cloth almost meet under the right shoulder. The structure of this garment becomes confusing at this point, however. It appears to be a mantle, but as such it would normally be fastened over the shoulder by the corners, yet there are no such corners to be seen on the right shoulder: the braid runs smoothly over it. In their study of the costume, Wang Xu and Wang Yarong propose that the mantle is attached with a sleeve at the right shoulder (Fig. 5.17g),276 but the sculpture does not supply any conclusive indication of this, as the would-be edge of the sleeve seems to coincide with the solder that joins the figure’s right arm to the shoulder.277 This outer garment is decorated with feathery, fantastically ornate dragons, each with two enormous clawed feet (Fig. 5.17j). Due to the way the clothing is tailored, the 276Wang & Wang 1993, p. 60. 277The casting of this figure will be discussed in detail in Section 6.2.2.3. 123 dragons are not easily visible from the front but are wrapped around the left side of the body. They are bordered on each side by a column of curly patterns and beyond it another column of patterns that look like the lozenge-and-prong design on the undergarment’s skirt reduced in size and turned sideways (see below). The middle robe is mostly covered by the outer garment, but it is exposed on the neck, shoulders, and arms. It has a V-shaped neckline on both front and back. A dragon on the back of the left shoulder must belong to it (Fig. 5.17e). The robe’s short plain sleeves end at the middle of the upper arm, and its open edges can be glimpsed under the right arm. This robe is shorter than the outer garment (Fig. 5.17h). It is not clear whether the two small openings under the right arm serve any purpose related to the garments.278 The inner garment is long, making its lower part visible, and it consists of two pieces. One skirtlike piece reaches just below the knee in the front, the other piece in the back partly overlaps the front piece. This back piece reaches to the calf in back and its dovetails descend to the ankle at the sides. The decoration on these two pieces is fairly simple: on the skirtlike part exposed below the hem of the mantle we see a horizontal band with a pair of large lozenges, probably eyes, the shape comparing closely with that on the figure’s crown. The eyes have a shield between them and are flanked by curly patterns. The same design appears on the back piece of the inner garment. Below this band is a much broader register decorated with two units of pattern, each composed of a pair of eyes with pronged devices hanging from it. This pattern seems to be the source from which the patterns along the edge of the outer garment derive. 278Wang Xu and Wang Yarong suggest that the upper opening was for fastening the outer garment while the lower opening for fastening the middle garment (Wang & Wang, pp. 60–1). 124 Nothing of this inner garment is visible at the neck, but two long sleeves are exposed under the sleeves of the middle garment and reach all the way to the figure’s wrists where they end with triple-band cuffs. The sleeves are boldly decorated, the designs again centered around eyes, one on each elbow. The boldness of the decoration forms a strong contrast with the designs on the lower part rendered in thin lines, suggesting that the sleeves belong to an upper part of the inner garment that is separate from the lower part, or to yet another layer of clothing (Fig. 5.17i).279 Braids at the ankles similar to that sewn along the upper edge of the outer garment suggest that the man might also wear a pair of pants under the undergarment. Alternatively, the braids might be anklets. The varying density of the patterns decorating the garments seems to suggest different textile techniques. The patterns on the long lower part of the inner garment appear simple enough to have been woven, but the dragons on the outer garment look like embroidery, as do the even bolder designs on the arms. The gesturing man stands on a two-part pedestal that adds a meter to his height. The upper, decorated part of the pedestal is composed of a block and an open four-legged support on which the block rests. The block has four rectangular sides, each decorated with a Style IV design featuring an eye centered among curls and diagonally extended filaments. This appears to be a peculiar rendition of the eye-and-diagonal patterns commonly seen on bronze vessels of the Erligang and transition periods. The design is set against a background of spirals and framed by a border of small circles—an Erligang invention. Below, the decorated support is apparently meant to be viewed not straight on 279Wang Xu and Wang Yarong propose that the inner garment has separate upper and lower parts (ibid., p.63). 125 but on a diagonal (Fig. 5.17k): in that view, prominent eyes left and right of the corner combine to suggest an animal face with a snout or elephant-like trunk. Above the eyes, the elements that touch the block suggest ears (compare the ears and eyes of the mask in Figure 5.28), while the notched borders along the trunk indicate teeth (compare the teeth of the plaques of taotie faces in Figure 5.51).280 The remainder of the pedestal is a trapezoidal box without bottom. It was left undecorated perhaps because it was sunk in the ground to secure the statue. In every respect this statue seems distinctly more ambitious than other Sanxingdui images. Few have nearly so much surface decoration. The shallowly drawn patterns on the clothing, much simpler than the intricate high relief of the imported zun and lei vessels reviewed in the previous chapter, probably take their inspiration from bronzes of the transition period.281 The rectangles of decoration on the block at the top of the pedestal belong to the same period. 5.1.5 Installation of the heads and their likely appearances In their original setting, how would the heads have been mounted? They obviously could not stand on their own, given their pointed necklines front and back.282 Now we may recall that the middle garment on the life-sized figure has a same V-shaped neckline, leaving bare an area corresponding to the pointed necks of the bronze heads. 280The identification of the creatures as elephants was first made by in Rawson 1996, p. 62. 281Bagley 1999, p. 217. Compare as regards draftsmanship the patterns on a bronze drum from southern Hubei (Bagley 1999, p. 151 fig. 3.5; better illustrations in Fong 1980, pp. 114–15, and Beijing 1998d, p. 171, pl. 178). 282Some of the heads have lost the pointed tips of their necks due to burning and/or breakage, e.g. Figs. 5.3, 5.4, 5.6, 5.10–5.12. 126 Given also the similarities in facial features between the heads and the figure, it is easy to imagine that the figure translates into bronze an image of which only the head normally was bronze. Put another way, we may speculate that all of the heads were originally mounted on bodies made of another material. Those bodies need not all have been standing like the statue; some might have been in the sitting or kneeling postures known from other Sanxingdui images (see e.g. Figs. 4.31, 5.19, 5.24, 5.40, 5.42, 5.43). Therefore, instead of one single statue, there probably were a large number.283 These figures would surely have been dressed, either in real garments or depicted ones, and those garments might vary according to status. Besides the complicated clothing described above, simpler types are depicted on other figures (Figs. 5.18–5.24, 5.40, 5.42, 5.43), but to some extent the simplicity might be due to the small size of those figures, which would make fine detail different or impossible. In general, the V-shaped neckline and the left lapel lying over the right seem to be standard (e.g. Figs. 5.18–5.20, 5.40, and the middle garment of the life-sized figure in Fig. 5.17h).284 One popular dress appears to be a robe that ends just above the knees and is tied at the waist by a girdle; in some images, a fastening device for tying the girdle is clearly depicted (Figs. 5.20, 5.40, 5.43).285 This type of robe appears to have either long sleeves reaching to the wrists or 283Goepper was the first Western scholar to suggest the possibility that the bronze heads and masks were installed on bodies made of perishable materials (Goepper 1995, pp. 254–5). The suggestion had earlier been made in Chen De’an 1992 (p. 42), the author noting the correspondence between the V-shaped necklines of the heads and of the statue and other figures; the same suggestion was made also in Zhao Dianzeng 1993 (p. 84) and Li Song 1993 (p. 103). 284To be precise, the backs of the figures in Figures 5.18 and 5.20 show an arc rather than a V, but this seems to be a minor variation. 285The same fastening device is seen on the kneeling figure in Figure 4.31 though the figure is bared to the waist. The device is put opposite in direction to the one seen in Figure 5.20. 127 no sleeves (Fig. 5.20 is the clearest example without sleeves). A somewhat different garment is worn by the headless figure in Figure 5.21, its lower part looking like a kilt, perhaps some form of protective armor. The figures engraved on the jade implement wear triangular skirts, but the design is too schematic to provide much detail. The robe worn by the figure with an animal headdress in Figure 5.18 must originally have extended below the waist as well, since the girdle is partially visible on the fragment. In some cases, figures might be unclothed above the waist like the figure bearing a zun on the head in Figure 4.31. In this figure, the emphasis on the nipples perhaps suggests a woman, but her facial features are generic, matching those of the other heads. Since we could not identify this person as a woman from her face alone, we must wonder whether some of the bronze heads were mounted on female bodies. Did women play a significant role in the Sanxingdui sacrifices? Since the kneeling figure is the only surviving example with a possible indication of gender, only future excavation can answer the question. Among the robed figures, a few appear to wear undecorated garments (Figs. 5.19, 5.24, and the kneeling figures on the altar in Fig. 5.43). The decorated garments commonly share the pattern of rectilinear volutes (Figs. 5.18, 5.20–5.23, 5.40), the most detailed version being seen on the hybrid figure with clawed feet, which is rendered in finely detailed Style IV. Two other patterns can be found: whorls decorate the robes worn by the standing figures on the miniature altar in Figure 5.43, and a motif centered on a single large eye appears at the top of each sleeve in Figure 5.18. This boldly drawn eyed motif is comparable to that on the arms of the life-sized figure in Figure 5.17. Interestingly, a similar design featuring eyes appears on the shanks of the hybrid figure with clawed feet (Fig. 5.22). As the design on the shanks does not 128 show a clear edge suggestive of the hems of a garment, it might represent tattoo instead of a pair of patterned trousers.286 The legs of a few other figures also bear patterns, but it is hard to judge whether they are tattoos or trousers (Figs. 5.20, 5.23, 5.40, 5.43). Assuredly, however, these figures share a motif connected with a type of diamond-shaped bronze plaque (Fig. 5.25). The hollow-backed shape and small perforations of these plaques indicate that they were appliqués once mounted on something else. Appearing either in one piece or in triangular parts, these appliqués commonly take the form of a diamond shape encasing a roundel. This form matches exactly the patterns seen on the shanks of the figures listed above, raising the possibility that the appliqués were placed on the legs of large figures made of some material now perished, some standing and some kneeling. This recognition in turns gives further support to the hypothesis that the bronze heads were mounted on bodies of another material. What the diamond-shaped appliqués represent is a puzzle, however. If they represent eyes, as the Sanxingdui excavators believe,287 it is curious that their form is geometrically so regular, differing sharply from the eyes on Sanxingdui images as well as from other appliqués found in K2 that are instantly recognizable as eyes for mounting on faces made of another material.288 Equally puzzling is the fact that some of them were made in one piece, some in two, and some in four. 286Liu & Capon 2000 (p. 86) also suggests that the patterns may indicate actual tattoos, without giving a reason. 287In Beijing 1999a (p. 207), these appliqués are classified as “eye-shaped objects” mounted as eyes on supernatural images. 288For these appliqués see Beijing 1999a (pp. 201–7), and Tokyo 1998 (p. 89, no. 40). Huang Jianhua has also observed the difference, and proposes that the roundels represent the sun; see Huang Jianhua 2001b (p. 45), and Huang Jianhua 2002 (pp. 163–4). 129 In addition to their garments and perhaps body tattoos, all the figures have pierced ears and probably wore earrings. Three types of earring are depicted among the Sanxingdui imagery. Two are shown in the picture on the jade implement, one shaped like a bell, the other made of two interlocked circular rings (Fig. 5.42). The third type is worn by the laughing faces on the gold sheath in Figure 7.22b: it is shell-shaped and marked with two horizontal bars. Among the many small bronze implements provided with suspension holes or suspension apparatus found in K2 are a number of small bells and shells (e.g. Figs. 5.33, 5.34). Some might have been worn by the figures, although it is also possible that they were meant instead for trees (see Section 5.2). Perhaps such ornaments were shared between the figures and the trees. Like the life-sized figure, the heads mounted on lost bodies might have gestured with their arms and hands. Again, several gestures are represented on smaller figures. The figure with an animal-shaped crown in Figure 5.18 resembles a reduced version of the life-sized man, but the hands are placed slightly differently, suggesting that they held two distinct objects instead of a single elephant tusk. A similar gesture was probably made by the kneeling figures on the tree pedestal in Figure 5.40.289 The standing figure in Figure 5.20 holds his hands in a similar way but with the palms facing down. Many other small figures hold the hands together before the chest, clearly to grasp a single item; these include the figure in Figure 5.21, the ones in the picture on the jade implement in Figure 5.42, and those on the miniature altar in Figure 5.43. The standing figures on the altar hold rope-like objects whose complete shape is unfortunately not known (might they represent tree branches?). Another figure that is 289Both the surviving figures on this pedestal have lost their arms. 130 very illuminating in this regard is the miniature kneeling figure presenting a forked blade (Fig. 5.24). It is the only evidence at Sanxingdui bearing on the manipulation of jade and stone forked blades like the one in Figure 7.10k. With their faces painted, earrings dangling from the ears, their bodies dressed in real or depicted robes or partly bared and perhaps tattooed, and some further crowned with headdresses, the figures of which now only the heads remain would have formed an awesome gathering. 5.1.6 Masks All but one of the human masks are uniform in shape (Figs. 5.26, 5.27, 6.11).290 The word “mask” is used here to describe a shell of bronze, U-shaped in horizontal section, with no top, back, or neck. The facial features of the masks match those of the heads exactly, though the proportions of the larger ones (Figs. 5.26, 5.27) are much broader. The largest intact one, shown in Figure 5.27, has a calm gaze made all the more impressive by facial features enlarged to the point of crowding the surface: the eyebrows touch the upper edge, the eyes and nose fill half the face. Presumably these shells would have been mounted differently from the heads, perhaps fitted around architectural members and secured in place through the openings usually seen in their sides or back corners.291 290For the one unusually shaped mask with the same facial features, K1:20, see Beijing 1999a, p. 29 fig. 19.2, p. 37 pl. 6.1, p. 525 color pl. 6. It is a hollow-backed plaque. 291For a summary of possible uses and installations of the masks proposed by scholars in China see Huang Jianhua 2002, pp. 119–21. 131 Besides these human masks, K2 also yielded three masks with the same shape but with monstrous facial features (Figs. 5.28, 5.29).292 Though rooted in the basic iconography of the human images, the features are so wildly exaggerated as to make them the most weirdly supernatural of all the Sanxingdui images. They have pointed ears raised alertly and telescopic pupils that burst out of the eye sockets, signifiers of perhaps acute hearing and sight. Like the human images, they have traces of black paint on the eyebrows and eyes and red paint on the lips.293 Other facial features are drawn in sweeping curves; the usual ridges on the cheekbones spiral elegantly into the corners of the beaklike nose. Unlike the soberly straight mouths of most of the other heads and masks, the mouth here smiles in a long arc that rises almost parallel to the cheekbone ridges. On the two smaller masks, the viewer’s attention goes first to the fantastic excrescence that rises periscope-like above the nose (Fig. 5.29). This trunk spirals inward at the top, has a double spiral at the bottom, and in the middle carries a quill-like element. A little below the spiral at the top are two tiny loops that might have held dangling ornaments. Given the curling top of the trunk, we might again wonder if some allusion to an elephant is intended. The opening in the forehead of the larger mask in Figure 5.28 must have been meant for an appendage, perhaps a similar trunk. Sun Hua further argues that these masks probably had birds’ bodies, for the noses, ears and eyes are similar to those of the bronze bird–human hybrids in Figures 5.32 and 5.43.294 The shape of the 292The third mask of this type, K2(2):144, is of the similar size and appearance as the one in Figure 5.29; see Beijing 1999a (p. 197 fig. 109), Hong Kong 1997 (pp. 92–3 no. 28), and Beijing 1999c (pp. 88–9). 293Beijing 1999a, p. 190. 294Sun Hua 2000, pp. 252–4. 132 masks and the square apertures at their back corners suggest that they were mounted architecturally, like the human masks. It is possible that additional masks of the same monstrous kind but made in a perishable material were destroyed during the sacrifice, for 33 columnar bronze pupils were found by themselves in K2.295 Alternatively the pupils may represent precast parts of masks that were never made. 5.1.7 Stylistic differences between images from K1 and K2 The only bronze images that K1 shares with K2 are human heads and a single Ushaped mask. The heads from the two pits are noticeably different. The most exceptional head is the one in Figure 5.3; it is set apart from all others by its soft, comparatively realistic modeling of the face. The lower jaw, nose, and cheekbones are all gently rounded rather than sharp-edged, and the mouth lacks the usual downward hooks at the sides. Nevertheless, the eyebrows, eyes and ears are still unnaturally large, connecting this head with the others. The reason for its soft features is not clear. Some scholars see a woman’s face296 or a face without a mask.297 The latter possibility seems more likely. Alternatively this head might be earlier than the others, belonging to a time when the facial features had not yet been standardized. In favor of this possibility is the fact that the heads from K1, thought fewer than those from K2, are more varied. The six heads from K1 illustrated here represent at least four distinct variations (Figs. 5.3; 5.4; 5.5; 5.10–5.12). The head in Figure 5.3 has already been discussed. The 295Beijing 1999a, pp. 209–17. 296E.g., Tokyo 1998, p. 152. 297Liu 2000, p. 3. Or perhaps it wears half a mask covering part of the face. 133 head in Figure 5.4 has a uniquely shaped nose, exceptionally long and hooked. The head swells slightly toward the top, distancing it from the stovepipe proportions of the heads in Figures 5.10–5.12 and other flat-topped heads from K2. This head probably set the precedent for the facial proportions of the heads with domed cranium in K2, which are distinctly less cylindrical than the flat-topped ones. The heads in Figures 5.5 and 5.10 share similar facial features, both having proportionally small eyes as compared to the others, and their triangular noses are short but especially sharp. This rendition of the facial features does not appear among the heads from K2. The heads in Figures 5.11 and 5.12 are tall and almost cylindrical, with proportions typical of the flat-topped heads from K1 as well as from K2. The nose of the head in Figure 5.11 is softer, rather fat, by comparison with those in Figures 5.4, 5.5, 5.10, and 5.12. Instead it is clearly related to the nose of the head in Figure 5.3, which is even more plump. This type of nose is also typical of the flat-topped heads in K2, and of the domed heads as well. On the other hand, the facial features of the head in Figure 5.12 are crowded toward the bottom of the cylinder, a trait seen again among the flat-topped heads from K2. The heads from K2 are far more numerous but much less varied in the rendering of the facial features. (The vessels, too, are more numerous and less varied than those in K1.) The heads may be broadly divided into two types according to their proportions: those with domed cranium are somewhat more natural, less cylindrical in shape, while the flat-topped ones appear like cylindrical stovepipes. The facial features tend to extend higher up the head on the domed ones, and to crowd towards the bottom on the flattopped heads. The head in Figure 5.14 is typical of the second type. The effect of compressing the mouth and chin emphasizes more than ever the enormous eyebrows and 134 bulging creased eyes on this head and similar ones from K2. Among the flat-topped heads from the pit, the one in Figure 5.13 is rare in having less emphatically modeled facial features, which give it a comparatively mild expression. This head shows a clear affinity with the head in Figure 5.11 from K1. The preceding comparisons strongly suggest that, in the interval separating K1 and K2, the Sanxingdui casters had standardized the designs of the heads. This possibility is strengthened by the weight of the heads. Again, the heads from K1 show a greater variation: excluding the four heads that had been severely damaged, the remaining nine heads weigh between 2.01 kg and 7.66 kg (two between 2 and 3 kg, two between 3 and 4 kg, three between 4 and 5 kg, one between 6 and 7 kg, and one between 7 and 8 kg). In contrast, the heads of similar sizes from K2 show more consistency. Excluding the three miniature heads and another three whose weights are unpublished, the majority of the remaining thirty–seven fall between 2 kg and 4 kg (twelve between 2 and 3 kg and sixteen between 3 and 4 kg); a minority of 8 heads weigh between 4 and 5 kg, and the heaviest head weighs 5.8 kg. Significantly, none of the heads weigh as much as the heaviest from K1, the one in Figure 5.10; at 7.66 kg, this head is twice the weight of most of the heads from K2. By the time of K2, it seems, the Sanxingdui founders had managed to make their castings more consistent and economically thin. Further evidence of standardization will be seen in the analysis of the casting technology in Section 6.2. On a separate note, the differences between K1 and K2 discussed here give additional support to the relative dating of the two pits discussed in the previous chapters. 5.2 Iconography and reconstruction: trees and associated images and implements 135 The second major category of bronze images consists of trees, creatures associated with trees, and ornaments for trees (Figs. 5.30–5.41). All come from K2, the largest being the four-meter-tall tree in Figure 5.30. However, none of the trees survive intact, all were severely battered and burned before burial, and parts of some of them are missing from the pit, never having been deposited there. Figure 5.2 shows some of the tree fragments. For obvious reasons it is difficult to restore the trees and impossible to return many ornaments and creatures to their proper places. We cannot even be sure how many trees were sacrificed. Nevertheless, careful reconstruction by the Sanxingdui restorers has revealed the basic features of the largest surviving tree (Fig. 5.30). This monumental example will be the focus of this section; clues from other objects will be utilized to help visualize details missing from it. The tree stands on a pedestal in the shape of a ring surmounted by a heavy coneshaped tripod. Smooth arches sweep from one leg to the next; a strip at the lower edge of each arch is decorated with an eye-and-diagonal pattern that we have seen several times already, an Erligang and transition period design. Above the base, nine branches are attached to the trunk in groups of three at three levels. Each branch produces a flower, or possibly a fruit, at the top of its arc, on which sits a bird with hooked beak; a second flower grows at the tip of the branch. It is the branches, flowers, and birds that unmistakably identify the object as a tree. In each group of three branches, one bifurcates and produces three flowers instead of two. The bifurcating branches are all on one side of the tree. On the other side, a dragon standing on the circular base sends a rope-like body undulating up the trunk; the dragon is secured to the trunk by two struts. The birds on the branches have large hooked beaks, a feature that characterizes all the locally made bird 136 images at Sanxingdui. Their low crests are more unobtrusive. The eyes are slightly oval, nearly circular (the eyes of other birds are more clearly oval-shaped with pointed eye sockets, e.g. Figs. 5.31, 5.44, and 5.45). The feathers on the body and tail are indicated by openwork patterns. Interestingly, all the wings are broken off, leaving behind large stumps; this looks deliberate and was perhaps part of the sacrifice. Magnificent as it is now, in its original condition the tree would have been an even more astonishing sight. Some sort of finial must have stood on top, perhaps a bird like one of the nine others perched on the tree, or a more showy one like the bird in Figure 5.31 (this has peacock-like plumes on the crest as well as on the upper and lower sections of its tail), or perhaps an even more fantastic creature like that in Figure 5.32, a bird whose head has human-like features. The creatures in Figures 5.31 and 5.32 are fragments of miniature trees that stood either on a branch or on top of a tree trunk. The rest of the large tree must have been heavily decked with pendants: small perforations from which ornaments could have been hung are seen in many places, such as the beaks of the birds and the openwork projections of the downward-facing flowers. Not surprisingly, K2 yielded many small bronze implements provided with suspension holes or suspension apparatus, among them little bells, thin shells, and foils of various shapes (e.g. Figs. 5.33–5.36). However, the locations on the tree where these objects should be hung, and indeed on which tree they belong, cannot be determined.298 Besides the various pendants, the tree probably also had ornaments that were movable but permanently attached, such as disks of jade or bronze strung on the trunk or the branches before the branches were attached to the trunk. K2 yielded three examples 298For a comprehensive list of the pendants see Beijing 1999a, pp. 289–325. 137 belonging to other trees: one bifurcated branch strung with two bronze rings (Fig. 5.40b), one trunk attached with a bronze collared disk (Fig. 5.37), and a third with a jade collared disk encircling the calyx of a flower (Fig. 5.38). These collared disks are similar in form to the fixed openwork rings encircling the calyxes at the tips of the branches of the large tree. K2 also contained nearly a hundred bronze collared disks, some circular in form, others roughly rectangular (Fig. 5.39), and ten jade collared disks (Fig. 7.3). Perhaps it is not too farfetched to imagine that at least some of these bronze and jade disks belonged to the trees, although none is exactly the same in shape as those still attached to tree fragments. K1 similarly contained more than a hundred collared disks in bronze and a few in jade. Perhaps in K1 real trees rather than bronze ones were burned and sacrificed, their branches hung with bronze and jade disks; perhaps the large amount of ash in the pit resulted from the burning of the trees (as well as, perhaps, other images such as human figures). The dragon’s body is incomplete, rising only about halfway up the tree. Originally it must have reached the top, doubling its present length, for the stumps of two more struts still survive, one of them near the top. This dragon—so called merely for convenience—is an odd creature, not obviously related to the animals on bronzes or in jade from elsewhere in China, much less to the Chinese dragons of more recent times. Its head is quite flat, with features depicted on both sides: a large round eye and an open mouth with circular markings representing teeth. Two rhinoceros-like horns billow above the snout, but the ear behind the horns is now half missing. The dragon seems to have a small wing and beneath it a small leg on which it balances. Although only one wing and one front leg are shown, pairs should probably be understood. The dragon’s long body 138 issues from beneath the wing. In its undulating course toward the top of the tree, the body carries a couple of small hooks or prongs, and it sprouts some extraordinary pendants, two shaped like feathers or knives, one shaped like a human hand with long graceful fingers. This dragon image recalls the dragons decorating the outer garment of the lifesized figure in Figure 5.17j. The comparison helps us imagine what the broken ear on the present dragon must have looked like: a curve extending backwards and ending with a prominent hook, similar to the ears of the human-bird hybrid in Figure 5.32 or to those of the dragon standing on a tube in Figure 5.47. Interestingly, the dragons on the mantle have claws remarkable for their disproportionately large size and their shape, resembling clenched human fists. The long-fingered hand of the present dragon must similarly stand for a claw as well; and the arched part from which the hand emerges must represent a back leg (it looks remarkably like the leg of the small bird in Figure 5.31). Immediately below the hand and on the other side of the dragon’s S-curved body is what looks like the remnant of the other hind leg. Given the smaller space there, the missing leg and the hand-shaped claw would clearly be smaller, just as the knife-shaped pendant on this side is smaller. Like the heads, figures, and masks, the tree must in its original state have been brightly colored. Examination reveals a few traces of red paint surviving here and there, and the newly cast bronze would have had a golden sheen. Other items on which colors still survive include the large bird’s head in Figure 5.44 and the taotie faces in Figure 5.51. It is possible that all the Sanxingdui bronzes were originally painted, including the trees. 139 The large tree helps identify other tree fragments among the bronzes from K2. We have already seen the hybrid in Figure 5.32 and the bird in Figure 5.31. There are more small birds and small dragon heads (some resembling the dragon on the large tree) as well as flowers and fruits.299 Another group of fragments, shown in a reconstruction drawing in Figure 5.41 and probably belonging to one tree, is very strange—a sort of bush with rope-like branches on which perch human-bird hybrids like the one in Figure 5.32. However, the support on which this bush must have stood cannot be identified among the bronzes from K2. The cone-shaped base in Figure 5.40 is clearly part of another tree. Judging by the ratio of the base diameters, this tree is likely to have been about half the size of the larger one. The fragment shown here comprises two sections, one comprising of the base, a portion of the trunk, and the stumps of three branches, the other comprising a short section of the trunk and a complete bifurcated branch attached to it. The branch has two flowers with a bird on one of them and two movable bronze disks. The remaining parts of the tree were apparently not interred in the pit. Like that of the largest tree, this base is a tripod of boldly arched forms, but here the openings in the arches are occupied by panels of openwork, in front of which three small figures kneel or sit on their heels in a formal seated posture. The arched shape so prominent on these tree bases is a recurrent motif on objects from K2, including the base of another tree (Fig. 5.2) and the jade implement in Figure 5.42. In view of the contexts in which the motif occurs, we might wonder whether it is meant to signify hills as 299Beijing 1999a, pp. 227–9 fig. 126 and pls. 84–5, p. 328 fig. 182, pp. 335–41. 140 proposed by Jenny So,300 hills where trees grow and where the figures on the tree base in Figure 5.40 perform a ritual. On a separate note, an interesting similarity may be remarked between the arches and the decorated part of the pedestal on which the lifesized figure stands (Fig. 5.17). The bases of the large tree and its smaller cousin in Figure 5.40 have upturned ends similar to the elephant trunks on the statute’s pedestal. Moreover, each elephant in that pedestal joins the next one to form an arch. Like the life-sized figure, trees were undoubtedly central to the sacrifice at Sanxingdui, even though we cannot be sure of their significance.301 5.3 Original ensemble of the images If the preceding discussions go some distance toward reconstructing the original appearance of the individual images, what might their display as an ensemble actually have looked like? In contemplating this section, we may first remember the varied sizes of the images, which range from monumental to miniature. The human heads are the most consistent in size: except for two miniatures from K2 (Figs. 5.9, 5.15), all are more or less near life-sized. The human masks are more varied. The complete ones can be sorted into three sizes: five are about 15 cm high, making them similar in scale to the lifesized heads; eight are about 26 cm high (Fig. 5.26); and a single large one measures about 40 cm high (Fig. 5.27). But an example of which only fragments survive must have been dramatically larger: a piece of its right ear measures 48 cm high, arguing that the mask as 300Bagley 2001, p. 162. 301Rawson comments on trees and associated images, “While it is possible that these images were the focus of worship, it is also possible that they were part of the depiction of an imaginary or spiritual landscape” (Rawson 1996, p. 76). 141 a whole was at least 80 cm high.302 The three monstrous masks come in two sizes. The human figures and the trees show the most drastic variations. The monumental size of the tree in Figure 5.30 raises the possibility of installation outdoors, whereas the altar in Figure 5.43 suggests that miniature items of all kinds were components of small models. Perhaps the temple(s) had chapels for such models; or some of the models might be portable shrines. In any event, the different sizes indicate ensembles at several different scales and probably serving different purposes. The carved design on the jade implement in Figure 5.42 and the bronze altar in Figure 5.43 provide revealing evidence suggestive of ways in which some of the images might have been arranged. The design on the jade implement, repeated four times on both sides, shows two units stacked one on top of the other in each end of the implement. One unit consists of a row of standing gesturing figures supported by a register of arches; the other unit is similar but the figures are kneeling. The format of figures standing on arches corresponds closely to that of the life-sized statue, while the combination of kneeling figures and arches recalls the tree base in Figure 5.40. All the figures clasp their hands in 302Chen De’an estimates that the mask was 80 cm high and 120 cm wide (Chen De’an 1992, pp. 38–9). 142 front.303 The two rows of figures and arches are partitioned by a register of rectilinear volutes. Obviously what is shown on the jade implement is a ritual ensemble.304 The ensemble in the altar (Fig. 5.43) is much more complicated than that shown on the jade implement, but its primary components are the same: a row of standing figures and a row of kneeling figures holding our their clasped hands in front.305 Similarities may also be found in other details: a row of rectilinear spirals (densely packed and beneath the kneeling figures this time) and a row of arches (on top of the standing figures). Another detail worthy of comment in this regard is the two quadrupeds with fabulous tails and wings carrying the platform in support of the standing figures. Again, the format reminds us of the life-sized statue. Moreover, if the heads of the quadrupeds are those of elephants, as suggested earlier in the discussion of Figure 5.18, we might see a further connection with the arches and, thus, with the design on the jade implement. 303The register below the standing figures contains between the arches a mysterious emblem whose meaning eludes us, but at each end of the register a giant fist can be seen to reach down from above to touch with its thumb the side of the arch. In the register below the kneeling figures, the motif between the arches might represent an elephant tusk, as Chen De’an 1990 (p. 86) suggests, while the motif at each end is clearly a forked blade standing on its end, presumably on the ground, suggesting that the headless figure in Figure 5.24 does not represent the only way in which forked blades were manipulated at Sanxingdui. 304If the design portrays live participants in a ritual ceremony, we might even detect a rare indication of action during the performance. All the figures wear earrings, the standing ones bell-shaped earrings that are flying almost horizontally, the kneeling ones interlocked circular rings spread out at 45-degree angles. By contrast, the shell-shaped earrings worn by the laughing faces on the gold sheath in Figure 7.22 hang vertically beside the faces. Perhaps a variety of different head movements during the ritual ceremony are depicted. 305Falkenhausen suggests that, “The placement of smaller kneeling figures and larger standing ones in superimposed tiers may indicate hierarchical distinctions, or different stages in the ceremony.” (Falkenhausen 2002, p. 75). 143 Apparently the design on the jade implement and the bronze altar represent two ensembles varying on a common theme. Together they strongly suggest the possibility of similar ensembles in large scale formed by bronze human heads, masks and figures. Furthermore, the altar also has birds standing in the corners, and human-bird hybrids perched atop the row of kneeling figures. Out of the headdresses of the standing figures rise masks. Though without protruding pupils, their facial features such as the eye sockets and the ears are similar in shape to those of the monstrous masks than the human ones. These details provide useful clues for the involvement of those creatures in large-scale ensemble. Besides the ensembles described above, the kneeling figures in Figures 4.31 and 5.23 show the tantalizing possibility of roofs topped with kneeling figures. The coneshaped pedestal on which the woman in Figure 4.31 kneels has three small extensions below the circular foot ring, implying that it was perhaps to be mounted on something else, possibly as the roof of another miniature altar or temple. The miniature in Figure 5.23 repeats the same theme; the diamond-shaped motif on the figure’s leg is an image of the actual appliqués meant for life-sized figures as discussed before. These images encourage us to imagine the scenario of actual buildings topped with kneeling figures at Sanxingdui. The fragmentary tree in Figure 5.40 shows the relationship between the tree and human figures in ritual ensemble. Unfortunately, little else among the Sanxingdui imagery has similarly explicit depiction. Nevertheless, two recurring motifs shared by these two groups of images may be useful for future reconstruction when more material is excavated at Sanxingdui. One of them is the tree’s cone-shaped pedestal with sweeping 144 arches. On the altar, the tripod of sweeping arches is expanded to four to suit the square form of the altar (Fig. 5.43). On the jade implement, the arches are unfolded to become a horizontal row (Fig. 5.42). In Figure 4.31, the lid of the tall zun borne on the head of the woman is a cone with undulating patterns suggesting four arches. A small lump of metal attached to the top of the cone might have supported a miniature tree. The pedestal on which the woman kneels is also cone-shaped. Although its openwork patterns differ from those on arches of the tree bases, the three knife-shaped flanges clearly allude to a tripod. Another recurring motif is the oversize hand-shaped claws of the dragons. We have earlier equated the graceful human hand seen with the dragon on the monumental tree with the clenched claws of the dragons decorating the outer garment of the lifesized figure. The giant hands reaching down to touch the sides of the arches seen on the jade implement may belong to dragons as well, for they are clearly the same as the dragons’ fists on the garment. In these elaborate ensembles, how were those imported bronze vessels involved? The importance of the vessels is best illustrated by the image of a kneeling woman bearing a zun on the head in Figure 4.31. Perhaps it is now worth reiterating that quite a few bronze vessels from the pits had small holes drilled around the edge of the foot rim, probably for the purpose of mounting them and perhaps for mounting on some of the bronze human images. In short, it is evident that the vessels, images of trees, images of human, and many other jade and bronze objects together constitute paraphernalia of ritual ensembles. The preceding discussions are unavoidably vague in delineating ritual ensembles at Sanxingdui, but they at least begin to suggest possible scenarios. Among 145 the evidences present available, the layout comprised of standing and kneeling figures common to the design on the jade implement and the bronze altar is particularly important and it could not have been accidental. In all probability the layout represents a basic structure of ensemble in the local ritual practice. 