University of Southern California Dissertations and Theses

Projecting wishes onto flying banners: materiality and painted banners from Cave 17 in the Mogao Caves of Dunhuang, China, in the ninth and tenth centuries

(USC Thesis Other)

Projecting wishes onto flying banners: materiality and painted banners from Cave 17 in the Mogao Caves of Dunhuang, China, in the ninth and tenth centuries

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

PROJECTING WISHES ONTO FLYING BANNERS:
MATERIALITY AND PAINTED BANNERS FROM CAVE 17 IN THE MOGAO CAVES OF
DUNHUANG, CHINA, IN THE NINTH AND TENTH CENTURIES

by
Yoon Ah Hwang

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
ART HISTORY

August 2021

Copyright 2021 Yoon Ah Hwang
ii

Acknowledgements

I could not have completed this journey without the guidance of my dissertation
committee members. Professor Sonya Lee as my dissertation advisor has truly been supportive of
and enthusiastic about my work. Her expertise in the field of Chinese art, her on-point questions,
and her useful critique have helped me to cross the finish line. Professors Rongdao Lai and Ann
Marie Yasin have inspired me to think outside of the box by bringing their expertise and
introducing innovative ways of thinking and methodologies.
I would also like to take a moment to thank those who helped me along the way,
especially during the research stage. While I stayed in London for the Stein Collection at the
British Museum, Professor Roderick Whitfield and Professor Youngsook Pak shared their
valuable comments and experiences during my in-person viewing session. Yu-ping Luk, curator
at the British Museum, was willing to help in checking conservation records for Stein Painting
52. Valerie Zaleski and Pierre Cambon at the Musée Guimet not only helped me to examine
paintings from Cave 17 but also answered my endless questions during my visit. Susan Whitfield
also helped me to examine the paintings on hemp in the British Library. My special thanks are
also extended to Haewon Kim at the National Museum of Korea, who shared her insights with
me during my in-person viewing session of paintings from the Ōtani Collection. I furthermore
owe a debt of gratitude for the kindness and support I received from Professor Sha Wutian in
Dunhuang, who arranged my on-site visit and gave me the opportunity to present my research at
a conference at the Dunhuang Academy. I also thank Neil Schmid for his helpful advice and
questions regarding my dissertation. His on-point questions and suggestions contributed to
shaping my dissertation. Wang Ruilei also shared her knowledge and research on Tibetan art.
iii

Conservators and experts in textile, including Jianlan Wang and Jing Han, as well as Professor
Zhao Feng, shared their expertise on the Dunhuang textiles and dye materials used in Dunhuang.
Tang Li, the Chinese studies librarian at USC, has additionally been a reliable resource for
securing any reference I had difficulty accessing.
All in all, I benefitted from the help of my former teachers, professors, and colleagues
both inside and outside of USC. Professors Juhyung Rhie and Chin-sung Chang at Seoul
National University helped structure my critical thinking in the fields of art history and Buddhist
studies. In addition, Professor Lori Meeks and Lisa Kochinski, a fellow teaching assistant, were
greatly supportive during my final year at USC. Rika Hiro and Xuejing Sun enlivened my
graduate experience with their emotional support and cheers.
I was additionally able to travel for research and write this dissertation thanks to the
fellowships and grants I received from the USC Graduate School and Department of Art History.
I particularly thank the faculty and staff at the USC Art History Department for their generous
support and encouragement during my past few years at USC.
Last but not least, I would like to dedicate this dissertation to my family: my
grandparents; my parents and in-laws; my brother; my brother- and sisters-in-law; my husband,
Taehwan Kim; and our two lovely daughters, Claire and Ellin. I am forever indebted to them for
their endurance and unparalleled love.

iv

TABLE OF CONTENTS

Acknowledgements.........................................................................................................................ii
List of Figures .................................................................................................................................v
Abbreviations.................................................................................................................................xii
Abstract ..........................................................................................................................................vi
Introduction .....................................................................................................................................1
0.1. Lifecycle and Definition of a Painted Banner..........................................................5
0.2. Methodology..........................................................................................................17
0.3. Definition of Materiality........................................................................................18
0.4. Previous Scholarship..............................................................................................21
0.5. Outline of Chapters................................................................................................25

Chapter 1: Artistic Labor and Materials of Stein Painting 52.......................................................27
1.1. Introduction............................................................................................................27
1.2. Stein Painting 52 and its Support Material............................................................31
1.3. Silk’s Characteristics and Painting Techniques.....................................................37
1.4. Stein Painting 52’s Pigments and Red Lines.........................................................39
1.5. Framing and Mounting..........................................................................................43
1.6. Displaying and Hanging Method...........................................................................50
1.7. Skilled Labor in Dunhuang and Artistic Creativity...............................................52
1.8. Conclusion.............................................................................................................59

Chapter 2: Stylistic Variations and Materialistic Differences in Avalokiteśvara Sutra Paintings 61
2.1. Introduction............................................................................................................61
2.2. Cave 112: Changes in the Cave’s Design and Decoration................................... 67
2.3. Wall Paintings of the Avalokiteśvara Sutra in the Mogao Caves..........................77
2.4. Avalokiteśvara Sutra Painting in a Painted Banner Format...................................84
2.5. Stylistic Variations Based on Material Characteristics..........................................88
2.6. Conclusion..............................................................................................................94

Chapter 3: Painted Long Banner’s Meaning and Function........................................................... 96
3.1. Introduction............................................................................................................96
3.2. Stein Painting 216 and its Reconstruction...........................................................101
3.3. Sharing the Form: A Comparison with Daoist Examples....................................108
3.4. Buddhist Banners for Longevity and Soul-Saving: Stein Painting 216...............115
v

3.5. Religious Efficacy and Its Display as Double-Sided Painting.............................124
3.6. Conclusion...........................................................................................................129
Chapter 4: Silk as Devotional Offerings and the Tibetan Impact on Painted Banners................132
4.1. Introduction..........................................................................................................132
4.2. Textiles as an Offering.........................................................................................142
4.3. Silk Used in a Funeral Practice and the Meaning of Silk Thread........................140
4.4. The Origin of a Painted Banner and the Tibetan Influence.................................146
4.5. Funerary Practice in Tibet and Dunhuang and the Yinlu Bodhisattva................162
4.6. Conclusion...........................................................................................................177
Conclusion ..................................................................................................................................179
Figures ...................................................................................................................................….184
Bibliography ...............................................................................................................................261
Appendix......................................................................................................................................276

vi

List of Figures

Figure 1: Map of Northwest China (Image from Hansen, The Silk Road, 22–23). .................... 184
Figure 2: Stein’s Plan of Cave 16 and Cave 17 (gray area labeled as “Walled-up chapel”) (Image
from Stein, Serindia 3, Plans: 43). .................................................................................. 185
Figure 3: Bodhisattva with Glass Beaker, Stein Painting 120 (1919,0101,0.120), Tang dynasty,
9th or 10th century CE, ink and colors on silk, H. 58.0 cm, W. 18.0 cm, British Museum
(Image from Whitfield, The Art of Central Asia 1, Plate 28.). ....................................... 186
Figure 4: Portrayal of the Life of Buddha, Maya Cave, Kizil, Xinjiang (Image from von Le Coq,
Die Buddhistische Spätantike in Mittelasien 3, Tafel A.a.). ........................................... 187
Figure 5: Drawing of dhāraṇī pillars as a cloth banner, south wall of Cave 217, Mogao Caves,
Dunhuang, Gansu Province, China (Image from Shimono, “Tonkō Bakukōkutsu dai 217
kutsu,” figures 10e-f). ..................................................................................................... 188
Figure 6: Panel with confronting horses (a) and its lining of dark-red/brown silk (b), late 8th–9th
century CE, length (a): 67.5 cm; width (a): 52.8 cm, length (b): 66 cm; width (b): 55 cm,
MAS. 885 a-b, British Museum (Image from Zhao, Textiles from Dunhuang in UK
Collections, 196). ............................................................................................................ 189
Figure 7: Seated Avalokiteśvara, Stein Painting 52 (1919,0101,0.52), 972 CE, ink and colors in
silk, H. 91.5 cm x W. 59.1 cm (painted area), H. 102.2 cm x W. 69.0cm (with frame),
British Museum (Image downloaded from http://britishmuseum.org). .......................... 190
Figure 8: Comparison of Stein Painting 52 and Stein Painting 103 showing different directions of
the silk across painting workshops (illustration made by author). .................................. 191
Figure 9: Painted Banner of Avalokiteśvara (detail), Stein Painting 154 (1919,0101,0.154), 10th
century CE, ink and colors on hemp, H. 94 cm x W. 27 cm, British Museum (Image
downloaded from http://britishmuseum.org). ................................................................. 192
Figure 10: Left: Standing Avalokiteśvara, 984 CE, ink and colors on hemp, W. 64.5 cm x H. 191
cm, MG. 22799, Musée Guimet; Right: Reconstructed image of the painting framed on a
wooden structure (Image from International Dunhuang Project,
http://idp.bl.uk/database/oo_scroll_h.a4d?uid=615818696;recnum=91351;index=1,
accessed November 2, 2016; Garcia, “‘La quête des mérites’,” 55). ............................. 193
Figure 11: Representation of a painting mounted on a drawing frame, line drawing of the south
wall of Cave 72, Dunhuang, mid-9th to 10th century CE (Image from Huo, “Mogaoku Di
72 Ku,” 38, fig. 6). .......................................................................................................... 194
vii

Figure 12: Seated Avalokiteśvara, Stein Painting 54 (1919,0101,0.54), 983 CE, ink and colors on
silk, H. 99.0 cm x W. 59.5 cm (painted area), H. 102.0 cm x W. 75.5 cm (with frame),
British Museum (Image from Whitfield, The Art of Central Asia 2, Pl. 27). ................. 195
Figure 13: Avalokiteśvara as a Savior of Perils, Stein Painting 2 (1919,0101,0.2), 10th century
CE, ink and colors on silk, H. 84.4 cm x W. 61.7 cm, British Museum (artwork in the
public domain; photograph all rights reserved, British Museum). ................................. 196
Figure 14: Avalokiteśvara of the Water Moon, 968 CE, ink and colors on silk, H. 106.8 cm x W.
58.9 cm, Freer | Sackler Gallery, Washington DC (Image from
https://www.freersackler.si.edu/object/F1930.36/#txtMetaData). .................................. 197
Figure 15: Embroidery from the original framing of Stein Painting 52, British Museum (Image
from http://britishmuseum.org). ...................................................................................... 198
Figure 16: The original framing of Stein Painting 52 (Image from Stein, Serindia 4, Pl. LXI). 199
Figure 17: Hanging scroll in typical Chinese style, front and back (Image from Valerie Lee,
Xiangmei Gu, and Yuan-Li Hou, “The Treatment of Chinese Ancestor Portraits,” 465,
fig. 1). .............................................................................................................................. 200
Figure 18: Cross section of a traditional Chinese scroll showing the layers of backing (Image
from Valerie Lee, Xiangmei Gu, and Yuan-Li Hou, “The Treatment of Chinese Ancestor
Portraits,” 466, fig. 2). .................................................................................................... 201
Figure 19: Typical Tibetan thangka frame (Image from Jackson and Jackson, Tibetan Thangka
Painting, 143). ................................................................................................................ 202
Figure 20: Padmasambhava and its reverse showing how the painting is framed without covering
the backside of the support, ground mineral pigment, red background on cotton. 17th
century CE, Field Museum of Natural History (Image from Himalayan Art Resources,
https://www. himalayanart.org/items/54358, accessed January 3, 2018). ...................... 203
Figure 21: Large patchwork, late 8th–9th century CE, various weaves in silk, 150.5 cm x 111 cm,
MAS. 856, British Museum, London (Image from Zhao ed., Textiles from Dunhunag in
the UK Collections, 113, Pl. 79). .................................................................................... 204
Figure 22: Border for a votive painting, 7th–10th century CE, silk, outside frame: 84.0 cm x
63.5), inside frame: 72.5 cm x 48.0 cm, MG. 26789, Musée Guimet, Paris (Image from
Zhao ed. Textiles from Dunhuang in French Collections, 124, Pl. 72). ......................... 205
Figure 23: The Medicine Buddha sutra transformation tableaux, early Tang dynasty, south wall
of Cave 220, Mogao Caves, Dunhuang (Image from Dunhuang wenwu yanjiusuo,
Zhongguo shiku: Dunhuang Mogaoku 3, Pl. 24). ........................................................... 206
viii

Figure 24: Central niche on the west wall of Cave 112, middle Tang dynasty, Mogao Caves,
Dunhuang (Image from Dunhuang yanjiu yuan, and Jiangsu meishu chubanshe, eds.
Mogao ku di 112 ku, 18, Pl. 1). ....................................................................................... 207
Figure 25: Layout of sutra paintings in Cave 112 (illustration by author). ................................ 208
Figure 26: Visualization Sutra Painting (left section) and Diamond Sutra Painting (right section),
south wall of Cave 112, Middle Tang dynasty, Mogao Caves, Dunhuang (Image from
Dunhuang yanjiu yuan, and Jiangsu meishu chubanshe, Mogao Ku Di 112 Ku, 59, Pl.
50). .................................................................................................................................. 209
Figure 27: Repaying Kindness Sutra (left section) and the Bhaisajyaguru Sutra (right section),
north wall of Cave 112, Middle Tang dynasty, Mogao Caves, Dunhuang, China (Image
from Dunhuang yanjiu yuan, and Jiangsu meishu chubanshe, Mogao ku di 112 ku, 130,
Pl. 128). ........................................................................................................................... 210
Figure 28: Avalokiteśvara Sutra (left section), Maitreya’s Subjugation of the Demons (above the
entrance), and Maitreya (right section), east wall of Cave 112, Middle Tang dynasty,
Mogao Caves, Dunhuang, China (Image from https://www.e-
dunhuang.com/cave/10.0001/0001.0001.0112, accessed March 1, 2015). .................... 211
Figure 29: Top: pomegranates with vines (shiliu juancao wen !"#$%), detail of Figure 26,
Middle Tang dynasty, Cave 112; Bottom: the same pattern from Cave 158, Middle Tang
dynasty (Images from Dunhuang yanjiuyuan, Dunhuang shiku quanji 14, Pl. 133). ..... 212
Figure 30: Camellia flowers with rolling vines (chanzhi chahua wen &'()%) (Image from
Dunhuang yanjiuyuan, Dunhuang shiku quanji 14, Pl. 136). ......................................... 212
Figure 31: Diamonds/rhombi (lingge wen *+%), detail of Figure 24, Middle Tang dynasty,
Cave 112. ........................................................................................................................ 212
Figure 32: Top: tortoiseshells (guijia wen ,-%), detail of Figure 27, Middle Tang dynasty,
Cave 112; Bottom: the same pattern from Cave 323, west niche, High Tang dynasty
(Image from Dunhuang yanjiuyuan, Dunhuang shiku quanji 14, Pl. 74). ...................... 213
Figure 33: Top: quatrefoil alternating with paired halved flowers (lingxing siye wen *./0
%), detail of Figure 27, Middle Tang dynasty, Cave 112); Bottom: the same pattern from
Cave 26, Yulin Caves, Song dynasty (Image from Dunhuang yanjiuyuan, Dunhuang
shiku quanji 14, Pl. 219). ................................................................................................ 213
Figure 34: Six-petal flower pattern alternating with a pair of halved flowers (liuyehua tuan 10
)23), detail of Figure 26, Middle Tang dynasty, Cave 112. ..................................... 213
Figure 35: Schematic illustration of decorative patterns in Cave 112 (illustration by author). .. 214
ix

Figure 36: Photograph showing diamond pattern around the western niche (left) and no enclosing
patterned borders around the Amitābha Sutra painting (right) on the north wall of Cave
322, Early Tang dynasty, Mogao Caves, Dunhuang (Image downloaded from
https://www.e-dunhuang.com). ....................................................................................... 215
Figure 37: Avalokiteśvara as Savior of Perils, Stein Painting 28* (1919,0101,0.28*), ink and
colors on silk, H. 83.3 cm x W. 63.1 cm, British Museum (Image downloaded from
http://idp.bl.uk/database/oo_scroll_h.a4d?uid=7675208;bst=1;recnum=40339;index=1;im
g=1). ................................................................................................................................ 216
Figure 38: Paradise of Amitābha, Stein Painting 203 (1919,0101,0.203), ink and color on hemp,
H. 95.5 cm x W. 49.5 cm, British Museum (Image downloaded from
http://britishmuseum.org). .............................................................................................. 217
Figure 39: Fragments with concentric lozenge design, дx46, damask on plain-weave silk, Tang to
Five Dynasties period, a. 7.7 cm x 9.5 cm, b. 10.0 cm x 17.0 cm, State Hermitage
Museum, Russia (Image from Zhao, Textiles from Dunhuang in Russian Collections, Cat.
46.). ................................................................................................................................. 218
Figure 40: Ribbon with diamond pattern, MAS 909, silk braid, High Tang dynasty, late 7th to
early 8th century, 19 cm x 2.5 cm, British Museum (Image from Zhao, Textiles from
Dunhuang in UK Collections, Cat. 62). .......................................................................... 218
Figure 41: Banner head with florets, Дx49, plain weave in silk, clamp-resist dyed, late Tang to
Five Dynasties period, 9th to 10th century, length: 19.7 cm, width: 36.5 cm, State
Hermitage Museum (Image from Zhao, Textiles from Dunhuang in Russian Collections,
61). .................................................................................................................................. 219
Figure 42: Bodhisattva Avalokiteśvara as a Savior of Perils, 1919,0101,0.243, Five Dynasties
period, 10th century CE, woodblock print, 32 cm x 20 cm, British Museum (Image
downloaded from http://www.britishmuseum.org/). ...................................................... 220
Figure 43: Avalokiteśvara Sutra painting (detail), Middle Tang dynasty, east wall of Cave 112
(Image from Dunhuang yanjiu yuan, and Jiangsu meishu chubanshe, Mogao ku di 112 ku,
196, Pl. 196). ................................................................................................................... 221
Figure 44: Identification of the thirteen peril scenes, Avalokiteśvara Sutra painting, east wall of
Cave 112, Mogao Caves, Dunhuang (The background image’s sharpness, contrast, and
brightness have been modified by the author to increase visibility). .............................. 222
Figure 45: Avalokiteśvara Sutra painting, High Tang dynasty, south wall of Cave 45, Mogao
Caves, Dunhuang (Image from Dunhuang wenwu yanjiusuo, Zhongguo shiku: Dunhuang
Mogaoku 3, Pl. 131). ....................................................................................................... 223
x

Figure 46: Avalokiteśvara Sutra painting, High Tang dynasty, photography taken 1943–1944,
south wall (center, upper section) of Cave 205, Mogao Caves, Dunhuang (Image
downloaded from James and Lucy Lo Photograph Archive, www.artstor.org). ............ 224
Figure 47: Avalokiteśvara as a Savior of Perils, Stein Painting 24 (1919,0101,0.24), 963 CE, ink
and colors on silk, H. 107.0 cm x W. 61.5 cm, British Museum (Image downloaded from
https://www.britishmuseum.org/collection/object/A_1919-0101-0-24). ....................... 225
Figure 48: Avalokiteśvara as Savior of Perils, 950–1000 CE, ink and colors on silk, H. 84.1 cm x
W. 61.2 cm, MG. 17665, Musée Guimet (Image downloaded from
http://idp.bl.uk/database/oo_scroll_h.a4d?uid=132366109;recnum=91290;index=1). .. 226
Figure 49: Eleven-headed Avalokiteśvara, 985 CE, ink and colors on silk, H. 99.6 cm x W. 62.8
cm, 1943.57.14, Harvard Art Museums (Image downloaded from
http://www.harvardartmuseums.org/art/204072). ........................................................... 227
Figure 50: Standing Bodhisattva, Stein Painting 22 (1919,0101,0.22), Tang dynasty, ink and
color on silk, H. 103.1 cm x W. 71.2 cm, British Museum (Image from Whitfield, The Art
of Central Asia 2, 3). ....................................................................................................... 228
Figure 51: Avalokiteśvara, Stein 391, 925 or 865 CE, ink and colors on silk. Nm 2003-17-352,
National Museum, New Delhi (Image from Chandra and Sharma, Buddhist Paintings of
Tun-Huang in the National Museum, New Delhi, 123 Pl. 34). ....................................... 229
Figure 52: Vajrapāni, Stein Painting 103, early 9th century CE, ink and colors on silk, H. 55 cm,
W. 14.5 cm, British Museum (Image downloaded from http://britishmuseum.org). ..... 230
Figure 53: Avalokiteśvara in different media: details of Figures 42, 49, and 50. ....................... 231
Figure 54: Banners being utilized in a temple (red arrows indicate a pair of banners being
displayed within a shrine), east wall of Cave 148, Mogao Caves, High Tang dynasty
(Image from Dunhuang yanjiuyuan, Dunhuang shiku quanji 21, 168, Pl. 159). ............ 232
Figure 55: Long Banner of Bodhisattvas, Stein Painting 216 (1919,0101,0.216), 956 CE, ink and
colors on silk, H. 396.0 cm x W. 59.0 cm, British Museum (Image downloaded from
http://britishmuseum.org). .............................................................................................. 233
Figure 56: Detail of Figure 55 showing thick threads to prevent the cut silk from fraying
(photograph by author). .................................................................................................. 234
Figure 57: Detail of Figure 55 showing the inscription on Stein Painting 216 (Image downloaded
from http://britishmuseum.org). ...................................................................................... 235
xi

Figure 58: Preaching Shakyamuni, Stein Painting 6, 1919,0101,0.6, early 8th century CE, ink
and colors on silk, H. 139.0 cm x W. 101.7 cm, British Museum (Image downloaded
from http://britishmuseum.org). ...................................................................................... 236
Figure 59: Long Banner with Bodhisattvas, Stein Painting 196 (1919,0101,0.196), 956 CE (?),
ink and colors on silk, H. 341.0 cm x W. 28.5 cm, British Museum (Image downloaded
from http://britishmuseum.org). ...................................................................................... 237
Figure 60: Long Banner of Bodhisattvas, Bon 4025, 10th century CE, ink and colors on silk,
National Museum of Korea (Image from Kim, Kungnip chungang bangmulgwan sojang
chungang ashia chonggyo hoehwa, 153, Pl 43). ............................................................ 238
Figure 61: Left to right: Comparison of the painting styles of Bon 4025 (fig. 60), Stein Painting
196 (fig. 59), and Stein Painting 216 (fig. 55). ............................................................... 239
Figure 62: Reconstructions of Stein Paintings 196 and 216, and Bon 4025 (reconstructed by
author). ............................................................................................................................ 240
Figure 63: Diagram of a Daoist “spirit-moving banner” (Image from Huang, “Daoist Uses of
Color in Visualization and Ritual Practices,” 230, fig. 7). ............................................. 241
Figure 64: Long Banner, Stein Painting 217 (1919,0101,0.217), silver on woven damask red silk,
H. 386.0 cm x W. 68.1 cm, British Museum (Image downloaded from
http://britishmuseum.org). .............................................................................................. 242
Figure 65: Streamer of Avalokiteśvara, Д x 291, plain-weave in silk damask on plain weave in
silk; painted in yellow, late Tang to Five Dynasties period, H. 48.5 cm x W. 37.6 cm,
State Hermitage Museum of Russia (Image from Zhao, Textiles from Dunhuang in Russia
Collections, 76, Pl. 9). ..................................................................................................... 243
Figure 66: Unidentified name of a bodhisattva on Bon 4025 (fig. 60), National Museum of Korea
(photograph by author). .................................................................................................. 244
Figure 67: Long Banner, EO 3647, 10th century CE, ink and colors on blue silk, H. 580.0 cm x
W. 32.9 cm, Musée Guimet (Image from Giès, Art of Central Asia 2, Pl. 46). .............. 245
Figure 68: Banner with Bodhisattva, possibly Mahamayuri, Tang dynasty or Five Dynasties
period, 9th–10th century, China (Dunhuang area, Gansu Province), ink and pigment on
silk, H. 57.0 cm x W. 28.0 cm, Metropolitan Museum of Art (2007.294 a, b) (Image from
https://www.metmuseum.org) ........................................................................................ 246
Figure 69: Painting on north wall of Cave 231 showing a flying banner hung on a dragon-headed
staff, 8th to 9th century CE, Mogao Caves, Dunhuang, China (Image from Dunhuang
yanjiuyuan, Dunhuang shiku quanji 21, 168, Pl. 159). .................................................. 247
xii

Figure 70: Dragon-headed banner stand (Image downloaded from
http://data.kdata.kr/resource/Yongdubodang, accessed 2018). ....................................... 248
Figure 71: Silk banner from Mawangdui Tomb No. 1, Western Han dynasty, ink and colors on
silk, H. 205.0 cm x W. 92.0 cm, Hunan Provincial Museum, China (Image downloaded
from Hunan Provincial Museum,
http://61.187.53.122/collection.aspx?id=1348&lang=zh-CN). ...................................... 249
Figure 72: Banner showing Fuxi and Nüwa, Ast(ana) ix. 2. 054, Turfan (Image from Stein,
Innermost Asia 3, Pl. CIX). ............................................................................................. 250
Figure 73: Image showing a wooden peg installed in a hole, south wall of Cave 290, Northern
Zhou dynasty, Mogao Caves, Dunhuang, China (Image from Dunhuang wenwu
yanjiusuo, Zhongguo shiku: Dunhuang Mogaoku 1, plate 181). .................................... 251
Figure 74: Bodhisattva with Lotus, Stein Painting 102 (1919,0101,0.102), early 9th century CE,
ink and colors on silk, H. 51.0 cm x W. 14.0 cm, British Museum (Image downloaded
from http://britishmuseum.org). ...................................................................................... 252
Figure 75: Bodhisattva with Lotus, Stein 535, ink and color on silk, National Museum, New
Delhi (Image from Chandra and Sharma, Buddhist Paintings of Tun-Huang in the
National Museum, New Delhi, 223, Pl. 97). ................................................................... 253
Figure 76: Avalokiteśvara, 11
th
century CE (?), H. 77.5 cm x W. 23.5 cm, Yarlung Museum,
Tsethang (Image from Lojda, Klimburg-Salter, and Strinu, “The Tibetan Himalayan
Style,” 141, figure 4.9). ................................................................................................... 254
Figure 77: Painted cloth textile placed between beams, Main Temple, Assembly Hall, Alchi
Monastery (Image from Klimburg-Salter and Luczanits, Tabo, 174 and fig. 193). ....... 255
Figure 78: Painted cloth textiles placed between beams, Main Temple, ambulatory, ceiling, Alchi
Monastery (Image from Klimburg-Salter and Luczanits, Tabo, 175 and fig. 196). ....... 256
Figure 79: Paradise of Bhaisajyaguru, Stein Painting 32 (1919,0101,0.32), 836 CE, ink and
colors on silk, H. 152.3 cm x W. 177.8 cm, British Museum (Image downloaded from
http://idp.bl.uk/). ............................................................................................................. 257
Figure 80: Yinlu Bodhisattva (?), Tang dynasty, photograph taken 1943–1944, west wall (lower
right) of Cave 205 (Image downloaded from James and Lucy Lo Photograph Archive,
www.artstor.org). ............................................................................................................ 258
Figure 81: Yinlu Bodhisattva, EO 1133, 10th century CE, ink and colors on silk, H. 59.0 cm x W.
38.0 cm, Musée Guimet (Image downloaded from http://idp.bl.uk/). ............................ 259
xiii

Figure 82: Line drawings of funerary scenes on coffin plank no. 2, Tomb M1, 8th century CE,
Guolimu (Image from Bellezza, Death and Beyond in Ancient Tibet, 247 and fig. 16). 260
Figure 83: Ritual implement, H. 18.2 cm x W. 11 cm, glued paper layers on a wooden stick, IOL
Tib J 1364, British Library (Image downloaded from
http://idp.bl.uk/database/oo_scroll_h.a4d?uid=5719393746;recnum=22865;index=5). 261
xiv

Abbreviations

D. The Dunhuang Collection at Peking University’s library, Beijing, China.
DZ Daozang 45, Sanjia ben 678 ed. 36 vols. Beijing: Wenwu chubanshe, 1988.
P. Pelliot Collection of Chinese manuscripts from Dunhuang, now stored at the
Bibliothèque National in Paris, France.
S. Stein Collection of Chinese manuscripts from Dunhuang, now stored at the
British Library in London, United Kingdom
T. Taishō shinshū daizōkyō 9:;<95=, ed. Takakusu Junjirō>?@AB et
al. 85 vols. Tokyo: Taishō issaikyō kankōkai 95CDE, 1924–1933. Also in
CBETA electronic database http://www.cbeta.org/index.html.

xv

Abstract

This dissertation explores the ways in which Buddhist painted banners were produced
and utilized during the ninth and tenth centuries in northwestern China in relation to other oases
and kingdoms along the ancient Silk Road. The main objects of the study are painted banners
that were uncovered from the Library Cave (Cave 17) at the Mogao Caves of Dunhuang. These
pictorial materials provide powerful evidence of religious practices, artistic production, and
material culture, thereby complementing and supplementing the ancient manuscripts from
Dunhuang that have attracted far more attention from scholars in Buddhist studies. This project
focuses on the material aspects of the paintings, which reveal not only the circulation of textile
goods and pigments but artistic practices and labor, from weaving textiles to the display of the
final votive works. By concentrating on the works’ materiality, artistic labor, and design
schemes, I argue that the fundamental characteristics of painting materials contributed not only to
the works’ formal appearance but to their visual effects and religious efficacy. The dissertation
scrutinizes the artistic labor and material characteristics of the banners painted on silk and hemp
based on close observation of the paintings’ construction. In addition, it compares the murals of
the Mogao Caves with banners painted during the Tibetan occupation period (787–848 CE).
With mural and portable paintings on textile depicting an Avalokiteśvara Sutra painting, I
demonstrate that artisans often adjusted their styles and compositional designs to conform to the
distinctive material qualities of the medium at hand. My analysis of mural and banner specimens
sharing the same theme leads to the idea of the functionality of a long banner as a votive object.
Finally, the connotations of silk in the region of northwestern China shed light on how patrons’
wishes became materialized in the works’ formal appearances.
1
Introduction
The Mogao Caves of Dunhuang in Gansu Province, China, are one of the most important
repositories of Buddhist sculptures and wall paintings in the world (fig. 1). Consisting of 492
caves constructed from the fifth to fifteenth centuries CE, the site attracted the modern
academy’s attention with the historic and unexpected discovery of the hidden Cave 17 by a local
caretaker named Abbot Wang Yuanlu FGH (c. 1834–1931) in 1900. The small unit, which
came to be known as the “Library Cave” (fig. 2), yielded some of the earliest extant documents
and artistic works in Asia, including tens of thousands of Buddhist manuscripts, records from
everyday life, religious paintings, and textiles from China’s medieval period.
1
This historic find
has reshaped scholarship in Chinese art history, as many of the objects from the Library Cave of
Dunhuang represent different ways of life, mindset, and cultural practice. The discovery also
verifies the advanced Buddhist thoughts and ritual practices of the ninth to tenth centuries, which
had evolved from the earlier Buddhism in Dunhuang since the fifth century. This dissertation
spotlights painted banners and hanging paintings uncovered from Cave 17, most of which have
remained understudied since their discovery in 1900. These pictorial materials provide powerful
evidence of religious practices, artistic production, and material culture in northwestern China,

1. Following Abbot Wang’s discovery of Cave 17, words of the cave and its contents reached European
archaeologists who were eager to explore Mogao Caves despite the long distance and difficulties of travel
from Europe to the Silk Road region. In 1907, Hungarian-born British archaeologist Aurel Stein (1862–
1943) arrived in Dunhuang. He succeeded in persuading Wang to sell him over 300 bundles, comprised of
3,000 Chinese and Tibetan Buddhist texts. The French sinologist Paul Pelliot (1878–1945) was also able
to obtain a large quantity of manuscripts that Stein could not take with him. The manuscripts, textiles, and
documents these two scholar-explorers brought to Europe entered museum collections in London and
Paris. For more on the history the Stein Collections in the UK, see Helen Wang, Helen Persson, and
Frances Wood, “Dunhuang Textiles in London: A History of the Collection,” in Sir Aurel Stein:
Colleagues and Collections, ed. Helen Wang (London: British Museum Research Publication, 2012), 1–8.
2
thereby complementing and supplementing ancient manuscripts from Dunhuang that have
attracted far more attention from scholars in Buddhist studies.
The focus of this dissertation will be on the material aspects of the banner paintings
found in Cave 17, which reveal crucial information about the circulation of textile goods and
pigments as well as artistic labor and religious meaning. By concentrating on the works’
materiality and design schemes in relation to the ritual practice of painted banners from the ninth
and tenth centuries, I demonstrate that the banner paintings’ materials—both the textiles from
which the banners were constructed and the pigments with which they were painted—were
chosen not only because of how they were perceived to impact the works’ formal appearance but
also because of the beliefs in religious efficacy of the objects connected to the materials used.
My dissertation also sheds new light on how material culture in Dunhuang shaped the ways in
which the Buddhist laity, from the lower to middle classes, utilized textiles in their offerings. The
findings will contribute to understanding the merit of using textiles in Buddhist rituals and
related funerary practices, which has remained a vital part of the contemporary practice of gifting
textiles as offerings.
Devotional offerings made with silk textiles comprise the majority of the pictorial art
works from Cave 17. Why was silk considered particularly suitable for the making of these
devotional objects? Their overall structure and the ways in which they were framed are
distinguishable from two standard formats in traditional Chinese painting, namely, handscrolls
and hanging scrolls.
2
Some examples from the cave include a rectangular-shaped silk banner

2. In addition, albums, fans, and screens are also widely used. Jerome Silbergeld also adds “altered
format,” which includes “screen paintings turned into hanging scrolls” and “albums leaves [that] were
remounted as handscrolls.” Jerome Silbergeld, Chinese Painting Style: Media, Methods, and Principles of
Form (Seattle and London: University of Washington Press, 1982), 11–15.
3
framed with multiple pieces of textile and suitable for hanging. Others showcase a different type
that was decorated with a triangular top piece, side pieces, and streamers at the bottom. A few
examples are extremely long for hanging paintings. All these examples, in short, did not follow
conventional structures, display methods, and material usage. They also would have required
different concepts and practices of artistic labor and craftsmanship in their production. Therefore,
the pictorial art from Cave 17 exemplifies an altogether different understanding of painting in
China, one that was distinct from the more familiar genres of landscape, figure, or literati
painting. Moreover, the corpus of paintings from the Mogao Caves at Dunhuang were made and
offered within the context of Buddhist merit-making practices. The works were thought to be
able to perform karmic deeds that could bring favorable rewards to their donors. To what extent,
then, did the textile material contribute to its “merit” and even to the purported religious
efficacy? In other words, what material qualities make the textile suitable for such religious
offerings?
Even though silk was a common and primary painting material for banner paintings, the
geographical remoteness of Dunhuang from central China prompts one to wonder about their
sources. Where were the textiles—particularly silk—originally produced? Was such a unique
framing practice with multiple pieces of textiles local to Dunhuang, or was it adopted from
somewhere else? Lastly, what do these many textile offerings signify regarding Buddhist
material culture and rituals? I raise these fundamental issues regarding the origin of painted
banners in conjunction with the phenomenon of producing Buddhist images in banner format in
large quantities during the ninth and tenth centuries in Dunhuang. In short, the correlation
between the material characteristics of devotional objects and religious practice is this
dissertation’s main subject of inquiry.
4
To answer the questions raised above, I first focus on the essential painting materials,
namely, the textiles and their qualities. Textiles could serve multiple functions as painting
surfaces and as mounting materials. In addition, the portability, lightness, affordability, monetary
value, and symbolic meaning of silk textiles could satisfy the demands of devotee-patrons who
came from different economic and social backgrounds. My key argument is that textiles—silk in
particular—played a pivotal role in conveying and transmitting religious concepts and artistic
practices between Dunhuang and its neighboring oasis kingdoms and empires such as Turfan and
Tibet. This approach contrasts with previous scholarship that focuses on silk as a means of
economic exchange. As the modern term “Silk Road” indicates, the pivotal role of silk as a
commodity was highlighted by scholars in social or economic history.
3
Their focus on silk’s
economic value has downplayed the common perception of it as an object of luxury and rarity. In
their interpretations, they have instead redefined silk as a common and everyday object that was
traded for necessities and as a form of tax payment.
The significance of this dissertation becomes clear when one compares it to the study of
the manuscripts found in the same Library Cave of Dunhuang. Historian Hao Chunwen explains
the importance of the written materials from Cave 17:
The Buddhist texts from Dunhuang only account for 350 works in the Kaiyuan
Catalogue, and of these 350 works, only ninety-three scriptures survive completely
intact, with the manuscripts of the other 257 works containing missing portions.
This shows that, compared with the systematic organization of the Buddhist canon
in central China of the time, the Buddhist texts in the Library Cave at Dunhuang do
not represent a systematic collection of Buddhist scriptures. ... As noted above,
compared to the collection of Buddhist texts in central China of the time or the later

3. Helen Wang, Money on the Silk Road: The Evidence from Eastern Central Asia to c. AD 800 (London:
British Museum Press, 2004); “Textiles as Money on the Silk Road?,” Journal of the Royal Asiatic
Society 23, no. 2 (2013): 165–74; Xinru Liu, Silk and Religion: An Exploration of Material Life and the
Thought of People, AD 600-1200 (Oxford University Press, 1999); Richard von Glahn, The Economic
History of China: From Antiquity to the Nineteenth Century (Cambridge: Cambridge University Press,
2016).
5
transmitted canon, the corpus of Buddhist sutras at Dunhuang has numerous gaps,
but it also contains many Buddhist texts that were not included in the transmitted
Buddhist canon. Modern scholars typically call them apocryphal or indigenous
sutras or adopt traditional categories of analysis... such as “suspicious” or “doubtful
scriptures” (yijing I=), “forged classics” (weijing J=), or “lost classics” (yijing
K=). As Buddhist scriptures that were not admitted into the canon, they possess
very historical and scholarly value. ... These Dunhuang materials were not
transmitted to the central regions of China and were restricted to the Northwest.
These “lost classics” can be used to fill in some of the gaps in the extant Chinese
Buddhist canon.
4

This statement by Hao makes clear how important the discovery of the cave and the value
of the “lost classics” are, as they are expected to “fill in some of the gaps in the extant Chinese
Buddhist canon.” This statement would be equally valid if one replaces “Buddhist texts” with
“Buddhist paintings” from the same cave. The pictorial representations realized on the surfaces
of the textiles are equivalent to the “lost classics” of the early painting history in China, as extant
examples from the ninth to the tenth centuries are extremely rare. The main goal of this
dissertation is to present the little-known history of artistic practice and religious practice in
medieval Dunhuang, a remote oasis city in northwestern China.

0.1. Lifecycle and Definition of a Painted Banner
Before investigating the various types of painted banners in detail, it is necessary to
understand the general life of a painted banner prior to it being stored in Cave 17 with hundreds
of other paintings. Based on textual documents from Cave 17, it is believed that there were about
eighteen Buddhist temples in the Dunhuang area during the medieval period, including Jingtu
TempleLMN, Yong’an TempleOPN, Longxing TempleQRN, and Sanjie Temple!"

4. Chunwen Hao, Dunhuang Manuscripts: An Introduction to Texts from the Silk Road trans. Stephen F.
Teiser (Diamond Bar: Portico Publishing Company, 2020), 63, 67–68, and 70.
6
#.
5
They provided religious services and functioned as recipients of religious offerings given by
devotees. When a devotee decided to offer a painting on textile, he or she would approach a
monk at a temple with a workshop to make work to order. Another possible scenario is that a
donor brought his or her painting made elsewhere and presented it to a temple to serve a special
purpose or to accumulate merit.
6
A third possibility was the use of a ready-made piece by a
workshop. Vacant cartouches from the paintings of Cave 17 have supported this theory of
treating a painting as a ready-made votive object. The vacant cartouche was filled by
calligraphers—who were not painters—after being purchased by devotees.
7
Produced upon a
devotee’s request, such paintings were displayed in the interior of a temple or shown outdoors in
case of a long banner (see Chapter 3). Banners were often displayed temporarily for use in
ritualistic settings such as funerals and salvation rituals. Once a painting had served its
designated purpose, it then was stored in a temple’s library or repository.
There has been a heated debate about when and why Cave 17 was sealed in the eleventh
century CE, and, more importantly, about the cave’s main function.
8
If all of the written and
visual materials from Cave 17 originally belonged to the Sanjie Temple, as historian Rong
Xinjiang has argued, it does not explain why there were also documents (e.g., an inventory of a

5. For the complete list of the temples in the Dunhuang region, see Imre Galambos, Dunhuang
Manuscript Culture (Boston: De Gruyter, 2020), 96–97.

6. Stein Paintings 101 through 103 likely exemplify this scenario not only because they have different
styles of standing bodhisattva figures, but also because the paintings’ textile support was cut differently
from most of the other painted banners. These features reinforce those paintings as having been painted
somewhere else (possibly in Khotan) and entered into the collections of the paintings in the Library Cave.
See Chapter 1 for more information on their style and different treatment on the silk support.

7. Aurel Stein, Serindia: Detailed Report of Explorations in Central Asia and Westernmost China 2
(Oxford: The Clarendon Press, 1921), 842.

8. See note 37 in this Introduction.

7
temple) from other temples such as Jingtu, Longxing, or Bao’en. Going beyond both Rong’s
widely quoted theory of the cave being a storeroom of the Sanjie Temple’s library or as a waste
depository, historians Sam van Schaik and Imre Galambos propose that the cave be understood
as a multifunctional space that accommodated not only leftover documents and paintings from
the multiple monasteries but also manuscripts and paintings commissioned by monks, nuns, and
laity that were donated to these monasteries.
9
Based on the overall good condition of the
paintings, they were likely to be displayed for a short period of time or simply received by
donors for merit accumulation with the anticipation of no immediate use.
10

Regardless of what function the Library Cave of Dunhuang might have served, there
were clearly two main types of paintings found inside the cave.
11
The first type is a vertically
oriented, oblong painting consisting of a main body with a triangular “head” on the top and side
streamers attached to this head. It is also common to have long streamers and a weighting board
at the bottom, as in the case of Bodhisattva with Glass Beaker or Stein Painting 120 (fig. 3). The
most popular subject for this vertical oblong type is a single bodhisattva in a standing position.
The second type is a rectangular or square painting rendered on a single piece of cloth with

9. Sam van Schaik and Imre Galambos, Manuscripts and Travelers: The Sino-Tibetan Documents of a
Tenth-Century Buddhist Pilgrim (Boston: De Gruyter, 2011), 25–26.

10. Even if the commissioned paintings and manuscripts with no immediate use were stored in the
monasteries, it does not mean that they were “waste” and acceptable to be thrown away. The way the
paintings were stacked and stored in Cave 17 means they cannot be considered waste: In each bundle,
paintings, manuscripts, and fragments of textiles were bundled together and wrapped with a thick paper.
See Chapter 1 for Stein’s own observation on these bundles.

11. Bhattacharya-Haesner categorizes them into three groups. His first type is a necktie-like banner called
potaka. However, it also can be seen as a part of a valance that is hung around the altar. Thus, in this
paper, I will exclude this type of banner since it has a strong decorative function. See Chhaya
Bhattacharya-Haesner, Central Asian Temple Banners in the Turfan Collection of the Museum für
Indische Kunst, Berlin Painted Textiles from the Northern Silk Route (Berlin: Reimer, 2003), 39–40.

8
several pieces of framing textiles, as seen in Stein Painting 52 (see fig. 7 in Chapter 1). As
mentioned earlier, neither of these paintings’ configuration are compatible with the typical
formats of the hanging scroll or the handscroll in premodern Chinese painting.
Due to the unique formats of the paintings found in Cave 17, they have been named and
defined differently by scholars and researchers. Considering the lack of scholarly consensus on
terminology, it is therefore worth reviewing the few key terms used by scholars and analyzing
their underlying assumptions.
After making the purchase from Abbot Wang, Aurel Stein was the first to handle,
categorize, and study the pictorial materials coming out of Cave 17. In Serindia, Stein’s detailed
report of his findings and collections from his 1906–1908 expeditions, he uses the terms “painted
(silk) banner,” “silk banner,” “silk painting,” and “narrow hanging of silk” in the case of any
painting whose material is silk. The first two terms refer to vertically oriented oblong paintings
with the reference to the material used; the third indicates a rectangular painting often framed
with multiple pieces of textiles; and the last one, “narrow hanging of silk,” refers to a long,
painted banner type.
Stein uses the term “banner” to connote a vertically oriented, oblong textile with a
triangular headpiece at the top, assuming that it would have been hung differently than the other
“silk painting” types. The loop typically found on the triangular head indicates that it would have
been displayed by being hung on a hook at the top of a long staff, so that it could be seen from
multiple angles.
12
In contrast, the term “silk painting” is used for banners that appear to have
been hung against a wall and viewed from the front side only. Stein probably made this

12. Stein, Serindia 2, 844.
9
assumption based on the framing techniques used and multiple loops existing across the top
portion as supports for a static display method.
13

On the other hand, other scholars argue that these two types of paintings are innately
“banners” and derived from an Indian prototype.
14
Art historian Chhaya Haesner classifies them
into Type A and Type B depending on the streamers attached to each. Type A refers to sapatākā,
which has anthropomorphic form such as a headpiece, arms, and “legs” of streamers.
15
Type B,
or niṣpatākā, is in the shape of a simple square or rectangle.
16
Both types, according to Haesner,
were carried in processions and thus would have been observed from multiple perspectives.
17

These classifications are based on the model of Indian religious paintings on textile, or paṭa.
Paṭa are also believed to have been the prototype for Eastern Indian paintings, Tibetan tangkas,
and Korean t'aeng-hwa.
18
A good example of a representation of a paṭa painting can be found in
a mural inside Cave 224 of the Kizil Caves. It depicts Varṣakāra presenting a paṭa depicted with
the four major events in the life of the Buddha to King Ajâtaśatru (fig. 4). The purpose of a paṭa
painting, as visually represented in this mural, was to serve as a visual aid to transmit religious
knowledge.
In addition, some conservators such as Matthew Brack and Erin Mysak use the more

13. Stein, 843.
14. Robert Jera-Bezard and Monique Maillard, “Origine et montage des bannières peintes de Dunhuang,”
Arts asiatiques 40 (1985): 83–91; Chhaya Haesner, “Banners from Dunhuang and Turfan: A Comparative
Study and Their Indian Counterparts,” South Asian Archaeology 1995. Proceedings of the 13th
Conference of the European Association of South Asian Archaeologists 2 (1995): 715.
15. Haesner, “Banners from Dunhuang and Turfan,” 715.
16. Haesner, 715.
17. Haesner, 721.
18. Matthew Kapstein, “Weaving the World: The Ritual Art of the Pata in Pala Buddhism and Its Legacy
in Tibet,” History of Religions 34 (1995): 246.
10
general term “portable painting” to highlight the works’ portability rather than shape. Regardless
of their shapes, they focus on the ease of hanging those groups of paintings against wall or with
some hanging device, especially in contrast to wall paintings, which cannot be removed or
displayed freely.
In order to resolve the inconsistency of the terms being used by modern scholars, we can
review what terms were used during the ninth and tenth centuries to refer to the different types of
textiles as documented in manuscripts from Cave 17. Terminology used during the ninth and
tenth centuries to differentiate objects with painted religious images can be found mainly in
dedicatory inscriptions on paintings, as well as in written records such as inventory lists of assets
stored in temples (shiwuli STU). Inventory lists, in particular, often include the following
terms: fan V / W, chuang X, and xiang Y. The first two generally referred to a banner, whereas
the last term xiang, meaning “image” or “picture,” was more specifically applied to rectangular
painted images with their edges adorned or framed with multiple pieces of colorful textiles. The
term xiang was used following the name of the venerated one, the material used, and sometimes
the material’s length and quantity. For example, from the inventory of assets in Longxing
Temple, one finds the following descriptions: “four silk images of the Four Heavenly Kings
(Sitianwang juan xiang si /ZF[Y\)” and “one hemp image of Avalokiteśvara (hua bu
Guansiyin xiang yi ]^_`aYb)”.
19
Although it is highly probable that the term xiang here
refers to the rectangular type of painted banner, one might argue that there is no hard evidence
supporting this interpretation.