5.4 Other images from K1 and K2 A few images from the two pits seem to fall outside the two categories discussed above—heads, masks, and trees—and must be tentatively be put in a third category awaiting evidence from future fieldwork. Those images come in several kinds. (1) Birds. Although birds are primarily associated with the trees, the largest and the most imposing of them, shown in Figure 5.44, has no clear connection with a tree. As its neck is open at the bottom and has three small holes along the lower edge, it was probably mounted on something made of another material. Given its size, it is unlikely to have stood on a tree; it would be disproportionately large even for the largest of the trees from K2. It might rather have been mounted on a post, making a sort of totem pole. This bird must originally have been quite colorful. The eye sockets and the opening in the beak were painted with a red pigment, still visible on the left side of the head. The large opening at the front end of the crest might have held some appendage or elaboration or even actual plumes. Artistically the image is forceful. The principal features have been exaggerated in size and rendered as a few bold shapes, their contours sharply cut, like the features of the human heads and masks. These shapes are subtly repeated: the upward arc and downward hook of the beak’s grooved opening are echoed in the beak itself, the crest, and below the crest, in the depression that surrounds the eye. 146 The front edge of the eye socket parallels the front of the beak. The forceful carving of these large features, undiluted by surface decoration or other distracting detail, combines with their elegant repetition to give the head remarkable power. Other birds unlikely to have been installed on trees include a bird standing on a short domed tube of unknown purpose and a rooster that is easily the most realistic image among the Sanxingdui bronzes (Figs. 5.45, 5.46). The block on which the rooster stands is a shell open at the back, and the rooster’s tail is split on the same axis, giving the object a large slot in the back. The slot must have been used for mounting it on something else, such as a piece of furniture. Unlike the stereotyped, rather generic birds decorating contemporary bronzes made in the Central Plain, birds at Sanxingdui show great variety. The beak of the bird in Figure 5.45 is curved but not hooked; more typical of Sanxingdui is the parrot-like beak of the bird in Figure 5.31, as are the still larger hooked beaks of the birds on the monumental tree. Sometimes, at least, the intention was clearly to depict a real species (Fig. 5.46). Nevertheless the plumage of the bird in Figure 5.31, not to mention the human head on the bird in Figure 5.32, reminds us that ornithological correctness was not the chief concern of Sanxingdui casters. The bird head in Figure 5.44 bears about the same relation to a real bird’s head as the head in Figure 5.3 bears to a real human head. (2) Dragons. Like birds, dragons at Sanxingdui are mostly associated with trees, but the tube in Figure 5.47 might have been mounted on top of some sort of pole or post. The open bottom has four semicircles at the edge and small holes piercing the wall above them. The dragon clambers up the side of the tube, its forelegs planted on the top surface, its hind legs clinging to the back. On the back of the tube the body is little more than a 147 raised strip, except where the tail curls outward, but the creature that emerges triumphantly on top is very lively. Its mouth gapes wide, exposing sharp teeth, and it has a prominent beard like that of a goat. Its horns are much smaller than its floppy ears. A small hole at the tip of each ear was no doubt meant for hanging additional ornaments. An appendage on the front of the tube is faintly reminiscent of the one on the forehead of the monstrous mask in Figure 5.29. (3) Snakes. K2 yielded about a dozen fragments of snakes, including two heads, two middle sections, and two tails (Fig. 5.48). The broken ends do not match well enough to make it certain which pieces belong together, but the three shown here suffice to give an idea of what the snakes looked like. They are shells of metal and were probably mounted on something else with the assistance of the two loops a little way behind the head. The dished spiral of the snake’s tail resembles the upper end of the trunk on the monstrous mask in Figure 5.29. A small stump on the head may be the remnant of an attachment that has broken off; the excavators found a quill-shaped element that fits neatly onto the stump on the body, and the quill is a recurring motif among the images, seen again on the trunk of the monstrous mask in Figure 5.29, for example. The snake’s eyes are similar to those of the monstrous mask in Figure 5.28 as well as the elephants propping up the life-sized figure in Figure 5.17k. The body of the snake is decorated with rectilinear spirals. (4) Tiger and human figure. In addition to the two chance finds of tiger-shaped plaques discussed in the previous chapter (Fig. 4.1), K1 yielded a feline-like creature that may be loosely identified as a tiger (Fig. 5.49). This tiger might be connected with a kneeling figure from the same pit (Fig. 5.50). Zhang Minghua argues that the figure 148 originally rode on the back of the tiger.306 Interestingly, the figure differs in some respects from other human images. The goggles around the figure’s eyes formed by eyebrows and high cheekbones are a familiar feature, but the mouth full of teeth is entirely different, resembling instead the grinning mouths of the taotie faces (Fig. 5.51). The figure sports a unique hairdo that rolls backward, then sweeps forward again to stop in mid-air. He wears a robe that covers the upper part of the body and exposes a Vshaped area at the neck on both back and front; in front the left lapel lies over the right, as seems to be usual on clothed figures from the pits. It is not clear whether the robe extends below the belt: in front the figure is damaged, but his plump hindquarters seem bare apart from a loincloth. He does apparently wear socks or shoes.307 (5) Plaques of taotie faces. K2 yielded three types of animal-faces plaques, three examples of each type (Fig. 5.51). The first two types, represented by Figures 5.51a and 5.51b, average about 20 cm in height; the third is smaller (Fig. 5.51c). In design these faces are related not to the Sanxingdui heads and masks but to the taotie faces that decorate bronze vessels made elsewhere in China, in particular to the transition-period taotie that features a continuous mouth (e.g. Fig. 4.17).308 All the plaques have small holes for attachment to a backing.309 306Zhang Minghua 1998. He also proposes that the form is connected with the Liangzhu Culture of the third millennium, but the evidence for this argument is too tenuous at present to suggest any firm conclusion. 307In addition, K2 yielded a plaque, K2(3):103-27, in a shape vaguely resembling a human torso with two legs, its identity and use remaining thoroughly mysterious on present evidence (see Beijing 1999a, p. 185 fig. 102 rubbing no. 13, p. 187 pl. 65.3, p. 550 color pl. 51). 308Nanba Junko 2002 (pp. 146–51) reaches the same conclusion. 309Zhang Minghua 1999 suggests that they were originally mounted on shields. 149 The plaque in Figure 5.51a has large lozenges for eyes, a slender nose ending in flared nostrils, and a widely grinning mouth full of teeth. Tiny prongs indicate the canthi of the eyes. Directly above the eyes, spiraling crests or horns stand on the upper edge of the plaque. Although the large hooks at the top corners are more horn-like, similar hooks in taotie designs often seem instead to represent bodies duplicated for symmetry (compare Figure 4.11a). The eyebrows are stretched sideways almost to the tips of these bodies. Beneath them are ears. The plaque in Figure 5.51b is the same but for the addition of an element below the chin. This element is decipherable as a pair of large eyes flanked by elements representing the inner and outer canthi. Similar eyes are featured on the plaque in Figure 5.51c, but here the outer canthi are prolonged so that they rather than the eyebrows (which are missing) reach onto the hooks at the upper corners of the plaque. Still other plaques from K2 take the form of similar eyes in isolation.310 Elegantly drawn versions of the same eyes appear also on the pedestal of the life-sized statue as well as the snake described earlier. Beyond Sanxingdui, they are a standard design element on bronze vessels as early as the Erligang period, with antecedents at Erlitou.311 A remarkable feature common to all the three types is the downward hooks at the outer corners of the mouth, just the same as on most of the heads and masks. This is perhaps the clearest single indication that the plaques were cast at Sanxingdui. At the same time it raises the possibility that this Sanxingdui trademark might be a local exaggeration of a detail that Sanxingdui casters first encountered in the taotie of some imported transition-period bronze. 310See Beijing 1999a, p. 203 fig. 114. 311For an Erlitou-period example see the jade baton (VKM4:1) in Bagley 1987 (p. 70 fig. 21), Beijing 1999d (p. 256, p. 257 fig. 168.1, pl. 125.15). 150 (6) Circular appliqués. Though at an overall diameter of about 85 cm this type of object with five spokes framed by a circular rim looks tantalizingly like a wheel, it could not have been one, for it is a thin shell open at the back (Fig. 5.52). Small holes on the rim and at the center probably served for mounting; the size of the object suggests an architectural setting, or perhaps mounting on a shield.312 A few bronzes from K2 include in their surface decoration a motif that consists of a scalloped or pronged device enclosed in a circle, among them the animal on the headdress in Figure 5.18, the animal of the altar in Figure 5.43, and an item the excavators identify as the roof of a building (Fig. 5.53). The prongs vary from five to seven in number. 5.5 Sources of Sanxingdui imagery The preceding discussions have established that the most of the Sanxingdui bronze images were originally components of sculptures that had parts made of another material that has perished. With this understanding, it is easy to imagine that there was a time prior to the introduction of bronze images when the sculptures were made entirely of that perishable material; in other words, the bronze images had local forbears. Several pieces of evidence hint that the material was wood rather than clay. The life-sized figure again provides the most compelling clue. In contrast to the realistic depiction of the garments, the body is a strangely attenuated column, anatomically quite unreal. Its columnar form might well be owed to a tree trunk. The small holes on many images indicate the use of pegs or nails for fastening them; wood again would be a material 312Tokyo 1998 (p. 122) and Beijing 1999a (p. 235) speculate that the object is a solar symbol. The suggestion that it ornamented a shield is made in Lin Xiang 1991. 151 suitable for this purpose. Moreover, as suggested earlier, the bronze tree in K2 might have replaced a real tree consumed in the sacrifice represented by K1. Recently, Falkenhausen has noted similarities between the bronze heads at Sanxingdui and a number of miniature jade human heads of the Shijiahe Culture, and to a lesser degree in the animal imagery from the two cultures (Fig. 5.54); he proposes that certain crucial elements in Sanxingdui iconography might be derived from Shijiahe, though he also mentions the time gap between the Shijiahe and Sanxingdui images—on the order of nearly a millennium— and the apparent absence of intermediaries.313 Despite the time difference and the absence at Sanxingdui of the fanged faces so characteristic of Shijiahe,314 the similarities cited by him are close enough to make the connection plausible. The missing intervening stage that puzzles him might have been the period in which the Sanxingdui sculptures were made entirely in wood. Obviously, however, not all of the Sanxingdui imagery finds sources at Shijiahe, and beyond Shijiahe little direct comparison can be drawn with other cultures. Ultimately we must credit the Sanxingdui artists with considerable originality and inventiveness. A few scholars, puzzled by the apparent lack of an earlier local sculptural tradition, have looked beyond the landmass of China for possible sources, speculating that the Sanxingdui imagery might reflect influence from Mesopotamia, where imagemaking was more familiar. The suggestion was first made by Huo Wei and has been 313Falkenhausen 2003. An abbreviated version of the paper appears in the form of two separate contributions to Paris 2003 (pp. 47–59, 185–94). 314Besides Shijiahe, fanged faces exist widely in several regional Neolithic and Bronze Age cultures. 152 pursued in more detail by Duan Yu.315 The principal arguments are as follows: (1) Human sculpture had appeared in Mesopotamia by the beginning of the third millennium and was used there in ritual contexts. (2) The facial features of the Sanxingdui heads seem Caucasian; the exaggerated eyes recall Sumerian votive figurines.316 (3) Gold and gold foil were used in Mesopotamia, sometimes for sculptural heads (e.g. the head of a bull on a lyre from Ur),317 and masks (Tutankhamun’s funerary mask is the example cited). (4) The “ram in a thicket” from Ur is a possible precedent for the metal trees at Sanxingdui.318 (5) Finally, the gold-sheathed staff from K1 might be a symbol of authority comparable to the mace in Mesopotamia and Egypt.319 Some of these parallels seem superficial, none is very compelling; nor do the authors suggest a mechanism by which Mesopotamian prototypes could have come to the attention of Sanxingdui artists and patrons. For the moment the question of outside influence from Mesopotamia should be left open. 5.6 The function of the Sanxingdui images and the reason for their sacrifice320 Having examined the Sanxingdui images and their likely original appearance in some detail, we face two questions: why were the images made in the first place? And 315Huo Wei 1989, Duan Yu 1993 and a few other articles. See the bibliography in Chen et al. 1998 for references. 316See, for example, Strommenger & Hirmer 1964, pls. 50–60. 317Ibid. (pl. 76), and Zettler & Horne 1998 (no. 3). 318Strommenger & Hirmer 1964 (pl. 80), and Zettler & Horne 1998 (no. 8). 319One shared item that has not been mentioned is sun-dried brick, known much earlier in Mesopotamia and the Indus valley. 320Many opinions have been expressed regarding the identities of the Sanxingdui images and their purposes. Most are speculations with no basis in material evidence; some are based on much later texts whose relevance to the Sanxingdui Culture is far from proved. It seems unnecessary to review them here. 153 why were they sacrificed? Again the evidence does not afford sure answers, but a few clues may be obtained by asking what light the images throw on each other. Assuming that the Sanxingdui images were used for some time above ground before they were interred in the pits, we may first note that, whether their eventual sacrifice was foreseen or not, it was not the sole purpose for which they were made. Assuming that the life-sized figure represents a man rather than a god, then its crown, richly decorated robe, and high pedestal surely signify a man of high status. In the absence of any sign that he came from an alien group, we must suppose that he belonged to the local community at Sanxingdui, as a member of the elite or perhaps the ruler. As pointed out earlier, the image probably portrays him in the act of making an offering, perhaps holding out an elephant tusk. If the leader and other elite members at Sanxingdui would commission sculptural representations of themselves, perhaps not so much in the sense of individuals but as a class, in the act of presenting offerings, these images must have served as stand-ins for them in some sort of symbolic enactment of ritual offering to some higher power. Whether these images were also objects of adoration for the populace at large, we have no way of knowing. The idea of the figures as stand-ins helps account for their ultimate sacrifice and the lack of discernible human remains in the pits. Since the figures were found in the pits, they were themselves offerings. In some crisis, perhaps, the Sanxingdui ritual required the sacrifice not only of objects but also of the officiants. The images in this circumstance perhaps symbolically represented elite self-sacrifice for the purpose of placating or 154 gaining favors from the high power.321 At Sanxingdui the power in question might have been the one that controlled the floods. If symbolic sacrifice of persons of high status was indeed the function of the figures and heads, it is utterly different from anything known at Anyang. Human sacrifice at Anyang was real, and the victims were not of high status, to judge by their burial, but commoners or war captives sacrificed like cattle.322 The significance of the trees and their exact relationship with the human figures remain unclear, but it is unlikely that the figures made offerings to them, for like the figures they were sacrificed by breaking and burning. This impression is reinforced by the placement of the human figures on the tree in Figure 5.40. This image might represent an outdoor ceremony among hills, as mentioned earlier; significantly, the figures face away from rather than towards the tree. On the monumental tree in Figure 5.30, the dragon descending from above must have had a specific meaning for its Sanxingdui audience, and so must the snapped wings of the birds, but we are at loss to guess those meanings. Much future fieldwork is needed before we can hope to say much more about the purpose of the Sanxingdui images and the reason for sacrificing them. 321Fan Xiaoping has earlier voiced the idea that the bronze human heads and figures were surrogates for actual human sacrifices. He did so by drawing comparisons with sacrifices at Anyang, but without considering the particular features of the images themselves. His division of the images into various social classes, including slaves and common laborers, is hard to sustain on present evidence. See Fan Xiaoping 1993, pp. 107–9. 322Bagley has compared the sacrifices at Sanxingdui and Anyang in Bagley 1988, and in more detail in Bagley 1990a. See also Xu Ziqiang 1993. For a succinct account of Anyang sacrificial burials see Bagley 1999, pp. 192–4. The low-status victims consumed in the Anyang sacrifices must of course be distinguished from the concubines, military escorts, etc., found in tombs of the highest rank—specific people or classes of people who followed their lord in death. Hu Houxuan has supplied useful surveys of Anyang oracle inscriptions mentioning human sacrifices (Hu Houxuan 1974b) and of archaeological evidence for human sacrifice at Erlitou, Zhengzhou, and Anyang (Hu Houxuan 1974a). 155 Chapter 6. Bronze at Sanxingdui (III): Casting Technology This chapter continues the investigation of the unparalleled items from K1 and K2. The aspect to be addressed is fabrication. As elsewhere in ancient China, bronzes were made at Sanxingdui by casting in clay molds of two or more sections.323 The only exceptions are miniature ornaments—birds, fish, animals, leaves, forked blades—made from very thin (0.1–0.2 mm) hammered sheet bronze.324 The investigation will therefore focus on casting, to explore the techniques employed, their origin and cultural implications, and the particular mentality evident in the application of the techniques. 6.1 Casting technology in the Central Plain325 Before attempting to describe foundry practice at Sanxingdui, let us first review the basic process of casting in clay section molds. As we know, the technology originated at Erlitou and Erligang in the Central Plain, and was used there throughout the Bronze Age for making vessels. Therefore, vessels are fitting examples. 6.1.1 Basic techniques The bronze tripod (ding) in Figure 6.1a, a cooking vessel of the Erligang period, is a good starting point for discussing the casting process. The first step of the process was the making of an undecorated model, to which fine wet clay would be applied as an envelope and then removed in sections; the decoration would then be carved on the 323Many images still contained remnants of clay cores used in casting. 324Beijing 1999a, pp. 315, 319–25. 325For detailed studies of the casting technology in the Central Plain see Gettens 1967, Gettens 1969 (pp. 57–120), and Bagley 1987 (pp. 37–45). 156 sections, which constitute the outer parts of the mold.326 Curiously, it seems that from the Erlitou times on until about the fifth century BC no model was used more than once in Central Plain foundries, for no two vessels are known to be identical.327 The diagram in Figure 6.1b shows the relationship between the ding and the various parts of the clay mold. A telltale sign indicating the use of a multipart mold is marks of mold joins on the finished vessel. When molten bronze was poured into the assembled mold, it might leak into the crevices between adjacent mold sections and leave mold marks. The mold marks on the present vessel reveal that the mold comprised three outer sections. In addition, a core was necessary for the vessel interior. When assembling the mold together, an open space corresponding to the desired thickness of the vessel had to be left between the outer sections and the core; scrap pieces of bronze were routinely used as spacers for this purpose. With such a mold, a vessel of simple shape could be cast in one pour of metal. The ding of Figure 6.1 was cast complete with its loop handles. Besides casting in one pour, a vessel could be cast in a sequence of pours, a technique which was developed at an early stage. In most cases, this technique entails the joining of two pieces of metal in the process of casting the second piece. Let us call the casting that produced the bulk of a vessel the main pour. It is convenient to speak of precasting when a smaller piece is cast before the main pour, casting on when the smaller piece is cast after the main pour. A small part cast first would be embedded in the mold for the main pour in such a way that, when the main pour was done, the two would be locked together. The flange in Figure 6.2 was precast: it can be seen that the metal from 326Later the decoration would be carved directly on the model (see Bagley 1987, pp. 38– 9). 327Bagley 1995 points out the production of identical objects using the same model in the fifth century (p. 46). 157 the main pour overlaps the flange. Alternatively, parts could be added after the main pour by building small molds for them against the already cast object. The tiger in Figure 6.3b, for example, was cast onto the handle of a large tetrapod (fang ding) (Fig. 6.3a). Casting on was used particularly freely, because it was the standard way of repairing casting defects. Bagley has suggested that it was in fact first devised for making repairs, then exploited as a fabrication technique.328 The fang ding in Figure 6.4 shows several large cast-on patches. The use of casting to patch defects would be regarded as highly eccentric by metalworkers almost anywhere else in the world. Sometimes, two components of an object might be joined by casting a third piece between them. Two variants of this technique are known: running on and soldering/brazing. Figure 6.5 shows an example of running on: a capped post on a wine vessel. Arrows 1 and 2 point at the very faint join lines, which indicate that the cap was cast separately and that the joining was done by building a mold between the cap and post, and then casting a bit of metal between them (arrow 3 points at a mold mark). Running on required a mold. If the shape made it possible not to use a mold, the joining might be done by pouring metal directly onto the two components. This process is called soldering or brazing, depending on the melting temperature of the solder used.329 In the present discussion, because in many cases no chemical analysis of the joining metal has been performed, only the term “soldering” will be use. Soldering often leaves untidy and irregular outlines such as are seen in Figure 6.6. The principle underlying running on and soldering is shared with precasting: running on may be thought of as a variety of 328Bagley 1999, p. 144. 329“Soft solder” is the term for an alloy with a melting point much lower than that of the objects being joined; “hard solder” refers to an alloy with a melting point close to that of the objects being joined. “Brazing” means joining with hard solder. 158 precasting in which the first two components are the precast parts, and the run-on part is the tiny “main” pour between them; a soldered join may be thought of as a join in which the “main” pour was so small that it did not need a mold to enclose it. Finally, it is important to note that in joining by any of these methods, the components were held in place almost certainly not by metal fusion alone, which is not easy to achieve in bronze alloys, but by mechanical interlock. The founder either provided a locking device between the two components or holes for the metal of the second pour to flow into. 6.1.2 The “one-pour mentality” in foundry practice In the Central Plain, although the technique for casting in a sequence of pours was available from the Erligang period on, the emphasis was clearly put on casting in one pour. Founders evidently thrived on the challenge of building increasingly complex mold assemblies. The ding in Figure 6.7, still dating from the Erligang period, already required a more complex mold than the ding in Figure 6.1, for this vessel is about 1.5 times larger in size. The mold sections doubled in number. In addition to the tripartite vertical division, the mold was divided horizontally in the undecorated part of the belly. More often, molds would have even more outer sections, and more cores not just for the interior but also for a ring foot or for the legs of a tripod. For example, the wine container (fang yi) in Figure 6.8a, which dates from the Anyang period and is roughly contemporary with the Sanxingdui images, required a very complex mold. The vessel’s roof-shaped lid was a separate casting; let us concentrate only on the vessel proper (Fig. 6.8b). The mold for it included two cores, one for the interior and one for the ring foot. 159 On the exterior, the mold was divided vertically on the axes of the eight sets of flanges (the diagram, which shows only four vertical divisions, is incorrect). It may have been divided horizontally as well, along the boundaries of the registers. Including the cores, the mold assembly would thus have at least ten sections, possibly as many as twenty-six. Before the final assembly, the outer sections on each side might be joined into larger sections as shown in Figure 6.8b, and then those would be assembled around the cores. Both operations must have been delicate and difficult. Moreover, the mold-making was greatly complicated by the intricate high-relief decoration, which made it tricky to remove outer sections imprinted with the design from the decorated model. The difficulties did not stop here, for it cannot not have been easy to keep the mold sections in precise alignment during casting. That the vessel was cast in a single pour of metal is something of a tour de force. The mindset that strove to cast complicated vessels in single pours was probably in part a habit that grew gradually on the founders of Erligang times as they refined their techniques and undertook ever more ambitious castings. But it probably also became a practical necessity as the decoration spread from narrow bands to cover the whole exterior of a vessel: an area that is decorated is an area where joins would be obtrusive. The founders therefore strove to make an object in one pour, or in as few pours as possible, so as to minimize the number of joins; they preferred casting on and precasting to running on and soldering, which leave messier joins;330 and they confined those additional pours to elements like the animal heads in Figure 6.9a, which could be joined very unobtrusively: only the three heads of that complicated vessel were cast separately 330For discussion of soldered examples see Bagley 1987, p. 43. 160 from the main pour (in Figure 6.9b, the stump of a pouring inlet, or sprue, visible on the side, shows how the metal entered the mold for the head). I propose to characterize this particular emphasis on casting an object in as few pours as possible as an “one-pour mentality.” As we shall see, the Sanxingdui founders did not share it. 6.1.3 Nature of clay section-mold casting Over the last half century, the technology of casting vessels in clay section-molds has been a fruitful subject of investigation. Many bronze specialists and metallurgists have published detailed elucidations of technical details and the nature of the technology. Among them, Gettens and his collaborators produced the most detailed analyses of the actual techniques.331 Bagley has presented the most coherent description of the technology and its inherent relationship with vessel design.332 Experts have also studied the technology in a global perspective, attempting to account for its preeminence in and uniqueness to China. One readily recognizable factor is a developed ceramic technology which involved highly skillful manipulation of clay and mastery of high temperatures. Chase has pointed out that the ternary alloy of Chinese bronze—leaded tin bronze—lends itself well to casting but not to hot-working or cold-working in the solid state, or in another word, hammering.333 Bagley has noted an important economic prerequisite for the development of casting in the Central Plain: it was probably the abundant metal supply that fostered a reliance on casting; in the Near East by contrast, bronze was a much more scarce resource and therefore mostly worked by hammering, which is capable 331Gettens 1967, and Gettens 1969. 332Bagley 1987 (pp. 37–45), Bagley 1990b, and Bagley 1999 (pp. 141–6). 333Chase 1981 (p. 103), and Chase 1991 (p. 22). 161 of rendering the walls of an object extremely thin.334 The largest vessel so far known from the Central Plain—the Si Mu Wu fang ding excavated at Anyang—weighs 875 kg.335 It was cast around 1200 BC at the height of the Anyang bronze industry. Such extravagant use of metal is distinctively Chinese. Following Franklin, Bagley observes that the casting technology encouraged division of labor and the organization of efficient workshops.336 All these observations have contributed importantly to our understanding of the bronze technology at the Central Plain. It is clear that the technology must be understood as a cultural product: in other words, it was developed under a particular set of conditions for the purpose of fabricating a particular type of object—vessels. 6.2 Foundry practice at Sanxingdui337 Having briefly reviewed the casting technology developed in the Central Plain, let us return to Sanxingdui. Our understanding of foundry practice at Sanxingdui is just beginning, but it is clear that the bronze images, including such complicated sculptures as the life-sized statue in Figure 5.17 and the monumental tree in Figure 5.30, were made by casting in clay section-molds. And as in the Central Plain, casting was done sometimes in one pour, sometimes in a sequence of pours. 334Bagley 1987 (p. 17), and Bagley 1999 (p. 141). 335For an illustration see Beijing 1997a, pl. 47. 336Bagley 1999 (p. 141), Franklin 1981 (pp. 96–7), and Franklin 1999 (pp. 10–7). 337Earlier studies of the casting techniques at Sanxingdui include Barnard 1990, Zeng Zhongmao 1994, Chen et al. 1998 (pp. 37–8), Tokyo 1998, and Beijing 1999a. The most recent technical study is Mifune Haruhisa 2002. The present discussion draws on previous work, but relies mostly on my own observations. Since little scientific study has yet been undertaken, nor have radiographs been made, the discussion here is unavoidably tentative. 162 6.2.1 Objects cast in one pour of metal At Sanxingdui, the plaques of taotie faces (Fig. 5.51), appliqués (Figs. 5.25, 5.52), and some figures (e.g., Fig. 5.19, which is hollow at the back), being basically sheets of bronze, are among the simplest one-pour castings. The molds used to cast them had two sections, a front and a back, the back following the contours of the front so that the metal would be uniform in thickness.338 For these objects no core was needed. The molds for casting bronze collared disks like those shown in Figure 5.39 would require two sections and also a core for the central opening. The ge with serrated edges were cast similarly (Fig. 6.10a). Molds with only a few outer sections and a cored shaped to achieve even thickness seem also to have sufficed for casting human masks, heads, and other figural items in one pour.339 For instance, on the mask in Figure 6.11, only a single mold mark is visible on each side of the face: it runs down the axis of the ear. On the fragment of a mask in Figure 6.12, the mold mark is clearly visible both above and below the ear. Moreover, we may suppose that the mold for masks of this kind was also divided along the central axis of the face, for the shape of the face makes it unlikely that the mold could have been withdrawn from the model in only one front section.340 The mold was thus divided vertically into two back sections and possibly two front sections which met at the 338Uneven thickness results in uneven cooling of the metal, which can cause defects. 339Some of the objects discussed here evidently had or were meant to have additional components such as headgear added on by separate casting, and therefore, strictly speaking, should not be regarded as objects cast in one pour. However, they provide good examples of moldmaking for one-pour casting. 340Clear evidence for a mold division along the nose can be seen on the head in Figs. 5.10/6.16; see discussion below. 163 ears and the nose, and it required a core or rather a rear section that formed the open interior.341 The molds for the heads were similarly divided at the ears, as seen for example on the head in Figures 5.3/6.13a. On another head (Fig. 6.14), the mold mark that continues below the right ear and onto the neck is clearly visible, and the mold may have been divided at the nose also. The assembled mold thus had two or, more likely, three outer sections, and a shaped core responsible for the hollow interior.342 Among the heads with a flat top, two ways of dividing the mold between the top panel and the rest of the head can be observed. On the head in Figures 5.15/6.15, a conspicuous mold mark runs around the circumference of the flat top; the mold thus had three or four outer sections.343 On the head in Figures 5.10/6.16, two mold marks run across the top: one bisects the top and continues down both sides of the head in the plane of the ears; the other begins at the center of the first mold mark and runs down the central axis of the face.344 For the heads with domed crania, the head in Figures 5.6/6.17 provides clear evidence for mold divisions on the top.345 As in the case of the head in Figures 5.10 /6.16, 341Other human masks cast in one pour include K2(2):43, K2(3):57, K2(2):60, K2(3):65, K2(2):102, K2(3):119, and K2(2):153. The mask from K1 (K1:4) was cast in one pour as well. 342Other heads with open tops cast in similar molds include K1:6, K1:8, K1:26, K2(2):17, K2(2):41, K2(2):50, K2(2):52, K2(2):55, K2(2):59, K2(2):72, K2(2):77, K2(2):78, K2(2):90, and K2(2):113. 343Similar heads with clear circumferential mold marks on the top include K1:10, K2(2):22, K2(2):34, K2(2):53, and K2(2):121. 344The mold division seen on this head is so far unique among the flat-top heads cast in one pour. 345Other domed heads are K2(2):63, K2(2):137, and K2(2):214 (covered with gold foil). 164 a mold mark extends across the cranium from one ear to the other.346 The remnant of a second mold mark bisecting the front part of the head is visible at the top of the cranium (Fig. 6.17a). The mold for the head thus was composed of three outer sections. Some heads required a mold with more than three outer sections. The finned head in Figures 5.5/6.18 was probably cast in a mold with four or five outer sections. A prominent mold mark runs up one ear, over the edge of a fin, across the cranium, over the other fin, and down the other side. The mold probably had a front section that corresponded to everything in front of this line—the face, the fronts of the ears, and the front part of the cranial dome. Or possibly there may have been two front sections divided vertically along the median of the nose: on and above the nose are irregularities looking like the remnant of a filed-off mold mark. A rear section formed the back of the head and neck and the inside surface of each fin. A narrow side section was then needed to form the back of each ear and the adjacent (outside) part of the fin. Relatively few objects from the pits required molds more complicated than this. In molds that required cores, Sanxingdui founders often used spacers, sometimes profusely. Spacers are clearly visible on the head in Figures 5.15/6.15 and the figure with animal headdress in Figure 5.18 (left side of the face).347 6.2.2 Objects cast in several pours of metal 346This mold mark disappears at the point where it intersects the triangular tip of the gold mask: evidently the caster ground it off the surface to which he was going to apply the gold (see Figure 6.17b). 347Spacers are easily found on other Sanxingdui images as well, including the heads K2(2):34, K2(2):48, K2(2):50, K2(2):53, and K2(2):121, and the masks K2(2):57, K2(2):102, K2(2):111, and K2(2):153. 165 All the joining techniques used in the Central Plain were employed at Sanxingdui as well, and many of the Sanxingdui images involved more than one pour of metal. In the following discussion we will first examine the techniques individually and then discuss their application on several of the most complex images. 6.2.2.1. Precasting, casting on, running on, and soldering Some of the flat-topped heads appear to have precast top panels as seen, for example, in Figure 6.19. Irregularities that look like slight overflows of metal from the sides of the head onto the flat top suggest that the top was precast, then set into a mold for the rest of the head.348 The weirdly supernatural mask in Figures 1.3/5.28/6.20 shows extensive use of precasting. The mask was made in six components, five of which were precast and embedded in the mold for the sixth. The precast parts are the two ears, the two pupils of the eyes (Fig. 6.20a), and, unexpectedly, the triangular underside of the nose (Fig. 6.20b). Casting on is illustrated by the head in Figures 5.9/6.21. Traces of metal overflow running from the crown onto the head indicate that the head was cast first, the crown then cast onto it. The inward step encircling the head must have been intended to facilitate the attachment. On the flat-topped head in Figure 6.22, the top panel was cast on, the sprue being clearly visible.349 Among the heads with open tops, some have a ridge around the circumference that looks as though meant to receive a cast-on panel that was for some 348Other heads with top panels likely to have been precast include K1:11 and K2(2):48. 349The head K2(2):71 also has an obvious sprue for casting on the top panel. Other flat- topped heads with cast-on flat panels include K1:3 and K2:45. 166 reason never added (or that has been lost?) (Figs. 5.8/6.23a).350 Sometimes what is left open on the top is a large hole as seen in Figure 6.24, the clean contour of which argues that it is not a casting defect; perhaps it was meant to simplify removal of the core. The patches atop some of the heads are probably not repairs of casting defects but plugs covering such intentional openings (Fig. 6.25).351 Evidence of cast-on repairs is nevertheless plentiful. On the head in Figures 5.3/6.13, repairs were made next to the left nostril and below the outer corner of the right eye: Figure 6.13b shows two cast-on plugs as seen from the interior. On the head in Figures 5.8/6.23b, a large casting flaw in the back of the head was repaired with a crude cast-on patch. The monstrous mask in Figures 1.3/5.28/6.20 has numerous repairs. The large size must have made the main pour difficult; defects were repaired by casting on additional patches of metal. The repairs were smoothed by careful polishing, and corrosion has made them even harder to spot. However, one edge of a large repair below the nose is visible as a vertical crack that runs down the mouth and chin; the battering inflicted on the mask before burial may be responsible for the crack. The repair is clearly visible seen from behind (Fig. 6.20c).352 Joins made by running on are visible on the fragment of a bronze tree in Figures 5.40/6.26. Its base is a ring carrying a heavy tripod of three sweeping arches. Each of the three arches is a separate piece. The pieces were joined at the top by a rather crude patch 350Other heads having such ridges include K2(2):55 and K2(2):113. 351Similar cast-on patches appear on heads K2(2):2 (the patch itself has a hole), K2(2):15, K2(2):40, K2(2):82, K2(2):115, and K2(2):118 (the patch has three holes). 352Other images with cast-on repairs include K2(2):15 (head: left nostril), K2(2):58 (head: top), K2(2):121 (head: tip of the pigtail), K2(2):57 (mask: numerous repairs), K2(2):100 (mask: right nostril), and K2(3):94 and K2(3)174 (triangular appliqués: back side). 167 of run-on metal. Two horizontal stumps of metal, visible in Figure 6.26d, were the sprue for pouring the metal and the vent for air to escape. Their positions at the same level indicate that the tree was standing upright when bronze was being poured into the mold. This same pour also joined the base to the tree trunk. The three arches were joined also at the bottom; one join still survives and is visible in Figure 6.26a. This join could have been made by running on or, since it is not clear whether a mold was used for it, by soldering. The joins between the stumps of three branches and the trunk were also made by running on. Soldering was used extensively at Sanxingdui. Quite a few of the human masks of relatively simple shape, such as the one in Figure 5.26, had ears soldered on. Each mask was cast without ears, but with holes designed to receive them: the ears were cast separately, then inserted into the holes and secured with solder. Figure 6.27 shows the rear of the left ear on another mask. A strip of solder overlapping both the ear and the side of the face is clearly visible. On the hybrid figure in Figures 5.22/6.28, patches of solder join the figure’s feet to the birds on which they stand.353 On the tree fragment just discussed, soldering joined the kneeling figures to the rim of the pedestal (Fig. 6.26b). Soldering was used profusely on the monumental tree in Figures 5.30/6.32, as will be discussed later. 6.2.2.2 Concurrent use of joining techniques 353Other notable images with soldered joins include the woman in Figure 4.31 (she was joined to her pedestal by soldering), the tube with dragon in Figure 5.47 (the dragon, its tail, and the curling flange in the front were soldered onto the tube), and the snake in Figure 5.48 (put together from several components). 168 For complex images, Sanxingdui founders often resorted to concurrent use of different joining techniques to assemble separately cast components. The mask with protruding pupils and a raised trunk in Figures 5.29/6.29 was assembled from six components, namely the two ears, the two pupils, the elephant-like trunk, and the face to which these elements are attached. The pupils of the eyes were precast—from the rear it is easy to see that the metal of the face locks around them (Fig. 6.29a)—but on the front the join was reinforced with solder. The ears were soldered on: next to the right ear the metal of the join can be seen to flow onto the ear and also onto the face (Fig. 6.29b). Figure 6.29c shows the back side of the left ear: a patch of untidy solder is very distinct. The trunk was attached to the face, probably by casting on, through a hole left on the forehead. The figure with animal headdress in Figures 5.18/6.30 is another telling instance of this approach. The figure was made by joining together (not very neatly) seven components: the three projections from the hat, the head, the torso, and the two arms. The mold for the head had two sections, but the mold divisions were not aligned with the ears, as we might expect, but ninety degrees away, at the nose and the back of the head: vertical mold marks are visible in the hair on the back of the head (Fig. 6.30a) and, more faintly, between the eyebrows. A hole was left on top when the head was cast. The curling projection at the front of the head was then cast on (Fig. 6.30b); a patch of metal inside the animal’s open mouth is a result from this process. The two side projections were cast separately and perhaps embedded in the mold for the curling projection and locked in position when the latter was cast. On the outside, they were reinforced by solder (Fig. 6.30c). The torso was cast onto the head at the neck (Fig. 6.30d), and the two 169 arms of the figure were joined to the torso by soldering. Figure 6.30a shows clearly a strip of irregular solder on each arm. 6.2.2.3 The life-sized figure For even more complex shapes, Sanxingdui founders relied mostly on running on and soldering. The most spectacular examples of this manner of fabrication are, of course, the life-sized figure in Figure 5.17 and the monumental bronze tree in Figure 5.30. To meet the challenge of casting a nearly 3-meter tall statue, the Sanxingdui founder joined eight separate pieces.354 In the following description, the term running on will be used if the area or quantity of join metal is comparatively large, or if the mold mark or the stump of a sprue survives and thus indicates the use of a mold; otherwise the joins will be described as soldered. The eight pieces are: (1) The undecorated portion of the pedestal. Mold marks along the vertical edges and the top suggest that this piece was cast in a mold of five outer sections, as might be expected. Assuming that it was cast upside down, the opening in the top would have allowed for the support of the core from below. (2) The remainder of the pedestal, except the slightly raised plain board immediately beneath the figure’s feet. Mold marks are less apparent here, but partly visible along the two elephant trunks in the rear (in Fig. 6.31a, which shows these two trunks, the angle of the photograph makes visible the mold mark on the left trunk). With its complicated shape and surface decoration, this must have been one of the most 354 In Bagley 2001 (pp. 75–6), I stated that the statue was assembled from seven separate pieces. In subsequent examination I discovered another join. It is possible that laboratory examination will detect still others. 170 technically difficult castings undertaken at Sanxingdui. According to the excavator, tenons secured it to the undecorated element below and to the plain board beneath the figure’s feet (Fig. 5.17a).355 Whether the tenoned joins were soldered is uncertain. All the other joins in the statue were made by running on or soldering. (3) The feet of the figure, including the plain board on which they stand, and the lower part of the robed body. The horizontal division in the body comes not at the hem of the mantle but slightly higher up, interrupting a row of dragons on the mantle; the join at this height can be faintly seen in Figure 6.31b toward the bottom of the picture. The mold for this component probably had three sections, one for the back and two for the front. On the front, the vertical crease in the clothing seems to coincide with a mold division, for horizontal lines in the decoration are slightly disrupted where they cross it (Fig. 6.31c). At the sides, the mold marks interrupt the textile decoration where the two sides are about to turn into the flat back (Fig. 6.31d). A rectangular opening under the skirt was probably left by a device that supported the core during casting (Fig. 6.31e). (4) The short middle part of the robed body, whose upper edge appears as the blank space between the two rows of dragons on the mantle. The join at this height is also faintly visible in Figure 6.31b. The mold for this part was similarly divided into three sections: one for the back and two for the front. Although it is very small, this part is strategic, serving to link the lower and upper parts of the statue. It can be seen through 355The interior of the undecorated element is now filled with a modern pedestal designed to provide stability to the statue. I have therefore not been able to observe the joining device. Nor have I been able to observe the underside of the top of the decorated part of the pedestal. The information about the tenons given is inferred from Fig. 5.17a, originally published in Beijing 1999a. 171 the rectangular opening under the skirt that the interior of this part has a ledge at each end, presumably for the purpose of strengthening it. (5) The upper part of the body, including the head but excluding the arms and the upper portion of the crown. The mold for this part was divided on the axes of the ears. On the headband a vertical mold mark is visible above the left ear (Fig. 6.31f). Further down, the mold division is indicated by an interruption in the ribbon under the left armpit. Two small openings under the right arm and another in the back may have been left by core supports (Figs. 5.17b, e/6.31b).356 (6 & 7) The two arms. The point where each arm was attached to the shoulder is betrayed by the run-on metal that made the join on the figure’s shoulders (Fig. 6.31g). The joining must have been facilitated by stepped edges prepared for the purpose, similar to those seen on the broken arms of the kneeling figures on the tree fragment in Figures 5.40a and 6.26f. A clearly visible mold mark runs lengthwise along the outside of the left arm (Fig. 6.31f), indicating that the mold for each arm was composed of two sections. (8) The upper part of the headdress. A soldered join attached this to the headband beneath. 6.2.2.4 The monumental tree357 The monumental tree in Figure 5.30, measuring nearly 4 meters tall, was assembled from many more components. Even in its present only partially restored state, 356Wang and Wang 1993 (p. 60) propose that these openings were meant for attaching additional ornaments for the garments. 357The description of joins given here is based on the author’s inspection of the tree and conversations (October 1999) with Mr. Yang Xiaowu, the conservator who restored the tree. 172 it consists of at least 100 separate pieces, almost all of which seem to have been cast in two-part molds.358 Only the base—a ring with a heavy tripod on it—was cast in a threepart mold divided on the axes of the tripod legs (Fig. 6.32a). Most of the joins between components were made by running on or soldering. The extent to which the fabrication of the tree also involved precasting and casting on is uncertain. The trunk of the tree was joined to the base by running a separate pour of metal between them: a large irregular patch of metal that does not belong either to the trunk or to the base made the join (in Figure 6.32a the lower edge of the patch is particularly clear). The trunk was set in position on the base, a mold was built around the point of contact, and metal was poured in to make the join. The stump of a sprue remains in the patch, showing where the metal entered the mold (the tree was probably lying on its side when the join metal was poured). Running on may also have been used for major joins in the branches and for joining the branches to the trunk. Each branch was made in several pieces. In Figure 6.32b, a sleeve-like thickening in the branch is probably the run-on metal that joined the separate pieces, as evidenced by the mold mark and the stump of its sprue. Numerous other sleeves, all joining together two separate pieces, are visible in Figure 5.30. It is worth remarking that these run-on joins were much more neatly done than the join that connects the trunk to the base. At three points the branches are joined to the trunk; one is visible near the top of Figure 6.32c. At each point the trunk is encircled by a zone of thicker metal from which the branches emerge, looking as though they were held in tubes. 358The tree was found badly broken (Fig. 5.2) and cannot be completely restored. The figure of a hundred components is my very approximate count. 173 This zone of thicker metal is probably also a run-on pour; curiously, however, the mold mark on the thicker zone and the trunk are neatly aligned. Standing on the circular base of the tree, facing outward, is the peculiar dragon whose rope-like body and tail rise in an undulating curve up the trunk of the tree (in Figure 6.32d the dragon’s head is at the lower left). The upper half of the dragon’s body is lost. The surviving part is attached to the trunk by two struts, both included in Figure 6.32d though the upper one, partly obscured by a bird nearer the camera, is more fully visible in Figure 5.30. These struts were cast in two-part molds: Figure 6.32e shows the lower surviving strut with a clear mold mark on top of it, and a second mold mark can be found on its underside. The mold marks do not continue onto the tree trunk or the dragon’s body, showing that the strut was cast separately (at the right side of Figure 6.32e the metal of the strut can be seen to flow onto and overlie the metal of the trunk). This seems to be a clear case of running on. The procedure would have gone as follows: the dragon and the tree trunk were clamped in position the desired distance apart; the mold for the strut was built between them; and the strut was cast. Almost certainly it was held in place not by metal fusion but by mechanical interlock; presumably on the trunk and dragon the founder provided either tenons for the strut to lock over or holes for the metal of the strut to flow into. The body of the dragon was cast in several separate pieces, which were then joined by soldering; Figure 6.32f shows one of the soldered joins. A good many other smaller joins were made in this way. Figure 6.32g, which shows a bird standing on a flower that sprouts from one of the branches, provides a good illustration. The bird seems to have been assembled from four separately cast components: the head, the body and the 174 two wings. Each was cast using only a simple two-part mold divided longitudinally; the components were then joined with solder.359 In Figure 6.32g a collar of solder at the neck above the openwork body is clearly visible. The bird’s wings, now broken stumps, were probably joined to its shoulders by solder as well. Beneath the claws solder joined the bird to the flower on which it stands. Further down, below the petals of the flower, a flat openwork ring encircles the calyx. In Figure 6.32h, which shows the underside of this ring, a wad of metal (now much corroded) can be seen holding the ring in place; this too is probably some sort of soldered join. Still further down, where the flat underside of the calyx sits atop the stem projecting from the branch, calyx and stem were joined by another collar of solder. Thus within the boundaries of Figure 6.32h there are more than half a dozen soldered joins. 6.2.2.5 Devices of mechanical interlock Whatever the joining, all joins are likely to have depended less on metal fusion than on mechanical interlock, which the founders ensured by providing either tenons (as in the case of the life-sized figure), or stepped edges (as seen on the kneeling figures and the top of the broken trunk on the tree fragment in Figure 5.40), or holes for the metal to flow into and lock over. Small holes for this purpose are seen extensively among the Sanxingdui bronzes, for example, on the stepped-back upper edge of the head in Figure 359All nine birds on the tree were fabricated in the same way. Two-part molds longitudinally divided were responsible for other birds as well, including K2(2):141 (bird’s head; Fig. 5.44), K2(2):213 (the flower beneath the bird was cast in the same pour; the tail was cast separately and soldered on; Fig. 5.31), K2(3):301-3 (bird on a post: the post was cast in the same pour as the bird; the three-branched ornament on the back was cast separately, and probably soldered in place; Fig. 5.45), K2(3):103-8 (clapper bell in the shape of a bird; Fig. 5.33e), K2(3):107 (rooster; Fig. 5.46). 175 5.4. The circular appliqué in Figures 5.52/6.33, cast in six pieces, was put together using similar holes. The rim of this object is composed of five equal arcs, each of which was cast in one piece with the spoke that projects from it. The joins, very neat seen from the front, are located midway between spokes. They were made on the rear of the object by soldering; the solder was allowed to flow into the small holes at the ends of the arcs, faintly visible in Figure 6.33.360 Another remarkable example of joining done through small holes is the fragmentary tree in Figures 5.40/6.26. The surviving join in the rim of the tripod was clearly effected using holes; Figure 6.26c shows a hole on the damaged rim. A number of small holes in the arches were evidently intended for joining as well, though most of them were not actually used (Fig. 6.26d). However, halfway down the side of the arch in Figure 6.26d, one hole was filled by a pour of metal that formed a plug connecting two adjacent arches, like a sort of cast rivet (Fig. 6.26e). This plug, very neat, is also visible on the other arch, as seen on the left in Figure 6.26f, where it coincides with the eye of the eye-and-diagonal pattern. For masks with soldered ears, like the monstrous mask in Figures 5.29/6.29 or the human ones in Figures 5.26 and 6.27, as a rule two large holes were left on each side of the mask to receive separately cast ears. The holes are now covered on the inside wall by patches of solder as seen, for example, in Figure 6.27b. Some of the masks have a square perforation in the forehead, which was evidently intended to attach additional ornament. Curiously, those large holes as well as the small holes in many cases were sawn and 360The hub was cast onto the spokes in the last step. 176 drilled rather than cast.361 In the absence of steel tools, sawing and drilling bronze cannot have been easy; they may even have had to be done with abrasives. Why they were done in this way is a mystery. 6.2.3 Origin and characteristics of Sanxingdui foundry practice 6.2.3.1 The Central Plain origin The Sanxingdui images just discussed make clear that, as in the Central Plain, fabrication of bronzes at Sanxingdui relied heavily on casting. Moreover all the techniques used in the Central Plain were used at Sanxingdui as well. In Section 6.1, however, it was argued that the reliance on casting and on the section-mold technique in particular was a development conditioned by factors specific to the culture of the Central Plain. To find it in a very different cultural setting at Sanxingdui, therefore, strongly suggests that Sanxingdui acquired the technology from the Central Plain. In addition, we may observe several other pieces of evidence in support of this conclusion. (1) Like founders in the Central Plain, Sanxingdui founders cast even objects that would have been very easily hammered such as the many flat plaques and appliqués (e.g. Figs. 5.19, 5.51, 5.52). (2) The technique of patching defects by casting on is rarely encountered in the world. (3) No two Sanxingdui images of the same type, say heads, are identical, indicating the same curious avoidance of reuse of models. (4) A copper-tin-lead alloy is the norm at Sanxingdui as well (see Table 4). 361The holes in the back corners of the masks were often similarly sawn. In addition, on several other masks the square on the front is scored but the opening was never made (see e.g., Bagley 2001, no. 20). 177 The transmission of the technology implies an important cultural connection between Sanxingdui and the Central Plain, one that their dramatically different bronzes tend to disguise. Since a few Sanxingdui images have surface decoration apparently indebted to widespread transition-period styles, the introduction of the section-mold technology probably took place during the thirteenth century BC.362 In cultural terms, this means that the Sanxingdui bronze industry probably arose under the influence of the Erligang culture and thus that it was a development independent of the Shang bronze industry at Anyang. This understanding regarding the relationship (or lack thereof) between Sanxingdui and Anyang agrees with the conclusion drawn from the analysis of bronze vessels in Section 4.2.2.3. How long the Sanxingdui bronze industry survived after the time of the two great sacrificial pits of about 1200 BC is uncertain, as local bronze finds later than the pits are almost entirely lacking.363 However, the recent discovery of the site of Jinsha in a western suburb of Chengdu gives some indication of a continuing bronze industry in the general vicinity if not at Sanxingdui itself.364 362As discussed in Section 4.1, the few bronze plaques found elsewhere might be imports from Erlitou or later copies at Sanxingdui. They do not constitute evidence for local bronze casting earlier than the thirteenth century. The objects bearing transition-period designs include the life-sized statue in Figure 5.17 (the decorated pedestal), the hybrid figure standing on birds in Figure 5.22, the clapper bell in Figure 5.33b, and the plaques of taotie faces in Figure 5.51. 363A bronze zun with a slender body, found outside the west gate of the old Guanghan city, appears to date from the 11th century BC (see Beijing 1994c, pl. 91, and Li Xueqin 1998). 364For bronzes from the Jinsha site see Beijing 2002c (pp. 37–72), and discussion in Chapter 10 of the present dissertation. In addition, two bronze zhi and a zun with a slender body unearthed in 1959 from a hoard found at Zhuwajie in Peng Xian are probably late Shang or early western Zhou, but they are commonly regarded as imports to Sichuan. See Wang Jiayou 1961 (p. 29, p. 5 pl. 3), Xu Zhongshu 1962, Feng Hanji 1980, and Bagley 2001 (p. 178 figs. 2–3, p. 179). 178 Nevertheless the bronze industries at Sanxingdui and in the Central Plain were in some respect very different. Here, we may first notice that metal supplies at Sanxingdui may not have been as abundant as in the Central Plain. The hundreds of bronzes from the two pits weigh about 1000 kg altogether—no small quantity of metal. Yet it is not much by Anyang standards; only a little more, in fact, than the 875 kg Si Mu Wu fang ding, a single Anyang vessel, though the largest known. As discussed in Section 5.1.7, the bronze heads from K2 are thinner and lighter than those from K1, and more numerous but less varied. Perhaps Sanxingdui founders learned to standardize their designs and to make more objects out of a same amount of metal. They succeeded in making extremely impressive objects, often very large, while maintaining remarkable thinness. The monumental bronze tree in Figures 5.30/6.32 is a good example. Figure 6.32i shows the cross section of one of its broken branches before restoration; the metal part is very thin. Even though we cannot rule out the possibility of technical reasons for casting the images with thin walls, it seems quite possible that Sanxingdui founders had more limited metal supplies but used the Central Plain technology anyway because it was what they had learned. 6.2.3.2 Characteristics of the foundry practice In Section 5.1.2, I proposed to identify an “one-pour mentality” in Central Plain foundries, or in other words, a preference for complex molds and few pours: the foundry invested most of its effort in the mold-building stage. Sanxingdui founders were less systematic, fabricating images of the same type in a variety of ways. Given their relatively simple shapes, the flat-topped heads could all have been cast in one pour, but 179 instead they were made in at least three different ways, as though they founders were experimenting. Of those from K1, the head in Figures 5.10/6.16 was cast in one pour, but another two heads have separately cast top panels (K1:3 with cast-on top, and K1:11 with precast top), while two others are without tops but must originally have had or been intended to have a top panel (K1:8 and K1:72). For a foundry that was capable of casting the head in Figure 5.5/6.18 in one piece, the decision to make simpler heads in several pours is curious. Perhaps it means that the founders made so many objects for which multiple pours were necessary (e.g. the bronze tube with dragon in Figure 5.47) that they became accustomed to making joins, and were more comfortable with multiple joins than with complex mold-building. While some of the flat-topped heads from K2 were cast in one pour, a larger number were made by separate casting and joining, some having precast top panels but more having cast-on panels or patches. Still more have no top panel.365 Other types of images, with few or no equivalents in K1, show the same variation. The mask in Figure 5.27, the largest of the intact human masks, was cast in one piece, as were a group of small masks.366 However, of more than a dozen masks of medium size, only two were cast in one pour, while the others were made by casting the ears separately and then joining them to the faces, as seen in Figures 5.26 and 6.27.367 365According to my tentative examination, the thirty nine flat-topped heads from K2 include 5 cast in one pour, 13 cast in separate pours (2 with precast top panels, 3 with cast-on top panels, 7 with cast-on patches on top, and 1 with a large hole clearly meant to be covered with a cast-on patch like the other seven), and 18 without top panels but probably having had them in their original state or intended to have them. 366The small-size masks, on average 15 cm high, comprise K2(2):43, K2(3):57, K2(3):65, K2(2):102, K2(3):119, and K2(2):331. 367The medium-size masks are about 25–26 cm in height. The two one-pour masks are K2(3):14 and K2(2):60. Those with separately cast ears include, in addition to the two in 180 The bases of the two trees in Figures 5.30/6.32 and 5.40/6.26 provide another interesting contrast. The structures of the two bases are similar, but curiously, though the former is more than 1.5 times the size of the latter, it was cast in one piece, in a mold divided on the axes of the tripod legs. The smaller base was joined together from three arches. One gets the impression that the founders had no fixed way of making these objects—perhaps because they had never made such objects before—and were exploring the options. The two supernatural masks in Figures 1.3/5.28/6.20 and 5.29/6.29 are again similar objects made differently, the first by precasting, the second by a mixture of methods, as discussed already. The ears of the human masks also appear to have been joined by several techniques chosen apparently at random. The mask whose ears are shown in Figure 6.34 had its left ear soldered onto the face (Fig. 6.34a); on its right ear, however, is a back piece on which a sprue still survives (Fig. 6.34b), indicating that the back piece should be considered either as a run-on join or a cast-on component.368 Both ears were soldered on the front side. Although this mask seems to be a unique case in which two joining techniques were used simultaneously on the same mask, the free choice of joining techniques is apparent in other masks as well. In the flexible selection of casting techniques, we may detect a tendency for fabricating an object by joining separately cast components rather than casting in one pour, though Sanxingdui founders used both options. This tendency is manifest also in Figures 5.26 and 6.27, K2(2):33, K2(2):53-1, K2(2):57, K2(3):58, K2(2):100, K2(2):1091, K2(2):111, K2(2):114 (Fig. 6.34), and K2(2):131. 368Perhaps the characterization of the back piece as a run-on join is more appropriate, because the piece on other masks is usually an irregular patch though it occasionally follows the contour of the ear quite closely. In any event, the back piece served to lock the ear to the face. 181 the fabrication of complex images like the life-sized figure and the monumental tree. Unlike their Central Plain counterparts, Sanxingdui founders show no inclination to minimize the number of pours and thus of joins. A case in point is the bird in Figure 6.32g, one of the nine perched on the tree. A Central Plain founder would probably have cast this bird in one piece, using a rather complicated mold. The Sanxingdui founder instead assembled it from multiple components, all cast in simple two-section molds. In fact, the Sanxingdui founder built up the whole tree from components cast in the simplest possible way, apparently quite untroubled by the number of joins that this approach entailed. The practice at the Sanxingdui foundry described above is thus at the opposite extreme from the Central Plain founders’ meticulous attention to complex moldmaking—their “one-pour mentality.” I therefore propose to call it a “simple-mold mentality”. To understand this reliance on simple molds, we must notice several related factors making it possible. The different surfaces of Sanxingdui images and the Central Plain vessels is one such factor. The Sanxingdui images usually have large, bold features, many of them capable of impressing the viewer from a distance; vessels have intricately detailed surface decoration that invites close scrutiny and thus requires fine finish. Conspicuous joins are far more acceptable on the images than on vessels. For example, the run-on join between the base and the trunk on the monumental tree is quite sloppy (Fig. 6.32a), but viewers confronting the tree would be very unlikely to notice—whether in its original setting or in a museum gallery today. Similar untidy joins appear on other images. In this context, casting components in simple molds and then joining them 182 together was probably significantly less demanding and painstaking than building complicated molds and eliminating joins. Even though K2 contained objects far more complicated than anything from K1, few required molds more complex than that for the K1 head in Figures 5.5/6.18. Evidently, founders at Sanxingdui never learned or had the need to learn the complex mold building needed for elaborate vessels, and they would probably have been quite unable to cast the sort of bronze vessels that their patrons imported from the middle Changjiang region.369 Another factor favoring separate casting and joining might be limited metal supplies, since this method made it easier to cast components with thin walls, and thus save metal. The preceding discussion leads us to conclude that the Sanxingdui bronze industry is technically distinctive from Central Plain foundries not because it used different techniques but because it used the same techniques with a very different emphasis, an emphasis that was probably fostered by aesthetic, technical, and economical considerations. To cast an object so wildly complicated as the monumental tree, Sanxingdui founders dissected it into a multiplicity of components that, individually, were easy to cast. Complexity was achieved by relentless joining. 369It should be noted that the quality of Sanxingdui castings is often poor. Casting defects in the forms of holes are common, and repairs by casting on are extensive. Many defects were not repaired, as seen, for example, on K1:5 (head: top and left side; Figs. 5.5/6.18), K2(2):22 (head: top), K2(2):77 (head: right side), K2(2):78 (head: back), K2(2):107 (head: pigtail), K2(2):121 (head: top), K2(2):57 (mask: nostrils), K2(3):264 (figure with an animal headdress: right side of the face; Fig. 6.30d). 183 Chapter 7. Lithic and Gold Artifacts at Sanxingdui In contrast to bronzes, few of which have been found outside K1 and K2, ritual artifacts made of jade and stone have been unearthed at numerous locales at the Sanxingdui site since the first discovery at Yueliangwan in 1927. Stone artifacts have been found in cultural strata as well as in burials, jades mostly in burials. The burials include all the nine pits as well as some of the tombs at the Rensheng village.370 Current statistics put the count of lithic artifacts identifiable as ritual implements at more than 1000 of various types.371 Beyond the Sanxingdui site, finds were also made at related satellite sites such as Gaopian and Mayang.372 The present chapter examines these lithic ritual implements: their types and typological sources, their functions, and the production of such implements at Sanxingdui. The chapter will also examine gold artifacts. These have been found only in K1 and K2, once again testifying to the extraordinary status of the two pits. Though far fewer than bronzes and lithic artifacts, they constitute the largest amount of gold known from Bronze Age China. 7.1 Lithic artifacts 7.1.1 Types, typological sources, and functions 370For artifacts from cultural strata in major excavations see reports referred to in Section 2.1.5. For general descriptions of the burials containing jade and stone artifacts and their relevant bibliographic references see Sections 2.3 and 3.2. 371Chen et al. 1998, p. 40. This figure apparently does not include nearly 400 small jade beads from K2. On the other hand, functional stone tools unearthed at Sanxingdui number in thousands (see ibid.). They lie outside the scope of the present dissertation. 372For general descriptions of the finds at Gaopian and Mayang and their bibliographic references see Section 2.3.2. 184 On the following list the types included in each find are enumerated.373 The artifacts may be roughly divided into stone and jade, the latter word being understood as a conventional designation for fine stones not necessarily limited to nephrite.374 (1) 1927 Yueliangwan pit Stone: bi disk (20 or so of graduated sizes). Jade: collared ring, cong tube, forked blade, axe, chisel, bead.375 (2) 1974 Suozitian pit Jade: “whetstone” (several dozens). (3) 1976 Gaopian pit 373Some pits are not listed here because the types have not been published. 374The excavation report for K1 and K2 (Beijing 1999a) is confusing and contradictory in its classification of lithic materials. Its two basic categories are “jade” and “stone,” the first understood in the general sense of “fine hardstone” rather than meaning nephrite specifically. Artifacts were assigned to one category or the other by visual inspection alone, without instruments; the more specific mineral identifications given in tables 5–8 and 26–8 of the report were apparently made in the same way. However, the authors of the report’s petrographic appendix (pp. 500–14) point out that nineteen samples studied under a microscope proved in almost every case to have petrographic structures at odds with the identifications made without microscopic examination, and they conclude that simple visual inspection is unreliable. The implication seems to be that the report publishes mineral identifications in which the investigators have little confidence. What is clear is only that a wide variety of lithic materials was used at Sanxingdui, ranging from fine nephrite to coarse sandy rock. Whereas coarse sandy rock is easy to identify, it is often impossible to differentiate jades from other fine stones. Therefore, even though counts of lithic artifacts given in the following discussion are divided into “jade” and “stone”, they should be understood as approximations designed to give an impression as to the ratios between objects made of fine and coarser stones. On a separate note, Beijing 1999a’s sorting of the lithic types by supposed functions, i.e., objects for ritual use, objects for ceremonial display, and tools, is equally dubious, because these functions are not clearly defined and could very well overlap. Moreover, the actual use of those lithic types is far from clear on present evidence. The reader will therefore find no simple correspondence between counts given in the present dissertation and counts given in Beijing 1999a. 375As pointed out in note 100, the exact inventory of the pit is not known. The terms “axe”, “chisel”, “spearhead”, “adze”, and “hatchet” used here refer to artifacts that have those shapes but should not be taken to imply any confident functional identification. 185 Jade: axe (one), axe-shaped blade with concave edge (one), spearhead (one). (4) 1986 K1 Stone: ge blade (27+), spearhead (two), axe (37), adze (two), chisel (one), rectangular slab (one). Jade: collared ring (three), irregularly shaped implement with collared perforation (three), cong (one), trapezoidal implement (one), forked blade (five), ge (18), hybrid of forked blade and ge (34), hatchet (one), sword-shaped implement (one), axe (12), adze (11), chisel (35), “whetstone” (two).376 (5) 1986 K2 Stone: ge (10), bi (one), collared disk (two), collared ring (two). Jade: collared disk (two), collared ring (eight), parallelogram-shaped implement (four), forked blade (13), miniature forked blade (three), ge (21), knife (one), adze (two), chisel (43), “whetstone” (six), globular or tubular bead (380).377 (6) 1987 Cangbaobao pit 376The categories used here are strictly by morphology and differ in some cases from the classification in Beijing 1999a. The correspondence between the categories in the two schemes is as follows (in each pair, the first term is the one employed here, the second the one employed in Beijing 1999a). For stone types: spearhead = mao; adze = jin; axe = chan, fu; chisel = zuo. For jade types: trapezoidal implement = zhang type A; forked blade = zhang types B, C, E; hybrid of forked blade and ge = zhang type D; hatchet = qixing pei; sword-shaped implement = jian; axe = fu; adze = ben, jin, aoren zuo; chisel = zuo; collared ring = huan, yuan; irregularly shaped implement with collared opening = qixing bi; “whetstone” = moshi. 377The correspondence between the categories here and those in Beijing 1999a is as follows. For stone types: collared disk = bi; collared ring = yuan. For jade types: parallelogram-shaped implement = zhang type A; forked blade = zhang types B, C; miniature forked blade = zhangxing shijian; knife = dao; adze = fu, jin; chisel = zuo; collared disk = bi; collared ring = huan, yuan; “whetstone” = moshi; globular and tubular bead = zhu, guan. 186 Stone: bi (21 of graduated sizes, including some reworked from the centers of large disks), miniature cong (one), axe (three). Jade: ring (eight), bracelet-like ring (one), chisel (one). (7) 1987 Mayang pit Stone: bi (10 of graduated sizes). Among the types enumerated above, many were made in both stone and jade, including collared disks, collared rings, cong, ge, spearheads, axes, adzes, and chisels. Other types occur only in jade, while sets of bi of graduated sizes were made exclusively in coarse stone. The materials and the history of lithic production at Sanxingdui will be discussed in Section 7.1.2. The range of size among the artifacts, though less dramatic than that of the bronze imagery, is still pronounced. The largest stone bi, from the 1927 pit at Yueliangwan, measures 70 cm in diameter and weighs more than 50 kg (Fig. 2.4); the longest object is the trapezoidal implement from K1, measuring in its damaged state 1.62 m long (Fig. 7.6a). At the other end of the range, the smallest bi, from Cangbaobao, has a diameter of a little over 3 cm (Fig. 7.1b), and the miniature cong, also from Cangbaobao, is about 3 cm tall.378 The miniature forked blades from K2 are less than 5 cm long.379 In a broad sense, all the lithic artifacts from the pits were implements of ritual offerings, judging from their burial context and the kinds of treatment they were subjected to before burial.380 Their particular uses are far from clear in most cases, 378The cong was found by the farmers who came upon the pit and was discarded because it was badly broken. 379For illustrations see Beijing 1999a, p. 369 fig. 201, p. 373 pl. 142.2–4. 380See discussion in Section 2.3.2. 187 however, though the following examination of individual types may suggest some possibilities. 7.1.1.1 Sets of stone bi of graduated sizes Three sets of stone bi of graduated sizes have appeared so far, one from the 1927 Yueliangwan pit, one from the 1987 Cangbaobao pits (Fig. 7.1), and one from the 1987 Mayang pit. Their number, sometimes exceptionally large size, and their coarse material hint at local manufacture. Additional fragments of bi were unearthed from cultural strata in both the 1963 excavations at Yueliangwan and the 1980–81 excavations at Sanxingdui, further strengthening the probability of local production and affording important clues to date. At Yueliangwan the fragments were excavated from the upper cultural stratum, datable to Phases III and IV, and at Sanxingdui from strata datable to Phases II and III. Significantly, none was found in Phase I strata at either location.381 Moreover, no such bi have appeared in any of the Baodun settlements. The bi therefore most likely appeared in the Chengdu Plain for the first time during Phase II, and it probably lasted through Phase IV. The quality of these disks varies in regularity and polish. The perforation, often off center, was drilled from one side, leaving behind clear traces and slanting walls. Among the bi from the Cangbaobao pit, the ten smaller ones were re-worked from the centers of the large ones (Fig. 7.1b).382 These typological features and manufacture peculiarities point to a close tie with the Qijia Culture (ca. 2000 BC) in northwest China, 381Ma Jixian 1992 (pp. 321–2), and Kaogu xuebao 1987.2 (pp. 230–1, 248–9). 382Beijing 1998a, pp. 85–7. 188 as several scholars have suggested (Fig. 7.2).383 On the other hand, large bi as a type seem to have ultimately derived from the Liangzhu Culture (ca. 3200–2000 BC), where on present evidence it first appeared. In another direction, the use of disks in graduated sizes finds a parallel in the tomb of Fu Hao, from which a set of eight jade bi of graduated size were excavated,384 but it is not clear how closely related these two contexts of occurrence are. Equally puzzling is the question of how these sets of bi functioned, a question that has generated a wide range of different opinions. Zheng Dekun proposed that they might have been used as potter’s wheels, as currency, or as wheels for grinding purposes.385 Zhang Xunliao, followed by Feng Hanji and Tong Enzheng, regarded them as weights.386 Shen Zhongchang and Huang Jiaxiang conjectured that the disks constituted symbols of wealth and status.387 And recently Chen Jiangfeng and Zhou Tiexiang have expressed the view that the disks, when stacked, formed an axis mundi, by which souls of the deceased could ascend to heaven.388 Unfortunately all these ideas find little corroborating evidence in the field. The sets of bi may certainly be identified as ritual implements, judging from the fact that they were smeared with a red pigment, always neatly arranged, and placed 383Yang Meili 1994 (pp. 69–70, pp. 77–9), Yang Meili 1999 (p. 33), and Jenny So in Bagley 2001 (pp. 159–60). However, Yang’s assertion that the custom of burying a large number of bi in the Qijia Culture was a Liangzhu influence transmitted via Sanxingdui is debatable. 384Beijing 1984a, pp. 118–9. Moreover, at least one small disk from the tomb of Fu Hao (artifact no. 1100) appears to have been re-worked from the center of a larger bi (p. 119). The relationship between the Qijia Culture, the Sanxingdui Culture, and the Shang culture of Anyang is a topic awaiting future study. 385Zheng Dekun 1946 (p. 34), Cheng 1982 (p. 58). 386Feng & Tong 1979, p. 35. 387Shen & Huang 1986, pp. 4–5. 388Chen & Zhou 1995, pp. 15–7. 189 together with burnt bone and ash in the 1987 Yueliangwan pit. But no specific function or ritual purpose can be asserted on present evidence. Interestingly, stone bi, in sets or otherwise, are conspicuously absent from the lithic inventory of K1 or K2. The absence cannot be explained by the difference in time or material. As noted earlier, stone bi were probably manufactured from Phase II through Phase IV, and other artifacts of coarse stone certainly appeared in the two pits in considerable numbers. It seems that bi were not required for the kind of ritual represented by K1 and K2. 7.1.1.2 Collared types, and other circular types Collared types at Sanxingdui comprise ring, disk, and a curiously irregular shape (Fig. 7.3). The disks and rings have about the same inner diameters (i.e. central perforations), only their outer diameters differ significantly. In the disks of Figures 7.3d– e, the outer diameter is more than double the inner diameter. When the diameter is smaller than this, as in Figures 7.3a–c, we will call the object a ring. The distinction between the two types is obviously arbitrary and a matter of convenience, since we have no evidence to show that the two objects served different functions or to show that their makers thought of them as different objects. In either type, some have plain surfaces, others are decorated with incised concentric lines alternating with shallow grooves. Jade collared rings were contained in the 1927 Yueliangwan pit, and in K1 and K2, but jade collared disks appeared only in K2. K2 indeed included jade and stone versions of both types. Collared rings and disks have been excavated at a number of contemporary sites across long distances in other regions. Several finds have been made at Anyang. The 21 190 pieces from the tomb of Fu Hao there are the largest group found anywhere until the find at the Jinsha site.389 Several sites in the middle Changjiang region have yielded examples, including the tomb at Xin’gan Dayangzhou and three burials in Hunan province.390 The objects are all similar in proportions and surface treatment; they also seem to form graduated sets with similar ranges of size.391 Undoubtedly, the collared rings and disks in these three regions had a common origin. 389Beijing 1984a, pp. 119–22. Nearly 200 jade collared disks and rings have recently been unearthed at Jinsha (Beijing 2006b, pp. 22–3), but their burial context and date are uncertain. See Section 10.2 for discussion on the Jinsha site. 390For the Xin’gan find, which comprised nine jade collared rings and disks, see Beijing 1997e, pp. 141–5. The Hunan jades seem to include only rings. One group, including 14 collared rings, was contained in a bronze you unearthed at Huangcai in Ningxiang county in 1970 (see Fong 1980, pp. 129–31 no. 25, and Yu Yanjiao 2002, pp. 43–4, p. 44 fig. 2). Another group, including five collared rings, was found in 1973, also at Ningxiang Huangcai, near a large bronze nao bell (see Yu Yanjiao 2002, p. 44 fig. 6, p. 45). The third group, including four collared rings, was found inside a bronze you unearthed at Xinghuacun in Hengyang city in 1985 (see Zheng & Tang 2000, and Yu Yanjiao 2002, p. 46). 391The three collared rings from K1 have the following overall diameters/perforation diameters (all measurements are given in centimeters): 14/6.6, 7.9/5.4, 7.2/5.1. The collared disks and rings from K2 have the following measurements: (two jade disks) 17.8/6.7, 17.5/6.7. (two stone disks) 17.5/7.5, 13.2/5.8. (eight jade rings) 13.7/6.8, 12.2/6.5, 11.7/6.5, 11.6/6.5, 10.6/6.4, 8.6/5.8, 8.5/5.5, 8.2/6.2. (two stone rings) 12.2/6.6, 9.5/5.5. The collared disks and rings from the tomb of Fu Hao have the following measurements: (four disks) 18.9/6.2, 18.6/6, 16.2/6, 15.2/7.5. (16 rings, one other was significant altered and not included in the statistics given here): 12.8/5.7, 11.4/5.7, 10.8/5.8, 10.5/5.1, 10.3/5.3, 10.2/5.5, 9.8/5.4, 9.5/5.7, 9.3/5.4, 8.8/5.6, 8.4/5.3, 8.1/5.6, 8/5.5, 7.8/5.4, 7.7/5.3, 7.2/4.7. The collared disks and rings from the tomb at Xin’gan Dayangzhou have the following measurements: (two disks) 18.4/7.5, 16.8/7.2. (seven rings) 11.5/7.1, 10/6.8, 10/6.5, 8.9/6.6, 8.3/6, 8/5.5, 7.8/5.5. The individual sizes of the collared rings from Hunan Ningxiang Huangcai are not known, but it is clear that they all vary in outer diameter, from 10.8 cm to 6.8 cm (the fourteen rings found in 1970), and from 10.2 cm to 7.2 cm (the five rings found in 1973). 191 On present information, collared rings appear sporadically but persistently in the archaeological record of the Central Plain from the Neolithic Longshan Culture through the Erlitou period; in addition, a single example is known from Erligang.392 As for the collared disk, however, no example has been found in the Central Plain, nor is the concentric surface pattern known there. While it may be reasonable to suppose that both types arose somewhere in the Central Plain,393 on present evidence it seems impossible to pinpoint the place of origin. Some of the jade disks and rings found at Sanxingdui may have been imported from elsewhere. At least two of the bronze vessels from K2 contained jade collared rings at the time of excavation, lei K2(2):88 in Figure 4.17 and zun K2(2):127 in Figure 4.20 (the lei contained also a large number of jade beads and tubes as well). The placement of jades inside vessels immediately recalls two of the Hunan finds, and as Section 4.2.2.4 has argued, the vessels themselves were almost certainly imported from the middle Changjiang region. It is possible that the Sanxingdui users imported not only the vessels The ranges are comparable to those of the collared rings from K2, the tomb of Fu Hao, and the tomb at Xin’gan Dayangzhou. Of the four collared rings from Hunan Hengyang Xinghuacun, one has a teethed outer edge, so far unique, and another two are of the same size. 392Yang Jianfang 1994, Yoshikai Masato 1999. Yoshikai Masato 1999 provides a comprehensive though not exhaustive list of collared rings and disks dating from the late Neolithic to Western Han. All the examples from the Longshan and Erlitou culture sites in Henan are fragments (no example has been found at Erlitou itself). An intact example was collected in field survey at the Simatai site, Haiyang county in Shandong province (Wang Hongming 1985, p. 1062). Whether it comes from the Shandong Longshan Culture is under debate. Accepting it as a Shandong Longshan product, Yang Jianfang 1994 further postulates that the type may have originated in the Dawenkou Culture of Shandong, where an ivory ring with collar on one side was excavated from a tomb dating from the early phase of the Dawenkou Culture (ca. 4300–3500 BC). For an illustration of the Dawenkou ring see Beijing 1974, p. 102 fig. 87.4, pl. 94. The single Erligang collared ring is recently published in Beijing 2001f, vol. 2, p. 929 fig. 620.4. 393Yoshikai Masato 1990 holds the same opinion (p. 89). 