19. Longxing Temple $%# was an official temple in Dunhuang during the Tibetan Occupation Period
(787–848 CE). The transliteration of its inventory document (P. 3432) can be found in Ma De &',
Dunhuang gudai gongjiang yanjiu ()*+,-./ [Dunhuang ancient craftsman research] (Beijing:
Wenwu chubanshe, 2018), 102–04.
11
Objects whose inventory names include the term fan V / W share particular
characteristics. First, they were in most cases made of raw or juan silk, a material commonly
used for painting.
20
Second, if the object was painted with a pigment that had noteworthy value,
this pigment would be mentioned in the inventory description. Third, not all object names
including fan specified the name of the icon represented on them. However, there are cases in
which the banner is said to have included a “bodhisattva” without specifying the name. Two such
examples of fan V / W come from the Dunhuang manuscript P. 2613, another inventory list of
assets in Longxing Temple: “24 life-size banners painted with silver (dengshen yinni fan ershi si
kou cdefWgh\i)” and “18 juan silk banners of a bodhisattva (pusa juan fan yishi ba
jk[Wbhl).”
21
Dunhuang manuscript S. 1774 (942), a list of assets of Dacheng Temple
(9mN), includes similar entries, such as “28 small banners of bodhisattva (xiao pusa fan ershi
ba kou njkVghli)” and “six banners in (made of) juan silk for painting (huajuan fan
liu kou ][Voi).”
22
In many cases, an object whose record includes the term fan V / W
does not record the specific name of the Buddhist icon shown on the banner. Instead, what was
recorded were its material, length, and its broader category of painted image (if any). Thus, it is
highly reasonable to assume that the term fan corresponds to the oblong type of banner with a
single representation of divinity—a bodhisattva in most cases. As several matching banners
painted with silver were found from Cave 17, it is safe to conclude that the “24 of life-size
banners painted with silver” would have had visual representations of either Buddhas or

20. Chapter 1 will discuss the special characteristics of juan silk.
21. P. 2613; Ma, Dunhuang gudai gongjiang yanjiu, 405–08.
22. S.1774; Ma, 376–79.
12
bodhisattvas. If this is the case for fan, it is also within reason to argue that the term xiang
indicates the other type of a banner in a rectangular shape.
The last term that requires our attention is chuang X, which is often considered a
synonym of fan and hence translated as “banner.”
23
With regard to the translation and definition
of chuang, Michelle Wang points out that the term is understood as “pillar” as well as
“banner.”
24
Why was a textile banner referred to with a character indicating a stone pillar in its
name? Wang correctly argues that painted banners identified as chuang X share a similar
tripartite structure as a stone or dhāraṇī pillar, which also has a head, a body, and a base that is
the equivalent of a banner’s legs or bottom streamers.
25
A rare example of a chuang as a cloth
banner is referenced in several scenes on the south wall of Cave 217 in the Mogao Caves (fig. 5).
When explaining this image, art historian Akiko Shimono asserts that the practice of writing a
dhāraṇī spell on a cloth banner and hanging it on a tall structure began during the Tang period.
However, this argument cannot be proven simply by the painted representations of “cloth
banners” from Cave 217.
26

23. There is no clear distinction between these terms when they indicate a banner. In fact, the compound
word chuangfan 01 is a collective term that means all types of banners.
24. To differentiate it from meaning “banner,” she points out that shi 2 (stone) was often added to the
term chuang when used to mean “pillar.” Michelle C. Wang, “From Dhāraṇī to Maṇḍala: A Study of
Mogao Cave 14 and Esoteric Buddhist Art of the Tang Dynasty (618-907),” (PhD diss., Harvard
University, 2008), 37.
25. Michelle C. Wang, Xin Wen, and Susan Whitfield, “Buddhism and Silk: Reassessing a Painted
Banner from Medieval Central Asia in The Met,” The Metropolitan Museum Journal 55 (2020): 13.
26. Shimono argues that the dharani, which was written on the representation of a cloth banner in Cave
217, is from The Superlative Dhāraṇī of the Buddha’s Crown (Foding zunsheng tuoluoni jing 3456
789:). However, the square shape of the object which she claims is a cloth banner is fixed onto a
thin holder with a base. The square object represented is not likely a textile object, because it would not
be possible to keep it erect and steady without any bending or distortion. Shimono Akiko ;<=>,
“Tonkō Bakukōkutsu dai 217 kutsu nanheki kyōhen no shin kaishaku” ()?@A BCDEAFG:
HIJKL [New interpretation on south wall of Cave 217 of Mogao Caves, Dunhuang], Bijutsushi MN
O 54, no. 1 (2004): 96–115.
13
The use of the term chuang in ninth- and tenth-century written records helps clarify what
this term might have referred to. In the inventories of the Longxing Temple (P. 2613), seventeen
records of chuang objects are found. Whereas the term fan in the same document seems to have
been applied mainly to juan silk textiles or plain-weave silk (a popular support material for
painting), cases in which chuang was used meant refined, decorative textiles such as damask silk
(ling p) and compound weave silk (jin q).
27
Examples include “purple-damask banner (ziling
chuang rpX),” “white-damask banner (bailing chuang spX),” and “dark-red-damask
banner (feiling chuang tpX),” all of which are from the inventory of the Longxing Temple of
873 CE.
28

Temple-owned objects with the description of chuang also mention the lining material, if
any, on the back side of the chuang object. From the same inventory record from Longxing
Temple (P. 2613) is a description of “two banners (made of) white-damask (with) dark-red silk
lining.”
29
This descriptor offers a crucial clue to determining what the term chuang might have
referred to, as the banners with religious images described as fan do not typically include linings
inside the painted body. This leads me to the tentative conclusion that the term chuang might
have indicated a banner type that does not bear iconic representations of any religious deities––as

27. For the definitions of different types of silk, see Feng Zhao and Le Wang, “Glossary of Textile
Terminology (Based on the Documents from Dunhuang and Turfan),” Journal of the Royal Asiatic
Society 23, no. 2 (2013): 349–87. Also one of the notable types of silk from the list is the so-called Gaoli
jin @P(or Q)R, which refers to “brocade (imported) from Korea especially from the Three Kingdoms
Period (57 BCE–668 CE) to Koryŏ or Goryeo (918–1392 CE).” See Zhao Feng ST, Dunhuang sichou
yu sichou zhi lu ()UVWUVXY[Dunhuang silk and Silk Road] (Beijing: Zhonghua shuju, 2009),
234–36.
28. P. 2613; Ma, Dunhuang gudai gongjiang yanjiu, 406.
29. “Z[0\ ]^_” P. 2613 For a transcription of the original document, see Ma, Dunhuang gudai
gongjiang yanjiu, 406.
14
in the case of the banners from Cave 17––but rather banners that used textiles woven with
complex weaving technique or dyed and clamp-resisted patterned textile, all of which would fall
within the category of “complex” or “fancy” silk.
30
Several banners and fragments from the Stein
Collection confirm this written record.
The most compelling case of what a chuang object might have looked like is the Panel
with Confronting Horses (MAS. 885.a-b) from the British Museum, which is backed with a red
silk lining (fig. 6).
31
Even though the weaving structure of the panel is that of plain weave, it is
the pattern of the panel that made it “fancy.” In other words, the weaving structure of the panel is
plain-woven silk but its confronting-horses pattern, made with a clamp-resist dyeing technique,
was what rendered the work a “fancy” type of silk. The lining silk is also plain-weave but has no
pattern on it. Although the panel is not white damask as recorded in the inventory, the dark red
silk agrees with the short written description of the lining material for the two banners in P.
2613.
32

In short, the seemingly mixed use of the terms fan, chuang, and xiang found in the
inventories of temples’ assets can be better understood by comparing the written records with
actual artifacts from Cave 17, which has helped classify the types of artifacts with their original
names. The terms fan and chuang were used interchangeably, but in Dunhuang a fan banner

30. In short, “simple textile” refers to the plainest binding structure without any decorative woven,
painted, or dyed patterns. The term “complex silk” indicates a complicated weave structure, which
includes but is not limited to twill, satin, gauze, and compound weaves. Lastly, the term “fancy textile”
refers to any textile adorned with dyeing, painting, or weaving techniques. For the detailed definitions of
“simple,” “complex,” and “fancy” silk, see Angela Sheng, “Determining the Value of Textiles in the Tang
Dynasty in Memory of Professor Denis Twitchett (1925–2006),” Journal of the Royal Asiatic Society 23,
no. 2 (2013): 175–76.
31. Feng Zhao, ed. Textiles from Dunhuang in UK Collections (Shanghai: Donghua University, 2007),
196.
32. Another example of a lining of a panel is Hir.14Oct04/6 from the Stein Collection. Its structure is
plain weave, from the seventh to tenth century. Zhao, Textiles from Dunhuang in UK Collections, 241.
15
tends to connote an iconic banner with a painted religious figure (usually in a standing position)
that had no lining and was viewed from multiple angles. On the other hand, chuang was
understood as an aniconic banner (meaning with no painted religious figures) that was made of
damask silk backed with another monochrome silk. The fact that the main panel was backed not
with undyed natural plain-weave silk but rather with a colored/dyed plain-weave silk implies a
mode of display that would allow the viewers to see the image on the banner from the front and
the back.
33

This leads me back to the meaning of the term xiang, which indicates religious figures
within a rectangular frame without a triangular headpiece or streamers. It can be translated as an
iconic, portable painting that was meant to be seen frontally only, unlike the fan and chuang
types. A fan banner can be converted to a painted (iconic) banner, whereas a chuang banner
represents an aniconic/decorative banner.
This connects to another critical term bianxiang uv, which is normally translated as
transformation tableau. The term is the prime written signifier for compositions found on walls
of caves and monasteries as Sinologist Victor Mair and art historian Wu Hung argue, whereas
the fan and chuang are exclusive to portable banners.
34
The differentiation between the

33. The colors of silk used for lining requires further attention, in my opinion, because they are primary
colors that might have represented the “five-colors (wuse `a).” This, however, does not explain why a
banner with “five-color” panels, which has no iconic images, is called five-colored fan banner (wuse fan
`a1) with no use of chuang. Despite this pitfall, this analysis strongly suggests different functions of
and corresponding structures of the banners’ body panels, ones that would differentiate a fan banner from
a chuang banner. MAS.860 and MAS.861 from the Stein Collection are the closest examples to the “five-
colored” banner. Comprised of three panels stitched together, these aniconic, decorative banners have
blue, white, and red panels from the top to the bottom of the body. For the textiles of the head and its
infill, see Zhao, Textiles from Dunhuang in UK Collections, 59–63.
34. Victor Mair, “Records of Transformation Tableaux (Pien-Hsiang),” T’oung-pao 72, no. 3 (1986): 3–
43; Wu Hung, “What is Bianxiang?: On the Relationship Between Dunhuang Art and Dunhuang
Literature,” Harvard Journal of Asiatic Studies 52, no. 1 (1992): 111–92.
16
bianxiang and fan/chuang as determined by portability and medium is also confirmed by the fact
that it is rare to find a bianxiang in those temple inventory documents. On the other hand, fan
was characterized by its support material, huajuan ][, which is silk for painting. In A Record
of the Famous Painters in Succeeding Dynasties (Lidai minghua ji Uwx]y), Zhang
Yanyuan z{| (c. 815–c. 877) mentions the following:
Thirteen fan banners on silk for painting [They have feet of gilt bronze. The length is one
fathom two feet. There are copies (of paintings) made by Zhang Liba z}~ which are
also finished in colors. There are four other banners also one zhang two feet in length,
but they were finished by various different workmen].
35

In regard to the definition of the term fan, the record yields two important facts. First, the term
hua juan specifically refers to a silk support for painting, suggesting that the fan banners were
painted objects. Second, the painter Zhang Liba drew and completed it by coloring the drawing;
this confirms that the fan indicates a pictorial object.
This raises an important question in terms of the relationship between bianxiang as a
fixed wall painting and fan and xiang as movable, portable pictorial paintings. Were painters
aware of the different materiality between wall and textile? How did they take advantage of
different media in depicting the same subject? Were there any preferred materials for specific
compositions to convey the underlying meanings? In what ways did the donors contribute to the
selection of materials and the creative process? The discussion of these issues in Chapter 2 will
shed new light on different strategies and painting practices in Buddhist visual culture in
Dunhuang.

35. The original text is: “bc1d!efgDhCifjklmno. “pq1d!e bcrfgD
hCifjklmnostuefvgDhCifwxo.” The translation is slightly modified based
on William Reynolds Beal Acker, Some T’ang and Pre-T’ang Texts on Chinese Painting (Leiden: E.J.
Brill, 1954), 326–7.
17
0.2. Methodology
This dissertation will mainly focus on the fan banner type and xiang painting type, as it
will unpack the meanings of Buddhist figural images in relation to their physical configurations
and the painting materials used in their creation. Artistic practice and circulation of the painted
materials, especially for textiles as support materials, and their religious meaning and practice
will be the main focus of the dissertation. To analyze these aspects, I have focused on fifty
painted banners from Dunhuang held at the British Museum, the Musée Guimet, and the
National Museum of Korea. Even though the total number of paintings from Cave 17 is nearly
1,050, including the three major collections (the Stein Collection in the British Museum; the
National Museum of India, New Delhi; and the Pelliot Collection in the Musée Guimet in Paris),
the paintings under discussion here not only represent popular yet peculiar religious ideas but
also show significant material qualities as well as artistic labor and techniques.
36
Among them, I
have further narrowed down and selected both types of paintings, including an oblong painted
banner and a rectangular, portable liturgical painting with dedicatory inscriptions and dates, both
of which encompass religious practices associated with ordinary devotees from lower to middle
social classes as a more affordable offering than a set of wall paintings within a cave.
Pictorial materials from Cave 17 cannot be considered separately from the pictorial art of
the Mogao Caves. (see fig. 2). Indeed, Buddhist texts and documents found therein supplement

36. Yuan Ting came up with this number by counting each institution’s collection spread all over the
world. However, this number needs to be updated because it does not include fragments of paintings. In
addition, it should be pointed out that a discrepancy exists between Stein’s report from his expedition and
the actual number of collections in the British Museum, British Library, and the National Museum of
India, New Delhi. For example, he reported that the number of banners and large hanging paintings he
had brought from the cave was 398, but the museums’ collections have 562 works: 278 in the British
Museum, four in the British Library, and 280 in the National Museum of India, New Delhi. Stein,
Serindia 2, 843–845; Yuan Ting yz, Dunhuang zangjingdong chutu huihuanpin yanjiushi (){:|
}~ÄÅ./O [Historiography of paintings from the library cave of Dunhuang] (Lanzhou: Gansu
jiaoyu chubanshe, 2016), 4–5.
18
and complement the wall paintings inside cave-temples and also portable paintings, providing
liturgical sources and material supply situation during the ninth to tenth centuries in Dunhuang.
37

For example, identical religious icons in a similar style are found in these two different artistic
formats. A single bodhisattva in a standing position or a series of bodhisattvas in standing
positions were two of the most popular motifs at Mogao Caves across the centuries; they were
painted in corridors, on the sides of the western niche, and on the sides of a sutra painting (or
bianxiang) on the walls. Bodhisattva figures with the same posture are a common pictorial
subject of the oblong painted banner format.
38
In addition, sutra paintings were also commonly
painted on textiles, but due to the limited sizes of textile and the difference in materiality
between textile and wall, the compositions on painted banners and portable paintings are not the
same as those for mural paintings. Chapter 2 compares textile support materials and murals,
revealing different and similar strategies that were applied to the textile paintings.

0.3. Definition of Materiality

37. For defining the function of the cave, and the characteristics of the objects and sacred sutras in Cave
17, two major arguments have been proposed. Stein argues that the cave functioned as “waste repository,”
while others such as Rong Xinjiang oppose this idea and instead contest that the entire contents of the
cave should be understood as a monastic library, specifically a library from the Sanjie Temple, which was
believed to be located in the vicinity of the caves. Van Schaik and Galambos in their 2011 publication
suggest abandoning the point of view on the “split between two incompatible options” and argue that the
cave had multiple functions that included use as “a repository for donated and commissioned manuscripts
and, some paintings, that could not be stored in the monastic libraries” and “a storeroom for the monastic
library of the Sanjie Temple.” Aurel Stein, Serindia: Detailed Report of Explorations in Central Asia and
Westernmost China 2 (Oxford: The Clarendon Press, 1921), 791–840; Xinjiang Rong, “The Nature of the
Dunhuang Library Cave and the Reasons for its Sealing,” Cahiers d’Extrême-Asie 11, no. 1 (1999): 247–
75. Recent debate and the summary of each argument is well summarized in van Schaik and Galambos,
Manuscripts and Travelers, 13–34, especially 18–28.
38. For a description of the artistic practice of transferring a sketch to a mural plan, see Sarah E. Fraser,
“Formulas of Creativity: Artists' Sketches and Techniques of Copying at Dunhuang,” Artibus Asiae 59,
no. 3/4 (2000): 189–224.
19
The fundamental idea underlying a materiality-centered approach is rooted in thing
theory, which acknowledges the agency of material and its cultural value.
39
A banner’s length,
the way it is displayed, its movement, and the transparency of the textiles all contribute to its
autonomous agency, creating individual, unique experiences for its donors and viewers.
40

Because a painted banner was regarded as a religious object, the relationships with its donor, its
painter, and the viewer become crucial. The paintings’ efficacy as a religious offering as well as
the painted images’ capacity to bring about epiphany are what the viewers and painters try to
achieve along with personal wishes and goals.
41
For this reason, this dissertation examines the
collective material qualities of painted banners and portable sutra paintings in terms of their
micro and macro structures; these include the weaving structure of the support textile, pigments,
and the painting process (for the micro structure) as well as the mode of display, painting style,
length, and ritualistic context.
To analyze strategies used by artisans/painters to elicit specific viewer responses to their
works, this dissertation will investigate the painted banners’ material qualities as active agents in
their function as religious objects. What generates religious efficacy is not only what is painted
on a banner (i.e., religious iconography) but also what is used to paint with (i.e., painting
materials such as certain silks and pigments). A painting’s materials—such as the textile used for

39. Daniel Miller, “Materiality: An Introduction,” in Materiality, ed. Daniel Miller (Durham: Duke
University Press, 2005), 4.
40. For the concept of “autonomous agency” of things and social relations between persons and things,
see Alfred Gell, Art and Agency: An Anthropological Theory (New York: Oxford University Press, 1998),
12–27.
41. When he defines the term bian H or “transformation” in the context of religious narrative literature
culture in Dunhuang, Victor Mair discusses the issues of epiphany, which was expected to be created by a
bian storyteller. See Victor H. Mair, T’ang Transformation Texts: A Study of the Buddhist Contributions
to the Rise of Vernacular Fictions and Drama in China (Cambridge: Harvard University Press, 1989),
36–72, especially 60.
20
the body of a painting or for framing and the pigments used—are deliberate choices that can
directly contribute to its perceived religious efficacy.
This approach is relevant to broader issues about how Buddhist material culture and
devotional practices mutually informed each other. Understanding how a mundane object such as
a textile was transformed into a votive object through the choice of material is key to elucidating
what devotees considered to be suitable and effective materials as offerings for Buddhist deities.
My research helps illuminate the long-standing tradition of offering textiles to monasteries to
facilitate rituals (e.g., funerary practices) and adorning iconic paintings, sculptures, stupas, and
other architectural monuments with various types of textiles, a tradition which remains
widespread today. Tracing the beginning of this tradition to the specific context of Dunhuang in
the ninth and tenth centuries allows us to comprehend the enduring appeal of making and
offering paintings as a form of religious devotion across the centuries. This will help further our
understanding of the merit of using textile materials in Buddhist rituals and in funerary practices
then and now.
To understand all these cultural practices, it is important to see a painted banner not as a
flat composition but rather as an object in multi-dimensions, adorned with multiple pieces of
textile. Unlike a landscape or literati painting, a painted banner was both produced and viewed
by making full use of its front and back sides. This mode of display leads me to argue that
donating a painted banner could have meant a “textile” offering. In fact, the large cache of
textiles and their fragments from the cave attest to the local popularity of offering textiles in
temples. I explain why textiles and painted textile banners were considered proper objects for
Buddhist offerings by focusing specifically on a funerary ritual in which textiles were generously
and popularly offered. Why was silk chosen as an “appropriate” material for offering? This study
21
ultimately seeks to uncover the historical and religious connotations behind silk material, with
examples from tombs and cave-temples.

0.4. Previous Scholarship
A review of the scholarship on the painted banners from the Library Cave would not be
complete without discussing art historian Roderick Whitfield’s pioneering works of cataloging
over five hundred painted banners that were brought by Stein to the British Museum.
42
Each
catalogue entry by Whitfield provides a thorough analysis of a banner’s iconography and
painting style, thereby laying the groundwork for future scholarship in this area.
43
In the early
2000s, art historians Sarah Fraser and Sha Wutian brought the Library Cave’s visual materials,
including sketches and hanging banners, into the discussion of artistic practice and artisan
organizations in the tenth century in the Dunhuang area.
44
The different stances on artisans’
creativity that they respectively took were based on a different understanding of their social
status and their ability to perform their work.
On the origins and formats of Buddhist banners in general, Buddhist banners’ versatile
functionality as temple and stupa decorations and in processions were discussed.
45
Bhattacharya-

42. Roderick Whitfield, The Art of Central Asia: The Stein Collection in the British Museum 1 (Tokyo:
Kodansha International, 1982).
43. Sonya Lee’s interview with Whitfield is an invaluable resource to understand his role and contribution
to the Stein Collection. Sonya S. Lee, “An Interview with Roderick Whitfield on the Stein Collection in
the British Museum,” The Silk Road 17 (2019): 12–25.
44. Sarah E. Fraser, Performing the Visual: The Practice of Buddhist Wall Painting in China and Central
Asia, 618-960 (Stanford: Stanford University Press, 2004); Sha Wutian ÇÉÑ, Dunhuang huagao yanjiu
()pÖ./ [Studies on the sketches of Dunhuang] (Beijing: Minzu chubanshe, 2006).
45. Money L. Hickman, “Notes on Buddhist Banners,” Boston Museum Bulletin 71, no. 363 (1973): 4–20;
Jera-Bezard and Maillard, “Origine et montage des bannières peintes de Dunhuang,” 83–91; Robert Jera-
Bezard and Monique Maillard, “Le rôle des bannières et des peintures mobiles dans les rituels du
bouddhisme d’Asie centrale,” Arts asiatiques 44, no. 1 (1989): 57–67; Yim Yŏng-ae, “Chungguk kodae
22
Haesner, in particular, did a comparative study of the shape, material, and style of banners from
Dunhuang and Turfan.
46
This compendium shows that cotton was the most popular material for
painted banners in the Turfan region.
The academic field that most actively investigates the Dunhuang painted banners is
textile studies. In 2004–2015, museums in Europe began cataloging their textile collections from
Dunhuang. While these catalogues are useful sources for understanding the basic characteristics
of silk and other textiles, including their weaving structure, dyeing material, and variety of
patterns, they mostly deal with decorative banners, sutra covers, and canopies, none of which
have any iconic representations.
47
In addition to these museum catalogues, Chinese publications
in the 2000s have shown increased interest in the Dunhuang banners from a textile studies
perspective. They focus on the fundamentals of a banner and provide an understanding of the
material qualities of silk.
48
Overall, however, there is little historical explanation regarding the
material significance of banners and their ritualistic usage.
My approach to textiles as an important cultural agent echoes the scholarship by historian
BuYun Chen and art historian Mariachiara Gasparini. In her 2019 book, Chen convincingly
demonstrates how the cultural values embedded in attire and textiles manifested in Tang dynasty

pulgyobŏnŭi yangshik pyŏnch’ŏn’go” *+ Üá 3à1의 âäHãå [Stylistic changes in Buddhist
banners from the anceint China], Misul sahak yŏn’gu 189, no. 3 (1991): 69–109.
46. Haesner, “Banners from Dunhuang and Turfan,” 715–30; Bhattacharya-Haesner, Central Asian
Temple Banners.
47. In his edited volume about the French collections, Zhao Feng includes several painted banners unlike
the earlier one in the UK collections. Zhao, Textiles from Dunhuang in UK Collections; Feng Zhao ed.,
Textiles from Dunhuang in French Collections (Shanghai: Donghua University Press, 2011); Feng Zhao,
ed. Textiles from Dunhuang in Russian Collections (Shanghai: Donghua University Press, 2014).

48 .Wang Le çé and Zhao Feng ST, “Dunhuang fan de shiwu yu fenxi ” ()1èêëíìî./
[Study and analysis of existing fabric of the Dunhuang banners], Dunhuang yanjiu ()./ 1 (2008): 1–
9.
23
fashion.
49
Gasparini argues that tracing “the movement of textiles and related materials” leads to
discoveries of the “repertoire of Central Asian imagery on textile surface” a development in
which the transmission and sharing of motifs and shared were not limited by traditionally fixed
physical boundaries and borders.
50
Her argument is based on her view of a textile as a three-
dimensional object:
The three-dimensionality of a textile, created by the fibers interwoven in its structure, is
an intrinsic quality that assures it a place in the field of material culture, but because it is
almost invisible to the naked eye, this automatically shifts textile into the category of
two-dimensional surfaces, such as paintings. … Because of its three-dimensional nature,
a textile is a medium of material culture but because of its two-dimensional surface it can
be also included in the field of visual culture.
51

My interpretation of a textile’s intrinsic three-dimensionality is in accord with
Gasparini’s. The framework of this dissertation corresponds with current thought, one which
rediscovers the role of textile and emphasizes its materiality, which in turn transforms the two-
dimensional painting into three-dimensional object.
These scholars show that there is a general consensus regarding the categorization of
painted works and textile objects. It is true that weaving textiles requires techniques and skills
that are completely different from those required for paintings. Thus, different expertise and
methodology is required for these two fields of study. However, painted banners and iconic
liturgical paintings in Dunhuang are likely to fall into a gray area if one follows the trajectory of
existing scholarship on banners. This is because both types of Dunhuang paintings were painted
on textiles and framed/adorned with textiles. Without a holistic understanding of these painted

49. BuYun Chen, Empire of Style: Silk and Fashion in Tang China (Seattle: University of Washington
Press, 2019).
50. Mariachiara Gasparini, Transcending Patterns: Silk Road Cultural and Artistic Interactions Through
Central Asian Textile Images (Honolulu: University of Hawai’i Press, 2020), 1–3.
51. Gasparini, Transcending Patterns, 4–5.
24
banners regarding their material aspects, it is difficult to define their role in the material culture
of medieval Dunhuang.
In this regard, conservation studies can fill the gap between textile-oriented scholarship
and iconography-focused research. In recent years of 2000–2020, the importance of conservation
studies of the murals and painted banners of the Mogao Caves has been highlighted. The Getty
Conservation Institute has a long-term partnership with the Dunhuang Academy for site
management of the Mogao Caves, including conservation of wall paintings and the study of the
types of pigments used, the process of painting and sculptures, and the environmental conditions
that have impacted the condition of the wall paintings therein,.
52
While the Getty–Dunhuang
collaboration does not include any painted banners or textile objects, its outcome functions as a
stepping stone, as progress has been made in conservation research conducted on the painted
banners and textiles from Cave 17. One notable study, by Brack and Mysak, focuses on the
banner paintings in museums collections in the United States, including the Harvard Art
Museums; the Museum of Fine Arts, Boston; and the Freer Gallery of Art, Smithsonian
Institution. Brack and Mysak shed new light on the resemblance between artistic practices (i.e.,
painting techniques and preparatory processes) executed on both mural paintings and portable
paintings in the caves (especially the ones painted on hemp).
53
Conservation scientist John
Winter’s extensive work on East Asian paintings also provides a reliable resource for pigments,

52. For the outline of the project and its outcome, see Neville Agnew, Martha Demas, and Wang Xudong,
“The Enduring Collaboration of the Getty Conservation Institute and the Dunhuang Academy in
Conservation and Management at the Buddhist Cave Temples of Dunhuang, China,” The Public
Historian 34 (2012), 7–20; Lori Wong and Neville Agnew, eds. The Conservation of Cave 85 at the
Mogao Grottoes, Dunhuang: A Collaborative Project of the Getty Conservation Institute and the
Dunhuang Academy (Los Angeles: Getty Conservation Institute, 2014).
53. Matthew Brack and Erin Mysak, A Technical Study of Portable Tenth-Century Paintings from
Dunhuang in US Collections (Boston: Harvard Art Museums, 2010), 49–51.
25
sizing, mounting, and framing.
54

By synthesizing what has come to light through conservation studies, this dissertation
acknowledges both painted banners and liturgical paintings as a complete pictorial art by
focusing more on their textual aspects than previous studies, and by finding meanings inherent in
their chosen materials. In doing so, this dissertation aims to increase understanding of the genre
of the painted banner and rectangular liturgical/portable painting, as a votive object, emphasizing
its three-dimensionality, physicality, and interactivity with the viewer.
55
The portability and
agency innate in textile as material would highlight the role of the pictorial object as an active
agent or recipient of multicultural exchanges. As this study will demonstrate, both preexisting
Buddhist visual and ritual programs and other cultural stimuli from within and outside Dunhuang
provided resources for artistic practice, pictorial arrangement within a cave, and ritual practices.
The material culture of the neighboring regions such as Turfan and Chinese Daoist practices are
two main sources of outside stimuli to be discussed in the dissertation.

0.5. Outline of Chapters
The dissertation consists of four chapters, each devoted to visual design, ritualistic

54. In addition to Winter’s work, Robert H. van Gulik’s publication is worth reviewing as it provides
detailed description on mounting and framing. John Winter, East Asian Paintings: Materials, Structures
and Deterioration Mechanisms (London: Archetype Publications, 2008); Robert Hans van Gulik, Chinese
Pictorial Art as Viewed by the Connoisseur: Notes on the Means and Methods of Traditional Chinese
Connoisseurship of Pictorial Art, Based upon a Study of the Art of Mounting Scrolls in China and Japan
(Rome: Istituto italiano per il Medio ed Estremo Oriente, 1958). For a history of the modern mounting
practice on the famous the Admonition Scroll in the British Museum, see Joanna Kosek, et al., “The
Admonitions Scroll: Condition, Treatment and Housing 1903–2014,” The British Museum Technical
Research Bulletin 9 (2015): 25–43.
55. A textile’s three dimensionality is intrinsic in nature, and is generated through its weaving process
(fibers being interwoven). However, because weaving structure is often unrecognized and unappreciated
by a viewer, textiles are often conceived of as being two-dimensional objects. Gasparini, Transcending
Patterns, 4–5.
26
function, and artistic practice in different types of painted banners. Chapter 1 scrutinizes the
artistic labor and materialistic characteristics of painted banners on silk. Using Stein Painting 52
as the key example, the chapter attempts to reconstruct the production process that began with
weaving the support to mounting it with multiple silk fabrics. By analyzing the microstructure of
the painting, I explore the types of weaving and artistic labor that contributed to the entire
process. Chapter 2 focuses on the relationship between the murals of the Mogao Caves and
painted banners. Focusing on a Avalokiteśvara Sutra painting (Guanyin jingbian _a=u), I
demonstrate that artisans often adjusted styles and compositional designs to conform to that
medium’s distinctive material qualities. Chapter 3 delves into the functionality of a long banner
(Stein Painting 216) as a votive object, showing that multiple bolts of silk were sewn together to
emphasize its extreme length and numerous images of bodhisattvas. The multiplicity of the
textile and its materiality visually signify the donor’s wishes such as prolonging life and gaining
rebirth in the Western Paradise. Chapter 4 further investigates the signification of silk in
northwestern China and the dissemination of such textile votive objects throughout Asia. By
looking through the inventories of tombs in Turfan, an oasis city near Dunhuang where silk
threads were a common burial goods, I discuss the practice of using silk as votive offerings in
tombs and cave-temples. The chapter not only broadens the meaning of textile material but also
deepens the understanding of the belief that tangible material virtually connects to what one
wishes for—such as reaching heaven.

27
Chapter 1. Artistic Labor and Materials of Stein Painting 52 from Cave 17

1.1. Introduction
Much of the pictorial and written materials from Cave 17 remained undisturbed for nearly
a millennium until their discovery in the early twentieth century. Stein readily observed their
overall well-preserved condition during his 1907 visit. He described the original condition of the
bundles from which painted banners and large rectangular paintings were found as follows:
The importance of the exceptionally favourable conditions of preservation enjoyed by
this great deposit impressed itself upon me even more when, on opening a large packet
wrapped in a sheet of stout discoloured canvas, I found in it, mixed up with
miscellaneous papers, paintings on fine gauze-like silk and on linen, as well as a mass of
textile pieces in all kinds of silk and brocade, suggesting ex-votos [my italics]. Most of
the paintings first found were narrow pieces from 2 to 3 feet in length, and could, by their
triangular tops and floating streamers, be recognized at once as having been intended for
temple banners.
1

Some of the narrow painted banners first recovered from ‘miscellaneous’ bundles were
found, indeed, neatly rolled up, and the pliable and soft condition that their silk material
had retained allowed them to be unfolded with ease.
2

In this excerpt, one can find crucial information on the way each painting was deposited
in what Stein named “miscellaneous” bundles. Unlike regular bundles, in which a dozen or so
Buddhist manuscripts in either Chinese or Tibetan were rolled up individually and then covered
with a separate wrapper, a “miscellaneous bundle” had mix materials: manuscripts written in
Sanskrit, Khotanese, Tibetan, Uighur, and Sogdian; painted banners as well as fragments of

1. Stein, Serindia 2, 811.

2. Stein, 831.

28
paintings and textiles; and papers.
3
Stein noted the painted banners were “neatly rolled up” and
easy to unfold because of the condition of the silk used in the paintings.
On the other hand, in a case involving a rectangular silk painting, which normally has a
wider width than that of an oblong silk banner, the painting was likely folded in half both
vertically and horizontally, which was observed by Stein as follows:
The risks attending the use of a very fine fabric were demonstrated only too clearly when
subsequently I came upon convolutes containing silk paintings much larger in size. They
must have been closely and often carelessly folded up at the time of their deposition, and
were much creased and crumpled in consequence [my italics]. After centuries of
compression any attempt to open them out completely would have entailed obvious risks
of damage to the thin material. But by lifting a fold here and there it was possible to see
that the scenes represented were often as elaborate and crowded with figures as the fresco
panels on the walls of some of the largest temple cellas.
4

Because this type of a painting was stored folded and then pressed by other bundles,
damages such as creases and crumples were inevitable. Today such heavily damaged bundles are
intentionally left untreated in order to avoid further deterioration.
5

Stein Painting 52, which features a seated Avalokiteśvara or Guanyin Bodhisattva
(Guanyin pusa_ajk), is the one of the best-preserved artifacts from Cave 17 (fig. 7). Based
on the clear crease line along the vertical axis, as well as several horizontal creases, the painting
was likely folded multiple times.
6
The reason for the superb condition of the painting could be

3. Valerie Hansen, The Silk Road: A New History (New York: Oxford University Press, 2015), 180–181.

4. Stein, Serindia 2, 811.

5. 1919,0101,0.225 shows the original condition of bundle in which a large painting of the Paradise of
Amitābha was found. For Stein’s description on this particular bundle, see Stein, 1083 (Ch.lviii.006).
Another example showing the original state of a bundle can be found in Stein, Serindia 4, LXXVI (Ch.
00350).

6. However, according to Stein, very few paintings were not folded but rolled up. Examples Stein
mentioned include Stein Paintings 7, 15, 45, 46, and 47. See Stein, Serindia 2, 843 and no. 3.

29
that the bundle in which the painting was found was not placed under other heavy bundles.
Painted on silk and framed in borders of fine silk, this painting’s vivid pigments, brushstrokes,
and dedicatory inscriptions have been preserved since their creation in the tenth century CE.
Considering that over five hundred painted banners and hanging paintings were recovered during
the respective expeditions of Stein and Pelliot, a significant concentration of skilled labor and
materials undoubtedly existed in the oasis city of Dunhuang. Making even one such painting
would have required artisans with a range of skills: weaving and dyeing the textile, making and
applying color pigments, writing dedicatory inscriptions, and sewing and mounting the finished
painting. In this chapter, I reconstruct this process by examining the microstructure of Stein
Painting 52 and identifying the various types of artisans who participated in the production of
this painting.
The artistic practices involved in creating the painted banners from Cave 17 have not
been fully explored since the cave’s discovery in 1900, as research published on the Mogao
Caves to date has focused primarily on the caves’ wall paintings. However, since the beginning
of the twenty-first century, other pictorial materials from Cave 17 have been studied.
Two prominent scholars, Fraser and Sha, initiated the discussion of the artistic practices
used to create the wall paintings and painted banners from Library Cave. Fraser first utilized
sketches, drawings, pounced papers—papers with a design outlined with tiny holes that enabled
colored powder to pass through and transfer the design to a surface underneath—and banner
paintings from this cave to discuss how the paintings were copied from sketches and made into
wall paintings or banners; she emphasized how the painters exercised creativity in their

30
arrangement of these pictorial units.
7
Fraser’s ideas were refuted by Sha, who argued that the
extent of the painting practice was rather limited and that the painters were not independent
artists but instead skilled professional craftsmen whom the patrons had hired.
8
He further argued
that imitation, rather than creation, was the painters’ foremost task, and that preparatory
materials were merely tools used to increase efficiency and expedite the task’s completion.
9

These authors’ emphasis on the role that painters played in ninth- and tenth-century
Dunhuang has shed new light on those behind the scenes who had not previously received proper
assessments in studies of Dunhuang’s visual culture. Fundamental studies on who these artisans
actually were have been undertaken by historian Ma De in 2018. Ma, in his in-depth research on
the artisans of Dunhuang, examined the types of craftsmen (gongjiang ÄÅ) there were, as well
as their organizations, social status, and workmanship.
10
His work was based on inscriptions on
painted banners, wall paintings, and documents found in the Library Cave, which also included
contracts, payroll lists, temple inventories, and lists of offerings (shiru shu ÇÉÑ).
11
Ma’s study
can provide a context for the discussion of who was actually responsible for the visual culture of
Dunhuang.
Meanwhile, it is still unclear which types of artistic labor and how many workers were
involved in the creation of the painted banners and hanging paintings such as Stein Painting 52.

7. Fraser, Performing the Visual, 206–12.
8. Sha, Dunhuang huagao yanjiu, 493–501.
9.

Sha, 499.
10. Ma, Dunhuang gudai gongjiang yanjiu, 22–84.
11. Ma’s work is based on his earlier compendium of related historical documents in 1997, and the most
recent publication in 2018 is a complete study on those documents he collected. Ma De &', Dunhuang
gonjiang shiliao (),-Oï [historical document on craftsman in Dunhuang] (Lanzhou: Gansu
renmin chubanshe, 1997).

31
Based on technical studies of silk paintings conducted in 2000–2020, I argue that a painting’s
support, raw silk, is an essential material that provides significant information about a banner’s
origin, workshop, and the ways in which textile production impacted painting styles and the
artistic techniques. The delicate quality of the support made it possible for artisans to observe the
sequence of strokes, and the relatively small pictorial plane would not require as much artistic
labor as mural paintings. This means that silk paintings were more accessible and affordable for
the common, ordinary laity. Furthermore, by reconstructing how Stein Painting 52 was
constructed, I contend that the silk painting reflects an unusual artistic practice that is not
common in typical Chinese painting’s mounting method.
1.2. Stein Painting 52 and its Support Material
Dated to 972, Stein Painting 52 has a seated Avalokiteśvara as its central icon,
surrounded by six smaller bodhisattvas. The lotus flower bud in his right hand and the kundika
(sprinkle bottle) in his left hand help identify him as Avalokiteśvara, the bodhisattva of
compassion and mercy. In a register below him, divided by a thick, dark-red line, six patrons are
shown, with three females on the left and three males on the right.
In the creation of a work such as Stein Painting 52, the most fundamental process would
have been to find or produce the painting’s support material and prepare it for painting. Since the
Han Dynasty, textiles had been the primary painting surfaces for banners, with silk the most
popular material followed by paper. Silk was preferred for its durability. The availability of the
materials and their religious connotations, which will be scrutinized in a later chapter,
determined the materials chosen for a painting’s support, except in the case of a textile that was
woven from scratch. The support material of Stein Painting 52 is raw silk, or shengjuan Ö[, a

32
tabby-patterned or plain-woven silk that is created using a specific preparation process. The silk
is “raw” in the sense that it is not processed (e.g., boiled or beaten) to remove its sericin—a
natural, gluey protein that adheres to the silk thread. Thus, shengjuan is also called raw or un-
degummed silk. Its natural color is a pale yellow, and when woven it has the characteristic of
relatively large spaces between threads. The thread count is 55 threads/cm in warp and 20
threads/cm in weft, with a very thin warp thread of 0.05 to 0.1 mm.
12
In contrast, another type of
plain silk, juan, which has been degummed, has a thread count ranging from 40 to 60 threads/cm
in both warp and weft. As the difference in thread count indicates, shengjuan silk is rather
loosely bound, leaving relatively large rectangular openings of 2.5 mm between warps.
13

Shengjuan was used in particular for pictorial materials such as hanging paintings and painted
banners,
14
as the coarsely woven silk was considered more suitable as a base material for
paintings than the juan silk.
15

There are some cases from Cave 17 in which non-shengjuan silk was used as a painting
support for painted banners and hangings. According to Tasaki, the nonconventional/traditional

12. Tasaki Chiharu Ññóò, Tōrufan, Tonkō no bukkyō kaiga no kiteīzai ni tsuite ôöõúùû ・ (
)Iü†°pI¢£§•¶ß® [On the base material of Turfan and Dunhuang Buddhist paintings]
(Tokyō: Fujizerokkusu kobayashisetsu Tarō kinen kikin, 2012), 18.
13. Zhao, Textiles from Dunhuang in French Collections, 43. In her study, Tasaki Chiharu particularly
focuses on the silk woven by this weaving structure. It is called moro date (©™) in Japanese, which
means “doubled warp.” Tasaki, Tōrufan, Tonkō no bukkyō kaiga, 4.
14. Zhao, Textiles from Dunhuang in French Collections, 43.
15. In the case of Stein Painting 52, the thickness of the warp thread is 0.05–0.1 mm, and that of the weft
thread is 0.15–0.2 mm. Tasaki, Tōrufan, Tonkō no bukkyō kaiga no kiteīzai ni tsuite, 18. According to
Brack and Mysak, the terms “coarse” and “fine” in relation to silk are determined not by the thickness of
the silk threads but by thread count and the density of the silk. However, it is worth noting that the warp
thread is particularly thin—less than 0.1 mm—in the case of Stein Painting 52. Matthew Brack, and Erin
Mysak, A Technical Study of Portable Tenth-Century Paintings from Dunhuang in US Collections (2010),
30 and no. 63.

33
silk or tightly woven silk as opposed to plainly woven with double warp-threads (vertical),
became prevalent in the tenth century.
16
For example, Stein Painting 19 and Stein Painting 54 are
painted on tightly woven silk without leaving any porous spacing in between the threads. The
other two examples Tasaki mentions are two paintings from the National Museum of India at
New Delhi. There are no inscriptions on these two paintings, both of Thousand-Armed
Avalokiteśvara, but the styles of the donor figures’ clothing and hair are contemporary with the
late-tenth-century paintings discussed earlier, including Stein Painting 19 and Stein Painting
54.
17

New evidence of closed-woven-structure silk being used as a painting material indicates
the influx of silk in the Dunhuang region, which was once collected as tax payments, and the
repurposing of such textiles as temple offerings. In the Tang dynasty, because textiles such as
silk and hemp are lighter than coins, they were considered proper monetary means for paying
taxes and corvée duties.
18
Due to military campaigns with Tibetans in the seventh to eighth
centuries, armies received tax-textiles as part of their salaries and to use as necessities.
19
The
weaving structure of tax-textile during the tenth century can be characterized by its closed,
simple plain-weave structure. Recycling such tax-textile in silk, once used as tax payments or
salaries to military forces, was a common practice. There are several supporting cases in which
tax-textiles were recycled for funerary materials. For example, the tomb (Ast. ix. 2) of Fan

16. Tasaki, Tōrufan, Tonkō no bukkyō kaiga, 10–11.
17. Tasaki, 10–11.
18. Binghua Wang, “A Study of the Tang Dynasty Tax Textiles (Yongdiao Bu) from Turfan,” Journal of
the Royal Asiatic Society 23, no. 2 (2013): 263.

19. For the seventh to eighth centuries’ tax textiles shipment to northwestern China, see Masahiro
Arakawa, “The Transportation of Tax Textiles to the North-West as Part of the Tang-Dynasty Military
Shipment System,” Journal of the Royal Asiatic Society 23, no. 2 (2013): 245–261.

34
Yanshi Üáà (d. 698) from the Astana graveyard includes hemp shrouds that were originally
tax-textiles from the seventh and eighth centuries.
20
Even though in this case hemp cloth was
used, it suggests that similar practices of recycling silk used for tax were prevalent in
northwestern China where resources were not abundant as in central China.
In terms of the availability of silk, there are multiple ways to understand the circulation of
silk and the recycling of the material, which could have been once used as currency and later
repurposed as a painting material. Considering the circumstances of circulating bolts of silk
especially in the Dunhuang area, the major sourcing of silk materials in general was inevitably
from the central China, namely from the state-run textile workshops in the lower part of the
Yangzi River (Henan circuit) and southern Sichuan and central Yunnan (Jiannan circuit) regions,
as well as the coastal region of Jiangnan (Jinagxi, Hunan, southern Anhui, southern Hubei,
southern Jiangsu, Zhejiang, Fujian, Shanghai).
21
In addition, there was probably a local supply of
silk produced by private and local workshops as well.
22

20. For the reconstructed history of the tax-textiles, see Valerie Hansen and Helen Wang, “Introduction,”
Journal of the Royal Asiatic Society 23, no. 2 (2013): 155–63.
21. Chen, Empire of Style, 122–30; Zhao, Dunhuang sichou yu sichou zhi lu, 208–15.
22. In order to identify and discuss where the silk supports for painted banners from the Library Cave
actually originated, there are two possible options: analyzing weaving structure and examining silk
thread. The first option, analyzing weaving structure, has been considered a possible way to find its
origin. However, this is an effective and accurate method when the weaving structure is unique and
complex enough to identify. The majority of silk supports for painted banners are woven plainly in basket
weaving style (i.e., one thread over and the other under), a structure that is too common to determine any
origins. Theoretically, to distinguish and determine the origin and quality of a plain weave silk, analyzing
the silk threads’ thickness and coarseness, which relates to the types of sericulture that silkworms would
have eaten, would be ideal. Sheng notes that in most cases the leaves of mulberry trees were fed to
silkworms, but there are other cases in which silkworms are fed leaves from other type of trees. For
example, if silkworms eat the leaves of oak, they produce a coarser silk. In Dunhuang, where “coarse”
silk was relatively common, the local production of silk and the feeding of the local silkworms is one
option to consider. Sheng, “Determining the Value of Textiles,” 179.

35
Where did Stein Painting 52’s silk come from then? This question is difficult to answer
because the silk support of Stein Painting 52 has such a simple weaving structure and lacks any
decorative motifs.
23
Even though the silk was probably transported from central China, the many
references to artisans linked to textile production, such as “mulberry tree artisan” (sangjiang â
Å) and “expert in dyes” (ranbu jiang ä^Å), that appear in documents from Dunhuang do
point to a significant local production center.
24
Moreover, the paintings from caves that are
contemporaneous with Cave 17—Caves 6, 98, and 196—show images of objects related to
textile production, such as a loom and a spinning wheel, thus signifying these processes’
valuable contributions to not only the local economy but also Dunhuang’s religious life.
The orientation of the silk threads can differ across painting workshops. For example,
Stein Paintings 101, 102, and 103 not only have a different painting style than Stein Painting 52,
their silk is also oriented in a different direction; their the selvedges (the edges produced during
the manufacturing process to prevent the woven fabric from fraying) are located on the top and
bottom, instead of at the sides as in Stein Painting 52.
25
Therefore, the height (51 cm) of Stein
Painting 103 is similar to that of the width of Stein Painting 52 (fig. 8).
The different preparatory practices for cutting silk material implies the existence of
multiple painting workshops, with each having different inventories of textiles for painting. Even
though not much research has been done regarding how painting materials were supplied to

23. A unique weaving structure can be a useful tool to determine the textile’s period and region. For
example, a jin ´ silk in compound weaves with a “chess-board” abstract pattern of tree-leaves is unique
to Turfan during the sixth to seventh centuries. Angela Sheng, “Innovations in Textile Techniques on
China’s Northwest Frontier, 500–700 AD,” Asia Major 11, no. 2 (1998): 117–60.
24. In P.5032v, sangjiang, Guo Yier (¨v≠) received two dou Æ of wine, which is equivalent to 12 or 4
liters depending on the size of the dou. Ma, Dunhuang gudai gongjiang yanjiu, 342.
25.

Tasaki, Tōrufan, Tonkō no bukkyō kaiga, 9.

36
painting workshops, it is plausible that monks from nearby temples would function as mediators.
Previously, it was suggested that there were two types of painting workshops in tenth-century
Dunhuang: the folk painting guild (huahang ]ã) and the government-run painting academy
(huayuan ]å) by Fraser.
26
Ma De added that the last type of workshops was managed by
temples.
27
When they needed to repair Buddhist statues or open a new cave, they hired artisans
on a per-project basis.
28
The lists of donors’ offerings and inventories at temples such as Jingtu
Temple and Longxing Temple indicate the presence at these temples of dye materials, glue, and
silk that had been dyed bright red (fei t) and purple (zi r).
29
The Stein Collection’s long red
and indigo painted silk banners, such as Stein Painting 154 (fig. 9), and the purple damask loops
on Stein Painting 52 support the idea that these types of dyes were used for painting materials.
When a workshop needed loosely woven raw silk (shengjuan) for a banner, therefore, they could
recycle previously owned textiles received as offering or payment or purchase the raw silks then
dyed them themselves as part of the preparatory process.
30

26. Fraser, Performing the Visual, 22–34.

27. Ma, Dunhuang gudai gongjiang yanjiu, 65–82.
28. For example, S.6829 shows that the temple paid for 60 chi of hemp in order to repair a throne for a
Buddhist statue. Ma, Dunhuang gudai gongjiang yanjiu, 75, 77.
29. Eric Trombert, “Cooking, Dyeing, and Worship: The Uses of Safflower in Medieval China as
Reflected in Dunhuang Documents,” Asia Major 17, no. 1 (2004): 67. In a different article, Trombert
argues that under monastic circumstances, hemp was more common as a payment method than raw silk
because silk was regarded as a consumer commodity. Along with hemp, I found that coins were also used
to purchase everyday items or raw materials such as glue. Eric Trombert, “The Demise of Silk on the Silk
Road: Textiles as Money at Dunhuang from the Late Eighth Century to the Thirteenth Century,” Journal
of the Royal Asiatic Society 23 (2013): 329; Ma, Dunhuang gudai gongjiang yanjiu, 75.
30. Trombert, “The Demise of Silk on the Silk Road,” 329.

37
1.3. Silk’s Characteristics and Painting Techniques
Regardless of whether a painting workshop purchased raw silk or utilized its own
material, preparatory treatment was required before silk could be painted on. The textile would
first be fixed onto a wooden frame with a thick thread or rope that wrapped the mount and
wound through the perforations along the border of the material. A recent conservation study on
Standing Avalokiteśvara (MG. 22799), painted on hemp, from the Pelliot Collection at the
Musée Guimet, reveals that the border of the painting has at least eighteen holes that were meant
to be fixed to the wooden frame with a thread (fig. 10).
31
Once the silk was mounted, outlines
were drawn. An image from Cave 72 shows what an in-process painting would look like and
how the painting support would be fixed a wooden structure when the outlines were finished (fig.
11).
Regarding the properties of raw silk, the interstitial space between the thin threads made
the support material porous and even gave it a sheer look. When considering the effectiveness of
the pigments, the support material does not provide ideal conditions for painters, as the pigment
can penetrate the intersectional spaces, thus wasting resources. Given this inconvenience, why
was this type of silk acceptable—and even preferable—as a painting surface in the pictorial art of
Dunhuang?
A close examination of Stein Painting 52 reveals the characteristics of the painting’s
support material and the practices that the artisans used. The openly woven silk enables a
reconstruction of the painting’s creation (in the order that the tasks were performed) because the

31. Violaine Garcia, “‘La quête des mérites’: Conservation-restauration d’une banniere votive peinte du
10th siècle provenant de Dunhuang (grottes de Mogao), conservée au Musée national des arts Asiatiques-
Guitmet” (dissertation, Institut national de patrimoine, 2015), 52.