192 but also some of their contents.394 The rings and disks made of coarser material were likely local copies after imported originals. Moreover, this lithic type was copied at Sanxingdui in bronze also, a unique phenomenon (Fig. 5.39). Opinions vary regarding the function of the collared ring and disk, with bracelet being the most commonly proposed, though the evidence for this or any other use is decidedly weak.395 It is important to note that objects used in different cultural settings need not have had the same function in each setting. The approach advocated by Yoshiaki Masato, that functions must be analyzed individually with regard to region and time, is commendable.396 At Sanxingdui, the lithic and bronze disks and rings may have been used as ornaments for bronze trees, as suggested in Section 5.2. It is not clear if they were ever used as bracelets. On the other hand, the type of circular artifact with a tall but very narrow wall, represented in the 1987 Cangbaobao pit (Fig. 7.4a), may well have functioned as bracelet, for a miniature bronze figure of a gesturing person recently unearthed at the site of Jinsha wears bracelets of similar shape (Fig. 10.12). The figure’s physiognomy leaves no doubt of its connection with Sanxingdui. The Jinsha jades also include eight similar braceletlike rings, the two published examples appearing somewhat taller than the one from Cangbaobao (Fig. 7.4).397 A still taller example was unearthed at Erlitou but the basic 394Jenny So has hypothesized that some jade disks and rings at Sanxingdui may have been imports because their fine material is exceptional (Bagley 2001, p. 154). Sun Hua 1993d, however, proposes that the collared disk and ring originated at Sanxingdui. As Yang Jianfang 1994 points out, Sun’s idea is factually and logically untenable. 395Feng & Tong 1979 (p. 34), Yang Jianfang 1994, Hayashi Minao 1997 (p. 83 note 135), and Yu Yanjiao 2002 (pp. 47–8). 396Yoshiaki Masato 1999, p. 90. 397Beijing 2002c, pp. 89–91; Beijing 2006b, pp. 141–2. 193 shape is consistent.398 The possibility of an Erlitou connection is reinforced by the bronze rectangular plaques found in the same pit at Cangbaobao, for those surely derived from Erlitou originals (see Section 4.1). The irregularly shaped implement with collared perforation seen in Figure 7.3f is a type unique to Sanxingdui, but the collar argues for a relationship of some kind with the collared ring and disk. Another local peculiarity is the small rectangular hole cut through the collarless jade rings excavated from the 1987 Cangbaobao pit (Fig. 7.4b). But the rings themselves are of a type common since Neolithic times and found in many regions. 7.1.1.3 Cong tubes Conventionally called cong, this jade type has the form of a cylindrical tube contained in a square prism. The bore of the tube projects above and below the square to form two collars. Cong were found in the 1927 Yueliangwan pit, in K1, and in the 1987 Cangbaobao pit. The one from Cangbaobao was discarded by the farmers who discovered the pit, but allegedly it was a miniature 3 cm tall.399 Three are known to have come from the 1927 pit, and K1 yielded one (Fig. 7.5).400 The four cong at Sanxingdui are far removed from their Liangzhu forebears in surface treatment. With its pairs of circles at the corners, the one in Figure 7.5a still recalls the Liangzhu face motif. The one in Figure 7.5b is decorated only with incised straight lines, however, and the other two are plain. Cong with simple or no surface decoration seem to belong to a lithic industry active in the northwest region from the end 398See Beijing 1999d, p. 257 fig. 168.8. 399Beijing 1998a, p. 78. 40024 cong have been unearthed at the Jinsha site; see Section 10.2 for discussion. 194 of the third millennium to about the twelfth century BC.401 The Sanxingdui examples were part of this tradition,402 but it is not clear whether they were local products or imports, nor is their function known. The plain ones are similar to cong from the tomb of Fu Hao, once again demonstrating a cultural network in which artifact types were shared across long distances.403 7.1.1.4 Trapezoidal and parallelogram-shaped implements Although damaged at both ends, the object from K1 in Figure 7.6a originally was probably an isosceles trapezoid. So far the only representative of its type at Sanxingdui, it is also the longest among the lithic artifacts, measuring 1.62 m in length. The shape is well known elsewhere as a type of blade derived from a Neolithic harvesting knife.404 The closest parallel is a blade from a tomb at Erlitou decorated with similar incised patterns, a network of crisscross lines bordered by parallel lines (Fig. 7.6b).405 The Sanxingdui object appears to be an enlarged version of the Erlitou type and was probably a local product. Certain differences from the Erlitou prototype suggest that the Sanxingdui implement was not thought of as a knife. Its longer edge is not sharpened or 401Dai Yingxing 1993 (part 1), Deng Shuping 1993 (part 1), and Rawson 1995 (p. 150– 1). 402The cong in Figure 7.5a, stylistically the earliest among the four, finds close comparisons with two cong collected at Lushanmao in Yan’an, Shaanxi province (see Mou & Yun 1992, nos. 43, 44). For other comparable examples see the references given in the preceding note. 403For similar cong from the tomb of Fu Hao, classified by the excavators as Type II, see Beijing 1984a, p. 115. 404For accounts of this type of knife see Fong 1980 (pp. 76–7), Dai Yingxing 1993 (part 3), Rawson 1995 (pp. 184–5), Hayashi Minao 1997 (pp. 286–349), and Yang Meili 1997. 405Kaogu 1978.4, Beijing 1999d (pp. 341–2). Two other knives unearthed at Erlitou have the same shape, attesting that the type was well established in the Erlitou lithic inventory, but they do not have surface decoration (see Beijing 1999d, p. 250 fig. 162.6; Chen & Fang 1993, pl. 11.) 195 beveled, and it lacks the holes that were originally meant for hafting the reaping knife to a handle. A similar but much shorter implement, also without the cutting edge, was unearthed from a tomb at Weigang in Shimen county, Hunan province (Fig. 7.6c).406 The date of the Sanxingdui implement may be inferred from the dates of the Erlitou and Weigang tombs: the former is assigned to the end of the Erlitou occupation, the latter to late Erlitou or early Erligang, putting both of them around 1500 BC.407 The production date for the Sanxingdui object is therefore likely to fall between 1500 and the time of K1, just before 1200 BC. Such a large artifact would surely have been treasured and probably occupied a conspicuous position in the local ritual. While the long object from K1 resembles items from Erlitou and the middle Changjiang region, the four parallelograms from K2 are rather different from anything known elsewhere (Figs. 5.42, 7.7). The surface decoration seen on the two examples in Figure 7.7 consists of lightly incised patterns: parallel lines flank rectilinear spirals similar to those seen on Erligang bronzes and on some of the Sanxingdui images.408 The objects can therefore be dated between the Erligang period and the time of K2, i.e., 1500– 1200 BC. The type represented by the implement in Figure 7.7b is, strictly speaking, not a parallelogram. It has a short tang-like portion that is slightly narrower than the rest of the object; a hole is drilled at the border between the tang and the body. These features argue 406Wang & Long 1987, p. 17 fig. 8.1; also Yang Meili 1999 (part 1), p. 33. 407For the date of the Weigang tomb see Wang & Long 1987, p. 17. 408A parallelogram identical in shape as Figure 7.7a was recently unearthed at Jinsha (Fig. 7.7c). Its surface bears an extraordinary design, four times repeated, showing a kneeling human figure carrying on the shoulder what looks like a curved elephant tusk (its forked end tantalizingly resembling that of a forked blade). Beneath the figure is a border composed of parallel lines flanked by rectilinear spirals on both sides, an interesting reverse of what appears in Figure 7.7a. 196 that the shape is related in some way with the forked blades to be discussed below. The implement with incised design in Figure 5.42 has a similar shape, but again, it cannot be categorized as a blade, as it does not have a cutting edge at the top. The perfectly parallelogram-shaped type has no such edge either (Fig. 7.7a). 7.1.1.5 Forked blades Blades of various types constituted the bulk of the lithic inventory in K1 and K2 (not counting the beads in K2), and they were present in other pits also. None were functional; all were made specifically for use in ritual activities. We will first discuss forked blades. The forked blade is one of the most widely distributed lithic types in ancient China (Fig. 7.8), but also one of the most enigmatic in origin, history, and usage.409 Many scholars have studied its origin and evolution.410 Here I will first sketch my understanding of the stylistic development so as to provide a context for the ensuing analysis of the blades at Sanxingdui. The defining feature of the type is a spreading blade that culminates in an inwardcurved, sharpened, always asymmetrical end. The curvature of the end varies, and so do the proportions of the blade, though less obviously. The greatest variation is in the flanged zone that separates the blade proper from the tang. The stylistic development of 409The blades from Sanxingdui shown in Figure 7.8 include hybrids of forked blade and ge, which are treated separately in the present dissertation. 410A rich array of studies has been published on the subject. For example, nearly half of the papers published in Deng Cong 1994 are devoted to forked blades. A review of them is not possible here, but relevant studies or arguments will be referred to in the following discussion. 197 the forked blade may be analyzed in terms of these features. A hole in the tang is customary. Among all the known subtypes of the forked blade, the two from Shaanxi Shenmu Shimao and Shandong Linyi Dafanzhuang seem to represent the most primitive, each featuring a simple, plain surface, a gently curved end, and a pair of short projections that suggest a crosspiece separating blade from tang (Figs. 7.9a, b).411 On another example from Shimao (Fig. 7.9c), the flanges have become more elaborate, with notches cut into the sides. Yet another example from Shimao shows further elaboration in the form of the flanges: with a spiked silhouette (Fig. 7.9d). Similar spikes appear on the sides of the blade, anticipating the fully developed flanged zone seen in the example from Erlitou in Figure 7.9e. On the Erlitou piece, the crosspiece has become a decorated zone framed by two sets of flanges. The flanges at bottom of the blade proper retain the simple shape of the earliest form, while the ones marking the tang compare closely with the flanges of Figure 7.9d; moreover, the teeth in between the two sets of flanges are similar to those on the sides of the blade in Figure 7.9d. Such comparisons suggest the origin of the Erlitou form. From this point on, forked blades commonly have a well-defined zone between the blade proper and the tang, marked usually with two sets of projecting flanges. Much variation would happen in the details of this zone but the basic structure would stay 411For forked blades from Shimao and studies of them see Dai Yingxing 1993 (part 2), Dai Yingxing 1994, and Zhang Changshou 1994. For forked blades from various locales in Shandong see Yu Qiuwei 1998. The origin of the forked blade remains an unsolved problem. The argument by Hayashi, first presented in 1969, that this jade type was derived from the bone plows of the Neolithic period is influential among scholars (see Hayashi Minao 1997, p. 61), but its likelihood is not strong. Bagley argues for a metal ancestor and cites a long blunt-ended axe at Erlitou as representing a possible prototype (see Fong 1980, pp. 75–6). It is important to observe that the jade forked blade and the Erlitou bronze axes share an essential feature of tang whereas plows do not have tangs. 198 constant. The Erlitou form, seen also in Figures 7.9f–h, therefore represents a crucial moment in the history of the forked blade. The most sophisticated treatment is seen on the Erlitou blade in Figure 7.9h, which features a series of about forty parallel grooves grouped by fours and fives. All the edges—those of the flanges as well as the area between them—are notched with deep cuts to produce a jagged silhouette.412 Furthermore, each of the two flanges at the tang consists of two units, the one nearer the tang looking tantalizingly like a downward-facing head of some sort of creature. The sharpened end on the Erlitou blades does not change much from that of the earlier examples, remaining a gentle inward curve.413 After the Erlitou period, the forked blade 412For a detailed description of this blade see Fong 1980, pp. 75–6. The tombs from which the Erlitou blades came have been dated according to the four-phase periodization of the Erlitou site: the tomb containing the blade in Figures 7.9e, f belongs to Phase III (Kaogu 1983.3, p. 202), so does the tomb for Figures 7.9g (Beijing 1999d, p. 241); and the tomb for Figure 7.9h belongs to Phase IV (ibid., p. 341). This sequence is consistent with the idea that the blade in Figure 7.9h represents the latest form of the zone between the blade proper and the tang. 413A seminal study of the typological sequence of the forked blade is Li Xueqin 1992, which identifies the forked blades from Shandong Linyi Dafanzhuang as the simplest, i.e. earliest, in form. Forked blades from Shimao are deemed somewhat later, followed by the Erlitou blades. Li, however, does not produce a detailed stylistic analysis to support his sequence. My understanding of the typological development basically concurs with his, but I do not think that the blade from Dafanzhuang (Fig. 7.9b) is necessarily earlier than the Shimao blade (Fig. 7.9a). It is impossible at present to determine the relative date between the Shandong ones and the stylistically earliest examples from Shimao. Hayashi has produced so far the most meticulous study of typological development of the forked blade (see Hayashi Minao 1997, chapter 6), but his methodology of relying almost exclusively on the form of the flange is questionable, and his classification is excessively minute, taking almost every change in the detail of the flange as chronologically significant. The resulting typological sequence and absolute dating are therefore fraught with problems or evidently wrong, for example, in the case of the forked blades from Yueliangwan (to be discussed below), which he dates to the late Neolithic period of the Longshan Culture. Caution should also be voiced regarding the typological sequence described here. As Robert Bagley remarked to me (personal communication, May 2003), it is in principle dangerous to construct a typological sequence out of examples from all over China, as one may be shuffling together random survivors from a whole series of loosely related local sequences. The present typological sequence should therefore be 199 seems to have virtually disappeared in the Central Plain; only one stray example is reported from the Erligang area (Fig. 7.9i), and not a single complete one from Anyang.414 At Sanxingdui, however, forked blades appeared in large numbers in Bronze Age burials. Five were included in K1, 13 plus three miniatures in K2 (Figs. 7.10d–k).415 An unknown number were unearthed from the Yueliangwan pit in 1927, but seven have so far been published (Figs. 7.10b, c).416 Besides these finds from pits, a surface find was considered nothing more than a working hypothesis that is strictly stylistic and that does not necessarily bear on the actual date of the individual examples. 414For the Erligang find see Zhao Xinlai 1966 and Chen & Fang 1993, pl. 20. At Anyang, the tomb of Fu Hao contained the upper half of a broken blade. It has an inwardcurving end with sharpened edge, clearly that of a forked blade (see Beijing 1984a, pl. 84.4). In regions outside the Central Plain, a similar disappearance of forked blades may be observed, e.g. in the tomb at Jiangxi Xin’gan Dayangzhou, which contained no example. However, a few sporadic finds have been made deep in southern China as well as in northern Vietnam. Their date and nature are topics of ongoing debate. 415In addition, a picture of a forked blade with elaborate scrollwork along the long sides is incised on a hybrid of forked blade and ge, K1:235-5 (Fig. 7.13c). No actual forked blade has been found to have similar decoration, either at Sanxingdui or anywhere else. This image perhaps represents a fanciful elaboration of a real blade. However, the model must have been a Sanxingdui one because the style of the flanged zone in the image is unmistakably local (see the following discussion of the forked blades from K1 and K2, and compare the depicted blade with those in Figures 7.10h, j). 416In the strictest sense, only two, including the blade in Figure 7.10b, are known for sure to have come from the 1927 Yueliangwan pit, as they are published in Graham 1934. They are now in the collection of the Sichuan University Museum and published in color plates in Gao & Xing 1988, pls. 277–8. Five other forked blades, in the collection of the Sichuan Provincial Museum (three) and the Palace Museum, Beijing (two), are close parallels. There is little doubt that they were unearthed at Yueliangwan as well. The Sichuan Provincial Museums ones, including the one in Figure 7.10c, are published in color plates in Gao & Xing 1988, pls. 279–81; the Palace Museum ones in Zhou Nanquan 1995, nos. 43–4 (no. 43 was earlier published in Chen De’an 1994, p. 99 pl. 13-1). In addition to these blades, Gao & Xing 1995 and Gao & Xing 1998 publish two more forked blades in the collection of the Sichuan University Museum (Gao & Xing 1995, p. 71 pl. 10, p. 72 pl. 13; Gao & Xing 1998, p. 26 fig. 40.1, color pl. 285), but to judge by their style, material, and workmanship, they are forgeries. I reached this conclusion through examination on several occasions, the latest in November 2003, and through 200 made at Cangbaobao in 1984 (Fig. 7.10a), and four more miniature forked blades were found in 1988 at a spot nearby the Yazihe River outside the west wall of the Sanxingdui city.417 The surface find at Cangbaobao is stylistically probably the earliest among all the known pieces. The form of its notched crosspiece, consisting of two rectangular projections with densely serrated edges interrupted in the median point by Y-shaped slits, is similar to that of the forked blade from Shimao in Figure 7.9c and has not yet expanded into a flanged zone like that of the Erlitou examples. Also comparable are the lightly incised horizontal and crisscross patterns, which are similar to those on a trapezoidal blade from Erlitou (Fig. 7.6b). The sharpened end of the Cangbaobao blade is remarkable for its steep angle and barely curved edge, but the overall shape is generally comparable with that of forked blades from Shimao and Shandong (Fig. 7.9a–d).418 It is possible that this blade dates no later than the Erlitou period. No other forked blade so far known at Sanxingdui compares closely to this one.419 consultation with the Sichuan University Museum staff; the research is now published in Xu Jie 2006. 417The four miniatures are reported in Chen De’an 1994. For color illustrations see Tokyo 1998, p. 178 no. 157. The Jinsha site has now yielded about 200 forked blades as reported in Beijing 2006b (pp. 20–1) and fourteen have been published (pp. 62–75); their stylistic range is covered by the finds at Sanxingdui. 418The barely curved end of the Cangbaobao blade is similar to that of the Erligang forked blade in Figure 7.9i, but the flanged zone of the Erligang blade is much more elaborate. 419Jenny So thinks that the blade was likely manufactured locally because the material is typical of Sanxingdui (Bagley 2001, p. 161). Based on my experience with lithic artifacts at Sanxingdui, it is sometimes hard to comment on the material with certainty. 201 The forked blades from the 1927 Yueliangwan pit point instead to close relationship with Erlitou (Figs. 7.10b, c).420 The blade in Figure 7.10b retains a relatively broad proportion and a gentle curve at the sharpened end. The flanged zone is well developed, marked by a pair of simple short protrusions and a pair of elaborately shaped projections with notches cut into the profile. The area in between has spiked sides, each side bearing two groups of two spikes. This blade compares closely with the Erlitou blade in Figure 7.9e in shape and proportions, particularly in the structure of the flanged zone. The difference from Figure 7.9e lies mainly in the shape of the flanges at the tang; those on the Yueliangwan blade appear like animal-heads and thus more similar to those on another Erlitou blade, the one in Figure 7.9h. Also similar to Figure 7.9h is the use of horizontal grooves aligned with the spikes in the flanged zone. Another Yueliangwan blade, the one in Figure 7.10c, demonstrates more marked differences from the Erlitou prototypes. Proportionally this blade is exceptionally slender and long. Its sharpened end is deeply curved, its flanged zone oblique. The slanting form of the flanged zone appears already in the blade in Figure 7.10b though it is not as noticeable there. It becomes pronounced in Figure 7.10c and other forked blades from Yueliangwan. These were undoubtedly a local type derived from Erlitou prototypes. In contrast to the Yueliangwan blades, the forked blades from K1 and K2 are much further removed from Erlitou prototypes (Figs. 7.10d–k). A few appear to be variations of Erlitou examples with obvious differences (compare Figures 7.10f, i with 7.9f, g): the blade in Figure 7.10f has an exceptionally narrow and long tang; the deep 420The Erlitou connection manifested by the Yueliangwan blades was first suggested by Bagley in Fong 1980, p. 76. Since the excavation of K1 in 1986, scholars have often noted the connection, which is now supported by more blades from the two pits. 202 cleft end of the blade in Figure 7.10i is no longer a continuous curve but a curved triangle.421 Other forked blades from the two pits are even more distinctly local in style, finding few comparisons among forked blades in other regions. Most of them have deeply cleft ends (Figs. 7.10g–k), and the hole in the tang is often moved to the flanged zone. The flanged zones may be divided into two classes, one bordered with evenly arranged teeth or notches but without projecting flanges bracketing the zone (Figs. 7.10g, h, j), the other featuring confronting scrolls at both ends in addition to the even teeth (Fig. 7.10k). Of the two types, the second was probably developed later, for it appeared only in K2 while the first occurs in both pits. Both types seem to represent standardization of forked blades in the local production. On another forked blade, the flanged zone has clearly been simplified into a blank area that slightly bulges out (Fig. 7.10d). This kind of simpler, sometimes cursory, form should not be considered contemporary with the Neolithic ones, which lack the expanded zone between blade and tang. Comparing forked blades from K1 and K2, we may notice once again a difference between the two pits that has already been observed in the discussion of their bronzes: K1 had fewer pieces, but all are different from one another (Figs. 7.10d–h), whereas K2 had many more, but they may be grouped into three subtypes according to the treatment of the flanged zone (Figs. 7.10i–k). This difference gives further support to the relative dating of the two pits. Forked blades are the only lithic type at Sanxingdui whose use is illuminated by other evidence. The miniature bronze figure in Figure 5.24 demonstrates one way of manipulation: the figure holds out, with some formality, a forked blade by grasping the 421This type of deep cleft end appeared also among Shimao blades (see e.g. Fig. 7.9j). 203 tang with both hands and holding the blade upright. Another usage is shown in the incised design on the jade implement in Figure 5.42: forked blades are depicted standing on the tang on the ground.422 However, while it is generally agreed that the blades were involved in ritual sacrifices, their particular symbolism or the purpose of the sacrifice in which they were consumed remains an open question. 7.1.1.6 Ge blades, and hybrids of forked blade and ge Unlike the lithic types so far discussed, ge blades made of jade and stone seem to have been based on the bronze ge, a functional weapon that first appeared during the Erlitou period and remained one of the most enduring combat weapons throughout China’s Bronze Age (Figs. 7.11, 7.12).423 In actual use, a bronze ge would be mounted perpendicular to a haft. The hole in the tang might help secure the blade or might hold a decorative tassel. A mounted ge could thus be wielded like a battle-axe to hack at the 422Moreover, as mentioned in note 408, the pallellogram from Jinsha in Figure 7.7c show has a design showing a kneeling person carrying on the shoulder what looks like a curved elephant tusk but its end tantalizingly resembles that of a forked blade. There is a slight chance that the person might carry a forked blade mounted on something else. Outside the Chengdu Plain, no evidence on the use of forked blades has been found anywhere where they are unearthed. The Sanxingdui evidence may well apply only to the ways this mysterious shape was used in the Chengdu Plain, since Sanxingdui rituals were surely different from those practiced in other regions. 423For Erlitou ge blades, see Kaogu 1976.4, p. 260 fig. 3, pl. 5.5 (bronze), p. 262 fig. 6.6, pl. 6.1 (jade); Bagley 1999, figs. 3.12a, 3.14a. Ge also remained a favorite jade shape in the Central Plain in the Erligang and Anyang periods; the tomb of Fu Hao contained 39 pieces. For a lucid description of the relationship between the lithic and bronze ge, and the stylistic evolution of the lithic ge in the Central Plain, see Wilson 1990. However, some scholars argue that the lithic ge already appeared in the late Neolithic period (see e.g. Yang Meili 1998, p. 156). The evidence cited by Yang is three ge-shaped blades from Shaanxi Shenmu Shimao, one of which appears quite similar to the Erlitou one illustrated in Chen & Fang 1993, pl. 9. As the Shimao jades cover a chronological range as seen clearly in the case of forked blades, some of the objects may be as late as the Erlitou times. It is not clear if the Shimao examples actually date earlier than the Erlitou one. At any rate, the ge was a lithic type popular only in the Bronze Age. 204 adversary, but unlike a battle-axe it had two cutting edges tapering to a point. The extended horizontal cutting edges made it possible to use the weapon like a sickle in a sweeping motion; perhaps for this reason the blades sometimes have a slight downward curvature. At Sanxingdui, no bronze ge similar to those known from the Central Plain has been found,424 but lithic versions were included in K1 and K2: K1 contained 17 jade ge blades and about 26 stone ones; K2 had 21 jade ge and 10 stone ones.425 Most can be broadly classified into two subtypes. One features a long and slender blade proper (Figs. 7.11a, b). The section of the blade nearest the rectangular tang takes up about two thirds of the length, tapering slightly in the direction of the blade tip. The next section narrows more perceptibly, its demarcation from the first section being made particularly clear by the contour of the lower edge, which curves inward through much of the first section but then reverses the direction before it curves in again. The change of direction in the curvature forms a pointed tip separating the two sections. This second section then narrows to form a triangular head with a pointed tip that droops downward. Sometimes the triangular is rather pronounced and may be considered a separate third section. The 424As mentioned in note 192, a burial at Xinfan Shuiguanyin yielded three bronze ge similar to Erligang types, and another similar one has been found at Chengdu Jinsha (Fig. 8.2). 425The stone ge blades from both pits are all fragments beyond restoration and will not be addressed here. Not included in the numbers given here is another jade ge-shaped blade, K1:155-1. It is listed among the ge in Beijing 1999a but should instead be classified as a hybrid of ge and forked blade (Fig. 7.14b, see discussion below). Similarly K1:222, a stone blade classified as ge in Beijing 1999a, should also be viewed as a hybrid (see Beijing 1999a, p. 128 fig. 66.4, p. 131 pl. 43.1). On a separate note, ge blades co-existed with forked blades in K1 and K2, whereas the known inventory of the 1927 Yueliangwan pit includes no ge. This difference seems to support the possibility that the Yueliangwan pit was earlier in date than K1 and K2, as suggested in Section 2.3.2. The Jinsha site has now yielded more than 50 examples (including hybrids of forked blade and ge) (Beijing 2006b, pp. 19–20, 78–89). 205 hint of a median crest runs through the tang and the blade proper. All the blades of this type have plain surfaces except for two with shallowly incised parallel lines next to the tang (Fig. 7.11b).426 The other type of ge is much broader but shorter in proportion, and the triangular head is particularly prominent (Figs. 7.11c–e). The raised median ridge is pronounced; next to it are two broad bevels and then two narrow facets along the cutting edges. The facets end at about the point where the edges turn sharply inward toward the tip.427 Blades of this type are either plain or decorated in a zone adjacent to the tang with incised parallel lines from edge to edge, or short raised parallel ridges along the edges, or more complex patterns such as the rectangular panel filled with crisscrossed lines bracketed by short ridges seen in Figure 7.11d.428 Both types can be traced back to prototypes in the Central Plain and clearly descended from a common origin at Erlitou (Figs. 7.12a, b). The slender version may have had a more recent source at the Erligang times, as evinced by similar blades found 426This type of ge seems to appear only in jade: K1 had three among its 17 jade blades; K2 had 11 among its 21 jade blades. 42712 jade ge from K1 and 10 from K2 belong to this type. 428Eight blades of this type have plain surfaces: five in K1 and three in K2. As in the case of forked blades, a perforation is the norm on ge blades whatever the type. The perforation is mostly placed in the tang, but about half of the first type have it in the blade proper instead. Moreover, one blade of the first type, K2(3):248, has two holes, one placed in the tang, the other in the blade proper (see Beijing 1999a, p. 384 fig. 208.4, p. 387 pl. 150.1, and color pl. 103). The blade in Figure 6.10a also has two holes, both in the tang. In contrast, a blade of the second type, K2(3):322-2, has no perforation (see Beijing 1999a, p. 378 fig. 205.5, p. 381 pl. 147.1). The wandering perforations remind us that those blades were almost certainly never hafted; their design was not governed by functional constraints. 206 at Panlongcheng in Hubei, an Erligang outpost (Fig. 7.12c).429 The broader and more richly decoration ge were popular in the Anyang period (Figs. 7.12d, e). It is likely that the Central Plain styles were transmitted to Sanxingdui at different points of time, but the Sanxingdui blades also display distinctive local features, particularly in the case of the slender type, whose clearly articulated form appears almost unique to Sanxingdui with only two examples known outside the Sichuan Basin.430 Blades of the broad type are less distinctive in shape but the decoration adjacent to the tang seems to express a local identity in its use of patterns of earlier times such as the crisscross seen in Figure 7.11d.431 Two of the ge blades from K1 fall outside the two types discussed above (Figs. 7.11f, g). The blade in Figure 7.11f appears to be a miniature version of the first type but with its scalloped contours multiplied.432 The blade in Figure 7.11g may have been 429Jenny So in Bagley 2001, p. 167. The example used by So is not the same as illustrated here in Figure 7.12c, which is even closer a comparison, but this example was not yet published at the time of Bagley 2001. 430The tomb of Fu Hao contained two such ge blades that are comparable to the Sanxingdui ones (see Beijing 1984a, color pls. 17.1 [upper], 18.1 [upper]). 431The crisscross pattern in Figure 7.11d continues an earlier feature: the pattern is similar to that on the Erlitou jade trapezoidal knife in Figure 7.6b as well as on some of the Erligang period ge. For Erligang period ge blades with crisscross patterns see Chen & Fang 1993, pl. 18 (from Zhengzhou Baijiazhuang); Fong 1980, no. 10, and Beijing 2001b, p. 180 fig. 119.3 (from Hubei Huangpi Panlongcheng). The raised ridges along the edges are first seen on ge blades from Panlongcheng (Beijing 2001b, p. 222 fig. 159.9, p. 294 figs. 216.2, 216.9). Of the many ge blades from the tomb of Fu Hao, only one retains this feature (Fig. 7.12e) whereas it is common at Sanxingdui. The particular combination of crisscross pattern and short raised ridges seen in Figure 7.11d finds a close comparison in a ge fragment from Panlongcheng (Beijing 2001b, p. 435 fig. 319.1). 432The contours call to mind bronze ge with serrated edges (Fig. 6.10). However, the resemblance to the bronze ge might be accidental, for the bronze blade is much more emphatically scalloped as pointed out by So in Bagley 2001, p. 170. Perhaps more importantly, it takes the form of an isosceles triangle with a broad crosspiece in the shape of a rectangle. This feature is distinctly different from the jade scalloped blade discussed here. 207 similar to Figure 7.11f in its original condition, its contours comparing closely. This blade now has a flat tip sharpened to a cutting edge, an oddity that may have resulted from re-working a piece that had lost its pointed tip. These two blades are peculiar to Sanxingdui, and they remind us of the remarkable typological variety of the artifacts from K1.433 A conspicuous peculiarity in the lithic inventory of Sanxingdui is the blade type seen in Figures 7.13a–c, which has appeared only in K1 at the Sanxingdui site: a total of 35 examples.434 As observed by Jenny So, this type of blade is evidently a hybrid of forked blade and ge, marrying the notched flanged zone of the forked blade, its cleft tip, and its unfaceted blade with the asymmetrically tapering profile of the ge.435 The flanged zones on these hybrids are remarkably consistent. The upper flanges are notched once, while the lower flanges have a broad face marked with multiple notches, similar to that on the forked blade in Figure 7.10e (some similar to Figure 7.10f). Between the two sets of flanges are always four teeth, some evenly spaced (Fig. 7.13a), others forming two distinct sets (Fig. 7.13b). Shallowly incised parallel lines aligned with the notches run across the blade from edge to edge on many of the hybrids, including the three illustrated here, whereas a small number of them are left plain. Variation among these hybrids is most clearly expressed in the form of the Vshaped mouth: nearly a third of the blades have exceedingly minute cleft tips (Fig. 7.13a); more have large ones (Fig. 7.13b); and one blade has a bird perched in the gaping mouth 433Again, this peculiar type finds presence at the Jinsha site, where one example has been published (Fig. 7.12g). 434The only other site outside Sanxingdui that has yielded similar blades is Jinsha; see Beijing 2002c (pp. 148–51), Beijing 2006b (pp. 70–1, 82–6). 435Bagley 2001, p. 156. 208 (Fig. 7.13c).436 The differences may be viewed as contemporaneous variations if we take stock of the constant form of the flanged zone. However, they may also have a chronological significance; this latter possibility is suggested by several unusual blades from K1 (Fig. 7.14). The blade in Figure 7.15a is a typical ge except for its faintly cleft tip, which compares closely with the mouth of the hybrid in Figure 7.14a.437 The one in Figure 7.14b, on the other hand, has an area near the tang converted into a flanged zone typical of the hybrids.438 These blades may very well have been the starting point for the development of the hybrid, and a progression from the sub-types represented by Figure 7.13a through 7.13b to 7.13c is easily comprehensible. Moreover, the stylistic evolution must have already been completed by the time the Sanxingdui forked blades developed a distinctive form of flanged zone as seen in Figures 7.10g, h, j, for none of the hybrids bears this local feature. The hybrids may have stopped being produced by the time of K1. The function of the hybrid blade is unknown, as is the reason for their short life (though speculation abounds).439 The hybrid in Figure 7.13c does, however, remind us once more of the importance of birds in Sanxingdui ritual, a matter we will return to in Section 7.2. 7.1.1.7 Other blades and implements Various other lithic types were found in the pits at Sanxingdui and related sites, including several “weapons”—spearhead and sword-shaped implement—and many 436Four other blades, their mouths damaged, may have had birds too, as suggested by the remnants of the mouths (see Beijing 1999a, p. 81 figs. 41.2–5). 437Another blade (K1:353) has a similar cleft tip also (see Beijing 1999a, p. 91 fig. 46.4, p. 92 pl. 26.3). 438The other similarly altered blade (K1:222), made of stone, is referred to in note 425. 439For references to a sampling of such speculation see Bagley 2001, pp. 168–9. 209 “tools”—knife, hatchet, axe, adze, and chisel (Figs. 4.4a–c, 4.5a–b, 7.15, 7.16, 7.17a, 7.18a).440 Most of them find comparisons in other cultures. For examples, the spearheads and the majority of the axes or adzes have precedents both in the Baodun Culture and in Neolithic cultures in middle Changjiang region (Figs. 4.4, 4.5a–b, 7.15–16).441 The jade axe from Gaopian and the jade hatchet from K1 have notched edges similar to those on Erlitou jades (Fig. 4.5), which in turn have Neolithic prototypes in the Shandong Longshan Culture.442 The jade knife from K2 resembles a knife from the tomb of Fu Hao in shape though the decorative patterns on the two pieces are different (Fig. 7.17). On the other hand, the chisels seem to be a type developed during the Sanxingdui Culture, perhaps in Phase II, as examples were found in the 1927 Yueliangwan pit (Fig. 7.18a). The jade sword-shaped implement in Figure 7.19a, a unique item from K1, presents a shape typical of the bronze sword prevalent in the Sichuan Basin and neighboring regions in later times.443 Two bronze examples from the Shi’erqiao site are identified by Jiang Zhanghua as the earliest in the stylistic sequence of this weapon type and earlier than the jade example from K1.444 440Unlike other implements, the chisels may not have had a clear prototype among functional tools. 441For comparisons see Tokyo 2000 (p. 161 fig. 96, p. 163 fig. 97, p. 169 fig. 99, pp. 171–4 figs. 100–103), Beijing 1999d (pls. 109–111, 115, 147, 148). Particularly noteworthy is the common absence of perforation on these types. The claim in Yang Jianfang 1993, an influential study of Sanxingdui jades, that these types and some others are unique to the Sichuan region appears to be less than accurate. The Jinsha site has recently yielded one similar jade spearhead (Fig. 4.4c), and many axes, adzes and chisels totaling about 200 objects (Beijing 2002c, 126–40) and Beijing 2006b (pp. 24, 100–37). 442For a detailed study of the history of the notched edges see Hayashi Minao 1997, pp. 232–85. Similar notched edges can be observed on several Jinsha jades; see Beijing 2006b, pp. 90–1. 443This sword-shaped object is now joined by a second example from Jinsha, a fragment corresponding to the lower half of the K1 example (Fig. 7.19b). 444Jiang Zhanghua 1992, p. 82 and fig. 1. 210 Interestingly, the jade axes and chisels all have unworked rough ends in contrast to other finely shaped jades. The large quantities of these implements, particularly of the chisels (35 in K1 and 43 in K2), make clear that the rough ends were not accidental, and none of them shows traces of wear from use. 7.1.2 History and material of local production The preceding discussion of individual lithic types shows that the majority of the artifacts were locally produced. Besides finished artifacts, the lithic inventory at Sanxingdui includes raw material—large boulders of jade and stone, partly worked artifacts, and waste materials.445 These provide further evidence for the existence of a local lithic industry for ritual implements (Figs. 7.20–21). It is not clear, however, when production started.446 The cemetery at the Rensheng village, dating from the period of the Baodun Culture, yielded 22 small jade objects, which represent the earliest 445Burials containing such materials include M5 at the Rensheng village (one partly worked object; personal communication with Chen De’an in 1999), the 1964 Yueliangwan pit (about 300 objects including partly worked objects and raw materials; see Section 2.3.2), K1 (one piece of unworked jade; see Beijing 1999a, p. 117), K2 (two pieces of unworked jade; see Beijing 1999a, pp. 405–6), and the 1987 Cangbaobao pit (cores drilled in perforating stone bi; see Section 2.3.2). The several dozens so-called “whetstones” from the 1974 Suozitian pit and eight from K1 and K2 (two and six respectively) might be slightly worked raw materials (see Section 2.3.2; Beijing 1999a, pp. 117, 399–400). Besides these, large boulders of jade and stone, some with cut marks, were found in 1974 by the Yazihe River near the Suozitian pit (Xiao et al. 2001, p. 24). In 1984, the excavation at Xiquankan yielded a large number of stone bi including partly worked disks and waste materials (see Chen Xiandan 1988, p. 10; Chen Xiandan 1989a, p. 216). In 1986, the excavation at the Sanxingdui Locus III also yielded a large number of stone raw materials, some of which had been sliced or partly sliced (see Chen Xiandan 1988, p. 10). 446Production of functional stone tools of course went on since the beginning of the occupation of the site in Phase I. 211 appearance of jades in the Chengdu Plain on present information.447 However, their publication is of such limited detail that it is impossible to review the place of production and other related issues for those objects beyond the few sketchy description in Section 2.1.4. Lithic workshops for ritual artifacts must have come into existence by the time of Phase II, as evidenced by the large number of finished and partly worked stone bi as well as waste materials excavated in 1984 at Xiquankan near a house foundation in the stratum dating from that period.448 Fragments of similar stone disks were also excavated in 1980–81 at Sanxingdui Locus III from the second cultural stratum, corresponding to late Phase II, and from the stratum of Phase III.449 At Yueliangwan they were excavated from strata corresponding to Phases III and IV.450 More than one workshop probably operated at Sanxingdui. Production must have attained a large volume during Phases III and IV to judge by the remarkable quantity of finished artifacts so far unearthed, particularly the riches represented in K1 and K2. The minerals worked by Sanxingdui lapidaries are not known since only a few objects have been subjected to scientific mineralogical examination. A first study of 10 fragments of blades from K1 involving microscopic examination and chemical analysis established that all of them are nephrite, but they are of poor grade with relatively low 447Kaogu 2004.10. 448Chen Xiandan 1988, p. 10. Chen points out that the disks were excavated from the fourth cultural stratum at Xiquankan. According to the correlation in Table 1a, this stratum seems to cover the interval between the Report Scheme’s Phases I and II, as well as Phase II itself. 449Kaogu xuebao 1987.2, p. 230. 450Ma Jixian 1992, pp. 321–2. 212 structural density and hardness (on average 5.58 on the Mohs scale).451 A second microscopic examination of 18 artifacts from K1 and one from K2 (11 axes, one adze, one chisel, four ge, one trapezoidal implement, and one unidentified lump) revealed that only the adze, chisel, and trapezoid are nephrite while the others were made of softer materials—varieties of marble or limestone.452 Though far less than adequate, the mineralogical studies thus suggest that relatively soft stones constituted the bulk of the material for local lithic production. This must have conditioned the fabrication of artifacts in technical terms. For example, some of the largest forked blades from K2 are remarkably thin, about half a centimeter thick although measuring nearly 70 cm long (Fig. 7.10k). They testify to great skill in cutting and slicing the stones, probably with the help of metal tools.453 In excavation reports and studies, lithic artifacts have been sorted into the categories of jade and stone. As pointed out earlier, many artifact types at Sanxingdui appear in materials, but others appear in jade alone: jade types outnumber the types executed in stone. Moreover, the so-called jades are apparently from various sources, judging from their different textures and appearances. This perhaps suggests that the local lithic industry did not have consistent supplies of raw material and used whatever fine stones were available, while at the same time supplementing with coarser stones. 451Su Yongjiang 1996, Beijing 1999a (pp. 515–21). Of the ten fragments, actually only seven were examined under microscope for their petrographic structures; the other three were identified as nephrite by comparison with them. Chemical analysis of elemental composition was performed on two of the seven fragments. All ten fragments were tested for hardness and density. 452Beijing 1999a, pp. 500–14. 453Jenny So noted the probable use of metal tools in the Sanxingdui lithic production (see Bagley 2001, pp. 154–5). 213 As regards actual sources of raw material, Su Yongjiang, the mineralogist responsible for the first study, observes that the samples examined by him resemble jade materials from elsewhere in petrographic structure and elemental composition: jades from the tomb of Fu Hao that have been recognized as made of nephrite from Hetian in Xinjiang Autonomous Region; the nephrite mine at Longxi in Wenchuan county in the middle Minjiang River to the northwest of Guanghan; and the nephrite mine at Xiaomeiling in Liyang county, Jiangsu province in the lower Changjiang region.454 The authors of the second study also take note of the extensive distribution of rocks in the hills of Wenchuan and adjacent areas and regard them as a possible source for the lithic production at Sanxingdui.455 Given the large boulders that survive in the Sanxingdui site, supplies from nearby areas transported on waterways are a strong probability.456 Beyond that, little can be said about actual sources of raw material, and, as observed by So, there is no secure way to connect ancient lithic artifacts with mineral sources known today.457 7.2 Gold artifacts Gold artifacts have been found at Sanxingdui only in K1 and K2.458 Although fewer in number than bronzes and lithic artifacts, they sometimes appear on the most conspicuous part of an image (e.g. on the faces of bronze heads). Among the gold objects 454Su Yongjiang 1996, pp. 87–9. 455Beijing 1999a, p. 514. 456Curiously, the boulders have not been sampled and analyzed. Recently, mineralogical analysis of jades from Jinsha points to the same source in the Wenchuan Longxi (see Beijing 2002c, pp. 194–8). 457Bagley 2001, p. 159 note 29. 458Jinsha is the only other site in the Chengdu Plain that has yielded gold, about 200 items; nearly all are thin, small foils; see Beijing 2002c (pp. 16–26), Chengdu 2005 (pp. 10–1, 20–7). The finds at Sanxingdui and Jinsha constitute the largest groups of gold so far recovered from any Bronze Age site in China. 214 from K1 was an ingot weighing170.44 g, included in the sacrifice perhaps simply as a valuable.459 All the other pieces, judging from their appearance and condition, were part of composite objects that had components made of other materials. The clearest example is the four bronze heads from K2 with masks of gold foil (Figs. 5.6, 5.16). Three were found with their masks in place at the time of excavation while in the case of the fourth, the mask was detached but could be returned to its proper head.460 In addition, there are one fragmentary gold mask from K1 and two more from K2 that have not been matched with heads.461 Aside from the masks, the gold inventories of the two pits have little overlap. K1 yielded merely four items, including the gold mask, yet it also included one of the most celebrated pieces at Sanxingdui, a unique decorated tube of gold 1.42 m long weighing 463 g (Fig. 7.22). The tube must originally have served as a sheath, wrapping around a wooden staff, burnt traces of which still remain inside. It belonged no doubt to the regalia of a person of the highest status. The decorated area amounts to a third of the sheath’s length (Fig. 7.22b). The main motif, repeated four times, consists of an arrow whose 459K1:39 (see Beijing 1999a, p. 61 fig. 34.4, p. 62, pl. 15.4). The ingot is 11.9 cm long, 4.4 cm wide, and 0.2–0.5 cm thick. The stump of a pouring inlet remains visible on it. Louis proposes that the ingot was cast from gold dust extracted from the river sand (Louis 2003, p. 159). 460One of the three heads is illustrated here in Figure 5.16; for the other two see Bagley 2001, nos. 12, 13. The fourth head with detached mask is the one illustrated in Figure 5.6. The information about the condition of this head at the time of excavation is provided by Yang Xiaowu, a conservator who participated in the restoration of the objects from the two pits (personal communication, July 2000). 461For the gold mask from K1 see Beijing 1999a, p. 61 fig. 34.2, p. 62 pl. 15.2; Tokyo 1998, p. 148 no. 113. For the two masks from K2 see Beijing 1999a, p. 352 fig. 194, p. 355 pl. 135.2, and p. 355 pl. 135.3. 215 feathered shaft runs past a bird to strike the forehead of a fish.462 Below them is a narrow band containing a pair of smiling faces wearing earrings and pronged crowns or headdresses (Fig. 7.22c). The bird in the design resembles those on the monumental tree and several other bird images from K2 (Figs. 5.30, 5.31). An essentially identical design has recently been found at the site of Jinsha, decorating a gold band that might have been part of a headdress (Fig. 7.23). This iconography must have been important to the ideology of the ruling class at Sanxingdui, but we lack the information to decipher it at this point. The sheath was made by hammering an ingot into a rectangular sheet about 7.2 cm by 142 cm. As Louis observes, the sheet was then wrapped around the wooden staff before it was decorated with the images, for some of the designs run neatly across the seam, making it clear that they were applied as a final step.463 The designs, which have the appearance of thread relief, were executed by pushing the metal down on either side of the desired line so that the line itself remains level with the surface as a whole (Fig. 7.22b). This laborious technique is not known on gold objects from anywhere else in ancient China. The fourth gold piece from K1 is also unique among the gold artifacts from the two pits: a foil in the shape of an animal (Fig. 7.24). Rounded ears, lithe body, clawed feet, and coiled tail identify the creature as a feline, and the markings on the body and tail establish that a tiger was intended. In fact, the markings are rather close to those of a real tiger, certainly if measured against the stylized stripes seen on the tiger-shaped bronze 462The commonly accepted identification of the shaft that penetrates the fish’s head as an arrow is based on the feather-like part at the end of the shaft. The possibility exists, though not very likely in my opinion, that the shaft belonged to a decorated spear. 463Louis 2003, p. 163. 216 plaques in Figure 4.1, or on bronzes from Anyang and the middle Changjiang region.464 The piece was raised into relief by working from both the front and the back, the technique representing an early, simple form of the repoussé technique in the opinion of Louis.465 Originally the foil must have been wrapped around some sort of support, for its edge turns backward and the residue of an unidentified material remains on the back. The gold artifacts from K2 comprise the six masks for bronze human-like heads mentioned above and more than five dozen miniature foils of various shapes (Fig. 7.25). In addition, surviving fragments of miniature trees include a few wrapped in gold foil.466 The other foils must originally have been attached to objects as well, the holes seen on many of them apparently being meant for attachment. For once, K2 was not as rich as K1, for the total weight of its gold, at about 200 g, amounts to little more than the ingot from K1 and far less than the weight of the gold sheath. Altogether the two pits contained nearly 900 grams of gold, by far the largest amount from any one site in Bronze Age China. Elsewhere in China gold was little used. The existing gold artifacts from all finds can be largely grouped into two categories: foils for attachment to objects made of other materials, and self-contained items, mostly jewelry.467 Before the discovery at Sanxingdui, gold foils had been unearthed in both the Central Plain and the Northern Zone of China, whereas jewelry was found almost 464See for example the tiger on the zun in Figure 4.11. For typical Anyang tigers see Bagley 1987, figs. 148–50. For a northern Hunan version see Beijing 1997d, p. 134 (there incorrectly dated to the Warring States period). For a tiger from the Xin’gan tomb see Bagley 1993, fig. 44; Beijing 1997e, color pl. 38; and Yang 1999, no. 58. 465Louis 2003, p. 164. 466For examples see Beijing 1999a, p. 227. 467For a survey of Bronze Age gold artifacts see Gong Guoqiang 1997. See also relevant parts in Bunker 1993, Liang-Lee & Louis 1996, Huang Shengzhang 1996, and Qi Dongfang 2000. 217 exclusively in the Northern Zone.468 The Sanxingdui gold artifacts obviously fall into the first category, but their techniques, involving both hammering and cutting of the hammered gold (e.g. in the case of masks), were among the most sophisticated of the time. The origin of such techniques and the relationship with gold production in the Central Plain and the Northern Zone are by no means clear.469 Also to be answered are questions of elemental composition and source of supplies. For this latter issue, attention has been called to the rich gold resources in the Sichuan Basin.470 468The Northern Zone refers to areas largely in today’s Inner Mongolia, Liaoning, and northern parts of Shanxi, Shaanxi, and Hebei provinces. 469Scholars have commented on the absence of gold jewelry at Sanxingdui as a major distinction between the Chengdu Plain and the Northern Zone (Beijing 2002c, p. 17). Yet we need to take note of the difference in the burial context: K1 and K2 being sacrificial burials, and the burials in the Northern Zone being tombs. The possibility that gold jewelry was included in elite tombs at Sanxingdui cannot be ruled out at the present time. The same is true for the lithic artifacts: the absence of jewelry from K1 and K2 does not necessarily mean that jade ornaments were not included in tombs. 470Beijing 2002c, p. 18. The remark is specifically about the source of the gold for artifacts discovered at Jinsha, but it is naturally applicable to those at Sanxingdui as well. 218 Chapter 8. The Sanxingdui Site in the Context of Ancient China Few communities have ever existed in total isolation, the trajectory of their cultural development unaffected by interaction with other communities and cultures. Sanxingdui was not cut off from the rest of the world. Even though the contents of K1 and K2, probably produced at the height of the local civilization, are astonishingly original, leaving no doubt that they represent a material culture radically different from that of the Central Plain or anywhere else, traces of interaction are discernible. Some have been discussed in the previous chapters. Even the extraordinary imagery hints at inspiration from an external source.471 This chapter will now pool together the material mentioned earlier so as to summarize cultural contacts between Sanxingdui and places outside the Sichuan Basin.472 In some cases, external contacts can be described fairly precisely. In others the inter-regional relationship is hard to pinpoint, as objects of similar shapes appear in more than one region. Given that the Sanxingdui site may be archaeologically divided into two large developmental stages, Phase I (Baodun Culture) and Phases II to IV (Sanxingdui Culture), our discussion will be divided into two sections for clarity. But we naturally expect a certain measure of continuity, as some cultural influences might persist and routes of communication might remain open even if the people and communities involved in exchange changed over time. 471see Section 5.5. 472Since much length has been given in the previous chapters to discussing individual features and artifact types and their sources of origin and influence, only brief summaries will be offered here with references to pertinent sections of the previous chapters. 219 8.1 Cultural contacts in Phase I Cultural contacts in Phase I may be detected in two directions. The Baodun Culture is primarily defined by a set of distinctive pottery types that gives only limited hint of contact with Neolithic cultures outside the Sichuan Basin. Nevertheless, some traits in fabrication technique, vessel type and decorative style are found to be shared with pottery recently excavated at the fourth-millennium BC Yingpanshan site at Mao Xian county in the mountainous area along the northwestern edge of the Sichuan Basin. The nature of the commonality awaits for future fieldwork, however, as the large discrepancy in time between Baodun and Yingpanshan on present information makes tentative any claim of direction association. Closer in time but farther in distance, similarities with Baodun pottery have been observed in the late Neolithic site at Chenggu Baoshan in the Hanzhong Basin.473 It is significant that the Chengdu–Hanzhong connection can now be traced back to the Baodun times. Contact in another direction appears more tangible in archaeological record. Pottery vessels with tall ring feet decorated with fine openwork, including the one from M10 at the Rensheng cemetery, resemble those of the Shijiahe Culture of the third millennium downriver in the middle Changjiang region (Figs. 2.12a, 2.16). The three jade awls from Rensheng M5 find comparisons in Shijiahe also (Fig. 2.17) (Section 2.3.1),474 as do the axes or adzes excavated in the Baodun site.475 473See Section 2.1.6. 474More distantly, the awls are similar to those of the third-millennium Liangzhu Culture of the coastal area of the lower Changjiang (Fig. 2.17c). Contact with Liangzhu, if there was any, would presumably have gone through the middle Changjiang. As will be mentioned in the discussion below, several other lithic types at Sanxingdui trace their origin to Liangzhu, but find closer affinities elsewhere and closer in time. Recently, the presence of Liangzhu originals in the Chengdu Plain has been confirmed by a typical 220 Contact with the middle Changjiang was most conspicuously expressed in the walls of the Baodun settlements, which were probably built with know-how acquired from the Shijiahe Culture. Even though the Sanxingdui site was not walled at the time, the large walled settlement that arose in Phase II was undoubtedly based on the Baodun and Shijiahe precedents.476 8.2 Cultural contacts in Phases II–IV Walls erected during Phase II and the large space they enclosed are powerful testimony that the Sanxingdui site had changed into a metropolis. Concomitant with this development is large-scale production of lithic ritual implements and broader longdistance contacts manifested by lithic and pottery types. Later, during Phases III and IV, cultural contacts were further expressed in the medium of bronze. Those contacts direct our attention to three regions in particular: the Central Plain, the middle Changjiang region of Hubei and northern Hunan provinces, and the northern region comprising parts of modern Gansu and Shaanxi provinces. 8.2.1 Central Plain contacts Contact with the Central Plain may be viewed in three phases: Erlitou, Erligang, and Anyang periods, as the media or nature of the contact changed over time. In each of Liangzhu cong discovered at the Jinsha site (Fig. 7.5e), but it was evidently an heirloom piece handled down over ages and perhaps through several intermediaries; see discussion in Section 10.2. 475See Section 7.1.1.7. 476See Sections 2.1.6 and 2.3.2. 221 these periods, contacts probably went through the northern region or the middle Changjiang region, as will be discussed in Sections 8.2.2 and 8.2.3. The influence of the Erlitou Culture appears in pottery, lithic, and bronze types. The evidence in lithic types is extensive. Shapes related to, and possibly derived from, Erlitou include the circular collar with tall wall (Fig. 7.4a), the large trapezoidal knife (Fig. 7.6), and at least one subtype of forked blade (Figs. 7.9e–h, 7.10b–c). The crisscross pattern on the trapezoidal knife was also an Erlitou influence. The jade axe in Figure 4.5a and the jade hatchet in Figure 4.5b have notched edges similar to those on Erlitou jades (Fig. 4.5c). The Erlitou site has also yielded the oldest known examples of jade ge blade, which remained a favorite jade shape in the Central Plain in the Erligang and Anyang periods with changes in stylistic details. It seems that Sanxingdui absorbed successive innovations from outside while developing its own distinctive traits.477 Among the new pottery types at Sanxingdui, the tripod he finds the closest comparison at Erlitou, where it was a signature type (Fig. 8.1).478 Erlitou’s influence on Sanxingdui pottery is a complicated issue, however, involving interaction between the Central Plain and the middle Changjiang region before and during Erlitou times and the presence in both regions of shared types. Further discussion on this topic will be given in the following Section 8.2.2 on the middle Changjiang. The bronze plaques found at Sanxingdui are related to the inlaid bronze plaques at Erlitou (Fig. 4.2–3), where a bronze foundry capable of casting those artifacts with section molds had come into existence by 1500 BC.479 However, without assured 477Sections 7.1.1.2, 7.1.1.4–7. 478For a comprehensive study of the vessel type see Du Jinpeng 1992. 479See Section 4.1. 222 understanding of their date and place of production, it is premature to suggest that the necessary bronze casting technology was transmitted to the Chengdu Plain in Erlitou times. Presence of the Erligang bronze tradition in the Chengdu Plain was first suggested by three ge blades from a tomb at Xinfan Shuiguanyin, and a similar blade has recently been excavated at the Jinsha site (Fig. 8.2).480 The bronze vessels in K1 and the style-III ones in K2 were clearly products of the widespread Erligang Culture and its successors of the transition period (Figs. 4.6-10, 4.12-14). Possibly imports from elsewhere, they do not necessarily attest a local bronze industry.481 A local bronze industry apparently started in Phase III, and the clearest evidence of it comes, rather surprisingly, in the form of the extraordinary bronze images cast with a section-mold technique evidently acquired from the Central Plain. The particular versatility in the use of joining methods in fabricating the images owed its origin to the Erligang Culture where such methods were developed. Since a few Sanxingdui images have surface decoration apparently indebted to widespread transition-period styles, the introduction of the casting technology probably took place during the thirteenth century BC. In cultural terms, this means that the Sanxingdui bronze industry arose under the influence of the Erligang expansion.482 Erligang was responsible for certain lithic types as well. The ge blades at Sanxingdui seem to have particular affinity with Erligang examples (Figs. 7.11–12).483 480See note 192. 481Section 4.2.1. 482Section 6.2.3.1. 483Section 7.1.1.6. 223 Erligang might also be responsible for the widespread distribution of similar collared rings and disks at Sanxingdui, Anyang, and Xin’gan (Fig. 7.3).484 In the Anyang period, the relationship between Sanxingdui and the Central Plain must have changed. With the northward retreat of Erligang power around 1300 BC, distinctive regional cultures arose on the common Erligang foundation. In this sense Sanxingdui and Anyang were parallel developments. It is not clear whether the two centers engaged in direct contact, but shared traits are observable. For example, some of the Sanxingdui jade ge blades resemble Anyang examples (Figs. 7.12d, e), and the knife in Figure 7.17a is similar to an example from the tomb of Fu Hao (Fig. 7.17b). Interestingly, the type of long and slender ge blade distinctive at Sanxingdui (Figs. 7.11a, b) finds similar examples only in Fu Hao’s tomb, giving a rare suggestion that they may have been exports from Sanxingdui.485 Overall, the Anyang connection seems to have been rather limited, however, as no bronze vessels of the hallmark Anyang style have been found at Sanxingdui, while vessels of the same period produced in the middle Changjiang region were imported to Sanxingdui. 8.2.2 The middle Changjiang contacts Cultural contact with the middle Changjiang region persisted from Phase I through the rest of the occupational history of the Sanxingdui site. Distinctive pottery types that emerged at Sanxingdui such as tripod he, high-stem dou, and gu present a rather complicated situation (Fig. 2.12). Prototypes for all of them can be traced to both the middle Changjiang region and the Central Plain among late Neolithic cultures. The 484Section 7.1.1.2. 485Sections 7.1.1.6–7. 224 present archaeological record indicates that, towards the end of the Shijiahe Culture around the turn of the second millennium (corresponding to the transition from Phase I to Phase II at the Sanxingdui site), intensive interaction took place between the cultures of the Central Plain and the middle Changjiang region, resulting in an intermingling of cultural traits and in an Erlitou presence in the middle Changjiang.486 As already mentioned, the tripod he at Sanxingdui finds its closest parallels at the Erlitou site, while the high-stem dou appears to find particular affinity in the middle Changjiang (Figs. 8.1, 8.3). In addition, a type of zun-shaped vessel with flaring mouth and a bulging midsection sharply tapering off to a small base may have originated in middle Changjiang prototypes (Fig. 8.4). Moreover, miniature animal sculptures distinctive of the Shijiahe Culture probably inspired similar examples said to have been unearthed at Sanxingdui (Fig. 8.5).487 On present information, it seems plausible that new pottery types arrived at Sanxingdui from the middle Changjiang, perhaps as a mixture of Erlitou and middle Changjiang influences. Among the lithic types, spearheads and the majority of the axes or adzes of the Sanxingdui Culture carried on the Baodun/Shijiahe tradition (Figs. 4.4, 7.15, 7.16),488 and the Shijiahe jade carvings possibly provided iconographic inspiration for the extraordinary images at Sanxingdui.489 The middle Changjiang region was probably also one of the routes by which Erligang influences, as discussed in the previous section, reached the Sichuan Basin. The 486Asahara 1984 (p. 25), Meng Huaping 1997 (pp. 165–6), Zhang & Wei 2004 (pp. 250– 1). Among traces of an Erlitou presence is the trapezoidal knife decorated with crisscross pattern found at Hunan Shimen Weigang in Hunan in Figure 7.6c. 487This is the only such Sanxingdui miniature published so far. 488Section 7.1.1.7. 489Section 5.5. 225 Erligang Culture had at least one notable outpost in the region, a walled city at Panlongcheng in Hubei, and probably more.490 Given this presence, diffusion of Erligang influences upriver to Sichuan is easy to imagine. The distinctive local cultures that arose in the middle Changjiang region in the wake of the Erligang retreat gave the Sichuan Basin new neighbors there.491 Among those new neighbors, it was metallurgical centers in northern Hunan that probably supplied most of the bronze vessels found in K2 at Sanxingdui, and it is possible that the Sanxingdui users imported not only the vessels but also some of their contents such as jade collared rings and small beads and tubes.492 Evidently, throughout the time of the Sanxingdui site, contact with the middle Changjiang region was active, with successive waves of diffusion upstream. 8.2.3 The northern region contacts Contacts reaching northward into Shaanxi and Gansu are first suggested by shared traits with Yingpanshan pottery as mentioned earlier. During the time of the Sanxingdui Culture, they are expressed principally in lithic types, including sets of stone bi of graduated sizes whose typological features and fabrication peculiarities indicate a close tie with the Qijia Culture (ca. 2000 BC) in northwest China (Fig. 7.1).493 The squat cong tubes at Sanxingdui, plain or decorated with simple incised lines (Fig. 7.5), have been 490On Panlongcheng see Bagley 1999 (pp. 168–71), Bagley 1977. On the site of Shimen Zaoshi in northern Hunan, which testifies to Erligang influence south of the Changjiang, see Kaogu xuebao 1992.2, pp. 185–218. 491For these regional developments see Bagley 1999, pp. 171–5, 208–12. 492Section 4.2.2.4. 493Section 7.1.1.1. As mentioned in that section, the use of stone bi of graduated sizes that was characteristic of the Qijia Culture found its presence not only at Sanxingdui, but also at Anyang in Fu Hao’s tomb. 226 similarly associated with a lithic industry in the northern region that was active from the late Neolithic period, around 2000 BC, through at least the twelfth century BC.494 Among the forked blades at Sanxingdui, some find particularly close comparisons at Shenmu in northern Shaanxi province (Figs. 7.9c, 7.10a). Another northern connection, already established in Baodun times, as shown by the pottery from Chenggu Baoshan in the upper Hanshui region mentioned in Section 8.1, is now supplemented by an early second millennium site at Ankang, downriver along the Hanshui near the eastern end of the Hanzhong Basin, where some of the pottery is said to resemble Sanxingdui Period II pottery.495 The remains at the Baoshan site corresponding to Phases III and IV show similarities in building technology and pottery traits, but at the same time strong local characteristics.496 The bronze vessels of course are the strongest evidence of extensive connections between the Chengdu Plain and the Hanzhong Basin, as discussed in Section 4.2. The Hanzhong Basin deserves special mention as a crossroads where traffic from many regions met. Lying immediately north of the Sichuan Basin, across the Qinling Mountains, it was in contact not only with regions further north (as far as the steppes) and east (including the Central Plain) but also with the middle Changjiang region by way of the Hanshui, a major Changjiang tributary. Although so far no obvious traces of the Erlitou Culture have been observed in this region, repeated finds of bronzes in the neighborhood of Chenggu have yielded Erligang and transition-period types, bird and tiger images typical of the middle Changjiang region, and vessels from the middle 494Section 7.1.1.3. 495Wang & Sun 1992. 496Beijing 2002b, pp. 176–80. 227 Changjiang region. While the Changjiang was an obvious route of transmission of bronze vessels from the middle Changjiang to the Sichuan Basin, the Hanzhong Basin probably provided a simultaneous route.497 Perhaps even vessels from western Anhui in the Huaihe River valley reached Sichuan through the Hanzhong Basin. Sanxingdui’s contacts with any of these regions could have gone by way of the Hanzhong Basin as well as by more direct routes along the Changjiang.498 8.3 Large network of exchange, limited participation The preceding discussions have identified, with varying degrees of precision, cultural influences that Sanxingdui received from other regions outside the Sichuan Basin. They make clear that Sanxingdui was part of a large network of exchange, in which many artifacts circulated across long distances. Metal provenance studies have recently helped to reveal that during the early Bronze Age a trade in metals not only covered Sanxingdui and the aforementioned three regions but also reached far to the south: lead of an unusual isotopic composition that has been traced to mines in Yunnan province is found in many of the Sanxingdui bronzes (including heads, masks, trees, and the life-sized figure), as well as in bronzes from the Hanzhong Basin, from Fu Hao’s tomb at Anyang, and from the Xin’gan tomb.499 497Recent fieldwork has revealed shared traits in pottery between the Baoshan site and the sites in the Three Gorges area and western Hubei, indicating that the upper Hanshui and the middle Changjiang were culturally connected. This connection probably involved other smaller rivers such as the Daninghe River, a small tributary of the Changjiang in the Three Gorges area that provided another linkage with the Hanshui drainage; see Beijing 2002b, pp. 179–83. 498Section 4.2.2.4. 499See Appendix. 228 However, as extensive as its external contacts appear to be, Sanxingdui’s participation in the exchange was selective or limited. There was never wholesale introduction of a cultural assemblage. The Erligang expansion did bring the bronze technology to Sanxingdui, but its influence was limited in other areas. When bronze vessels from the middle Changjiang region were imported to Sanxingdui, not all the vessel types existing in that region were included, and the large bells for which the region was known were either unavailable or not needed. Likewise, Sanxingdui’s own cultural traits seldom found expression or external influence outside the Sichuan Basin. So far, traces of a Sanxingdui influence have been found only in the Three Gorges area and in the northern region.500 All these, along with the extraordinary bronze images, manifest Sanxingdui’s distinctive and unusual position in ancient China. One factor responsible for this distinctive and unusual position was surely the unique ritual practices at Sanxingdui 500Sanxingdui’s influence in the Three Gorges area appear primarily in pottery, where vessel types similar to those distinctive of Sanxingdui such as the ladle in the shape of a bird with a hooked beak were found; see Wang & Zhang 1999, Sun Hua 2000 (p. 129), Jiang Zhanghua 1998a (p. 5), Jiang et al. 2001b (p. 56), Jiang Zhanghua 2005 (p. 19). In the northern region, Sanxingdui influence can be observed in pottery and bronzes. Pottery resembling Phase II types at Sanxingdui was found at Ankang in the Hanzhong Basin as mentioned earlier. In later times, the bronze forked blades found in a hoard at Yang Xian Fanba in Hangzhong (Zhao Congcang 2006 p. 171, pls. 261–271) might represent an influence of the Sanxingdui Culture. A little further to the north, at Baoji by the Weihe River valley in Shaanxi, the early Western Zhou cemetery of the Yu state has yielded two small bronze figures that look like miniature versions of the statue from Sanxingdui K2 (Bagley 1990a, figs. 33–4; Lu & Hu 1988, vol. 2, color pl. 23). In addition, pottery cups with pointed bottom related to a Sanxingdui prototype were found, and interestingly, they were also made in bronze, presenting another case of bronze copying artifacts that originated in another material (the cups are published in various parts of Lu & Hu 1988). Dating from around 1000 BC, those objects suggest an extensive Sanxingdui legacy in the Yu state. Eastward along the Weihe River, at Xi’an, a fragmentary tiger-shaped foil resembling the two tiger plaques at Sanxingdui was unearthed from a tomb at Laoniupo, but it does not have turquoise inlay (see Liu Shi’e 2002, p. 297, p. 298 fig. 258, pl. 155.1). Falkenhausen considers the Laoniupo foil as possibly inspired by the Sanxingdui tigers (Falkenhausen 2003, p. 212). 229 as embodied by the pits, particularly K1 and K2, and their contents. Another factor, as we will turn to in the next section, could be the particular geographical situation of the Sichuan Basin, which I propose to define as “qualified isolation.” These two factors must have been part of a set of conditions that set Sanxingdui apart from the rest of ancient China and determined its limited participation in the network of exchange. 8.4 Qualified isolation: geography and the formation of the Sanxingdui Culture As briefly mentioned at the beginning of Chapter 2, the Sichuan Basin is a land sheltered by topography (Fig. 1.1). Land barriers surround it on all sides: mountains or high plateaux mark the basin off from the outside world. In the west the Tibetan Plateau rises more than 3000 m above sea level; the only avenues of westward communication are river valleys that dissect the plateau. In other directions the highlands are 1000 to 3000 m in elevation. In the south is the Yunnan-Guizhou Plateau, which extends in an arc toward the northeast to embrace the basin. In the north the Qinling Mountains run from the Tibetan Plateau in the west to the Yunnan-Guizhou Plateau in the east. These highlands and mountains were daunting obstacles to human traffic; the difficulty of reaching the basin was proverbial. Describing the journey to Sichuan from the north, the eighth century poet Li Bo (AD 701–62) famously said, “The road to Shu is harder than the road ascending to the blue sky.” Another popular saying labeled the Sichuan Basin “the land in heaven,” a description that evokes its combination of abundance and inaccessibility. Yet despite this forbidding topography, mountain paths and waterways offered routes of communication with the outside world (Fig. 1.2). One such route was the 230 Changjiang, which runs south across the Tibetan Plateau to enter the Sichuan Basin at its southwest corner. Once inside the basin, the river flows east along the edges of the Yunnan-Guizhou Plateau to exit through the Three Gorges, a series of narrow passages between precipitous cliffs, where navigation is made perilous by rapid currents and submerged reefs. The Changjiang gave access both to the western high plateau and to regions downriver to the east. In the south, the Changjiang and mountain paths led into the Yunnan-Guizhou Plateau and beyond, as far as Southeast Asia and India. Tributaries of the Changjiang, running southward to it from the north of Sichuan, enabled its traffic to spread throughout the basin and invited communication with regions to the north. The Minjiang River led northwest to the Tibetan Plateau. The Jialingjiang River, which originates in the Qinling Mountains, connects the Sichuan Basin with the Hanzhong Basin in the upper Hanshui region and with the Weihe River valley farther north. The Qin armies that conquered the Sichuan Basin in 316 BC and eventually brought it into the orbit of a unified China came from this direction; later Li Bo traveled at least part of the route and captured its difficulty in his poem. The Hanshui River runs southeast to join the Changjiang in the Jianghan Plain of southern Hubei. As mentioned in the previous section, the Sichuan Basin could communicate with southern Hubei and regions further south or east either along the Changjiang or by way of the Hanshui River. Moreover, Sichuan is also connected with southern Hubei by the Qingjiang, a Changjiang tributary that originates on the Sichuan-Hubei border and runs parallel to the Changjiang (south of it) until meeting it at Yidu, east of the Three Gorges. Tributaries of the Changjiang in the Three Gorges area such as the Daninghe River provide further linkage between the Changjiang and the Hanshui. 231 Thus while mountains isolated the Sichuan Basin, limiting communication to a trickle, rivers brought those trickles from all directions. This condition, which I proposed to call “qualified isolation,” shaped the course of Sichuan’s cultural development. The rise of a distinctive bronze-using civilization in the Chengdu Plain in the second half of the second millennium BC may well have depended crucially on the combination of geographical isolation and far-flung contacts. The plain was isolated enough to escape being overwhelmed by any one outside culture, but open enough to be stimulated by outside contacts. 232 Chapter 9. Conclusions The archaeological record presently at our disposal shows that the occupational history at the Sanxingdui site spans nearly two thousand years, covering both the Neolithic period and the early Bronze Age. In the days of Phase I in the third millennium BC, the site was a modest settlement of the Neolithic Baodun Culture, but it rose to preeminence during Phase II in the first half of the second millennium when a walled settlement enclosing an area of 3.5 sq km (350 hectares) came into existence, one of the largest in East Asia. A new epoch then opened in the Chengdu Plain, as the landscape came to be dominated by a metropolis overshadowing by far the other settlements. In the second half of the second millennium BC during Phases III and IV, the Sanxingdui site unequivocally entered the Bronze Age, with its elite material culture dramatically revealed to us in the bronze images buried in the two sacrificial pits (K1 and K2), their peculiar iconography and casting technology attesting an artistic and cultural tradition like none other in ancient China. Representational sculpture in bronze sets Sanxingdui most conspicuously apart from the elaborately ornamented ritual vessels that define bronze cultures elsewhere, marking the civilization on the Chengdu Plain as radically different. Around 1000 BC, however, Sanxingdui was abandoned, for reasons that remain to be ascertained.501 501Flooding remains a possible cause for the abandonment of the Sanxingdui settlement, as mentioned in Section 2.3.2. The cultural strata at the Shi’erqiao site in Chengdu whose early stages overlap with the end of the Sanxingdui settlement often include layers of river pebbles and alluvial silt and sand (Wang Yi 1988), and some houses at the Shi’erqiao site show signs of destruction by flooding (Wenwu 1987.12). At the Fangchijie site in Chengdu, strata as late as Warring States have yielded actual flood-control apparatus, in the form of dikes made by filling bamboo cages with pebbles (Wang et al. 233 Focusing on internal evidence and inter-regional comparison, the present dissertation has reviewed the Sanxingdui site’s various components, and discussed at length its elite material culture as embodied in the contents of the two sacrificial pits, particularly its bronzes. Several conclusions were reached in the course of this exercise. In Chapter 2, the temporal span of the Sanxingdui Culture as an archaeological construct was re-defined to begin with Phase II and end with Shi’erqiao’s twelfth stratum. The Sanxingdui Culture so delimited is not defined only by a core pottery assemblage but also takes into account the artifacts from the two pits and the city walls. Moreover, this redefinition expresses a hierarchical relationship in the settlement pattern of the Chengdu Plain, with the Sanxingdui site as the primary center. In Chapter 3, the two pits were re-affirmed as sacrificial in nature, representing two performances of the same type of ritual action across a time span of perhaps a few decades, during which significant changes took place in the material expression of the ritual iconography. This understanding was arrived at by comparing the contents of the two pits and their burial context, and by a detailed review of suggestions advanced by other scholars. The next three chapters focus exclusively on the bronzes, which constitute the principal evidence for the extraordinary character of the Sanxingdui elite material culture. Chapter 4 was devoted largely to the vessels, as they first of all serve as the primary dating criteria for the two pits and also demonstrate cultural contacts that Sanxingdui had 1999, p. 5; Xu & Wang 1999). Analysis of pollen samples taken at the Zhihuijie site in Chengdu indicates that the area was dotted with lakes and swamps (Nanfang minzu kaogu 1987). The disconnected pattern of the Shi’erqiao sites known today (see Section 10.1) might thus reflect an ancient settlement pattern that built houses on patches of higher ground. 234 with other regions. Based on this evidence, the two pits were dated just before and just after 1200 BC. The vessels themselves were identified as imports from the middle Changjiang region through a detailed examination from multiple perspectives including style, casting technology, and alterations inflicted on them. The change in vessel types between the two pits was construed as representing a standardization in ritual over the time. This interpretation is further reinforced by the stylistic differences between the bronze images from the two pits discussed in Chapter 5. In Chapter 5, the second chapter on Sanxingdui’s bronzes, the majority of the images were broadly divided into two categories in terms of iconography. The first consists of heads, figures, and masks, most of which share a striking physiognomy, while the second comprises trees, ornaments for trees, and creatures associated with trees, birds especially. The classification produces a mechanism by which to explore two levels of reconstruction: the original appearance of the individual images, and the original appearance of the ensembles that they composed. It can be reasonably imagined that all the Sanxingdui images were brightly colored in their original state. The heads and masks would have had bodies of another material, perhaps wood, or would have been mounted as architectural fittings. The individual human figures would have had their facial features painted, earrings dangling from the ears, their bodies dressed in real or depicted robes or partly bared and perhaps tattooed, and some further crowned with headdresses. The monumental tree would have had a fantastic bird-like creature perched on the top of the trunk in addition to the nine birds now standing on the three layers of branches, and the branches would be decked with detachable as well as permanently attached but movable pendants such as bells, rings and shells. Several possible ways in which some of 235 the images might have been arranged in ensembles or to form parts of actual architecture are conjured up. It is argued that perishable materials such as real trees were involved in the ritual sacrifices, and that wooden sculptures may have provided the prototypes that the bronzes copied or adapted. Lastly, the human images are identified as symbolically representing elite self-sacrifice for the purpose of placating or gaining favors from a high power. The results summarized here can best be described as sketchy and tentative, partly due to the limitation of the evidence at hand. In my opinion, however, this exercise is prerequisite to any further study of the ritualistic art and ritual system at Sanxingdui. The effort made in this dissertation is the first in this area. Chapter 6 presented the first ever comprehensive study of the bronze fabrication techniques at Sanxingdui. A detailed description of the techniques and comparison with Central Plain practice suggested that Sanxingdui acquired bronze fabrication technology from the Central Plain and that the local industry probably arose under the influence of the Erligang Culture. However, in contrast to the “one-pour mentality” evident in the Central Plain, foundry practice at Sanxingdui is characterized by a “simple-mold mentality”: a preference for simple molds, and for making complicated objects by joining components cast in those simple molds. To cast an object so wildly complicated as the monumental tree (Fig. 5.30), Sanxingdui founders dissected it into a multiplicity of components that, individually, were easy to cast. Complexity was achieved by relentless joining. This “simple-mold mentality” is explained by considering various aspects of the Sanxingdui bronze industry: the kind of artifacts to be cast, the technical challenges they presented, their aesthetic and visual characteristics, and possible economical factor such as metal supply. 