38
material is so sheer. The faces of the central and surrounding bodhisattvas clearly indicate that
the outlining was drawn first with black ink. On top of the black ink, red ink traces these
guidelines—but in a hastier manner. This reveals that the outlines did not run or spread and that
underlining was the first step of the painting process.
The fact that the outlines are not blurred indicates that the silk underwent a preparation
process called sizing, a basic preparatory practice that prevents ink from running. Applying alum
or a glue solution not only stiffens the fabric but fills the interstices of the silk, making the
support material less porous and more manageable for painters.
32
Intriguingly, the natural protein
substance in shengjuan silk, sericin, may function as a sizer because of its gluey characteristics,
which may explain why shengjuan was the preferred painting material.
33
The natural tacky
surface enhances the support material’s ability to hold pigments with only a small amount of
alum and glue.
At this point, it is worth comparing shengjuan silk with another type of support: hemp. It
is well known that silk was popularly used for painted banners and hanging paintings but hemp
was another preferred material depending upon its availability. The choice of material was
determined either by the donor or the painter.
Hemp is known as an alternative to silk as a painting surface because the texture is
relatively crude and not delicate enough to deliver subtle painted details or expressions. It is

32. Winter, East Asian Paintings, 78. Other methods that would result in the same outcome as applying
alum and glue, are compaction techniques. By hammering or beating the silk, the surface of the fabric
becomes smoother, and the space in between the threads becomes smaller. This method does not apply to
Stein Painting 52 because of the large size of the interstices.
33. Winter, 78.

39
often claimed to be “much cheaper material than silk” and thus being suitable for “a less wealthy
donor.”
34
Contrary to the common perception, however, recent conservation and material culture
studies have determined that hemp was also considered valuable as it signified wealth.
35
In fact,
wealthy donor groups often preferred hemp as a main material for their votive object as it was
deemed suitable for the application of lapis lazuli, a rare and expensive pigment during the tenth
century CE in Dunhuang.
36
Hemp cloth is much more densely woven than raw silk or silk, and
due to its comparatively thick threads, the weaving structure does not leave intersectional spaces
between the warp and weft. As a result, pigments tend to stay on the side of the support they
were applied on rather than penetrate through to the reverse side of the cloth. Moreover, even
though the hemp thread is crude and rough, it is much more durable and resistant to
environmental conditions than silk thread.
1.4. Stein Painting 52’s Pigments and Red Lines
After sizing the material supports of a painting such as Stein Painting 52, outlines were
drawn with black ink, followed by coloring with pigments. The work’s main color palette is of
orange-red and darkened green, the latter of which has been altered in appearance from its
original state due to its reaction with oxygen. Three representative paintings, such as Seated
Avalokiteśvara from 983 CE (Stein Painting 54, fig. 12), Avalokiteśvara as a Savior of Perils

34. Zhao, Textiles from Dunhuang in UK Collections, 41.

35. Brack and Mysak, A Technical Study of Portable Tenth-Century Paintings from Dunhuang in US
Collections, 42–44; Garcia, “‘La quête des mérites’,” 65–66.

36. Garcia, 65–66.

40
(Stein Painting 2, fig. 13), and Seated Mañjuśirī from Cave 220, show stylistic similarities.
37

Even though there are variations in the appearance of the surrounding deities and the altars,
many choices—the colors, the canopies overhead, the facial expressions, the ways in which the
bodhisattva’s body is adorned with necklaces and scarves, the order of the painted colors in the
halo, and the use of a mandorla with a small-flame pattern—provide visual evidence that these
paintings and Stein Painting 52 were completed by the same painting workshop.
The painted portions of the paintings were finalized by emphasizing details and body
lines with red lines done by a different hand, evidence that a painting such as Stein Painting 52
was completed via a collaborative process that is partially illustrated in a painting on the south
wall of Cave 72. There we see a depiction of two artisans, each holding one side of a giant mount
that is placed next to a standing Buddha statue (see fig. 11). The painting on that mount, which is
not colored—only outlined—is waiting for artisans to complete it. Even though the
representation of the Buddha does not fit on the painting’s support—which could be symbolic of
the difficulties in capturing the Buddha’s image—it still captures an important moment in the
painting process. Next to the mount in the Cave 72 painting, an inscription is written in a
cartouche: “The moment when skilled portrait painters are invited to copy the image of his holy
face and true body (Qing danqing qiaojiang miao shengrong zhenshen shi).”
38

37. For donor images of Stein Painting 54 and their identifications, see Henrik H. Sørensen, “Donors and
Image in Dunhuang: A Case Study of OA 1919,0101,0.54,” Buddhistroad Paper 4, no. 1 (2019): 3–25.
38. The original text is “Ø∞±≤-≥¥µ∂∑∏” from Ma, Dunhuang gudai gongjiang yanjiu, 217;
Xu Ziqiang π∫ª, Chen Yongqiang ºΩª, and Zhang Jing æø, eds. Dunhuang Mogao Ku ti ji hui
bian ()?@A¿¡¬√ [Dunhuang Mogao Caves inscription compilation] (Beijing: Wenwu
chubanshe, 2014), 62. I modified Wu Hung’s translation in 1996 based on definitions by Ma De. Ma
clarifies that this inscription is an example showing mixed usage of a general term with a specialty term.
Wu Hung, “Rethinking Liu Sahe: The Creation of a Buddhist Saint and the Invention of a ‘Miraculous
Image’,” Orientations 27, no. 10 (1996): 42; Ma, Dunhuang gudai gongjiang yanjiu, 212–17.

41
This inscription and visual representation are important because they demonstrate that
painting in this era was not the work of a single painter. Moreover, the description indicates that
“a skilled portrait painter” completed the Buddha’s facial features, which had been left undrawn
until almost the final stage.
39
The inscription and the scene from Cave 72 can be interpreted to
mean that after a general-level painter completed the outlines using the nearby statue for
reference, two artisans brought the work to an advanced-level painter (i.e., a portrait painter) to
complete the image of the Buddha.
40

Cave 72’s inscription and images show that outlines were drawn by lower-level artisans
in this workshop. Once the outline was finished, a more advanced painter filled in the colors and
added accents. In Stein Painting 52 we see that red lines overlap with the body and face of
Avalokiteśvara. The red lines not only differentiate the figure of Avalokiteśvara from the
subordinate figures but also adds extra details to the painting.
41
These confident lines also correct

39. A ritual performance such as an “opening eye ceremony,” which is dotting the eyes, could have been
conducted by another specialist. This ritual functions as the transformation of a physical object into a live
divinity by activating “the eyes.”
40. This scene is part of the “Liu Sahe and Miraculous Image of Liangzhou” transformation tableaux
from Cave 72. The condition of the wall is not superb because the cave was half-filled with sand. Huo
Xiliang copied the entire south wall in his article in 1993. Figure 10 in this dissertation is part of his
drawing. That the story is originated from Linagzhou in which Dunhuang located and about the famous
Buddha image/statue at Fanhe County makes the painting extremely unique and valuable. To the right of
the painting, the Fan Buddha statue is measured by two craftsmen under supervision of a monk. These
two imageries represent how people wanted to duplicate the famous image in a two- or three- dimensional
form. Based on Huo’s line drawing, Wu Hung examines two modes of representation of the Fanhe
Buddha statue: conceptualized (a real Buddha) and materialized presentation (the copied image of the
Buddha). For the various editions and copies of the story, see Chen Zuolong æƒ$, “Liu Sahe Yanjiu ≈
∆«./” in Dunhuang ziliao kaoxue()»ïå… [Detailed Examination of Dunhuang Research
Materials] (Taipei: Shangwu Yinshuguan,1979), 212–52; Wu, “Rethinking Liu Sahe,” 41–43; Huo
Xiliang, “Mogaoku di 72 ku jiqi nanbi Liu Sahe yu Liangzhou Shengrongfo ruixiang shiji Bian” ?@A
B 72A ÀFG≈∆«WÃÕ¥Œ3œ–O—H[Cave 72 in the Mogao Grottoes and the mural on
its south wall depicting Liu Sahe and the history of the miraculous image of the holy-visage Buddha of
Liangzhou], Wenwu 2 (1993): 32–47.
41. Wong and Agnew, The Conservation of Cave 85, 169.

42
errors that were originally in the outline drawing, so we can deduce that they must have been
drawn by a senior member of the painting workshop.
42
After the red lines were added, final
touches were made in black ink, particularly to the figure’s brows, eyes, and mustache. After
being dismounting from the frame, the fabric’s perforated holes were covered with a different
type of silk that was sewn tightly around the borders to increase the painting’s durability. The
border of Stein Painting 52 is a dense, plain-woven silk as well as a piece of embroidered gauze
silk. Three loops of durable silk were sewn together so that the painting could be hung on a
horizontal rod.
Finally, a dedicatory inscription, dated to 972, was written on a stone tablet at the bottom
center.
43
The male patrons are wearing official uniforms, and the female patrons are wearing
bright yellow headdresses and elaborate garments with sleeves embroidered with floral
patterns.
44
According to the inscription, the main donor, Zhang Laza zç!—the middle of the
three male figures—was an infantry captain during the Return to Righteous Army (Guiyi jun é
èê) Period (848–1036 CE) under the Cao regime.
45
From the inscription, we also learn that

42. Wong and Agnew, 169.
43. The inscription can be translated as follows: "The Buddhist disciple, Captain of Infantry Troops
Zhang Laza, reverently painted an image of Avalokiteśvara with the wish that his deceased parents be
newly reborn in the [western] paradise, not falling into the Three Paths [!“, literally “unhappy roads”—
in other words, hells]. With the wish that the whole family—the old and the young—will live long and be
happy, the donor offers this with a whole heart: I will never blow out long-lasting incense smoke from a
lamp so that I can use the fire for another year. This dedicatory inscription was written on the sixth day of
the ninth month in the fourth year of Kaibo, a renshen year." Translation in English is mine, based on
partial translation by Arthur Waley. Whitfield, The Art of Central Asia 2 (Tokyō: Kodansha International,
1983), Pl. 26.
44. Whitfield, Pl. 26.
45. Imre Galambos reads the main donor’s name as “Zhang Laza.” According to Galambos, the given
name means “heap of garbage” and was not rare at all in Dunhuang. Previously, Whitfield read the
donor’s name as Zhang Geqiao, whereas Xu and Zhang identifies it as Zhang Geya (º”‘). I follow
Galambos’s reading. Whitfield, Pl. 26; Xu and Zhang, Dunhuang Mogao Ku ti ji hui bian, 442–43;
Galambos, Dunhuang Manuscript Culture, 190.

43
Zhang dedicated the image of Avalokiteśvara to his deceased parents so that they would be
reborn in the Pure Land and to ensure his family’s longevity and prosperity. The inscription was
written in a rather hasty manner; the last two characters of Avalokiteśvara’s name extend to the
edge of the altar cloth. The inscription was likely the final step of the painting and evidently the
calligrapher did not see any problem with writing outside of designated spaces.
Both the manner of the writing and the outlines of the figures in this work are comparable
to those found in other works commissioned by members of the highest social levels of society.
The Freer and Sackler Gallery’s Avalokiteśvara of the Water Moon (968 CE), which was
sponsored by Lady Li and the mother of Cao Yanqing ëáí under the Cao regime, is another
example of work by well-directed, skillful, professional artisans and calligraphers judging from
its carefully drawn brushstrokes as well as its colorings not extending to the outlines (fig. 14).
This exemplary painting must have been finalized by high-ranking painters—such as a chief of
materials or head designer (duliao ìî) or masters (boshi ïñ)—from the painting
workshop.
46
In contrast, Stein Painting 52 illustrates less delicate brushstrokes, which
corroborate the idea that highly skilled professional painters were not involved in small-scale
projects requested by commoners.

1.5. Framing and Mounting
Stein Painting 52 is framed with multiple pieces of silk textiles—patchwork—without
backing material.
47
One of the border fabrics is embroidered with satin stitching on dark green

46. For the definition of the terms as well as the structure of the Painting Academy, see Fraser,
Performing the Visual, 34–42.
47. Yu-Ping Luk, curator, British Museum, email message to author, January 28, 2019.

44
gauze (fig. 15). This fragment was originally part of the frame at the upper left-hand corner of
the painting as seen from Stein’s original photograph (fig. 16). In around 1920, the embroidery
was removed by the British Museum when Stein Painting 52 was remounted onto a panel.
48

Unlike the embroidery, which is backed with plainly woven dark green silk due to the gauze’s
fragility, the painting panel itself is not backed with any other material; it is only framed on those
multiple textile borders.
49

Framing a textile with another textile is a method that is certainly different from
conventional ways of framing and mounting; a brief review of Chinese methods of mounting will
help the reader understand the differences (fig. 17). A painting on silk is first backed with either
paper or silk, and then mounted with thin, monochrome silk borders (fig. 18).
50
The painting and
the silk borders at the middle, upper and lower middle, and top and bottom are separately backed
with paper or silk, followed by being joined together. They are backed again with additional
layers of paper.
51
A colored or patterned silk is often used for framing, and was usually placed in

48. Roderick Whitfield, The Art of Central Asia: The Stein Collection in the British Museum, Textiles,
Sculpture and Other Arts 3 (Tokyo; New York: Kodansha International in cooperation with the Trustees
of the British Museum, 1985), Pl. 5.
49. Dr. Luk, Basil Gray Curator of Chinese Painting, Prints and Central Asia at the British Museum,
confirms that the detailed image of the plain weave silk on the website for Stein Painting 52, in fact,
belongs to Stein Painting 52*. In other words, Stein Painting 52 does not have the plain weave backing
material but Stein Painting 52* (embroidered gauze fragment) does. Yu-Ping Luk, curator, British
Museum, email message to author, January 28, 2019.
50. Winter also noted that paintings on silk sometimes were backed with another piece of silk. Winter,
East Asian Paintings, 131.
51. Robert Hans van Gulik, Chinese Pictorial Art as Viewed by the Connoisseur: Notes on the Means and
Methods of Traditional Chinese Connoisseurship of Pictorial Art, Based upon a Study of the Art of
Mounting Scrolls in China and Japan (Rome: Istituto italiano per il Medio ed Estremo Oriente, 1958),
57–67; Valerie Lee, Xiangmei Gu, and Yuan-Li Hou, “The Treatment of Chinese Ancestor Portraits: An
Introduction to Chinese Painting Conservation Techniques,” Journal of the American Institute for
Conservation 42, no. 3 (2003): 463–64.

45
the upper and bottom part of the mounting to increase the work’s overall durability.
52
A pair of
decorative thin strips are pasted on the upper border. Finally, the upper and bottom rods are
wrapped with the borders.
The most notable difference between Stein Painting 52’s mounting and framing and
traditional Chinese mounting and framing is how the backing was accomplished. In Chinese
mounting, the edge of the top textile for the borders is attached and pasted to a painting. The
border textile is folded and sewn together with the edge of the painting, eliminating the extra step
of adding additional backing material to support the borders. Another possible reason for the
folded borders is to hide perforations or holes along the edges of the silk painting.
53

If the painted banner framing practice was not adopted from or inspired by the Chinese
method for framing, how can we understand such a unique method? One possible source can be
seen in the Tibetan Empire’s visual culture, which dominated the Dunhuang area in the eighth
and ninth centuries. During the occupation period, fifty-seven new caves were built and thirty-
one were renovated at the Mogao Caves.
54
These numbers clearly suggest that Tibetan resources
and presence at the site were prominent. Interestingly, covering a painting’s four edges with
colored textiles, as well as the way the painting was hung on a stretcher during its painting
process, resembles Tibetan religious painting practices for creating thangka.
55
Unlike the
mounting of a traditional Chinese painting, which basically consists of pasting a painting onto a

52. Silbergeld, Chinese Painting Style, 13.
53. As discussed earlier in this chapter, a painting support was hung over a stretcher. In order for it to be
fixed onto the stretcher, the support had to be perforated. Such holes can be easily hidden away by
wrapping them with another piece of fabric such as a decorative patterned silk.
54. Chen Shuangyin ’÷◊, “Tubo tongzhi dunhuang shiqi dongku xiujian jingji yuanyin chutan” ÿŸ
⁄¤()‹›|Afifl‡·‚„‰Â [the preliminary study on the economic reason of building caves
during the Tibetan occupation period in Dunhuang], Shijie zongjiao yanjiu Ê"Á†./ 4 (2009): 32.
55. Brack and Mysak, A Technical Study of Portable Tenth-Century Paintings, 30.

46
dense long silk cloth either for a hanging scroll or a handscroll, a Tibetan thangka is sewn and
framed with several pieces of cloth as is also the case with Stein Painting 52.
56

Reconstructing the Tibetan painting method based on the current practice is reliable to a
certain degree because it is regarded as what art historian John Huntington calls “a product of an
extremely conservative tradition.”
57
To take a closer look, let us examine the typical mounting
process of a Tibetan painting,
58
illustrated by contemporary Tibetan artist Wangdrak from
Shekar Dzong.
59
Once he finishes a painting, which is consecrated through writing sacred
syllables on the back of the work, he takes the painting out of its wooden stretcher or frame by
cutting all four sides and then the four corners. The four borders are then sewn with a colored
brocade (normally red is used first), which is then sewn with another brocade (yellow this time),
making a double frame. The framing can stop here, or the painting can be mounted onto a larger
patterned brocade. Embellishments can include a pair of long strips of fabric coming down from
the top of the largest brocade. The central image is supposed to be covered with a curtain, which

56. Because traditional artistic practices in Tibet have been preserved, the original way of constructing a
thangka painting exists today, but with some modern twists such as using Indian cotton muslin instead of
silk. David Jackson, and Janice A. Jackson, Tibetan Thangka Painting: Methods and Materials (Boulder:
Shambhala, 1984), 16.
57. He even mentions, “It is quite likely that there has been very little change in method since the
introduction of painting from India in the 8th century.” John C. Huntington, “The Technique of Tibetan
Paintings,” Studies in Conservation 15, no. 2 (1970): 132.
58. Even though Tibetan painting practice is generally considered conservative, this does not mean that
all painting practices from different parts of Tibet involve the same methods. There are, indeed, variations
and modifications to a minor degree depending on region and iconography. An example from Eastern
Tibet shows that because of the particular subject matter and style, there is a “rainbow” surrounding the
silk frame and a “door” in the lower section of the mounting. See Ann Shaftel, “Notes on the Technique
of Tibetan Thangkas,” Journal of the American Institute for Conservation 25 (1986): 102, figure 2.
59. David Jackson and Janice Jackson believe that tracing and reconstructing traditional Tibetan painting
methods from living Tibetan painters is still valid to understand the painting tradition in Old Tibet
because of the fact that the painters have used/relied on traditional pigments and painting treatises.
Jackson and Jackson, Tibetan Thangka Painting, 2.

47
is gathered on the top if the painting is on display or being used in a service. Thus, the Tibetan
mounting that creates the painting as a three-dimensional object is an essential part of the entire
religious practice.
60

The main difference between this and the Chinese mounting method is the use of colored
textiles as a frame. Tibetan Buddhist paintings are framed with brocade, a polychrome, patterned
silk, which is more elaborate and durable than plain-weave silk (fig. 19), while the Chinese way
of mounting prefers a muted color palette or monochrome silk that focuses more on
functionality, namely keeping the silk support from fraying and ensuring longer durability.
61
A
Chinese painting on silk is mounted or pasted on a larger monochrome silk without sewing
additional fabric, and thin, monochrome silk is used as a surrounding material.
62
The biggest
difference is that there is no sewing process involved in the typical process of mounting a
Chinese painting. Even when using silk brocade as panels, it is cut into tailored pieces and then
pasted on the upper and lower parts of a hanging scroll.
63

In contrast, Tibetan paintings involve a sewing process once a painting is completed.
Painted banners uncovered from Cave 17 show a similar mounting practice to the Tibetan
method. Judging from the extent of painted banners, the majority were framed and sewn with
multiple fabrics. Stein Painting 54, Seated Avalokiteśvara (983), maintains its original

60. Ann Shaftel, “Conservation Treatment of Tibetan thangkas,” Journal of the American Institute for
Conservation 30, no. 1 (1991): 9.
61. Brocade is called jin ´ in Dunhuang documents and defined as “textiles with discontinuous
supplementary weft yarns of different colours woven into the foundation weave in specific areas as
required by the pattern.” Brocade is often found in borders of paintings from Cave 17 and also streamers
of an oblong banner. Zhao, Textiles from Dunhuang in UK Collections, 122–23.
62. Winter, East Asian Paintings, 131.
63. A hanging scroll in typical Chinese mounting style can be found in Winter, 131, figure 8.6;
Silbergeld, Chinese Painting Style, figure 1.b.

48
appearance and is in a fairly good condition with the painting’s fabric frame and the five loops
on the top all preserved (see fig. 12).
64
It is framed with four different pieces of colored silk:
indigo blue, white, purple, and red. Their weaving structures are distinct from each other,
meaning they range from a plain-weave silk to a diamond-patterned damask.
65
These fabric
pieces are sewn around the painting with an overlap of a centimeter at the edges.
66

The order of and types of fabrics and their colors are designated: red brocade is sewn first
along the edges of a cotton support, and then yellow brocade is sewn on forming an outer border
without covering the back side of the main painting area. An example from the collection of the
Field Museum in Chicago demonstrates this (fig. 20).
67
For the last mounting step, which can be

64. Except for it being backed with modern Japanese paper, the painting has barely been treated since it
entered the museum’s collection. Whitfield noted that “a stave and roller have been added,” but it seems
that they must have been removed at some point (probably when the painting was mounted on a wooden
frame for storing purposes at the museum). See Whitfield, The Art of Central China, plate 27.
65. Based on my in-person study of the work, the indigo silk on the top and on one quartered side is plain
weave silk. Based on the white spot on the top right corner from the viewer’s perspective, the original
textile was clamp-resist dyed silk. It might be the case that a dyeing artisan failed to have the desired
pattern using the clamp-resist dye technique, so the fabric was repurposed as a border for the painting.
The yellowish, white silk, as well as the red silk on the bottom, are plain-weave silk. The purple silk is
damask with a diamond pattern, meaning it has 1/3Z twill for the pattern and 1/1 plain weave for the
foundation. The damask with a diamond pattern is widely used as a streamer for oblong banners and also
a border for a rectangular painted banner. For a similar type of damask with a diamond pattern, see Wang,
Persson, and Wood, “Dunhuang Textiles in London,” MAS 938, MAS 940, MAS 941, MAS 947, MAS
948, and MAS 949.
66. Whitfield, The Art of Central China, plate 27.
67. The reason why the backside of the painting, where more protective supports should be needed, was
not covered is uncertain. One possible reason could be for the consecration ritual, which includes
inscribing the sacred syllables, oṃ āḥ hūṃ, on the backside of the image. The other reason could be that
the painting had its own cover, mounted at the top. In other words, the front of the painting was perhaps
covered when not in use, and this being considered a more direct and reliable protection for the painting,
the backing was not covered. The top and the bottom parts of the textiles wrap a wooden slat or rod to
hold the painting firmly. In the case of the Dunhuang banners, color-blocked banners, and a banner with a
single iconic figure, wooden rods help keep it shape. Jackson and Jackson, Tibetan Thangka Painting:
Methods and Materials, 143; Field Museum, “Thangka Preservation,” Field Museum, accessed on
February 12, 2019, https://www.fieldmuseum.org/science/research/area/conserving-our-
collections/thangka-preservation.

49
often omitted, another cotton or brocade silk piece is used for mounting.
Above all, the biggest difference between Stein Painting 52’s mounting and a Tibetan
thangka’s mounting is in the use of a large textile to mount the double-framed painting at the
center. As seen in Stein Painting 52, it omits the second frame and the large brocade mounting
that would have been placed in a thangka (see fig. 19 for comparison). Instead of a wooden rod
and hanging strings, three to four loops of damask silk are attached on the frame directly without
having an additional mounting. Based on the Tibetan example, one possible way to think of the
original mounting method for Stein Painting 52 is that the painting could have been attached or
sewn with a larger textile to enhance its durability. In the Stein and Pelliot Collections, there are
a large number of fabrics fragments and patchwork areas. In the Stein Collection, a large
patchwork (MAS. 856) is a possible candidate for purposes of mounting (fig. 21). Ranging from
delicate embroidery to twill damask, the patchwork shows an elaborate effort with its clamp-
resist dyed border. Several possible purposes and functions of this piece have been proposed,
including it being a simple votive object, a kasaya (a patchwork robe for a monk), a wrapping
scroll, or a bed cover.
68
Considering its size, 150.5 cm long by 111.0 cm wide, it is too small for
a monk to wear.
69
However, it is large enough to mount a votive painting such as Stein Painting
52 as its size (102.2 x 69.0 cm) would fit within the patchwork. According to thangka
conservationist Ann Shaftel, a painting can be removed from the mounting and displayed
separately; the mounting still is “an inherent part of the thangka.”
70

Among the Stein and Pelliot Collections, on the other hand, there are a few examples of
textile frame borders. For example, a frame border (MG. 26789) from the Musée Guimet is a

68. Zhao, Textiles from Dunhuang in UK Collections, 113.
69. Zhao, 113.
70. Shaftel, “Conservation Treatment of Tibetan Thangkas,” 4.

50
more relatable and direct option for a painting mounting (fig. 22). Even though one loop on the
right corner of this piece is missing, the border has two loops on the top, which strongly implies
that the border functioned as both a method of framing and of mounting. In other words,
Dunhuang’s hanging paintings are often framed with a separate cloth and hung directly using the
loops. Framing with textile borders is similar to Tibetan thangka practice with a larger textile for
the provisional mounting. Simplified mounting was considered suffice for travelers and
practitioners who valued portability and compact size. It also reduced production cost by
omitting the mounting cloth.

1.6. Displaying and Hanging Method
The question to follow is how Stein Painting 52 would have been displayed and hung.
Was there a separate piece of textile on which such a sheer painting was sewn for each use? Or
was it supposed to be hung in its current condition on a hanging device, if such an implement
ever existed? Furthermore, was the painting displayed inside or outside of a cave structure?
These questions are crucial to a full reconstruction of its functional context and original use.
A piece of plain-woven silk with large intersectional spaces does not provide enough
stability for hanging a silk banner even if its edges were covered with other silks. The first
possible scenario for hanging Stein Painting 52 was that it was sewn onto a large piece of textile
only when the painting was in use. This would prevent the painting’s support, a mere single layer
of plain-weave silk, from being damaged, and would also provide some weight to the painting so
that it will not be folded or flipped while hanging. For this possibility, however, one would still
need to explain why there are loops on the top of the banner. If the loops were solely decorative,
their size would not need to be large enough for them to function as hangers. I would argue that

51
the loops indeed had a practical function, as a wooden rod or similar hanging implement could
pass them through.
Pictorial examples of a wooden stick used as a hanging implement can be found in
several mural paintings, including those in Cave 125 and Cave 220. Even though these
depictions are limited to aniconic, color-blocked, oblong banners, they provide compelling
evidence that wooden rods were used as hanging implements. In the case of the color-block
banner, two different methods of hanging are visible. In one, the top of the oblong banner is hung
directly on a dragon-headed pole, and, second, a dragon-headed pole is tangled with thin white
clothes, which are fixed to a long, wooden rod. The top of a triangular head of a banner is tied
tightly at the middle point of the rod (see figs. 36–37). A clear representation of the second case
can be seen on the north wall of Cave 220, where a pair of very tall poles with swaying banners
is depicted on either side of the Eight Buddhas (fig. 23).
Unfortunately, a visual rendering of a rectangular painting like Stein Painting 52 does not
exist among the murals of the Mogao Caves.
71
One scene from Cave 148, however, proposes
how a painting of icons on silk (or other textile) could have been displayed. On the east wall of
the cave, which is from the High Tang period, a ritual in progress is depicted, and a wooden pole
with a flying banner is seen placed in the courtyard of temple buildings (see fig. 54). The temple
is a space enclosed by a corridor and two shrines. Inside of the main building, painted images of
the Buddha and two bodhisattvas are displayed. A pair of color-block banners at the sides of the

71. A possible reason for the absence of the actual description of hanging paintings on Buddhist deities is
that they are normally described as a non-material form, except for a few instances in which they have to
be presented in either a painting or sculptural form based on a narrative sequence. In these cases they are
depicted as lively beings that are interactive with viewers. A representative example of the Buddha being
in sculptural or pictorial (painting) form is Liu Sahe’s miraculous images at Fanhe from Cave 72. Wu,
“Rethinking Liu Sahe,” 32–43.

52
bodhisattvas appear to be in motion, as if moving because the Buddhist deities, summoned
during the ritual, have become animated. Monks and a donor are seen sitting next to the altar.
Based on the candle-lighting structure placed in another shrine on the opposite side, it is believed
that the scene shows a ritual related to the Bhaisajyaguru.
72
From this scene, we can see that a
painting with iconic figures was often accompanied by decorative banners at its sides when used
in a ritual.

1.7. Skilled Labor in Dunhuang and Artistic Creativity
Stein Painting 52 and its production process reveal the ways that skilled labor was
organized in Dunhuang, which expands the discussion of production to include the ways artisans
interacted with others in different regions or the extent to which a monastic authority affected the
overall painting process. I have suggested a close tie with the Tibetan Empire’s visual culture in
creating a mounting and hanging method. This scenario is even more sensible because of the role
of Khotan. As one of the largest oasis settlements along the Silk Road, Khotan was located in
southwestern Xinjiang, approximately 1,325 km to the east of Dunhuang (see fig. 1).
73
To the
south, the Kunlun Mountains form a natural boundary with Tibet. The Tibetan influence on
Khotan was more direct and immediate than it was on Dunhuang. It is more accurate to say
Khotan was dominated by Chinese culture in the eighth century CE because its official
documents were often written in both Chinese and Khotanese. However, because the Tibetan
Empire occupied Khotan twice (see Chapter 4), Khotan was—similar to Dunhuang—one of the

72. Ning Qiang, Art, Religion, and Politics in Medieval China: the Dunhuang Cave of the Zhai Family
(Honolulu: University of Hawaiʻi Press, 2004).
73. Hansen, The Silk Road, 199.

53
oases that had close affiliation to the Tibetan culture.
While much of this dissertation focuses on the Tibetan influence on visual culture, it also
aims to highlight other multicultural resources that would have affected the visual culture in the
city of Dunhuang. Narrowing the dissertation down to the close relationship between Khotan and
Dunhuang provides a snapshot of artistic practice, especially between these two oasis cities. The
close relationship between Dunhuang and Khotan reached its apex via sending envoys to
Dunhuang during the tenth century CE, when the Cao regime ruled the Dunhuang area. The
marriage between the daughter of Cao Yijin and the Khotanese King Visa Sambhava (or Li
Shengtian }óZ, r. 912–966) reinforced their royal ties. Even their crown prince took
residency in the Sanjie Temple.
The close ties between the two regions accelerated and connected their artistic
communities. As Chapter 4 will discuss, Khotan played a pivotal role because it was in close
proximity to a lapis lazuli mine in Badakhshan, thus providing precious mineral pigment to
Dunhuang even when the direct trading route between Dunhuang and the mines was blocked
during the Muslim occupation in the ninth century CE.
74
In addition, the localized Buddhism in
Khotan related to Vaiśravaṇa in the north and “Auspicious Images” (ruixiang òY) meant these
subjects were popular for paintings in Dunhuang.
75

Under these circumstances, one might ask whether artisans were allowed to travel from
one region to another to share their artistic knowledge and practice. In this regard, a Khotanese

74. Brack and Mysak, A Technical Study of Portable Tenth-Century Paintings, 25–26.

75. Zhang Xiaogang ºËÈ, “Dunhuang ruxiang tu zhong de Yutian huguo shen wang” ()œ–ÍÜ
èÎÏÌáÓç [Protection gods of Khotan found in auspicious image paintings of Dunhuang],
Dunhuang yanjiu 1 (2005); Christoph Anderl, “Linking Khotan and Dūnhuáng: Buddhist Narratives in
Text and Image,” Entangled Religions 5 (2018): 250–311.

54
letter written by Khotanese envoys has an interesting remark on a monk-painter: “We came here
to Su-chou [Suzhou]. Here there are two ‘painter-monks’ of Phema. From these monks, we heard
that: ‘From China, a Ta-shih [Da-shi] departed (towards us), Rajarimai Yaugandharayana by
name.’”
76

Suzhou is one of the towns in the Dunhuang area, and the envoys met two painter-monks
of “Phema.” According to Hiroshi Kumamoto, Phema refers to a town between Khotan and
Niya, meaning the painter-monks were able to travel.
77
It is unclear why they traveled, but it is a
rare example that shows the high possibility of their mobility. Based on the example, artisans
could also likely travel between Dunhuang and Khotan. By traveling from one region to another,
these artisans had a chance of becoming familiar with new motifs and iconography, even if they
were less likely to have the authority to import their practice, especially when they were at the
novice level.
Monks who had the dual duties of painting and monastic practices lived in Dunhuang,
which can be confirmed by tenth-century inscriptions found in Dunhuang. Such inscriptions are
found on the north wall under the eaves of Cave 444, one of which indicates that Fan Dingquan
Üôö, a monk-painter (huaxiang huashi õú]ù), visited the cave in 975 CE.
78
Another
inscription written on Stein Painting 3 is also an interesting example, part of which reads as

76. Hiroshi Kumamoto, “Khotanese Official Documents in the Tenth Century A.D.,” (PhD diss.,
University of Pennsylvania, 1982), 119.

77. Kumamoto, 184.

78. In the inscription, the character, shi (Ô), is not found, but Ma reads it as a painter, huashi. The
inscription is transcribed as follows: “Ò%á!ÚÛÙ(ı)ˆ˜¯‰!˘˙˚ÄA!¸˝˛ˇ!.”
Ma, Dunhuang gudai gongjiang yanjiu, 345, 459; Xu and Zhang, Dunhuang Mogaoku ti ji hui bian, 385.

55
follows: “Cili ûü, elder of the Yong’an Temple, piously and reverently painted Avalokiteśvara
Bodhisattva, on behalf of his departed father.”
79
This line of the inscription is often understood
as a merely formulaic and conventional phrase that indicates the monk was not actually
responsible for the actual act of painting, even though the verb hua ] (to paint) was written. It is
also because another verb, “to make” or zao †, was used in the other donor’s name. Among the
donors in the inscription, the “believer and performer Dong Wenyuan °¢£” was considered a
painter, according to Ma.
80
However, the inscription does not clearly say that Dong Wenyuan
painted the picture either. If the former expressions associated with Cili were merely
conventional, then there is no reason to consider Dong Wenyuan as the main painter. In this case,
the painting might well have been painted by completely different artisans and the actual donors
merely sponsored it. If we read the inscription differently, however, the situation becomes more
straightforward and sensible. In other words, Cili might have participated in the production of the
mural as an actual monk-painter alongside other painter or as a supervisor in the painting
process.
In addition, analyzing surnames of artisans found in written documents and inscriptions
in wall paintings and portable paintings yields an interesting snapshot of how diverse and
multiethnic the artistic community in Dunhuang was. In Ma’s survey of artisanship and artistic
labor in Dunhuang, he made an index of names of artisans that have written references. The
number of the artisans was 136, which could be two or three times higher because unknown

79. This painting, which features a mirror image of two standing bodhisattvas, also has a mirrored
inscription at the upper center. For this peculiar writing method, see Galambos, Dunhuang Manuscript
Culture, 186–189.

80. Ma, Dunhuang gudai gongjiang yanjiu, 462.

56
artisans were not recorded in these documents and references. It is undeniable that Han Chinese
surnames such as Wang F, Ma §, Zhang z, and Fan Ü make up the majority of the group.
However, non-Chinese surnames such as Shi •, An P, Kang ¶, Mi ß, and Li } also attract
attention.
81
The first three are well-known surnames closely related to the Sogdians. In the case
of Li, it is undeniable that it is one of the most common Chinese surnames. At the same time, it
can be a Khotanese surname, which was predominantly used by the Khotanese kings (e.g., Li
Shengtian) after adopting Tang’s imperial surname.
82

In Ma’s table, a considerable number of these Central Asian names are found; for
example, four artisans have the surname Li, five artisans have the surname An, eleven artisans
have Shi as a surname, and six artisans have the surname Kang. Even though it is not possible to
conclude the artisans’ cultural affiliation solely by their surnames, especially in such a
multicultural region as Dunhuang, it still provides referential information on their ethnical origin
and cultural background. Therefore, whether private, public, or monastic, a painting workshop
had ethnic diversity, which may explain the variety of styles in the pictorial materials from Cave
17. Judging by styles, it is difficult to count all of them. Stein conveniently categorizes “Indian”
or “Chinese” styles mainly determined by clothing styles, but it is common for the two styles to
be mixed within a single picture frame. In Stein Painting 5 (864), the four bodhisattvas in the
upper section are painted in “Indian” style, whereas Mañjuśirī and Samantabhadra in the lower
part represent the “Chinese” style.
83

81. Galambos, Dunhuang Manuscript Culture, 222.

82. Xin Wen, “What’s in a Surname? Central Asian Participation in the Culture of Naming of Medieval
China,” Tang Studies 34, no. 1 (2016): 92.

83. Whitfield, Art of Central Asia I, pl. 23.

57
In a broader perspective on artistic practice, diverse painting styles were acceptable as
much as ethnic diversity in the organization of painting workshops was allowed. It also relates to
how a painting was ordered and produced. As discussed earlier, two modes of production
coexisted: made-to-order and ready-made. The ready-made or prefabrication practice would
allow a workshop to replicate the same style in a large quantity. The groups of standing
bodhisattvas in profile, such as Stein Paintings 104, 120, Stein 156, and EO 1170, exemplify this
scenario. The empty cartouche in these paintings testifies that they were prefabricated. On the
other hand, if a painting were made to order, then a monastery would either hire a professional
painter or ask a monk-painter if he were available. It then depended on a workshop’s manual
predetermined process that would follow. In contrast to the ready-made process, the availability
of artisans would affect a painting’s style, even though they might not have had the permission to
draw something completely new. I believe Stein Painting 52 displays the made-to-order process
executed not by a freelance painter but by a private, organized workshop hired by a monastery,
because a very similar copy, Stein Painting 54, was made eleven years (in 983) after the
dedication of Stein Painting 52 in 972.
This leads me back to the debate on the artistic creativity of the artisans in Dunhuang that
I introduced earlier in the chapter. The painting process of Stein Painting 52 shows that the
process was rather straightforward. In fact, the main goal for the artisans was to produce similar
images efficiently and effectively, as indicated by the similar style, color palette, and
composition shared by Stein Paintings 2, 52, and 54, all of which were completed in the same
workshop. Even though Stein Painting 52 and Stein Painting 54 were painted eleven years apart,
they still maintained nearly identical styles, color palettes, and sizes of the silk support. The size
of Stein Painting 52 is 91.5 cm in height with 59.1 cm in width, without the textile frame. The

58
painted area of Stein Painting 54 measures 99 cm in height and 59.5 cm in width. Moreover, to
include the bottom row of the donors’ portraits, an additional piece of textile was sewn below the
upper register of the donor figures. In addition, the large gap between the eyebrows and eyelids,
as well as the crown from which a seated Amitābha is depicted, share similar stylistic
characteristics. The halo and mandorlas are nearly identical by alternating green and orange. The
outermost mandorla from both paintings is painted with bright white pigment, which indicates an
emission of light and flame as the stylized flame pattern surrounded the outermost circle. This
pictorial evidence corroborates that the artisans were from the same workshop and followed the
workshop’s artistic norms without exerting individual creativity and free will.
The last important factor related to the issue of artistic creativity is the role of patronage,
as art historian Sonya Lee has pointed out. The fact that another goal artisans had was to satisfy
their patrons’ wishes and requirements should not be overlooked. Lee demonstrates that the three
earlier Cao family caves influenced the later construction of Cave 61 in terms of its architecture,
pictorial program, and compositional styles.
84
Specific motifs and arrangements were tailored to
individuals’ needs and wishes. This case also shows artisans had relatively little room to create a
new composition or a novel visual language freely to express artistic creativity. It is crucial,
therefore, to reassess patrons’ requests or intentions and the ways they were reflected in the final
pictorial language. In case of Stein Painting 52, it is impossible to conjecture the patron’s
detailed wishes—other than what was written in the inscription—which they would have
delivered verbally to the artisans. The choice of iconography was determined by the patron:
Zhang Laza’s wishes were that his deceased parents would be reborn in the Pure Land and that

84. Sonya S. Lee, “Repository of Ingenuity: Cave 61 and Artistic Appropriation in Tenth-Century
Dunhuang,” Art Bulletin 94 (2012): 210–213.

59
his family would attain longevity and prosperity. Stein Painting 54 has the same deity at the
center but uses a simpler composition for the surrounding attendant figures: The Youth of Good
and the Youth of Evil, who are more popularly associated with and represented by Kṣitigarbha in
relation to keeping a record of the karmic behaviors of the dead. This rather unique combination
might reflect the patron’s request. The accompanying inscription does not contain typical wishes,
such as for being reborn in the Pure Land, longevity, or prosperity. Sørensen also notices this
peculiarity, noting that “the presence of the Youths of Good and Evil in the painting also makes
it clear that Mi Yande, the principal donor, and his relatives were keen to have their good deed,
i.e., the donation, recorded in the divine registers.”
85
His statement reflects that the pictorial
arrangement can be deeply associated with a donor’s personal preference and specific choice of
pictorial element. Thus, it can be safely concluded that the workshop was not the place for
artisans to exercise free will and artistic creativity, as they were limited by workshop hierarchy
and the demand to satisfy patrons’ requests.

1.8. Conclusion
To reconstruct the ways in which Stein Painting 52 was created, I have examined the
types of artistic labor that went into the painting—not only the main pictorial subject but also the
surface materials behind the pigments. The natural stickiness of the painting’s raw silk made the
material suitable for painting even though it still required a preparatory process such as sizing. Its
delicate surface does not hide its outline drawings, which transparently reveal the labors of
several painters and the hierarchical levels of their workmanship. The availability of raw silk, as

85. Sørensen, “Donors and Image in Dunhuang,” 20.

60
well as its relative affordability, made it a popular material for painting.
By combining historical aspects of textile and the painting’s microstructure, this chapter
paper attempts to understand the role of materials in the painting process and the types of artistic
considerations that had to be given to the materials’ characteristics. The types of textile
workshops, their relation to painting workshops, and the circulation of preparatory materials
among the painting academy and local painting workshops are topics that require further
research. Nevertheless, this study contributes to a focus on the painting’s support and the artistic
labors that made Stein Painting 52 possible, as well as its mounting and hanging issues. The
artisanship of Dunhuang’s skilled artisans reflected in Stein Painting 52 has been weighed from
diverse perspectives, such as artistic exchange between neighboring regions, diversity in
workshop organization, and patrons’ role in the production process.

61
Chapter 2. Stylistic Variations and Materialistic Differences in Avalokiteśvara Sutra
Paintings
2.1. Introduction
Throughout the history of the Mogao Caves, the design of the caves’ structure and the
subjects depicted in its murals changed in accordance with not only religious beliefs but also
artistic practices. From the Middle Tang period or mid-eighth century onward, changes in the
design and layout of pictorial elements, including sutra paintings, decorative patterns, and their
presentation in the central niche in the west wall, became prominent.
1
On the south and north
walls of a cave, the upper part of the walls contain at least three sutra paintings, and the lower
parts contain narrative scenes from chapters of these sutras. Departing from the presentation
norm of the earlier period, which emphasized one or two large main sutra paintings on the north
and south walls, decorative patterns along with an increased number of sutra paintings were
widely applied from the Middle Tang period onward. The complicated decoration scheme and
the display of multiple images in one space certainly implies that the visual paradigm for
constructing a cave-temple had shifted from previous times. New religious interests and practices
that had been introduced to and widely accepted in Dunhuang prompted this increase in the
number of distinct images and greater diversity in subject matter. The production of portable
painted images on textiles also increased during the period, as proven by the large number of

1. Dunhuang wenwu yanjiusuo, Zhongguo shiku: Dunhuang Mogaoku 4Ü"2A:()?@A [Chinese
caves: Dunhuang Mogao Caves] (Beijing: Wenwu, 1987), 161–62.

62
paintings from the ninth to tenth centuries found in the Library Cave.
2
These banners included
sutra paintings, single and multiple bodhisattvas, heavenly guardians, mandala-like paintings,
and decorative banners.
Wall paintings in the Mogao Caves have been popularly spotlighted for their artistic
value, religious and political meaning, and spatial function. The vibrant colors of the pigments
and spatial illusion created by the murals and the placement of each painting in an artificial cave
space certainly captivate public and professional attention. Despite their large quantity, painted
banners found in the Library Cave of the Mogao Caves have not received as much attention as
the wall paintings. The concurrent increase in the number of painted banners and wall painting
motifs at Mogao raises questions about their possible interrelationship, especially as they often
shared identical pictorial subjects, embellishments, and painting methods and techniques.
Comparing these two forms yields not only a possible correlation between the painting
techniques but also parallel design principles and artistic ideas that were popular during the
Middle and Late Tang dynasty. This chapter will also ask whether painted banners were
produced as counterparts to the mural paintings or as condensed/miniaturized versions of the
murals, and also whether they were intended to be used as pictorial alternatives to the mural
paintings.

2. This statement is debatable because the paintings from the Library Cave are mostly from the ninth to
tenth centuries. This does not mean, of course, that no painting was made in the previous period. Indeed, a
few paintings and embroidery textiles from the Northern Wei period were found during archaeological
excavations in front of Caves 125 and 126 in 1965. Even though they are fragmentary, such findings are
important for visual representations of a dedicatory inscription and donor figures of the earlier period. See
Dunhuang yanjiuyuan ()./#, “Xin faxian de Beiwei cixiu” J$%è&'() [newly discovered
embroideries from the Northern Wei], Wenwu 2 (1972): 54–60.

63
The murals and the banners can be analyzed from multiple perspectives. First, they can
be compared in terms of their subject matter. Based on the contents of Cave 17, there are
considerable thematic overlaps among murals and banners, just as certain motifs were specific to
each medium. What were the most popular themes shared by both murals and banners? Did the
artists show any preferences for one or the other when deciding on subject matter? If so, what
were the reasons behind their selection? Conversely, if a particular theme appeared in both
formats, how were they different or similar in terms of composition and arrangement of pictorial
languages?
Based on a comparative analysis of subject matter for murals and banners, this chapter
further explores how their artists worked with each medium, as well as the strategies and
pictorial devices they applied to achieve the desired visual effects. I show that the cave-temple’s
space was filled with patterns referencing textile examples, which include not only paintings on
textiles but also woven textile patterns. A close analysis of the paintings in Cave 112, one of the
mid-sized caves from the Middle Tang period, highlights the ways in which sutra paintings were
arranged and painted in the main chamber (fig. 24). Based on an observation of decorative motifs
in the cave, I argue that textile products provided an inspiration for the border patterns framing
the wall paintings.
Textile production and its influence on wall paintings raises questions about whether the
cave was intended to recreate the ways in which hanging paintings would customarily be
arranged inside an architectural setting. Characterizing the space itself would help us understand
the cave’s function, which has been a debated topic in Dunhuang studies. Art historian Winston
Kyan argues that the murals’ decorative borders testify that the space was reproduced to mirror

64
domesticity: a space for familial rituals and ancestor worship.
3
Because he focuses primarily on
the western niche’s architectural element (tent-shaped niche, zhang ®) as the representation of
“domestic couch,” my approach here is to explore the possibility of the borders inside of cave
and portable paintings’ frames in textiles sharing a parallel function. In other words, the
overlapping textile patterns exploited both by mural paintings and portable paintings suggest that
how the mural paintings and portable paintings are visually related.
Cave 112 also includes an image of Avalokiteśvara as a savior of perils on the left side of
the east wall, an image that was also depicted on several painted banners. A comparison of two
versions of this scene exemplify the differences that were created to fulfill each painting’s
designated function. In this regard, with Cave 112’s visual style as a main example, the
following section compares four Avalokiteśvara Sutra paintings (showing Avalokiteśvara as a
savior of perils) in terms of their composition and design (based on their material) as well as
their stylistic characteristics. The painting style of the image of Avalokiteśvara in Cave 112 is
what is often called as “Indian” style, based on their physical appearance and the way of wearing
garments. The custom of labeling “Indian” cast questions about the identity of the artisans. Does
the painting style associate with artisan’s ethnical identity? What does these different painting
style and labeling imply about the painter’s group?
Comparing the paintings’ color schemes is also helpful in determining whether a
relationship exists between the painting techniques used in the murals and the painted banners.

3. Winston Kyan, “Family Space: Buddhist Materiality and Ancestral Fashioning in Mogao Cave 231,”
The Art Bulletin 92, no. 1–2 (2010): 65–68.