236 Chapter 7 was a typological exercise in classifying the jade and stone artifacts from the two pits and other locales of the Sanxingdui site. All the lithic types had origins outside the Sichuan Basin, but they were often developed into distinctive local shapes. Stylistic sources, evolution, and functions of individual types were traced to the degrees possible. The classifications proposed are sometimes different from what are published in the formal excavation report of the two pits (Beijing 1999a), and thus offer a new frame of reference for future study. Compared to the bronze images, the lithic artifacts afford richer evidence for contacts between Sichuan and other regions, and these were discussed in detail. Gold appears more conspicuous at Sanxingdui than elsewhere in ancient China. Mostly foils, the gold artifacts were made by techniques that involved both hammering and cutting of the hammered gold and that ranked among the most sophisticated of the time. Chapter 8 summarized cultural contacts between Sanxingdui and external sources outside the Sichuan Basin, many of which have been observed and discussed in various sections of the preceding chapters. Three regions were recognized as particularly important sources of influence or as avenues by which influences originating elsewhere traveled: the Central Plain, the middle Changjiang region of Hubei and northern Hunan provinces, and the northern region comprising parts of modern Gansu and Shaanxi provinces. Those regions appear to have been principal nodes of a large network of communication across much of ancient China. Sanxingdui’s participation in this network was limited, however, and it was more recipient than donor. This phenomenon, whether a matter of choice or otherwise, may be partly explained by the unique ritual practice at Sanxingdui and by the particular geography of the Sichuan Basin, what I define as 237 “qualified isolation”: the land was isolated enough to escape being overwhelmed by any one outside culture, but open enough to be stimulated by outside contacts. It is hoped that the present dissertation has fulfilled to some extent the goal it set out to achieve: to sort out the basics of the Sanxingdui site and its elite material culture. However, throughout the dissertation, I have said time and again that much more fieldwork is needed, for the evidence at our disposal is very limited and specialized. At this point, we have little idea about the layout of the city and the dwellings of its inhabitants, elite or common. We know something about elite material culture, yet really cannot guess the construction and inventory of a major tomb. So large and rich a city is likely to have buried the dead of its ruling class ostentatiously. As to the bronze industry responsible for the extraordinary sculptural tradition, there are few traces of foundry and casting debris, and there has been little metallographic study and laboratory examination of fabrication techniques. For these and many other problems, we have to wait for future fieldwork and collaboration across disciplines. On the other hand, the mere existence of a dissertation on the Sanxingdui site attests how much our knowledge of the Chengdu Plain during the Neolithic and early Bronze Age has grown since the discovery of K1 and K2 twenty two years ago.502 No longer perceived as a cultural backwater, it has risen to the exalted status of the home of a major civilization. Our sense of caution in acknowledging the limited nature of the available evidence must be balanced with an undeterred optimism in exploiting that evidence. Sanxingdui remains a fertile ground for original research. For the immediate future, I hope to continue the effort in re-constructing the original appearance of the 502The present dissertation is of course preceded by Ge 1997. 238 Sanxingdui images and their ensembles and to continue research on the bronze fabrication technology. 239 Chapter 10. Epilogue: Expanding Horizon and Legacy of the Sanxingdui Culture At the time of the two sacrificial pits, Sanxingdui seems to have been the main urban center on the Chengdu Plain; the only other sites known with similar pottery and stone tools are unwalled and much smaller.503 This suggests a considerable change since the days of the Baodun Culture, when the landscape had been dotted with settlements of similarly moderate size rather than dominated by a single large one. Toward the end of Sanxingdui Phase III, the situation apparently began to change again, as a settlement at Chengdu about 40 km southwest of Sanxingdui began to grow. This would seem eventually to have become the regional urban center succeeding Sanxingdui (Fig. 2.1).504 This epilogue will provide a brief survey of the Chengdu settlement so as to attain a measure of understanding of Sanxingdui’s legacy. 10.1 The Shi’erqiao and related sites Unlike the Sanxingdui site, the ancient settlement in Chengdu lies beneath the modern city, making systematic exploration impossible. The settlement as known presently comprises a dozen separate sites spread over about 10 km in the west and south of modern Chengdu, basically along the old course of the Pijiang River and its branches (Fig. 10.1).505 The largest of them, also the most recently known, is Jinsha, discovered in 2001 (Fig. 10.2). As in Sanxingdui, it is mainly the elite material culture at Jinsha, mostly 503Section 2.1.6. 504In later times Chengdu would become the dominant power in the plain and the capital of the state of Shu. 505Zhu et al. 2003, p. 127. 240 luxury goods, that underscores the site’s probable high status in a political hierarchy. The Jinsha site will be discussed in the next section. Prior to the discovery of Jinsha, the Chengdu settlement was represented principally by the type site at Shi’erqiao on the western side of the modern city, the site with the deepest cultural strata and the richest remains (Fig. 10.2). In its early stages Shi’erqiao appears to be an offshoot from Sanxingdui: its stone tools are very similar, and its pottery repertoire, in which vessels with pointed bottoms are prominent, resembles that of Sanxingdui Phases III (late) and IV (Figs. 2.12c–d, 2.13, 2.20). In late stages, from around 1000 BC on, the Shi’erqiao site developed distinctive features. Similar remains tie the other Chengdu sites to different stages of Shi’erqiao.506 An earthen mound explored decades ago at Yangzishan in the northern part of Chengdu about 8 km northeast of Jinsha might be a Shi’erqiao construction of about 1000 BC.507 The mound was square and terraced, with stairs leading from one terrace to the next (Fig. 10.3). It was constructed by building a wall around a square area and filling the enclosure with pounded earth, then building a larger and lower wall around the first and filling the new enclosure with earth, and finally repeating the operation one last time. The walls were of adobe brick, recalling the brickwork found atop the east wall at Sanxingdui, 506For Shi’erqiao and related sites in Chengdu and their material culture see Wenwu 1987.12, Sun Hua 1996, Jiang Zhanghua 1998b, Wang et al. 1999 (pp. 4–5), and Jiang et al. 2002 (pp. 11–6). Archaeologists in Sichuan commonly call the early stages at the Shi’erqiao site “Shi’erqiao Culture,” a terminology that is problematic in my view. See Section 2.4 for discussion on the relationship between Shi’erqiao and Sanxingdui in pottery typology and the problems with the definition of the “Shi’erqiao Culture”. 507The excavation report is Kaogu xuebao 1957.4. For discussion of the mound and its date see Lin Xiang 1988, Sun Hua 1993f, and Li Mingbin 2003. The mound no longer exists, having been dug away for clay to make bricks in the 1950s. 241 and the mound was oriented with its corners to the cardinal points, like the Sanxingdui pits. At the Shi’erqiao site, archaeologists have excavated a cluster of large and small houses.508 Wooden beams and planks, bamboo slats from walls, and straw from roofs were preserved well enough to shed much light on building techniques. In one case, long beams had been laid parallel on the ground and drilled with holes at regular intervals, evidently to receive wooden pillars. In some of the smaller buildings the floor was kept off the ground, perhaps for protection against damp or flooding, by driving wooden stakes into the ground, lashing a grid of poles to the stakes, and laying a plank floor on top of the grid (Fig. 10.4). Although the buildings at Shi’erqiao were contemporary with Sanxingdui Phases III and IV, they seem architecturally more sophisticated than anything so far known at the Sanxingdui site. Another difference from Sanxingdui is the turtle plastrons found at several Shi’erqiao sites. These have bored or chiselled hollows and were obviously used for divination. Divination plastrons reportedly are found at Chengdu all the way from the twelfth to the eighth century BC,509 but no example has yet been found at Sanxingdui. Although Shi’erqiao and related sites yielded remarkable finds, until 2001 the Chengdu settlement looked rather impoverished by comparison with Sanxingdui and its sacrificial pits. This picture started to change when the Jinsha site was discovered and a large number of precious materials such as bronzes and jades came to light. 508Wenwu 1987.12. 509Luo Erhu 1988, and Wang Yi 1988. Similar plastrons have been found at Jinsha, too; see below. The Chengdu plastrons are significantly larger than those commonly seen at Anyang. 242 10.2. The Jinsha site Like most major finds in China, the Jinsha site was discovered as a by-product of capital construction. In the afternoon of 8 February 2001, construction workers digging a ditch in the Jinsha village in the western suburb of Chengdu (Fig. 10.2), bulldozed out some fragments of elephant tusks and a few jade artifacts. Archaeologists of the Chengdu Municipal Institute of Archaeology quickly moved in and started working at the site the next day.510 It immediately became clear to them that the find was important, and they decided to conduct large-scale excavation, which has been continuing to this day and continuously expanding in scope.511 As of now, the Jinsha site is estimated to cover more than 5 sq km (nearly 2 sq mi),512 but the archaeologists have yet to determine its limits. Excavation has now been undertaken at more than 20 loci, with the total area excavated approaching 10 ha (100,000 sq m or 25 a) (Fig. 10.5).513 Like Shi’erqiao, the 510Fieldwork at the Jinsha site had actually begun earlier in 1995 when just across a small river from the Jinsha village an ancient site was discovered at Huangzhongcun. Occupation there is now known to have begun at the end of the Sanxingdui civilization and apparently attained a height in the Western Zhou period (Beijing 2002c, p.7). The Huangzhongcun site is now grouped as part of the Jinsha site. 511The preliminary excavation report of the 2001 discovery is Wenwu 2004.4. The volumes of Chengdu kaogu faxian (Archaeological discoveries in Chengdu), a collection of excavation reports published annually by the Chengdu Municipal Institute of Archaeology since 2001, have continuously provided latest results from the site (it is also an invaluable source on fieldwork at other sites carried out by the Institute in the Chengdu Plain, including Baodun sites). Zhu et al. (2003), written by some of the field archaeologists working at the Jinsha site, provides the latest survey of archaeological finds and the latest understanding from their perspective. The initial discovery in 2001 yielded the most remarkable objects so far known from the Jinsha site; a selection of the most outstanding ones are published in an exhibition catalogue (Beijing 2002c). Several other catalogues devoted to the Jinsha site have also appeared, including Chengdu 2005 and Beijing 2006b. Two more exhibition catalogues, Paris 2003 and Hong Kong 2007, include sections on Jinsha. The present survey is based on information published in all these sources as well as my own study at the site, particularly of the artifacts excavated. 512Chengdu 2005, p. 2. 513Ibid. 243 Jinsha site has thick cultural deposits, measuring 4 to 5 m (13 to 16 ft) deep. A total of sixteen cultural strata was exposed at the locus of the initial discovery in the central eastern sector of the site, now called Meiyuan. Stratigraphic evidence and typological comparisons in pottery, bronze and jade connect the site with Sanxingdui and with Shi’erqiao, and indicate that it was occupied continuously for a relatively long time, roughly from the second half of the second millennium through the beginning of the fifth century BC. The site was at its apogee from the mid eleventh through early ninth century, as cultural remains are the most abundant and widespread during this period, including most of the major features and precious materials. Remains datable to the earlier times, corresponding to the last phase of the Sanxingdui site, are found extensively as well. Earlier remains dating from Baodun times have been found only in one locale so far, at Jinshayuan south of Locus Meiyuan.514 In the ninth century, however, the Jinsha site apparently went into a sharp decline from which it did not recover, as suggested by greatly reduced remains. The remains at Jinsha comprise features, artifacts, and organic materials. The features include building foundations on the ground, pottery kilns, burial grounds, refuse pits, and pits containing precious materials. The large building foundations concentrate in the Huangzhongcun area in the northeast part of the site, the largest located at Locus Sanhe Huayuan measuring over 430 sq m (514 sq yd) (Figs. 10.5, 10.6). Smaller foundations are widespread, found at numerous loci. All had foundation trenches filled with clusters of small postholes and featured wattle-and-daub walls, as at Shi’erqiao. Associated with the building foundations were refuse pits and pottery kilns. Several 514Chengdu 2004. As mentioned in Section 2.16, the Baodun remains in Chengdu have also been found at Shijiefang south of the city and Huachengcun west of the city. 244 burial grounds, usually in areas of abandoned buildings, have been excavated, exposing altogether over 1,000 tombs, though almost all of them were poorly furnished, having at most a few pottery and lithic artifacts, and sometimes no grave goods at all. The precious materials that made the Jinsha site an archaeological sensation came largely from the initial find at Locus Meiyuan (Fig. 10.5). A small number have also been found in other loci in subsequent excavations. At present, some 5,000 luxury artifacts have been unearthed, including hundreds of objects made of bronze (1,200), gold (200), jade (2,000), stone (1,000), and animal bones, in addition to approximately 1,000 elephant tusks, thousands of boar tusks and deer antlers, and some 10,000 pottery vessels and shards. The luxury artifacts may be broadly divided into two categories: implements and three-dimensional images. As at Sanxingdui, the artifacts at Jinsha include a number of lithic implements widely shared across long distances, such as collared disk and ring (Fig. 7.3), ge blade (Figs. 7.12–13), and forked blade (Figs. 7.9–10), to name a few. The cong tubes present a particularly interesting case of connections over long distances and times and of local adaptation. Twenty four examples have so far been unearthed at Jinsha, comprising two large ones and 22 small ones with plain surface (Figs. 7.5e–f).515 As we know, the jade cong is characteristic of the Neolithic Liangzhu Culture, which flourished in the third millennium BC. A typical late Liangzhu example has a slender shape that tapers from top to bottom and is decorated with tiers of faces whose eyes are supplied by pairs of circles folded around the corners, as seen in Figure 7.5e. Although unearthed at Jinsha, this cong must be a Liangzhu original handed down over the ages and no doubt passing through the 515For a study of the cong tubes from Jinsha see Zhu & Wang 2004. 245 hands of many intermediaries. It surface decoration has become blurred from handling, yet the fine greenish nephrite sheen is not much altered, as it normally is in the Liangzhu burial context. Such a cong was probably treasured exotica at Jinsha and also by the intermediaries through whose hands it passed. After Liangzhu times, the shape lived on in various simplified versions in many regions over a span of two thousand years, the ones with plain surface being associated particularly with the lithic industry of the northwest region.516 The plain ones at Jinsha, like those at Sanxingdui, must have been part of this northwest tradition. However, the example in Figure 7.5f stands out for its fine workmanship and remarkable size and weight. Though unique, it was probably a local product: a similar material was used to fabricate many other jade objects, and mineralogical analysis suggests that the source of the material might be nearby in the hills to the northwest of the Chengdu Plain.517 The diversity in the type of cong seen at Jinsha reinforces the scenario of extensive cultural contacts with the outside regions that has already been recognized at Sanxingdui. Types shared exclusively with Sanxingdui stand out among the objects at Jinsha. Heavy consumption of elephant tusks as seen in several deposits immediately recalls the elephant tusks buried in the two sacrificial pits at Sanxingdui (Figs. 2.7, 3.4, 10.7). Lithic types unique to the two sites include bracelet-like rings (Fig. 7.4), bi of exceptionally large sizes (Fig. 10.8), parallelograms (Fig. 7.7), scalloped ge-blades (Figs. 7.11f–g, 7.12g), hybrids of ge and forked blade decorated with jagged or toothed notches above the tang (Fig. 7.13), sword-shaped implement (Fig. 7.19), and elongated chisels (Fig. 7.18). Local peculiarities appear also in surface decoration, as seen for example in the use 516Section 7.1.1.3. 517Beijing 2002c, p. 198. 246 of crisscross lines contained in rectangles on ge blades (Figs. 6.11d, 6.12f). The incised crisscross pattern dates back to the Erlitou times, while the ge blades were of the shape widely shared among regions during the Anyang period. Neither was thus unique to Jinsha, but the imposition of an earlier decorative pattern on a later form was peculiar. Bronze types shared between Jinsha and Sanxingdui include a type of flimsy triangular ge blade with serrated edges (Fig. 6.10), and the square collared disk and circular collared ring, the latter apparently imitating the lithic version (Fig. 5.39). The close relationship between Jinsha and Sanxingdui is most vividly seen in the three-dimensional images and related surface decoration. At Jinsha, a good number of stone images of animals and human figures have been excavated, exceeding the number at Sanxingdui, the types including crouching tiger (Fig. 10.9), coiled snake (Fig. 10.10), and kneeling human figure (Fig.10.11).518 Naked and probably male, the stone figure in Figure 10.11 kneels and leans forward with his buttocks resting on his heels. He sports a hairstyle that resembles an open book: the neatly combed hair represented by incised parallel lines on top of the head is parted in the middle and combed to the sides. In the back, a double-braid pigtail can be seen descending down the head, and his oversized hands are bound by a thick rope, all indicated by incising. The facial features are depicted by modeling and incising. The man 518The Sanxingdui tiger illustrated in Figure 10.9a is so far the only one known to me from there, while ten have been excavated at Jinsha (Chengdu 2005, p. 18), all sharing the same appearance with minor variations in detail and ranging in size from 15–22 cm in height and 18–28 cm in length. Two of the tigers, including the present one, are published in Beijing 2002c (nos. 55–56) and in Chengdu 2005 (nos. 164–165), which also publishes a third one (no. 166). For discussion of their style and carving technique see Paris 2003, pp. 175–6. Three stone snakes from Sanxingdui are known to me while eight have been excavated at Jinsha (Chengdu 2005, p. 19). In addition to the snake illustrated here, another snake from Jinsha is published in Chengdu 2005, no. 168. By comparison, the Jinsha snakes are more realistic in rendering. 247 has raised eyebrows, somewhat sunken eye sockets, a slightly hooked nose, high cheekbones, and large ears that stick out perpendicularly and that are pierced, presumably for earrings. The eye sockets are defined by incised lines, as are the eyeballs and the mouth. The effect is of a pair of widely open eyes staring straight ahead and a mouth gaping as if in shock.519 The pose and appearance of this figure surely suggest a subdued person awaiting some terrifying and painful act to be inflicted upon him. This figure is one of at least twelve such figures so far unearthed at the Jinsha site.520 All share the same pose and hairstyle, and on all of them the facial features may originally have been highlighted in colors (as on Sanxingdui bronze figures). The residues on some of the figures show that the ears would have been delineated in red, while the pupils and the contours of the eyeballs would have been colored in black and the irises painted in white.521 These figures vary in size, from 17 to 26 cm tall, and differ somewhat in proportions and in the amount of detail included. They also have different facial expressions, ranging from dazed indifference to dramatic amazement, but this apparent difference may partly result from the present eroded and discolored conditions of the 519It appears that all the other similar figures at Jinsha have a mouth depicted by a single straight line. 520Four of these kneeling figures, including the present one, are published in Beijing 2002c, nos. 51–54. The same four are published in Chengdu 2005, nos. 159–162, in addition to a fifth one (no. 163). A rare in-situ combination of a stone kneeling figure and a stone tiger is published in Chengdu 2005, p. 17. Whether the combination was intentional or accidental is yet to be determined. 521Prior to the discovery at Jinsha, Wu Hung, citing finds from Chengdu and Sanxingdui as well as an unprovenanced figure in the collection of the Art Institute of Chicago, asserted that such figures were portrayed as being without eyes and used their supposed eyelessness to construct a difference between them and the Sanxingdui bronze figures (Wu 1997). The stone figures now unearthed at Jinsha prove his observation to be mistaken. For the Art Institute figure see Casadio et al. 2007. Facial features of the stone tigers and snakes are similarly highlighted in colors. 248 figures. Beyond the Jinsha site, a badly eroded figure, similar but much larger in size, measuring 50 cm tall, had been unearthed at another locus in Chengdu, and two small fragmentary kneeling figures with their heads missing were unearthed at the Sanxingdui site.522 No example has been found outside the Chengdu Plain so far; it seems that they may have been confined to the cultures in the Chengdu Plain. These stone kneeling figures are very different from the Sanxingdui bronze figures. For example, no Sanxingdui figure seems to be shown naked except for a topless, perhaps female kneeling figure carrying a zun vessel on her head (Fig. 4.31). The differences perhaps conveyed ethnic and/or social distinctions (perhaps the stone figures were captured aliens?). That similar stone figures were found at Sanxingdui further highlights the distinction between the two types. Yet, certain features such as the braided pigtail seems to suggest a connection, and certain facial features—eyebrows and eyes— are comparable to another, albeit unique, bronze kneeling or seated figure at Sanxingdui (Fig. 5.50). Still more impressive examples demonstrate the close relationship between Jinsha and Sanxingdui. One is a small bronze human figure from Jinsha about 20 centimeters tall (Fig. 10.12). This figure sports a fantastic crown composed of whirling strands. A different coiffure formed of similar strands can be seen on the seated figure from Sanxingdui in Figure 5.50. Though different in appearance, these headdresses are similar in their attention-catching quality. The Jinsha bronze figure also wears a long pigtail, more elaborate than the one worn by the stone figures and the single braid seen on many 522For the Chengdu figure see Wu Yi 1988 (p. 21 fig. 3), and Wu 1997 (p. 59 fig. 3). The Sanxingdui figures have not been published, but they are on display at the Sanxingdui Museum at Sanxingdui, Guanghan county, Sichuan province. 249 bronze heads at Sanxingdui. The most revealing feature is the figure’s gesturing hands. Apparently meant to hold something, these hands are familiar from the Sanxingdui lifesized stature (Fig. 5.17). A second piece of evidence for a close Jinsha–Sanxingdui relationship is a gold face (Fig. 10.13). A third is a gold band whose function is not clear (Fig. 10.14a). Now broken, it was originally a continuous belt, measuring about 60 cm in length. As noted in Section 7.2, the motif showing an arrow flying past a bird to penetrate the head of a fish is nearly identical with the one on the gold sheath from K1 at Sanxingdui. The comparison suggests that the circular motif next to the shooting scene on the Jinsha band might be understood as a simplified version of a face. In view of these comparisons we cannot doubt that Jinsha belonged to the same elite culture as Sanxingdui. Another prominent motif in the Sanxingdui–Jinsha iconography is the bird, as seen for instance on the circular gold ornament from Jinsha or in a more elaborate version on a handled collar disk of bronze (Fig. 10.14b). It seems to be related to birds at Sanxingdui such as those perching on the branches of the monumental tree (Fig. 6.32g). The whirling pattern in the center of the gold ornament reminds us of the hairdo of the standing figure in Figure 10.12. The conspicuous use of gold at Jinsha is another link with Sanxingdui, a trait that distances them from other cultures in early Bronze Age China. Despite the close similiarities with Sanxingdui, the Jinsha site also shows some important differences and raises puzzling questions. While the stone images are consistent in style, the jade implements, much more numerous than at Sanxingdui, appear to be a random gathering of types and materials from diverse sources. Bronze images on 250 the other hand are rare and small, on the scale of miniature images at Sanxingdui, with only one complete figure, which was discussed earlier. The other few images include human heads, parts of birds and other creatures, and foils in the shape of eyes and such. The most of the bronze artifacts are similar to the ones that at Sanxingdui have been identified as objects associated with trees. No complete vessel is known at Jinsha at this point, only a few small fragments ranging in date from around the twelfth to the tenth century BC (Figs. 10.15a, b). The fragment in Figure 10.15a is clearly part of a vessel (zun or lei)’s foot ring, bearing Style Va decoration familiar among vessels from K2 at Sanxingdui (compare zun K2[2]:129 in Figure 4.21). The fragment in Figure 10.15b consists of a tapir-like animal head locked in a semi-circular loop, surely the remnant of a swing handle for a vessel (probably a you).523 A tiger-shaped plaque similar to the two stray finds at Sanxingdui and an Erligang-style bronze ge blade similar to those found at the Xinfan Shuiguanyin site might be the earliest items among the bronzes (Figs. 4.1c, 8.2b). Again, the bronzes appear to be a random accumulation. Surely this impression is partly owed to the fact that the objects of the initial find were disturbed by the bulldozer and lack a clear stratigraphic relationship. The excavation done later suggested that they may have come from ten or so strata and features.524 The rapidly progressing fieldwork at Jinsha since 2001 suggests that while objects are often found scattered over the ground in a distribution quite unlike the usual pattern of a cemetery or a sacrificial deposit, objects of related types tend to concentrate in the same areas. Besides the bronzes and jades of the initial find, the archaeologists have 523For actual examples of you with swing handles ended in tapir-like animal heads see Beijing 1997c, pls. 56, 115, 175, and 192. 524Wenwu 2004.4, p. 6. 251 discerned an area of unfinished lithic implements and possible manufacturing preforms, a dense concentration of boar tusks and deer antlers, an area of elephant tusks and jades, and a small area of turtle shells that were possibly used in ritual prognostication. Occasionally objects were deposited in pits, as in the case of a cache of elephant tusks (Fig. 10.7). The nature of the Jinsha site is far from clear at this early stage of fieldwork. The cluster of large-scale buildings might represent a palatial compound, as the archaeologists conjecture. The site was probably occupied also by a large population of lesser social standing, as indicated by the burial grounds with poorly furnished tombs and the clusters of modest buildings. The archaeologists have interpreted the scattered yet patterned distribution of objects as the result of sacrificial activities, but this hypothesis may not apply in every situation. It is possible that some areas might be workshop surfaces, such as the area of lithic implements that includes raw materials, half-finished or waste products in addition to finished ones (Fig. 10.8a). As for the elephant tusks, those at Jinsha were not burned, while the tusks at Sanxingdui were apparently consumed in a burning sacrifice; moreover, in addition to whole tusks, at Jinsha there were neatly cut chunks that were perhaps on the way to being made into ivory artifacts. Also unclear is Jinsha’s exact relationship with the Sanxingdui city, now that we know that they overlapped in time and have many similarities in elite material culture. At present, no wall and no large-scale bronzes have been found at Jinsha, and the chronology of cultural development at the site is yet to be firmly established. Nevertheless, Jinsha, Shi’erqiao and related sites begin to present in broad strokes the trajectory of development of the Chengdu settlement. It looks as though during the time 252 of their co-existence, comprising primarily the early phases of the Jinsha and Shi’erqiao sites, the Chengdu settlement was secondary to Sanxingdui. After the Sanxingdui site was abandoned, around 1000 BC, the situation appears to have changed greatly, with an expansion of the Chengdu settlement, whose remains from this stage are known at a dozen or so separate sites spread over about 10 km. The luxury objects and large-scale buildings set Jinsha apart from all the other sites, and they begin to fill a frustrating blank in the archaeological record for the six or seven centuries after the time of the Sanxingdui pits.525 525Prior to the discovery at Jinsha, the only notable find of elite artifacts in the Chengdu Plain for the entire first half of the first millennium BC was an isolated pair of bronze hoards, of unknown archaeological context, found at Peng Xian (now Pengzhou) Zhuwajie, about 10 km south of Sanxingdui. These hoards are not related in any obvious way to the Sanxingdui civilization; on the contrary, the most notable items in them are conspicuous intrusions of early Western Zhou material culture, and they point to contact with the Weihe River valley of Shaanxi to the north. See Xu 2001 (pp. 35–6), Falkenhausen 2001, and Sun Hua 2003. 253 Appendix. Elemental Analysis and Lead-Isotope Ratios at Sanxingdui and other Early Bronze Age Sites Throughout the Bronze Age in China, both binary (copper-tin or copper-lead) and ternary (copper-tin-lead) alloys are commonly encountered. Elemental analyses have now been reported for a considerable number of archaeologically provenanced bronzes. They show a wide range of compositions among objects from a single site, indeed from a single tomb. Table 4 gives some idea of the range of compositions reported from the Sanxingdui pits (24 bronzes, vessels excluded) and from sites earlier than the pits (Erlitou and Erligang), contemporary with them (Fu Hao’s tomb at Anyang and the Xin’gan tomb in Jiangxi), and slightly later (a twelfth century tomb at Anyang Guojiazhuang and more modest twelfth century burials in the Yinxu West Sector at Anyang). Only the Fu Hao bronzes show any appreciable control of alloy composition. Otherwise the limits are very wide, at Sanxingdui as elsewhere. This should not be surprising. Even assuming an unrestricted supply of the constituent metals, alloy composition is very difficult to control, and no ancient bronze founder would be likely to take the trouble to purify his metals and mix them in specific proportions unless he had some good reason to do so. In the case of metal used for weapons, where mechanical properties are important, alloy control might have been attempted (though analytical data show little sign of it). For vessels and statuary, however, all that was needed was an alloy that would cast well, and this is not a severe constraint. More important, surely, was the need to recycle a material as valuable as bronze, and the founder who tossed miscellaneous artifacts (for instance captured bronze weapons) into the crucible 254 relinquished control of composition. It seems likely that the only control normally exercised came at the stage when the bronze in the crucible was molten: if the color or viscosity of the molten metal did not seem right, the founder added copper or tin or lead as required to achieve a look that his experience told him would pour well. It is not clear that elemental analysis has much of archaeological interest to reveal. Lead-isotope analysis by contrast does seem to say something useful about the trade in metals. The lead from a given lead mine has a distinctive isotopic composition that does not change during the smelting and casting processes. (Nor does recycling change the isotopic composition, provided that all the bronzes being melted together drew their lead from a single mine.) Unless it mixes leads from two or more distinct mines, therefore, the lead in a bronze artifact can in principle be matched with the lead mine it came from, or with other bronzes whose lead came from the same mine.526 The isotopic analyses performed so far show patterns consistent enough to suggest that the mixing of lead from different sources did not happen so often as to make the typing of leads uninformative. In fact, very interesting results have been obtained. Lead-isotope analyses of bronzes from Sanxingdui (imported vessels as well as local castings), bronzes from the Xin’gan tomb, bronzes from Chenggu and Yang Xian in the Hanzhong Basin, and early bronzes from Anyang (i.e. those from Fu Hao’s tomb, not later ones) show that they all contain lead of the same unusual isotopic composition, suggesting that the same lead source was supplying foundries in all three places.527 Analysis of samples from lead 526For more detail see Bagley 1987, pp. 558–60 (appendix 2) and references cited there. 527Beijing 1997e (pp. 245–50), Jin et al. 1998, Beijing 1999a (pp. 490–9), Zhao Congcang 2006 (pp. 250–9). Lead-isotope ratios for 53 samples from Sanxingdui bronzes are reported in Beijing 1999a, pp. 498–9. Of the 53 samples, 20 come from fifteen vessels; the remaining 33 are from objects likely to be local castings. 255 mines strongly suggests that the source was in Yunnan province.528 It seems likely that Yunnan lead was shipped north to Sichuan, eastward down the Changjiang River, and then to destinations both south (Xin’gan) and north (Anyang) of the Changjiang (after the time of Fu Hao’s tomb Anyang must have switched to some other source). The trade was presumably in lead ingots, but we should remember that lead could also travel in the form of finished bronze artifacts. Since the recycling of wandering artifacts might be expected to mix leads from different mines, it is surprising that the lead-isotope analyses performed so far show as much regularity as they do. 528Li Xiaocen 1993 reports analyses from several hundred lead mines and makes a convincing case for Yunnan as the source. 256 Chinese Glossary A Ankang Anyang aoren zuo 安康安陽凹刃鑿 B Ba Baijiazhuang Baodun Baoduncun Baoji Baoshan ben bi Bianduishan 巴白家荘寶墩寶墩村寶雞寶山錛璧邊堆山 C Cangbaobao chan Changjiang Changsha Chengdu Chenggu Chengziya Chongqing Chongzhou cong 倉包包鏟長江長沙成都城固城子崖重慶崇州琮 D Dafanzhuang Daninghe dao Dawenkou Dayan Dayangzhou Dengjiawan ding dou Dujiangyan 大范荘大寧河刀大汶口大堰大洋洲鄧家灣鼎豆都江堰 257 E Erligang Erlitou 二里岡二里頭 F Fanba Fangchijie fang ding fang lei fang yi fang zun fu Fu Hao Funan Fuqinxiaoqu Fuquanshan 范壩方池街方鼎方罍方彝方尊斧婦好阜南抚琴小區福泉山 G Gaocheng Gaopian ge gu guan (vessel) guan (beads) guang Guanghan Guangming ribao Gucheng gui Guilinxiang Guojiazhuang Guo Moruo 藁城髙駢戈觚罐管觥廣漢光明日報古城簋桂林鄉郭家荘郭沫若 H Haiyang 海陽 Hanchow: same as Hanzhou Hanshui 漢水 Hanzhong 漢中 Hanzhou 漢州 he 盉 Hengyang 衡陽 Hetaocun 核桃村 258 Hetian hu Huachengcun Huaihe huan Huangcai Huangpi Huangzhongcun Huarong Hui Huilong Hui Xian 和田壶化成村淮河環黃材黃陂黃忠村華容回迴龍輝縣 J jia Jialingjiang jian jiandi zhan Jianghan Jiangling Jiangning Jin jin Jingzhou Jinsha Jinshayuan 斝嘉陵江劍尖底盞江漢江陵江寧晋斤荊州金沙金沙園 L Laoniupo lei leiwen Liangzhu Li Bo Lingbao Linyi Liuan Liuyang Liyang Longshan Longxi Lushanmao 老牛坡罍雷纹良渚李白靈寶臨沂六安瀏陽溧陽龍山龍溪蘆山峁 259 M Majiayao Mamuhe Mangcheng mao Mao Xian Mayang Meiyuan Mianyang Minjiang Minshan moshi Mu shi 馬家窑马牧河芒城矛茂縣麻秧梅園綿陽岷江岷山磨石牧誓 N Nanchang Nanjing nao Ningxiang 南昌南京鐃寧鄉 P pan Panlongcheng Peng Xian Pengzhou Pijiang Pinglu Pi Xian pou 盘盘龍城彭縣彭州郫江平陸郫縣瓿 Q Qianzhuang Qijia Qin Qingjian Qingjiang Qingjiangcun Qinling qixing bi qixing pei 前荘齊家秦清澗清江清江村秦嶺戚形璧戚形珮 R 260 Rensheng 仁勝 S Sanhe Huayuan Sanxing Sanxing ban yue Sanxingdui Shang Shang shu Shashi Shaxi Shenmu Shi’erqiao Shifang Shijiahe Shijiefang Shilou Shimao Shimen Shizinao Shu Shuanghe Shuiguanyin Simatai Si Mu Wu Suozitian 三合花園三星三星伴月三星堆商尚書沙市沙溪神木十二橋什邡石家河十街坊石楼石峁石門獅子闹蜀雙河水觀音司馬臺司母戊梭子田 T taotie Tianshui Tuojiang 饕餮天水沱江 W Weigang Weihe Wenchuan Wenjiang Wuhan Wushan 桅崗渭河汶川溫江武漢巫山 X Xi’an 西安 261 Xiaojiawuji Xiaomeiling Xiaotun Xinfan Xin’gan Xinghuacun Xinjin Xiquankan 肖家屋脊小梅嶺小屯新繁新淦杏花村新津西泉坎 Y Ya’an Yan yan Yan’an Yang Xian Yang Xiaowu Yangzishan Yanting Yazihe Yidu Yinchuan Yingpanshan you Yu yuan Yueliangwan Yueyang Yufu 雅安燕甗延安洋縣楊晓鄔羊子山鹽亭鴨子河宜都银川营盘山卣弓+魚瑗月亮灣岳陽魚鳧 Z Zaoshi Zaoyang zhang zhangxing shijian Zhengzhou Zhenwu zhi Zhihuijie Zhongxing Zhou Zhou li Zhouyuan 皂市棗陽璋璋形饰件鄭州真武觶指揮街中興周周禮周原 262 zhu Zhuwajie Zizhu zun zuo Zuo zhuan 珠竹瓦街紫竹尊鑿左傳 263 Works Cited 1. Periodicals Antiquity. Gloucester, England Artibus Asiae. Zurich, Switzerland Arts Asiatiques. Paris Beiträge zur allgemeinen und vergleichenden Archäologie. Munich, Germany The Bulletin of the Cleveland Museum of Art. Cleveland Chengdu kaogu faxian 成都考古發現. Beijing Chengdu wenwu 成都文物. Chengdu Dongnan wenhua 東南文化. Nanjing Gudai wenming 古代文明. Beijing Gugong wenwu yuekan 故宮文物月刊. Taibei Gugong xueshu jikan 故宮學術季刊. Taibei Hsieh-ta Journal of Chinese Studies. Chengdu Huaxia kaogu 華夏考古. Zhengzhou Hunan kaogu jikan 湖南考古辑刊. Changsha Jianghan kaogu 江漢考古. Wuhan Journal of the West China Border Research Society. Chengdu Kaogu 考古. Beijing Kaogu tongxun 考古通訊. Beijing Kaogu xuebao 考古學報. Beijing Kaoguxue jikan 考古學集刊. Beijing 264 Kaoguxue yanjiu 考古學研究. Beijing Kaogu yu wenwu 考古與文物. Xi’an Koshi shunjū 古史春秋. Kyoto Lishi yuekan 歷史月刊. Taibei Nanfang minzu kaogu 南方民族考古. Chengdu Nanfang wenwu 南方文物. Nanchang Oriental Art. London Orientations. Hong Kong Shirin 史林. Kyoto Shuowen yuekan 說文月刊 (Chongqing edition). Chongqing Sichuan Daxue xuebao 四川大學學報. Chengdu Sichuan wenwu 四川文物. Chengdu Transactions of the Oriental Ceramic Society. London Wenbo 文博. Xi’an Wenwu 文物. Beijing Wenwu cankao ziliao 文物參考資料. Beijing Wenwu ziliao congkan 文物資料叢刊. Beijing Zhongguo Lishi Bowuguan guankan 中國歷史博物館館刊. Beijing Zhongguo wenwu bao 中國文物報. Beijing Zhonghua wenhua luntan 中華文化論壇. Chengdu Zhongyuan wenwu 中原文物. Zhengzhou Ziran kexue shi yanjiu 自然科學史研究 265 2. Titles Ao & Liu 1991 Ao Tianzhao 敖天照 and Liu Yutao 劉雨濤. “Guanghan Sanxingdui kaogu jilue 廣漢三星堆考古記略”. In Li et al. 1991, pp. 331–8. Ao & Wang 1980 Ao Tianzhao and Wang Youpeng 王有鹏. “Sichuan Guanghan chutu Shang dai yuqi 四川廣漢出土商代玉器” . Wenwu 1980.9, p. 76. Asahara 1984 Asahara Tatsurō 淺原達郎. “Ka bunka tansakuno michi 夏文化探索の道”. Koshi shunjū 1 (1984), pp. 18–28. Bagley 1977 Robert W. Bagley. “P’an-lung-ch’eng: A Shang City in Hupei”. Artibus Asiae 39 (1977), pp. 165–219. Bagley 1987 _____. Shang Ritual Bronzes in the Arthur M. Sackler Collections. Washington, DC: Arthur M. Sackler Foundation, 1987. Bagley 1988 _____. “Sacrificial Pits of the Shang Period at Sanxingdui in Guanghan county, Sichuan province”. Arts Asiatiques 43 (1988), pp. 78–86. Bagley 1990a _____. “A Shang City in Sichuan Province”. Orientations, November 1990, pp. 52–67. A Chinese translation published in Li et al. 1993, pp. 69–75. Bagley 1990b _____. “Shang Ritual Bronzes: Casting Technique and Vessel Design”. Archives of Asian Art 43 (1990), pp. 6–20. Bagley 1992 _____. “Changjiang Bronzes and Shang Archaeology”. In Taibei 1992, pp. 209–55. Bagley 1993 _____. “An Early Bronze Age Tomb in Jiangxi Province”. Orientations, July 1993, pp. 20–36. Bagley 1995 _____. “What the bronzes from Hunyuan tell us about the foundry at Houma”. Orientations, January 1995, pp. 46–54. Bagley 1999 _____. “Shang Archaeology”. In Michael Loewe and Edward L. Shaughnessy, eds. The Cambridge History of Ancient China, chapter 3 (pp. 124–231). Cambridge: Cambridge University Press, 1999. Bagley 2001 Robert Bagley, ed. Ancient Sichuan: Treasures from a Lost Civilization. Seattle: Seattle Art Museum in association with Princeton University Press, 2001. Barnard 1990 Noel Barnard. “Some Preliminary Thoughts on the Significance of the Kuang-han Pit-burial Bronzes and Other Artifacts”. Beiträge zur allgemeinen und 266 vergleichenden Archäologie 9–10 (1990), pp. 249–79. A Chinese translation published in Nanfang minzu kaogu 5 (1993), pp. 25–66. Beijing 1972 Chutu Wenwu Zhanlan Gongzuozu 出土文物展覽工作組, ed. Wenhua Dageming qijian chutu wenwu (vol. 1) 文化大革命期間出土文物（第一辑）. Beijing: Wenw Chubanshe, 1972. Beijing 1974 Shandong Sheng Wenwu Guanlichu 山東省文物管理處, Ji’nan Shi Bowuguan 濟南市博物館, ed. Dawenkou: Xinshiqi shidai muzang fajue baogao 大汶口：新石器時代墓葬發掘報告. Beijing: Wenwu Chubanshe, 1974. Beijing 1979 Shaanxi Sheng Kaogu Yanjiusuo 陝西省考古研究所, Shaanxi Sheng Wenwu Guanli Weiyuanhui 陝西省文物管理委員會, Shaanxi Sheng Bowuguan 陝西省博物館, ed. Shaanxi chutu Shang Zhou qingtongqi (1) 陝西出土商周青銅器 (1). Beijing: Wenwu Chubanshe, 1979. Beijing 1982 Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo 中國社會科學院考古研究所, ed. Kaogu gongzuo shouce 考古工作手冊. Beijing: Wenwu Chubanshe, 1982. Beijing 1983 Hunan Sheng Bowuguan 湖南省博物館, ed. Hunan Sheng Bowuguan 湖南省博物館. Beijing: Wenwu Chubanshe, 1983 (Japanese edition: Konan-shô Hakubutsukan 湖南省博物館. Chûgoku no hakubutsukan 中國博物館 first series, vol. 2. Tokyo: Kodansha, 1981). Beijing 1984a Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo, ed. Yinxu Fuhao mu 殷墟婦好墓(2nd ed.). Beijing: Wenwu Chubanshe, 1984. Beijing 1984b Shanghai Bowuguan Qingtongqi Yanjiuzu 上海博物館青銅器研究組, ed. Shang Zhou qingtongqi wenshi 商周青銅器纹饰. Beijing: Wenwu Chubanshe, 1984. Beijing 1985a Hebei Sheng Wenwu Yanjiusuo 河北省文物研究所, ed. Gaocheng Taixi Shang dai yizhi 藁城臺西商代遺址. Beijing: Wenwu Chubanshe, 1985. Beijing 1985b Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo. Yinxu qingtongqi 殷墟青銅器. Beijing: Wenwu Chubanshe, 1985. Beijing 1991 Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo. Zhongguo Kaoguxue zhong tanshisi niandai shujuji, 1965–1991 中國考古學中碳十四年代數據集 1965– 1991. Beijing: Wenwu Chubanshe, 1991. Beijing 1993 Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo, ed. Kaogu jinghua— Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo jiansuo sishinian jinian 考古精華—中國社會科學院考古研究所建所四十年紀念. Beijing: Kexue Chubanshe, 1993. 267 Beijing 1994a Wenwu Chubanshe 文物出版社, ed. Shangdai Shu ren mibao, Sichuan Guanghan Sanxingdui yizhi 商代蜀人秘寶，四川廣漢三星堆遺址 (Zhongguo kaogu wenwu zhi mei 中國考古文物之美, vol. 3). Beijing: Wenwu Chubanshe, 1994. Beijing 1994b Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo. Yinxu de faxian yu yanjiu 殷墟的發現與研究. Beijing: Kexue Chubanshe, 1994. Beijing 1994c Zhongguo Qingtongqi Quanji Bianji Weiyuanhui 中國青銅器全集編輯委員會, ed. Zhongguo qingtongqi quanji 13, Ba Shu 中國青銅器全集 13, 巴蜀. Beijing: Wenwu Chubanshe, 1994. Beijing 1996a Sichuan Sheng Wenwu Kaogu Yanjiusuo 四川省文物考古研究所, ed. Sichuan kaogu lunwenji 四川考古論文集. Beijing: Wenwu Chubanshe, 1996. Beijing 1996b Zhongguo Qingtongqi Quanji Bianji Weiyuanhui, ed. Zhongguo qingtongqi quanji 1, Xia, Shang 1 中國青銅器全集 1, 夏、商 1. Beijing: Wenwu Chubanshe, 1996. Beijing 1997a _____. Zhongguo qingtongqi quanji 2, Shang 2 中國青銅器全集 2, 商 2. Beijing: Wenwu Chubanshe, 1997. Beijing 1997b _____. Zhongguo qingtongqi quanji 3, Shang 3 中國青銅器全集 3, 商 3. Beijing: Wenwu Chubanshe, 1997. Beijing 1997c _____. Zhongguo qingtongqi quanji 6, Xi Zhou 2 中國青銅器全集 6, 西周 2. Beijing: Wenwu Chubanshe, 1997. Beijing 1997d _____. Zhongguo qingtongqi quanji 11, Dong Zhou 5 中國青銅器全集 11, 東周 5. Beijing: Wenwu Chubanshe, 1997. Beijing 1997e Jiangxi Sheng Wenwu Kaogu Yanjiusuo 江西省文物考古研究所, Jiangxi Sheng Bowuguan 江西省博物館, Xin’gan Xian Bowuguan 新淦縣博物館. Xin’gan Shang dai damu 新淦商代大墓. Beijing: Wenwu Chubanshe, 1997. Beijing 1997f Zhongguo Wenwu Jinghua Bianji Weiyuanhui 中國文物精華編輯委員會, ed. Zhongguo wenwu jinghua 1997 中國文物精華 1997. Beijing: Wenwu Chubanshe, 1997. Beijing 1998a Sichuan Sheng Wenwu Kaogu Yanjiusuo Sanxingdui Gongzuozhan 四川省文物考古研究所三星堆工作站, Guanghan Shi Wenwu Guanlisuo 廣漢市文物管理所. “Sanxingdui yizhi Zhenwu Cangbaobao jisikeng diaocha jianbao 三星堆遺址真武倉包包祭祀坑調查簡報”. In Beijing 1998b, pp. 78–90. 268 Beijing 1998b Sichuan Sheng Wenwu Kaogu Yanjiusuo. Sichuan kaogu baogaoji 四川考古報告集. Beijing: Wenwu Chubanshe, 1998. Beijing 1998c Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo, ed. Zhongguo Shang wenhua guoji xueshu taolunhui lunwenji 中國商文化國際學術討論會論文集. Beijing: Dabaike Quanshu Chubanshe, 1998. Beijing 1998d Zhongguo Qingtongqi Quanji Bianji Weiyuanhui, ed. Zhongguo qingtongqi quanji 4, Shang 4 中國青銅器全集 4, 商 4. Beijing: Wenwu Chubanshe, 1998. Beijing 1998e Sichuan Sheng Wenwu Guanli Weiyuanhui 四川省文物管理委員會, Sichuan Sheng Wenwu Kaogu Yanjiusuo, Wushan Xian Wenhuaguan 巫山縣文化館. “Wushan jingnei Changjiang Daninghe liuyu guyizhi diaocha jianbao 巫山境內長江大寧河流域古遺址調查簡報”. In Beijing 1998b. pp. 1–10. Beijing 1998f Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo, Anyang Yinxu Guojiazhuang Shang dai muzang, 1982 nian–1992 nian kaogu fajue baogao 安陽殷墟郭家荘商代墓葬，1982 年代–1992 年考古發掘報告. Beijing: Zhongguo Dabaike Quanshu Chubanshe, 1998. Beijing 1999a Sichuan Sheng Wenwu Kaogu Yanjiusuo, ed., Sanxingdui jisikeng 三星堆祭祀坑. Beijing: Wenwu Chubanshe, 1999. Beijing 1999b Gugong Bowuyuan 故宮博物院, ed. Gugong qingtongqi 故宮青銅器. Beijing: Zijincheng Chubanshe, 1999. Beijing 1999c Guojia Wenwuju 國家文物局, Zhongguo Lishi Bowuguan 中國歷史博物館, Zhongguo Geming Bowuguan 中國革命博物館, ed. Guo zhi guibao, Zhongguo wenwu shiye wushi nian, 1949–1999 國之瑰寶，中國文物事業五十年，1949-1999. Beijing: Chaohua Chubanshe, 1999. Beijing 1999d Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo. Yanshi Erlitou 1959 nian–1978 nian kaogu fajue baogao 偃師二里頭 1959 年-1978 年考古發掘報告. Beijing: Zhongguo Dabaike Quanshu Chubanshe, 1999. Beijing 1999e Hubei Sheng Jingzhou Bowuguan, Hubei Sheng Wenwu Kaogu Yanjiusuo, Beijing Daxue Kaoguxi Shijiahe Kaogudui 湖北省荊州博物館、湖北省文物考古研究所、北京大學考古系石家河考古隊. Xiaojiawuji 肖家屋脊. 2 vols. Beijing: Wenwu Chubanshe, 1999. 269 Beijing 1999f Henan Sheng Wenwu Kaogu Yanjiusuo 河南省文物考古研究所, Zhengzhou Shi Wenwu Kaogu Yanjiusuo 鄭州市文物考古研究所. Zhengzhou Shang dai tongqi jiaocang 鄭州商代銅器窖藏. Beijing: Kexue Chubanshe, 1999. Beijing 2001a Chengdu Shi Wenwu Kaogu Yanjiusuo 成都市文物考古研究所, Pi Xian Bowuguan 郫縣博物館. “Sichuan Sheng Pi Xian Gucheng yizhi 1998–1999 niandu fajue shouhuo 四川郫縣古城遺址 1998–1999 年度發掘收獲”. Chengdu kaogu faxian 1999 (2001), pp. 29–39. Beijing 2001b Zhong Ri Lianhe Kaogu Diaochadui 中日聯合考古調查隊. “Dujiangyan Shi Mangcheng yizhi 1998 niandu fajue gongzuo jianbao 都江堰市芒城遺址 1998 年度發掘工作簡報”. Chengdu kaogu faxian 1999 (2001), pp. 54–98. Beijing 2001c Zhong Ri Lianhe Kaogu Diaochadui. “Dujiangyan Shi Mangcheng yizhi 1999 niandu fajue gongzuo jianbao 都江堰市芒城遺址 1999 年度發掘工作簡報”. Chengdu kaogu faxian 1999 (2001), pp. 99–126. Beijing 2001d Chengdu Shi Wenwu Kaogu Yanjiusuo, Pi Xian Bowuguan. “Sichuan Sheng Pi Xian Qingjiangcun yizhi diaocha fajue shouhuo 四川省郫縣清江村遺址調查發掘收獲”. Chengdu kaogu faxian 1999 (2001), pp. 146–63. Beijing 2001e Hubei Sheng Wenwu Kaogu Yanjiusuo 湖北省文物考古研究所, ed. Panlongcheng—1963–1994 nian kaogu fajue baogao 盤龍城—1963–1994 年考古發掘報告. 2 vols. Beijing: Wenwu Chubanshe, 2001. Beijing 2001f Henan Sheng Wenwu Kaogu Yanjiusuo, ed. Zhengzhou Shang cheng— 1953–1985 nian kaogu fajue baogao 鄭州商城—1953–1985 年考古發掘報告. 3 vols. Beijing: Wenwu Chubanshe, 2001. Beijing 2002a Chengdu Shi Wenwu Kaogu Yanjiusuo, Aba Zangzu Qiangzu Zizhizhou Wenguansuo 阿壩藏族羌族自治州文管所, Mao Xian Bowuguan 茂縣博物館. “Sichuan Mao Xian Yingpanshan yizhi shijue baogao 四川茂縣營盤山遺址試掘報告”. Chengdu kaogu faxian 2000 (2002), pp. 1–77. Beijing 2002b Xibei Daxue Wenbo Xueyuan 西北大學文博學院. Chenggu Baoshan, 1998 nian fajue baogao 城固寶山, 1998 年發掘報告. Beijing: Wenwu Chubanshe, 2002. 270 Beijing 2002c Chengdu Shi Wenwu Kaogu Yanjiusuo, Beijing Daxue Kaogu Wenboyuan 北京大學考古文博院. Jinsha taozhen—Chengdu Shi Jinshacun yizhi chutu wenwu 金沙淘珍—成都市金沙村遺址出土文物. Beijing: Wenwu Chubanshe, 2002. Beijing 2003 Hubei Sheng Wenwu Kaogu Yanjiusuo, Beijing Daxue Kaoguxuexi, Hubei Sheng Jingzhou Bowuguan Shijiahe Kaogudui 湖北省文物考古研究所、北京大學考古學系、湖北省荊州博物館石家河考古隊. Dengjiawan 鄧家灣. Beijing: Wenwu Chubanshe, 2003. Beijing 2004 Chengdu Shi Wenwu Kaogu Yanjiusuo. “Chengdu Jinsha yizhi ‘Zhixin Jinshayuan yiqi’ didian fajue jianbao 成都金沙遺址‘置信金沙園一期’地點發掘簡報”. Chengdu kaogu faxian 2002 (2004), pp. 1–41. Beijing 2005 Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo, ed., Anyang Yinxu chutu yuqi 安陽殷墟出土玉器. Beijing: Kexue Chubanshe, 2005. Beijing 2006a Zhongguo Guojia Bowuguan 中國國家博物館, Jiangxi Sheng Wenhuating 江西省文化廳. Shang dai Jiangnan—Jiangxi Xin’gan Dayangzhou chutu wenwu jicui 商代江南—江西新淦大洋洲出土文物辑粹. Beijing: Zhongguo Shehui Kexue Chubanshe, 2006. Beijing 2006b Chengdu Shi Wenwu Kaogu Yanjiusuo, ed. Jinsha yuqi 金沙玉器. Beijing: Kexue Chubanshe, 2006. Bowman 1990 Sheridan Bowman. Radiocarbon Dating. London: British Museum Publications, 1990. Bunker 1993 Emma C. Bunker. “Gold in the ancient Chinese world: a cultural puzzle”. Artibus Asiae 53.1/2 (1993), pp. 27–50. Cai Ge 1995 Cai Ge 蔡革. “Cong Guanghan Sanxingdui jisikeng chutu wenwu kan dangshi Shu ren de fushi tezheng 從廣漢三星堆祭祀坑出土文物看當時蜀人的服飾特徵”. Sichuan wenwu 1995.2, pp.18–24. Casadio et al. 2007 Francesca Casadio, Jay Xu, Elinor Pearlstein, Katherine T. Faber, Ariel Knowles, and Jing Zhichun, “A Stone Kneeling Figure in the Art Institute of Chicago: New Evidence from Scientific Investigations and Archaeological Finds in China.” In Janet G. Douglas, Paul Jett, and John Winter, ed., Scientific Research on the Sculptural Arts of Asia: Proceedings of the Third Forbes Symposium at the Freer Gallery of Art, pp. 3–11. London: Archetype Publications, 2007. Chase 1981 W. T. Chase. “Bronze Casting in China: A Short Technical History”. In Kuwayama 1981, pp. 100–23. 271 Chase 1991 _____. Ancient Chinese Bronze Art: Casting the Precious Sacral Vessel. New York: China House Gallery, 1991. Chen et al. 1998 Chen De’an 陈德安, Wei Xuefeng 魏學峰, Li Weigang 李偉綱. Sanxingdui: Changjiang shangyou wenming zhongxin tansuo 三星堆：長江上游文明中心探索. Chengdu: Sichuan Renmin Chubanshe, 1998. Chen & Chen 1987 Chen Xiandan 陈顯丹, Chen De’an. “Shixi Sanxingdui yizhi Shang dai yihao keng de xingzhi ji youguan wenti 試析三星堆遺址商代一號坑的性質及有關問題”. Sichuan wenwu 1987.4, pp. 27–9. Chen & Fang 1993 Chen Zhida 陳志逹, Fang Guojin 方國錦, ed. Zhongguo yuqi quanji 2, Shang, Xi Zhou 中國玉器全集 2，商西周. Shijiazhuang: Hebei Meishu Chubanshe, 1993. Chen & Liu 2002 Chen Xiandan, Liu Jiasheng 劉家勝. “Lun Sanxingdui wenhua yu Baodun wenhua zhi guanxi 論三星堆文化與寶墩文化之關係”. Sichuan wenwu 2002.4, pp. 3–6. Chen & Zhou 1995 Chen Jiangfeng 陳江風, Zhou Tiexiang 周铁項. “Ye tan Guanghan Sanxingdui yubi de wenhua gongneng 也談廣漢三星堆玉璧的文化功能”. Sichuan wenwu 1995.2, pp. 14–7. Chen De’an 1990 Chen De’an. “Qian shi Sanxingdui erhao jisikeng chutu de ‘bianzhang’ tu’an 浅釋三星堆二號祭祀坑出土的‘邊璋’圖案”. Nanfang minzu kaogu 2(1990), pp. 85–90. Chen De'an 1992 _____. “Sanxingdui jisikeng chutu qingtong mianju yanjiu 三星堆祭祀坑出土青銅面具研究”. Sichuan wenwu 1992 (special issue on studies of ancient Shu culture at Sanxingdui), pp. 38–44. Chen De'an 1994 _____. “Shilun Sanxingdui yuzhang de zhonglei, yuanyuan jiqi zongjiao yiyi 試論三星堆玉璋的種類、渊源及其宗教意義”. In Deng Cong 1994, pp. 87–99. Chen De'an 1998 _____. “Sanxingdui yizhi de faxian yu yanjiu 三星堆遺址的發現與研究”. Zhonghua wenhua luntan 1998.2, pp. 57–63. Chen De'an 2000 _____. Sanxingdui: gu Shu wangguo de shengdi 三星堆：古蜀王國的聖地. Chengdu: Sichuan Renmin Chubanshe, 2000. Chen Fangmei 1991 Chen Fangmei 陳芳妹. “Xiaotun wuzuomu de qingtong rongqi 小屯五座墓的青銅容器”. In Song Wenxun 宋文薰, Li Yiyuan 李亦園, Xu Zhuoyun 許倬 272 雲, Zhang Guangzhi 張光直, ed., Kaogu yu lishi wenhua: qingzhu Gao Quxun xiansheng bashi dashou lunwenji 考古與歷史文化：慶祝髙去尋先生八十大壽論文集, pp. 181– 232. Taibei: Zhengzhong Shuju, 1991. Chen Fangmei 1998 _____. Gugong Shang dai qingtong liqi tulu 故宮商代青銅禮器圖錄 (Shang Ritual Bronzes in the National Palace Museum Collection). Taibei: Guoli Gugong Bowuyuan, 1998. Cheng 1949 Cheng Te-k’un (=Zheng Dekun). “The T’ai-p’ing-ch’ang Culture, Szechwan”. Hsieh-ta Journal of Chinese Studies 1 (1949), pp. 67–81. Republished in Cheng Te-k’un, Studies in Chinese Archaeology, pp. 55–65. Hong Kong: The Chinese University Press, 1982. Chengdu 2004 Chengdu Wenwu Kaogu Yanjiusuo 成都文物考古研究所, ed., “Chengdu Jinsha yizhi ‘Zhixin Jinshayuan Yiqi’ didian fajue jianbao 成都金沙遺址‘置信金沙園一期’地點發掘簡報”. Chengdu kaogu faxian 2002 (2004), pp. 1–41. Chengdu 2005 Chengdu Wenwu Kaogu Yanjiusuo, ed., Jinsha, zaixian huihuang de gu Shu wangdu 金沙,再現輝煌的古蜀王都. Chengdu: Sichuan Renmin Chubanshe, 2005. Chengdu 2006 Sichuan Daxue Bowuguan 四川大學博物館, ed. Sichuan Daxue Bowuguan cangpin jicui 四川大學博物館藏品集萃. Chengdu: Sichuan Daxue Chubanshe, 2006. Chengdu wenwu 2001.1 Chengdu Shi Wenwu Kaogu Yanjiusuo, “Chengdu Shi 2000 nian tianye kaogu gongzuo shuyao 成都市 2000 年田野考古工作述要”. Chengdu wenwu 2001.1, pp. 62–8. Chen Xiandan 1988 Chen Xiandan. “Lun Guanghan Sanxingdui yizhi de xingzhi 論廣漢三星堆遺址的性質”. Sichuan wenwu 1988.4, pp. 9–11, 8. Chen Xiandan 1989a _____. “Guanghan Sanxingdui yizhi fajue gaikuang chubu fenqi—jianlun ‘zao Shu wenhua’ de tezheng jiqi fazhan 廣漢三星堆遺址發掘概況、初步分期—兼論‘早蜀文化’的特徵及其發展”. Nanfang minzu kaogu 2 (1989), pp. 213–29. Chen Xiandan 1989b _____. “Guanghan Sanxingdui yi er hao keng liangge wenti de tantao 廣漢三星堆一二號坑兩個問題的探討”. Wenwu 1989.5, pp. 36–8. Chen Xiandan 1989c _____. “Sanxingdui yi er hao keng jige wenti de yanjiu 三星堆一二號坑幾個問題的研究”. Sichuan wenwu 1989 (special issue on the Guanghan Sanxingdui site), pp. 11–22. 273 Chen Xiandan 1997 _____. “Guanghan Sanxingdui yizhi yi er hao keng de shidai xingzhi de zai taolun 廣漢三星堆遺址一二號坑的時代性質的再討論”. Sichuan wenwu 1997.4, pp. 8–12. Chen Xiandan 2001 _____. “Guanghan Sanxingdui Dashiji 廣漢三星堆大事記(1929– 2000.2)”. Zhonghua wenhua luntan 2001.1, pp. 41–5. Dai Yingxin 1980 Dai Yingxing 戴應新. “Shanbei Qingjian, Mizhi, Jia Xian chutu gudai tongqi 陝北清澗、米脂、佳縣出土古代銅器”. Kaogu 1980.1, pp. 95, 70. Dai Yingxin 1993 _____. “Shenmu Shimao Longshan wenhua yuqi tansuo 神木石峁龍山文化玉器探索”. Gugong wenwu yuekan 1993.8, pp. 44–55 (part 1); 1993.9, pp. 46– 61 (part 2); 1993.10, pp. 78–84 (part 3); 1993.11, pp. 44–51 (part 4); 1993.12, pp. 26–35 (part 5). Dai Yingxin 1994 _____. “Shimao yazhang jiqi gaizuo—Shimao Longshan wenhua yuqi yanjiu zhaji 石峁牙璋及其改作—石峁龍山文化玉器研究札記”. In Deng Cong 1994, pp. 79–85. Deng Cong 1994 Deng Cong [=Tang Chung] 鄧聰, ed. Nan Zhongguo ji linjin diqu guwenhua yanjiu: qingzhu Zheng Dekun jiaoshou congshi xueshu huodong liushi zhounian lunwenji 南中國及鄰近地區古文化研究：慶祝鄭德坤教授從事學術活動六十周年論文集 (Ancient Cultures of South China and Neighboring Regions: Essays in Honor of Professor Cheng Te-K’un on the Occasion of the Sixtieth Anniversary of His Academic Career). Hong Kong: The Chinese University of Hong Kong, 1994. Deng Shuping 1993 Deng Shuping 鄧淑萍. “Yetan Huaxi xitong de yuqi 也談華西系統的玉器”. Gugong wenwu yuekan 1993.8, pp. 56–65 (part 1); 1993.9, pp. 62–71 (part 2); 1993.10, pp. 86–93 (part 3); 1993.11, pp. 52–63 (part 4); 1993.12, pp. 36–45 (part 5). Dong Qixiang 1983 Dong Qixiang 董其祥. Ba shi xinkao 巴史新考. Chongqing: Chongqing Chubanshe, 1983. Duan Yu 1991 Duan Yu 段瑜. “Ba Shu guwenzi de liangxi jiqi qiyuan 巴蜀古文字的两系及其起源”. Chengdu wenwu 1991.3, pp. 20–33. Duan Yu 1993 _____. “Gudai Ba Shu yu Jindong wenming 古代巴蜀與近東文明”. Lishi yuekan 1993.2, pp. 92–100. Du Jinpeng 1992 Du Jinpeng 杜金鹏. “Fengdinghe yanjiu 封頂盉研究”. Kaogu xuebao, 1992.1, pp. 1–34. Du Jinpeng 1995 _____. “Guanghan Sanxingdui chutu Shang dai tongpai qianshuo 廣漢三星堆出土商代铜牌浅說”. Zhongguo wenwu bao, April 9, 1995. 274 Dye 1931 Daniel S. Dye. “Some ancient circles, squares, angles and curves in earth and in stone in Szechwan, China”. Journal of the West China Border Research Society 4 (1930–31), pp. 97–105. Ebrey 1996 Patricia Buckley Ebrey. The Cambridge Illustrated History China. Cambridge: Cambridge University Press, 1996. Falkenhausen 2001 Lothar von Falkenhausen. “The Chengdu Plain in the Early First Millennium B.C.: Zhuwajie.” In Bagley 2001, pp. 177–201. Falkenhausen 2002 _____. "Some Reflections on Sanxingdui”. I-tien Hsing, ed. Regional Culture, Religion, and Arts before the Seventh Century: Papers from the Third International Conference on Sinology, History Section, pp. 59–97. Taibei: Institute of History and Philology, Academia Sinica, 2002. Falkenhausen 2003 _____. “The External Connections of Sanxingdui”. Journal of East Asian Archaeology, 5.1–4 (2003), pp. 191–245. Fan Tuoyu 1992 Fan Tuoyu 樊拓宇. “Dujiangyan shi Mangcheng yizhi diaocha jiankuang 都江堰市芒城遺址調查簡况”. Chengdu wenwu 1992.4, pp. 107–15. Fan Xiaoping 1993 Fan Xiaoping 范小平. “Sanxingdui qingtong renxiangqun de shehui neirong he yishu xingshi chutan—jianyu Zhongdong diqu shanggu diaosu yishu zhi bijiao 三星堆青銅人像羣的社會內容和藝術形式初探—兼與中東地區上古雕塑藝術之比較”. In Li et al. 1993, pp. 107–15. Feng & Tong 1979 Feng Hanji 馮漢骥, Tong Enzheng 童恩正. “Ji Guanghan chutu de yushiqi 記廣漢出土的玉石器”. Wenwu 1979.2, pp. 31–7, 30; and Sichuan Daxue xuebao 1979.1, pp. 79–85. Reprinted in Feng Hanji 1985, pp. 11–8. Feng Guanghong 1999 Feng Guanghong 馮廣宏. “Chengdu guchengzhiqun tigong de gushi xinxi 成都古城址羣提供的古史信息”. Chengdu wenwu 1999.2, pp. 36–40. Feng Hanji 1980 Feng Hanji. “Sichuan Peng Xian chutu de tongqi 四川彭縣出土的銅器”. Wenwu 1980.12, pp. 38–47. Reprinted in Feng Hanji 1985, pp. 19–27. Feng Hanji 1985 _____. Feng Hanji kaoguxue lunwenji 馮漢骥考古學論文集. Beijing: Wenwu Chubanshe, 1985. Fong 1980 Wen Fong, ed. The Great Bronze Age of China. New York: Metropolitan Museum, 1980. Franklin 1981 Ursula M. Franklin. “The Beginnings of Metallurgy in China: A Comparative Approach”. In Kuwayama 1981, pp. 94–8. 275 Franklin 1999 _____. The Real World of Technology (revised ed.). Toronto: Anansi, 1999. Gao & Xing 1995 Gao Dalun 髙大倫, Xing Jinyuan 邢進原. “Sichuan Daxue Bowuguan shoucang de Han yiqian bufen yushiqi 四川大學博物館収藏的漢以前部分玉石器”. Wenwu 1995.4, pp. 68–75. Gao & Xing 1998 _____. “Sichuan liangchu bowuguan cang Sanxingdui yushiqi de xinrenshi 四川两處博物館藏三星堆玉石器的新認識”. In Deng Cong, ed. Dong Ya yuqi 東亞玉器 (East Asian Jade: Symbol of Excellence) (3 vols.), vol. 2, pp. 25–9. Shatin, Hong Kong: Chinese University of Hong Kong Press, 1998. Ge 1997 Yan Ge. “The Coexistence of artistic styles and the pattern of interaction: Sanxingdui during the second millennium BC”. Ph D dissertation. Pittsburgh: University of Pittsburgh, 1997. Ge & Linduff 1990 Yan Ge and Katheryn M. Linduff. “Sanxingdui: a new Bronze Age site in southwest China”. Antiquity 64 (1990), pp. 505–13. Ge Jieping 1959 Ge Jieping 葛介屏. “Anhui Funan faxian Yin Shang shidai de qingtongqi 安徽阜南發現殷商時代的青銅器”. Wenwu 1959.1, inside front cover. Gettens 1967 Rutherford John Gettens. “Joining Methods in the Fabrication of Ancient Chinese Bronze Ceremonial Vessels”. In William J. Young, ed. Application of Science in Examination of Works of Art, pp. 205–17. Boston: Museum of Fine Arts, 1967. Gettens 1969 _____. The Freer Chinese Bronzes, Volume II, Technical Studies. Washington, D.C.: Freer Gallery of Art, 1969. Goepper 1995 Roger Goepper, ed. Das alte China: Menschen und Götter im Reich der Mitte. Essen: Kulturstiftung Ruhr, 1995. Gong Guoqiang 1997 Gong Guoqiang 龔國強. “Jianlun Shang Zhou wangguo jiqi zhoubian diqu de huangjin qishi 簡論商周王國及其週邊地區的黃金器饰”. In Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo, ed., Kaogu qiuzhi ji, ‘96 Kaogu Yanjiusuo zhongqingnian xueshu taolunhui wenji 考古求知集，‘96 考古研究所中青年學術討論會文集. Beijing: Zhongguo Shehui Kexue Chubanshe, 1997, pp. 353–60. Graham 1934 David C. Graham. “A Preliminary Report of the Hanchow Excavation”. Journal of the West China Border Research Society 6 (1934), pp. 114–31. Gu Jiegang 1981 Gu Jiegang 顧頡刚. Lun Ba Shu yu Zhongyuan de guanxi 論巴蜀與中原的關係. Chengdu: Sichuan Renmin Chubanshe, 1981. 276 Hayashi Minao 1997 Hayashi Minao 林巳奈夫. Zhongguo guyu yanjiu 中國古玉研究. Taibei: Yishu Tushu Gongsi, 1997. Chinese translation by Yang Meili 楊美莉 of Hayashi Minao. Chūgoku kogyoku no kenkyū 中国古玉の研究. Tokyo: Yoshikawa Kōbunkan, 1991 Hong Kong 1994 Jiangxi Sheng Bowuguan, Shanghai Bowuguan 上海博物館, eds., Changjiang zhongyou qingtong wangguo—Jiangxi Xin'gan chutu qingtong yishu (The Bronze Kingdom in Mid Yangtze—The Art of Bronze Unearthed at Xingan in Jiangxi). Hong Kong: Liangmu Chubanshe, 1994. Hong Kong 1997 State Bureau of Cultural Relics Catalogue Editorial Board, Guobao— Zhongguo lishi wenwu jinghua zhan 國寶—中國歷史文物精華展(National Treasures— Gems of China’s Cultural Relics). Hong Kong: the Provisional Urban Council of Hong Kong, 1997 (in Chinese and English). Hong Kong 2007 Xianggang Wenhua Bowuguan 香港文化博物館 (Hong Kong Heritage Museum). Sanxing shanshuo jinsha liucai—shenmi de gu Shu wenmin 三星閃爍金沙流采—神秘的古蜀文明 (Splendour and Mystery of Ancient Shu—Cultural Relics from Sanxingdui and Jinsha). Hong Kong: Kangle Ji Wenhua Shiwushu (Leisure and Cultural Services Department), 2007 (in Chinese and English). Huang Jianhua 2001a Huang Jianhua 黃建華. “Sanxingdui fushi wenhua tantao 三星堆服飾文化探討”. Sichuan wenwu 2001.2, pp. 3–12. Huang Jianhua 2001b _____. “Sanxingdui taiyang chongbai tantao 三星堆太陽崇拜探討”. Zhonghua wenhua luntan 2001.2, pp. 42–7. Huang Jianhua 2002 _____. Gu Shu de huihuang—Sanxingdui wenhua yu gu Shu wenming de xiaxiang 古蜀的輝煌—三星堆文化與古蜀文明的遐想. Chengdu: Ba Shu Shushe, 2002. Huang Jiaxiang 1990 Huang Jiaxiang 黃家祥. “‘Guanghan Sanxingdui yizhi’ de chubu fenxi ‘廣漢三星堆遺址’的初步分析”. Kaogu 1990.11, pp. 1030–6. Huang Shengzhang 1996 Huang Shengzhang 黃盛璋. “Lun Zhongguo zaoqi (tong tie yiwai) de jinshu gongyi 論中國早期（銅鐵以外）的金屬工藝”. Kaogu xuebao 1996.2, pp. 143–64. Huang Xuanpei 2000 Huang Xuanpei 黃宣佩, ed., Fuquanshan—xinshiqi shidai yizhi fajue baogao 福泉山—新石器時代遺址發掘報告. Beijing: Wenwu Chubanshe, 2000. Huaxia kaogu 2002.3 Chengdu Shi Wenwu Kaogu Gongzuodui 成都市文物考古工作隊, Chengdu Shi Wenwu Kaogu Yanjiusuo. “Chengdu Shi xijiao Jinshacun Longshan 277 shidai yizhi shijue 成都市西郊金沙村龍山時代遺址試掘”. Huaxia kaogu 2002.3, pp. 3–6, 72. Hu Houxuan 1974a Hu Houxuan 胡厚宣. “Zhongguo nuli shehui de renxun he renji (shangpian) 中國奴隸社會的人殉和人祭（上篇）”. Wenwu 1974.7, pp. 74–84. Hu Houxuan 1974b _____. “Zhongguo nuli shehui de renxun he renji (xiapian) 中國奴隸社會的人殉和人祭（下篇）”. Wenwu 1974.8, pp. 56–67, 72. Hunan kaogu jikan 1984 Yueyang Shi Wenwu Guanlisuo 岳陽市文物管理所. “Yueyang shi xin chutu de Shang Zhou qingtongqi 岳陽市新出土的商周青銅器”. Hunan kaogu jikan 2 (1984), pp. 26–7. Hunan kaogu jikan 1986 Hunan Sheng Bowuguan, Changsha Shi Wenwu Gongzuodui 長沙市文物工作隊. “Xinshao, Liuyang, Zhuzhou, Zixing chutu Shang Zhou qingtongqi 新邵、瀏陽、株洲、資興出土商周青銅器”. Hunan kaogu jikan 3 (1986), pp. 27–30. Huo Wei 1989 Huo Wei 霍巍. “Guanghan Sanxingdui qingtong wenhua yu gudai Xiya wenming 廣漢三星堆青銅文化與古代西亞文明”. Sichuan wenwu 1989 (special issue on the Guanghan Sanxingdui site), pp. 37–43. Jiang et al. 1997 Jiang Zhanghua 江章華, Yan Jinsong 顏勁松, Li Mingbin 李明斌. “Chengdu pingyuan de zaoqi guchengzhiqun—Baodun wenhua chulun 成都平原的早期古城址羣—寶墩文化初論”. Zhonghua wenhua luntan 1997.4, pp. 8–14. Jiang et al. 2001a Jiang Zhanghua, Wang Yi 王毅, Zhang Qing 張擎. “Chengdu pingyuan zaoqi chengzhi jiqi kaoguxue wenhua chulun 成都平原早期城址及其考古學文化初論”. In Su Bai 2001, pp. 699–721. Jiang et al. 2001b Jiang Zhanghua, Yin Jianhua 尹建華, Xie Hui 謝輝. “Ba Shu wenhuaqu de xingcheng jiqi jinyibu qutong fazhan de lishi guocheng 巴蜀文化區的形成及其進一步趨同發展的歷史過程”. Zhonghua wenhua luntan 2001.4, pp. 54–8. Jiang et al. 2002 _____. “Chengdu pingyuan xian Qin wenhua chulun 成都平原先秦文化初論”. Kaogu xuebao 2002.1, pp. 1–22. Jiang & Chen 2001 Jiang Cheng 蔣成, Chen Jian 陳劍. “Minjiang shangyou kaogu xinfaxian shuxi 岷江上游考古新發現述析”. Zhonghua wenhua luntan, 2001.3, pp. 27– 31. Jiang & Li 1998 Jiang Cheng, Li Mingbin. “Sichuan Wenjiang xian Yufu cun yizhi fenxi 四川溫江縣魚鳧村遺址分析”. Dongnan wenhua, 1998.4, pp. 15–29. 278 Jiang & Li 2002 Jiang Zhanghua, Li Mingbin. Guguo xunzong, Sanxingdui wenhua de xingqi jiqi yingxiang 古國尋踪，三星堆文化的興起及其影響. Chengdu: Ba Shu Shushe, 2002. Jiang Chunfang 1986 Jiang Chunfang 姜椿芳 ed,. Zhongguo dabaike quanshu kaoguxue 中國大百科全書考古學. Beijing: Zhongguo Dabaike Quanshu Chubanshe, 1986. Jiang Zhanghua 1992 Jiang Zhanghua. “Ba Shu liuyexingjian yuanyuan shitan 巴蜀柳葉形劍渊源試探”. Sichuan wenwu 1992 (special issue on studies of ancient Shu culture at Sanxingdui), pp. 81–4. Jiang Zhanghua 1998a _____. “Shilun Sanxingdui wenhua Shi’erqiao wenhua yu zhoulin wenhua de guanxi 試論三星堆文化十二橋文化與周鄰文化的關係”. Chengdu wenwu 1998.1, pp. 4–11. Jiang Zhanghua 1998b _____. “Chengdu Shi’erqiao yizhi de wenhua xingzhi ji fenqi yanjiu 成都十二橋遺址的文化性質及分期研究”. In Sichuan Daxue Kaogu Zhuanye 四川大學考古專業, ed. Sichuan Daxue Kaogu Zhuanye chuangjian sanshiwu zhounian jinian wenji 四川大學考古專業創建三十五周年紀念文集, pp. 146–64. Chengdu: Sichuan Daxue Chubanshe, 1998. Jiang Zhanghua 2005 _____. “Cong kaogu cailiao kan Sichuan pendi zai Zhonghua wenming xingcheng yu fazhan guocheng zhong de diwei 從考古材料看四川盆地在中華文明形成與發展過程中的地位”. Zhonghua wenhua luntan 2005.4, pp. 16–20. Jin et al. 1998 Jin Zhengyao 金正耀, Y. Hirao 平尾良光, Yang Xizhang 楊錫璋, W. T. Chase 齊思, H. Mabuchi 馬渊久夫, K. Miwa 三輪嘉六. “Zhongguo lianghe liuyu qingtong wenming zhijian de lianxi—yi chutu Shang qingtongqi de qian tongweisu bizhi yanjiu jieguo wei kaocha zhongxin 中國两河流域青銅文明之間的聯係—以出土商青銅器的鉛同位素比值研究結果為考察中心.” Beijing 1998c, pp. 425–33. Kaogu 1959.8 Sichuan Sheng Bowuguan 四川省博物館. “Sichuan Xinfan Shuiguanyin yizhi fajue jianbao 四川新繁水觀音遺址發掘簡報”. Kaogu 1959.8, pp. 404–10. Kaogu 1973.1 Hebei Sheng Bowuguan Wenwu Guanlichu 河北省博物館文物管理處. “Hebei Gaocheng xian Shang dai yizhi he muzang de diaocha 河北藁城縣商代遺址和墓葬的調查”. Kaogu 1973.1, pp. 25–9. Kaogu 1976.4 Zhongguo Kexueyuan Kaogu Yanjiusuo Erlitou Gongzuodui 中國科學院考古研究所二里頭工作隊. “Yanshi Erlitou yizhi xinfaxian de tongqi he yuqi 偃師二里頭遺址新發現的銅器和玉器”. Kaogu 1976.4, pp. 259–63. 279 Kaogu 1978.4 Yanshi Xian Wenhuaguan 偃師縣文化館. “Erlitou yizhi chutu de tongqi he yuqi 二里頭遺址出土的銅器和玉器”. Kaogu 1978.4, p. 270. Kaogu 1979.1 Henan Sheng Bowuguan 河南省博物館, Lingbao Xian Wenhuaguan 靈寶縣文化館. “Henan Lingbao chutu yipi Shang dai qingtongqi 河南靈寶出土一批商代青銅器”. Kaogu 1979.1, pp. 20–2. Kaogu 1983.3 Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo Erlitou Dui 中國社會科學院考古研究所二里頭隊. “1980 nian qiu Henan Yanshi Erlitou yizhi fajue jianbao 1980 年秋河南偃師二里頭遺址發掘簡報”. Kaogu 1983.3, pp. 199–205, and 219. Kaogu 1984.1 Zhongguo Kexueyuan Kaogu Yanjiusuo Erlitou Gongzuodui. “1981 nian Henan Yanshi Erlitou muzang fajue jianbao 1981 年河南偃師二里頭墓葬發掘簡報”. Kaogu 1984.1, pp. 37–40. Kaogu 1986.4 Zhongguo Kexueyuan Kaogu Yanjiusuo Erlitou Gongzuodui. “1984 nian qiu Henan Yanshi Erlitou yizhi faxian de jizuo muzang 1984 年秋河南偃師二里頭遺址發現的幾座墓葬”. Kaogu 1986.4, pp. 318–23. Kaogu 1990.4 Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo Sichuan Gongzuodui 中國社會科學院考古研究所四川工作隊. “Sichuan Mianyang shi Bianduishan xinshiqi shidai yizhi diaocha jianbao 四川綿陽市邊堆山新石器時代遺址調查簡報”. Kaogu 1990.4, pp. 307–13. Kaogu 1992.4 Zhongguo Kexueyuan Kaogu Yanjiusuo Erlitou Gongzuodui. “1987 nian Yanshi Erlitou yizhi muzang fajue jianbao 1987 年偃師二里頭遺址墓葬發掘簡報”. Kaogu 1992.4, pp. 294–303. Kaogu 1997.1 Chengdu Shi Wenwu Kaogu Gongzuodui, Sichuan Lianhe Daxue Kaogu Jiaoyanshi 四川聯合大學考古教研室, Xinjin Xian Wenguansuo 新津縣文管所. “Sichuan Xinjin xian Baodun yizhi diaocha yu shijue 四川新津縣寶墩遺址調查與試掘”. Kaogu 1997.1, pp. 40–52. Kaogu 1998.1 Zhong Ri Lianhe Kaogu Diaochadui. “Sichuan Xinjin Xian Baodun yizhi 1996 nian fajue jianbao 四川新津縣寶墩遺址 1996 年發掘簡報”. Kaogu 1998.1, pp. 29–50. Kaogu 1999.7 Chengdu Shi Wenwu Kaogu Gongzuodui, Dujiangyan Shi Wenwuju 都江堰市文物局. “Sichuan Dujiangyan Mangcheng yizhi diaocha yu shijue 四川都江堰芒城遺址調查與試掘”. Kaogu 1999.7, pp. 14–27. Kaogu 2002.11 Chengdu Shi Wenwu Kaogu Gongzuodui. “Sichuan Chongzhou shi Shuanghe shiqian chengzhi shijue jianbao 四川崇州市雙河史前城址試掘簡報”. Kaogu 2002.11, pp. 3–19. 280 Kaogu 2004.10 Sichuan Sheng Wenwu Kaogu Yanjiusuo Sanxingdui Yizhi Gongzuozhan 四川省文物考古研究所三星堆遺址工作站. “Sichuan Guanghan shi Sanxingdui yizhi Rensheng cun tukengmu 四川廣漢市三星堆遺址仁勝村土坑墓”. Kaogu 2004.10, pp. 14–22. Kaogu xuebao 1957.4 Sichuan Sheng Wenwu Guanli Weiyuanhui. “Chengdu Yangzishan tutai yizhi qingli baogao 成都羊子山土臺遺址清理報告”. Kaogu xuebao 1957.4, pp. 17–31. Kaogu xuebao 1981.4 Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo Anyang Gongzuodui 中國社會科學院考古研究所安陽工作隊. “Anyang Xiaotun cunbei de liangzuo Yin dai mu 安陽小屯村北的两座殷代墓”. Kaogu xuebao 1981.4, pp. 491–518. Kaogu xuebao 1987.2 Sichuan Sheng Wenwu Guanli Weiyuanhui, Sichuan Sheng Bowuguan, Guanghan Xian Wenhuaguan 廣漢縣文化館. “Guanghan Sanxingdui yizhi 廣漢三星堆遺址”. Kaogu xuebao 1987.2, pp. 227–54. Kaogu xuebao 1992.2 Hu’nan Sheng Wenwu Kaogu Yanjiusuo 湖南省文物考古研究所. “Hu’nan Shimen Zaoshi Shang dai yicun 湖南石門皂市商代遺存”. Kaogu xuebao 1992.2, pp. 185–218. Kuwayama 1981 George Kuwayama, ed. The Great Bronze Age of China: A Symposium. Los Angeles: Los Angeles County Museum of Art, 1981. Laufer 1927 Berthold Laufer. Archaic Chinese Jades Collected in China by A. W. Bahr, Now in the Field Museum of Natural History, Chicago. New York, 1927. Li et al. 1984 Li Minsheng 李敏生, Huang Suying 黃素英, Ji Lianqi 季連琪. “Yinxu jinshu qiwu chengfen de ceding baogao (2)—Yinxu xiqu tongqi he qianqi ceding 殷墟金屬器物成份的測定報告（2）—殷墟西區銅器和鉛器測定.” Kaoguxue jikan 4 (1984), pp. 328–33, 341. Li et al. 1991 Li Shaoming 李绍明, Lin Xiang 林向, Xu Nanzhou 徐南洲, eds. Ba Shu lishi, minzu, kaogu, wenhua 巴蜀歷史、民族、考古、文化. Chengdu: Ba Shu Shushe, 1991. Li et al. 1993 Li Shaoming, Lin Xiang, Zhao Dianzeng 趙殿增, ed. Sanxingdui yu Ba Shu wenhua 三星堆與巴蜀文化. Chengdu: Ba Shu Shushe, 1993. Li & Chen 2001 Li Mingbin, Chen Yunhong 陈雲洪. “Wenjiang xian Yufu cun yizhi 1999 niandu fajue 溫江縣魚鳧村遺址 1999 年度發掘”. Chengdu kaogu faxian 1999 (2001). pp. 40–53. 281 Liang-Lee & Louis 1996 Yeajen Liang–Lee and François Louis. An Index of Gold and Silver Artifacts Unearthed in the People’s Republic of China. Zürich: Museum Rietberg, 1996. Li Boqian 1997 Li Boqian 李伯谦. “Dui Sanxingdui wenhua ruogan wenti de renshi 對三星堆文化若干問題的認識”. Kaoguxue yanjiu 3 (1997), pp. 84–94. Li Mingbin 1999a Li Mingbin. “Sichuan Ya’an Shaxi yizhi taoqi ji xiangguan wenti de chubu yanjiu 四川雅安沙溪遺址陶器及相關問題的初步研究”. Kaogu 1999.2, pp. 75– 81. Li Mingbin 1999b _____. “Guanghan Yueliangwan yicun shixi 廣漢月亮灣遺存試析”. Huaxia kaogu 1999.1, pp. 26–35, 59. Li Mingbin 2001a _____. “Luelun Xinjin Baodun yizhi de fenqi 略論新津寶墩遺址的分期”. Huaxia kaogu 2001.1, pp. 37–40. Li Mingbin 2001b _____. “Shilun Yufu cun yizhi disanqi yicun 試論魚鳧村遺址第三期遺存”. Kaogu yu wenwu 2001.1, pp. 40–1, 48. Li Mingbin 2001c _____. “Guanyu Sanxingdui yizhi diyiqi wenhua yicun de jige wenti 關於三星堆遺址第一期文化遺存的幾個問題”. Chengdu wenwu 2001.2, pp. 56–61. Li Mingbin 2003 _____. “Yangzishan tutai zai kao 羊子山土臺再考” . Gudai wenming 2 (2003), pp. 241–51. Lin Mingjun 1942 Lin Mingjun 林名均. “Guanghan gudai yiwu zhi faxian jiqi fajue 廣漢古代遺物之發現及其發掘”. Shuowen yuekan (Chongqing edition), 3.7 (1942), pp. 93–101. Lin Xiang 1985 Lin Xiang. “Zhouyuan buci zhong de ‘Shu’ 周原卜辭中的‘蜀’”. Kaogu yu wenwu, 1985.6, pp. 66–74. Re-published with minor changes in Lin Xiang 1995, pp. 76–95. Lin Xiang 1987 _____. “Shu jiu tanyuan—Ba Shu de ‘Samanjiao wenhua’ yanjiu zhiyi 蜀酒探源—巴蜀的‘薩满教文化’研究之一”. Nanfang minzu kaogu, 1 (1987), pp. 73–86. Re-published with minor changes in Lin Xiang 1995, pp. 96–119. Lin Xiang 1988 _____. “Yangzishan jianzhu yizhi xinkao 羊子山建築遺址新考”. Sichuan wenwu 1988.5, pp. 3–8. Lin Xiang 1989 _____. “Sanxingdui yizhi yu Yin Shang de xitu—jian shi Yinxu buci zhong de ‘Shu’ de dili weizhi 三星堆遺址與殷商的西土—兼釋殷墟卜辭中的‘蜀’ 的地理位置”. Sichuan wenwu 1989 (special issue on the Guanghan Sanxingdui site), pp. 282 23–30. Re-published with minor changes and a different title in Lin Xiang 1995, pp. 57– 75. Lin Xiang 1991 _____. “Shuo dun 說盾”. Chengdu wenwu 1991.1, pp. 1–10. Republished with minor changes and under different titles in Sichuan wenwu 1992 (special issue on studies of ancient Shu culture at Sanxingdui) (pp. 18–25), and in Lin Xiang 1995 (pp. 147–66). Lin Xiang 1995 _____. Ba Shu wenhua xinlun 巴蜀文化新論. Chengdu: Chengdu Chubanshe, 1995. Lin Xiang 2001 _____. “Shixi Baodun wenhua guchengzhiqun 試析寶墩文化古城址羣”. Chengdu wenwu 2001.4, pp. 4–7. Li Song 1993 Li Song 李松. “Guanghan qingtong renwu qundiao de meishushi jiazhi 廣漢青銅人物羣雕的美術史價值”. In Li et al. 1993, pp. 101–6. Liu 2000 Liu Yang. “The False Face of an Ancient Society”. Oriental Art, 48.3 (2002), pp. 2–16. Liu & Capon 2000 Liu Yang and Edmund Capon. Masks of Mystery: Ancient Chinese Bronzes from Sanxingdui. Sydney: Art Gallery of New South Wales, 2000. Liu & Rong 2001 Liu Yumao 劉雨茂, Rong Yuanda 榮遠大. “Chengdu shi xijiao Huachengcun yizhi 1999 niandu fajue baogao 成都市西郊化成村遺址 1999 年度發掘報告”. Chengdu kaogu faxian 1999 (2001), pp. 127–45. Liu Shi’e 2001 Liu Shi’e 劉士莪. Laoniupo 老牛坡. Xi’an: Shaanxi Renmin Chubanshe, 2001. Liu Xingshi 1998 Liu Xingshi 劉興詩. “Chengdu pingyuan guchengqun xingfei yu gu qihou wenti 成都平原古城羣興廢與古氣候問題”. Sichuan wenwu 1998.4, pp. 34–7. Li Xiandeng 1994 Li Xiandeng 李先登. “Guanghan Sanxingdui qiwukeng zhi zai yanjiu 廣漢三星堆器物坑之再研究”. Zhongguo Lishi Bowuguan guankan 23 (1994.2), pp. 41–9. Reprinted in Li Xiandeng, Xia Shang Zhou qingtong wenming tanyan 夏商周青銅文明探源, pp. 239–49. Beijing: Kexue Chubanshe, 2001. Li Xiaocen 1993 Li Xiaocen 李晓岑. “Shang Zhou Zhongyuan qingtongqi kuangliao laiyuan de zai yanjiu 商周中原青銅器礦料來源的再研究.” Ziran kexue shi yanjiu 12 (1993), pp. 264–7. 283 Li Xueqin 1989 Li Xueqin 李學勤. “Shang wenhua zenyang chuanru Sichuan 商文化怎樣傳入四川”. Zhongguo wenwu bao, July 21, 1989. Reprinted in Li Xueqin 1997a, pp. 228–32. Li Xueqin 1991 _____. “Lun Erlitou wenhua de taotiewen tongpaishi 論二里頭文化的饕餮纹铜牌饰”. Zhongguo wenwu bao, October 20, 1991. Reprinted in Li Xueqin 1997a, pp. 149–52. Li Xueqin 1992 _____. “Ji Xianggang Dawan xinchu yazhang ji youguan wenti 記香港大灣新出牙璋及有關問題”. Nanfang wenwu 1992.1, pp. 25–9, 18. Reprinted in Li Xueqin 1997a, pp. 125–41. Li Xueqin 1993 _____. “Sanxingdui taotiewen de fenxi 三星堆饕餮纹的分析”. In Li et al. 1993, pp. 76–80. Reprinted in Li Xueqin 1997a, pp. 232–8. Li Xueqin 1997a _____. Zouchu yigu shidai 走出疑古時代 (revised ed.). Shenyang: Liaoning Daxue Chubanshe, 1997. Li Xueqin 1997b _____. “Cong yijian xincailiao kan Guanghan tongpaishi 從一件新材料看廣漢铜牌饰”. Zhongguo wenwu bao, November 30, 1997. Li Xueqin 1998 _____. “Lun Guanghan ximen wai chutu de Shang dai qingtong zun 論廣漢西門外出土的商代青銅器”. In Beijing 1998c, pp. 307–9. Li Xueqin 1999 _____. “Changjiang liuyu Shang dai qingtong wenhua 長江流域商代青銅文化”. In Li Jinyun 李縉雲, ed. Li Xueqin xueshu wenhua suibi 李學勤學術文化隨筆, pp. 384–93. Beijing: Zhongguo Qingnian Chubanshe, 1999. Loewe 1993 Michael Loewe, ed. Early Chinese Texts: A Bibliographical Guide. Berkeley: The Society for the Study of Early China and the Institute of East Asian Studies, University of California, Berkeley, 1993. Louis 2003 François Louis. “L’or au Sichuan à l’âge du bronze”. In Paris 2003, pp. 157– 65. Lu & Hu 1988 Lu Liancheng 盧連成, Hu Zhisheng 胡智生. Baoji Yuguo mudi 寶雞魚國墓地. Beijing: Wenwu Chubanshe, 1988. Luo Erhu 1988 Luo Erhu 羅二虎. “Chengdu diqu bujia de chubu yanjiu 成都地區卜甲的初步研究.” Kaogu 1988.12, pp. 1122–9. Ma Jixian 1992 Ma Jixian 馬繼賢. “Guanghan Yueliangwan yizhi fajue zhuiji 廣漢月亮灣遺址發掘追記”. Nanfang minzu kaogu 5 (1992), pp. 310–23. 284 Ma Xinxin 1993 Ma Xinxin 馬辛辛. “Daba shanmai yu chuanbei shiqian wenhua de tantao 大巴山脈與川北史前文化的探討”. Sichuan wenwu 1993.5, pp. 8–14. Meng Huaping 1997 Meng Huaping 孟華平. Changjiang zhongyou shiqian wenhua jiegou 長江中游史前文化結構. Wuhan: Changjiang Wenyi Chubanshe, 1997. Mifune Haruhisa 2002 Mifune Haruhisa 三船溫尚. “Sanxingdui de qingtong zhuzao jishu 三星堆的青銅鑄造技術”. In Nishie Kiyotaka 2002, pp. 55–89, tr. into Chinese by Zhang Shenglan 張勝蘭. Mou & Yun 1992 Mou Yongkang 牟永抗, Yun Xizheng 雲希正, ed. Zhongguo yuqi quanji 1, Yuanshi shehui 中國玉器全集 1，原始社會. Shijiazhuang: Hebei Meishu Chubanshe, 1992. Nanba Junko 1990 Nanpa Junko 難波纯子. “Inkyo zenhanki no seidô iki no hennen to ryǔha no ninshiki 殷墟前半期の青銅彝器の編年と流派の認識”. Shirin 73.6 (1990), pp. 1–43. Nanba Junko 2002 _____. “Sanxingdui chutu qingtongqi shoumianwen de laiyuan 三星堆出土青銅器兽面纹的來源”. In Nishie Kiyotaka 2002, pp. 127–57, tr. into Chinese by Han Guohe 韩國河 and Hua Guoqiang 華國強. Nanfang minzu kaogu 1987 Sichuan Daxue Bowuguan, Chengdu Shi Bowuguan 成都市博物館. “Chengdu Zhihuijie Zhou dai yizhi fajue baogao 成都指揮街周代遺址發掘報告”. Nanfang minzu kaogu 1 (1987), pp. 171–210. Nanfang minzu kaogu 1990 Sichuan Sheng Wenwu Guanli Weiyuanhui, Sichuan Sheng Wenwu Kaogu Yanjiusuo, Sichuan Sheng Ya’an Diqu Wenwu Guanlisuo 四川省雅安地區文物管理所. “Ya’an Shaxi yizhi fajue ji diaocha baogao 雅安沙溪遺址發掘及調查報告”. Nanfang minzu kaogu 3 (1990), pp. 293–339. Nanfang minzu kaogu 1992a Sichuan Sheng Wenwu Kaogu Yanjiusuo Sanxingdui Gongzuozhan 四川省文物考古研究所三星堆工作站, Sichuan Sheng Guanghan Shi Wenguansuo 四川省廣漢市文管所, Shifang Xian Wenguansuo 什邡縣文管所. “Sichuan Guanghan, Shifang Shang Zhou yizhi diaocha baogao 四川廣漢、什邡商周遺址調查報告”. Nanfang minzu kaogu 5 (1992), pp. 295–309. Nanfang minzu kaogu 1992b Beijing Daxue Kaoguxi, Hubei Sheng Wenwu Kaogu Yanjiusuo, Hubei Jingzhou Diqu Bowuguan Shijiahe Kaogudui 北京大學考古系、湖北省文物考古研究所、湖北荊州地區博物館石家河考古隊. “Shijiahe yizhi diaocha baogao 石家河遺址調查報告”. Nanfang minzu kaogu 5 (1992), pp. 213–94. 285 Nishie Kiyotaka 2002 Nishie Kiyotaka 西江清髙, ed. Fusang yu Ruomu—Riben xuezhe dui Sanxingdui wenming de xinrenshi 扶桑與若木—日本學者對三星堆文明的新認識. Chengdu: Ba Shu Shushe, 2002. Okamura Hidenori 1998 Okamura Hidenori 岡村秀典. “Sanseitai 三星堆”. In Tokyo 1998, pp. 200–7. Chinese tr. in Nishie Kiyotaka 2002, pp. 36–51. Paris 2003 Chine, l’énigme de l’homme de bronze: Archéologie du Sichuan. Paris: Paris Musées, 2003. Peng Jinhua 1987 Peng Jinhua 彭錦華. “Shashi jinjiao chutu de Shang dai daxing tongzun 沙市近郊出土的商代大型铜尊”. Jianghan kaogu 1987.4, pp. 89–93. Pope et al. 1967 John Alexander Pope, Rutherford John Gettens, James Cahill, and Noel Barnard. The Freer Chinese Bronzes, Volume I, Catalogue. Washington, D.C.: Freer Gallery of Art, 1967. Qian Yuzhi 1992 Qian Yuzhi 錢玉趾. “Sanxingdui qingtong lirenxiang kao 三星堆青銅立人像考”. Sichuan wenwu (special issue on Sanxingdui studies), pp. 50–5. Qi Dongfang 2000 Qi Dongfang 齊東方. “Zhongguo zaoqi jinyinqi yanjiu 中國早期金銀器研究”. In Beijing Daxue Zhongguo Chuantong Wenhua Yanjiuzhongxin 北京大學中國傳統文化研究中心, ed., Wenhua de kuizeng—Hanxue yanjiu guoji huiyi lunwenji, kaoguxue juan 文化的饋贈—漢學研究國際會議論文集，考古學卷, pp. 222–39. Beijing: Beijing Daxue Chubanshe, 2000. Rawson 1992 Jessica Rawson. “Shang and Western Zhou Designs in Jade and Bronze”. In Taibei 1992, pp. 73–105. Rawson 1993 _____. “Contact between Southern China and Henan during the Shang period”. Transactions of the Oriental Ceramic Society, 57 (1992–93), pp. 1–24. Rawson 1995 _____. Chinese Jade from the Neolithic to the Qing. London: British Museum Press, 1995. Rawson 1996 _____, ed. Mysteries of Ancient China. London: British Museum, 1996. Sage 1992 Steven F. Sage. Ancient Sichuan and the Unification of China. Albany: SUNY Press, 1992. Shen & Huang 1986 Shen Zhongchang 沈仲常, Huang Jiaxiang. “Guanyu Guanghan tukeng chutu shibi de renshi 關於廣漢土坑出土石璧的認識”. Chengdu wenwu 1986.4, pp. 1–5. 286 Shi Jinsong 1998a Shi Jinsong 施劲松. “Lun dai hushiren muti de Shang Zhou qingtongqi 論帶虎食人母題的商周青銅器”. Kaogu 1998.3, pp. 56–63. Shi Jinsong 1998b _____. “Lun woguo nanfang chutu de Shang dai qingtong dakouzun 論我國南方出土的商代青銅大口尊”. Wenwu 1998.10, pp. 47–54. Shi Zhilian 1972 Shi Zhilian 石志廉. “Tantan longhu zun de jige wenti 談談龍虎尊的幾個問題”. Wenwu 1972.11, pp. 64–6. Song Zhimin 1983 Song Zhimin 宋治民. “Guanyu Shu wenhua de jige wenti 關於蜀文化的幾個問題”. Kaogu yu wenwu 1983.2, pp. 71–80. Song Zhimin 1991 _____. “Cong Sanxingdui de xinfaxian kan zaoqi Shu wenhua 從三星堆的新發現看早期蜀文化”. In Li et al. 1991, pp. 207–23. Song Zhimin 1998 _____. Shu wenhua yu Ba wenhua 蜀文化與巴文化. Chengdu: Sichuan Daxue Chubanshe, 1998. Song Zhimin 2000 _____. “Shilun Sichuan Wenjiang Yufu cun yizhi, Xinjin Baodun yizhi he Pi Xian Gucheng yizhi 試論四川溫江魚鳧村遺址，新津寶墩遺址和郫縣古城遺址”. Sichuan wenwu 2000.2, pp. 9–18. Strommenger & Hirmer 1964 Eva Strommenger and Max Hirmer. 5000 Years of the Art of Mesopotamia. New York: Abrams, 1964. Su et al. 1995 Su Rongyu 蘇榮譽, Hua Jueming 華覺明, Li Kemin 李克敏, Lu Benshan 盧本珊. Zhongguo shanggu jinshu jishu 中國上古金屬技術. Ji’nan: Shandong Kexue Jishu Chubanshe, 1995. Su Bai 2001 Su Bai 宿白, ed., Su Bingqi yu dangdai Zhongguo kaoguxue 蘇秉琦與當代中國考古學. Beijing: Kexue Chubanshe, 2001. Sun Hua 1992 Sun Hua 孫華. “Shilun Guanghan Sanxingdui yizhi de fenqi 試論廣漢三星堆遺址的分期”. Nanfang minzu kaogu 5 (1992), pp. 10–24. Reprinted in Sun Hua 2000, pp. 49–67. Sun Hua 1993a _____. “Guanyu Sanxingdui qiwukeng ruogan wenti de bianzheng 關於三星堆器物坑若干問題的辯證”. Sichuan wenwu 1993.4, pp. 3–11. Sun Hua 1993b _____. “Guanyu Sanxingdui qiwukeng ruogan wenti de bianzheng (xu) 關於三星堆器物坑若干問題的辯證（續）”. Sichuan wenwu 1993.5, pp. 3–7. Sun Hua 1993c _____. “Sanxingdui qiwukeng de niandai ji xingzhi fenxi 三星堆器物坑的年代及性質分析”. Wenwu 1993.11, pp. 71–6. 287 Sun Hua 1993d _____. “Tuhaoguoqi de yuanyuan 凸好郭器的渊源”. Zhongguo wenwu bao, 40 (1993). Sun Hua 1993e _____. “Shilun Guanghan Sanxingdui yizhi de fenqi 試論廣漢三星堆遺址的分期”. Nanfang minzu kaogu 5 (1993), pp. 10–24. Also in Sun Hua 2000, pp. 49– 67. Sun Hua 1993f _____. “Yangzishan tutai kao 羊子山土臺考.” Sichuan wenwu 1993.1, pp. 3–7. Sun Hua 1996 _____. “Chengdu Shi’erqiao yizhiqun fenqi chulun 成都十二橋遺址群分期初論.” In Beijing 1996a, pp. 123–44. Also in Sun Hua 2000, pp. 68–88. Sun Hua 2000 _____. Sichuan Pendi de qingtong shidai 四川盆地的青銅時代. Beijing: Kexue Chubanshe, 2000. Sun Hua 2001 _____. “Chengdu pingyuan de xian Qin wenhua 成都平原的先秦文化”. In Su Bai 2001, pp. 470–94. Also in Sun Hua 2000, pp. 91–115. Sun Hua 2003 _____. “The Zhuwajie Bronzes,” Journal of East Asian Archaeology, 5.1–4 (2003), pp. 277–336. Su Yongjiang 1996 Su Yongjiang 蘇永江. “Guanghan Sanxingdui chutu yuqi kaogu dizhixue yanjiu 廣漢三星堆出土玉器考古地質學研究”. In Beijing 1996a, pp. 79–90. Taibei 1992 Guoli Gugong Bowuyuan Bianji Weiyuanhui 國立故宮博物院編輯委員會, ed. Zhongguo yishu wenwu taolunhui lunwenji, qiwu (shang) 中國藝術文物討論會論文集，器物（上）(Proceedings, International Colloquium on Chinese Art History, 1991, Antiquities, Part 1). Taibei: National Palace Museum, 1992. Taibei 1999 Guoli Gugong Bowuyuan 國立故宮博物院. Sanxingdui chuanqi, Huaxia gu wenming de tansuo 三星堆傳奇，華夏古文明的探索. Taibei: Taipingyang Wenhua Jijinhui, 1999. Tang et al. 1980 Tang Jinyu 唐金裕, Wang Shouzhi 王壽芝, Guo Changjiang 郭長江. “Shaanxi Sheng Chenggu Xian chutu Yin Shang tongqi zhengli jianbao 陝西省城固縣出土殷商銅器整理簡報”. Kaogu 1980.3, pp. 211–8. Tang Jigen 1999 Tang Jigen 唐繼根. “Zhong Shang wenhua yanjiu 中商文化研究”. Kaogu xuebao 1999.4, pp. 393–420. Tokyo 1986 Kôga bunmei ten. Tokyo: The Chunichi Shimbun, 1986. 288 Tokyo 1998 Asashi Shimbun 朝日新聞社, ed. Sanseitai—Chûgoku 5000 nen no nazo, kyôi no kamen ôkoku 三星堆—中国 5000 年の謎*驚異の仮面王国. Tokyo: Asahi Shimbunsha, 1998. Tokyo 2000 Chengdu Shi Wenwu Kaogu Yanjiusuo, Sichuan Daxue Lishixi Kaogu Jiaoyanshi 四川大學歷史系考古教研室, Zaodaotian Daxue Changjiang Liuyu Wenhua Yanjiusuo 早稻田大學長江流域文化研究所. Baodun yizhi: Xinjin Baodun yizhi fajue he yanjiu 寶墩遺址：新津寶墩遺址發掘和研究. Tokyo: ARP, 2000. Wang 2003 Wang Yi. “Prehistoric Walled Settlements in the Chengdu Plain”. Journal of East Asian Archaeology, 5.1–4 (2003), pp. 109–48. Wang et al. 1997 Wang Yi, Jiang Zhanghua, Jiang Cheng, Lu Ding 盧丁. “Zhongguo Changjiang wenming qiyuan yanjiu de xinchengguo—Chengdu pingyuan shiqian chengzhiqun faxianji 中國長江文明起源研究的新成果—成都平原史前城址羣發現記”. Chengdu wenwu 1997.2, pp. 23–9. Wang et al. 1999 Wang Yi, Jiang Cheng, Jiang Zhanghua. “Chengdu diqu jinnian kaogu zongshu 成都地區近年考古綜述.” Sichuan wenwu 1999.3, pp. 3–12. Wang & Jiang 1958 Wang Jiayou 王家佑, Jiang Dianchao 江甸潮. “Sichuan Xinfan Guanghan guyizhi diaochaji 四川新繁廣漢古遺址調查記”. Kaogu tongxun 1958.8, pp. 27–31. Wang & Jiang 2000 Wang Yi, Jiang Cheng. “Chengdu pingyuan zaoqi chengzhi de faxian yu chubu yanjiu 成都平原早期城址的發現與初步研究”. In Yan Wenming 嚴文明, Yasuda Yoshinori 安田喜宪, eds. Daozuo taoqi he dushi de qiyuan 稻作、陶器和都市的起源, pp. 143–65. Beijing: Wenwu Chubanshe, 2000. Wang & Long 1987 Wang Wenjian 王文建, Long Xibin 龍西斌. “Shimen xian Shang shiqi yicun diaocha—Baota yizhi yu Weigang muzang 石門縣商時期遺存調查—寶塔遺址與桅崗墓葬”. Hunan kaogu jikan 4 (1987), pp. 11–8. Wang & Sun 1992 Wang Weilin 王炜林, Sun Bingjun 孫秉君. “Hanshui shangyou Ba Shu wenhua de zongji 漢水上游巴蜀文化的蹤跡.” In Zhongguo Kaogu Xuehui 中國考古學會, ed. Zhongguo Kaogu Xuehui diqici nianhui lunwenji 1989 中國考古學會第七次年會論文集 1989, pp. 236–48. Beijing: Wenwu Chubanshe, 1992. Wang & Sun 1999 Wang Yi, Sun Hua. “Baoduncun wenhua de chubu renshi 寶墩村文化的初步認識”. Kaogu 1999.8, pp. 60–73. Wang & Wang 1993 Wang Xu 王 , Wang Yarong 王亞蓉. “Guanghan chutu qingtong lirenxiang fushi guanjian 廣漢出土青銅立人像服飾管見”. Wenwu 1993.9, pp. 60–8. 289 Wang & Ye 1993 Wang Renxiang 王仁湘, Ye Maolin 葉茂林. “Sichuan pendi beiyuan xinshiqi shidai kaogu xinshouhuo 四川盆地北緣新石器時代考古新收獲”. In Li et al. 1993, pp. 257–65. Wang & Zhang 1999 Wang Yi, Zhang Qing. “Sanxingdui wenhua yanjiu 三星堆文化研究”. Sichuan wenwu 1999.3, pp. 13–22. Wang Congli 1993 Wang Congli 王從禮. “Ji Jiangling Cenhe Miaoxing Bagutai chutu Shang dai tongzun 記江陵岑河廟興八姑臺出土商代铜尊”. Wenwu 1993.8, pp. 67–8, 31. Wang Cunwu 1998 Wang Cunwu 王純五. “Chengdu pingyuan gucheng yizhi yu guhedao de guanxi 成都平原古城遺址與古河道的關係”. Chengdu wenwu 1998.4, pp. 39–41. Wang Hongming 1985 Wang Hongming 王洪明. “Shangdong sheng Haiyang xian shiqian yizhi diaocha 山東省海陽縣史前遺址調查”. Kaogu 1985.12, pp. 1057–67. Wang Jiayou 1961 Wang Jiayou. “Ji Sichuan Peng Xian Zhuwajie chutu de tongqi 記四川彭縣竹瓦街出土的銅器”. Wenwu 1961.11, pp. 28–31. Wang Jinqiu 2001 Wang Jinqiu 王金秋. “Tan Erlitou yizhi chutu de tongpaishi 談二里頭遺址出土的铜牌饰”. Zhongyuan wenwu 2001.3, pp. 18–20 and 27. Wang Shouzhi 1988 Wang Shouzhi 王壽芝. “Shaanxi Chenggu chutu de Shang dai qingtongqi 陝西城固出土的商代青銅器”, Wenbo 1988.6, pp. 3–9. Wang Yi 1988 Wang Yi. “Chengdu shi Shu wenhua yizhi de fanxian ji yiyi 成都市蜀文化遺址的發現及意義”. Chengdu wenwu 1988.1, pp. 10–6. Wang Yi 1991 _____. “Chengdu shi Ba Shu wenhua yizhi de xinfaxian 成都市巴蜀文化遺址的新發現.” In Li et al. 1991, pp. 295–309. Wenwu 1961.11 Sichuan Daxue Lishixi Kaoguxue Jiaoyanzu 四川大學歷史系考古學教研組. “Guanghan Zhongxing gongshe gu yizhi diaocha jianbao 廣漢中興公社古遺址調查簡報”. Wenwu 1961.11, pp. 22–7. Wenwu 1972.1 Hunan Sheng Bowuguan. “Hunan sheng gongnongbing qunzhong reai zuguo wenhua yichan 湖南省工農兵羣眾熱愛祖國文化遺產”, pp. 6–7. 290 Wenwu 1983.3 Henan Sheng Wenwu Yanjiusuo, Zhengzhou Shi Bowuguan 鄭州市博物館. “Zhengzhou xinfaxian Shang dai jiaocang qingtongqi 鄭州新發現商代窖藏青銅器”. Wenwu 1983.3, pp. 49–59. Wenwu 1987.10 Sichuan Sheng Wenwu Guanli Weiyuanhui, Sichuan Sheng Wenwu Kaogu Yanjiusuo, Sichuan Sheng Guanghan Xian Wenhuaju 四川省廣漢縣文化局. “Guanghan Sanxingdui yizhi yihao jisikeng fajue jianbao 廣漢三星堆遺址一號祭祀坑發掘簡報”. Wenwu 1987.10, pp. 1–15. Wenwu 1987.12 Sichuan Sheng Wenwu Guanli Weiyuanhui, Sichuan Sheng Wenwu Kaogu Yanjiusuo, Chengdu Shi Bowuguan 成都市博物館. “Chengdu Shi’erqiao Shang dai jianzhu yizhi diyiqi fajue jianbao 成都十二橋商代建築遺址第一期發掘簡報”. Wenwu 1987.12, pp. 1–23, 37. Wenwu 1989.5 Sichuan Sheng Wenwu Guanli Weiyuanhui, Sichuan Sheng Wenwu Kaogu Yanjiusuo, Guanghan Shi Wenhuaju Wenguansuo 廣漢市文化局文管所. “Guanghan Sanxingdui yizhi erhao jisikeng fajue jianbao 廣漢三星堆遺址二號祭祀坑發掘簡報”. Wenwu 1989.5, pp. 1–20. Wenwu 1997.3 Chengdu Shi Wenwu Kaogu Gongzuodui, Xindu Xian Wenwu Guanlisuo 新都縣文物管理所. “Sichuan Xindu xian Guilinxiang Shang dai yizhi fajue jianbao 四川新都縣桂林鄉商代遺址發掘簡報”. Wenwu 1997.3, pp. 24–34. Wenwu 1998.12 Chengdu Shi Wenwu Kaogu Gongzuodui, Sichuan Lianhe Daxue Lishixi Kaogu Jiaoyanshi 四川聯合大學歷史系考古教研室, Wenjiang Xian Wenguansuo 溫江縣文管所. “Sichuan sheng Wenjiang xian Yufu cun yizhi diaocha yu shijue 四川省溫江縣魚鳧村遺址調查與試掘”. Wenwu 1998.12, pp. 38–56. Wenwu 1999.1 Chengdu Shi Wenwu Kaogu Gongzuodui, Pi Xian Bowuguan. “Sichuan Sheng Pi Xian Gucheng yizhi diaocha yu shijue 四川省郫縣古城遺址調查與試掘”. Wenwu 1999.1, pp. 32–42. Wenwu 2000.12 Anhui Sheng Wanxi Bowuguan 安徽省皖西博物館. “Anhui Liuan chutu yijian daxing Shang dai tongzun 安徽六安出土一件大型商代铜尊”. Wenwu 2000.12, pp. 65–8. Wenwu 2001.3 Chengdu Shi Wenwu Kaogu Yanjiusuo, Pi Xian Bowuguan. “Sichuan Sheng Pi Xian Gucheng yizhi 1997 nian fajue jianbao 四川省郫縣古城遺址 1997 年發掘簡報”. Wenwu 2001.3, pp. 52–68. Wenwu 2003.1 _____. “Sichuan Pi Xian Qingjiangcun yizhi fajue jianbao 四川郫縣清江村遺址發掘簡報”. Wenwu 2003.1, pp. 56–64. 291 Wenwu 2003.4 Chengdu Shi Wenwu Kaogu Gongzuodui. “Chengdu Shi Hetaocun Shang dai yizhi fajue jianbao 成都市核桃村商代遺址發掘簡報”. Wenwu 2003.4, pp. 21–5. Wenwu 2004.4 Chengdu Shi Wenwu Kaogu Yanjiusuo. “Chengdu Jinsha yizhi I qu ‘Meiyuan’ didian fajue yiqi jianbao 成都金沙遺址 I 區‘梅苑’地點發掘一期簡報”. Wenwu 2004.4, pp. 4–65. Wenwu cankao ziliao 1954.3 Xi’nan Bowuyuan Choubeichu 西南博物院籌備處. “Baocheng tielu xiuzhu gongcheng zhong faxian de wenwu jianjie 寶成鐵路修築工程中發現的文物簡介”. Wenwu cankao ziliao 1954.3, pp. 10–34 (p. 12 on field survey at the Sanxingdui site). Wilson 1990 J. Keith Wilson. “Lithic art in the Bronze Age: a jade dagger-axe”. The Bulletin of the Cleveland Museum of Art 77.1 (1990), pp. 1–35. Wu 1997 Wu Hung. “All About the Eyes: Two Groups of Sculptures from the Sanxingdui Culture.” Orientations 28.8 (1997), pp. 58–66. Wu Yi 1988 Wu Yi 吴怡. "Chengdu Fangchijie chutu shidiao renxiang ji xiangguan wenti 成都方池街出土石雕人像及相關問題”. Sichuan wenwu 1988.6, pp. 18–22. Xiao et al. 2001 Xiao Xianjin 肖天進, Ao Tianzhao, Liu Jiasheng, Bao Yuzhi 包育智. Sanxingdui faxian fajue shimo 三星堆發現發掘始末. Chengdu: Sichuan Renmin Chubanshe, 2001. Xiong Chuanxin 1981 Xiong Chuanxin 熊傳新, “Hunan xinfaxian de qingtongqi 湖南新發現的青銅器”. Wenwu ziliao congkan 5 (1981), pp. 103–5. Xu 1998 Jay Xu. “The Diamond-back Dragon of the Late Shang Period”. Orientations, May 1998, pp. 42–54. Xu 2001 _____. “Introduction Part 1: Sichuan before the Warring States Period,” “Chapter 1: Bronze at Sanxingdui.” In Bagley 2001, pp. 21–37, and 59–151. Xu 2003 _____. “Defining the archaeological cultures at Sanxingdui”. Journal of East Asian Archaeology, 5.1–4 (2003), pp. 149–90. Xu & Wang 1999 Xu Pengzhang 徐鵬章, Wang Yi. “Fangchijie guwenhua yizhi de chutu wenwu 方池街古文化遺址的出土文物”. Chengdu wenwu 1999.2, pp. 45–6. Xu Chaolong 1992a Xu Chaolong 徐朝龍. “Sanxingdui ‘jisikeng shuo’ changyi— jiantan Yufu he Duyu zhi guanxi 三星堆‘祭祀坑說’唱异—兼谈魚鳧和杜宇之關係”. Sichuan wenwu 1992.5, pp. 32–8. 