65
An analysis of the colors provides a better sense of the types of pigments that were used in both
types of paintings.
To situate the Avalokiteśvara Sutra paintings in the Dunhuang visual culture during the
Middle Tang period, it is important to briefly review major changes in layouts and arrangements
of cave paintings that occurred in this period. In general, the pictorial planes of the walls on the
south, north, and east sides of caves were divided into upper and lower registers and the variety
of sutra paintings increased. In the upper sections, scenes of a Buddha preaching before an
assembly of disciples and bodhisattvas and paradise scenes became the most frequently selected
subjects in sutra paintings, including paradise scenes of Śakyamuni, Amitābha, and
Bhaisajyaguru (the Medicine Buddha), as well as panoramic views of Mount Wutai or Five-
Terrace Mountain (Wutai Shan ©™´). In the bottom sections, either details from the above
themes or stories from the life of the Buddha and his past lives were included.
4

Except for Mount Wutai, many of the paradise scenes were also depicted on painted
banners. Specifically, shared pictorial subjects include the life of the Buddha, a single
anonymous bodhisattva, the Guardian Kings (lokapāla), the Savior Bodhisattva (jiuku pusa ¨≠
jk), the Thousand-Armed and Thousand-Handed Avalokiteśvara, Mañjuśirī, Kṣitigarbha, the
Pure Land of Bhaisajyaguru, and the Pure Land of Amitābha.
The single bodhisattva is the most popular motif, which can be easily located in both
painted banners and murals in the Mogao Caves. The major locations of the overlapping subject
matter are the sides of central niche on the west wall and the lower parts of the north and south

4. Dunhuang wenwu yanjiusuo, Zhongguo shiku: Dunhuang Mogaoku 4, 161–65.

66
walls, forming a screen panel. The former is the most frequent and the easiest to locate. In a
narrow and paired space due to the central niche, the paired bodhisattvas are a suitable option for
filling the space. For example, Cave 14 has twelve more bodhisattvas, which are painted on the
lower parts of the south, north, and east walls. The flanked bodhisattvas fill the pillars’ front
planes, introducing the main chamber on the east. The painted banners of a single bodhisattva are
comparable to this type and the location, which indicate that one of the possible placements of
the painted banners is the flanking sides of the main composition. The second most frequent
placement is the lower parts of the north, south, and west walls. In this case, multiple
bodhisattvas are arranged along the walls. They are depicted on a smaller scale, leaving an
impression of multiple screen paintings in a row, with each section containing between five and
fifteen standing bodhisattvas. With slight variations in mudras, directions of the face, and
decorative ornaments, each bodhisattva has a distinctive pictorial description but not significant
enough to enable identification of its iconography in detail.
5

Thematic subjects that center on a dialogue between two figures or a topographical
landscape without a major iconic image, were less frequently drawn on the banners. These
include the “Magic Competition” between Raudrāksha and Mañjuśirī, Mount Wutai, and
transformation tableau of the Deva’s Inquiries Sutra (Tian qingwen jing ZÆØ=). At the same
time, some subjects were not used in the mural paintings but were commonly employed in the
painted banners, including the Bodhisattva Who Leads the Way or Yinlu Bodhisattva (Yinlu pusa

5. These comparisons are based on the Stein Collection and the Pelliot Collections, both of which are
catalogued in Whitfield, The Art of Central Asia: The Stein Collection in the British Museum (Tokyo:
Kodansha International, 1982); Jacques Giès, Hero Friessen, and Roderick Whitfield, Les arts de l’Asie
centrale: la collection Paul Pelliot du Musée National des Arts Asiatique-Guimet (London: Serindia,
1994); and Dunhuang yanjiuyuan ()./#, ed. Dunhuang shiku neirong zonglu ()2A*µ+,
[Catalog of pictorial contents in cave temples of Dunhuang] (Beijing: Wenwu chubanshe, 1996).

67
∞±jk) and the Descent of Amitābha accompanied by Avalokiteśvara and/or Kṣitigarbha.
6

While my intention is not to strictly categorize pictorial subjects as “banner-exclusive” and
“mural-exclusive,” a rough sketch of popular topics for painted banners helps pave the way for a
scrutiny of their purpose within the context of Buddhist visual culture in medieval China.
Finally, Savior Bodhisattva (jiuku pusa) is a subject that appears frequently on both
caves’ walls and in textiles. In both media, this bodhisattva is often represented either
independently as a single unit or with surrounding peril scenes from the Avalokiteśvara Sutra,
which was originally part of the Lotus Sutra. The jiuku pusa is an interesting example because it
appears in two different versions depending on the material. Thus, it provides an important
example of how artisans considered material a factor in conveying a subject’s religious
connotation. Constant dialogues between mural paintings, textiles, and portable paintings exist,
as the Savior Avalokiteśvara painting in Cave 112 shows.

2.2. Cave 112: Changes in the Cave’s Design and Decoration

6. The Bodhisattva Who Leads the Way, which first appeared in the late Tang and early Song periods, is a
motif that depicts the bodhisattva taking a recently deceased person (or one on his or her deathbed) to the
Western Pure Land where Amitābha Buddha resides. The name of Yinlu pusa is not found in canonical
sutras but several painted banners from Cave 17 such as Stein Painting 47, MG. 17657, and MG. 17762,
are identified as “Yinlu pusa” based on the inscriptions. One major iconographical feature is that the
bodhisattva holds a long staff with a swaying banner. Sha, Dunhuang huagao yanjiu, 373. That some
subjects were not used in the mural paintings but were commonly employed in the painted banners is
quite an interesting phenomenon when one considers the full range of visual topics in the Mogao Caves’
wall paintings. The paintings related to Bodhisattva Who Leads the Way are called Laiying paintings,
which are associated with a deathbed ritual. The tactile experience—physical touch using a thread
attached to the painting—of the person on the deathbed was considered the most important quality of the
special use. A painted banner was a more suitable medium than a wall painting because it provided
portability and easy accessibility.

68
Cave 112’s structure and design plan are representative of the Middle Tang period, as it
has a truncated pyramidal ceiling and a niche in the west wall in which the tented canopy (zhang
®) structure and the couch-shaped platform are installed. This cave exhibits six sutra paintings
in total (on the south, north, and east walls), six screen paintings inside the western niche, and
paintings of Samantabhadra and Mañjuśirī on the sides of the niche. The cave’s display of
multiple sutra paintings along with screen paintings in the lower registers shows the changes that
took place during the Tibetan occupation in the Middle Tang period, roughly from late eighth to
mid-ninth centuries. Instead of having one large sutra painting on the south or north wall, which
was the preferred way of decorating the walls in the previous period, the cave exhibits a larger
number of sutra paintings, with two each on the south and north walls (fig. 25). While the visual
program became more complex, at the same time it adopted several visual aids to make all the
images coherent in style and recognizable to the viewer.
The south wall of Cave 112 contains a pair of sutra paintings on the south wall: the
Visualization Sutra (Guanwuliangshou jing _≤≥¥=) to the east or the viewer’s left and the
Diamond Sutra (Jingang jing µ∂=) to the west or the viewer’s right (fig. 26). On the north
wall, paintings of the Repaying Kindness Sutra (Baoen jing ∑∏=) and the Bhaisajyaguru Sutra
(Yaoshifo jing πù∫=) are presented (fig. 27).
7
On the east wall, sutras depicting
Avalokiteśvara and Maitreya are painted on the left and right sides of the entrance, respectively,
and a scene of Maitreya’s subjugation of the demons is above the entrance (fig. 28).
8

7. The identifications of the sutra paintings except the one on the east wall are based on Dunhuang
yanjiuyuan, Dunhuang shiku neirong zonglu, 42–43.
8. For the Maitreya sutra mural’s identification and iconography, Fan Jinshi argues that the representation
of the painting on the south side of the east wall should be interpreted as a Maitreya sutra painting

69
Like other caves of the ninth century, Cave 112 exhibits religious themes alongside non-
religious decorative motifs in the cave’s interior. Analyzing the decorative pattern that runs along
the borders of sutra paintings clearly shows the use of painted textile designs as picture frames.
This is in fact one of the artistic strategies the artists used to organize a variety of sutra paintings
into a confined space while maintaining visual coherency within the cave.
In Cave 112, six patterns are found at the edges of mural compositions: pomegranates
with vines (shiliu juancao wen !"#$%, fig. 29), camellia flowers with rolling vines
(chanzhi chahua wen &'()%, fig. 30), diamonds/rhombi (lingge wen *+%, fig. 31),
tortoiseshells (guijia wen ,-%, fig. 32), quatrefoil alternating with paired halved flowers
(lingxing siye wen *./0%, fig. 33), and a six-petal flower pattern alternating with a pair of
halved flowers (liuyehua tuan 10)23, fig. 34).
9
Each design serves as a picture frame to
encase individual sutra painting units. I will focus on how each painting unit is presented in a
cave setting.
10
One of the main features of these designs is that one dominant pattern covers the

because the scenes’ narratives of scenes are relevant to their textual resources, which also can be
corroborated by the image above the entrance, which depicts the subjugation of the demons by Maitreya.
Fan Jinshi -´., and Mei Lin /0, “Mogaoku di 112 ku tuxiang zakao” ?@AB112A1–wå
[miscellaneous studies on the iconography of cave 112, Mogao caves], Dunhuang yanjiu ()./ 4
(1996): 6–7.
9. To better illustrate the patterns, the image of each pattern is presented with the comparable patterns
found in the caves of the Mogao Caves. The second pattern, camellia flowers with rolling vine, is difficult
to identify. Based on the current remains, it is comparable to the pattern from Cave 361, which is figure
29 in this dissertation. One note to make is that here, I mainly focus on the patterns surrounding the sutra
paintings. Other patterns, of course, are used throughout the cave. For example, the western niche’s
ceiling is covered with six-petal lotus flowers. The inside of the angled roof wall has seated Buddha
patterns in a vertical row. Each row is separated by a decorative rhombus-patterned strip.
10. Besides the surroundings of sutra paintings, these types of decorative patterns are also found on a
mandorla and halo of Buddhas and bodhisattvas, the ceiling, and a central niche on the west wall. Youhui
Guan, Dunhuang zhuangshi tu’an ()2314 [Dunhuang decorative patterns] (Shanghai: Huadong
shifan daxue chubanshe, 2010), 140.

70
top part of all four walls to create visual unity while other patterns are deployed for various
thematic paintings. For example, the design of pomegranate with rolling vines, which is typical
of the early Tang period, runs horizontally along the top of all the walls, visually unifying the
space underneath.
11
The strip in which the pomegranate pattern is presented is the widest; the
other four patterns are narrower and run vertically along the height of the sutra paintings (fig.
35).
A clear example of a pattern used as a picture frame can be found on the west wall where
the central niche is located. A painting of Samantabhadra riding on an elephant appears on the
left and one of Mañjuśirī riding a lion is on the right (see fig. 24). Each painting is enclosed
within a lozenge border pattern; a red outline defines the pattern with the smaller diamonds with
four floral petals inside. The artisan also added a half-lozenge pattern outside the main diamond
(see fig. 31). Instead of wrapping the whole wall from bottom to the ceiling, the pattern starts
along the top of the west niche and ends a few inches above the floor. This pattern forms the
main figures’ pictorial plane on the west wall. Although the two paintings can be identified
iconographically as a pair, the same lozenge pattern also contributes to the viewer’s perception
of them as a matched set (see fig. 35).
On the south and north walls, the patterns in strips proceed in a more straightforward
fashion; instead of defining pictorial planes using three separate patterns, the painters employed
them efficiently by extending the strips across multiple compositions. Each wall has two sutra
paintings, and the top strip—with the pomegranate-combined-with-vines pattern—horizontally

11
.
An earlier example of the pomegranates with vines design can be found in the caves from the Early
Tang period. In contrast to the pattern in Cave 120, the earlier design exhibits an elaborately represented
pomegranate as well as simpler stems and fewer leaves. See Dunhuang yanjiuyuan, Dunhuang shiku
quanji 14 (Hong Kong: Shangwu renshuguan, 2003), Pl. 22.

71
runs along the top part of the picture plane. At the bottom of the sutra paintings runs a pattern of
rolling vines with camellia flowers, which is difficult to identify due to the mural’s poor
condition (see fig. 30).
12
Little remains of the part on the north wall, thus making it impossible to
identify its pattern. In addition to two main bands on the top and bottom of the paintings, the
south and the north walls are divided into two pictorial planes by three vertical strips in various
patterns. They are, in order, the quatrefoil alternating with paired flowers in halves, the
tortoiseshell, and the six-petal flower pattern alternating with a pair of halved flowers, starting
from the main niche on the west to the east wall so that the same strips are placed symmetrically
along the opposite walls (see fig. 35). The very first vertical strip maintains a lozenge-shaped
four-petal flower at the center followed by half of the same floral pattern. The tortoiseshell
design, which is located in the middle among the vertical strips, bears a four-petal flower within
each shell. The third pattern, the six-petal flower pattern alternating with paired halved flowers,
wraps up the last sutra paintings’ edges toward the east wall.
Finally, on the east wall, where the entrance to the cave is located, the identical
combinations of patterns were applied. The pattern of pomegranates with rolling vines runs
across the top, and a simplified version of the same pattern—with smaller/thinner leaves and
flowers than the one on the top—continues on both sides of the entrance. Right above the
entrance is a crude representation of flowers with five or six petals alternating with a pair of
flowers in halves. Both ends of the wall share the same patterns used in the south and the north
walls (see fig. 35).
13

12. A similar pattern can be found in Cave 361, another cave from the Middle Tang period. See
Dunhuang yanjiuyuan, Dunhuang shiku quanji 14, Pl. 136.
13. It is interesting to observe that the east wall shares the same pattern with the south and the north walls,
but the west wall does not. On the west wall, as I mentioned above, diamond strips clearly define the

72
The use of patterns to demarcate pictorial planes was a feature that first appeared in the
Middle Tang period, but the patterned strip was not a new practice. In older caves, the niches on
the west walls were often decorated with similar or sometimes flaming patterns. Representative
older caves, such as Cave 322 from the early Tang period or seventh century and Cave 217 from
the High Tang period or eighth century, clearly show that sutra paintings and iconic images were
presented on the west and north walls. It was normal to see a diamond pattern or a lozenge with a
flower around the western niche as evident in Cave 322 (fig. 36). Significantly, the main images
on the south and north walls did not have the aforementioned patterned borders. Instead, black
lines defined the borders, with additional lines inside these borders. The lined border style was
later replaced by the patterned borders, which mimicked contemporary textile patterns, as seen in
Cave 112. On the other hand, the eighth-century Cave 217, shows that the patterned borders
began to be used to demarcate iconic images from narrative scenes in sutra paintings. The
panoramic background of the south wall, on which a box frames a preaching scene, contrasts
with the north wall’s paintings of the Amitābha Sutra and Visualization Sutra.
14
The grandeur of
the preaching scene was placed at the center, and the other narrative scenes from the
visualization sutra were assigned to the left and right sides of the central painting. An alternating
halved lotus flower pattern runs along the borders between the central and side scenes.
Interestingly, this border does not extend all the way to the end but instead stops right above the

pictorial plane that is exclusive to the west wall. On the east wall, the shared bands provide viewers a
smooth transition in their viewing experience from one wall to another.
14. Eugene Wang argues that this sutra painting is based on the Lotus Sutra, but Akiko Shimono suggests
that it represents, in fact, the Usnisa Vijaya Dharani Sutra painting. Scholars have largely accepted the
latter claim, including those from the Dunhuang Academy. See Eugene Wang, Shaping the Lotus Sutra:
Buddhist Visual Culture in Medieval China (Seattle and London: University of Washington Press, 2002),
67–121; Shimono, “Tonkō Bakukōkutsu dai 217 kutsu nanheki kyōhen no shin kaishaku,” 96–115.

73
lower narrative scenes. The main function of this border is, therefore, to delineate and highlight
the central section of the painting.
The strategic use of the patterned borders was more apparent in defining pictorial planes
when more subjects were involved. This means that a wall that formerly had conveyed only one
or two major themes could contain multiple canvases with a series of sutra paintings (and fewer
iconic paintings). The use of decorative patterns to present the various sutra paintings was one of
the critical artistic strategies for increasing the number of mural compositions on the same wall.
Coincidently, the effective use of the decorative borders for a mural painting corresponds
to the way the Buddhist textile paintings from the Library Cave are framed, as Chapter 1 has
explored. Even though many paintings have lost their original textile mountings, a significant
number still retain their original materials, which in turn help show that paintings in medieval
China were presented and framed using a single or multiple types of textiles. Stein Painting 28*
(Avalokiteśvara as Savior of Perils) shows a separate piece of textile—a clamp-resist dyed
damask silk with a blue floral pattern with a hint of yellow—being sewn around the borders of
the central panel to create a frame (fig. 37).
15
The materials used in textile frames varied with the
availability of materials and could be a combination of fabrics. The four borders of Stein
Painting 54 consist of four overlapping, solid-colored silks in blue, yellowish white, brown
black, and deep red (see fig. 12). The extra effort taken to make symmetrical color blocks is

15. Clamp-resist dye (Jiaxie 56) refers to a dyeing technique popularly applied on textiles in the
Dunhuang area and also to a fabric made by this technique. The technique was invented in the first half of
the eighth century and eventually became prevalent. To make a pattern, it was carved on woodblocks
symmetrically. Then the blocks were “clamped” around the folded textile at the desired locations and
dyed. The clamped area would resist the dye, hence forming the desired pattern. Most clamp-resist dyed
textiles from Dunhuang were achieved with multiple colors. Zhao, Textiles from Dunhuang in UK
Collections, 192–95; Zhao and Wang, “Glossary of Textile Terminology,” 379–80.

74
notable. For example, the blue silk on the top extends down the left and right sides, occupying
one quarter of the sides’ lengths if measuring from the corners of the central image. Even though
no patterns are found in this example, painted banners were mostly finalized with a set of textiles
as a mounting. This example clearly shows that a votive painting in Dunhuang area in the Middle
Tang period would not be deemed complete without an actual frame or at least a painted
replication of it. Murals from the same period were likewise defined by decorative patterned
strips. In the painted banners, because a wider choice of materials was available, the frames were
either patterned or dyed textiles or combinations of mono- or multicolored fabrics.
When there was no actual textile for a banner’s mounting, a replication of a patterned
strip in paint was the alternative, marking a parallel situation in the production of mural
paintings. Stein Painting 203 (Paradise of Amitābha) in the British Museum has a lozenge
pattern identical to the one in Cave 112 on all four of its borders, with the top border also having
an additional piece of hemp cloth to which three loops were also attached for the purpose of
hanging (fig. 38).
16
Stein Painting 71 is another example that has a patterned border: alternating
lotus flowers in halves are repeated along all the banner’s remaining borders.
17

Interestingly, these patterns are comparable with those on the extant textile fragments
from Cave 17, among which the diamond pattern is one of the predominant patterns. Fragments
with a concentric lozenge design (дx46) in the State Hermitage Museum in St. Petersburg, for

16. Stein Painting 16 from the British Museum also retains the same lozenge patterns on either side of the
painting.
17. Even though the painting’s condition is poor and the top part does not remain, it is highly possible that
it should have had the same pattern as the other three, if there was any.

75
example, display such a motif in damask silk (fig. 39).
18
The fragmentary silk bears a small
lozenge in a larger diamond pattern on purple silk.
19
In addition, a pattern may be woven directly
into a textile rather than applied with paint. An example from the British Museum, a ribbon with
a diamond pattern (MAS. 909) reveals a motif that is achieved by braiding silk thread in a 1/1
structure (fig. 40).
20
The alternating floral pattern, which was found on the east end of Cave
112’s south wall, is comparable to a pattern dyed on the head border part of a banner from the
Hermitage (Дx49, fig. 41). Another set of similar examples are textile fragments from the Stein
Collections of both the British Museum (MAS. 924) and the Victoria and Albert Museum
(LOAN:STEIN 651), each of which highlights a pattern similar to the rosette pattern applied to
the wall painting.
21

These examples of actual textiles and replications of painted patterns prompt a
reevaluation of the “traditional” mounting method for painted banners and ultimately lead to a
new understanding of the function of such patterned strips or fabrics. In contrast to the mounting
method associated with traditional Chinese painting—pasting a painting onto fabric (usually silk)
of which both ends are wrapped around a wooden rod, making either a hanging scroll or
handscroll—this type of mounting likely originated as a way of displaying paintings in Buddhist
visual culture. Many of the paintings from the Library Cave were found with between two and

18. Zhao, Textiles from Dunhuang in Russian Collections, 123.
19. The original use of the pieces of damask silk is unknown. Fragment B shows that two separate
rectangular pieces of red silk in plain weave are stitched diagonally relative to the purple damask. Based
on how the red silk is sewn, I suggest they could have been part of a valance or canopy.
20. Zhao, Textiles from Dunhuang in UK Collections, 104.
21. The fragments from the British Museum and the Victoria and Albert Museum were originally
assigned the same number by Stein. See, Zhao, 156–57.

76
five loops attached to their top edges. A wooden rod would have passed through these loops to
hang the paintings vertically.
22
Even a small paper prayer sheet has two to three loops attached to
the top, as seen in a woodblock print (1919,0101,0.243) sponsored by Cao Yuanzhong ëªº of
a prayer sheet in the British Museum (fig. 42). The loops indicate that these types of paintings
and prayer sheets were meant to be hung vertically with a rod passing through the top. The most
common method for mounting and hanging a scroll in historical Chinese paintings was attaching
the main pictorial plane of fine silk or paper to a textile, most frequently silk. This textile was
then wrapped around a wooden stick to make a handle for the painting. A thread was attached at
the top of the painting so it could be hung on the wall. In contrast, the loops on the rectangular
painted banners in the Library Cave—Stein Painting 54, for example—suggest that they must
have been hung on independent implements such as a wooden shaft or stick (see fig. 12).
23
It is
also plausible that the wooden rod later became part of the mounting, similar to today’s
mounting practice.
24

22. Unfortunately, no actual device or installation for hanging a prayer sheet or a painting has been found
in the Library Cave.
23. No research has been conducted about how this type of paintings was hung and what types of
implements were used because artifacts have not been found. Future research is required.
24. How a painted banner was mounted prompts questions about the beginning of this practice as well as
its impact on the practice of Chinese painting. Visual similarities between a wall painting and a painted
banner might be key to understanding the hanging scroll’s origin in Chinese painting history. It also
connects to a cave space’s functionality and the intended type of space. As for the origin of the mounting
of Chinese paintings, it has been proposed that the Central Asian practice of hanging paintings was a
possible source. The T-shaped silk banner found in the famous Han tomb at Mawangdui might suggest
another possible way of hanging a scroll. In my opinion, visual evidence related to the former—especially
with Tibetan practice of hanging and mounting pictures—is more direct than the Mawangdui banner
when one compares their shape and mounting method. For a brief discussion of the origin of hanging
scrolls and mounting practices, see Robert Hans van Gulik, Chinese Pictorial Art as Viewed by the
Connoisseur: Notes on the Means and Methods of Traditional Chinese Connoisseurship of Pictorial Art,
Based upon a Study of the Art of Mounting Scrolls in China and Japan (Rome: Istituto italiano per il
Medio ed Estremo Oriente, 1958), 135–225, especially 169–172; Silbergeld, Chinese Painting Style, 13.

77
Furthermore, the practice of replicating all sides of a silk painting with textile design
motifs in paint, which mirrors what we find in the decorative framing of murals in the Mogao
Caves, implies a function beyond protecting the main painting but also of defining a range of
pictorial planes for the figures of veneration. This practice is particularly important in the case of
multiple images being presented in proximity of each other in a cave, where there was little light
inside; a visual demarcation helps viewers identify each object of veneration. The advantage of
this artistic practice would have been particularly important when the color palette was limited.
In the case of Cave 112, the decorative strips’ color palette is not far from that of the main sutra
paintings on the four walls. In other words, green, reddish brown, white, and black are the major
colors of Cave 112, both figural representations and all the patterns mentioned above. The
painters’ limited access to pigments challenged them to create a visually appealing space with
distinguishable areas. Adequate application of the decorative design allowed them to host
multiple images in one space.

2.3. Wall Paintings of the Avalokiteśvara Sutra in the Mogao Caves
In addition to dividing its pictorial plane using textile-patterned strips, the Avalokiteśvara
Sutra paintings provide another opportunity for a discussion of how the pictorial plane was
created in textiles as opposed to on the walls. The foremost reason the Avalokiteśvara Sutra
painting was chosen for this discussion is that the theme was a popular theme in both formats. Its
popularity enables us to observe the ways in which the shared pictorial subjects were visualized
in different media in terms of strategies for accommodating necessary compositional units such

78
as a main central figure for worship and subsidiary narrative scenes supplementing the main
image of worship.
The basic composition of an Avalokiteśvara Sutra painting has two parts: the main iconic
image at the center and the narrative scenes on two sides. The main image is usually a standing
Avalokiteśvara. Surrounding this main image are narrative scenes composed of scenes from the
Avalokiteśvara Sutra, numbering from four to thirty-three in total.
25
Representative scenes of
peril from this sutra include those being pushed from the summit of Mount Sumeru, falling down
the Diamond Mountain, and burning by a serpent’s poisonous breath.
26
Each section is
accompanied by a cartouche that bears excerpts from the sutra. In addition, topographic elements
such as mountains, rivers, and ponds were used not only to divide each scene but also to
encompass the whole composition, making it a unified composition.
On the east wall of Cave 112, the Avalokiteśvara Sutra painting is depicted on the north
side of the wall. The Sutra of Maitreya’s Attaining Buddhahood (Fu shuo mile da chengfo jing
∫Ωæø9¿∫=) and the Sutra on the Descent of Maitreya (Mile xiasheng jing æø¡Ö=)
are on the south side of the entrance, with the scene of Maitreya’s subjugation of the demons

25. More than 200 copies of the Avalokiteśvara Sutra were found in the Library Cave, some with
illustrations including S.6542 and S.6983 in London and P.4513 and P.2010 in Paris. Akira Fujieda,
“Sutain Tonkō shūshū e-iri Kannongyō sasshi (S.6983): Tonkō ni okeru mokuhitsu no shiyō,” Bokubi 177
(1968): 8; Jacques Giès, ed. The Arts of Central Asia: the Pelliot Collection in the Musée Guimet
(London: Serindia Publications, 1996), 80.

26. The Avalokiteśvara Sutra was originally the 25th chapter of the Lotus Sutra. The scenes of peril are
described in prose and verse. The details of the perils are almost identical except in a few instances. For
the complete list of the perils in the verse section, see Miyeko Murase, “Kuan-Yin as Savior of Men:
Illustration of the Twenty-Fifth Chapter of the Lotus Sūtra in Chinese Painting,” Artibus Asiae 33, no. 1/2
(1971): 60–61.

79
above the entrance and between the walls (see fig. 28).
27
Much of the east wall is heavily
damaged, as the bottom half of the north part of the wall has crumbled. The condition is worse
on the opposite side, where the bottom section and the left side of the Maitreya have mostly
deteriorated. However, the remaining part at least shows that the east wall was sectioned into
three domains by painted decorative strips with a geometric pattern and has a layout identical to
that of the two sutra paintings, placing the main bodhisattva figure at the center in a painted
banner format and narrative scenes at either side of the central banner format. The east wall’s
layout was carefully designed in order to accommodate the two main bodhisattvas,
Avalokiteśvara and Maitreya, and their accompanying stories, which are similar to the Buddha
preaching scenes on the neighboring south and north walls. The overarching design scheme in
this cave, apparently, was meant to demarcate a main iconic figure from the surrounding
narrative scenes with patterned decorative bands as borders.
The Avalokiteśvara Sutra painting on the north side of the east wall exemplifies the
layout of the peril scenes from the Avalokiteśvara Sutra and the main iconic figure (fig. 43).
Signified by the blank (or later erased) cartouches, there were twelve perils altogether, six on the
right side and the other six on the left (fig. 44). Even though not every scene is identifiable, at
least eight peril scenes are, including falling from Sumeru Mountain and Diamond Mountain,

27. The identification of the south part of the east wall was initially known as Mahāsthāmaprāpta, as
described in Dunhuang shiku neirong zonglu. However, the identification was rebuked by Fan Jinshi and
Mei Lin, who closely analyzed four narrative scenes painted on the right side of the bodhisattva, based on
the Dashengfo jing and Mile xiasheng jing. In my opinion, the combination of Avalokiteśvara on the
north part and Maitreya on the south is a very interesting pair and also calls for further research in terms
of their overarching subject, salvation of the suffering. See Dunhuang yanjiuyuan, Dunhuang shiku
neirong zonglu, 43; Fan and Mei, “Mogaoku di 112 ku tuxiang zakao,” 6–7.

80
being attacked by evil beast and bandits, falling into pit of fire followed by the lake of water
being turned from the fire, being attacked by snakes and thunder, and being punished by fetters.
One particular scene that attracts the viewer’s attention is the falling into a pit of fire on
the right side of the painting. A man in a white robe is being pushed into the fire pit as being
depicted in the middle of the flame. Below the scene is the representation of a pond in which the
man stands peacefully, thus indicating him being saved by Avalokiteśvara. In the Avalokiteśvara
Sutra, the corresponding verse reads as follows:
Suppose someone with harmful intent, Casts you into a great pit of fire;
Keep in mind Avalokiteśvara powers, And the pit of fire will change into a pond.
28

This is the first verse preaching the benefits of calling out Avalokiteśvara’s name. The miracle is
visualized in detail with two separate scenes, which indicate the cause and effect of “being
pushed by someone.” This is one of the major differences from the representation of the painted
banner counterpart, where the two scenes of showing a figure being pushed into a pit of fire and
then saved by Avalokiteśvara, who appears in front of the person in peril, are merged into one. In
other words, depicting the person being pushed into the water surrounded by fire is in fact the
shortcut to meeting Avalokiteśvara.
How did the design of the Avalokiteśvara Sutra painting evolve? Paintings of the
Avalokiteśvara Sutra can be found in thirteen units at Mogao Caves (Caves 8, 14, 45, 55, 112,

28. T. 262 57:17–18c, “78%9:, ;<=>?f@ABCDf>?HoE.” Translation is from Fo
Guang Shan International Translation Center, The Lotus Sutra’s Universal Gate Chapter on
Avalokiteśvara Bodhisattva (Hacienda Heights, Fo Guang Shan International Translation Center, 2016),
43.

81
126, 128, 185, 205, 217, 342, 395, and 472).
29
The earliest examples of the Avalokiteśvara Sutra
paintings found in Caves 45 and 205, constructed during the High Tang period, yield two
different compositional types. One used an entire wall by featuring Avalokiteśvara at the center
and then placing scenes of perils in a large topographical setting without having strict boundaries
between the scenes and the central Avalokiteśvara figure. The other one is to have the
Avalokiteśvara motif painted in a vertically oriented rectangular format.
30

The Avalokiteśvara Sutra painting of the eighth century CE found in Cave 45 exemplifies
the first type and contains the most scenes of perils (fig. 45).
31
A monumental image of
Avalokiteśvara stands at the center, but unlike in the later example of Cave 112, there is no clear
border on either side of the figure. Instead of patterned borders, Avalokiteśvara here is separated
from other narrative scenes only by two loose-leafed trees (one on either side) that are in
harmony with the background. The central figure gives the impression that he is standing in the
middle of a landscape, while all of the supplementary scenes are blended into the composition
through landscape motifs such as panoramic mountains in green. This artistic choice strongly
implies the omnipresence of Avalokiteśvara no matter what hazardous situation occurs. This

29. This theme is found in an additional two caves from Cave 2 of the Yulin Caves and Cave 14 of the
Western Thousand Caves FG3|. For a complete list of the sutra paintings, see Dunhuang yanjiuyuan,
Dunhuang shiku neirong zonglu, 284. I added Cave 217 to the list because it had been excluded.
30. Screen paintings inside the western niche of Cave 359 are illustrations of Avalokiteśvara saving lives
from sufferings. Zhang Yuanlin ºH0, and Xia Shengping IJÒ, “‘Guanyin jiunan’ de xingxiang
tushi: Mogaoku di 359 ku xibi kan nei pingfeng hua neirong shidu” ‘BCKL’èMN1O: MNÍO:
?@AB359 AFGPQRSÄQµLT [Illustration to Guanyin’s story of saving life from
sufferings: An interpretation of the screen paintings inside the niche of Cave 359 Mogao Grottoes],
Dunhuang yanjiu ()./ 5 (2010): 36–46.
31. Murase, “Kuan-Yin as Savior of Men,” 64–65.

82
early example of sutra painting was painted on the most prominent location of a cave, namely the
north or the south wall.
At the same time, there is a second type of Avalokiteśvara Sutra painting as seen on the
south wall of Cave 205 from the eighth century (fig. 46). This type became more prevalent in the
later periods. Aside from the earliest examples (Caves 45 and 205, which are from the High Tang
period), eleven other caves are from the Middle or Late Tang, Five Dynasties, Song, and
Western Xia periods. In contrast to the Buddha preaching scenes, which required wide,
horizontal composition to accommodate the large groups of Buddhist deities, bodhisattvas, and
Buddhas involved, the Avalokiteśvara Sutra was usually painted in a vertically oriented
rectangular format. The vertical orientation often allowed the scene to be painted on two sides of
the east wall where the entrance to the cave was located. Six caves (112, 126, 128, 217, 342, and
472) have a painting of the same theme on their east walls. Because of the entrance, the east
wall’s pictorial plane is more limited than those of the south and north walls. The east wall is,
however, suitable for paired scenes (e.g., Vimalakīrti and Mañjuśirī) or for a large, single-themed
painting such as one showing the Avalokiteśvara Sutra. The second most popular location is the
north or south wall. Even though these walls are wider and with more space than the east wall,
the Avalokiteśvara Sutra was still depicted as part of a compartmentalized composition that
features multiple sutra paintings. In short, this theme is not one that occupies the whole wall but
rather forms part of a larger group of paintings. For example, in Cave 205, a painting of the
Avalokiteśvara Sutra is situated in the central section of the south wall along with five other

83
paintings, including two preaching Buddha scenes, a standing Buddha, a preaching Amitābha,
and a Bhaisajyaguru flanked by Avalokiteśvara and Kṣitigarbha.
32

Depending on the size of the designated wall and the cave’s overall layout, painters might
choose from two types of Avalokiteśvara Sutra representations for wall paintings: a narrow,
sectionalized format and a wider version. From the Middle Tang period onward, however, the
narrow and sectionalized format became more popular. This should be understood within the
context of changes in the pictorial contents at the Mogao Caves since the ninth century CE. The
compact version of the sutra painting became a preferable option to accommodate the increased
number of slots for a sutra painting in a cave by selecting representative scenes of peril from the
sutra and placing them on either side of the main iconic image of the standing Avalokiteśvara
(see fig. 43).
The narrow version of the sutra painting is fairly comparable to a banner format in terms
of composition, size, and range of pigments, with the Avalokiteśvara Sutra paintings in Cave 112
(east wall) and Cave 205 (south wall) providing compelling evidence for this similarity. Unlike
the murals from the earlier times (such as Cave 45 and Cave 217), those from these two caves
show Avalokiteśvara in a demarcated plane at the center. In Cave 112, the standing
Avalokiteśvara is situated independently in the middle and does not interact with the scenes of
peril to the sides (see fig. 43). The pictorial plane where Avalokiteśvara stands is bordered with
the same decorative strip motif—a diamond pattern—as in the Samantabhadra and Mañjuśirī
wall paintings on the west wall next to the central niche.

32. For iconographies of the rest of the entire pictorial program in Cave 205, see Dunhuang yanjiuyuan,
Dunhuang shiku neirong zonglu, 81–82.

84
A mural with resemblance to a hanging painting can also be found on the south wall of
Cave 205 (see fig. 46), which was created before Cave 112. The iconic image of Avalokiteśvara
is identical to that of a single bodhisattva in a painted banner. The pictorial plane to which
Avalokiteśvara belongs is visually distinct due to the application of contrasting colors in the
central part. This scheme recalls the design of the narrow painted banners; the use of tiles at the
bottom of the painting is also a key pictorial element in the painted banner with the single
bodhisattva.
The depictions of the Avalokiteśvara Sutra in Caves 112 and 205 contrast with that of
Cave 45 in terms of how the images of the sutra were meant to be viewed. The painting style
changed from showing Avalokiteśvara in the middle of a panoramic or topographic view to a
smaller, banner-like wall painting. The latter is more convenient to mesh with other sutra
paintings; this type is also suggestive of depicting the actual portable version of a painting that
was used during religious rituals.

2.4. Avalokiteśvara Sutra Painting in a Painted Banner Format
When depicted in a portable painting format, compositions with the Avalokiteśvara Sutra
were not executed in the same format as in the mural versions. These compositions vary
depending on Avalokiteśvara’s posture, for instance. In the portable format, he is surrounded by
fewer scenes from the sutra than in the wall version, with only a five or six select scenes
included. When standing, he extends out of the frame and is situated in the center. When seated,
his circular mandorla and throne take up most of the central space along with two attendant
figures.

85
The extant painted banners of the Avalokiteśvara Sutra include at least seven pieces that
are kept in various museums worldwide. The British Museum has the largest number, including
Stein Paintings 24, 28, and 2. The Pelliot Collection in the Musée Guimet has two banners: EO.
1142 and MG.17665. A painting of the eleven-headed Avalokiteśvara is held by the Harvard Art
Museums, whereas the Museum of Fine Arts, Boston, has a painting with Avalokiteśvara as a
savior from perils. The least-known painting, which shows Avalokiteśvara with water and the
moon, is in the Sichuan Museum in Chengdu.
33
Among these, Stein Painting 24 (fig. 47) from
the British Museum is a representative example of a standing Avalokiteśvara. Even though the
overall quality of the painting is not high, the application of various pigments, delicate outlines
of the figures, and the dynamic pictorial depiction of hazardous incidents make the painting
noteworthy.
MG 17665 from the Musée Guimet (fig. 4) is dramatically different from the composition
of a wall painting of the Avalokiteśvara Sutra. The central figure of Avalokiteśvara does not
have a mandorla behind him, thus yielding more space for other pictorial elements.
Avalokiteśvara is also surrounded by four stories from the Avalokiteśvara Sutra and the twenty-
fifth chapter of the Lotus Sutra. Each scene is accompanied by an inscription of gathas (verses)
from the sutra.
34
At the upper right, a man wearing a red robe and official’s headgear is falling
from Mount Sumeru but is safely rescued by Avalokiteśvara, who enables the man to float on a

33. A picture of the painting can be found in Yiyuan duoying UVWX 9 (1980), 4. It bears a dedicatory
inscription that tell us the donor’s name, Fan Zaisheng. His name also appears in a rental contract (S.6452
Vo). Giès, The Arts of Central Asia, 100.
34. The verse sections are a summary of the previous prose section. According to Murase, the verse’s
succinct form would have appealed to the public and convinced them of instant salvation by
Avalokiteśvara. Most inscriptions to the Avalokiteśvara Sutra painting both on murals and on painted
banners are quoted from the verse. See Murase, “Kuan-Yin as Savior of Men,” 60.

86
cloud-shaped sun.
35
In the next scene, below and to the right, a man is threatened by poisonous
animals that spew intense fire from their mouths. The inscription on the far right reads, “Should
you be burnt by the flames of the poisonous breath of vipers, of venomous serpents and of
scorpions, invoke the power of Avalokiteśvara and [these beasts] will instantly depart of their
own will.”
36
Unlike the upper scene, this image does not show what happens after one invokes
the power of Avalokiteśvara. The fact that the man in this scene remains calm, with his hands
folded together, indicates that Avalokiteśvara is already protecting him. On the left side of the
painting, the same figure is falling from the Diamond Mountain. This scene is described in the
Avalokiteśvara Sutra as follows: “If, being pursued by wicked people, you should fall to the
bottom of the Diamond Mountain, invoke the power of Avalokiteśvara, and no harm will come
to you, not even to a single hair of your head.”
37
In the final scene in the lower left, a man in a
red robe pushes another man in a dark navy robe into a fire. The fire turns into water, and
Avalokiteśvara stands upon a lotus throne. The fire’s transformation into water is described in
the inscription: “If [people] intent on evil push him into a great pit of fire, let him think of the

35. The cloud motif should be identified with the sun based on the inscription and its color—red. The
inscription states that “If you are pushed from the summit of Mount Sumeru, invoke the power of
Avalokiteśvara and you will be supported in space like the sun.” (Italics are mine.) T262 57:21c-22c “Y
Z[\], ^¸_;`f@ABCD, a˘bcd.” For translations, see Giès, The Arts of Central
Asia, 80.
36. T. 262 58:08a-09a “ef ghfijk>lf@ABCDfmn∫op” Translation is from
Giès, 80.
37. T. 262 57:23c-24c “Yqr¸sf`?f@ABCDf>?HoE.” Fujieda, “E-iri
Kannon-kya Sasshi,” 8. Translation is from Giès, 80.
39. The Harvard Art Museums labeled this painting as “Eleven-Headed Guanyin,” as opposed to Ma De’s
identification, “Twelve-Headed Guanyin.” I followed Ma’s naming because the main face of
Avalokiteśvara should be included in counting of the number of heads.

88
sequence—as the viewer was expected to read the scenes from top to bottom and then left to
right—but it is in fact the first peril described in the Avalokiteśvara Sutra. Placing the first
described peril as the last destination of the viewing sequence is one of the strongest visual
devices that remind the viewer of Avalokiteśvara’s power and his uninterrupted and direct
rescue.

2.5. Stylistic Variations Based on Material Characteristics
An analysis comparing the Avalokiteśvara Sutra murals and portable paintings from Cave
112 (see fig. 43) with other painted banners (figs. 47–49) in terms of style and color palettes
reveals the work of a group of painters who shared a similar painting style. These comparisons
show the extent to which the base material for a painting would have affected the style or
iconography of the Avalokiteśvara Sutra paintings.
The standing Avalokiteśvara of the Avalokiteśvara Sutra painting from Cave 112
showcases the “Indian” style of representations of figures, which was first defined by Stein in
order to differentiate the Chinese (Han Chinese) style of painting from the other.
40
He defines
that the bodhisattva in “Indian” style as follows:
The slender-waisted body of the standing figure is usually shown curving at the hips; its
upper part is nude except for rich ornaments and a winding narrow stole, while the
hanging drapery of the skirt has folds arranged generally in close conformity with the
style of Gandhara sculpture. The painting of the flesh in different colours according to

40. Stein, Serindia 2, 862–3 and no. 11. It should be noted that the term “Han (Chinese) style” was
actually used when listing up a temple’s assets in Dunhuang. For example, Bao’en Temple’s document
(S.4215) indicates a painting by labeling it as the “Han-style Kṣitigarbha painting,” whereas other
paintings were simply called without the name of a style.

89
the deity represented is a characteristic feature which this class derives from its Indian
models.
41

As Stein points out in reference to “the painting of the flesh,” the major difference
between the two styles is in the way dressing garments are depicted on the bodhisattvas’ or
Buddhas’ bodies and the representation of facial features. If an upper torso is covered by a
Chinese-style garment, then the painting is categorized as Chinese style. In the case of the
Avalokiteśvara on the east wall, the torso is naked and instead adorned with necklaces and
scarves (see fig. 43). Regarding his facial features, the arches of his eyebrows are raised toward
the forehead. Crease lines between the eyes and eyebrows are not completely straight but drawn
halfway up the brow line. The eyes are half closed, gazing downward. He has a wavy mustache
and a small beard on his chin. Chinese-style bodhisattvas are usually represented with directly
gazing eyes and less often with raised eyebrows. The tip of the nose is outlined without the ridge.
These facial features are found in several other painted banners, which will be discussed below.
This painting style, however, does not directly reveal the painters’ ethnicity. In other
words, if a painting is drawn in the “Indian” style, it does not necessarily mean that the painters
were Indian. In fact, it is common to observe that many paintings present a style that is a mix of
“Chinese style” and “Indian style.” In the case of the Avalokiteśvara Sutra painting in Cave 112,
even though the main bodhisattva is depicted in “Indian” style, the figures in the subsidiary
scenes at either side of the main image wear Chinese garments. Moreover, even if a painting
such as Stein Painting 141 is labeled as “wholly Indian” style, as Arthur Waley argues, another
painting (EO. 1399) regarded as a “Chinese-style painting that displays a very similar
representation of a lion can counter that claim. Whitfield strongly suggests that the two drawing

41. Stein, 862.

90
modes “flourished side by side at Dunhuang and may even have been executed by the same
painters, following different models.”
42
Therefore, it does not make sense to infer the ethnic
identity of a painters’ group or workshop from the style of the figures. In the context of visual
culture in Dunhuang, style can be useful but should not be regarded as a reliable criterion for
cultural identity.
Among other painted banners, several show facial features of Avalokiteśvara that are
similar to the Avalokiteśvara from Cave 112 and therefore can be grouped together. These
examples include: Stein Painting 22 (fig. 50), Stein Painting 5, Stein Painting 108*, Stein
Painting 109*, EO 1189, and Stein 391 (fig. 51).
43
Stylistic features such as the naked torso,
raised arches of eyebrows, wavy mustache, and three-peak headdress, are typically found in the
paintings from the Middle Tang period, which corresponds to the Tibetan occupation period.
Even though they do not exhibit conventional Chinese attributes, and can hardly be categorized
as exclusively non-Chinese-style, they are certainly different from a group of paintings including
Stein Painting 103 of the Bodhisattva Vajrapāni (fig. 52). Representing the Khotanese tradition,
this painting bears completely different stylistic attributes, including the bodhisattva’s greenish
body, striped dhoti, and a direct gaze toward the viewer.
44

42. Arthur Waley, A Catalogue of Paintings Recovered from Tun-huang by Sir Aurel Stein, K.C.I.E
(London: British Museum and the Government of India, 1931), 153; Whitfield, The Art of Central Asia 1,
Pl. 54.
43. Except Stein 391, which is now in the National Museum in New Delhi, India, all paintings are in the
British Museum. For the New Delhi collection, the paintings are labeled as “Stein #” without adding
“painting” after “Stein,” which is different from the naming system in the British Museum.
44. Whitfield, The Art of Central Asia 1, Pl. 48.

91
Among the paintings in this group, Stein Paintings 22 and 391 can be directly compared
to the Avalokiteśvara painted on the east wall of Cave 112. From this one may conclude that they
might have been executed by the same painters (fig. 53). Judging from their styles, color palettes,
and painting techniques, the three examples share similarities that set them apart from other
groups of paintings. In terms of iconography, each of the three bodhisattvas wears a mountain-
shaped crowns with Amitābha Buddha, which helps identify them as Avalokiteśvara, together
with a kundika (sprinkling bottle) in each of their left hands.
45
Their elaborate, elevated crowns
are regarded as Himalayan or Southeast Asian style. Their right hands make a mudra with a
thumb and an index finger forming a circle. An inscription on Stein 391 confirms that he is a
“Great Merciful Great Passionate Savior from Pain Avalokiteśvara Bodhisattva” (daci dabei
jiuku guanyin pusa 9û9¬¨≠_ajk), which is the bodhisattva that is identical to the one
from the east wall of Cave 112.
46
The way of representing a halo is also noticeably similar; green
and red concentric circles echo the paintings’ major colors of dark red or orange (red ocher from
iron oxide and lead), green (malachite), and white (white lead). In addition, each figure has lotus
flowers behind their ears and above the shoulders. Another peculiar feature is that a piece of long
scarf or textile is affixed to the backs of both arms, forming a heart-shaped folds at the top of the
shoulders.
47

45. Because the bottom half of the bodhisattva in Stein 391 is missing, it cannot be confirmed whether he
holds the bottle. Nevertheless, it is highly possible that he might have been held it due to striking
similarities with the other bodhisattvas from the mural and Stein Painting 22.
46. Lokesh Chandra, and Nirmala Sharma, Buddhist Paintings of Tun-Huang in the National Museum,
New Delhi (New Delhi: Niyogi Books, 2012), 122.
47. Although it is difficult to say that the folds are an exclusive feature to this group of paintings, they are
not found in those drawn in the “Chinese” mode of bodhisattvas whose upper bodies are wrapped around
with several scarves instead of a garment.

92
In terms of painting technique, one prominent aspect shared by the wall paintings and the
painted banners is that the skin of the main iconic figure is painted with lead white in order to
make the figure distinctive from the background.
48
Even though this painting practice is not
exclusive to this particular group of paintings, the seeming consistency in the shared artistic
intention points to the possibility of their production by an identical group of painters. In
comparison, other painters often left the flesh part unpainted so that the background color
became the skin color, with added darker shades for contouring. Contouring becomes more
prominent if the background color is brighter.
The discussion above has shown that painting workshop had shared style, painting
practice, and iconography between hanging paintings and wall paintings regardless of the base
material. Yet at the same time, painters also took into consideration the qualities of the base
material when developing a design unique to the work. Based on visual analysis, I have observed
that in a textile format the bodhisattva figures show a more accentuated sense of movement than
the ones on walls. For example, the Avalokiteśvara from Cave 112 is standing upright—his head
is still and not tilted. He brings his right hand to his chest, which does not break a viewer’s gaze
but keeps it within his body. In contrast, the Avalokiteśvaras from Stein 391 and Stein Painting

48. White lead is widely found among painted banners from Cave 17. It was a popular pigment for pure
white or was mixed with red pigment for pink color as skin color. White lead, or basic lead white
carbonate, 2PbCO₃·Pb(OH)₂, is the most common pigment used for pure white or mixing with other
pigments to control color tone; Brack and Mysak, A Technical Study of Portable Tenth-Century
Paintings, 20. See the following conservational reports for white lead pigment used in painted banners
from Dunhuang: Suzy R. Delbourgo, “Two Far Eastern Artefacts Examined by Scientific Methods,”
International Symposium on the Conservation and Restoration of Cultural Property. Conservation of Far
Eastern Art Objects. (1980), 178; John Winter, “‘Lead White’ In Japanese Paintings,” Studies in
Conservation 26, no. 3 (1981): 89–90; East Asian Painting, 36–37; Brack and Mysak, A Technical Study
of Portable Tenth-Century Paintings, 18, Table 1 and 20; Garcia, “‘La quête des mérites’,” 61.

93
22 are depicted as moving; their bodies are S-shaped and their heads slightly tilted head as each
figure’s central axis from head to feet is off center.
In the case of a single bodhisattva in the banner format, the emphasis on movement is
frequently represented by using multiple pictorial devices such as the canopy above the
bodhisattva. If a bodhisattva is represented as moving forward or looking back, the canopy above
them and its dangling ornaments also move in that direction.
49

One might argue that the two representations of standing Avalokiteśvara (Stein Painting
22 and Stein 391) might be the outcome of using different sketches or models. Even though this
statement is plausible, it is also critical to remember that flexibility, movement, and mobility are
closely related to a painted banner format. In visual representations of banners on the walls of the
Mogao Caves, banners are shown in motion. For example, the Bhaisajyaguru sutra painting on
the east wall of Cave 148, shows two ways of using a banner in a Buddhist ritual (fig. 54). One is
to indicate a sacred space such as a Buddhist temple. In this usage, a long banner-pole (fangan
V√) is placed in the middle of a temple’s courtyard where a table with food offerings is set. On
top of the pole, a long banner waves in the air. The second use can be seen inside of the shrine,
where a group of monks are worshipping in front of an altar with paintings of a Buddha flanked
by two bodhisattvas. Each side of the set of paintings has one decorative banner, which seems to
be in motion. The presence of the banners certainly promotes the sacred atmosphere and possibly

49. In addition, several banners with a single bodhisattva evoke multiple senses with round bells added to
the canopy so that a viewer can elicit not only a sense of visual movement but also of hearing The
representation of bells attached to a canopy is not exclusive to the painted banner format. For instance, the
combination of a canopy with bells is seen on the ceiling of Cave 196, built in the Late Tang period.
However, it is more common in the painted banner. Dunhuang yanjiuyuan, Dunhuang shiku quanji 14, Pl.
156.