292 Xu Chaolong 1992b _____. “Sanxingdui ‘jisikeng shuo’ changyi—jiantan Yufu he Tuyu zhi guanxi (xu) 三星堆‘祭祀坑說’唱异—兼谈魚鳧和杜宇之關係（續）”. Sichuan wenwu 1992.6, pp. 40–7. Xu Jie 2006 Xu Jie 許傑 (=Jay Xu). “Sichuan Guanghan Yueliangwan chutu yushiqi erze 四川廣漢月亮灣出土玉石器二則”. Sichuan wenwu 2006.5, pp. 51–7. Xu Lianggao 1991 Xu Lianggao 徐良髙. “Shang Zhou qingtongqi ‘renshou muti’ wenshi kaoshi 商周青銅器‘人兽母題’纹饰考釋”. Kaogu 1991.5, pp. 442–7, 404. Xu Lianggao 1998 _____. “Wenhua yinsu dingxing fenxi yu Shang dai ‘qingtong liqi wenhuaquan’ yanjiu 文化因素定性分析與商代‘青銅禮器文化圈’研究”. In Beijing 1998c, pp. 227–36. Xu Zhengguo 1988 Xu Zhengguo 徐正國. “Zaoyang shouci faxian Shang dai tongzun 棗陽首次發現商代铜尊”. Zhongguo wenwu bao, July 8, 1988. Xu Zhengguo 1990 _____. “Hubei Zaoyang faxian yijian Shang dai tongzun 湖北棗陽發現一件商代铜尊”. Wenwu 1990.6, p. 57. Xu Zhongshu 1962 Xu Zhongshu 徐仲舒. “Sichuan Peng Xian Mengyang zhen chutu de Yin dai er zhi 四川彭縣濛陽鎮出土的殷代二觶”. Wenwu 1962.6, pp. 15–8, 23. Xu Zhongshu 1982 _____. Lun Ba Shu wenhua 論巴蜀文化. Chengdu: Sichuan Renmin Chubanshe, 1982. Xu Ziqiang 1993 Xu Ziqiang 徐自强. “ Guanghan Anyang jisikeng bijiao yanjiu 廣漢安陽祭祀坑比較研究”. In Li et al. 1993, pp. 136–44. Yang 1999 Yang Xiaoneng, ed. The Golden Age of Chinese Archaeology, Celebrated Discoveries from the People’s Republic of China. New Haven and London: Yale University Press, 1999. Yang & Ding 1959 Yang Gen 楊根, Ding Jiaying 丁家盈. “Si Mu Wu dading de hejin chengfen jiqi zhuzhao jishu de chubu yanjiu 司母戊大鼎的合金成份及其鑄造技術的初步研究.” Wenwu 1959.12, pp. 27–9. Yang Jianfang 1993 Yang Jianfang 楊建芳. “Zaoqi Shu guo yudiao chutan—Shang dai fangguo yuqi yanjiu zhiyi 早期蜀國玉雕初探—商代方國玉器研究之一”. Gugong wenwu yuekan 120 (1993), pp. 54–65. Reprinted in Li et al. 1993 (pp. 160–9), and Yang Jianfang 2001 (vol. 1, pp. 121–29). 293 Yang Jianfang 1994. _____. “Luelun youlinghuan de qiyuan, chuanbo yu yongtu 略論有领環的起源、傳播與用途”. Zhongguo wenwu bao, January 9, 1994. Reprinted in Yang Jianfang 2001, vol. 2, pp. 168–9. Yang Jianfang 2001 _____. Zhongguo guyu yanjiu lunwenji 中國古玉研究論文集 (2 vols.). Taibei: Zhongzhi Meishu Chubanshe 2001. Yang Meili 1994 Yang Meili. “Damo guyan zhi, changhe luori yuan—gudai xibei diqu de huanxing yushiqi xilie zhiyi—Qijia wenhua fengge de huanxingqi 大漠孤煙直、長河落日圓—古代西北地區的環形玉石器系列之一—齊家文化風格的環形器”. Gugong wenwu yuekan 131 (1999), pp. 66–79. Yang Meili 1997 _____. “Duokongshi yudao de yanjiu 多孔式玉刀的研究”. Gugong xueshu jikan 15.3 (1997), pp. 17–74. Yang Meili 1998 _____. “Shi, yuge de yanjiu (shang) 石、玉戈的研究（上）”. Gugong xueshu jikan 16.1 (1998), pp. 155–82. Yang Meili 1999 _____. “Ji ‘Sanxingdui chuanqi’ de yushiqi 記‘三星堆傳奇’的玉石器”. Gugong wenwu yuekan 194 (1999), pp. 32–45 (part 1); 195 (1999), pp. 22–35 (part 2). Ye & Li 2001 Ye Wansong 葉萬松, Li Defang 李德方. “Yanshi Erlitou yizhi shouwen tongpai kaoshi 偃師二里頭遺址兽纹铜牌考釋”. Kaogu yu wenwu 2001.5, pp. 40–8. Yoshikai Masato 1999 Yoshikai Masato 吉開將人. “Zhongguo yu Dongnanya de ‘T’ zixing huan 中國與東南亞的‘T’字形環”. Sichuan wenwu 1999.2, pp. 81–96, tr. into Chinese by Chen De’an and Shi Yingping 石應平. An abbreviated version in Deng Cong 1994, pp. 255–68. Yu Qiuwei 1998 Yu Qiuwei 于秋偉. “Shandong Yi’nan xinfaxian de yazhang he yuqi 山東沂南新發現的牙璋和玉器”. Gugong wenwu yuekan 1998.2, pp. 78–85. Yu Weichao 1996 Yu Weichao 俞偉超. “Sanxingdui wenhua zai woguo wenhua zongpuxi zhong de weizhi, diwang jiqi tudi chongbai 三星堆文化在我國文化總譜系中的位置，地望及其土地崇拜”. In Beijing 1996a, pp. 59–63. Partly reprinted, with revision, in Yu Weichao, Gushi de kaoguxue tansuo 古史的考古學探索, pp. 276–80. Beijing: Wenwu Chubanshe, 2002. Yu Yanjiao 2002 Yu Yanjiao 喻燕嬌. “Luelun Hunan chutu de Shang dai yuqi 略論湖南出土的商代玉器”. Zhongyuan wenwu 2002.5, pp. 43–50. 294 Zeng Zhongmao 1989 Zeng Zhongmao 曾中懋. “Guanghan Sanxingdui yi er hao jisikeng chutu tongqi chengfen de fenxi 廣漢三星堆一、二號祭祀坑出土銅器成份的分析”. Sichuan wenwu 1989 (Sanxingdui yanjiu zhuanji 三星堆研究專辑), pp. 76–80. Zeng Zhongmao 1991 Zeng Zhongmao. “Guanghan Sanxingdui erhao jisikeng tongqi chengfen de fenxi 廣漢三星堆二號祭祀坑銅器成份的分析”. Sichuan wenwu 1991.1, pp. 72–4. Zeng Zhongmao 1994 Zeng Zhongmao. “Sanxingdui chutu tongqi de zhuzao jishu 三星堆出土銅器的鑄造技術”. Sichuan wenwu 1994.6, pp. 68–9, 77. Zettler & Horne 1998 Richard L. Zettler and Lee Horne, ed. Treasures from the Royal Tombs of Ur. Philadelphia: University of Pennsylvania Museum of Archaeology and Anthropology, 1998. Zhang & Wei 2004 Zhang Jiangkai 張江凱, Wei Jun 魏峻. Xinshiqi shidai kaogu 新石器時代考古. Beijing: Wenwu Chubanshe, 2004. Zhang Changping 2004 Zhang Changping 張昌平. “Lun Yinxu shiqi nanfang de zun he lei 論殷墟時期南方的尊和罍”. Kaoguxue jikan 15 (2004), pp. 116–28. Zhang Changshou 1994 Zhang Changshou 張長壽. “Lun Shenmu chutu de daoxing duanren yuqi 論神木出土的刀形端刃玉器”. In Deng Cong 1994, pp. 59–63. Zhang Minghua 1989 Zhang Minghua 張明華. “Sanxingdui jisikeng hui fou shi muzang 三星堆祭祀坑會否是墓葬? ” Zhongguo wenwu bao, June 2, 1989. Zhang Minghua 1998 _____. “Liangzhu shoumian wei huwen de youyi zhongyao lizheng 良渚兽面為虎纹的又一重要例證”. Zhongguo wenwu bao, September 9, 1998. Zhang Minghua 1999 _____. “Sanxingdui ‘shoumianju’ 三星堆‘兽面具’”. Zhongguo wenwu bao, March 17, 1999. Zhang Tian’en 2002 Zhang Tian’en 張天恩. “Tianshui chutu de shoumian tongpaishi ji youguan wenti 天水出土的兽面铜牌饰及有關問題”. Zhongyuan wenwu 2002.1, pp. 43–6. Zhang Xiaoma 1996 Zhang Xiaoma 張肖馬. “‘Jisikeng shuo’ bianxi ‘祭祀坑說’辨析”. In Beijing 1996a, pp. 70–8. Zhao & Chen 2001 Zhao Dianzeng, Chen De’an. “Yige chongman huoli de xueke shengzhangdian—Su Bingqi xiansheng zhidao xia de Sanxingdui kaogu 一個充滿活力的學科生長點—蘇秉琦先生指導下的三星堆考古”. In Su Bai 2001, pp. 451–69. 295 Zhao Congcang 1996 Zhao Congcang 趙叢蒼. “Chenggu Yang Xian tongqiqun zonghe yanjiu 城固洋縣銅器羣綜合研究”. Wenbo 1996.4, pp. 3–26. Zhao Congcang 2006 Zhao Congcang, ed., Chengyang qingtongqi 城洋青銅器. Beijing: Kexue Chubanshe, 2006. Zhao Dianzeng 1987 Zhao Dianzeng. “Ba Shu wenhua jige wenti de tantao 巴蜀文化幾個問題的探討”. Wenwu 1987.10, pp. 18–21. Zhao Dianzeng 1991 _____. “Mianyang wenwu kaogu zhaji 綿陽文物考古札記”. Sichuan wenwu 1991.5, pp. 35–8. Zhao Dianzeng 1993 _____. “Sanxingdui jisikeng wenwu yanjiu 三星堆祭祀坑文物研究”. In Li et al. 1993, pp. 81–92. Zhao Dianzeng 1996a _____. “The Sacrificial Pits at Sanxingdui”. In Rawson 1996, pp. 232–9. Zhao Dianzeng 1996b _____. “Renshen jiaowang de tujing—Sanxingdui wenwu yanjiu 人神交往的途徑—三星堆文物研究”. In Beijing 1996a, pp. 91–103. Zhao Zike 1991 Zhao Zike 趙紫科. “Yanting xian chutu gudai shibi 鹽亭縣出土古代石璧”. Sichuan wenwu 1991.5, p. 69. Zhao Xinlai 1966 Zhao Xinlai 趙新来. “Zhengzhou Erligang faxian de Shang dai yuzhang 鄭州二里岡發現的商代玉璋”. Wenwu 1966.1, p. 58. Zheng 1996 Zheng Zhenxiang. “The Royal Consort Fu Hao and Her Tomb”. In Rawson 1996, pp. 240–7. Zheng & Tang 2000 Zheng Junsheng 鄭均生, Tang Xianhua 唐先華. “Hunan Hengyang faxian Shang dai tongyou 湖南衡陽發現商代铜卣”. Wenwu 2000.10, pp. 58– 60, 11. Zheng Dekun 1946 Zheng Dekun 鄭德坤. Sichuan gudai wenhua shi 四川古代文化史 West China Union University Museum Monograph Series No. 1, 1946. Chengdu. Zheng Zhenxiang 1993 Zheng Zhenxiang 鄭振香. “Zaoqi Shu wenhua yu Shang wenhua de guanxi 早期蜀文化與商文化的關係”. Zhongyuan wenwu 1993.1, pp. 6–11, 46. Zhou Nanquan 1995 Zhou Nanquan 周南泉, ed., Yüqi (shang) 玉器（上）(Jades, Vol. 1). Hong Kong: the Commercial Press, 1995. 296 Zhu et al. 1966 Zhu Peizhang 祝培章, Bo Zhemin 卜哲民, Cheng Xuehua 程學華. “Shaanxi Chenggu xian faxian de qingtongqi 陝西城固縣發現的青銅器”. Wenwu, 1966.1, pp. 1–3. Zhu et al. 2003 Zhu Zhangyi, Zhang Qing, and Wang Fang. “The Jinsha Site: An Introduction,” Journal of East Asian Archaeology 5.1–4 (2003), pp. 247–76. Zhu & Wang 2004 Zhu Zhangyi 朱章義, Wang Fang 王方. “Chengdu Jinsha yizhi chutu yucong chubu yanjiu 成都金沙遺址出土玉琮初步研究”. Wenwu, 2004.4, pp.66– 70. Zhu Zhangyi 2001 Zhu Zhangyi. “Chengdu Shi nanjiao Shijiefang yizhi niandu fajue jiyao 成都市南郊十街坊遺址年度發掘紀要”. Chengdu kaogu faxian 1999 (2001), pp. 1–28. 297 THE SANXINGDUI SITE: ART AND ARCHAEOLOGY VOL. 2 (TABLES AND FIGURES) JAY JIE XU A DISSERTATION PRESENTED TO THE FACULTY OF PRINCETON UNIVERSITY IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY RECOMMENDED FOR ACCEPTANCE BY THE DEPARTMENT OF ART AND ARCHAEOLOGY Advisor: Robert Bagley November 2008 © Copyright by Jay Jie Xu, 2008. All rights reserved 298 Table 1a. Stratigraphy at the Sanxingdui Site from 1980-1986a (Adapted from Chen Xiandan 1989a, p. 217 table) Periodization by Chen Xiandan Xiquankan 1984 80–81 III (1) (1) Han to modern times (3) (4) (5) (6) 4th Phase 3rd Phase (2) (2) 2nd Phase (3) (3) (4) 1st Phase Loci and Strata Sanxingdui 1982 1984 I III I (1) (1) (1) (2) (2) (3) (4) (5) (6) (7) (8) (2) (3) (4) (5) (6) (4) (5) (6) (7) (8) 1986 II (1) (2) (3) (4) (5) (6) (7) (8) III (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) 299 a Adapted from Chen Xiandan 1989a, p. 217 table. While this table usefully correlates the strata excavated from 1980 to 1986 at Sanxingdui, one must exercise caution in using the correlation because it represents Chen Xiandan’s own problematic periodization scheme as discussed in the present chapter. 300 Table 1b. Periodization Schemes and Cultural Identifications of the Sanxingdui Site Sun Hua Report Scheme Chen Xiandan Scheme 1st Scheme Sun Scheme Sub-phases Phases Identity Phases Identity Phases Identity 6 4 3 4 Phases Identity 3 Shi’erqiao 2 Sanxingdui 1 Baodun Shi’erqiao 5 4 3 Sanxingdui 2 1 3 2 1 Sanxingdui 3 2 Sanxingdui 1 Baodun 2 1 301 302 303 304 305 306 307 Table 3. Inventories of K1 and K2. After Falkenhausen 2002, pp. 90–2 table 1. 308 309 Table 4. Elemental Composition of Bronzes from Early Bronze Age Sites (Adapted from Xu 2001, p. 69, table 1) Site Date (BC) Number of samples analyzed (number of objects sampled) Erlitou1 c. 1500 32 (32) 35–99+ 0.04–23 0.03–61 Zhengzhou2 c. 1500–1300 5 (5) 53–80 0.53–18 6–41 Sanxingdui3 c. 1200 27 (24) 64–98 0.03–12 0.03–33 Tomb at Jiangxi Xin’gan4 c. 1200 6 (6) 75–84 4.6–18.4 0–7.8 Fu Hao tomb at Anyang5 c. 1200 89 (89) 72–88 9–20 <8 Guojiazhuang M160 at Anyang6 12th c. 19 (19) 69–99 0–19 0.41–22 Yinxu West Sector tombs of Periods II and III at Anyang7 12th c. 18 (18) 72–94 0–20 0.5–22 Range of percent Cu Range of percent Sn Range of percent Pb 310 1 Beijing 1999d, p. 399 (table 2). 2 Beijing 1999f, pp. 125–7. Another object from Zhengzhou is reported to be 91 percent copper, 7 percent tin, and 1 percent lead (Wenwu 1959.12, p. 28). 3 The analyses tabulated here come from Zeng Zhongmao 1989, Zeng Zhongmao 1991, and Beijing 1999a (p. 495). Copper contents are fairly evenly distributed over the range from 64 percent to 98 percent. 4 Beijing 1997d, pp. 241–4. Twenty analyses are reported, but fourteen are disregarded here as unreliable (probably because of corrosion products in the samples, their metal contents totaled less than 95 percent). 5 Beijing 1984a, pp. 267–81. Ninety-one analyses are reported, but two are disregarded here as their metal contents total less than 95 percent. The composition of the Si Mu Wu fang ding, which is likely to be of about the same date, falls into the same range (Wenwu 1959.12, p. 28). 6 Beijing 1998f, p. 178. Twenty-six analyses are reported, but seven are disregarded here as their metal contents total less than 95 percent. 7 Li et al. 1984. Twenty-four analyses of Periods II and III samples are reported, but six are disregarded here as their metal contents total less than 95 percent. 311 Figure 1.1. Topographical map of China. After Ebrey 1996, p. 11. 312 Figure 1.2. Sichuan province (including the Chongqing municipality). After Hong Kong 2007, p. 23 fig. 1. 313 Figure 1.3. Bronze mask with protruding pupils, SXD, K2(2):148, h. 66 cm, w. 138 cm, d. 73 cm. After Bagley 2001, p. 109 no. 22. Figure 2.1. Location of the Sanxingdui site. After Paris 2003, p. 28. 314 315 Figure 2.2. Extent of the Sanxingdui site. After Beijing 1999a, p. 11 fig. 2. 316 Figure 2.3. The walled city, major finds and fieldwork of the Sanxingdui site. After Xu 2003, p. 150, fig. 1. Figure 2.4. Objects unearthed in 1927 at Yueliangwan. After Dye 1931. 317 Figure 2.5. Excavation loci at Sanxingdui from 1980 to 1986. After Xu 2003 p. 154, fig. 2. 318 Figure 2.6. K1 under excavation. After Beijing 1994a, p. 106. 319 Figure 2.7. K2 under excavation. After Bagley 1990a, p. 55 fig. 9. 320 321 Figure 2.8. The west wall, looking south. After Tokyo 1998, p. 42 fig. 3. 322 Figure 2.9. Layout of the Rensheng cemetery. After Kaogu 2004.10, p. 15 fig. 2. 323 Figure 2.10. Locations of walled settlements of the Baodun Culture, and other Neolithic sites in the Sichuan Basin. After Tokyo 2000, p. 101 fig. 66. 324 Figure 2.11. F5: the largest house foundation at the Gucheng site. After Wenwu 2001.3, p. 58 fig. 12. 325 a: Phase I. b: Phase II. c: Phase III. d: Phase IV. Figure 2.12. Pottery types at the Sanxingdui site. After Beijing 1999a, p. 425 figs. 225–6, p. 426 figs. 227–8. Figure 2.13. Jiandi zhan from K1. After Beijing 1999a, p. 146 fig. 76. Figure 2.15. Building foundations at Sanxingdui Locus III. After Kaogu xuebao 1987.2, p. 233 fig. 6. Figure 2.14. A reconstructed layout of the Sanxingdui city. After Sun Hua 2000, p. 164 fig. 6.9. 326 a. Figure 2.16. b. Pottery dou with openwork ring foot. a: M10:8, Rensheng cemetery, overall h. 24 cm. After Kaogu 2004.10, p. 19 fig. 11:5. b: AT2806(3):4, ring foot, Xiaojiawuji site of the Shijiahe Culture, existing h. 16.5 cm. After Beijing 1999e, p. 191 fig. 148:5. a. Figure 2.17. b. c. Jade awl-shaped implement. a: M5:6, Rensheng cemetery, l. 29.2 cm. After Kaogu 2004.10, p. 20 fig. 12:22. b: AT1215(2):2, Xiaojiawuji, l. 13.5 cm. After Beijing 1999e, p. 329 fig. 260:2. c. M9:4, Fuquanshan site of the Liangzhu Culture, l. 27 cm. After Huang Xuanpei 2000, p. 84 fig. 63.12. 327 Figure 2.18. Cross-section of the Yueliangwan wall excavated in 1999. Photograph by author, February 27, 2000. Figure 2.19. Adobe brick atop the east wall. Photograph courtesy of Chen De’an. 328 Figure 2.20. Pottery types from Shi’erqiao. After Sun Hua 1996, pp. 125–6 figs. 2–3. 329 330 Figure 3.1. K1, line drawing. After Beijing 1999a, p. 20 fig. 8. Figure 3.2. K2, line drawing. After Beijing 1999a, p. 159 fig. 81. Figure 3.3. K2, middle layer. After Taibei 1999, p. 196. 331 Figure 3.4. K2, top layer. After Taibei 1999, p. 194. The trench at the top of the photograph is not a ramp but an intrusive ditch dug in a later period. 332 Figure 4.1. Bronze tiger-shaped plaques. a: 1984 Rensheng village near the Yazihe River, l. 38 cm. After Beijing 1994a, pl. 66. b: Rensheng village, l. 43.4 cm, h. 13.2 cm. After Bagley 2001, no. 40. c: Jinsha, l. 26.5 cm. After Chengdu 2005, p. 48 no. 45. 333 a. Figure 4.2. c. Rectangular bronze plaques from the 1987 Cangbaobao pit. After Beijing 1998a, p. 81 fig. 3. a: 87GSZJ:16, l. 14 cm, w. 4.9–5.3 cm. b: 87GSZJ:36, l. 13.8 cm, w. 5.2–5.6 cm. c: 87GSZJ:17, l. 13.8 cm, w. 5.2–5.8 cm. a. Figure 4.3. b. b. Rectangular bronze plaques from other sites. a: Gaopian, l. 12.3 cm, w. 4.3–5 cm. After Ao & Wang 1980, p. 76 fig. 2.4. b: Erlitou, 87M57:4, l. 15.9 cm, w. 7.5–8.9 cm, ca. 16th century BC. After Kaogu 1992.4, p. 296 fig. 2.1. 334 335 a. Figure 4.4. b. c. Lithic spearheads. a: Gaopian, jade, h. 23 cm. After Ao & Wang 1980, p. 76 fig. 2.3. b: SXD, K1:141, stone, h. 34.2 cm. After Beijing 1999a, p. 128 fig. 67.1. c: SXD, K1:137, stone, h. 23 cm. After Beijing 1999a, p. 128 fig. 67.2. d: Jinsha, 2001CQJC:18, jade, h. 24.51 cm. After Beijing 2002c, p, 153. d. 336 a. Figure 4.5. b. c. Jades with notched edges. a: axe, Gaopian, h. 18 cm. After Ao & Wang 1980, p. 76 fig. 2.1. b: hatchet-shaped implement, SXD, K1:235, h. 12.4 cm. After Beijing 1999a, p. 84 fig. 44.3. c: axe, Erlitou, h. 21 cm, ca. 16th century BC. After Kaogu 1984.1, p. 38 fig. 5.2. Figure 4.6. Bronze pan, K1:53, h. 10.4 cm. Reconstruction drawing after Beijing 1999a, p. 43 fig. 26.1. Figure 4.8. Bronze zun, K1:163/K1:59. After Beijing 1999a, p. 41 fig. 24. Figure 4.7. Bronze lid, K1:135, h. 11.2 cm. Reconstruction drawing after Beijing 1999a, p. 43 fig. 26.2. Figure 4.9. Bronze pou, K1:130. After Beijing 1999a, p. 45 rubbing no. 4.1. 337 Figure 4.10. Bronze zun, K1:158/258, h. 43.3 cm. After Beijing 1999a, p. 35 fig. 23. Figure 4.11. Bronze zun from Anhui Funan, h. 50.5 cm, 13th century BC. After Beijing 1996b, pp. 117–8. a: side view centered on flange. b: detail of tiger and human. 338 Figure 4.12. Bronze zun, K2(2):112, h. 31.5 cm. After Beijing 1999a, p. 239 fig. 135. Figure 4.14. Figure 4.13. Bronze zun, K2(2):135. After Beijing 1999a, p. 241 fig. 137. Bronze zun, K2(2):109. Reconstruction drawing after Beijing 1999a, p. 240 fig. 136. 339 340 Figure 4.15. Bronze zun, K2(2):79, h. 44.2 cm. a: after Beijing 1999a, p. 242 fig. 138. b: after ibid., p. 246 rubbing no. 18. Figure 4.16. Bronze lei, K2(2):70, h. 33.4 cm. After Beijing 1999a, p. 263 fig. 145. Figure 4.18. Bronze fang lei, K2(3):205/K2(3):205-1, h. 35.6 cm. After Beijing 1999a, p. 279 rubbing no. 29. Figure 4.17. Bronze lei, K2(2):88, h. 35.4 cm. After Beijing 1999a, p. 264 fig. 146. Figure 4.19. Bronze lid, K2(2):32, h. 8.1 cm. After Beijing 1999a, p. 275 fig. 149. 341 Figure 4.20. Bronze zun, K2(2):127, h. 41.6 cm. After Beijing 1999a, p. 253 fig. 140. Figure 4.21. Bronze zun, K2(2):129, h. 45.5 cm. a: after Beijing 1999a, p. 254 fig. 141. b: after ibid., p. 257 rubbing no. 20. 342 343 Figure 4.22. Bronze zun, K2(2):146, h. 52.6 cm. After Beijing 1999a, p. 252 fig. 139. Figure 4.23. Bronze zun, K2(2):151, h. 56.5 cm. After Beijing 1999a, p. 255 fig. 142. Figure 4.24. Bronze lei, K2(2):159, h. 54 cm. After Beijing 1999a, p. 265 fig. 147. Figure 4.25. Bronze lei, K2(2):39/K2(2):39-1. After Beijing 1999a, p. 274 fig. 148.2–3. Figure 4.26. Bronze lei, K2(2):103/K2(2):103-1/K2(2):103-2. After Beijing 1999a, p. 278 rubbing no. 27. 344 Figure 4.27. Bronze bird for a vessel lid, K2(3):193-1, h. 34 cm. After Beijing 1999a, p. 334 fig. 184. Figure 4.29. Figure 4.28. Bronze fragment, probably of a vessel lid, K2(3):23, h. of bird, 13.3 cm. After Beijing 1999a, p. 266 rubbing no. 23. Bronze lid (MA1612) said to come from Hunan Changsha, h. 28 cm, late 13th century BC, Musée Guimet, Paris. After Bagley 2001, p. 150 fig. 50.1. 345 Figure 4.30. Bronze zun from Anyang, M18:13, h. 53 cm, early 12th century BC. After Beijing 1985b, fig. 53. Figure 4.31. Bronze kneeling figure bearing a zun, K2(3):48, h. 15.6 cm. After Beijing 1999a, p. 170 fig. 86. 346 Figure 5.1. Bronze head covered with gold foil, K2(2):45. Before restoration. Photograph courtesy of Yang Xiaowu. See Figure 5.16 for the restored head. Figure 5.2. Fragments of bronze tree before restoration (the photograph shows the trees in Figures 5.30 and 5.40, and two smaller bases). After Bagley 2001, p. 116 fig. 27.1. 347 Figure 5.3. Bronze head, K1:2, h. 29 cm, greatest width 20.6 cm, weight 4.48 kg. After Beijing 1999a, p. 24 fig. 9. Figure 5.4. Bronze head, K1:6, h. 25 cm, greatest width 20.4 cm, weight 3.36 kg. After Beijing 1999a, p. 24 fig. 10. Figure 5.5. Bronze head, K1:5, h. 45.6 cm, greatest width 22 cm, weight 4.54 kg. After Beijing 1999a, p. 29 fig. 17. 348 Figure 5.6. Bronze head covered with gold foil, K2(2):214, h. 48.1 cm, greatest width 22 cm. After Beijing 1999a, p. 184 fig. 101. Figure 5.7. Bronze head, K2(2):58, h. 51.6 cm, greatest width 23.8 cm, weight 5.8 kg. After Beijing 1999a, p. 178 fig. 96. 349 Figure 5.8. Bronze head, K2(2):90, h. 34.8 cm, greatest width 17.2 cm, weight 2.081 kg. After Beijing 1999a, p. 177 fig. 95. Figure 5.9. Bronze head, K2(2):83, h. 13.6 cm, greatest width 10.8 cm, weight 0.71 kg. After Beijing 1999a, p. 172 fig. 88. 350 Figure 5.10. Bronze head, K1:7, h. 27 cm, greatest width 22.8 cm, weight 7.66 kg. After Beijing 1999a, p. 27 fig. 12. Figure 5.11. Bronze head, K1:72, h. 30.8 cm, greatest width 17.8 cm, weight 2.01 kg. After Beijing 1999a, p. 27 fig. 14. Figure 5.12. Bronze head, K1:10, h. 27.8 cm, greatest width 19.4 cm, weight 4.36 kg. After Beijing 1999a, p. 28 fig. 15.2. 351 Figure 5.13. Bronze head, K2(2):17, h. 40 cm, greatest width 18.2 cm, weight 4 kg. After Beijing 1999a, p. 176 fig. 93. Figure 5.14. Bronze head, K2(2):34, h. 36.8 cm, greatest width 17.4 cm, weight 2.4 kg. After Rawson 1996, p. 65 fig. 24.1. Figure 5.16. Figure 5.15. Bronze head, K2(2):154, h. 17.6 cm, greatest width 10.8 cm, weight 0.691 kg. After Beijing 1999a, p. 175 fig. 89. Bronze head covered with gold foil, K2(2):45, h. 42.5 cm, greatest width 19.6 cm, weight 2.55 kg. After Beijing 1999a, p. 183 fig. 98. 352 f 353 g, h, i. j. Figure 5.17. k. Bronze figure on pedestal, K2(2):149, 150, overall h. 260.8 cm, h. of figure 172 cm. a-e: line drawings from various angles. After Beijing 1999a, foldout following p. 162. f: design of headdress. After Beijing 1999a, p. 164 rubbing 11. g: outer garment. After Wang & Wang 1993, p. 62 fig. 2. h: middle garment. After ibid. i: inner garment. After ibid. j: dragons on the mantle. After Bagley 2001, p. 74 fig. 2.4. k: detail of decorated pedestal. Photograph by author. 354 Figure 5.18. Bronze figure with an animal headdress, K2(3):264, h. 40.2 cm. After Beijing 1999a, p. 167 fig. 84. Figure 5.19. Bronze plaque in the form of a kneeling figure, K2(3):04, h. 13.3 cm. After Beijing 1999a, p. 170 fig. 85.3. Figure 5.20. Bronze figure, K2(3):296-1, h. of figure 10.8 cm. After Beijing 1999a, p. 234 fig. 130.1. Figure. 5.21. Bronze figure, K2(3):292-2, h. 8.3 cm. After Beijing 1999a, p. 167 fig. 83. 355 Figure 5.22. Bronze hybrid figure standing on birds, K2(3):327, overall h. 81.4 cm, h. of figure 30 cm. After Beijing 1999a, p. 171 fig. 87. Figure 5.24. Bronze kneeling figure holding a forked blade, K2(3):325, h. 4.7 cm. After Beijing 1999a, p. 235 fig. 133. Figure 5.23. Bronze pedestal or miniature building with kneeling figure, K2(2):143-1, fragment, h. 31 cm. After Beijing 1999a, p. 234 fig. 131. Figure 5.25. Bronze diamond-shaped appliqués. After Beijing 1999a, p. 208 fig. 115. a: integral appliqué, K2(3):202, l. 57.2 cm, w. 23.6 cm. b: two halves, K2(3):197, l. 54.8 cm, w. 12.7 cm; K2(3):8, l. 54.8 cm, w. 12.8 cm. c: four quarters, K2(3):101, l. 29 cm, w. 12.4 cm; K2(3):106, l. 28 cm, w. 12.4 cm; K2(3):8-1, l. 28.6 cm, w. 12.4 cm; K2(3):99, l. 27.8 cm, w. 13.2 cm. 356 Figure 5.26. Bronze mask, K2(2):293, h. 25.5 cm, w. 37.8 cm. After Beijing 1999a, p. 189 fig. 104.2. Figure 5.27. Bronze mask, K2(2):153, h. 40.3 cm, w. 60.5 cm. After Beijing 1999a, p. 189 fig. 103.2. 357 Figure 5.28. Bronze mask with protruding pupils, K2(2):148, h. 66 cm, w. 138 cm, depth 73 cm. After Beijing 1999a, p. 197 fig. 110. Figure 5.29. Bronze mask with protruding pupils and a trunk, K2(2):142, overall h. 82.5 cm, h. of mask 31.5 cm, w. 77 cm. After Beijing 1999a, p. 197 fig. 108. 358 Figure 5.30. Bronze tree, K2(2):94, overall h. 3.96 m, h. of trunk 3.59 m, d. of base 0.93 m. After Beijing 1999a, foldout page facing p. 218. Figure 5.31. Bronze bird from a tree, K2(2):213, h. 8 cm. After Beijing 1999a, p. 227 fig. 126.1. Figure 5.32. Bronze bird with human head, K2(2):154, h. 12 cm. Reconstruction drawing after Taibei 1999, p. 99. . 359 Figure 5.33. Bronze bells. After Beijing 1999a, p. 292 fig. 161.2, p. 293 figs. 162.1– 162.2, p. 299 figs. 163.1–163.2, 164.1–164.2. a: K2(3):103-31, h. 10.2 cm. b: K2(3):103-28, h. 7.35 cm. c: K2(3):274, h. 8.3 cm. d: K2(3):149, h. 14.3 cm. e: K2(2):103-8, h. 14 cm. f: K2(3):70-7, h. 7.6 cm. g: K2(3):78, h. 12.2 cm. 360 a. c. d. b. Figure 5.34. Bronze shells. After Beijing 1999a, p. 300 figs. 165.1, 165.4, p. 301 fig. 166.5, p. 310 fig. 169.1. a: K2(2):79-6, d. 9.1 cm. b: K2(3):103-22, d. 6.9 cm. c: K2(3):6, h. 12.7 cm. d: K2(3):265-1, h. 9.5 cm. Figure 5.35. Bronze hanging apparatus, K2(3):124, h. 6.2 cm. After Beijing 1999a, p. 314 fig. 171.1. . Figure 5.36. Bronze foils. After Beijing 1999a, p. 320 figs. 173.3, 174.2, 175.2. a: fish-shaped foil, K2(3):194-6, l. 9.8 cm. b: forked-blade-shaped foil, K2(3):194-2, l. 6.5 cm. c: foil with vine pattern, K2(3):194-13, size unknown. 361 Figure 5.37. Fragment of a bronze tree with a bronze collared disk attached to the trunk, K2(3):204, 261, overall h. 59.7 cm, diameter of disk 8.8 cm. After Beijing 1999a, p. 222 fig. 123.1. Figure 5.39. Figure 5.38. Fragment of a bronze tree with a jade collared disk attached to the calyx of a flower, K2(3):20, overall height 50 cm. After Beijing 1999a, p. 225 fig. 124. . Bronze collared disks. a: SXD K2(3):134, d. 11.2 cm. After Beijing 1999a, p. 282 fig. 154.2. b: SXD K2(2):99, l. 8.9 cm. After Beijing 1999a, p. 285 fig. 157.5. 362 a. b. Figure 5.40. Bronze tree (fragment), K2(2):194, overall h. 193.6 cm. a: after Tokyo 1998, p. 99. b: line drawing after Beijing 1999a, p. 220 fig. 121. Figure 5.41. Bronze tree, K2(3):272, overall h. 50 cm. Reconstruction drawing after Beijing 1999a, p. 226 fig. 125. 363 364 Figure 5.42. Jade implement with incised figures, K2(3):201-4, l. 54.2 cm. After Beijing 1999a, p. 361 fig. 197.1. Figure 5.43. Bronze altar?, K2(3):296, h. 54 cm. Reconstruction drawing based on surviving fragments, after Beijing 1999a, p. 233 fig. 129. Figure 5.44. Bronze bird’s head, K2(2):141, h. 40.3 cm. After Beijing 1999a, p. 333 fig. 183.1. . Figure 5.45. Bronze bird on a post, K2(3):301-3, h. 27.8 cm. After Beijing 1999a, p. 335 fig. 185.3. Figure 5.46. Bronze rooster, K2(3):107, h. 14.2 cm. After Beijing 1999a, p. 333 fig. 183.2. 365 Figure 5.47. Bronze tube with dragon, K1:36, h. 41 cm. After Beijing 1999a, p. 34 fig. 20. Figure 5.48. Bronze snake. After Beijing 1999a, p. 326 fig. 178.1, p. 327 figs. 179.1, 179.3. a: head, K2(3):87, l. 54.8 cm. b: middle section, K2(3):56, l. 35.6 cm. c: tail, K2(3): 44, l. 21.2 cm. Figure 5.49. Bronze tiger-like creature, K1:62, l. 11.4 cm, h. 10.8 cm. After Beijing 1999a, p. 35 fig. 22. Figure 5.50. Bronze seated figure, K1:293, h. 14.6 cm. After Beijing 1999a, p. 29 fig. 18. 366 Figure 5.51. Bronze plaques of taotie faces. After Beijing 1999a, p. 198 figs. 111.2 and 112.1, p. 203 fig. 113.3. a: K2(3):221, h. 21.2 cm, w. 35 cm. b: K2(3):231, h. 20.8 cm, w. 26.4 cm. c: K2(3):231-1, h. 12.3 cm, w. 27.8 cm. Figure 5.52. Bronze circular appliqué, K2(3):1, d. 84 cm. After Beijing 1999a, p. 239 fig. 134.1. Figure 5.53. Bronze fragment of a miniature roof, K2(2):143, h. 15.8 cm. After Beijing 1999a, p. 235 fig. 132. 367 Figure 5.54. Comparisons of Shijiahe and Sanxingdui heads. After Falkenhausen 2003, p. 194 fig. 2. 368 b. a. Figure 6.1. a: bronze ding, PLZM2:36, Hubei Huangpi Panlongcheng, h. 30.1 cm, 15th–14th century BC. After Beijing 1996b, pl. 32. b: diagram showing the relationship between the Panlongcheng ding and the mold used to cast it. After Fong 1980, p. 72 fig. 16. Figure 6.2. Bronze gui, Freer Gallery of Art (31.10). Detail of precast flange. After Gettens 1969, p. 93 fig. 94. 369 a. b. Figure 6.3. a: bronze fang ding, XDM:8, Jiangxi Xi’gan Dayangzhou, h. 97 cm, 14th century BC. After Beijing 1997d, p. 33 fig. 20(A). b: detail of cast-on tiger. After Hong Kong 1994, pl. 35-3. Figure 6.4. Bronze fang ding from Shanxi Pinglu Qianzhuang, h. 82 cm, 14th century BC. After Beijing 1996b, pl. 37. 370 Figure 6.5. Bronze jia, Freer Gallery of Art (35.12). Detail of capped post with run-on join. After Gettens 1969, p. 94 fig. 98. Figure 6.7. Figure 6.6. Bronze guang, Freer Gallery of Art (39.53). Detail of brazed handle. After Gettens 1969, p. 87 fig. 83. Bronze ding from Henan Zhengzhou, h. 77.3 cm, 14th century BC. After Tokyo 1986, p. 62 no. 32. 371 a. Figure 6.8. b. a: bronze fang yi, h. 29.8 cm, ca. 12th century BC, Harvard University Art Museums (1943.52.109). After Bagley 1990b, p. 7 fig. 1B. b: diagram showing the relationship between the Harvard fang yi and the mold used to cast it. After Bagley 1990b, p. 9 fig. 5. a. Figure 6.9. b. a: bronze zun, h. 36.8 cm, ca. 12th century BC. Freer Gallery of Art (51.19). After Pope et al. 1967, p. 99 pl. 16. b: detail of animal head. After Pope et al. 1967, p. 101 fig. 11. 372 a. Figure 6.10. b. Bronze ge with serrated edges a: Sanxingdui, K1:247-2, l. 20.5 cm. After Beijing 1999a, p. 56 fig. 32.5. b: Jinsha, 2001CQJC:169, l. 21.5 cm. After Beijing 2002c, p. 56. Figure 6.11. Bronze mask, K2(2):331, h. 15.4 cm, w. 18.2 cm. After Tokyo 1998, p. 79 (below). Figure 6.12. Bronze mask, K2(3):14 (fragment), h. 25.6 cm. Left side. Photograph by author. 373 b. a. Figure 6.13. Bronze head, K1:2, in Figure 5.3. Photographs by author. a: detail of area above right ear. b: casting-on repairs viewed from interior. Figure 6.14. Bronze head, K2(2):82, h. 39.6 cm. Right side. Photograph by author. Figure 6.15. Bronze head, K2(2):154, in Figure 5.15. Detail of top. After Tokyo 1998, p. 57 (left). 374 Figure 6.16. Bronze head, K1:7, in Figure 5.10. Detail of top. Photograph by author. a. b. Figure 6.17. Bronze head covered with gold foil, K2(2):214, in Figure 5.6. a: detail of top. Photograph by author. b: right side. After Yang 1999, p. 211. 375 a. b. Figure 6.18. Bronze head, K1:5, in Figure 5.5. a: detail of top. After Tokyo 1998, p. 154 (lower). b: left side. Photograph by Paul Macapia. Figure 6.19. Bronze head, K2(2):51, h. 40.4 cm. Detail of top. Photograph by author. 376 a. b. c. Figure 6.20. Bronze mask with protruding pupils, K2(2):148, in Figures 1.3/5.28. a: detail of left eye. Photograph by author. b: detail of nostrils. Photograph by author. c: rear view. Photograph by Paul Macapia. 377 Figure 6.22. Bronze head, K2(2):107, h. 36.6 cm. Detail of top. Photograph by author. Figure 6.21. Bronze head, K2(2):83, in Figure 5.9. Left side. After Tokyo 1998, p. 117 (lower). b. a. Figure 6.23. Bronze head, K2(2):90, in Figure 5.8. a: detail of top. Photograph by author. b: rear view. After Tokyo 1998, p. 60 (right). Figure 6.24. Bronze head, K2(2):14, h. 39.5 cm. Detail of top. Photograph by author. Figure 6.25. Bronze head covered with gold foil, K2(2):115, h. 41 cm. Detail of top. Photograph by author. 378 b. a. c. d. e. f. Figure 6.26. Bronze tree (fragment), K2(2):194, in Figure 5.40, a–e, photographs by author; f: after Beijing 1994a, pl. 41. a: detail of soldered or run-on rim. b. detail of a kneeling figure. c. detail of a hole on the rim. d: detail of damaged side. e: cast rivet seen from inside. f: view of another side. 379 a. b. Figure 6.27. Bronze mask, K2(2):128, h. 25.4 cm. Photographs by author. a: left ear, back side. b: detail of right inside. Figure 6.28. Bronze hybrid figure standing on birds, K2(3):327, in Figure 5.22. Detail of soldered join between the figure’s foot and the bird. After Bagley 2001, p. 128 fig. 36.3. 380 a. b. c. Figure 6.29. Bronze mask with protruding pupils and a trunk, K2(2):142, in Figure 5.29. a: rear view. After Rawson 1996, p. 67 fig. 25.2. b: right side. After Taibei 1999, p. 80. c: detail of left ear, back side. Photograph by author. 381 a. b. c. d. Figure 6.30. Bronze figure with an animal headdress, K2(3):264, in Figure 5.18. a: rear view. After Tokyo 1998, p. 118 (right). b: detail of top. Photograph by author. c: detail of left projection with solder reinforcement. Photograph by author. d: detail of neck, right side. Photograph by author. 382 a. b. c. d. 383 384 e. Figure 6.31. f. Bronze figure on pedestal, K2(2):149, 150, in Figure 5.17. a: detail of the decorated part of the pedestal. After Yang 1999, p. 210 (right). b: detail of back. After Tokyo 1998, p. 48 (left). c: detail of lower part. After Tokyo 1998, p. 48 (right). d: detail of lower right side. Photograph by author. e: before restoration. After Beijing 1999a, p. 165 pl. 57. f: detail of left side. After Bagley 2001, p. 74 Figure 2. g: detail of upper part. After Tokyo 1998, p. 49. g. a. b. c. d. 385 e. f. g. h. i. Figure 6.32. Bronze tree, K2(2):94, in Figure 5.30. a: detail of base. After Tokyo 1998, p. 97 (below). b: detail of a branch join. Photograph by author. c: detail of join of branches to trunk. After Tokyo 1998, p. 97 (above). d: detail showing the dragon. After Tokyo 1998, p. 95. e: detail of one of the four struts joining the dragon to the trunk. Photograph by author. f: detail of a soldered join on the dragon. Photograph by author. g: detail of a bird. After Tokyo 1998, p. 96. h: detail of a calyx. Photograph by author. i: cross section of a broken branch. Photograph by author. 386 Figure 6.33. Bronze circular appliqué, K2(3):1, in Figure 5.52. Detail of join at rim. After Bagley 2001, p. 135 fig. 41.2 a. Figure 6.34. b. Bronze mask, K2(2):114, h. 26.6 cm. Photographs by author. a: detail of left ear, back side. b: detail of right ear, back side. 387 Figure 7.1. Stone bi from the 1987 Cangbaobao pit. a: larger ones, diameter of the largest one (87GSZJ:21), 20.3 cm. After Beijing 1998a, p. 84 fig. 7. b: smaller ones, diameter of the smallest one (87GSZJ:9), 3.1–3.5 cm. After Beijing 1998a, p. 86 fig. 9. Figure 7.2. Stone bi with undrilled center, collected in 1976 at Yinchuan in Ningxia Hui Autonomous Region. Qijia Culture, early 2nd millennium BC. After Bagley 2001, p. 159 fig. 8. 388 a. b. c. d. e. f. 389 390 g. Figure 7.3. h. i. Jade collared types. a: Sanxingdui, ring, K2(2):5-4, overall d. 8.5 cm, d. of perforation 5.5 cm. After Beijing 1999a, p. 371 fig. 204.3. b: Sanxingdui, ring, K2(2):146-3, overall d. 13.7 cm, d. of perforation 6.8 cm. After Beijing 1999a, p. 370 fig. 203.1. c: Sanxingdui, ring, from the 1927 Yueliangwan pit, overall d. 11.2 cm, d. of perforation 7 cm. Sichuan Provincial Museum (110483). After Xiao et al. 2001, p. 15. d: Sanxingdui, disk, K2(2):57-2, overall d. 17.5 cm, d. of perforation 6.7 cm. After Beijing 1999a, p. 369 fig. 202.1. e: Sanxingdui, disk, K2(2):146-2, overall d. 17.8 cm, d. of perforation 6.7 cm. After Beijing 1999a, p. 369 fig. 202.2. f: Sanxingdui, irregularly shaped implement with collaring perforation, K1:204, l. 20.8, d. of perforation 3.3 cm. After Beijing 1999a, p. 84 fig. 43.1. g. Xin’gan, XDM:651, overall d. 16.8 cm, d. of perforation 7.2 cm. After Beijing 1997d, p. 144 fig. 75.1. h. Anyang, 01HDM54:352, overall d. 17.6 cm, d. of perforation 5.7 cm. After Beijing 2005, p. 6. i: Jinsha, 2001CQJC:2, overall d. 16.96 cm, d. of perforation 6.2–6.42 cm. After Beijing 2002c, p. 94. 391 a. Figure 7.4. b. c. Jade circular types. a: Sanxingdui, 1987 Cangbaobao pit, bracelet-like ring, 87GSZJ:33, d. 5.6 cm, h. 2.2 cm. After Beijing 1998a, p. 83 fig. 6.4. b: Sanxingdui, 1987 Cangbaobao pit, ring, 87GSZJ:11, d. 9 cm, d. of perforation 6.4 cm. After Beijing 1998a, p. 82 fig. 5.1. c: Jinsha, bracelet-like ring, 2001CQJC:172, d. 6.92–7.10 cm, h. 3.68 cm. After Beijing 2002c, p. 90. a. b. c. d. 392 e. Figure 7.5. f. Jade cong tubes. a: 1927 Yueliangwan pit, h. 5.5 cm. After Graham 1934, plate following p. 128. b: 1927 Yueliangwan pit, h. 7.2 cm. After Tokyo 1998, p. 176 no. 152. c: 1927 Yueliangwan pit, AK:2.2, 110485, h. 11 cm. Sichuan Provincial Museum. Photograph by author. d: Sanxingdui, K1:11-2, h. 7.6 cm. After Beijing 1999a, p. 81 fig. 42.1. e: Jinsha, 2001CQJC:61, h. 22.26 cm. After Beijing 2002c, p. 84. f: Jinsha, 2001CQJC:1, h. 16.57 cm. After Beijing 2002c, p. 86. 393 a. b. c. Figure 7.6. a: Jade trapezoidal implement, Sanxingdui, K1:81, 97, l. 1.62 m. After Beijing 1999a, p. 64 fig. 35. b: Jade trapezoidal knife, unearthed at Erlitou in 1975, late 16th century BC, l. 60.4–65 cm. After Kaogu 1978.4, p. 270 fig. 1.3. c: Jade rectangular implement, from Hunan Shimen Weigang, M1:07, ca. 1500 BC, l. 48 cm. After Wang & Long 1987, p. 17 fig. 8.1. 394 395 a. Figure 7.7. b. Jade parallelogram-shaped implements. a: Sanxingdui, K2(3):150, l. 36.1 cm. After Beijing 1999a, p. 360 fig. 196.1. b: Sanxingdui, K2(3):194, l. 66.4 cm. After Beijing 1999a, p. 361 fig. 197.2. c: Jinsha, 2001CQJL10:16, l. 18.3 cm. After Beijing 2006b, p. 76. c. 396 Figure 7.8. Map of the distribution of forked blades. After Deng Cong 1994, end papers. Figure 7.9. k. Jade forked blades from sites other than Sanxingdui. a: SSY7, from Shaanxi Shenmu Shimao, l. 26.5 cm. Shaanxi Provincial Museum. After Deng Cong 1994, fig. A:6. b: from Shandong Linyi Dafanzhuang, l. 32.5 cm. After Deng Cong 1994, fig. D:7. c: SSY16, from Shaanxi Shenmu Shimao, l. 34.5 cm. Shaanxi Provincial Museum. After Deng Cong 1994, fig. A:14. d: SSY15, from Shaanxi Shenmu Shimao, l. 30.6 cm. Shaanxi Provincial Museum. After Deng Cong 1994, fig. A:13. e: VM3:4, from Erlitou, l. 54 cm. After Kaogu 1983.3, p. 204 fig. 10.6. f: VM3:5, from Erlitou, l. 48.1 cm. After Kaogu 1983.3, p. 204 fig. 10.5. g: III KM6:8, from Erlitou, l. 49.5 cm. After Beijing 1999c, p. 250 fig. 162.3. h: VII KM7:5, from Erlitou, l. 46–48 cm. After Deng Cong 1994, fig. D:12. i: unearthed in 1958 at Erligang, l. 66 cm. After Deng Cong 1994, fig. B:1. j: SSY17, from Shaanxi Shenmu Shimao, l. 49 cm. Shaanxi Provincial Museum. After Deng Cong 1994, fig. A:7. k: 2001CQJC:955, from Jinsha, l. 42.25 cm. After Beijing 2002c, p. 109. 397 a. b. g. Figure 7.10. c. h. d. i. e. j. f. k. Jade forked blades at Sanxingdui. a: surface find in 1984 at Cangbaobao, l. 55.5 cm. After Bagley 2001, no. 52. b: 1927 Yueliangwan pit, no. (3.1)260, l. 39.3 cm. Sichuan University Museum. After Deng Cong 1998, vol. 3, color pl. 277. c: 1927 Yueliangwan pit, no. 313, l. 61 cm. Sichuan Provincial Museum. After Deng Cong 1994, fig. D:8. d: K1:170, l. 48.4 cm. After Beijing 1999a, p. 64 fig. 36.1. e: K1:23, l. 27 cm. After Beijing 1999a, p. 64 fig. 36.2. f: K1:01, l. 25.2 cm. After Beijing 1999a, p. 64 fig. 36.3. g: K1:02, l. 28.2 cm. After Beijing 1999a, p. 64 fig. 36.4. h: K1:275, l. 23.2 cm. After Beijing 1999a, p. 64 fig. 36.5. i: K2(3):167, l. 30.5 After Beijing 1999a, p. 362 fig. 199.2. j: K2(3):322-7, l. 33.6 cm. After Beijing 1999a, p. 362 fig. 198.4. k: K2(3):320, l. 67.8 cm, thickness 0.6 cm. After Beijing 1999a, p. 363 fig. 200.5. 398 a. b. c. d. e. f. g. Figure 7.11. Jade ge blades from K1 and K2. a: K2(3):322-8, jade, l. 55.4 cm. After Beijing 1999a, p. 385 fig. 209.1. b: K1:97-8, jade, l. 47.1 cm. After Beijing 1999a, p. 94 fig. 48.2. c: K2(3):227-1, jade, l. 33.9 cm. After Beijing 1999a, p. 378 fig. 205.3. d: K1:141-1, 155-2, jade, l. 40 cm. After Beijing 1999a, p. 91 fig. 46.1. e: K2(3):314-6, jade, l. 27 cm. After Beijing 1999a, p. 379 fig. 206.5. f: K1:136, jade, l. 17.8 cm. After Beijing 1999a, p. 97 fig. 49.1. g: K1:246, jade, l. 21.1 cm. After Beijing 1999a, p. 97 fig. 49.2. 399 a. b. c. d. e. f. g. Figure 7.12. Jade ge blades from other sites. a: ELT Phase III, IIIKM1:2, l. 30.2 cm. After Beijing 1999d, p. 250 fig. 162.5. b: ELT Phase III, bronze, IIIcai:60, l. 27.5 cm. After Beijing 1999d, p. 169 fig. 103. c: PLZM2:14, from Hubei Huangpi Panlongchen, l. 70 cm. After Beijing 2001e, p. 180 fig. 119.1. d: from a tomb at Henan Zhengzhou Baijiazhuang excavated in 1955, l. 37.9 cm. After Chen & Fang 1993, pl. 19. e: M5:444, from the tomb of Fu Hao, l. 27.4 cm. After Chen & Fang 1993, pl. 32. f. Jinsha, 2001CQJC:60, l. 50 cm. After Beijing 2002c, p. 144. g. Jinsha, 2001CQJC:478, l. 16.2 cm. After Beijing 2002c, p. 142. 400 401 a. Figure 7.13. b. c. Jade hybrids of forked blade and ge. a: Sanxingdui, K1:151, l. 43.2 cm. After Beijing 1999a, p. 68 fig. 37.1. b: Sanxingdui, K1:75, l. 34.8 cm. After Beijing 1999a, p. 75 fig. 39.1. c: Sanxingdui, K1:235-5, l. 38.2 cm. After Beijing 1999a, p. 81 fig. 41.1. d: Jinsha, 2001CQJC:27, l. 34.6 cm. After Beijing 2006b, p. 85. d. a. b. Figure 7.14. a: Jade ge, K1:23-1, l. 29.5 cm. After Beijing 1999a, p. 94 fig. 48.5. b: Jade hybrid, K1:155-1, l. 39 cm. After Beijing 1999a, p. 97 fig. 49.3. Figure 7.15. Jade axe, K1:266, h. 16.5 cm. After Beijing 1999a, p. 98 fig. 51.3. Figure 7.16. Jade adze, K1:261, h. 18.3 cm. After Beijing 1999a, p. 118 fig. 60.2 402 a. b. Figure 7.17. Jade knives. a: Sanxingdui, K2(2):314-5, l. 27 cm. After Beijing 1999a, p. 390 fig. 210.1. b: Tomb of Fu Hao, l. 33.5 cm. After Beijing 1984a, p. 134 fig. 74 (middle). a. Figure 7.18. b. Jade chisels a: Sanxingdui, K1:150, h. 27.1 cm. After Beijing 1999a, p. 107 fig. 55.6. b: Jinsha, 2001CQJC:29, h. 22.61 cm. After Beijing 2002c, p. 138. 403 a. b. Figure 7.19. Jade sword-shaped implements. a: K1:280, h. 28.2 cm. After Beijing 1999a, p. 97 fig. 50. b: Jinsha, 2001CQJC:583, h. 10.8 cm. After 2006b, p. 98. . Figure 7.20. Boulders of jade found in 1974 by the Yazihe River at Suozitian. After Xiao et al. 2001, p. 24. Figure 7.21. Partly-worked and waste stone bi disks from Sanxingdui, exact locale of excavation unknown. After Tokyo 1998, p. 182. 404 a. b. c. Figure 7.22. Gold sheath, K1:1, l. 142 cm, diameter 2.3 cm, weight 463 g (including carbonized wood). a: after Bagley 2001, p. 71 no. 1. b: after ibid., fig. 1.1. c: after Beijing 1999a, p. 61 fig. 34.1. Figure 7.23. Gold band from Jinsha, 2001CQJC:688, diameter 19.6–19.9 cm, weight 44 g. After Paris 2003, p. 163 fig. 3. 405 Figure 7.24. Gold appliqué in the shape of a tiger, K1:11–1, h. 6.7 cm, l. 11.6 cm, weight 7.27 g. After Beijing 1999a, p. 61 fig. 34.3. Figure 7.25. Gold foils from K2. After Beijing 1999a, p. 354 fig. 195. 406 Figure 8.1. Pottery gu and tripod he from Sanxingdui (left) and Erlitou (right). After Sun Hua 2000, p. 154 fig. 6.5. a. Figure 8.2. b. Bronze ge blades. a: Xinfan Shuiguanyin, , l. 15–19.8 cm. Photography by author. b: Jinsha, 2001CQJC:646, l. 22 cm. After Chengdu 2005, p. 33. 407 a. Figure 8.3. Pottery high-stem dou. a: Sanxingdui. After Beijing 1999a, p. 425 fig. 226:7. b: from Xiaojiawuji, H538:15, h. 39.6. After Beijing 1999e, p. 253 fig. 191:17. a. Figure 8.4 b. b. Pottery zun-shaped vessels. a: from Sanxingdui. After Beijing 1999a, p. 426 fig. 227:4. b: from the Dengjiawan site of the Shijiahe Culture, H48:12, h. 14 cm. After Beijing 2003, p. 170 fig. 141:6. 408 a. Figure 8.5. b. Pottery hens. a: from Sanxingdui. After Chen Xiandan 1989a, pl. 3:4. b: from Dengjiawan, AT203(2a):1, existing h. 8.1 cm. After Beijing 2003, p. 210 fig. 169:3. 409 Figure 10.1. Shi’erqiao and related sites in Chengdu. After Sun Hua 1996, p. 124 fig. 1. 410 Figure 10.2. Location of the Jinsha site in Chengdu city. After Zhu et al. 2003, p. 250 fig. 2. Figure 10.3. Terraced mound at Yangzishan. Reconstruction drawing. After Kaogu xuebao 1957.4, p. 20 fig. 3. 411 Figure 10.4. House elevated on stakes at Shi’erqiao. Reconstruction drawing. After Wenwu 1987.11, p. 8 fig. 11. Figure 10.5. Excavation loci in the Jinsha site. After Zhu et al. 2003, p. 253 fig. 3. 412 Figure 10.6. Aerial view of exposed wall foundations of the large row buildings excavated at Locus Sanhe Huayuan. After Chengdu 2005, p. 2. . Figure 10.7. A pit of elephant tusks at Jinsha, K1. After Chengdu 2005, p. 3. 413 a. b. Figure 10.8. a: Archaeologist Zhu Zhangyi at the locus with a large distribution of stone bi, the Jinsha site. Photograph by author, June 27, 2001. b: large stone bi unearthed in 1927 at Yueliangwan, the Sanxingdui site, outer d. 51.4 cm, inner d. 14 cm. Sichuan University Museum (2.3) 29. After Chengdu 2006, p. 13 pl. 8. 414 . a. Figure 10.9. b. Stone tigers. a: Sanxingdui, 99GSZYT111(17), excavated in 1999 from a ditch next to a wall at Yueliangwan. Photography by author. b: Jinsha, 2001CQJC:684, h. 21.5 cm, l. 28.8 cm. After Beijing 2002c, p. 185. a. b. Figure 10.10. Stone coiled snakes. a: Sanxingdui, locus unknown. Photograph by author, July 29, 2002. b: Jinsha, 2001CQJC:719, l. 41.8 cm. After Beijing 2002c, p. 189. 415 Figure 10.11. Stone kneeling human figure from Jinsha, 2001CQJC:716, h. 21.72 cm. After Beijing 2002c, p. 177. Figure 10.12. Bronze standing human figure from Jinsha, 2001CQJC:17, h. 19.6 cm. After Beijing 2002c, p. 43. 416 a. b. Figure 10.13. Gold masks. a: Sanxingdui, K1:282, h. 11.3 cm, weight 10.62 g. After Beijing 1999a, p. 61 fig. 34.2. b: Jinsha, 2001CQJC:465, h. 3.74 cm, weight 5 g. After Beijing 2002c, p. 22. a. b. Figure 10.14. a: Jinsha, gold circular ornament, 2001CQJC:477, d. 12.5 cm, weight 20 g. After Beijing 2002c, p. 30. b: Jinsha, bronze handled collared disk, 2001CQJC:588, d. 10.24 10.36 cm. After Beijing 2002c, p. 61. 417 Figure 10.15. a: bronze fragment, part of a vessel (zun or lei)’s foot ring, from Jinsha, 2001CQJT8105(7):34, h. 12.3 cm. Style Va, 12th century BC. Photograph by author, July 1, 2002. b: bronze fragment, animal head for the swing handle of a vessel (probably you), from Jinsha, 2001CQJT8406(6):2, h. 4 cm, l. 4 cm. early 10th century BC. Photograph by author, July 1, 2002. 418

Log In

THE SANXINGDUI SITE: ART AND ARCHAEOLOGY VOL. 1 (TEXTS) A DISSERTATION PRESENTED TO THE FACULTY OF PRINCETON UNIVERSITY IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY RECOMMENDED FOR ACCEPTANCE BY THE DEPARTMENT OF ART AND ARCHAEOLOGY

THE SANXINGDUI SITE: ART AND ARCHAEOLOGY VOL. 1 (TEXTS) A DISSERTATION PRESENTED TO THE FACULTY OF PRINCETON UNIVERSITY IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY RECOMMENDED FOR ACCEPTANCE BY THE DEPARTMENT OF ART AND ARCHAEOLOGY

Related Papers

RELATED PAPERS