94
symbolizes the sacred beings’ “arrival” at the ritual site by visualizing the banners flipped
outward. Chapter 4 will analyze this scene in greater detail.
These representations of banners in use suggest that movability was one of the main
qualities that donors and painters expected in these objects. Bodhisattvas who were captured in
motion, such as in Stein Paintings 22 and 391, created more dynamic imagery than the static type
seen in Cave 112. That is to say, depending on the type of support material for painting, a cave
wall versus a textile banner, painters would create a depiction of the bodhisattva that was most
suitable to the material.

2.6. Conclusion
Paintings in the two media—mural and textile—serve different visual effects and
functionality while sharing decorative motifs. Wall paintings in the Mogao Caves utilized
painted depictions of contemporaneous textile patterns, which hint at the original mounting
method for banners: using colored/patterned textile pieces on all four borders. One of the widely
read sutras, the Avalokiteśvara Sutra, or the twenty-fifth chapter of the Lotus Sutra, is
represented in two modes on wall paintings: freely arranged rescue scenes within panoramic
views with a standing Avalokiteśvara and vertically organized perils with a standing
Avalokiteśvara who is separated from the dangers and represented more as an iconic image.
While the composition type featuring freely arranged rescue scenes has not yet been discovered
among painted banners, there are specimens with vertically organized peril scenes, a genre that
emphasizes a direct relationship between Avalokiteśvara and the endangered. Removing
marginal borders from Avalokiteśvara’s iconic representation makes him a major protagonist in

95
the stories. Compositions emphasizing a more direct relationship between Avalokiteśvara and the
endangered would also have been experienced more intimately by the viewer because they were
on banners that are light and thus portable than an immovable cave wall.
In addition to the contrasting stylistic features between identical thematic subjects,
materialistic difference was another pivotal factor for the painters to determine a more effective
style that incorporates the material quality. Emphasis on movement, vitality, and the body
through the application of light colors can be understood as a result of manifesting material
qualities, which would enhance the viewer’s experience. I have demonstrated that the banner
format was designed to create a more intimate relationship with the viewer than a wall painting,
which is part of a bounty of images in a closed space. When the Avalokiteśvara as a Savior of
Perils is depicted solo, his relaxed posture rather than stiff posture creates an enlivened tone
together with a thin silk, which maximizes the movement if the banner were hung from the top of
a pole.

96

Chapter 3: Long Banner’s Meaning and Function
3.1. Introduction
One unusual type of painted banner found in Cave 17 is the so-called “long banner.” As
the name indicates, these banners are extremely long in length—up to 915 cm—in contrast with
most of the other banners, which range generally from 30 cm to 120 cm in length. In addition,
the way in which each long banner was constructed, using unusual materials such as blood and
silver, differentiates them from the rest of the painted banners in the group. One example is Stein
Painting 216, which at 396 cm in length is four to ten times longer than the conventional painted
banners found in Cave 17 (fig. 55). In addition, the configuration of this banner painting is
unconventional: Across the top of the banner, a small piece of silk with a written inscription is
sewn horizontally, and the rest of the long silk is cut in half vertically starting from the right and
below this inscription. On the left and right sections of the cut silk, five and four standing
bodhisattvas are in columns, respectively. The name of each bodhisattva is written in a cartouche
close to the corresponding figure, although not every identification is legible. This type of banner
was popular during the latter half of the ninth century and into the tenth century, even though
they were first produced in the eighth century.
1

Including Stein Painting 216, there are nearly thirty paintings sharing these stylistic and
formalistic features (see Appendix for the complete list of the long banners).
2
Not every example

1. Yim, “Chungguk kodae pulgyobŏnŭi yangshik,” 83.
2. They are now in museums worldwide, including the Stein Collection at the British Museum in
England, the Pelliot Collection at the Musée Guimet in France, the Otani Collection at the National
Museum in Korea, and the State Hermitage Museum in Russia.

97
in this group was painted on split silk; some are painted on a single piece of textile that was not
cut. Regardless of whether its supporting textile is split, the painted banners in this group share
several key features. First, their subject matter is multiple standing bodhisattvas arranged in a
column, with some exceptions having Buddha figures.
3
Second, there is a minimal application of
color; the Buddha or bodhisattva figures are drawn with lines that are not filled in with color.
Where color is used, the shading scheme is simple and features red, yellow, and green. Third, the
spontaneous brushstrokes are simple and inconsistent in terms of thickness. All the outlines were
done in rather a hasty manner; each figure is delineated with bold lines and the painting style is
not refined. Finally, the paint has been applied to dyed silk, with red, yellow, and dark blue the
three main dye colors used in this type of long banner. In most cases, a dark paint such as black
or cinnabar red was applied, especially on a light-colored dyed silk such as yellow, but there are
examples in which silver, gold, or similar pigments were used to create a contrast of light on a
dark silk background.
It is interesting to observe that there is a discrepancy between the painting style and the
materials that were applied to these long banners. Contrary to the common belief that when
luxurious, precious painting materials are used the level of execution is high, this group of long
banners—made with valuable pigments and textiles—were executed in a rather hasty manner.
The paradox of this crude, unsophisticated painting style being executed with valuable applied
materials implies that the banners’ materiality outweighed their style; more important than the
completeness and precision of its painted subjects was a banner’s full use of precious pigments,
which were limited in quantity, as a form of offering. This raises the questions of who the donors

3. These exceptions are Stein Painting 195 at the British Museum, and MG 17675, MG 17791, and EO
1165 at the Musée Guimet.

98
were, what they prayed for using these banners, and what symbolic meanings and functions can
we interpret from the odd pictorial layout of these long painted banners? More importantly, is the
crude painting style associated with the mode of display at outside setting?
These long banners have five different combinations of silk colors and pigment colors.
4

In each case, one of three primary colors was used for the silk support: yellow, red, or blue. The
rule of a thumb seems to have been that if the support was a dark color, the painting pigment was
either white (white lead), silver, or yellow (orpiment) to distinguish the painted subject from its
dark background. In cases in which a more brightly colored support, such as yellow, was used,
dark ink is the primary pigment. All in all, the five combinations seen are white on dark blue,
yellow on blue, silver on red, red on yellow, and black ink on yellow.
The application of white and yellow pigments is particularly interesting because in these
cases the painter was likely trying to imitate more precious metallic pigments such as gold and
silver. At least five long banners are known to have been drawn with actual silver pigment,
including Stein Painting 217, Stein Painting 214 (1), EO 1137, EO 1418 (?), and Д x 59, 115, 39.
These are all paintings in which a single bodhisattva was depicted. In cases where there appears
to be gold, yellow pigment—usually orpiment, as in the case of Stein Painting 214 (2)—was
actually used. It is extremely rare to find an example with actual gold pigment applied; an
exception is a small painting on paper (EO 1228) from the Pelliot Collection.
5
However, given
its size (26.5 x 12.5 cm), it is unlikely to be part of a long banner. If this was a frontispiece of a
sutra, as Whitfield argues, it is safe to say that any example of the actual application of gold
pigment on banners has not been discovered yet.

4. Whitfield, The Art of Central China, Plate 33.
5. Whitfield believes that this example was intended to be a sutra’s frontispiece because of its small size
and carefully executed drawing. Whitfield, Plate 35.

99
The use of bright colors against a dark, contrasting background such as dark blue or
vermilion is only observable in the long banner format and signifies a unique devotional practice
in artistic form. This suggests that the materiality of the pigments applied on long-format banners
was not trivial at all, but significant to enhance the meaning of the banner, and idea that I will
examine in detail later in this chapter.
As for the original meaning of the painted long banner form, Whitfield argues that the
long length of the painting coincides with the “forty-nine-chi ƒ” banner, referencing the
Consecration Sutra (Guanding jing ≈∆=, T. 1331), a section that explains and sermonizes
about the benefits of having the banner.
6
Because the sutra is the earliest Bhaisajyaguru Sutra,
Whitfield argues that its function should be understood in relation to the Bhaisajyaguru Buddha
cult. He briefly points out, quoting Alexander Soper’s work, that the section of the Consecration
Sutra mentions the forty-nine-chi banner.
7
Even though this is an important finding with regard
to the original function of the group of long banners, in-depth research on the relationship
between the text and the actual object has not yet been done. For this reason, questions including
the context(s) in which a long banner was utilized and what values or donor’s wishes were
promoted by such a peculiar format remain unanswered.
Based on preliminary research, this chapter challenges some fundamental questions
regarding the format, composition, and materiality of the long banner. Why were they made so
extremely long? What was the intended or expected function of this unique format? What types
of visual and textual resources led to the creation of these objects? How was this type of object

6. Whitfield, Plate 33.
7. Whitfield, Plate 33; Alexander C. Soper, “Literary Evidence for Early Buddhist Art in China,” Artibus
Asiae. Supplementum 19 (1959): 171–178.

100
intended to be displayed? Are there any meaningful correlations between the painted subjects—
mostly depictions of Avalokiteśvara labeled in various names—and the objects’ long format?
One of challenges of studying the Dunhuang long banners is that very few among the
thirty examples of them have votive inscriptions. In most cases, they tend to bear cartouches in
which the name of each bodhisattva is written, except for a few cases with blank cartouches. The
aforementioned long banner, Stein Painting 216, is an exceptional example that contains
significant information, as well as possible implications and functions imposed on the painting.
This chapter will reconstruct Stein Painting 216 using other possible original pieces and will
propose its original designated function and usage while stressing the embedded materiality that
contributes to this function.
Beginning with a review of textual resources, I suggest diverse ways to understand such a
unique object type. Its extreme length and configuration should be examined first, with possible
cultural encounters and borrowings between Daoism and Buddhism considered. The salvation
ritual and its implements in relation to the material characteristics of the banner as an offering
will help us understand the ritualistic use of the long banner. Attempting to reconstruct the
scattered pieces of long banners, including Stein Painting 216, this chapter first reviews a
possible way to recreate the original formation of the painting. Paintings from the National
Museum of Korea and the State Hermitage Museum of Russia that have been neglected by
scholars are included in the discussion. Analyzing the inscription on Stein Painting 216, this
chapter ultimately argues that long banners were intentionally made as votive offerings
according to the belief that their materiality could enhance the efficiency of a donor’s wish,
along with a bodhisattva’s own auspicious abilities. Hence, this paper further claims that the
banner was a suitable medium for connecting believers and the subject of worship.

101

3.2. Stein Painting 216 and Its Reconstruction
Stein Painting 216 is a key example in this study, not only because of the inscription on
its top, but also because of its configuration. This painted banner is drawn on the entire width of
the silk.
8
The top portion, bearing an inscription, preserves the original full width of the silk,
which was divided into two pieces below it, like streamers. Selvedges—a term referring to the
edges of woven textile—on the sides of the cloth are clearly visible. Measuring between
selvedges is a convenient way to determine the original width of a piece of silk, which in China
in the tenth century typically measured 60 cm.
9
The edges of the trimmed silk were hemmed
with a thick, coiled thread to prevent them from unraveling (fig. 56). Instead of attaching two
pieces of silk together, Stein Painting 216 was conveniently cut up the middle. It therefore
provides a reference to an original configuration for similar paintings on which a single column
of bodhisattvas is painted on a narrow width of silk, such as Stein Painting 205, MG 17675, and
MG 17791. From the banner’s current condition, one can find four standing bodhisattvas on the
right strip, and four or five on the left, some of which are based on partial representations.
The inscription on the top presents the donor’s name, the purpose of the banner’s
offering, and the date it was offered (fig. 57). The style of writing is erratic, but Waley suggests a
possible reading as follows:
Controller of the Guiyi army…member of the Order of Silver-blue Luminous
Salary, additional President of the Board of Works, Censor, Upper Pillar to the
State…of Xihe jun, Ren Yanchao drawing blood respectfully (caused to be) painted
this forty-nine chi banner in one strip. This banner suspended on high from a dragon
hook…reach straight to… twisting about and flapping in the wind like a bird in

8. Stein, Serindia, 1029.
9. This painting is very fragmentary. It was restored in the 1920s and mounted on contemporary silk.

102
flight, like the coloured (hangings) in the Western Apartments of the Palace. May
his Excellency’s life be as that of the hills, his salary vast as the sea. May his Lady
Wife long be spared; may her flower-like countenance forever bloom. Next, it is
the object of this offering that his father and mother in the plain may long continue
to announce themselves in health and security, and for them are desired the same
blessings as for their son and his bride. The time being Da Zhou, third year of
Xiande...
10

From the inscription, we learn specific things about Stein Painting 216: the pigment used for the
painting (“blood”), the specific measurement of the length (“forty-nine chi”), an implement for
hanging the banner (“a dragon hook”), and the donor’s detailed wishes, including longevity,
good healthy, and security.
Since the long banner is characterized by its length, it is worthwhile to review a possible
way to reconstruct it in relation to other banners share a similar painting style. One might argue
that the “forty-nine chi” (fourteen meter) length is a notional and rhetorical expression, and I
agree with this idea to some extent.
11
It may not be possible to reconstruct the whole length of
Stein Painting 216 since each banner could have been painted for different donors by the same
workshop. However, there are no overlapping bodhisattvas’ names among other painted banners
similar to Stein Painting 216. In addition, a reconstruction would be beneficial in seeking to
understand dominant painting styles among the group of the painted banners comparable to Stein
Painting 216; as Whitfield argues, it could be a real measurement of the banner.
It is well known that the number forty-nine is commonly found in Buddhist sutras,

10. The inscription was originally transcribed by Arthur Waley and adapted by Whitfield in 1983. The
author has altered “forty-nine foot” to “forty-nine chi” for more accurate identification. Waley, A
Catalogue of Paintings Recovered from Tun-huang, 186–87; Whitfield, The Art of Central China, Plate
33.
11. Jacques Giès and Haewon Kim argue that it is not the actual length but a symbolic expression. Giès,
Les arts de l’Asie centrale, 295; Haewon Kim, ed. Kungnip chungang bangmulgwan sojang chungang
ashia chonggyo hoehwa 국립중앙박물관 소장 중앙아시아 종교 회화 [Central Asian religious paintings in
the National Museum of Korea] (Seoul: Kungnip chungang bangmulgwan, 2013), 154.

103
especially in relation to funerary rituals. For example, in Buddhist practice, memorial services
are held every seven days for forty-nine days (qiqi ri ««») after a person’s death. The number
forty-nine can also refer to the seventh and final ceremony (qiqi zhai ««…) that marks the
forty-ninth day after a person’s death, at which time the deceased has been judged for their sins
and good karma and has reached their destiny with the help of family members who have
contributed offerings during a series of ceremonies over the forty-nine days. Given the fact that
this is merely one of many common usages of this number in rituals and sutras within the
Buddhist community, it is reasonable to argue that the “forty-nine” of the “forty-nine-chi” banner
can simply symbolize its long length.
However, reviewing how this long banner was documented in the inventory of its
temple’s assets indicates the actual way in which the object was perceived and named. In the
“List of Permanent Objects (changzhu shiwuli ÀÃSTU)” also discovered in Cave 17, not
only is a “forty-nine-chi banner” listed but also a “forty-three-chi, large silk banner (sishisan chi
da juan fan \hÕƒ9[V)” and a “forty-five-chi, large silk banner (sishi wuchi da juan fan
\hŒƒ9[V).” The fact that the lengths vary but are all close to forty-nine chi underscores
the idea that an accurate measurement was sought and applied to this type of painting. To be
specific, Longxing Temple had thirty-one forty-nine-chi silk banners and eight forty-nine-chi
hemp banners, as well as three forty-five-chi silk banners and four forty-three-chi silk banners. It
is notable that the thirty-one silk banners identified as “forty-nine-chi” were found to actually be
this physical length when measured. It is uncertain whether Stein Painting 216 came from
Longxing Temple because its inscription does not provide any clue as to where the banner was
dedicated. However, based on this record, it is reasonable to suggest that the total length of the
original painting to which Stein Painting 216 belongs was in fact close to forty-nine-chi, or

104
fourteen meters.
In fact, within the group of long banners from Cave 17, there are a few that have painting
styles, color palettes, sizes, and support materials similar to those of Stein Painting 216. Quoting
Waley’s description, Whitfield argued that Stein Painting 216 and Stein Painting 196 were
“apparently painted by the same hand,” and therefore originally belonging to the same single
banner.
12
Agreeing with his hypothesis, I would add that a banner from the National Museum of
Korea should be considered in this reconstruction. A proposal of an original construction cannot
be relied upon based solely on evidence of shared artistic or iconographic features; one must also
analyze physical elements including pigments and weaving structure. However, it is still
worthwhile reconstructing a plausible original piece in order to confirm the shared painting
techniques and to fully understand this lesser-known and unfamiliar format of painting.
Looking closely at Stein Painting 216, one can see that the outlines of its bodhisattvas are
drawn in red, which is the dominant color in the painting. Between these outlines, red ink-
washed strokes are used to form the bodhisattvas’ flesh, garments, and draperies. Even though
red is the major color, the painter added some subtle accents by using gray, yellow, and black
ink. In terms of the overall artistic level of the painting, however, it is not comparable to other
finer paintings such as Stein Painting 6, Preaching Shakyamuni (fig. 58); the consistency of the
figures’ outlines and the wider range of colors in the latter point to it having been executed using
more labor and with superior craftsmanship. In other words, Stein Painting 216 may have been
drawn by an apprentice painter under less favorable conditions, such as an uneven surface, as
proposed by Whitfield.
13
The hasty mode of painting is evident in the bodhisattvas, through

12. Whitfield, The Art of Central China, Plate 33; Waley, A Catalogue of Paintings Recovered from Tun-
huang, 186–187 and no. 1.
13. Whitfield, Plate 32.

105
choices such as a single brushstroke delineating a figure’s lips and facial expressions that are
almost identical, despite the figures’ postures, mudra gestures, and angles of faces being slightly
different from each other. The anonymous painter or painters had to find efficient ways to
complete such a long double-sided banner; the use of thin silk and paint-wash techniques would
have expedited the process and saved time during production.
This painting style can be also found in Stein Painting 196 in the British Museum (fig.
59), indicating that Stein Paintings 216 and 196 could have been originally part of one single
banner as Waley first asserted.
14
The manner in which each bodhisattva in Stein Painting 196
was painted is almost identical to that of Stein Painting 216. The outlines and shading of the
figures is rudimentary, as if the painter dashed them off with single brushstrokes. Two
bodhisattvas and Ānanda, plus one fragmentary bodhisattva at the top of Stein Painting 196, are
drawn with red outlines that are filled in with a watery reddish-brown shade. The pupils of the
figures’ eyes and the figures’ necklaces are filled with black ink. The only difference between
the two is that with Stein Painting 196 one cannot find other hues, such as blue, green, or yellow,
which may be because these colors have faded. The thickness of the brushstrokes is also

14. This specific painting style is also found in Long Banner with Five Apsaras (EO 1166) at the Musée
Guimet in Paris, France. In contrast to the preceding examples, it features flying five apsaras (female
spirit of the clouds and waters in Hindu and Buddhist culture) on a piece of silk that is 19.6 cm wide.
Even though the subject matter is different, there are some stylistic similarities: red outlines, red ink
washes, and a spontaneous drawing style. However, its width is shorter than those of the three long
banners such as Stein Paintings 196 and 217, and Bon 4025. In addition, there is no hint of the application
of yellow and blue hues. Thus, it is safe to conclude that this painting must have been part of other
painting. With the except of the second apsara from the top, the four other apsaras face the right side of
the viewer, which suggests that this banner might have been paired with another with flying apsaras
facing the opposite direction. In my opinion, however, we cannot assume that EO 1166 was a piece from
the left based on the apsaras’ orientation. It could have been the far-right piece because of the location of
its selvedge. Looking closely, one can find the selvedge on the right side, whereas the opposite side
should have been cut and sewn. In instances where the silk was split for making streamers at the bottom,
there are some cases in which the far right or left streamer keeps its original selvedge. Giès, Les arts de
l’Asie centrale, plate 85 and figure 50.

106
consistently inconsistent in both. In Stein Painting 196, not only are the three lines of Ānanda’s
halo irregularly shaped as circles, but some parts of the lines are thick, while others are thin and
faded.
A banner of similar measurement with a painting style identical to that of both Stein
Painting 216 and Stein Painting 196 is Bon 4025 in the Ōtani Collection of the National Museum
of Korea (fig. 60).
15
In each banner, the figures’ outlines are achieved mainly with red pigments,
except for the eyelids, pupils, eyebrows, and lips where additional brushstrokes in black were
added in order to accentuate shapes. The volume and contour of the faces are all achieved by red
ink wash in most areas except the highest part of the faces such as nose, cheeks, and chins. Light
blue wash was applied to the figures’ hair in all three paintings. Application of a light-yellow ink
wash to the embellishment of the bodhisattvas is also evident in each banner, adding to the same
way that halos are depicted by red ink wash only. Comparing the bodhisattva faces in Stein
Painting 216, Stein Painting 196, and Bon 4025, it is undeniable that they were painted by the
same painter as a single piece (fig. 61).
In addition to the style and technique of Stein Painting 196, its measurement and the
ways in which the edges of the silk were treated are additional clues indicating that it could have
been part of Stein Painting 216. Stein Painting 196 measures 28.5 cm in width, exactly half the
width of Stein Painting 216 (59.0 cm) and close to the width of a bolt of silk produced in the
Tang period. Textile historians confirm that the standard size of a bolt of raw silk (shengjuan)

15. The Ōtani Collection refers to a group of objects from Central Asia that Ōtani Kozui (1876–1948), the
twenty-second Abbot of the Nishi Honganji Temple, and his colleagues excavated and obtained during
three expeditions to the region in 1902–1914. The collection is mostly comprised of mural paintings,
banner paintings, and sculptures. Part of the original Ōtani Collection is now displayed in the National
Museum of Korea. In the collection, there are more than fifteen paintings on paper or textile. Haewon
Kim, “Kungnip chungang pangmulgwan sojang chungang asia hoehwa,” in Kungnip chungang
bangmulgwan sojang chungang ashia chonggyo hoehwa, 32–43.

107
was approximately 12 m long by 56 cm wide.
16
Another way to check the width of silk is to find
selvedges on the edges. As discussed earlier, Stein Painting 216 reveals clear selvedges on the
left side of the left banner and on the right side of the right banner, confirming that the silk
support was originally one piece that was cut into two pieces. Bon 4025 at the National Museum
of Korea measures 28.2 cm wide—almost the same width as the banner strips of Stein Paintings
196 and 216. While this comparable width measurement cannot prove that these three banners
were originally part of the same painting, they are supportive evidence that allow us to create a
hypothetical reconstruction (fig. 62).
All in all, it is highly plausible that the long banners mentioned above could have been
painted by the same hand and even may be reconstructed as one whole painting. In this scenario,
there is no doubt that Stein Painting 216, which has a full inscription on the top, would be the
topmost section, while Stein Painting 196 would be a bottom part due to its donor portraits at the
bottom. One might assume that Stein Painting 196, based on the selvedge along the left edge of
its silk and from the location of the cartouches on the right, would have been part of the left
banner of the Stein Painting 216. Following this logic, Bon 4025 from the National Museum of
Korea would be from the opposite part of the banner because of its left-oriented cartouches and
selvedge on the right. However, I do not argue that Stein Painting 216 would be completely
whole with these banners added to it. Other fragments must have existed between Stein Painting
216 and Stein Painting 196, as the bodhisattva at the top of Stein Painting 196 is only depicted
below its waistline. Except for the fact that Stein Paintings 216 and 196 must have been the
beginning and end of the painting respectively, the order of these pieces as they would have been

16. Laure Feugère, “Introduction,” in Textiles from Dunhuang in French Collections, ed. Zhao Feng
(Shanghai: Donghua University, 2011), 18.

108
presented in the original painting is also uncertain.
As discussed earlier, Whitfield points out that the total length of the original long banner
would have been close to the forty-nine chi described in the inscription.
17
If we were to place
combine Stein Painting 196, Stein Painting 216, and Bon 4025 end to end, the full length would
be 917.4 cm, which is short of forty-nine chi—about 14 m long. Considering that there must
have been additional banner parts, the entire length could have been longer than the current
reconstructed size (see fig. 62).

3.3. Sharing the Form: A Comparison with Daoist Examples
A survey of the vocabulary that has historically been used to describe the Buddhist long
banner, both in sutras and historical documents, reveals its possible function. There are literal
descriptive expressions of this type, such as “long banner (changfan œV)” and “forty-nine-chi
banner (sishi jiu chi fan /–—ƒV).”
18
Other terms referring to this particular type of banner
describe a function or purpose, and include “spirit banner (shenfan “V)” and “life-extending
spirit banner (xuming shenfan ”‘“V).”
19
That the long banner is sometimes called a “spirit”
banner implies an auspicious functionality, possibly in conjunction with “life-extending” rituals
as the latter term indicates.

17. His argument is based on the reconstruction of Stein Painting 214 (1, 2) with EO 3647 and EO 3648.
Whitfield, The Art of Central China, Plate 35.
18. In the inscription of Stein Painting 216, “y,” which has the same phonetic pronunciation (fan) as that
of1, was used.
19. Some examples are Bhaisajyaguru Sutra (zÔ{|}a~ÄÅ':), T450, 404:20c;
Consecration Sutra (3ÇÉ4:), T1331 535:13b; and The Liturgy for Cultivating Dharani for the
Buddha Above the Leading General Atavaka (ÑÖÜáHà=âä3789:fiãåç), T1239
196:11–12a.

109
Whitfield claims that the sutra that explains how to use the long banner is the
Consecration Sutra.
20
The specific passage relevant to the long banner also informs us of a
specific target group who would benefit from having or making this type of banner. Specifically,
in fascicle 12 of the Consecration Sutra, the Buddha preaches when and how the forty-nine-chi
“spirit banner” is to be utilized:
The Bodhisattva Saving and Freeing said to the Buddha: “If there should be some
man or woman, gently born, who is desperately ill and lies on his bed in pain and
distress, without anyone to aid or to defend him; I now must urge and beg the
priesthood to fast wholeheartedly for seven days and nights, keeping to the Eight
Commandments, and carrying out ritual processions at the six hours of the day. Let
this sutra be read in its entirety forty-nine times. I urge them to light a seven-tiered
lamp, and to hang up parti-colored, life-lengthening spirit-banners... These should
be forty-nine feet long [my italics]. The seven-tiered lamps should have seven lights
per tier, following a form like a cart-wheel. Again, should [such a person] fall into
danger or be imprisoned, with fetters loading down his body, he should have parti-
colored spirit-banners made and forty-nine lamps lit, and should release various
kinds of living creatures, to the number of forty-nine [my italics] …”
21

This passage, in fact, prescribes the entire ritual process used for those who are extremely
ill and in distress as well as physically bound. The number forty-nine is repeated and emphasized
throughout the recitation of the sutra as it is an auspicious number, gives the length of the
banner, and enables living creatures to be released from danger. The cart-wheel-shaped lamp
mentioned in the text also has forty-nine lights, as each of its seven tiers is supposed to have
seven lights. If we read the passage more closely, the number forty-nine is also given as a length,
describing how long the “parti-colored, life-lengthening spirit banner” should be. That this

20. Whitfield, The Art of Central China, Plate 33.
21. Fascicle 12 of the Consecration Sutra is Consecration Sutra Spoken by the Buddha that Rescues
from Sin and Enables Salvation from Birth and Death (Fushuo guanding bachu guozui shengsi
dedu jing ∫Ω≈∆’÷◊ÿÖŸ⁄¤=, T. 1331). The translation is from Soper, “Literary
Evidence for Early Buddhist Art in China,” 171.

110
particular length is prescribed for a sick person specifically designates it as a “life-lengthening
spirit banner.”
Based on this terminology, I suggest that the length of the forty-nine-chi banner is
associated with “life-lengthening,” longevity, and healing from desperate illness. More
importantly, I argue that the wishes of the donor of Stein Painting 216—for the longevity of “His
Excellency,” his “Lady Wife,” and his parents—is materialized in the extremely long length of
the banner. It is debatable whether the term “forty-nine chi” in its inscription refers to the actual
length of the banner or is a rhetorical expression for a long length; either way, I would argue that
the term defines the banner’s meaning.
Interestingly, the same length of banner is found in Daoist ritual paraphernalia and is
called a “spirit-moving banner” (qianshen fan ‹“V) (fig. 62).
22
According to art historian
Shih-shan Susan Huang, the main function of this banner was to summon soul/“spirit”-
saving/“moving” deities, and it was specifically utilized in the Daoist salvation ritual called the
Yellow Register Purgation (huanglu zhai ›H…).
23
Instead of pictorial images, the banner has
scripted names and inscriptions as potent figures. For example, on the main body of the spirit-
moving banner, the names of Daoist soul-saving deities, such as the Heavenly Worthy Who
Rescues from Suffering (Taiyi jiuku tianzun fifl¨≠Z‡) and Heavenly Worthies of the Ten
Directions Who Save from Suffering (Shifang jiuku tianzun –·¨≠Z‡), are written. This
prioritization of writing and symbols over imagery is the major difference between the Daoist

22. Sometimes it is translated as a “spirit-removing banner.”
23. Shih-shan Susan Huang, Picturing the True Form: Daoist Visual Culture in Traditional China
(Cambridge: Harvard University Press, 2012), 218.

111
and Buddhist banners.
24
Based on these Daoist examples, a preliminary function of the long
banners or the spirit banners from the Library Cave might have been to evoke “spirits” or souls
of the deceased or Buddhist deities. Further comparison with Daoist spirit-moving banners
would provide more useful ideas about the Buddhist counterpart.
One of the main characteristics of the long banners in Cave 17 that differentiates them
from the rest of the painted banners is that their silk supports are dyed with one primary color
such as red, yellow, or dark blue. When bodhisattvas or Buddhas were painted on these dyed silk
supports, rather simple color palettes—such as silver, orpiment (as a substitute for gold), and
cinnabar or ocher for red—were used, all of which provide highly contrasting visual effects on
the supports. Daoist examples can provide clues to the symbolism behind these color
applications.
According to Huang, in Daoist salvation rituals, banners used were “mostly yellow,” with
some exceptions of blue-green, vermilion, and white.
25
Huang adds that the spirit-moving banner
is usually made with vermilion silk.
26
Interestingly, the colors of the long banners found in the
Library Cave are resonant with these Daoist salvation banners; Stein Painting 216, for example,
is painted on yellow dyed silk. Other examples, including Stein Painting 217, EO 1137, EO
1418, and EO 3657, were drawn on vermilion/red silk. Another group of long banners, including
Stein Painting 214, EO 3647, EO 3648, Bon 4019, and Bon 4043, are examples of long banners
drawn on navy blue silk.

24. Huang, 218 and 344–345.
25. Shih-Shan Susan Huang, “Daoist Uses of Color in Visualization and Ritual Practices,” in Color in
Ancient and Medieval East Asia, ed. Mary M. Dusenbury and Monica Bethe (Lawrence: Spencer
Museum of Art, the University of Kansas, 2015), 230.
26. Huang, “Daoist Uses of Color,” 230.

112
From these examples, one can find parallels between the Daoist ritual banners and the
Dunhuang long banners in terms of both color used and extreme length. A major difference,
however, exists with the configurations of the banners. According to a diagram in The Golden
Book of Salvation According to the Numinous Treasure Tradition (Lingbao lingjiao jidu jinshu),
the forty-nine-chi-long spirit-moving banner has a triangular “head” piece and a rectangular body
from which two “arms” and two “legs” are attached (see fig. 63).
27
This design is rather close to
those of shorter banners found in Dunhuang. While the majority of the long banners do not have
a triangular “head” and streamers along the sides, there exist a few instances in which a
triangular shape is included at the top, such as Stein Painting 217 at the British Museum (fig. 64),
with a standing bodhisattva, and Д x 291 at the State Hermitage Museum in Russia (fig. 65),
with Avalokiteśvara. Even though these examples do not have “arms,” their uppermost parts use
additional strips of textiles sewn to create pointy tops. For example, in Stein Painting 217, extra
effort was taken to decorate the headpiece by using three different colors—pale yellow, dark
blue, and red—of damask silk.
28
Looking at these examples, it seems possible that the “head”
part of Stein Painting 216 could have been torn off, as the whole upper part is missing along with
the red lines that should demarcate the inscription area.
It is still significant that there are echoing features between the Daoist and Buddhist long
banners. Does the similarity of the form follow the function? The mutual influence between
these two major religions is beyond the scope of this dissertation. However, it is worth pointing

27. DZ 466, 8:577c. See also Huang, “Daoist Uses of Color,” 230, figure 7.
28. In the case of the banner with Avalokitesvara (Д x 291) at the State Hermitage Museum in Russia,
only yellow silk strips were sewn on the top, which yields secure protection from fraying and includes a
loop. See Eluosi guoli Aiermita shi bowuguan and Shanghai guji chubanshe, Eluosi gu li Ai’ermita shi
bowuguan cang Dunhuang yishu pin éèê"ëíìîïñóëò{()ôöÅ [Dunhuang art relics
collected in the State Hermitage Museum of Russia] 1 (Shanghai: Shanghai guji chubanshe, 1997), plate
90.

113
out overlapping and similar concepts and practices, especially in the visual culture of the two
religions. Buddho-Daoist exchanges of materials, such as amulets and talisman writings found in
Dunhuang, have been discussed in depth.
29
I would expand this argument by contending that the
“forty-nine-chi banner” should be understood within the context of dynamic interactions between
Daoist and Buddhist ritual practices.
In this context, one religious practice in Dunhuang that is relevant to long banners such
as Stein Painting 216 is a healing ritual that utilizes a tiered lamp wheel along with a forty-nine-
chi banner.
30
According to art historian Ning Qiang, a painted scene on the northern wall of Cave
220 represents a healing ritual scene with Bhaisajyaguru, as a healing Buddha.
31
Revisiting the
previously discussed Consecration Sutra, its phrase “the seven-tiered lamps should have seven
lights per tier, following a form like a cart-wheel” is depicted pictorially on the northern wall.
Ning also argues that a multibranched lamp tree traditionally represents longevity, as it implies
the growth of life. This is why, he argues, the lamp fits into the function of the healing ritual,
pertinent to the extension of life.
32
The use of the lamp tree along with the forty-nine-chi banners
in the healing ritual is consequently rooted in longevity, healing, and salvation. This echoes the
Daoist salvation ritual, which aims to provide solace to deceased souls and to eventually save
them, as exemplified in the Daoist Yellow Register Purgation and the Buddhist healing ritual.
33

29. James Robson, “Signs of Power: Talismanic Writing in Chinese Buddhism,” History of Religions 48,
no. 2 (2008), 130–69; Christine Mollier, Buddhism and Taoism Face to Face: Scripture, Ritual, and
Iconographic Exchange in Medieval China (Honolulu: University of Hwai’i Press, 2008).
30. Ning Qiang argues that the multitiered lamp wheel probably originated from the Western Region and
was eventually incorporated to the Chinese Lantern Festival. For more information on the origin of the
festival and its local adaptation, see Ning, Art, Religion, and Politics in Medieval China, 122–33.
31. Ning, 20–23.
32. Ning, 128.
33. Huang, “Daoist Uses of Color,” 230; Huang, Picturing the True Form, 218.

114
Therefore, in the case of the hypothetically reconstructed long banner made up of Stein
Painting 216, Stein Painting 196, and Bon 4025, I contend that it is highly plausible that this
original banner would have been used in this type of salvation ritual—in particular, the healing
ritual—as it has features in common with its Daoist ritualistic counterpart.
34
As previously
mentioned, Ning states that the use of the light or lantern in the healing ritual possibly originated
from the Lantern Festival. According to Huang, the Lantern Festival coincided with the Festival
of the Upper Prime (Shangyuan jie ‚G„) during the Tang period.
35
Moreover, the specific
colors of red, yellow, and blue prescribed for the Buddhist salvation ritual banner share visual
norms with Daoist practice.
36

The term “forty-nine-chi banner” is indeed more commonly found in Daozang, Daoist
canon than the Buddhist canon, especially within Daoist scriptures related to salvation rituals.
The representative examples include the Marvelous Scripture of Salvation from Life and Death
in the Inferior Ways and the Five Sufferings (Taishang dongguan lingbao santu wuku badu
shengsi miaojing fi‚‰ÂÊÁ6Ë©≠’¤ÖŸÈ=, DZ 371) and Scripture for Saving

34. For the rest of the types of painted banners from Cave 17 (i.e., not long banners), Xie Shengbao and
Xie Jing propose that a large group of Buddhist paintings from the Library Cave could have been used for
the Water Land Ritual even though during this time period the paintings were not called “Water Land
Paintings.” I believe it is a worth reviewing their argument but further discussions are also needed on how
the paintings can be grouped and where the paintings were actually used. See Xie Shengbao õJú and
Xie Jing õù, “Dunhuang wenxian yu Shilu fahui: Dunhuang Tang Wudai shiqi Shuilu fahui yanjiu” (
)ûüî†°¢£: ()§`+‹›†°¢£./ [Dunhuang manuscripts and the study of the
ceremony of saving lives of the Water and the Land during the Tang Dynasty to the Five Dynasties],
Dunhuang yanjiu ()./ 2 (2006): 40–48; Xie Shengbao õJú and Xie Jing õù, “Dunhuang
wenxian yu Shilu fahui: Dunhuang Tang Wudai shiqi Shuilu fahui yanjiu zhi er” ()•pî†°p: (
)§`+‹›†°¢£./XC [Dunhuang manuscripts and the study of the ceremony of saving lives
of the Water and the Land during the Tang Dynasty to the Five Dynasties, second part], Dunhuang yanjiu
()./ 4 (2006): 1–7.
35. Huang, Picturing the True Form, 297.
36. Huang, “Daoist Uses of Color,” 230.

115
[Deceased Parents] from Distress in Future Lives (Taishang dongguan lingo wangsheng jiuku
miaojing fi‚‰ÂÊÍÎÖ¨≠È=, DZ 373). These Daoist scriptures from the Tang dynasty
repeatedly describe how to use the forty-nine-chi banner, as in the following:
Using colorful silk, make the “Spirit-Moving Precious banner,” which measures
forty-nine chi in height. Or one can make seven small-sized banners. Hang them on
the long staff facing wind, so that all sins are blown way and extinguished.
37

Comparing this with passages from the Consecration Sutra, one finds more similarities
than differences. While this Daoist prescription specifies the type of material for the banner (silk)
and the alternative format, the name of the banner (“Spirit-Moving”) and the length are identical.
More importantly, the dynamic movement of the banner is what activates the efficacy and
purpose of the banner: to remove all sins. In other words, the banner being blown and flipped by
the wind acts to eliminate the practitioner’s sins and thus guarantees his or her salvation.

3.4. Buddhist Banners for Longevity and Soul-Saving: Stein Painting 216
If the length and colors of the Buddhist long banners from Cave 17 such as Stein Painting
216 conform to what is prescribed in these Daoist sutras, the next question that must be answered
is what types of figures were depicted and what are their roles in the ritual process? In order to
answer this, it is critical to closely examine and identify the bodhisattvas depicted on these long
painted banners. Returning to Stein Painting 216, it has been mentioned that, besides the
inscription on the top, there are (from the viewer’s perspective) five bodhisattvas on the left
banner and four on the right banner. Five of these nine bodhisattvas have names that are legible;

37. The original text is: “t¶ß® ©™Ó´¨ ≠udÆi YØË¨E∞ ±≤≠≥¥S µ∂∑∏π
∫” (DZ 371: 7).

116
on the left side top to bottom are Lighting Virtue Bodhisattva (nanwu diande pusa Ï≤ÌÓj
k), Store of Virtuous Signs Bodhisattva (nanwu dexiangzang pusa Ï≤Óv5jk), Heavenly
King Bodhisattva (nanwu tianwang pusa Ï≤ZFjk), and Sound of Thunder Bodhisattva
(nanwu leiyin pusa Ï≤Ôajk). On the right side, only one bodhisattva, Precious Signs
Bodhisattva (nanwu baoxiang pusa Ï≤Ávjk), is labeled.
38
Stein Painting 196, which
includes donor portraits at the bottom, depicts Ānanda and two bodhisattvas whose names can be
found in a cartouche: Constant Tranquil Bodhisattva (nanwu changqingjing pusa Ï≤ íLj
k) and Flower Adorned Bodhisattva (nanwu huayan pusa Ï≤)jk).
39
Bon 4025, the long
banner at the National Museum of Korea, represents two bodhisattvas, of which only one can be
identified: Never Resting Bodhisattva (nanwu buxiuxi pusa Ï≤ÒÚÛjk).
40

It is rather uncommon for names of bodhisattvas to be depicted in painted banners from
Cave 17, and there is no single Buddhist text which gives all these bodhisattvas’ names together.
In other words, no sutra includes the names of all fifteen bodhisattvas from the reconstructed
banners with Stein Painting 216, Stein Painting 196, and Bon 4025. Although Waley identifies
them as different manifestations of Avalokiteśvara, there is one sutra in which five out of these
nine bodhisattvas’ names appear: Sutra on Solemn Attainment of Buddhahood by Means of
Repentance to Extinguish Sins in a Great, Thorough, and Broad Way (Datong fangguang

38. Waley reads this inscription as “Many Precious Signs ªºΩ.” However, when I examined the
inscription in person, I was not able to find the character ª. I believe he misread æø¿¡as ª. See
Waley, A Catalogue of Paintings Recovered from Tun-huang, 186.
39. The character of the Flower Adorned Bodhisattva written on Stein Painting 196 is read not as huá ¬
but as huā √.
40. The other one can be read partially as follows: FæZƒ(?)N(?)≈∆.

117
chanhui miezui zhuangyan chengfo jing 9Ù·ıˆ˜¯ÿ˘¿∫=, T. 2871).
41
This
apocryphal Chinese sutra celebrates all Buddhas and bodhisattvas through recitation of their
names as part of a confession ritual. By reciting the names, the practitioners can eliminate their
sins, including the most inexpiable and heinous sins such as the five ānantaryakarma and the
sins of the icchantika.
42
This sutra is specifically credited for a rather long list of names of
Buddhas and bodhisattvas.
43
The peculiarity and rarity of the bodhisattvas’ names featured in
Stein Painting 216, Stein Painting 196, and Bon 4025 can thus be possibly explained by or based
on this sutra.
If the Sutra on Solemn Attainment of Buddhahood provides all the bodhisattvas’ names, it
may provide a clue to the identity of a mysterious cartouche found on Bon 4025, the long banner
at the National Museum of Korea. An inscription that follows the character bai s meaning
“white” has not been positively identified because of the fragmentary condition of the letters (fig.

41. Waley, A Catalogue of Paintings Recovered from Tun-huang, 186; This sutra, previously lost, was
reconstructed with the manuscripts found in Cave 17 of the Mogao Caves of Dunhuang. Chou Po-Kan ∆
«», “Zhongshi ji zhongguo zaijia pusa zhi chanfa: dui …Datong fangguang chanhui miezui zhuangyan
chengfo jing de kaocha ” ÜÊÀÜáZÃ≈∆XÕ¢: Œ…=œ–—Õ“∫∏”‘o3: èå’
[An investigation of a confessional ceremony in medieval Chinese buddhism], Taida foxue yanjiu 18
(2009), 1–32.
42. Li-ying Kuo, Confession et contrition dans le bouddhisme chinois du Ve au Xe siècle (Paris:
Publications de l’École française d’Extrême-Orient, 1994), 140. The five ānantaryakarma includes
“patricide, matricide, killing an arhat, spilling the blood of a buddha, and causing schism in the monastic
order.” The icchantika means “incorrigible” in Sanskrit, referring to people “who have lost all potential to
achieve enlightenment.” See Robert E. Buswell and Donald S. Lopez, The Princeton Dictionary of
Buddhism (Princeton and Oxford: Princeton University Press, 2014), 40–41 and 370 respectively.
43. According to Dorothy C. Wong, another example of this sutra being the textual foundation of a
Buddhist sculpture is a stele called Chen Hailong zaoxiang bei æ÷$©–◊ from Shanxi, dated 562.
This stele, in contrast to Stein Painting 216 and its bodhisattvas, displays twenty-four small buddhas with
legible inscriptions of their names. Their visual representations are not distinguishable, a situation similar
to the bodhisattvas from the group of the long banners. For this stele and its textual resource, see Dorothy
C. Wong, “Guanyin Images in Medieval China, Fifth to Eighth Centuries,” Bodhisattva Avalokiteśvara
(Guanyin) and Modern Society (Taipei: Chung-Hwa Institute of Buddhist Studies, 2007), 266–70.

118
66). One possible reading, however, is xiang xiang or “fragrant elephant,” which may be part of
the name Bodhisattva of White Fragrant Elephant (nanwu bai xiang xiang pusa Ï≤s˙˚j
k), a name found in the Sutra on Solemn Attainment of Buddhahood.
44

Another interesting aspect of the Sutra on Solemn Attainment of Buddhahood is that it
was likely translated from Chinese to Tibetan and practiced or recited by Tibetan Buddhists for
forty-nine days when their family members or relatives died.
45
Even though there is no direct
reference in its current version to a Tibetan text, how this text was utilized in funerary rituals is
worth noting in terms of how its Chinese counterpart functioned in a ritual setting. It is not
surprising that this sutra has possible religious connections to funerary practices because reciting
the names of the Buddhas and bodhisattvas guarantees the elimination of even the most
egregious sins and, thus, appealed to dying individuals and their family members.
To sum up this point, the long banner form exemplifies reciprocal interactions between
multiple cultures. The format and configuration of the Daoist spirit-moving banner was
borrowed by Buddhists from what already existed in Daoist rituals and its practice as a spirit-
moving banner in forty-nine-chi as also adopted. The major difference is that the Daoist long
banner is aniconic—only the names of the Heavenly Worthy Who Rescues from Suffering and
Heavenly Worthies of the Ten Directions Who Save from Suffering appear and not their

44. T. 2871, 1344:14a–15a.
45. The name of the sutra in the Tibetan canon is hphags-pa thar-pa chen-po phyogs-suryas-pa hgyod-
tshans-kyis sdig-sbyans te sans-rgyas-su grub-par rnam-par-bkod-pa shes-bya-ba theg-pa chen-pohimdo
(vol. 37, no. 930) which can be translated as “Sheng da jietuo fangguang chanhui miezui chengfo
zhuangyan dacheng jing ¥=Kÿ–—Õ“∫∏o3”‘=Ÿ:.” See Chou, “Zhongshi ji zhongguo
zaijia pusa zhi chanfa: dui …Datong fangguang chanhui miezui zhuangyan chengfo jing de kaocha,”
8–9.

119
images.
46
The prescribed colors for the Daoist salvation banners provide further evidence for
some long banners from Cave 17 through their supports being dyed unusual solid colors dyed
such as red, yellow, and navy.
The application of visual representations of bodhisattvas in this format does help link
long banners with Buddhist ritualistic practices. The names of the bodhisattvas come from the
sutra used in confession and funerary rituals and its prescribed recitation of a long list of
Buddhas and bodhisattvas. The intrinsic nature of the healing, confession, and salvation rituals in
both Daoism and Buddhism are fundamentally indistinguishable, as the goals of both were to
assure and help the family of the deceased—and their parents in particular—to be reborn in the
Pure Land and save them from their sins from previous and present lives, thus salvaging their
souls. They are also closely related to the wish for the longevity and well-being of family
members as part of salvation.
47
This means that while the recitation of Buddha and bodhisattva
names is the intangible form of a ritualistic practice, painting the bodhisattvas is a way of
materializing the form of the ritual and its offering.
Practitioners believed that complete salvation and healing could not be completed
without acknowledging one’s past sins. This mindset of practitioners is well illustrated in one

46. The names of the Daoist deities are resonant with the aforementioned “jiuku pusa” or Savior
Bodhisattva from Suffering, who was one of the popular bodhisattvas featured in painted banners. This is
only example of many cases of Buddho-Daoist borrowings in visual materials.
47. Among the manuscripts from Cave 17, several apocryphal sutras related to lengthening one’s life span
were discovered. One example is Sutra on Prolonging Lifespan (fo shuo yan shouming jing 3Ç⁄¤‹
:, T2888). P. 2171 is the case in point. The practice of copying this type of texts was popular in the tenth
century. A colophon to P. 2805 (the Sutra of Goddess Marīcī) tells that a daughter from the Cao › family
clan copied four sutras, including Sutra on the Heart of Prajñā-pāramitā (Bore [boluomiduo] xin jing fi
fl[‡8·ª]‚:), Sutra on Extending Lifespan (Xuming jing „‹:), Sutra on Prolonging Lifespan
(Yanshouming jing ⁄¤‹:), and Sutra of Goddess Marīcī (Molizhitian jing ‰ÂÊG:), in order to
aid her sickness. For the full translation of the colophon, see Hao, Dunhuang Manuscripts, 87–88; Henrik
H. Sørensen, “Giving and the Creation of Merit: Buddhist Donors and Donor Dedications from 10th
Century Dunhuang,” Buddhist Road Paper 4, no. 3 (2020): 21–22.

120
colophon (P. 2805) to the Mārīcīdevīsūtra:
On the 13th day of the 10th month of the 6th xinchou year of the Tianfu [941], the
female disciple of pure faith, the young woman of the Cao family, commissioned
the copying of the Hṛdāyaprajñā- pāramitāsūtra in one roll, the Scripture on the
Extension of the Span of Life (Xuming jing ”‘=) in one roll, the Scripture on
Longevity and the Span of Life (Yanshou ming jing á¥‘=) in one roll, and the
Mārīcīdevīsūtra in one roll, respectfully offered on behalf of herself, as she suffers
from difficulties. Today she presents a number of scriptures, since the medicine
dumplings that were bestowed again and again in the morning still have not made
her well, and she now lies sick [in bed]. Beginning to realize her former misdeeds,
she humbly begs the Great Holy Ones to relieve her hardships and lift her out of
danger, and that the mirror will reflect the virtue of the copying of scriptures. She
[therefore] hopes to be protected, that this troublesome danger will be eliminated,
deceased family debtors will receive their capital [when] the merit is divided, and
that they will [subsequently] go for rebirth in the Western [Pure Land] [my italics].
With a mind full of prayer, she eternally supplies these [scriptures] as eternal
offerings [...].
48

After listing the titles of the four sutras that she copied (or commissioned to be copied),
“the young woman of the Cao family” explains the reason why she is offering the sutras: She has
taken medicine but it has not helped her recover from a sickness. Copying the sutra was one of
her last hopes, she states, and then she confesses her past sins. Ultimately, she was willing to
share any merit she gained from the copying of the sutra with those who had been harmed by
her. One of the copied manuscripts discovered in Cave 17 is the aforementioned Sutra on
Prolonging Lifespan (P. 2171), which includes the list of seventeen divinities (shen “) and the
benefits of reciting their names.
49
It is clear that these seemingly different rituals share an

48. The translation is slightly modified from Sørensen, “Giving and the Creation of Merit,” 21–22 and no.
34. The original inscription is “GÁËÚ [(941)] ÈÍ ˆ d¯d!˘ÎÏÌÓ>ËÔ>›ÒÚfifl
‚:DÛ, „‹:DÛ, ⁄¤‹:DÛ, ‰ÂÊG:DÛ, Ù^ıˆ˜L, ¯:˘˙, z˚¸˝u˛ˇ;
¯!"#, $%&', ()=¥*L+,, -.Ú:Å', /0,L12, 3Ã456»Áí, 7JF
–, 8À‚Ä, [...].”
49. P. 2171 is entitled Sutra on Prolonging Lifespan, but its content is different from S. 2428 (T2888),
which has the same title.

121
identical purpose: healing, eliminating sins, and, thus, being reborn in the Pure Land.
Returning to the inscription of Stein Painting 216, its donor, Ren Yanchao ¸á˝,
delineates his wishes: longevity for His Excellency and Lady Wife, as well as his family
members, including his parents, children, and wife. A wish for health and security follows. This
type of wish is not unique to this painting.
50
Nevertheless, this inscription’s sentiment echoes
both one of the names for a long banner, “Life-Extending Banner,” and the content of the sutras
discussed above.
Another notable element of the inscription of Stein Painting 216 is that it claims that the
painting was made in blood drawn from the donor, Ren Yanchao. Unfortunately, there has been
no conservation study conducted on either Stein Painting 216, so one wonders if this assertion is
true. A scientific analysis on the pigments and textile materials of Bon 4025 was conducted by
the National Museum of Korea in 2013. It revealed that the red pigment in Bon 4025 is cinnabar,
not actual blood, as there were no traces of iron.
51
If we assume that Bon 4025 was painted by
the same artist as painted Stein Painting 216, it is safe to conclude that the same type of pigment
must have been applied to Stein Painting 216 and that actual blood from the donor was not
used.
52

Even though the pigment used in the work does not contain actual blood, it is worthwhile
to consider what the inscription’s claim of using blood as an artistic medium connotes. There are

50. Stein Painting 217 bears a similar inscription: “May the land be peaceful and its people prosperous;
May the rural shrines continually flourish. May the whole house be clean and happy; May the lives (of the
inhabitants) be long extended.” Waley, A Catalogue of Paintings Recovered from Tun-huang, 188.
51. Before this result, Whitfield speculated that “the red pigment used was probably a local red earth,
tuhong ~9 for the washes and cinnabar, zhu : for the brighter outlines.” Whitfield, The Art of Central
China, plate 32; Kim, joongang Asia jonggyo hoehwa, 237.
52. John Kieschnick, “Blood Writing in Chinese Buddhism,” Journal of the International Association of
Buddhist Studies 23, no. 2 (2000): 177–94.

122
several examples of Buddhist manuscripts with inscriptions that claim they are written in the
donor’s blood. Owing to a recent conservational study, one Tibetan Buddhist manuscript (IOL
Tib J 308) was identified to have been written with actual blood: Homage to Aparimitāyus Sutra
(Aparimitāyurnāma-sūtra).
53
This particular sutra was popularly copied due to the belief that
doing so would bring longevity.
54
An X-ray fluorescence (XRF) analysis of the light brown link
of IOL Tib J 308 revealed that its ink contained iron (Fe), calcium (Ca), potassium (K), titanium
(Ti), sulfur (S), and zinc (Zn), without any traces of mineral pigment.
55
This means it is highly
probable that the main source of the light brown color was, in fact, blood.
It is known that the act of drawing blood as part of self-immolation was considered to be
a devotional and compassionate behavior as well as a sign of filial piety to parents since the late
fourth-century in China.
56
A notable number of monks were known for writing and copying
scriptures with their own blood.
57
For example, the tenth-century monk Yongming Yanshou O
˛á¥(904–975) wrote a death verse in blood and the ninth-century-monk Dinglan ôˇ (d.

53. Sam van Schaik, Agnieszka Helman-Ważny, and Renate Nöller, “Writing, Painting and Sketching at
Dunhuang: Assessing the Materiality and Function of Early Tibetan Manuscripts and Ritual Items,”
Journal of Archaeological Science 53 (2015): 117.
54. This sutra was copied numerous times, especially for the Tibetan emperor during the 830s–840s. Van
Schaik, Helman-Ważny, and Nöller, 117–18.
55. No indication of mineral pigment was based on the result of VIS-spectroscopy. Van Schaik, Helman-
Ważny, and Nöller, 118.
56. In early Buddhism in India, this idea is deeply rooted in jātaka tales, which are tales of previous lives
of the Buddha. They describe how the Buddha was compassionate enough to cut his own body parts to
save other sentient beings. For the history of self-immolation, see James A. Benn, Burning for the
Buddha: Self-Immolation in Chinese Buddhism (Honolulu: University of Hwai’i Press, 2007), 7–18.
57. For the practice’s scriptural basis, see Kieschnick, “Blood Writing in Chinese Buddhism,” 178–81.
According to Jimmy Yu’s publication on blood writing, the most popular sutras that preached the benefits
of self-sacrifice were written in blood. Such texts include the Lotus Sūtra, Brahma Net Sūtra, Flower
Garland Sūtra, and Diamond Sūtra. Jimmy Yu, Sanctity and Self-Inflicted Violence in Chinese Religions,
1500–1700 (New York; Oxford: Oxford University Press, 2012), 41.

123
852) copied sutras using blood.
58
It was not exceptional for lay people to do this as well.
Furthermore, it was believed that by doing so the practitioners could transfer merit to their
parents and, thus, repay them for their benevolence.
59
In one of the earliest known examples of
using blood as an artistic material in China, Yuan Dexiu ªÓ!, an eighth-century official,
pricked himself to draw blood upon his mother’s death, and painted images and copied scriptures
using this blood.
60
Daozhou 4" (864–941) used his own blood to paint an image of the
Thousand-Armed and Thousand-Eyed Avalokiteśvara, in addition to removing part of his own
ear.
61
In Tibet, a monk is said to have used the blood from his own nosebleed to make a painting
of an eminent monk.
62
Thus, in its claim of using blood as paint, Stein Painting 216 is not an
unusual case, but represents one way of proclaiming religious devotion and even expectations of
increasing religious efficacy.
Cinnabar, one of the oldest pigments used in China, is traditionally understood and
utilized as an alternative to blood or as a substance comparable to blood—thus it has symbolic
associations with death.
63
Moreover, cinnabar is also a critical material for alchemy in Daoist
rituals being thought to assist with obtaining immortality. Such rich symbolism means that the
choice of cinnabar for Stein Painting 216 reflects the pigment’s strong associations with blood. It

58. Benn, Burning for the Buddha, 123 and 139.
59. Kieschnick, “Blood Writing in Chinese Buddhism,” 183; Benn, Burning for the Buddha: Self-
Immolation in Chinese Buddhism, 153; Yu, Sanctity and Self-Inflicted Violence, 38–47.
60. Kieschnick, 183.
61. Kieschnick, 187.
62. Roberto Vitali, Early Temples of Central Tibet (London: Serindia, 1990), 51.
63. For the materiality of the colorant cinnabar in early Chinese visual history, see Guolong Lai, “Colors
and Color Symbolism in Early Chinese Ritual Art,” in Color in Ancient and Medieval East Asia, ed. Mary
M. Dusenbury (New Haven and London: Yale University Press, 2015), 25–43.

124
is also highly plausible that actual blood could have been added to the cinnabar at a ratio too
small to be detected in the scientific analysis that was conducted on some sample spots from Bon
4025.
To summarize, the artistic practice of drawing blood to be used as a painting pigment
originated from blood-writing practices in Buddhism. Sacrificing one’s blood would not only
represent one’s devotion and compassionate zeal in making an offering, it was also thought to
increase the offering’s efficacy. In particular, during the ninth century of the Tibetan occupation
period, the act of copying a sutra that was believed to guarantee one’s longevity was evidently
practiced by the Tibetan emperor, as thousands of copies of such sutras were discovered in Cave
17.
64
These sutras and images of Buddhist deities were written or painted in blood or its
symbolic replacement, cinnabar. Considering that the format of the long banner was possibly
adopted from Daoism, it is not surprising that cinnabar was chosen for the main pigment for
given its traditional materiality. Stein Painting 216’s use of cinnabar and its long length were
purposeful choices made to help this offered object achieve its purpose of extending the lives of
a deceased person’s parents, who would eventually be saved and reborn in the Western Pure
Land.

3.5. Religious Efficacy and Its Display as Double-Sided Painting
It is worth discussing how a hypothetical long painted banner reconstructed with Stein
Paintings 216 and 196 and Bon 4025 would have been displayed. Physical characteristics of the
original painting, such as its extremely long length and materials, strongly suggest that this

64. Sam van Schaik, “Towards a Tibetan Paleography: Developing a Typology of Writing Styles in Early
Tibet,” in Manuscript Cultures: Mapping the Field, ed. Jörg Quenzer, Dmitry Bondarev, and Jan-Ulrich
Sobisch (Berlin: De Gruyter, 2014), 309.

125
particular type of long banner was designed to be viewed from multiple angles. Long banners
functioned as landmarks at large gatherings such as rituals or festivals including the Lantern
Festival discussed earlier. This type of display would have provided a completely different
viewing experience from that of traditional Chinese paintings; the latter were displayed in a static
manner, enabling close looking by a limited number of viewers at one time.
My argument is based on paintings whose construction is similar to that of Stein Painting
216. Among the twenty-nine painted long banners from Cave 17, most were painted on both
sides. To be specific, in case of the Pelliot Collection in Paris, among the twelve examples of
painted long banners, seven are known to be painted on their verso. One comparable example is
EO 3647, a long banner featuring bodhisattvas painted with yellow pigments on blue-dyed silk.
Its length is 580.0 cm. Its overall shape is very similar to that of Stein Painting 216 and, more
importantly, the design on the front is visible on the verso (fig. 67). Other examples such as EO
1137, The Bodhisattva of Avalokiteśvara, Savior from Perils, and EO 3657, The Bodhisattva of
Avalokiteśvara who Prolongs Life, are also painted on both sides.
Based on a comparison with similar types of painted long banners, it is reasonable to
surmise that Stein Painting 216 was also intended to be viewed from both sides. Unfortunately, it
is now impossible to examine the verso due to how the painting was mounted in the 1910s.
65

65. When all the materials found during Stein’s Second Expedition (1906–08) were first entered into the
British Museum’s Department of Oriental Antiquities on August 5, 1909, the prime concern and foremost
task was to “flatten” the paintings in order to sort, research, and catalogue them. Sydney Colvin and
Laurence Binyon, the keeper and assistant keeper of the Stein Collection respectively, described the
condition of the paintings when they first arrived at the British Museum as follows: “The paintings were
in crumpled fragments crushed together, sometimes separated in different bundles, and sometimes a mass
of brittle fragments.” This was true of Stein Painting 216, as part of the painting was separated and
because of its fragmentary condition. For conservators such as Stanley William Littlejohn (1876–1916),
Urushibara Mokuchū (1888–1953), and W.J. Goodchild (?–?), who were involved in the Stein Collection
project, it was not ideal to leave the gauze-like silk in fragments. Instead, their choice was to mount and
flatten the works. Littlejohn devised “a method of backing the paintings with a neutral-tinted silk and
mounting them on light stretchers.” Without significant retouching done to the original paintings,
flattening and backing or mounting the paintings was the main focus of the conservation team. Stein

126
However, based on Stein’s description regarding the painting and looking at recent conservation
achievements, we can speculate a highly probable scenario. In his multi-volume post-expedition
report, Serindia, Stein noted the special quality of the material of Stein Painting 216, describing
the painting as a “roll made of single width of thin glazed silk.”
66
When I examined the painting
in person, I agreed with his observation because of the fragility and delicacy of the painting’s
silk support. One might wonder why thin silk was particularly preferred or even matters for this
type of work. The choice of material here is noteworthy because it was not arbitrarily chosen by
artisans; it had both economical and functional advantages.
As for the economic use of the material, Whitfield correctly remarks that “the material
itself was sheer enough to ensure that the image painted on the front would show through clearly
on the back.”
67
In case of Stein Painting 216, the banner’s sheer quality allows viewers to see the
image painted on the recto when viewing the verso. This choice is an efficient way of saving
both materials and time when producing the painting. In addition, the painter hemmed the inner
edges of the trimmed silk with a thick coiled thread to prevent fraying (unraveling) instead of
using a backing material, again saving materials.

Painting 216 was backed using this method. In addition, the backing and mounting processes were
completed rather quickly to facilitate the paintings’ display at the Crystal Palace for the Festival of
Empire exhibition in 1911. Within a year before the exhibition, over 200 paintings had been backed or
mounted, according to Stein. Another record states that Littlejohn had to devote more than two months to
framing and mounting for the show at the Festival. This source does not indicate the exact number of
paintings he worked on within that time frame, but it is not difficult to imagine that making the paintings
presentable was the main purpose of the flattening project. For the history of the Stein Collection in
London, see Helen Wang, Helen Persson, and Frances Wood, “Dunhuang Textiles in London: A History
of the Collection,” ed. Feng Zhao (Donghua University Press, 2007), 1–8; For the conservator Littlejohn,
see Laurence Binyon and Sidney Colvin, “The Late Stanley William Littlejohn,” The Burlington
Magazine for Connoisseurs 32, no. 178 (1918): 16–19.
66. Stein, Serindia, 1029.
67. Roderick Whitfield, “Four Unpublished Paintings From Dunhuang in the Oriental Collections of the
British Library,” The British Library Journal 24 (1998): 96.

127
A painting at the Metropolitan Museum of Art, Banner with Bodhisattva, possibly
Mahamayuri (fig. 68), demonstrates how this artistic practice would have worked in Stein
Painting 216. When the painting entered the Metropolitan Museum of Art’s collection in 2007, it
was mounted onto a new panel. In 2019, the museum’s conservation team removed the painting
from this panel, revealing that the image on the verso was—through tracing and applying colored
pigments inside the black outlines—the mirror image of the bodhisattva on the recto.
68
The
sheerness of the work’s silk allowed its artist to skip the outlining step on the verso because the
recto outlines showed through the material.
In Stein Painting 216, outlines are achieved with bright red lines only, and they are filled
with color washes. In terms of the brush technique and the number of colors in the overall
palette, it is much simpler than the Metropolitan Museum of Art banner. Like the Met banner,
Stein Painting 216’s silk supports are sheer enough for pigments to penetrate through to the other
side; its dominant red washes on the recto show through and reduce the number of brushstrokes
that had to be applied to the verso. In other words, the back of the painting might have been
completed by adding only a few touch-up brushstrokes to enhance the lines that penetrated from
the front.
The second reason for choosing light, thin silk for this type of long banner is closely
associated with its practical intended use. One interesting phrase from fascicle 12 of the
Consecration Sutra discussed earlier is relevant to the prescribed positioning of the life-
lengthening spirit banner. The author of the sutra instructs that one should light a “seven-tiered
lamp” before hanging up the banner. It is not entirely clear from the passage whether the spirit
banner was supposed to be hung on the top of the lamp or on a separate structure nearby.

68. Wang, Wen, and Whitfield, “Buddhism and Silk,” 8–25.

128
However, based on visual representations of banners in mural paintings and descriptions in
written records, one can assume the latter was most likely to be the case. Based on the pictorial
representation of a temple banner painted on the walls of the Mogao Caves, a banner would
likely be hung on a tall staff. The north wall of Cave 231, which is contemporaneous with Stein
Painting 216, shows how a temple banner would have been hung. On the top of a tall staff with a
dragon-headed hook, a color-blocked banner with streamers on the bottom and a triangular head
on the top sways dynamically in the air (fig. 69). Even though no actual object corresponding to
the visual representation on the wall paintings was discovered in Cave 17, the temple asset
inventory (P. 3638) of Jingtu Temple, written in 911 CE, indicates that the temple possessed a
“dragon-head staff for a banner (fan gan long touV√Q#).”
69
One relatable example from the
contemporary tenth century CE, a lacquered and gilt dragon-head banner-pole, is from the
Goryeo Dynasty (918–1392 CE) (fig. 70).
70
Even though this miniature-sized (104.3 cm)
decorative banner-pole was intended to be used inside a temple’s main hall, its overall shape
reflects the larger-sized ones that were placed outside of a temple or cave structure.
71

In addition to the long pole for the banner’s display, Whitfield also suggests that it could
have been hung without a particular implement by instead suspending it “outside of a cliff face
or inside a cave such as Cave 130, which shelters a twenty-six-meter figure of the Buddha
Amitābha from the eighth century.”
72
This is a highly plausible scenario, especially if a banner
prop was not readily available. It might not even have been required to have such a prop.

69. P. 3638; Ma, Dunhuang gudai gongjiang yanjiu, 431.
70. This is Korea’s National Treasure No. 136.
71. http://www.leeum.org/html_eng/collection/traditional.asp#zoom (Accessed April, 30, 2019)
72. Whitfield, The Art of Central China, plate 33.

129
Considering the height of the cliff face of the Mogao Caves, where even the nine-story temple of
Cave 96 was constructed, the outdoor setting of the Mogao Caves already offers the optimal
height for such long banners. Only when it was hung from the cliff face, it can be “twisting about
and flapping in the wind like a bird in flight” as described in the description. It signals that
movement and motion were key characteristics of the banner, as it represents a symbolic emblem
of the Western Paradise. The top part of Stein Painting 216 resembles post and lintel
construction. The dynamic movement would have been achieved by cutting the single width of
silk in half as the banner would have moved like a kite tail.
This mode of display did not require meticulous articulation of details especially when it
was viewed from afar. What was considered more important than the execution of details after
all was the length of the long banner. It corresponds to the ways in which temples’ inventories
described the long banners; what defines them was the total length of the objects.

3.6. Conclusion
Through an examination of similar painting styles seen in Chinese banners in museums
throughout the world, this discussion attempted to reconstruct a long banner with Stein Painting
216, Stein Painting 196, and Bon 4025, which would have been longer than ten meters. Even
though the estimated length does not literally match with the description of the “forty-nine-chi”
spirit banner, it indicates that there was an effort to offer a very long banner in order to follow
scriptural prescriptions. Dominant depictions of bodhisattvas on the banners signal their roles in
the rituals that the banners were used in; more in-depth research on Daoist and Buddhist
salvation rituals is needed to expand on this argument. And the concept of soul-saving, or soul-
carrying in the context of rescuing from calamity and death, should also be researched further, as

130
many of the painted banners other than the long banners feature the Rescuing Bodhisattva (¨≠
jk) and the Bodhisattva as Guide of the Soul (∞±jk).
The style of the long banners is completely different from that of the rest of the painted
banners in Cave 17. Only long banners have line-drawing paintings, a different genre from the
“regular” painted banners in which outlines are filled with multiple layers of pigments. In
addition, the dyed color of the long banners’ supports replaces hand-done background coloring,
which would save painters time and effort. This practice also suggests that artists would have
wanted to emphasize innate characteristics of the material such as the quality of the silk. Dyed
silks were used as the main painting support, in contrast to “regular” painted banners on which
the support was not normally dyed with pigments. Based on this observation, artistic intention
and effect were one of main considerations of the long banners’ artists. Applying contrasting
pigments to the bright or dark silk—such as gold, silver, black ink, and red—supports such an
assumption.
73

The long banner paintings unpack multiple stories that enrich our understanding of
artistic and religious practice in Dunhuang. Above all, a long banner should not be understood as
a simple devotional offering; rather, it was a product of rich Buddhist and Daoist traditions
meant for rituals practiced by the Chinese and Tibetan communities. The most common wishes
for ordinary lay people during the tenth century CE were to live long and well without having
serious health issues—this is true of people today, as well. Sutras and rituals related to
lengthening a donor’s and his/her family members’ lifespans gained significant popularity and

73. This painting method is reminiscent of an illuminated manuscript in which the first page is the painting
of a sermon scene. Sutras were often copied on paper that was dyed dark blue and then written and painted
on using gold foil and gold or silver pigments with hints of colors, much like techniques used with the
painted long banners. Moreover, by using gold to accentuate the outlines of the figures and architecture,
the illuminated painting is in sharp contrast with its dark background and the bright line drawings.

131
attention. Stein Painting 216 is a product of this phenomenon. This so-called spirit banner was
prescribed to be at least forty-nine-chi (or feet) as its length, and was destined to be used either in
a healing ritual, at a lantern festival, or as part of a confession ritual before death—all of which
require its donor to repent past sins. Moreover, together with its extremely long length, which
undoubtedly represents a long life span, the drawing of a donor’s blood to be used as the artistic
medium was one of the most significant aspects of the painted banner. Whether or not a few
drops of blood were mixed with cinnabar, the characteristics of this particular pigment still
signify blood, life, and viability, which conforms precisely to the purpose of this long banner: the
desire for longevity.

132

Chapter 4: Silk as a Devotional Offering and the Tibetan Impact on Painted Banners
4.1. Introduction
To fully understand ninth- and tenth-century painted banners from the Dunhuang region, is
important to think about them as not merely two-dimensional Buddhist paintings with central
pictorial representations, but rather as complete objects of offering adorned with multiple textile
pieces carrying symbolic meaning. Various types of silk, including recycled pieces of silk, were
used for framing and mounting these painted banners. With this in mind, giving a painted banner
to a temple for a ritual/merit-making purpose can also be understood as a textile offering.
Moreover, large caches of textiles found in Cave 17, including fragments whose purpose or
purposes have not yet been identified, testify to the high demand for and popularity among the
locals of textiles as offerings in temples. I intend to investigate why textiles and painted banners
framed adorned with multiple textiles were considered qualifying objects for Buddhist offering
by focusing specifically on a funerary ritual in which textiles were generously and popularly
used as offerings. Among different types of textiles used as offerings, the most popular one was
silk. In the painted banners from Cave 17, the supports and accessories (e.g., surrounding
borders, loops, and streamers) were mostly silk. Why was silk chosen as an appropriate material
for offering? Whereas Chapter 1 reviewed the artistic benefits of having silk as a painting
surface, this chapter seeks to determine the historical and religious connotations behind the use
of silk, with examples from tombs and cave-temples.
Exploring the material significance of these painted banners must begin with determining
their origins and visual sources, especially the rectangular format with surrounding borders and

133
loops.
1
What prototypes in visual culture inspired the Dunhuang painted banners’ format and
embellishment? That the mounting method for a Dunhuang painted banner is distinctive from
that of a traditional Chinese painting suggests different sources and inspiration gained from
outside Dunhuang and requires an understanding of the period-specific regional circumstances.
There is a sharp increase in the frequency of painted images appearing on textiles during and
after the Tibetan occupation period. It is a fact that the sixty years of the Tibetan conquest over
Dunhuang brought new religious thoughts, practices, icons, and artistic practices into the area. It
is possible that the introduction of the Tibetan Empire’s pre-Buddhist and Buddhist artistic
practices to Dunhuang’s preexisting visual culture was more widespread than is currently known.
In this respect, this chapter first outlines the significance of silk textiles as offerings and then
reviews possible sources that shaped the ninth- and tenth-century painted banner format. It
finally argues for the crucial role of Tibetans in transferring and delivering image-making
practice and the use of textiles in longevity, connectivity to the Pure Land, and successful
delivery of the soul.
The types of textiles from Cave 17 stored in the British Museum include but are not
limited to the following objects: canopies, valances, banners (both iconic and aniconic), banner
parts, covers and ties, patchwork, polychrome weaves, monochrome weaves, dyed textiles, and
embroidery.
2
It has been observed that the textile collection from Dunhuang is “almost

1. An early model of the elongated banner format—with a head, arms, streamers, and wooden
stretcher at the bottom—is easy to find in the murals of the Mogao Caves. This type of banner
is normally represented on the top of the stupa/pagoda as seen on the south wall of Northern
Wei Dynasty Cave 257. A fifth-century example shows a banner without any images of
Buddhist deities, hanging from the top of the stupa. Dunhuang wenwu yanjiusuo ()ûë.
/_, Zhongguo shiku: Dunhuang Mogaoku Ü"2A: ()?@A 1 (Beijing: Wenwu, 1982),
Pl. 40.
2. Wang, Persson, and Wood, “Dunhuang Textiles in London,” 18.

134
exclusively silk.”
3
While acknowledging the most popular material in Dunhuang is important, it
should be pointed out that the silk referred to here is traditionally categorized as “textile” silk. In
other words, “silk” only refers to non-painting materials, which means a painting support would
be excluded from the “textile” category. Ironically, a banner painting’s subordinate components
such as its textiles for framing, loops, and head and streamers are categorized as “textiles.” Thus,
these parts of the Stein Collection are separate from other painted materials and have been
studied in terms of their technical aspects.
4
Consequently, the exclusion of a painting as a textile
object depreciates both the painting’s value and its meaning as a votive object. Therefore,
revisiting and reconsidering a painted banner as a textile devotional offering will shed new light
on the banner’s underlying cultural values, symbolism, and the fundamental reason for its
donor’s preference toward a particular type of textile in the ninth to tenth centuries in Dunhuang.
There are several simple reasons why silk was mainly utilized for paintings in the
Dunhuang area. First, silk was widely available to purchase or obtain from local textile
workshops. Furthermore, a large influx of silk and hemp textiles from central China—as corvée
tax and military payment—contributed to the wide availability of silk.
5
In addition, extensive silk

3. Wang, Persson, and Wood, 18.
4. The “textiles” and paintings from Dunhuang, in particular, which were from his second and third
expeditions, were allocated to the British Museum and India Office mainly because the two expeditions
were funded by these institutions. For the Stein collection’s history and how the Stein’s textile findings
from his three expeditions were distributed to the British Museum, the India Office (later incorporated
into the British Library), and the Victoria and Albert Museum, see Wang, Persson, and Wood, 14–16.
5. According to Masahiro Arakawa, the military expenditure for the increased military commands in the
eighth century—especially after 713—was drastically escalated, based on the historical record from the
Jiu Tangshu ;§< [old Tang history]: “Before the Kaiyuan reign (713–741) the border costs did not
exceed 2,000,000; in the mid-Tianbao reign (742–755) the number reached this figure [of 12,100,000].”
Soldiers’ salaries are included in this “expenditure,” as another record by Tang scholar Du You (=>,
735–812) pointed out that “1,700,000 bolts of silk/hemp” were used to “pay soldiers’ salaries.” Many of
the textile payment included wide loom juan-silk (da sheng juan =J^), plain silk dyed red (manfei ?
]), plain silk dyed green (manlü ?@), and silk floss (damian =A). Those tax silks were produced in
the Hebei, Henan and Jiannan circuits. Masahiro Arakawa, “The Transportation of Tax Textiles to the
North-West as Part of the Tang-Dynasty Military Shipment System,” Journal of the Royal Asiatic Society

135
trade between China and Central Asia made possible the existence of diverse weave patterns and
types of silk. While these provide practical explanations for its abundance, cultural values
driving the preference for silk have been overlooked. In other words, it is probable that historical
connotations embedded in silk thread and silk might have contributed to boosting its ritualistic
use, for instance in funerary practice. The first known example in this regard is the famous T-
shaped silk banner excavated from the Han Dynasty’s (206 BCE–24 CE) Mawangdui §F$
Tomb Number 1 near Changsha œ% in Hunan Province (fig. 71).
6
This silk banner, which was
placed over Lady Dai’s the innermost coffin, has caused heated debates regarding the meanings
of its pictorial representations, its proper name, and its functional use in the context of ritualistic
use.
7
This banner is an example of how a mortuary practice was materialized and consolidated
with a pictorial representation symbolizing the deceased’s “permanent home” before Buddhism

23, no. 02 (2013): 247–249, 251, and 253. For additional discussion of how hemp cloth and silk arrived in
Xinjiang—especially in Turfan—see Binghua Wang and Helen Wang, “A Study of the Tang Dynasty Tax
Textiles (Yongdiao Bu) from Turfan,” Journal of the Royal Asiatic Society 23, no. 02 (2013): 276–79.
6. This tomb, undisturbed since its closing during the Western Han period (206 BCE–24 CE), was
excavated in 1972 and yielded more than a thousand of objects of silk, ceramic, lacquer, and wood. For its
archaeological reports, see Hunan sheng bowuguan BFCóëD, ed. Changsha Mawangdui yihao Han
mu ≠Ç&çEDFGH (Beijing: Wenwu, 1973). For a summary of the reports in English, see David
Buck, “The Han Dynasty Tomb at Ma-wang-tui,” World Archaeology 7, no. 1 (1975): 30–45. For the
most updated study on the site and on the silk manuscripts in particular, see Luke Waring, “Writing and
Materiality in the Three Han Dynasty Tombs at Mawangdui,” diss., Princeton University, 2019): 32–43.
7. Since the painting’s discovery, a number of studies have tried to identify its mysterious use. Yet as
Silbergeld stated in his article in 1982, “Despite the best efforts to collect and apply literary accounts, a
more specific function of the banner within a funerary context, much of its visual content and meaning,
the link between its iconography and funerary function, and even its name has not yet been determined.”
Ten years later, Wu Hung responded to this statement and denied what had been suggested in the previous
studies. He argued that the banner does not represent Lady Dai’s garment, nor is it associated with
summoning the deceased soul. It does not depict the soul’s journey to Heaven. He argued that the painting
should be understood in the context of burial art and interactions with the other three coffins, based on the
bin I/J ceremony. According to Wu, the banner should be called “Name Banner (mingjing K(L)M),”
which symbolizes the otherworldly existence of the dead. Jerome Silbergeld, “Mawangdui, Excavated
Materials, and Transmitted Texts: A Cautionary Note,” Early China 8 (1982): 79–92; Hung Wu, “Art in a
Ritual Context: Rethinking Mawangdui,” Early China 17 (1992): 111–144, especially 116–127.

136
was widespread in ancient China and flourishing nationwide.
8
In addition to this famous banner,
a great number of textiles were discovered in Mawangdui Tomb Number 1. Lady Dai, the wife
of the Marquis of Dai in the Han Dynasty during the period 206 BCE–220 CE, had her face
covered with two pieces of silk, whereas her body was covered and bound with multiple layers
of cloth and shrouds.
9
Mawangdui’s silk banner, however, was made more than a thousand years
before the Dunhuang painted banners. While a direct comparison between them would be
impractical because of this time difference, the overarching idea of using a silk banner/painting
as an essential part of a funerary ritual process leads us to assess painted banners not only as
objects made for Buddhist “merit making,” but also as textile offerings in a broader ritual
context.

4.2. Textile as an Offering
Donors who sponsored making Buddhist paintings and constructing cave-temples
believed that spreading and copying Buddhist sutras and imageries—in either a two- or three-
dimensional format—is equivalent to multiplying Buddhist merits and attaining Buddhahood.
10

Apart from paintings and sculptures that themselves contain figural imagery, the medium that
delivers the imagery also plays a critical role in merit-making practice. This is because the
materials for the artistic works sponsored or offered by donors have monetary, cultural, and

8. According to Wu, the silk banner, the “Name Banner,” represents the deceased in the jiu N coffin. The
character jiu means “a (wooden) encasement with the graph jiu O inside it supplying the meaning of
‘forever’ or ‘eternal.’” Wu, “Art in a Ritual Context: Rethinking Mawangdui,” 118 and 122.
9. For how Lady Dai’s body was wrapped and prepared, see Hunan sheng bowuguan BFCóëD,
Changsha Mawangdui yihao Han mu ≠Ç&çEDFGH, figs. 33–34; Wu, “Art in a Ritual Context:
Rethinking Mawangdui,” 119.
10. For example, see Lotus Sutra, T. 262, 58:10b–14b; 9:06a–11a.

137
personal value. Understanding and evaluating the materials’ inherent values and connotations has
hardly been discussed—especially offerings in the form of paintings—as opposed to statues and
relic caskets made out of expensive, precious metals.
Among documents recovered from the Library Cave, lists of the offerings (shishe shu Ç
&Ñ) accepted from devotees are worth reviewing to identify the types of textiles as offerings,
as well as their quantities, intended use, and purposes.
11
Haog examined thirty-eight of these
documents dated to the ninth and tenth centuries, including S. 6829 (806), P. 2583 (816), P.
2863, and D. 162 (836) to name a few, and classified them according to the donor’s name, type
of item, intended use, and purpose of offering. The donors ranged from monks, nuns, laity (both
male and female), and Tibetan monks to political authority figures such as a Tibetan official and
members of the Cao family. Among the most popularly donated objects were bolts of silk, hemp,
clothes, and grains; these items were intended to be used for monastic rituals, sutra recitation,
and the repair and construction of caves, paintings, and sculptures. For example, the verso of P.
2837 itemizes offerings by type, quantity, and intended use, followed by the donor’s wish. One
of the donors, named Li Xiaohu }n', donated “dark red silk in five chi (fei juan wu chi t[
©ƒ)” to be “used for repair and building [an image] (shiru xiu zhaoÇÉ(†).”
12
He wished
for his mother to overcome her sickness and be fully recovered.
From Hao’s list, we learn that donors’ top-ranked wishes were to have peace and well-

11. This genre of text was allegedly written by monks at the time of receiving listed offerings from
donors. P. 2583 and P. 2837, for instance, list the names of donors, the items they offered, the quantities,
and the purpose. The offerings were documented and collected either by one person or more people,
judging from the writing styles. Hao Chunwen Pòû, Tang houqi Wudai Songchu Dunhuang sengni de
shehui shenghuo §Q›`+R‰()S9èTUJV [The social life of monks and nuns at Dunhuang
in the late Tang, Five-Dynasties period, and early Song] (Beijing: Zhongguo shehui kexueyuan, 1998),
242–52.
12. A photograph of this document is available at the Bibliothèque nationale de France’s website:
https://gallica.bnf.fr/ark:/12148/btv1b8304038h/f11.item (accessed August 1, 2019.)

138
being for their families. The next most popular dedication purposes were to commemorate family
members’ deaths and for the deceased to be reborn in the Pure Land, followed by wishing for
fast healing from an illness. The items’ donors specified how they wanted their offerings to be
utilized, such as helping local monastic communities in Dunhuang or making Buddhist images
and spaces including cave and temple structures, paintings, and statues.
13
The cases of offerings
being meant to contribute to Buddhist pictorial art, in particular, constitute half of the ninety-
eight offerings.
The offerings dedicated to producing Buddhist imagery could have been spent on
purchasing relevant materials, such as painting supplies. Alternatively, the offerings themselves
could have been used as painting materials. In the example mentioned earlier, the case of Li
Xiaohu }n', it is possible that the five-chi length of dark red silk might have been used for a
painting’s support material; several examples of long banner-type paintings—such as EO 1137,
Stein Painting 217, EO 3651/1, and Д x 291—were specifically executed on red/vermilion silk.
Five chi is equivalent to around 150.0 cm, which is comparable to EO. 1137 (183.2 cm in height
and 58.0 cm in width).
14

How does a Buddhist scripture reveal the merit of silk? What values of silk, as both a
precious and mundane object, were noted? Let us examine the following Tibetan Buddhist rite
called ‘dod yon sna lnga, which can be translated as “five desire qualities,” “five strands of
desire,” or “five qualities of enjoyment”:
15

The Five Offerings of Sensory Enjoyment [Skt. Pancakāmagunā [sic], Tib. ‘dod
yon sna lnga] represent the most beautiful objects which attract the five senses. As

13. For the complete list of offerings, see Hao, 260.
14. See Appendix 1 for the complete list of “Long Banners” in Chapter 3.
15. Brian Duigan ed., “‘dod yon sna lnga,” Encyclopedia Britannica,
https://www.britannica.com/topic/dod-yon-sna-lnga (accessed August 10, 2019).

139
objects of the most delightful sensory pleasure [silk cloths] are presented as
offerings to the deities or gurus, symbolizing both the desire to please the
enlightened beings and as a gesture of sensual renunciation on the part of the
donor.
16

This rite involves offering five different objects, each representing one of the five senses: a
mirror (sight); a bell (hearing); incense, nutmeg, or a scented flower (smell); sugar or cake
(taste); and a piece of silk cloth (touch).
17
It is believed that the venerated being (a wrathful
tantric deity in this case) and his five senses will be pleased and satisfied by receiving these
objects. The smooth, soft texture of silk is appreciated and emphasized in the script. The five
offerings for the five senses are often represented in Tibetan thangka. Similarly, emphasis on the
offerings’ materialistic characteristics can be applied to the textile cache from Cave 17. To put it
differently, the purpose of offering silk or other textiles in the Dunhuang region can be explained
by their material quality that would satisfy the venerated deities.
Looking at a basic material for weaving silk, we find that silk threads used to be buried
along with a dead body at the ancient cemetery compounds in Astana, near Turfan. There are
also cases in which one can find a hanging painting on silk placed inside of a tomb chamber. The
following section will explore burial practices in Turfan, discuss the concepts that silk material
embodied, and investigate how they were expected to function in the context of funerary
practice. This will help us to understand the historical and regional significance of the material,
which eventually became a preferred material for offerings in funerary practice in Dunhuang.

16. Robert Beer, The Encyclopedia of Tibetan Symbols and Motifs (Boston and New York: Random
House, 1999), 194.
17. Brian Duigan ed., “‘dod yon sna lnga,” Encyclopedia Britannica,
https://www.britannica.com/topic/dod-yon-sna-lnga (accessed August 10, 2019).

140

4.3. Silk Used in a Funeral Practice and the Meaning of Silk Thread
From the ancient tomb complex in Astana near Turfan, over sixty burial inventories were
excavated since the 1900s.
18
They are crucial records because the types of objects listed/selected
by the people to be buried with them show how the afterlife was conceptualized in Turfan in the
sixth and seventh centuries. The inventories should be understood in connection with
accompanying funeral statements in order to glean more contextual information about the buried,
their identities, and their religious ideas. Some tombs yielded these types of inventories on paper
along with actual burials of different types of silk, garments, and other ordinary objects such as
bedsheets, combs, belts, and pillows.
One particular sixth-century tomb, 72TAM170, where one male and two female bodies
were buried, yielded three inventory lists, bolts of various silk and hemp, and everyday objects
that the buried used during their lifetime. I intend to highlight how the materiality of silk threads
was conceptualized to involve wishful thinking. From the inventory belonging to Xiao Zi )*
(d. 543), the first wife of the tomb’s male, Zhang Hong z+ (Buddhist name Xiao Yin ),, d.
562), one rather interesting item was listed: “Silk thread for climbing to heaven–100,009,000
decafeet,” that is, 3,110,279,900 cm of extremely long silk threads.
19
Placed on the final line of
the list, the silk thread is differentiated from other items by its description of its purpose; the
remaining listed items simply indicate the types of objects and their quantities. An identical item

18. Feng Zhao and Le Wang, “Reconciling Excavated Textiles with Contemporary Documentary
Evidence: a Closer Look at the Finds from a Sixth-Century Tomb at Astana,” Journal of the Royal Asiatic
Society 23, no. 02 (2013): 197. Excavations were conducted by various entities including Ōtani Kōzui
(1902), Albert Grünwedel (1902–1903), Albert von Le Coq (1904–7, 1913–4), and Sergey Oldenburg
(1909–10), and Stein (1907 and 1914).
19. “WGXYZZÆóh” Zhao and Wang, “Reconciling Excavated Textiles," 204.

141
was also in the burial inventory of Guang Fei -. (d. 548), the second wife to Zhang Hong,
from the same tomb.
20
This phrase with an inventory of extremely long silk threads is also found
in other funeral statements from the seventh-century tomb 86TAM386.
21

Surprisingly, the silk thread listed in the inventory was not found among the actual buried
objects. In 2013, Zhao Feng and Wang Le compared the inventory with the excavated objects
from the tomb labeled 72TAM170; they observed discrepancies between the excavated objects
and what is described on the burial list.
22
Such a discrepancy is not surprising phenomena
because the non-existing items on the list often represent the buried subject’s desire
conceptually.
What I would like to stress here is that the materiality of a long length of silk thread is
connected to a means of transporting the soul of the subject who is buried with it. Silk thread
from silkworms’ cocoons is very thin yet strong, elastic, and durable. These physical qualities
extend to the functionality of the object, even though it was not present in reality. The inclusion
of the hypothetical object in the funeral texts suggests that silk, at least under this regional
circumstance, had a significance in connecting with the afterlife and transporting the dead’s soul
to heaven. How does this silk thread’s imagined use contribute to other burials in terms of their
meaning and function? To be specific, in terms of silk’s physical advantages over other objects,
is it be legitimate to extend this concept to the cases of painted silk banners from Cave 17? Let

20. “WGXYZZÆóh” (Silk thread for climbing to heaven–100,009,000 decafeet), Zhao and Wang,
204.
21. For the translated text of this funeral statement, see Albert E. Dien, “Turfan Funereal Documents,”
Journal of Chinese Religions 30 (2002): 38–39. There are two more cases with slightly different figures.
Zhao and Wang, 39 and no. 85.
22. This tomb is one of a group of four tombs located in the middle of the northern section of the
cemetery complex. Three human bodies were found here: General Zhang Hong º[ (Buddhist name
Xiao Yin \ı, d. 562); his first wife, of the Jiao ] clan (Buddhist name Xiao Zi \^, d. 543); and his
second wife Guang Fei }_ (d. 548). Zhao and Wang, “Reconciling Excavated Textiles,” 198.

142
us look again at the examples of silk paintings that were uncovered intact from the Astana
cemetery site.
In Innermost Asia, an archaeological report published after his third expedition to Central
Asia (1913–1916), Stein reported some textile findings from Astana and analyzed multiple types
of textiles, among other things, being used as shrouds, face coverings, and hangings. The
hangings in particular attract our attention because of the relative uncertainty of their exact
purpose with regard to burials as compared to other more obviously functional objects, such as
face covers, clothes (shrouds), and body covers. In a tomb labeled Ast. i. 6, Stein observes the
following:
In the south-western corner of the chamber, which was only 4 1/2 feet in height,
there hung fastened to the ceiling by a twig a rag of completely rotten material,
apparently silk. It was interesting as a proof that the rags found clinging to the wall
of tomb i. 1 had been placed there on purpose.
23

Here, Stein conveys that a small number of hanging textiles were discovered intact with their
hanging material undisrupted. He speculates that the hangings would have been installed “on
purpose,” yet does not articulate or suggest exactly what this purpose might have been—
interesting considering that the practice was seemingly common when we look at seventh-
century tombs in Astana region near Turfan. Stein also found a hanging painting in nearly
complete condition in the painting identified as “ix. 2”:
But the most striking thing among the sepulchral deposits was the hanging ix. 2.
054 (Pl. CIX), still in its place on the back wall of the chamber, showing on ivory-
coloured silk the coarsely painted figures of the legendary sovereign Fu-hsi and his
consort with their lower serpentine bodies entwined.
24

23. Aurel Stein, Innermost Asia: Detailed Report of Explorations in Central Asia, Kan-su and Eastern
Īrān 2 (Oxford: The Clarendon Press, 1928), 649.
24. Stein, 664.

143
The painting he found was of the Chinese legendary figures of Fuxi and Nüwa (fig. 72). Stein
was surprised by the fact that not only was the condition of the painting good enough to describe
its pictorial subject, but also that it was discovered in situ at “the back wall of the chamber.”
25

The famous pair is well known for representing China’s traditional cosmic perspective, as they
are believed to be the creators of the universe. The female deity, Nüwa, is on the left with a
compass and the male deity, Fuxi, on the right is holding a ruler. Even though its iconography
has been identified, the reason this specific motif was painted and hung inside of a tomb is still to
be determined.
26

A further investigation on the meaning of Fuxi and Nüwa as part of burial object imagery
and embellishment of a tomb structure is out of the scope of this dissertation.
27
However, the fact
that the hanging painting with (possibly) religious/indigenous figural iconography was kept
inside of the tomb is noteworthy. It provides an interesting example of a confined space in which

25. The tomb, Ast. iv. 2, was discovered with three coffins; two (b, c) were found on a platform in the
north and the other (a) was on the ground. One on the west wall side, “a low pedestal” painted in red, was
placed with pottery jars, dishes, and saucers. Stein, 664.
26. Wu Hung argues that the cosmic figures transform a coffin (a stone box in the case of Lady Dai’s
funeral art) into the universe of the dead. Wu, “Art in a Ritual Context: Rethinking Mawangdui,” 142. Jae
Hee Seung argues that placing the hanging painting was an economic way of presenting the subject
instead of painting the theme on an entire coffin. During the Han and Qin periods, there was a tradition of
painting Fuxi and Nüwa, as well as the sun and moon, on a wooden coffin for decorative and talismanic
function. In other words, in the cemetery complex in Astana, it was rare for the deceased to have a
wooden coffin, which means that an embellishment needed to be somewhere else if ever planned. As a
result, a separate painting was simply installed in the space where the coffin was placed. This argument
sounds persuasive because the tombs in Astana have little or nonvisual decoration on coffins, if any.
However, one pitfall is that even if it was the “efficient” way of obtaining and maintaining the Chinese
tradition of the coffin decorative scheme, it still does not explain why the particular iconography of Fuxi
and Nüwa was chosen for such a mortuary setting. Seung Jae Hee, “T’urŭp’an Asŭt’ana, K’arahoja kobun
ch’ult’o pogŭi yŏwa do tosang yŏn’gu” [Iconography of Fuxi and Nüwa paintings from Astana, Turfan,
and Karahoja], Chungang ashia yŏn’gu 8 (2003): 133–34.
27. Another possible way of using the Fuxi and Nüwa painting in a tomb is to nail it to the ceiling of the
tomb, so that the buried subject can face the two legendary figures. This argument is based on wooden
nails and the holes made around the painting, as exemplified by several paintings from the Otani
Collection at the National Museum of Korea, such as Bon 4027 and Bon 4028. While this argument is
highly probable, another possibility to consider is that the holes were used for fixing the painting support
to a wooden frame during the painting process, as I discussed in Chapter 1.

144
a decorative or functional painting was installed. This Fuxi and Nüwa painting in a tomb setting
is comparable to the Dunhuang painted banners, which might have been installed inside of caves,
based on the holes for inserting wooden pegs that have been found within the caves.
It has been speculated but not confirmed that painted banners might have been installed
in caves temporarily in the event of a ritual or for temporary embellishment. The supporting
evidence that has been suggested includes long wooden sticks and holes for small-sized wooden
pegs, which were found below the ceiling as well as at the edges of walls, such as in Caves 248,
290, and 322.
Cave 322, one of the Early Tang caves, is an extremely rare situation in which the
original banner-hanging fixture remains; its thin wooden sticks are peculiar and rare because
they cross at the center of each wall. The ends of each stick are facing downward, as well as the
one coming from the bottom of the cave and crawling along the edges of the walls (see fig. 36).
Even though the fixtures themselves are impressive, it seems not very probable to me that
hanging paintings were hung on them. This is because the shape of the wooden “hook” cannot
properly hold a painting. A banner would have hung on a dragon-headed pole, as depicted in
multiple wall paintings at the Mogao Caves. The downward but uprising tip of the dragon’s
tongue provides a simple solution for a banner. The function of the wooden fixture is mysterious,
but at least its physical appearance is less likely to be associated with a painted banner.
The holes for wooden pegs or painting fixtures, on the other hand, might offer more a
convincing option than the previous one. A small hole in which a wooden peg was possibly
installed is located on the upper corner of the south wall of Cave 248. Another example is in
Cave 290. On the upper part of the south wall, paintings show a musical band of celestial beings
flying over the roof under which a Buddha is preaching. One can spot an in situ wooden peg at

145
the third figure from the left, who holds an hourglass-shaped drum (fig. 73).
28
When an
embroidered banner was found in front of Caves 125 and 126 in 1965, these holes and wooden
pegs were used as supporting evidence for a plausible way of hanging a banner inside of a
cave.
29
It should be noted that, however, the caves are from the Northern Wei (386–534) and
Northern Zhou (557–581) Dynasties, which makes it difficult to make a direct comparison. The
locations are also ambiguous: the upper corner of Cave 248 would have been suitable for a
decorative banner, such as a five-colored banner, but not for a rectangular painted banner. In case
of Cave 290, the wooden peg is located off-center on the wall. The peculiar remains of the
hanging fixture should be taken not as direct evidence but as a reference.
As seen from these examples, the ways in which painted banners would have been
displayed or hung are puzzling. Despite this, it should not be concluded that paintings were not
hung at all inside of cave structures. This assumption generalizes a variety of different types and
shapes of painted banners and makes them homogenous in terms of their function. In fact, the
issue of the original locations for painted banners is not simple because it relates to questions
regarding the function of both a cave and that of a painted banner. That is to say, the ultimate
function of the Mogao Caves is highly debatable. Arguments regarding the function of the caves
can be divided into two categories: One argues that caves would have been used frequently as
venues for ritual performance and maintained for generations, while the other sees the caves as if
they were tombs, and thus not actively in use as a mortuary space.
30
Depending on how a cave

28. The two examples from Dunhuang yanjiuyuan ()./#, “Xin faxian de Beiwei cixiu” J`aè
&'(b [Newly discovered embroideries from the Northern Wei], Wenwu 2 (1972): 58.
29. Dunhuang yanjiuyuan, 54.
30. This point of view is mainly held by Robert Sharf. See Robert H. Sharf, “Art in the Dark: The Ritual
Context of Buddhist Caves in Western China,” in Art of Merit: Studies in Buddhist Art and Its
Conservation, ed. David Park, Kuenga Wangmo, and Sharon Cather (London: Archetype Publications,
2013), 38–65.

146
was utilized by its devotees and monastic community, its painted banner’s function and
utilization can be revealed; conversely, if a painted banner’s utilization becomes known, the
functionality of that cave in Dunhuang can be revealed.
The practice of hanging a silk painting in a tomb can be compared to the treatment of
painted banners found in the Library Cave, in terms of a similar function of a pictorial hanging
and its construction. The width of the Fuxi and Nüwa painting is about 17 1/2 inches (44 cm),
which is wider than a similarly narrow type of silk banner, such as Stein Painting 120 (see fig.
3). The way the hanging is constructed is worth noting: three pieces of silk were sewn together to
make one pictorial plane, similar to hanging paintings from Dunhuang. The similarity of
construction between the Turfan and Dunhuang examples does not necessarily mean that they
came from the same painting workshop, but it may hint as to why the painting was stored in a
tomb.

4.4. The Origin of a Painted Banner and the Tibetan Influence
Despite the large quantity of painted banners from the Library Cave, inspiration for and
prototypes of the Dunhuang painted banner format are unclear. We know that the painted
banners are early examples of Chinese paintings on textiles, yet their precise origin and what
visual culture was responsible for their configuration, as well as their mounting and hanging
methods, are not yet known. Looking at both central China and the neighboring preexisting and
contemporaneous empires to the west, such as Tibet, provides interesting cases to consider.
Examples from the Library Cave have formats comparable to Chinese traditional hanging
paintings. As seen in Chapter 1, in contrast to a “traditional” mounting method, a painted banner,
especially one with an iconic image (e.g., the frontal view of seated Avalokiteśvara), is finished
with multiple pieces of textile without backing material. At first glance, it can be easily and

147
incorrectly assumed that the painting is pasted on a textile. However, in most cases, banners from
Dunhuang are not backed with additional textiles, and the four edges are covered with multiple
colored, dyed, or embroidered textiles, forming a frame. The availability of textiles for such a
frame would be the most considerable factor in creating such painted offerings as Stein Painting
52 from Chapter 1 shows that recycled textiles were also acceptable as a frame. Recycling
available fabrics was common when framing paintings, as it could reduce the donor’s
responsibility for the production cost.
31

As Chapter 1 argued, the similarity between the Dunhuang painted banners and the
Tibetan paintings, thangkas, suggests that the Tibetan impact on Dunhuang’s visual and religious
culture is worth revisiting. The visual culture of the Tibetan Empire from the eighth to the tenth
century has not been studied as extensively as the eleventh to thirteenth century CE—often
referred to as the “Tibetan Renaissance”—when Tibetan Buddhist practices and the arts are
flourished.
32
However, documents and sutras found in the Library Cave inform us that the
Tibetan culture had not completely vanished by the ninth century CE but rather was still
powerful even after they were defeated and officially expelled by the Han Chinese of the Return
to Righteous Army.
33

31. This was a common practice among painters and writers, as seen from a number of cases in which a
document was secondhand, such as a material of sutra copying, artistic sketch, or official document. For
example, on the verso of Tribute House and Camel (1919,0101,0.77) in the British Museum, there is a
draft of a eulogy of a cave restoration project sponsored by Cao Yuanzhong and his wife, Lady Zhai.
Secondhand textile or paper as writing or painting materials was inevitable, as resources were limited in
this remote city in Dunhuang. Whitfield, The Art of Central China, Pl. 56; Neville Agnew, Marcia Reed,
and Tevvy Ball, eds. Cave Temples of Dunhuang: Buddhist Art on China’s Silk Road (Los Angeles: The
Getty Conservation Institute, 2016), 198–199 and Cat. 7.
32. Carmen Meinert, ed. Transfer of Buddhism Across Central Asian Networks (7th to 13th Centuries)
(Leiden: Brill, 2016), 6.
33. For example, see Matthew T. Kapstein, “New Light in an Old Friend: PT 849 Revisited,” in Tibetan
Buddhist Literature and Praxis: Studies in Its Formative Period, 900-1400, ed. C. Wedermeyer and R.
Davidson (Leiden: Brill, 2006), 9–30.

148
In recent studies, the Tibetan role in the transfer and spread of Buddhist ideas, religious
practices, and visual culture has been acknowledged; it is becoming a major focus in Silk Road
and Dunhuang studies. Specifically, the recent publication in 2016 focuses on the dynamics of
political and religious environmental shifts in Central Asia and China between the sixth and
fourteenth centuries.
34
It finds the middle of the ninth century significant because it was the
period when major dynasties and empires rose and fell: The Tang dynasty collapsed, while the
East Uyghur and Tibetan Empires emerged in Central Asia.
35
The Tibetan occupation period also
falls into this time period, and, indeed, it brought not only social upheaval but also new religious
visual culture.
36
Recent studies focus on the new trends and changes brought to China by the
Tibetans and how their legacy remained and impacted subsequent generations. Art historian Karl
Debreczeny outlines how, during this specific period in Dunhuang, “Tibetan religious and
aesthetic interests began to assert themselves, and new visual models appeared alongside, or
even combined with, Chinese one.”
37
He further argues that the core characteristics of Tibetan
Buddhism, which is a political adaptation of Tibetan Buddhism in the form of emperor worship
(e.g., state protector of Mahakāla), developed from a mixing of and engagement between the
political ideologies of the descendent kingdoms, such as Xixia, Mongolia, and the Ming
Dynasty. Tibetan and Nepalese monks and artisans played a central role in these empires by
transferring and offering the necessary religious and political ideas and practices that would
enforce and ensure the emperor’s authority. Techniques and artistic practices related to Tibetan

34. Meinert, Transfer of Buddhism, 3–6.
35. Meinert, 3–6.
36. Meinert, 3–6.
37. Karl Debreczeny, “Faith and Empire: An Overview,” in Faith and Empire: Art and Politics in Tibetan
Buddhism, ed. Karl Debreczeny (New York: Rubin Museum of Art, 2019), 20.

149
Buddhist visual culture, such as kesi /0 with silk, continued from Tibet through the Tanghut to
the Mongol imperial court.
38

Scholars also focus on the visual transfer of the so-called “Tibetan Himalayan style.”
Exemplified in tenth-century sculptures at the Tabo Monastery and in Stein Painting 103—
argued to be one of the early examples of the style—the Tibetan Himalayan style can best be
characterized by a two-dimensional representation of the body with long arms and straight legs.
39

They argue that the Tibetan Himalayan style provides a possible scenario in which “Central
Tibet could still have served as a mediator in the process of cultural transfer from Central Asia to
Western Tibet” before the Kashimir influenced the Indo-Tibetan style that flourished in West
Tibet after the eleventh century.
40
I would add that the reverse statement could also be true,
especially during the earlier periods of the ninth century, as Stein Painting 103 clearly exhibits
the Tibetan Himalayan style and the actual Tibetan occupation of Dunhuang and their influence
is undeniable.
These approaches are critical to reevaluating the Tibetans’ role in painting practice in the
eighth and ninth centuries. Weighing the direct and indirect impacts of the Tibetan painting
practice during this period would contribute to tracing possible relationships to the early
formation of the painted banners from Dunhuang. I must note, however, this is not an exclusive
source for the Dunhuang painted banner, but rather one plausible inspiration for the format and
composition of the banners.

38. Debreczey, 34.
39. Linda Lojda, Deborah Klimburg-Salter, and Monica Strinu, “The Tibetan Himalayan Style:
Considering the Central Asian Connection,” in Transfer of Buddhism Across Central Asian Networks (7th
to 13th Centuries), ed. Carmen Meinert (Leiden: Brill, 2016), 126; Deborah Klimburg-Salter, “The
Tibetan Himalayan Style: The Art of the Western Domains, 8
th
–11
th
Centuries,” Cultural Flows Across
the Western Himalaya (2015), 436–54.
40. Lojda, Klimburg-Salter, and Strinu, “The Tibetan Himalayan Style,” 140.

150
In the Stein Collection held in the British Museum, London, and the National Museum,
New Delhi, nine painted banners are distinct from the rest of the painted banners from the
Library Cave.
41
These are Stein Paintings 101, 102 (fig. 74), and 103 (see fig. 52) in the British
Museum; and Stein 530, Stein 531, Stein 532, Stein 533, Stein 534, and Stein 535 in the National
Museum, New Delhi.
42
This group of paintings, all of which depict a single standing bodhisattva
shown in frontal posture and positioned on a lotus-flower throne, not only show unique artistic
styles with respect to the rest of the Stein Collection but also distinct materials and techniques.
43

The clumsy representation of feet in each case, for example, emphasizes the flatness of the
figure. Hands are drawn much longer than normal as exemplified in Stein Paintings 101 and 103,
and Stein 535. In contrast to the lack of interest in proper body representation, the painters paid
more attention to embellishments of attire and headdress. The bodhisattvas’ long hair coils and
hangs down onto the figures’ shoulders. They also wear multicolored, multi-patterned dhoti
skirts, leaving their torsos uncovered. As shown in the details of the clothes they are wearing, the

41. Marilyn Rhie mentions there are seven banner paintings of a standing bodhisattva painted in the
Tibetan style: Four are in the British Museum and three are in the National Museum in New Delhi.
However, based on my research there are nine paintings in this group: Six are in the National Museum in
New Delhi and three are in the British Museum. Marylin M. Rhie, “Tibetan Painting Styles, Sources, and
Schools,” in Worlds of Transformation: Tibetan Art of Wisdom and Compassion, ed. Marylin M. Rhie
and Robert A. Thurman (New York: Tibet House, 1999), 72 and no. 2. For original photo plates of Stein,
see Stein, Serindia, Pl. LXXXVII, upper. For the British Museum’s collection see Whitfield, The Art of
Central China, pls. 46–48. For the New Delhi collection see Pratapaditya Pal, The Art of Tibet: The
Meditation and the Rituals (New York: The Asia Society, 1969), 27 and figs. 11–12; Lokesh Chandra and
Nirmala Sharma, Buddhist Paintings of Tun-Huang in the National Museum, New Delhi (New Delhi:
Niyogi Books, 2012), plates 95–98.
42. Pal’s The Art of Tibet indicates that figure 11 is in the National Museum, New Delhi without
indicating any accession number. Based on the index published by Chandra and Sharma, the matching
number must be Stein 532. Confirmation is required. Chandra and Sharma, Buddhist Paintings of Tun-
Huang in the National Museum, New Delhi, 274; Pal, The Art of Tibet, 27, figure 11.
43. Comparable examples of the Tibetan Himalayan style include rock carvings and sculptures at Shey,
Ladakh, and Naupur near Gilgit; for the Five Buddhas at Shey and the standing Buddha at Naupur near
Gilgit, respectively, see Klimburg-Salter, “The Tibetan Himalayan Style,” 438, figure 15.8 and 440,
figure 15.9.

151
artists put more effort into detailing the patterns and cloths folds on the legs than the legs
themselves. In Stein Painting 103, for instance, the bodhisattva is wearing a striped skirt
alternating in red, yellow, blue, and green strips of fabrics with variations in patterns. In each
painting, a red, blue, or green scarf with a floral motif is covering both of the figures’ shoulders
and arms and wraps under the waistline. In terms of iconography, the bodhisattvas all wear three-
peaked crowns adorned only with colorful jewels (not with a miniature seated Buddha). Except
for Stein Painting 101, all the bodhisattvas are holding either a stem of a lotus flower, a sutra on
top of the lotus flower, a vajra, or a sword. Those with lotus flowers can be identified as
Padmapāṇi or Avalokiteśvara, but the Tibetan inscriptions confirm only two of the nine
paintings. Stein 531 is inscribed with a Tibetan word: “ārya,” meaning “noble,” “superior,” or
“noble one”—a common prefix to Avalokiteśvara.
44
Stein Painting 103 also bears a Tibetan
inscription on the left of the reverse of the painting.
45
The transcribed word in shorthand, “ba-ca-
ra-ban-ne,” is an “attempt to reproduce the name Vajrapāṇi.”
46

Even though scholars have agreed that these paintings might have been made after the
mid-ninth century, there are some disagreements about their origins. Some scholars think that
they have Dunhuang origins, particularly during the Tibetan occupation period, but were made

44. Pal, The Art of Tibet, 34. For the definition of “ārya”, see Buswell and Lopez, The Princeton
Dictionary of Buddhism, 64.
45. According to L. D. Barnett, who analyzed Tibetan inscriptions on Stein Painting 103 in Appendix K
of the Serindia publication in 1921, there are two inscriptions: one is over the head and the other one is on
his proper right. He says the first one is “gcen,” meaning “elder brother.” However, during my in-person
study session at the British Museum, I was unable to locate the inscription above the head. The second
inscription is actually placed on the right side of the painting in contrast to what Barnett described. The
plate (Pl. LXXXVII) of the painting in Serindia in volume 4 actually is the back side of the painting
judging from the way the eyes were painted. The pupils are not seen in this photo, as they were done on
the front side of the painting. See L. D. Barnett, “Notes on Tibetan Inscriptions of Buddhist Paintings
from Ch’ien-Fo-Tung, Tun-Huang,” in Serindia 4, 1474.
46. Barnett, “Notes on Tibetan Inscriptions,” 1474. Susan Whitfield reads the inscription as “ba ca ra
pange ne.” Susan Whitfield, The Silk Road: Trade, Travel, War and Faith (London: British Library,
2004), 210.

152
by a nonlocal painter. Others think that they might have come from outside of Dunhuang—that
is, Western Tibet or Khotan.
47
Archaeologist Gerd Gropp first argued that they came from
Khotan, an Iranian Buddhist kingdom.
48
On the other hand, historian Roberto Vitali was the first
scholar to point out that this group of painted banners is comparable with the sculptures of the
Kachu monastery in Central Tibet.
49
In addition, a recent conservation study found the use of
lapis lazuli in a painted banner from the Library Cave. In mural paintings, the same mineral
pigment was normally absent because trade routes were blocked by the Muslim invasion of the
Tarim Basin in the ninth century.
50
Stein 535 from the National Museum, New Delhi, clearly
shows bright, clear blue pigment on the bodhisattva’s crown jewels, the garments over his right
shoulder, and the lotus flower in his left hand (fig. 75). This extensive use of lapis lazuli supports
the theory that the group of paintings was probably produced in the Khotan area and brought to
the Dunhuang area before they were stored in the Library Cave. Even though it is difficult to pin
down the exact location of their original production, it is worth noting that Khotan must have
been benefitted from its geological location close to a large deposit of lapis lazuli in
Badakshan.
51
Thus, Khotan should have been involved in the paintings’ production by either
providing painting materials to Dunhuang—which would mean the paintings were made in
Dunhuang—or making the paintings directly at the local workshop in Khotan.

47. Whitfield, The Art of Central China, 22.
48. He mentions that one wall painting of the Samye Monastery was painted in the Khotanese style
around the ninth century. Gerd Gropp, Archäologische Funde aus Khotan Chinesisch-Ostturkestan: die
Trinkler-Sammlung im Ubersee-Museum, Bremen (Bremen: Röver, 1974), 94.
49. According to M. Rhie, the Kachu Monastery was patronized by Dro Trisumjey (Bro Khrisumje), who
was a successful military leader in the late eighth to the early ninth century. Rhie, “Tibetan Painting
Styles, Sources, and Schools,” 45–46.
50. Brack and Mysak, A Technical Study of Portable Tenth-Century Paintings, 21–26.
51. Brack and Mysak, 26.

153
The close affiliation between Khotan and Dunhuang has been acknowledged by
scholars.
52
One historical commonality between these two regions is that the Tibetan Empire
conquered both of them before the tenth century. Located immediately north of Central Tibet, the
Khotan Kingdom was under the Tibetans’ rule twice: once in the seventh century and again four
or five years after their occupation of Dunhuang (791/792–851).
53
The Tibetan Empire’s
ambitious territory expansion to the north was initiated by Tsenpo Sontsen Gampo (606/617–50)
and became successful during Tsenpo Trisong Detsen’s (r. 755–97) reign. From the seventh
century to the ninth century, Tibetan influence was ubiquitous in these regions of the Silk Road,
as the group of paintings, regardless of their original provenance, clearly shows Tibetan
Himalayan style, such as that seen in Stein Paintings 102 and 103.
One Tibetan thangka of Avalokiteśvara, believed to have been discovered in the Keru
Temple and is now stored in the Yarlung Museum in Tsethang, Tibet, exhibits early Tibetan
painting characteristics—ones that share formal similarities with the painted Dunhuang banners
(fig. 76).
54
The size of the thangka is slightly larger than any painting from the Tibetan

52. For a detailed discussion on this topic, see Rong Xinjiang cJd and Zhu Lishuang :ef, Yutian
yu Dunhuang ÎÏW() [Khotan and Dunhuang] (Lanzhou: Gansu jiaoyu chubanshe, 2013). Sha
Wutian focuses on the image of Khotan King painted in the Dunhuang caves. See Sha Wutian ÇÉÑ,
“Dunhuang shiku Yutian guowang huaxiang yanjiu” ()2AÎÏçÄ–./ [Studies in the image of
the King of Khotan at the Dunhuang caves, XInjiang shifan daxue xuebao JgÔh=iij 27/4
(2006): 22–30. Christoph Anderl also discusses their relationship in the context of “localized Buddhism,”
as well as the emergence of new iconographies. See Anderl, “Linking Khotan and Dūnhuáng, ” 250–301.
53. The first occupation is believed to have taken place between 665 and 670, as Tibet attacked against
countries under Chinese rule in 670 with the Khotanese troops. Tibetan rule ended when the Tang army,
led by Wang Xiaojie ç\k, defeated the Tibetan governor of Khotan in 694. Christopher I. Beckwith,
The Tibetan Empire in Central Asia: A History of the Struggle for Great Power Among Tibetans, Turks,
Arabs, and Chinese During the Early Middle Ages (Princeton, N.J.: Princeton University Press, 1987): 34
and 56. The second conquest was from 791 (or 792)–851, which overlaps with their rule over Shazhou Ç
l and the Dunhuang area from 787–848. For the Tibetan military leadership and their long-term
occupation of the Western Regions, see Beckwith, The Tibetan Empire, 155–172, esp., 155 and 171.
54. Lojda, Klimburg-Salter, and Strinu identify it as Mañjuśrī. Lojda, Klimburg-Salter, and Strinu, “The
Tibetan Himalayan Style,” 141, figure 4.9.

154
Himalayan group of banners from the Library Cave. The painting is 77.5 cm long by 23.5 cm
wide, and it is painted with color on hemp. Similarities with the Dunhuang Tibetan Himalayan
painting group include the bodhisattva’s standing, frontal posture and his holding a blue lotus
flower. The oblong painting is clearly divided into two parts: The upper part belongs to the
standing bodhisattva and the bottom part is devoted to a venerating monk and an attendant
figure, both in kneeling positions. The different scale of the monk (larger) and the attendant
(smaller) in accordance with their social statuses was also a common visual principle practiced in
the Dunhuang paintings.
In particular, the Keru Avalokiteśvara thangka displays one of primary characteristics of
the Dunhuang painted banners: the representation of donor figures at the bottom of the
painting.
55
The current condition of Stein Painting 103 suggests that it could also have had donor
figure(s) at the bottom. Because this particular painting was made by cutting a bolt of silk
parallel to the width, the height of the painting (55 cm) corresponds to the width of a bolt of silk.
In the bottom section of Stein Painting 103, one can observe with the naked eye double stitches
to attach a separate textile (see fig. 52). The upper stitches with white threads and the bottom
stitches with blue/gray threads are indications of holding extra silk onto the main silk support.
The additional silk portion is now ripped and is no longer in situ, but part of the sides’ hems are
extended, which means these parts could have covered the edges of the attached textile where
donor figures would have been painted.
Dating of the Keru Avalokiteśvara thangka has not been settled among researchers.
Lojda, Klimburg-Salter, and Strinu argue that it was painted during the early eleventh century,
whereas Wang Ruilei thinks that it is a much earlier example from the ninth century Tibetan

55. It is also said that the kneeling donors are “clearly depicted in the Western Tibetan mode and close to
the tenth century style of Tabo.” Lojda, Klimburg-Salter, and Strinu, 140.

155
Empire.
56
The former date theory is based on the inscription written in Tibetan cursive script on
the backside of the painting. It contains consecration formulas in Sanskrit verses, as writing an
inscription of verses on the back can be found in several other early Tibetan thangkas.
57

Inscribing names of Buddha and bodhisattvas as well as scriptural verses or devotional
writings, on the backside of a painting is not conventional in traditional Chinese painting
practices. Stein Painting 103 also has the bodhisattva’s name inscribed in Tibetan on the back.
What makes the Keru thangka painting different from the group of Tibetan Himalayan
paintings from Dunhuang is the use of a mandorla that surrounds the entire body. Instead of
having a canopy or depiction of a painting cover, the foliage stemmed from the lotus base on the
bottom forms a canopy over the bodhisattva. The corn-shape of his headdress is also distinct
from the three-peak crown. Judging from the delicate lines of the lotus stems and the better
proportioned bodies, the artistic skill and the understanding of the naturalistic body
representation of both divine and secular figures are achieved to a higher degree. Carefully
depicted floating floral motifs fill the space against the navy background.
This painting is an important piece in terms of its transitional components, as it can
connect the western Himalayas or West Tibet and Central Asia, including the northwestern
region of China. Even though its provenance has not been identified, the tall triangular shape of
the crown suggests strong affiliation to the Central Asian paintings.
58
Moreover, there is a strong
affiliation with Central Tibet’s political authorities and noble clan, specifically the Dro clan,
whose members possibly founded the West Tibet Kingdom. The Dro clan was in charge of the

56. Wang Ruilei, e-mail message to author, March 2, 2020.
57. Lojda, Klimburg-Salter, and Strinu, “The Tibetan Himalayan Style,” 140.
58. Keru is one possible provenance. Lojda, Klimburg-Salter, and Strinu, 140 and no. 39.

156
Dunhuang area and served as administrative governors during the Tibetan occupation period.
59

In addition to the style of the standing bodhisattva figures in this group of paintings, one
element I would like to point out is the representation of a painting cover on the top of the
paintings. Except Stein Painting 103 of Vajrapāni, all the other paintings in the group of the
Tibetan-Himalaya style paintings depict a patterned textile folded and dropping downward. The
pictorial representation deceives the viewer’s perception; it appears as if a real cloth is attached
on the top of the painting of Stein Painting 102 (see fig. 74). Attaching a painting cover is one of
the common features of Tibetan thangka paintings, indicating their close relationship to Tibet.
Even though Stein Painting 103 seems not to have the trompe l’oeil representation of the cover
textile, I would argue that the floral-patterned strip above the bodhisattva’s head represents a
physical object similar to a painting cover. In other words, it is an actual representation of a
canopy based on the four bells hanging down from the cloth. There is a pair of bells on the front,
and one can observe that bells with handles are attached with floral-shaped knobs under the
patterned strip. Another scenario is that the floral strip was originally meant to represent a
painting cover but a painter changed it to a canopy by adding bells. Stein 531 and Stein 533, on
the other hand, provide a mixed representation of painting covers and small ornaments (e.g.,
bells or miniature banners).
These paintings were painted on silk from which the roll was cut parallel to its width. In
other words, a roll of silk was unfolded like a handscroll and cut vertically, so that the width of
the roll becomes the height of the painting. This method of preparing support material is one of
the major differences from Chinese examples. By doing so, all nine paintings have similar
heights from 44–55 cm with almost identical widths of 14–15 cm. Common iconographical

59. Lojda, Klimburg-Salter, and Strinu, 140–142 and no. 41.

157
features—three-pronged crowns with jewels, figures wearing similarly patterned garments,
stylistic affiliations in body figure representation, and standing postures painted on the same type
of material (gray silk)—suggest that these paintings were made intentionally similar as opposed
to being a pantheon of various Buddhist bodhisattvas. I would argue that the similarity of the ten
bodhisattva paintings indicates that they were not supposed to be grouped or used all together for
a certain ritual purpose. It would be more reasonable to understand them as having been
produced at the same time by the same hand or workshop to be distributed or sold as individual
offerings.
The practice of having the pictorial representation of a textile element, such as a painting
cover and a canopy, also echoes a unique practice of decorating a sacred space in Tibet.
Monasteries in Western Tibet, such as the Tabo Monastery of the Spiti Valley and the Alchi
Monastery in the Leh District, exhibit an interesting architectural decorative scheme. They
include a trompe l’oeil representation of a strip of textile as a ceiling painting or an actual textile
glued and pasted on the ceiling. At the Tabo Monastery, both types are found. In the Assembly
Hall of the Main Temple, the mixed use of a painted textile and a real cloth pasted on a beam are
representative of the first example (figs. 77–78). The underlying idea, according to Klimburg-
Salter, is that the entire ceiling was considered as a “canopy,” as the top portion of the
surrounding walls were painted in the shape of valance.
60
The second ceiling painting type is
found at the Alchi Monastery, and also exhibits the idea of adorning a sacred space by creating
the representation of a luxury textile on the ceiling.
61
Instead of attaching an actual textile, the

60. Deborah E. Klimburg-Salter and Christian Luczanits, Tabo: A Lamp for the Kingdom (New York:
Thames & Hudson, 1997), 173.
61. The Alchi Monastery complex at Ladakh is believed to have been built by the great translator Rin
chen bzang po (958–1055) in Western Himalaya, but the inscription in ‘du khang revealed that it was
founded by a monk named Skal ldan shes rab. As the monk’s biography is unknown, the dating of the
ceiling painting has not been confirmed but only speculated, as “the paintings cannot be dated any earlier
than the beginning of the 13th century.” Christiane Papa-Kalantari, “The Ceiling Paintings of the Alchi

158
artist at the Alchi Monastery attempted to duplicate a specific type of textile modelling, one after
a clamp-resist-dyed, roundel-patterned cloth, which is a luxury textile in Central Asia.
62
The
textile-patterned work was painted directly on wooden panels in between beams of the Assembly
Hall at the Alchi Monastery (fig. 77). As the textile painting was placed in a location with the
shape of an oblong rectangle, the painting naturally follows the shape and resembles an unrolled
textile.
Interestingly, there are a number of silk textiles found in Cave 17 which share similar and
comparable motifs and dyeing techniques as the examples from Alchi and Tabo. Concentrating
on the clamp-resist-dyeing technique, a floral-roundel motif on silk is a popular motif among the
findings.
63
Of the examples of the clamp-resist-dyeing technique that are known, including
fragments at the British Museum and the Victoria and Albert Museum, there are at least six cases
with this motif: L:S. 592 Ch.xxii.0036 (18 cm in length and 37.5 cm in width), L:S. 591
Ch.00305 (45.5 cm in length and 12.2 cm in width), L:S. 682 Ch.xxii.0036 (31 cm in length and

gsum brtsegs: Problems of Style,” in Buddhist Art and Tibetan Patronage, ed. Deborah Klimburg-Salter
and Eva Alinger (Leiden and Boston: Brill, 2002), 86–87. The ceiling paintings at Tabo, dated to the tenth
century, on the other hand, are much earlier than those at Alchi.
62. The practice of attaching an actual textile was not always followed in Western Tibet. Ibid., 90 and no.
12. Christiane Papa-Kalantari argues that it was the clamp-resist dyeing technique the artisans tried to
reproduce based on her observation on how the painted textile motif mimics the outcome of the dyeing
process. See Ibid., 94. As for the beaded roundel motif, by the clamp-resist dyeing technique was
originated from Sassanian Persia. It, however, became a fashionable, popular, and desirable object, which
led China and Tibet to copy the pattern and spread it throughout the Asia during the sixth to eighth
centuries. This specific type of textile eventually gained religious and political authority, and they
appeared on ceiling paintings and official and royals’ costumes in Tibetan Buddhist monasteries, even
after its popularity drastically decreased in China by the eighth century, except for the main production
site in Chengdu of Sichuan, where an established trade route between Lhasa and Chengdu was stable
regardless of Song’s strict foreign policy. Valrae Reynolds, “Luxury Textiles in Tibet,” in Tibetan Art:
Towards a Definition of Style, ed. Jane C. Singer and Philip Denwood (London: Laurence King, 1997),
121–23.
63. For the chronology of the specific pattern and variations, see Wang Le Zhao Feng, Xu Zheng, “A
Typological and Chronological Study of Dunhuang Textiles,” in Textiles from Dunhuang in UK
Collections, ed. Zhao Feng (Shanghai: Donghua University Press, 2007), 23–25.

159
53 cm in width), L:S. 547 Ch.xxii.0036 (a,b,c), L:S. 552 Ch.00358 (banner part, as panel), and
L:S. 558 Ch.xxii.0036. If we include examples that have a triangular headpiece and a strip of
sutra cover, the number increases.
64

The incorporation of textile (or a pictorial representation of textile) as a compositional
element in a painted banner can be understood by recognizing the extensive use of textiles in
Tibetan Buddhist ritual settings and in Buddhist communities throughout Central Asia, including
Buddhist architecture. Besides the direct stylistic affiliations between the early Tibetan Empire’s
thangka and the group of the Tibetan Himalayan painted banners from Cave 17, their impact on
other paintings in different style groups is worth noting. The most interesting phenomenon is the
way an inscription was written. Traditional writing in China is done right to left. However, there
are a few examples that exhibit an inscription written left to right.
65
Hypothetical explanations
have been proposed, but it should be noted that Tibetan writing is done in the same direction as
the latter: from left to right. Examples of paintings with nonconventional writing direction
meaning left to right include Two Bodhisattvas (mid-eighth century, Stein Painting 3), Paradise
of Bhaisajyaguru (836, Stein Painting 32), Tejaprabha Buddha and the Five Planets (897, Stein
Painting 31), Eleven-Head Avalokiteśvara (959, MG. 25486), and Kṣitigarbha (963, Stein
Painting 19). The most interesting example in this context is Paradise of Bhaisajyaguru, which is
now in the Stein Collection at the British Museum (fig. 79). This painting exhibits the painting
styles of the Tibetan Himalayan and the Tang Chinese, and also bears an inscription at the center

64. In the State Museum of Hermitage, a number of examples contain the clamp-resist-dye floral pattern
silk used as a banner head and a border for a banner, such as Дx42, Дx49, and Дx169. Zhao, Textiles from
Dunhuang in Russian Collections, 58–75.
65. In his study on the writing practice in Dunhuang, Galambos argues that this practice of writing from
left to right seen from the paintings from Cave 17 “was directly connected to Guiyijun culture,” and also
“a local phenomenon.” Galambos, Dunhuang Manuscript Culture, 189.

160
of the painting written in two languages: Chinese and Tibetan.
66
The two inscriptions contain
almost identical content, and the writing is from left to right. It has been suggested that a
bilingual painter, Pelyang (Dpal dbyangs, ?–?), wrote both inscriptions as he painted the entire
painting.
67
With this example in mind, the Tibetan writing direction being left to right implies
how extensively Tibetan influence extended to the visual culture in Dunhuang.
Along the same line as writing practice, Tibetan thangka paintings often convey
inscriptions on their reverse. The practice is associated with the consecration ritual (Skr.
Abhiṣeka), which transforms a mundane object into an auspicious divine being by, for example,
conducting the “Opening/Dotting of the Eyes” ritual as soon as a painting is completed.
68
The
artist might inscribe sacred syllables, such as the six-syllabled mantra of “Oṃ Maṇi Padme
Hūṃ,” on different parts of the images’ bodies on the verso of a painting.
69
After being
consecrated with a ceremony, the images in a painting represent and become a living presence,
which means they are then qualified to be displayed and venerated by devotees.
70
During the
Tibetan consecration ritual, for example, the three fundamental seed syllables—oṃ, āh, and

66. Heather Stoddard and Michelle Wang call the style of Paradise of Bhaisajyaguru “Tibeto-Nepalese”
style and “Nepalese-derived Tibetan” style, respectively; Heather Stoddard, Early Sino-Tibetan Art
(Bangkok: Orchid Press, 2008), 11; Michelle C Wang, Maṇḍalas in the Making: The Visual Culture of
Esoteric Buddhism at Dunhuang (Leiden: Brill, 2018), 99–100.
67. Stoddard argues that the painting was made by two painters—one Chinese and one Tibetan—based on
the painting style. Whitfield and Wang, however, are against this analysis and suggest that one painter
was in charge of this painting. Their argument is based on the fact that the painting’s brushstrokes and
underlines, which are consistent throughout the painting, and it merely has two different “modes” of style
that the painter mastered.
68. Shaftel, “Notes on the Technique of Tibetan Thangkas,” 101.
69. On the other hand, Pelliot tibétain 389, a mandala of Vairocana with 33 deities, shows only the
syllable om inscribed on every deity. Christian Luczanits, “Ritual, Instruction and Experiment: Esoteric
Drawings from Dunhuang,” in The Art of Central Asia and the Indian Subcontinent in Cross Culture
Perspective, ed. Anupa Pande and Mandira Sharma (New Delhi: National Museum Institute-Aryan Books
International, 2009), 144–45.
70. Robert E. Buswell, Encyclopedia of Buddhism (2004), 178.

161
hūṃ—are inscribed on the head, throat, and heart, respectively,
71
followed by the dotting of the
eye as a climax of the ceremony.
72
Cases of a practice of writing syllables on the backside of a
painting have not yet been found among the painted banners from Cave 17, except on a few
mandalas, such as Pelliot tibétain 389, where the syllable oṃ was inscribed on the chest of every
deity.
I intend to focus on paintings that have inscriptions and writings on their versos. The
name of the painted image of the standing Vajrapāni seen in Stein Painting 103 is written on the
back of the work. Another case is the previous example of the Keru thangka, which has Sanskrit
consecration ritual verses are elaborately written on the reverse (see fig. 76). The final example
given here is Stein Painting 14, dated to 910 CE, which features a standing Avalokiteśvara
flanked by two smaller human figures—a deceased nun named Yanhui E from the Universal
Light Temple (Puguang si 1-N) on the left side and her younger brother, Zhang Youcheng z
2¿, on the right.
73
This painting notably has three inscriptions on the front and two additional
inscriptions on the back; the latter are written on paper and attached to the backside, in a rare
example of using a separate paper solely for the inscriptions.
74
The inscription on the right-hand
corner is written within a cartouche with a green background, and it is considered the principal

71. Each represents the Buddha’s body, speech and mind. Yael Bentor, “The Horseback Consecration
Ritual,” in Religions of Tibet in Practice, ed. Donald S. Lopez (Princeton: Princeton University Press,
1997), 236–37.
72. This statement is based on the Horseback Consecration Ritual manual, translated by Bentor. For the
process and the theological concepts of the ritual, see Bentor, 234–54.
73. The vivid and fresh application of pigments and careful outlines make the painting conspicuous
among other painted banners.
74. The painting does not retain the original mounting, which means that the attached inscriptions on
paper were repositioned when the painting was newly mounted in a frame during the 1920s. When the
current condition of the painting is compared with the reproduced image in Serindia, one can notice that
the original border parts on both sides, which were painted with brown, are removed. See Stein, Serindia,
plate LXIX. Roderick Whitfield thankfully showed me the back side of the painting during the viewing
session.

162
inscription among others in the painting. The three-line text starts from the left—another
example of the reversed sequence of writing—and praises the merciful “Bodhisattva
Avalokiteśvara of Savior from Suffering” and wishes that the Empire has peace and that the
deceased nun and her parents may be reborn in the Pure Land.
75
Similar to the Keru thangka, the
inscription on the back contains, after a dedicatory remark, verses praising the bodhisattva
painted on the front. Considering the date of the painting—910 CE, about fifty years after the
Tibetan control over the Dunhuang area—their impact was not completely vanished in
Dunhuang’s visual culture.

4.5. Funerary Practice in Tibet and Dunhuang and the Soul-Guiding Bodhisattva
As briefly discussed in Chapter 2, specific Buddhist themes and images were exclusive to
a particular artistic material. The Yinlu Bodhisattva or Soul/Road-guiding Bodhisattva (or
Bodhisattva Who Leads the Way) painting is a representative example of an imagery exclusively
seen on the painted banner format.
76
One way to understand this phenomenon is to consider the
iconography’s specific function on the textile surface rather than in the mural painting. For
painting, a textile’s physical qualities are preferred over cave walls. In other words, textiles’
superior flexibility, susceptibility, sensibility, mobility, and portability are required qualities to
fulfill certain purposes of the painting. In contrast, in the case of a wall painting, once painted,
the work cannot be moved or animated and cannot respond to any environmental stimuli. Simply

75. Whitfield, The Art of Central China, Pl.7; Waley, A Catalogue of Paintings Recovered from Tun-
huang, 26. It should be noted that the name of the brother is written differently: Zhang Youcheng (ºm
o) on the front and Zhang Youcheng (ºno) on the back. It suggests that it was not the donor himself
but an inscriber who actually wrote the inscription. For the transliteration of the inscriptions, see Xu,
Zhang, and Chen, Dunhuang Mogao Ku ti ji hui bian, 419–20.
76. Another name is Laiying ~o Bodhisattva painting, which is more commonly used in the Goryeo
period in Korea and the Kamakura period in Japan.

163
put, the physical qualities of a medium matter for effective use of certain iconographies and
subjects.
In this context, the Yinlu Bodhisattva painting is an example of how materialistic
qualities serve the depicted icon’s functionality in a ritual setting. To understand its utilization in
a ritual setting, a brief survey of the iconography is required. The Yinlu Bodhisattva does not
originate from India; it is a Chinese creation of a psychopomp in relation to Amitābha and
Kṣitigarbha in the afterlife.
77
As there is no scriptural base for this figure—meaning no backing
of a scripture or theological written element for the name of the bodhisattva—it has been
regarded as an apocryphal Buddhist deity invented in China.
The iconography of the bodhisattva certainly started developing from the late Tang
period to the Five Dynasties. The example from Cave 205 represents the prototype of the
iconography or at least the earlier type before a typical Yinlu Bodhisattva iconography was
established.
78
On the northern part of the south wall, the three-quarter-view face of this standing
bodhisattva is looking downward at a female devotee and one attendant (fig. 80).
79
Without a
cloud motif in this pictorial plane, the vacant background creates a transcendental atmosphere
between the devotees and the bodhisattva. The string of beads handed to the devotee—or was

77. Zhiru argues that the Yinlu Bodhisattva was invented at the time when a close connection between
death and Dizang (Kṣitigarbha) was established in China and became widespread during the ninth to tenth
centuries. After the tenth century, popularity of the Yinlu Bodhisattva declined owing to the fact that
Dizang and the Ten Kings of Hell were consolidated and that Dizang replaced Yinlu’s task of guiding a
soul. Zhiru Ng, The Making of a Savior Bodhisattva: Dizang in Medieval China (Honolulu: University of
Hawai’i Press, 2007), 159–60.
78. This bodhisattva is recognized/categorized not as Yinlu Bodhisattva but simply as Avalokiteśvara
Bodhisattva in Dunhuang shiku neirong zonglu. See Dunhuang yanjiuyuan, Dunhuang shiku neirong
zonglu, 82.
79. This combination of the bodhisattva and the devotees is also depicted on the other end of the same
wall (southern part of the west wall). The pair of bodhisattva paintings with the beneficiaries on the west
wall is placed respectively under the Samantabhadra and Mañjuśirī paintings painted later during the
Middle Tang period.

164
offered by her?—is not a conventional representation in later Yinlu Bodhisattva paintings, but it
promotes close contact between the sentient being and the laity. In other words, a one-on-one
relationship is emphasized by their eye contact and through the act of exchanging an object.
Normally, the donor figures, especially those who sponsored the construction of a cave-temple of
the Mogao Caves, are presented in a separate section below a main sutra or iconic Buddhist
painting on the wall. The inclusion of the donor figures within the main picture frame, together
with the bodhisattva, indeed requires special attention in terms of its exclusive development as a
portable painting format because this inclusion would have led devotees to interact with the
bodhisattva in a direct and personal way.
Qualities that have been added to the High Tang’s prototype of the Yinlu Bodhisattva, as
seen in Cave 205 and finally established as the iconography of the “Yinlu Bodhisattva painting,”
include a swaying banner being held by the bodhisattva, a moving cloud motif, and pavilion(s)
floating at the top of the painting. The nature of the Yinlu Bodhisattva in relation to his
identification is ambiguous, as supporting textual sources have yet to be found. Discussion
regarding its source and identification has been conducted mainly by Japanese scholars but
whether the bodhisattva is a manifestation of Avalokiteśvara or a completely different type of
bodhisattva has not been finalized.
80
Although it cannot be denied that he is a subordinate figure
to Amitābha Buddha—as he has the small Buddha figure (of Amitābha) in his crown in a few
Yinlu paintings—it does not necessarily mean that he should be identified as Avalokiteśvara.
The banner he is holding in his hand is called “road-guiding banner” (yinlu fen∞±V),
which is usually depicted with a color-blocked banner without any Buddhist imagery. The

80. Tsukamoto Zenryū, “Hikiji Bosatsu shinkō ni tsuite” pY≈∆Ïq•rß® [A study of Yinlu
Bodhisattva belief], Tōhō gakuhō s–ij 1 (1931); Matsumoto Eiichi tuD, “Tonko-ga no
kenkyu” ()pI./ [A study of Dunhuang wall paintings], (Tokyo: Toho bunka gakuin, 1937), 361–
67.

165
banner is hung on the tip of the dragon-mouth-shaped staff that the bodhisattva is holding in his
hand.
81
In his other hand, the attribute can vary from nothing to a handheld incense burner to a
willow branch.
Regarding the main attribute of the banner, Li Ling argues that the banner replaced a
lotus flower once held by Padmapāṇi. According to Li, if Avalokiteśvara is particularly
associated with “saving/rescuing” from hardships and obstacles, he is then called Padmapāṇi,
who was told by the Buddha of Immeasurable Light (Wuliangguang fo ≤≥-∫ ) to hold the
lotus flower forever and continue doing the beneficial things for all beings.
82
She further argues
that the Yinlu Bodhisattva paintings from the Library Cave are parallel to the Tibetan Padmapāṇi
in terms of their religious function of rescuing all sentient beings including human beings.
83
As a
result, Padmapāṇi’s main attribute, which is a lotus flower, was replaced by a life banner. Her
examples of Tibetan Avalokiteśvara with a lotus flower do not remain with a specific one but
rather cover multiple types of Cakravarti-cintāmaṇi-Avalokiteśvara (Ruyilun Guanyin 345_
a), Gandharāja (Xiangwang Guanyin ˙F_a), and Vajradharma (Jingangfa Guanyin µ∂6
_a). They are the manifestations of Avalokiteśvara whose vows are rescuing all sentient
beings who fall into the Six Destinies (liudao 14) sent as delegates of the Wuliangguang
Buddha.
84
While the shared function designated to the Padmapāṇis and Yinlu Bodhisattvas might
sound overly general, her comparison between the two is worth noting when pondering the

81. Stein Painting 47 (1919,0101,0.47) is an exception, where the Yinlu Bodhisattva is holding a lotus
flower’s stem, and the head of the yinlu banner is hung under the lotus leaf.
82. Li Ling kv, ““Yinlu pusa” yu “lianhua shou”: Han cang chi lianhua guanyin xiang bijiao ““pY≈
w”î“x√x”: y{zx√{C–|}” [“The soul-guiding bodhisattva” and “Padmapāṇi”], Xizang
yanjiu F{./ 3 (2006): 60.
83. Li, 59–60.
84. Li, 60.

166
popularity in Dunhuang of the Avalokiteśvara with a lotus flower. More than half of the group of
Tibetan Himalayan-style paintings discussed earlier are bodhisattvas holding a lotus flower.
85

Even though the iconographic transformation from the lotus flower to the banner requires further
investigations and the origin of the Yinlu Bodhisattva is still puzzling, the argument becomes
valid in terms of their shared religious functionality.
The Yinlu Bodhisattva’s posture is worth noting. The extant examples of the Yinlu pusa
paintings all show a three-quarter view of his face, as if he is looking back at the beneficiary
figures. He is standing in his signature posture, but his motion is emphasized by the addition of a
traveling cloud motif under his feet that is not static. The representation of the moving cloud
motif symbolizes “transformational creativity,” which emphasizes him as a transitional and
dynamic being.
86
Emphasizing the walking motion is a key to understanding the function of the
painting in relation to the choice of the physical application, which I will review later in this
chapter. In addition, most Yinlu Bodhisattva paintings feature heavenly pavilions represented on
the top of the paintings. The exaggerated descending cloud representation—especially in Stein
Painting 47— certainly implies that the Yinlu Bodhisattva is coming down from the pavilions.
87

Lastly, the beneficiary figures are depicted in a smaller scale, sometimes as a single or a familial
group in the right-hand corner of the painting.

85. Exceptions are Stein Painting 101, Stein 532, Stein 533, and Stein 534.
86. Victor H. Mair, “The Origins of an Iconographical Form of the Pilgrim Hsüan-tsang,” Tang Studies 4
(1986): 30.
87. Stein Painting 46 has rather a peculiar representation of the heavenly pavilion motifs at the top. The
top area is divided into three realms by three horizontal thick red lines. Three, two, and again three
pavilions are placed in each of the realms respectively. In contrast, Stein Painting 47 exhibits a three-
/two-storied building in the midst of a cloud. The building has a unique structure because, at the center, it
seems like a three-storied one, but at the end point of the complex, it comprises a two (in lieu of a three-)
storied structure. The 3-2-3 formation, which is seen in Stein painting 46, was something that the creator
of Stein Painting 47 aimed at reproducing. MG 17657 also depicts the complex, which showcases three
independent buildings: two of two-story buildings and one three-story structure.

167
Returning to the discussion of the popularity of such specific pictorial topics on textile
materials versus wall paintings, the extant number of the paintings corroborates such a practice.
The Yinlu Bodhisattva painting, which is usually depicted as an independent icon in a hanging
painting format, has eight examples: two in the Stein Collection at the British Museum and six in
the Pelliot Collection at Musée Guimet.
88
This number is significant when compared to the same
iconography on wall paintings. Besides the instance of the paired bodhisattva paintings from
Cave 205, which was, I argue, a prototype of Yinlu Bodhisattva iconography, only two cases are
known today according to Wang Ming: Cave 176 at Mogao Caves and Cave 16 at Yulin Caves.
89

These cases, however, are not parallel to the typical representation of the bodhisattva on textile

88. In his entry for The Bodhisattva Yinlu,“who shows the way” (EO1133), Jean-Pierre Drège identified
that the total number of Soul-Guiding Bodhisattva paintings is six: two in the Stein Collection and four in
the Pelliot Collection. However, in his recent study by Wang Ming, he included the bodhisattva holding a
soul-guiding flag in his hand without the representation of the beneficiary who is supposed to welcome
the bodhisattva being presented close to the bodhisattva or a descending cloud representation on which
the bodhisattva is normally standing. I also agree with Wang about including the bodhisattva with the flag
identified as the Yinlu Bodhisattva. See Giès, Les arts de l’Asie centrale: La collection Paul Pelliot du
musée national des arts asiatiques - Guimet, Pl. 65; Wang Ming çK, “Pusa yinlu: Tang Song shiqi
sangzang yishi zhong de yinhun fan” ≈∆pY:§R‹›~åäÜèpÄ1 [Road-guiding
Bodhisattva: Soul-guiding flags in funeral rituals in the Tang and Song dynasties], Dunhuang yanjiu ()
./ 1 (2014): 40–41 and Table 1. The eight examples are Stein Paintings 46 (1919,0101,0.46) and 47
(1919,0101,0.47) in the British Museum; MG 17657, MG 17697, MG 22795, MG 22796, EO 1133, and
EO 1398 in the Musée Guimet. In the other format, the Yinlu Bodhisattva also appears as a subsidiary
figure to the main iconic Kṣitigarbha painting (MG 17662), and the sutra painting in the Gansu Provincial
Museum, titled Sutra Painting on Deep Indebtedness to One’s Father and Mother (Fumu enzhong
jingbian ÅÇÉÑ:H), has the bodhisattva at the bottom-left section of the painting with the donor
figures on the right. In addition, a paper cut format of the bodhisattva also exists in the Pelliot Collection
(P.4517[1]). Wang adds a black horse rider wearing a black-striped garment and holding a black banner to
his list of the Yinlu Bodhisattva paintings from Dunhuang and Yulin. This figure is depicted at the
beginning of the Sutra of Ten Kings (Foshuo shiwang jing 3Çdç:) (Or.8210/S.3961) from the Stein
Collection. Although he is holding a similar banner to those of other Yinlu Bodhisattvas, the iconography
of the figure does not comply either with an iconic or with narrative representation of the Yinlu
Bodhisattva. In other words, the male figure on the horse wears an official’s garment and hat in black.
Yinlu Bodhisattva is represented not with a horse but with a cloud motif. Finally, the accompanying
inscription does not identify him as a Yinlu Bodhisattva. The figure should be identified as a “messenger
(shi 8)” who receives an order by King Yama to examine the dead’s sins and merits, as described and
written in the accompanying inscription. For the list of the Yinlu Bodhisattva paintings, see Wang, “Pusa
yinlu,” 40–41.
89. Wang, 41.

168
material as an independent subject being venerated.
From a broader perspective, a Yinlu Bodhisattva painting is part of Western Pure Land
Buddhism and a genre of the Descending Amitābha painting (Amituofo laiying tu 7æ8∫9:
2), which underwent stylized iconographic development during the thirteenth century in East
Asian countries and kingdoms such as the Xixia and Southern Song in China, the Goryeo in
Korea, and the Kamakura in Japan. In other words, rather than a solo representation of the Yinlu
Bodhisattva figure—as seen in earlier examples from Dunhuang—the later development expands
the band of guiding saviors, including Amitābha, Avalokiteśvara, and Kṣitigarbha, either in a
seated or standing position. The descending moment of a deity to a person on his or her deathbed
becomes even more dramatic with the depiction of an accompanying musical band that
celebrates and glorifies the Buddha coming from his Western Pure Land.
The observation I make here is that the exclusivity of the Yinlu Bodhisattva to being
depicted on silk might have been derived from a deathbed ritual that included a banner.
According to Daoxuan 4; (596–667), a seventh-century Chinese vinaya master who played a
pivotal role in establishing deathbed rituals, a Buddha imagery painted on a banner is
recommended in order to manage when one (usually a monk) is facing the “moment of death.”
90

In a chapter titled “Attending to the Sick and Sending off the Dead” (Zhanbing songzhong <=
>?) in his vinaya commentary, Commentary on monastic practices based on the
Dharmaguptaka vinaya (Sifenlü shanfan buque xingshi chao/@ABCDEãÃF, T. 804),
Daoxuan presents an example from an Indian monastery, the Jetavana Monastery, and illustrates
the ways in which a deathbed ritual for a dying monk was prepared as well as the types of

90. Koichi Shinohara, “The Moment of Death in Daoxuan’s Vinaya Commentary,” in The Buddhist
Dead: Practices, Discourses, Representations, ed. Bryan J. Cuevas and Jacqueline I. Stone (Honolulu:
University of Hawai’i Press, 2007), 105.

169
implements used to facilitate the ritual. This passage from the chapter describes the specific
space and direction one should face, detailed meditation topics one should contemplate during
the ritual, and how to properly use a set of a painting and sculpture:
In the northwest corner of the Jetavana [Monastery], at the place where the sun sets,
is the Hall of Impermanence. The sick are placed there. Sentient beings are affected
by greed, and when [dying monks] see their robes, bowls, and other implements in
their own cell, they often give rise to the feeling of attachment. [Then] the feeling
of repugnance [directed toward life in the world] disappears. For this reason, the
rule was established to place the sick in a separate hall. The hall is designated
“Impermanence.” Many go there, but few return. Facing the matter that confronts
them [death and rebirth], the dying monks single-mindedly meditate on the dharma.
Inside the hall is placed a standing [Buddha] image, covered with gold leaf. It faces
the direction of the west. The right hand of the image is raised, and to the left hand
is fastened a five-colored pennant, whose end trails on the ground. The sick monk
is to be placed behind the image. Holding the tail of the pennant with his left hand,
he is to think of following the Buddha to the buddha land. The attendant monk
(zhanbingzhe) burns incense and scatters flowers to adorn [the place]. If the sick
monk soils himself, urinates, throws up, or spits, it is to be cleaned up, and he is not
held responsible for any transgression.
91

After providing why the space is called “Impermanence” and how this would help to set a
sick monk’s mind before death, the most noticeable and relevant scene to the Yinlu Bodhisattva
painting comes in the second paragraph. According to the author, the sick monk is lying behind
the Buddha sculpture whose left hand is holding a “five-colored pennant (or banner).” Moreover,
the tail of the banner literally connects to the sick monk, and the tactile/physical connectivity
helps him visualize project himself in following the animated Buddha to the Buddha land.
92

The quoted passage, which later became popularly quoted by Japanese Pure Land

91. Shinohara, 108–09.
92. There are two different versions of this text regarding the direction one should face or the description
of the Hall of Impermanence, although they are attributed to be written by the same author or a person
related to himself, Daoxuan. One is Fayuan zhulin (T. 2122, ¢#Ö0, “Jade forest in the dharma
garden”) written by Daoshi ÜÊ (?–683), and the other is Zhong tianzhu sheweiguo qihuansi tujing (ÜG
áàâáäã#Í:, T. 1899, “Illustrated Sutra of the Jetavana Monastery in Śrāvastī, Central India”)
by Daoxuan in 667 CE. For a detailed discussion of the differences, see Shinohara, 130 and no. 6.

170
Buddhists such as Genshin GH (942–1017), clearly shows a unique method of using a banner in
the course of the deathbed ritual. The banner’s physical components, which are analogous to
human anatomy, such as the head, arms, body, and legs, function as a medium/instrument to
bridge a sculpture and a dying monk (the actual ritual participant). The banner in the above quote
is described as a “five-colored” banner with a “tail” long enough to be dragged on the ground.
The examples of the “five colors” of the “five-colored banner” are plenty among the Stein and
Pelliot Collections.
93
The banners made with color-blocked panels were certainly used on
various occasions such as the procession of a Buddha image (xingxiangãY), demarcating
sacred space, or one of the monk’s attributes.
94
However, the distinctive and direct uses of the
banner in relation to the deathbed ritual unpack not only the banner’s function that is
contributed/embedded/woven by its materiality but also its core function as a “proxy” for the
main object of veneration.
Reviewing the function of the five-colored banner yields the materiality of the banner
tailored to specific occasions. The foremost instance is that the banner was indicative of sacred
space. A wall painting in Cave 148 from the High Tang period captures the moment when the
ritual was being conducted (see fig. 54). The scene, which was briefly discussed in Chapter 1
regarding the types of painted banners from Cave 17, specifically encapsulates the middle of a
ritual using the five-colored banner inside and outside of a temple site with a wooden shrine. In

93. It should be noted that “five-colored” is more of a rhetorical expression meaning “multi-colored”
rather than a definitive descriptive one. Usually, the extant color-blocked banners have three colors of red,
blue, and white instead of the five colors of black, red, qing (blue and green), yellow, and white. Cases of
solid-colored silk and patterned (e.g., clamp-resisted pattern) silk were often used. Here, both terms will
be interchangeably used.
94. As for the processional scene of a Buddha image, in particular, a mural painting on the south wall of
Cave 72 depicts such a moment. In order to repair a Buddha statue, the head of the Buddha was carried in
a litter, following a group of people holding a color-blocked banner with a long staff. For the drawing of
the painting see Huo, “Mogaoku di 72 ku,” 38, figure 6.

171
the tile-roofed wooden shrine, a Buddha image—either a sculpture or a painting—is flanked with
two attentive bodhisattva images.
On the sides of the bodhisattva images, a pair of color-blocked banners are hung from the
ceiling, decorating and defining the sacred space. On the right side of the space, three seated
monks and one standing attendant are depicted together with the same banner. In the courtyard of
the temple, a flying banner is dynamically described on the tall metallic staff with an ornament
portraying the head of a dragon on the top. Although these banners are depicted in a swaying
motion, there are noticeable differences in their mode of depiction. The representation of the
banner hung outside is more vigorous and dynamic than those inside the shrine. It is also made
up of patterned textiles and multiple divided streamers attached to every two panels of the body,
which results in making the movement more conspicuous even from far away. It certainly
delivers a nonverbal message: Here is the sacred space where this banner belongs. The banners
inside the shrine are less decorative in terms of their type of textile and the number of streamers.
The pair of banners, in particular, is presented in a more subtle way because they are flipped
outward; their movement is not as dynamic as the outside banner’s. However, the way in which
the banners are flipped outwardly signifies not only the ritual in progress but also, more
importantly, the apparition of the invited beings to the ritual altar.
In other words, as important as being dedicated to and decorating the altar, another
expected effect of hanging a banner was to signify the auspicious being’s arrival by capturing the
moment of the banner being animated. To achieve such effects, material for this type of banner
was carefully chosen. Stein observes that the banner that has a triangular head with streamers is
made out of “gauze,” indicating the extremely thin and coarse weaving structure of silk.
However, it should be noted that Stein’s description does not refer to “gauze silk (luo I)” in

172
complex gauze with a twisted warp.
95
Stein refers to the material as gauze because it was “see
through” and “very thin,” as much as actual gauze silk. The weaving structure of the color-
blocked banner type is in plain weave but was woven more coarsely than other silk objects such
as sutra wrappers made out of plain-weave silk for painting.
96
Stein further notes that the choice
of this specific type of material— plain, coarsely woven silk—was not arbitrary but that the
opposite is true:
The use of this fine gauze-like silk for the banners was obviously intentional. It
offered the additional great advantage of causing the banners, when hung up in the
manner described, to interfere less with the lighting of the interior of the cave-
temples than would have been the case otherwise; for the latter, as explained above,
received light only through their porches and the antechapels usually found in front
of these.
97

Because he emphasizes that the thick silk was “intentional,” it leads us to consider
expected functions/advantages in addition to the banner’s adaptability to a darker surrounding, as
proposed by Stein. Another benefit would be sensitivity that is inherent to the material. The
sheerness of the material would not block light coming from outside—if it was displayed inside
the cave—but also make it possible to react to the natural environment, such as a breeze, thus
making the banner more animated and vivified. The fact that the “gauze-like silk” is prone to
bend and fly easily by a human’s or nature’s movements is thus pictorialized as a bent banner
format at the ritual scene of the east wall painting of Cave 148.
I argue, furthermore, that such a sensitive tendency of the material begets animated

95. For the definitions of the luo silk and its weaving structure, see Zhao and Wang, “Glossary of Textile
Terminology,” 365.
96. It should be noted that the “coarse” structure is based not on the thickness of the silk threads but rather
on thread count and density, which determine the quality of the textile. See Brack and Mysak, A
Technical Study of Portable Tenth-Century Paintings, 30 and no. 63.
97. Stein, Serindia, 844.

173
effects to the banners, particularly during a ritual that requires divinity’s visitation—the Yinlu
Bodhisattva, for example. The arrival of the bodhisattva is a crucial part of the deathbed ritual as
it ensures the dying (or sick) person’s access to the Pure Land where they will be reborn. In the
case of the deathbed ritual described in the above passage, it is highly likely that an aniconic,
color-blocked banner was used because it serves as a medium between the sculpture and the
benefactor. In the situation where a proper setting with a Buddha statue is not available, a painted
image would be substituted for the statue. This is the moment when the advantage of the silk
support becomes particularly prominent. The Yinlu Bodhisattva painting’s support material is
sheer, delicate, and sensitive enough to react even to a subtle breeze. Yinlu Bodhisattva (EO
1133) at the Musée Guimet exhibits the bodhisattva holding a banner in his right hand and a
willow branch in his left hand (fig. 81). The support is coarsely woven silk, which reveals both
black sketches and overlapping red outlines on the depiction of flesh. Riding on the cloud, the
family of benefactors—parents and a daughter all wearing formal attire—are escorted by the
bodhisattva who looks back to ensure they are following. Flipping and swaying the painting not
only would provide dynamic and animated effects to the viewer but also prompt the viewer to
engage with the ritual process.
Although there is no textual reference as to how this deathbed-related rite was performed
or what specific intended effects and functions were expected of the silk material, it does not
mean such a practice was not performed. Even the Yinlu Bodhisattva does not have any textual
reference, but his visual representation was prevalent and widely recognizable. In the same vein,
it is equally probable that a Yinlu Bodhisattva image painted on a specific type of material was
designed to provide an impression that he was actually descending (or ascending) to a person
who desires to be succored.
Interests in sensory illusion can explain why some single bodhisattva/Heavenly King

174
painted banners have a representation of bells at the top of the image. Yinlu Bodhisattva (MG.
17657) includes a set of musical instruments such as a flute, drum, and bells. There is no doubt
that they signify that the bodhisattva’s arrival is accompanied by a musical band. Other examples
of Vajrapāni and the Heavenly King of the North or Vaiśravaṇa—such as in Stein Painting 103,
Stein Painting 106, and EO 1172a—bear two to three bells at the top of the frame. What was
depicted above the bell representation in Stein Painting 106 can be surmised by looking at the
other example of the Vaiśravaṇa painting (EO 1172a) at the Musée Guimet: Three different
patterned strips are painted with two bells hanging from their edges. The strips, in fact, can be
identified as a canopy, which can be visually corroborated by the remaining ceiling painting in
Cave 196, which has the representation of a canopy with hanging bells painted in the truncated
pyramidal ceiling.
98
The bells function as a device indicative of the revered one and trigger one’s
visualization by making “sound” for his arrival.
The materiality of a painted banner’s support, as well as the inclusion of ritual
implements on the picture frame, enhance the overall potency of the visual object by creating
illusionary vision and hearing. These materialistic qualities and visual devices, which make the
object more powerful, would contribute to the painting’s talismanic function as argued by Paul
Copp. He states that the efficacy of a portable Buddhist image can be created by the
accompanying dedicatory statements and thus could similarly function like a xylography dhāraṇī
amulet.
99
His argument values and implies that possessing the banner itself could bear potency
regardless of its usage in an actual ritual.
Earlier in this chapter, I explored how painted banners from the Library Cave have shown

98. For the ceiling mural painting’s image reproduction, see Wenjie Duan, Dunhuang shiku quanji, plate
156.
99. Paul Copp, The Body Incantatory: Spells and the Ritual Imagination in Medieval Chinese Buddhism
(New York: Columbia University Press, 2014), 126–28.

175
visual affiliation with thangkas from the early Tibetan Empire from the perspectives of painting
style, mounting, and writing on the reverse, all of which make a painting a three-dimensional
object. I also discussed raw textile material (e.g., silk thread) for weaving being offered and
utilized in sacred spaces and funerary contexts such as a metaphorical burial, a hanging painting
in a tomb, and canopies represented in actual patterned textiles. Furthermore, the innate nature of
the silk material made silk an appropriate offering and implement for a funerary ritual.
A comparable case to the Dunhuang Yinlu Bodhisattva culture can be found thanks to the
2002 discovery of mid-eighth-century painted wooden coffins in Guolimu cemetery, in the
Tsaidam Basin, Qinghai. One coffin from Tomb M1 (fig. 82) exhibits pictorial representations of
funerary practices before Buddhist practices had become prevalent. As Bellezza briefly outlined,
the first panel describes six interrelated scenes of the buried couple, a man and a woman,
undergoing funerary rituals and eventually enjoying their idealistic afterlife. Among the six
scenes on coffin plank no. 2 from Tomb M1, one particular scene has iconographic similarities to
the Yinlu Bodhisattva as psychopomp and can relate to its discussion.
100

As seen in Figure 82, after the heavily caparisoned horses are brought into the ritual
space or in front of the funeral tent, the scene above the tent shows five people riding on horses.

100. The six scenes of the coffin plank no. 2 from Tomb MA are the following: 1) heavily adorned
horses, which are going to be ridden by the dead, transporting them to paradise through the provisional
hell; 2) a funerary tent surrounded by mourning crowds; 3) two bowmen mounted on unadorned horses;
4) an offering of a sacrificial cattle and horse; 5) a men riding a horse and holding a banner followed by
the dead couple riding the caparisoned horses; and 6) the idealized afterlife scene with a banquet and
lovemaking. For the full descriptions of the scenes, see Tao Tong and Patrick Wertmann, “The Coffin
Paintings of the Tubo Period from the Northern Tibetan Plateau,” in Bridging Eurasia, ed. M. Wagner
and Wei Wang (Mainz: German Archaeological Institute, 2010), 190–205; John Vincent Bellezza, Death
and Beyond in Ancient Tibet: Archaic Concepts and Practices in a Thousand-Year-Old Illuminated
Funerary Manuscript and Old Tibetan Funerary Documents of Gathang Bumpa and Dunhuang (Wien:
Verlag der Österreichischen Akademie der Wissenschaften, 2013), 243–47.

176
The last two are bowmen aiming at fully effaced objects.
101
What attracts our attention is the first
person wearing a turban, official garments and boots, as well as guiding the deceased couple.
Judged from his outfit and position, he is a funerary priest whose responsibility is to guide the
dead through the postmortem realm.
102
He holds in his right hand a pentagon-shaped object with
very long streamers. Bellezza argues that this is a nam-kha or thread-cross, “an important ritual
implement for housing and guiding the dead through the postmortem realm.”
103
He adds that the
streamers must be “the white cloth (dar-dkar) that shows the way (lam-ston) to heaven.”
104

The designated function of the funeral priest, to create a bridge to the otherworldly realm,
is strongly resonant with the Chinese version of a psychopomp, the Yinlu Bodhisattva, who is
also represented by a banner and destined to lead the deceased. A similar ritual object to the one
that the priest is holding in the painted panel of the coffin was found in the Library Cave and is
now in the British Library. IOL Tib J 1364 is an octagonal object made of layers of ramie and
hemp rag paper glued together and mounted on a wooden pointed stick (fig. 83).
105
Even if this
object represents a Buddhist tantric bodhisattva, Vajrasattva, and streamers were not discovered
with it, the multisided shape and short handle are comparable to that of the painted plank.
106
In
addition to the similar appearance of the white streamers from the painted panel to the banner of
the Yinlu Bodhisattva, the imposed functionality of the white cloth—which is showing “the way

101. Tong and Wertmann argue that these two bowmen represent the Tibetan indigenous bon ritual
slaying of any evil spirits. Tong and Wertmann, “The Coffin Paintings of the Tubo Period from the
Northern Tibetan Plateau,” 190–205.
102. Bellezza, Death and Beyond in Ancient Tibet, 245.
103. Bellezza, 245.
104. Bellezza, 245.
105. van Schaik, Helman-Ważny, and Nöller, “Writing, Painting and Sketching at Dunhuang: Assessing
the Materiality and Function of Early Tibetan Manuscripts and Ritual Items,” 123.
106. Van Shaik, Helman-Ważny, and Nöller, 123.

177
to heaven” and thus invoking the presence of the deified or the dead—is parallel to the
materiality of the painted banners from Dunhuang as discussed earlier. Therefore, it is safe to say
that the actual Tibetan ritual object from the Library Cave indirectly corroborates the fact that the
funerary practice in Dunhuang is possibly associated with that of the Tibetan Empire. In the
same vein, the Yinlu Bodhisattva’s guiding of the soul resembles the concept of the funerary
priest leading the soul. The fact that these Tibetan psychopomps and their representations also
predate the emergence of the Yinlu Bodhisattva, which appeared during the ninth and tenth
centuries, suggests the potential impact of the Tibetan practice and its visual culture.

4.6. Conclusion
In summary, the large cache of textiles from the Library Cave exhibit multiple
functionalities that fulfilled a wide spectrum of donors’ needs and wishes, ranging from
monetary means to religious offerings. Sacred spaces, such as a Buddhist cave-temple or a
represented sacred space within which a bodhisattva resides, are decorated with precious textiles.
This chapter focused on the latter, shedding new light on the embedded materiality of the
popularly chosen textiles. The inherent durability, extendibility, flexibility, and sensitivity of silk
were often connected to the metaphysical concept that silk would serve as an effective medium
between the dying/sick/dead and the otherworldly realms/beings. Underlying ideas behind the
association of the materiality of the textile and pertinent death rituals, as discussed in the context
of Astana tombs, Tibetan paintings, and the Yinlu Bodhisattva painting, are that textiles serve a
vital role in bestowing/forming a psychopomp in the funeral ritual performances. The materiality
of silk contributed to boosting the functionality of the psychopomp to serve its purpose.
While the physical characteristics of painted banners can be generously applied to other
contexts of visual culture, the impact of the Tibetan Empire cannot be overlooked. The

178
production of painted banners reached its apex during the ninth to tenth centuries after the
Tibetans occupied Dunhuang. The visual culture of the early Tibetan Empire has been less
highlighted than the later “Renaissance” period during the tenth to twelfth centuries CE. Actual
application of ritual implements, such as banners or handheld objects at funeral services, is
reminiscent of the Tibetan practices. Moreover, it was important for Tibetan donors to please
their deities by offering sensory objects such as soft silk.
107
Although I have attempted to
emphasize the aspects of Tibetan visual culture in terms of textile offerings, space
embellishment, and funerary practices utilizing the materiality of textile, what remains to be
investigated by future scholars is the origin of the Yinlu Bodhisattva in relation to the Old
Tibetan funeral practice and a funerary priest. The task that Yinlu was imposed overlaps with
that of Dizang, who is the Chinese version of Kṣitigarbha that takes on the appearance of a
hooded monk.
108
Given the fact that a Buddhist priest would conduct funerary services in
medieval Dunhuang, it is plausible that the monk became deified and appropriated to serve as a
psychopomp. In addition, more comparative visual analysis between the Tibetan bon art/practice
and the funerary representation in Dunhuang is needed to illuminate their direct/indirect
relationship.

107. Sam van Schaik, “The Naming of Tibetan Religion: Bon and Chos in the Tibetan Imperial Period,”
Journal of the International Association for Bon Research 1 (2013): 234.
108. Ng, The Making of a Savior Bodhisattva, 229–39.

179
Conclusion
Woven textile banners with painted images were significant religious offerings expressing
piety in ninth- and tenth-century China. Having been given less scholarly attention than Buddhist
texts and even the wall paintings of the Mogao Caves, the significance of the banner paintings
from Cave 17 should be emphasized because their physical materials were associated with
devotional practice. By analyzing and reconstructing the microstructure, style, and functionality
of select examples, I have discovered that artistic form and style were affected by the types of
materials used and that both painting format and the work’s perceived religious function were
largely shaped by the interactions between Dunhuang and neighboring regions such as Turfan,
Khotan, and Tibet. In addition, painted banners were relatively affordable devotional objects for
ordinary people, especially when they dedicated them in mortuary ritual settings. The extreme
long painted banner and silk threads being buried in a tomb in Astana near Turfan also exemplify
how patrons’ wishes (e.g., prolonging life and reaching heaven) became materialized in the
works’ formal appearance.
This dissertation has examined the material qualities of a banner painting’s support as
well as the structure of painted textiles that contributed to the efficacy of Buddhist rituals and
also of an offering itself. The lightweight and—in many cases—transparent silk materials of
painted banners were not only responsive to elements in their immediate environment such as
light sources and airflow, but also were made of expandable and flexible material, which enabled
them to be extended in length or width by sewing additional pieces onto them. These qualities of
the painted banners from Cave 17 are closely associated with artistic practice and religious life.
From the perspective of artistic practice, the banners’ material allowed the build-up of

180
multiple layers of pigments, which means that outlines and transgressing color fills were easily
noticeable if the pigment layers were not thick enough to conceal a mistake. The qualities of the
painted banners vary from highly sophisticated to roughly finished, depending on who the donors
were and what their social status was.
Unlike a traditional hanging painting, Stein Painting 52 was not backed with another
piece of textile. Instead, to increase its durability, the edges of the painting were covered and
patch-worked with yellow dyed silk of a denser, plain-weave structure and one small piece of
embroidery on green gauze silk (1919,0101,0.52*), which had been detached from the upper left-
hand corner when it was mounted on a new panel in the 1920s.
1
This strongly suggests that the
silk used for framing was in a short supply, so that a piece of embroidery was repurposed and
patch-worked to finish up the framing.
Unlike wall paintings and traditional Chinese hanging scrolls, the flexible material of all
the banner paintings examined in this dissertation also made them ideal to be displayed in ritual
processions, thereby impacting and enhancing the viewer’s experience of them. The thin silk
material is responsive to movement, unlike immovable wall paintings or even hanging scrolls. A
long banner and other rectangular-shaped iconic paintings were meant to be seen from both
sides. When displayed on top of a tall staff or even on a pole on the exterior, wooden
architectural structure outside cave-temples, the painted images from the recto were likely visible
on the verso due to the painting support’s transparent quality.
2
Such viewing setup was
intentional, as the banner was processed and utilized in the course of a ritual with large

1. Whitfield, The Art of Central Asia 3, plate 5; Zhao, Textiles from Dunhuang in UK Collections, 254.
2. The whole cave site was built and carved into the cliff side, which naturally created a tall height and
elevation throughout the site. For geographic information about the site, see Wong and Agnew, The
Conservation of Cave 85, 33–45.

181
gatherings as seen in visual representations of such scenes.
From the perspective of religious practice, furthermore, materiality generated by the
physical length and specific pigments reinforces the object’s expected functionality especially
during the ritual process. My analysis of Stein Painting 216 in Chapter 3, the Avalokiteśvara
Sutra paintings in Chapter 2, the textiles used as offerings and in tomb burials in Chapter 4 has
demonstrated that the objects’ physical characteristics (e.g., their length) and the symbolic
connotations associated with certain pigments and threads had allowed the users to imagine
longevity, salvation, and rebirth in the Pure Land when alive or when death was imminent. It is
thus no surprise that the savior Avalokiteśvara and Yinlu Bodhisattva were the most popular
pictorial subjects on textiles. Part of this phenomena can be explained by the fact that their
purported manifestations to devotees to lead them to the Pure Land were essential to the working
of the Avalokiteśvara cult during the medieval period.
As painted banners and liturgical framed paintings were meant to be portable and
accessible for a larger audience, their form and format had affected and were also affected by
other visual and religious practices. The visual culture of early Tibetan Empire and Daoist ritual
implements exhibit shared forms in terms of framing methods and ritual formats, ones that were
interchangeable with their Buddhist counterparts. As discussed in Chapter 4, the life-extending
ritual manuals in Chinese apocrypha that were based on Daoist practice and the relevant visual
paraphernalia in the “spirit-moving banner” showcase another example of Buddho-Daoist
interactions with the popular wishes among practitioners.
While this dissertation has demonstrated how the materiality of the painting offerings
enhanced the offerings’ religious efficacy in the ninth to tenth centuries in Dunhuang, more
research is needed to address outstanding questions in several areas. First, the Tibetan Empire’s

182
role in the broader context of esoteric Buddhist religious culture, especially in relation to the
Avalokiteśvara cult I mentioned above, needs to be clarified. In this project, I have focused on
the Avalokiteśvara Sutra painting and Yinlu Bodhisattva paintings, but there are other types of
texts from Cave 17 such as dharani (an incantation to be recited for a dying, deceased, or sick
person) and sādhana (the visualization and recitation practice of deity), both of which could
possibly corroborate and support my hypothesis of the Avalokiteśvara as a main agent in healing
and death-related rituals.
3
By broadening the pictorial imagery of Avalokiteśvara to include
esoteric forms as well as texts, the popularity of the deity in medieval China can be explained on
a more substantial level. Second, even though I have suggested painted banners’ use in ritualistic
settings, the exact venue for such rituals or even reconstructions of spatial setting have not been
discussed. With regard to the rectangular hanging painting in particular, any visual depiction of
such paintings has not been found among the wall paintings of Mogao Caves. Moreover, a lack
of actual physical remains of ritual implements such as banner holders, incense burners, and
sculptures from Cave 17 as they were mentioned in the temple’s inventory is an issue to reckon
with, possibly by referencing other visual materials even if they were from later periods. Lastly, I
also have proposed the possibility that the major components of the paintings could have been
repurposed from the tax-textiles. Given the fact that recycling preexisting materials owned by a
temple and patching them with old textiles or scraps of textiles were, in fact, common practices
among the corpus of paintings from Cave 17, what remains for further research is to find out to
what extent this practice was prevalent in this period.
4
Whereas raw silks, the major focus of

3. Sam van Schaik, “The Tibetan Avalokiteśvara Cult in the Tenth Century: Evidence from the Dunhuang
Manuscripts,” in Tibetan Buddhist Literature and Praxis: Studies in Its Formative Period, 900-1400, ed.
C. Wedermeyer and R Davidson (Leiden: Brill, 2006), 57–62.
4. For the practice of recycling textiles and the examples among the Dunhuang paintings in the US
collections, see Brack and Mysak, A Technical Study, 31–39.

183
Chapter 1, have the simplest weaving structure, which makes it difficult to determine their
original production location, there are cases in which complex weaving structures (e.g., damask)
were applied. Focusing on these types of textiles that are considered to be uniquely used for
painting would offer a clearer picture of what made the textiles “old” and the notion of suitability
with the recycled offerings. This would, in fact, help define the characteristics of all the materials
from Cave 17, as the scraps and fragments of textiles including a headpiece and a single
streamer, were all actually recorded as “permanent objects” in a temple’s inventory; this would
not have been done if these items were considered as waste. In contrast, they were regarded as
possible future resources which were repurposed and reappropriated as parts of banners’ frames,
loops, and patchwork valances.
In closing, the materiality of the textile part of the ritual offering played a crucial role in
promoting and conveying artistic practice, visual representation of Buddhist icons, and ritualistic
ideas inside and outside Dunhuang during the ninth and tenth centuries CE. I believe that the
banners added another layer of religious practice, providing more inclusive and accessible ways
to convey common devotees’ wishes. The landscape of the medieval Mogao Caves site was far
more dynamic than it appears now.

184
FIGURES

Figure 1: Map of Northwest China (Image from Hansen, The Silk Road, 22–23).

185

Figure 2: Stein’s Plan of Cave 16 and Cave 17 (gray area labeled as “Walled-up chapel”) (Image
from Stein, Serindia 3, Plans: 43).

186
Figure 3: Bodhisattva with Glass Beaker, Stein Painting 120 (1919,0101,0.120), Tang dynasty,
9th or 10th century CE, ink and colors on silk, H. 58.0 cm, W. 18.0 cm, British Museum (Image
from Whitfield, The Art of Central Asia 1, Plate 28.).

187

Figure 4: Portrayal of the Life of Buddha, Maya Cave, Kizil, Xinjiang (Image from von Le Coq,
Die Buddhistische Spätantike in Mittelasien 3, Tafel A.a.).

188

Figure 5: Drawing of dhāraṇī pillars as a cloth banner, south wall of Cave 217, Mogao Caves,
Dunhuang, Gansu Province, China (Image from Shimono, “Tonkō Bakukōkutsu dai 217 kutsu,”
figures 10e-f).

189

Figure 6: Panel with confronting horses (a) and its lining of dark-red/brown silk (b), late 8th–9th
century CE, length (a): 67.5 cm; width (a): 52.8 cm, length (b): 66 cm; width (b): 55 cm, MAS.
885 a-b, British Museum (Image from Zhao, Textiles from Dunhuang in UK Collections, 196).

190

Figure 7: Seated Avalokiteśvara, Stein Painting 52 (1919,0101,0.52), 972 CE, ink and colors in
silk, H. 91.5 cm x W. 59.1 cm (painted area), H. 102.2 cm x W. 69.0cm (with frame), British
Museum (Image downloaded from http://britishmuseum.org).

191

Figure 8: Comparison of Stein Painting 52 and Stein Painting 103 showing different directions of
the silk across painting workshops (illustration made by author).

192

Figure 9: Painted Banner of Avalokiteśvara (detail), Stein Painting 154 (1919,0101,0.154), 10th
century CE, ink and colors on hemp, H. 94 cm x W. 27 cm, British Museum (Image downloaded
from http://britishmuseum.org).

193

Figure 10: Left: Standing Avalokiteśvara, 984 CE, ink and colors on hemp, W. 64.5 cm x H. 191
cm, MG. 22799, Musée Guimet; Right: Reconstructed image of the painting framed on a wooden
structure (Image from International Dunhuang Project,
http://idp.bl.uk/database/oo_scroll_h.a4d?uid=615818696;recnum=91351;index=1, accessed
November 2, 2016; Garcia, “‘La quête des mérites’,” 55).

194

Figure 11: Representation of a painting mounted on a drawing frame, line drawing of the south
wall of Cave 72, Dunhuang, mid-9th to 10th century CE (Image from Huo, “Mogaoku Di 72
Ku,” 38, fig. 6).

195

Figure 12: Seated Avalokiteśvara, Stein Painting 54 (1919,0101,0.54), 983 CE, ink and colors on
silk, H. 99.0 cm x W. 59.5 cm (painted area), H. 102.0 cm x W. 75.5 cm (with frame), British
Museum (Image from Whitfield, The Art of Central Asia 2, Pl. 27).

196

Figure 13: Avalokiteśvara as a Savior of Perils, Stein Painting 2 (1919,0101,0.2), 10th century
CE, ink and colors on silk, H. 84.4 cm x W. 61.7 cm, British Museum (artwork in the public
domain; photograph all rights reserved, British Museum).

197

Figure 14: Avalokiteśvara of the Water Moon, 968 CE, ink and colors on silk, H. 106.8 cm x W.
58.9 cm, Freer | Sackler Gallery, Washington DC (Image from
https://www.freersackler.si.edu/object/F1930.36/#txtMetaData).

198

Figure 15: Embroidery from the original framing of Stein Painting 52, British Museum (Image
from http://britishmuseum.org).

199

Figure 16: The original framing of Stein Painting 52 (Image from Stein, Serindia 4, Pl. LXI).

200

Figure 17: Hanging scroll in typical Chinese style, front and back (Image from Valerie Lee,
Xiangmei Gu, and Yuan-Li Hou, “The Treatment of Chinese Ancestor Portraits,” 465, fig. 1).

201

Figure 18: Cross section of a traditional Chinese scroll showing the layers of backing (Image
from Valerie Lee, Xiangmei Gu, and Yuan-Li Hou, “The Treatment of Chinese Ancestor
Portraits,” 466, fig. 2).

202

Figure 19: Typical Tibetan thangka frame (Image from Jackson and Jackson, Tibetan Thangka
Painting, 143).

203

Figure 20: Padmasambhava and its reverse showing how the painting is framed without covering
the backside of the support, ground mineral pigment, red background on cotton. 17th century CE,
Field Museum of Natural History (Image from Himalayan Art Resources, https://www.
himalayanart.org/items/54358, accessed January 3, 2018).

204

Figure 21: Large patchwork, late 8th–9th century CE, various weaves in silk, 150.5 cm x 111
cm, MAS. 856, British Museum, London (Image from Zhao ed., Textiles from Dunhunag in the
UK Collections, 113, Pl. 79).

205

Figure 22: Border for a votive painting, 7th–10th century CE, silk, outside frame: 84.0 cm x
63.5), inside frame: 72.5 cm x 48.0 cm, MG. 26789, Musée Guimet, Paris (Image from Zhao ed.
Textiles from Dunhuang in French Collections, 124, Pl. 72).

206

Figure 23: The Medicine Buddha sutra transformation tableaux, early Tang dynasty, south wall
of Cave 220, Mogao Caves, Dunhuang (Image from Dunhuang wenwu yanjiusuo, Zhongguo
shiku: Dunhuang Mogaoku 3, Pl. 24).

207

Figure 24: Central niche on the west wall of Cave 112, middle Tang dynasty, Mogao Caves,
Dunhuang (Image from Dunhuang yanjiu yuan, and Jiangsu meishu chubanshe, eds. Mogao ku di
112 ku, 18, Pl. 1).

208

Figure 25: Layout of sutra paintings in Cave 112 (illustration by author).

209

Figure 26: Visualization Sutra Painting (left section) and Diamond Sutra Painting (right section),
south wall of Cave 112, Middle Tang dynasty, Mogao Caves, Dunhuang (Image from Dunhuang
yanjiu yuan, and Jiangsu meishu chubanshe, Mogao Ku Di 112 Ku, 59, Pl. 50).

210

Figure 27: Repaying Kindness Sutra (left section) and the Bhaisajyaguru Sutra (right section),
north wall of Cave 112, Middle Tang dynasty, Mogao Caves, Dunhuang, China (Image from
Dunhuang yanjiu yuan, and Jiangsu meishu chubanshe, Mogao ku di 112 ku, 130, Pl. 128).

211

Figure 28: Avalokiteśvara Sutra (left section), Maitreya’s Subjugation of the Demons (above the
entrance), and Maitreya (right section), east wall of Cave 112, Middle Tang dynasty, Mogao
Caves, Dunhuang, China (Image from https://www.e-
dunhuang.com/cave/10.0001/0001.0001.0112, accessed March 1, 2015).

212

Figure 29: Top: pomegranates with vines (shiliu juancao wen !"#$%), detail of Figure 26,
Middle Tang dynasty, Cave 112; Bottom: the same pattern from Cave 158, Middle Tang dynasty
(Images from Dunhuang yanjiuyuan, Dunhuang shiku quanji 14, Pl. 133).

Figure 30: Camellia flowers with rolling vines (chanzhi chahua wen &'()%) (Image from
Dunhuang yanjiuyuan, Dunhuang shiku quanji 14, Pl. 136).

Figure 31: Diamonds/rhombi (lingge wen *+%), detail of Figure 24, Middle Tang dynasty,
Cave 112.

213

Figure 32: Top: tortoiseshells (guijia wen ,-%), detail of Figure 27, Middle Tang dynasty,
Cave 112; Bottom: the same pattern from Cave 323, west niche, High Tang dynasty (Image from
Dunhuang yanjiuyuan, Dunhuang shiku quanji 14, Pl. 74).

Figure 33: Top: quatrefoil alternating with paired halved flowers (lingxing siye wen *./0
%), detail of Figure 27, Middle Tang dynasty, Cave 112); Bottom: the same pattern from Cave
26, Yulin Caves, Song dynasty (Image from Dunhuang yanjiuyuan, Dunhuang shiku quanji 14,
Pl. 219).

Figure 34: Six-petal flower pattern alternating with a pair of halved flowers (liuyehua tuan 10
)23), detail of Figure 26, Middle Tang dynasty, Cave 112.

214

Figure 35: Schematic illustration of decorative patterns in Cave 112 (illustration by author).

215

Figure 36: Photograph showing diamond pattern around the western niche (left) and no enclosing
patterned borders around the Amitābha Sutra painting (right) on the north wall of Cave 322,
Early Tang dynasty, Mogao Caves, Dunhuang (Image downloaded from https://www.e-
dunhuang.com).

216

Figure 37: Avalokiteśvara as Savior of Perils, Stein Painting 28* (1919,0101,0.28*), ink and
colors on silk, H. 83.3 cm x W. 63.1 cm, British Museum (Image downloaded from
http://idp.bl.uk/database/oo_scroll_h.a4d?uid=7675208;bst=1;recnum=40339;index=1;img=1).

217

Figure 38: Paradise of Amitābha, Stein Painting 203 (1919,0101,0.203), ink and color on hemp,
H. 95.5 cm x W. 49.5 cm, British Museum (Image downloaded from http://britishmuseum.org).

218

Figure 39: Fragments with concentric lozenge design, дx46, damask on plain-weave silk, Tang to
Five Dynasties period, a. 7.7 cm x 9.5 cm, b. 10.0 cm x 17.0 cm, State Hermitage Museum,
Russia (Image from Zhao, Textiles from Dunhuang in Russian Collections, Cat. 46.).

Figure 40: Ribbon with diamond pattern, MAS 909, silk braid, High Tang dynasty, late 7th to
early 8th century, 19 cm x 2.5 cm, British Museum (Image from Zhao, Textiles from Dunhuang
in UK Collections, Cat. 62).

219

Figure 41: Banner head with florets, Дx49, plain weave in silk, clamp-resist dyed, late Tang to
Five Dynasties period, 9th to 10th century, length: 19.7 cm, width: 36.5 cm, State Hermitage
Museum (Image from Zhao, Textiles from Dunhuang in Russian Collections, 61).

220

Figure 42: Bodhisattva Avalokiteśvara as a Savior of Perils, 1919,0101,0.243, Five Dynasties
period, 10th century CE, woodblock print, 32 cm x 20 cm, British Museum (Image downloaded
from http://www.britishmuseum.org/).

221

Figure 43: Avalokiteśvara Sutra painting (detail), Middle Tang dynasty, east wall of Cave 112
(Image from Dunhuang yanjiu yuan, and Jiangsu meishu chubanshe, Mogao ku di 112 ku, 196,
Pl. 196).

222

Figure 44: Identification of the thirteen peril scenes, Avalokiteśvara Sutra painting, east wall of
Cave 112, Mogao Caves, Dunhuang (The background image’s sharpness, contrast, and
brightness have been modified by the author to increase visibility).

223

Figure 45: Avalokiteśvara Sutra painting, High Tang dynasty, south wall of Cave 45, Mogao
Caves, Dunhuang (Image from Dunhuang wenwu yanjiusuo, Zhongguo shiku: Dunhuang
Mogaoku 3, Pl. 131).

224

Figure 46: Avalokiteśvara Sutra painting, High Tang dynasty, photography taken 1943–1944,
south wall (center, upper section) of Cave 205, Mogao Caves, Dunhuang (Image downloaded
from James and Lucy Lo Photograph Archive, www.artstor.org).

225

Figure 47: Avalokiteśvara as a Savior of Perils, Stein Painting 24 (1919,0101,0.24), 963 CE, ink
and colors on silk, H. 107.0 cm x W. 61.5 cm, British Museum (Image downloaded from
https://www.britishmuseum.org/collection/object/A_1919-0101-0-24).

226

Figure 48: Avalokiteśvara as Savior of Perils, 950–1000 CE, ink and colors on silk, H. 84.1 cm x
W. 61.2 cm, MG. 17665, Musée Guimet (Image downloaded from
http://idp.bl.uk/database/oo_scroll_h.a4d?uid=132366109;recnum=91290;index=1).

227

Figure 49: Eleven-headed Avalokiteśvara, 985 CE, ink and colors on silk, H. 99.6 cm x W. 62.8
cm, 1943.57.14, Harvard Art Museums (Image downloaded from
http://www.harvardartmuseums.org/art/204072).

228

Figure 50: Standing Bodhisattva, Stein Painting 22 (1919,0101,0.22), Tang dynasty, ink and
color on silk, H. 103.1 cm x W. 71.2 cm, British Museum (Image from Whitfield, The Art of
Central Asia 2, 3).

229

Figure 51: Avalokiteśvara, Stein 391, 925 or 865 CE, ink and colors on silk. Nm 2003-17-352,
National Museum, New Delhi (Image from Chandra and Sharma, Buddhist Paintings of Tun-
Huang in the National Museum, New Delhi, 123 Pl. 34).

230

Figure 52: Vajrapāni, Stein Painting 103, early 9th century CE, ink and colors on silk, H. 55 cm,
W. 14.5 cm, British Museum (Image downloaded from http://britishmuseum.org).

231

Figure 53: Avalokiteśvara in different media: details of Figures 42, 49, and 50.

232

Figure 54: Banners being utilized in a temple (red arrows indicate a pair of banners being
displayed within a shrine), east wall of Cave 148, Mogao Caves, High Tang dynasty (Image from
Dunhuang yanjiuyuan, Dunhuang shiku quanji 21, 168, Pl. 159).

233

Figure 55: Long Banner of Bodhisattvas, Stein Painting 216 (1919,0101,0.216), 956 CE, ink and
colors on silk, H. 396.0 cm x W. 59.0 cm, British Museum (Image downloaded from
http://britishmuseum.org).

234

Figure 56: Detail of Figure 55 showing thick threads to prevent the cut silk from fraying
(photograph by author).

235

Figure 57: Detail of Figure 55 showing the inscription on Stein Painting 216 (Image downloaded
from http://britishmuseum.org).

236

Figure 58: Preaching Shakyamuni, Stein Painting 6, 1919,0101,0.6, early 8th century CE, ink
and colors on silk, H. 139.0 cm x W. 101.7 cm, British Museum (Image downloaded from
http://britishmuseum.org).

237

Figure 59: Long Banner with Bodhisattvas, Stein Painting 196 (1919,0101,0.196), 956 CE (?),
ink and colors on silk, H. 341.0 cm x W. 28.5 cm, British Museum (Image downloaded from
http://britishmuseum.org).

238

Figure 60: Long Banner of Bodhisattvas, Bon 4025, 10th century CE, ink and colors on silk,
National Museum of Korea (Image from Kim, Kungnip chungang bangmulgwan sojang
chungang ashia chonggyo hoehwa, 153, Pl 43).

239

Figure 61: Left to right: Comparison of the painting styles of Bon 4025 (fig. 60), Stein Painting
196 (fig. 59), and Stein Painting 216 (fig. 55).

240

Figure 62: Reconstructions of Stein Paintings 196 and 216, and Bon 4025 (reconstructed by
author).

241

Figure 63: Diagram of a Daoist “spirit-moving banner” (Image from Huang, “Daoist Uses of
Color in Visualization and Ritual Practices,” 230, fig. 7).

242

Figure 64: Long Banner, Stein Painting 217 (1919,0101,0.217), silver on woven damask red silk,
H. 386.0 cm x W. 68.1 cm, British Museum (Image downloaded from http://britishmuseum.org).

243

Figure 65: Streamer of Avalokiteśvara, Д x 291, plain-weave in silk damask on plain weave in
silk; painted in yellow, late Tang to Five Dynasties period, H. 48.5 cm x W. 37.6 cm, State
Hermitage Museum of Russia (Image from Zhao, Textiles from Dunhuang in Russia Collections,
76, Pl. 9).

244

Figure 66: Unidentified name of a bodhisattva on Bon 4025 (fig. 60), National Museum of Korea
(photograph by author).

245

Figure 67: Long Banner, EO 3647, 10th century CE, ink and colors on blue silk, H. 580.0 cm x
W. 32.9 cm, Musée Guimet (Image from Giès, Art of Central Asia 2, Pl. 46).

246

Figure 68: Banner with Bodhisattva, possibly Mahamayuri, Tang dynasty or Five Dynasties
period, 9th–10th century, China (Dunhuang area, Gansu Province), ink and pigment on silk, H.
57.0 cm x W. 28.0 cm, Metropolitan Museum of Art (2007.294 a, b) (Image from
https://www.metmuseum.org)

247

Figure 69: Painting on north wall of Cave 231 showing a flying banner hung on a dragon-headed
staff, 8th to 9th century CE, Mogao Caves, Dunhuang, China (Image from Dunhuang
yanjiuyuan, Dunhuang shiku quanji 21, 168, Pl. 159).

248

Figure 70: Dragon-headed banner stand (Image downloaded from
http://data.kdata.kr/resource/Yongdubodang, accessed 2018).

249

Figure 71: Silk banner from Mawangdui Tomb No. 1, Western Han dynasty, ink and colors on
silk, H. 205.0 cm x W. 92.0 cm, Hunan Provincial Museum, China (Image downloaded from
Hunan Provincial Museum, http://61.187.53.122/collection.aspx?id=1348&lang=zh-CN).

250

Figure 72: Banner showing Fuxi and Nüwa, Ast(ana) ix. 2. 054, Turfan (Image from Stein,
Innermost Asia 3, Pl. CIX).

251

Figure 73: Image showing a wooden peg installed in a hole, south wall of Cave 290, Northern
Zhou dynasty, Mogao Caves, Dunhuang, China (Image from Dunhuang wenwu yanjiusuo,
Zhongguo shiku: Dunhuang Mogaoku 1, plate 181).

252

Figure 74: Bodhisattva with Lotus, Stein Painting 102 (1919,0101,0.102), early 9th century CE,
ink and colors on silk, H. 51.0 cm x W. 14.0 cm, British Museum (Image downloaded from
http://britishmuseum.org).

253

Figure 75: Bodhisattva with Lotus, Stein 535, ink and color on silk, National Museum, New
Delhi (Image from Chandra and Sharma, Buddhist Paintings of Tun-Huang in the National
Museum, New Delhi, 223, Pl. 97).

254

Figure 76: Avalokiteśvara, 11
th
century CE (?), H. 77.5 cm x W. 23.5 cm, Yarlung Museum,
Tsethang (Image from Lojda, Klimburg-Salter, and Strinu, “The Tibetan Himalayan Style,” 141,
figure 4.9).

255

Figure 77: Painted cloth textile placed between beams, Main Temple, Assembly Hall, Alchi
Monastery (Image from Klimburg-Salter and Luczanits, Tabo, 174 and fig. 193).

256

Figure 78: Painted cloth textiles placed between beams, Main Temple, ambulatory, ceiling, Alchi
Monastery (Image from Klimburg-Salter and Luczanits, Tabo, 175 and fig. 196).

257

Figure 79: Paradise of Bhaisajyaguru, Stein Painting 32 (1919,0101,0.32), 836 CE, ink and
colors on silk, H. 152.3 cm x W. 177.8 cm, British Museum (Image downloaded from
http://idp.bl.uk/).

258

Figure 80: Yinlu Bodhisattva (?), Tang dynasty, photograph taken 1943–1944, west wall (lower
right) of Cave 205 (Image downloaded from James and Lucy Lo Photograph Archive,
www.artstor.org).

259

Figure 81: Yinlu Bodhisattva, EO 1133, 10th century CE, ink and colors on silk, H. 59.0 cm x W.
38.0 cm, Musée Guimet (Image downloaded from http://idp.bl.uk/).

260

Figure 82: Line drawings of funerary scenes on coffin plank no. 2, Tomb M1, 8th century CE,
Guolimu (Image from Bellezza, Death and Beyond in Ancient Tibet, 247 and fig. 16).

261

Figure 83: Ritual implement, H. 18.2 cm x W. 11 cm, glued paper layers on a wooden stick, IOL
Tib J 1364, British Library (Image downloaded from
http://idp.bl.uk/database/oo_scroll_h.a4d?uid=5719393746;recnum=22865;index=5).

262
Bibliography

Dunhuang Manuscripts

Pelliot Collection
P. 2171
P. 2583
P. 2613
P. 2623
P. 2805
P. 2837
P. 2853
P. 3432
P. 3638
P. 5032v

Stein Collection
S. 1774
S. 2428
S. 4215
S. 6829

Pecking University Collection
D. 162

Primary Buddhist and Daoist Sources

Buddhist Sources
T. 262 Miaofa lianhua jing È6JK=.
T. 450 Yaoshi liuliguang rulai benyuan gongde jingπùLM-398NOÓ=.
T. 804 Sifenlü shanfan buque xingshi chao/@ABCDEãÃF.

263
T. 1239 Azhabaoju yuanshuai dajiang shang fo tuoluoni jing xiuxing yigui7PQRªS
9T‚∫8IU=(ãVW.
T. 1331 Guanding jing ≈∆=.
T. 1899 Zhong tianzhu sheweiguo qihuansi tujing ÜGáàâáäã#Í:.
T. 2122 Fayuan zhulin¢#Ö0.
T. 2871 Datong fangguang chanhui miezui zhuangyan chengfo jing 9Ù·ıˆ˜¯ÿ˘
¿∫=.
T. 2888 Fo shuo yan shouming jing ∫Ωá¥‘=.

Daoist Sources
DZ 371 Taishang dongguan lingbao santu wuku badu shengsi miaojing fi‚‰ÂÊÁ6
Ë©≠’¤ÖŸÈ=.
DZ 373 Taishang dongguan lingo wangsheng jiuku miaojing fi‚‰ÂÊÍÎÖ¨≠È
=.
DZ 466 Lingbao lingjiao jidu jinshu ÊÁXYZ¤µ[.

Secondary Sources
Acker, William Reynolds Beal. Some T’ang and Pre-T’ang Texts on Chinese Painting Hyperion
reprint ed. Westport, Conn: Hyperion Press, 1979.
Agnew, Neville, Marcia Reed, and Tevvy Ball, eds. Cave Temples of Dunhuang: Buddhist Art
on China’s Silk Road. Los Angeles: The Getty Conservation Institute, 2016.
Agnew, Neville, Martha Demas, and Wang Xudong. “The Enduring Collaboration of the Getty
Conservation Institute and the Dunhuang Academy in Conservation and Management at
the Buddhist Cave Temples of Dunhuang, China.” The Public Historian 34 (2012): 7–20.
Anderl, Christoph. “Linking Khotan and Dūnhuáng: Buddhist Narratives in Text and Image.”
Entangled Religions 5 (2018): 250–311.
Arakawa, Masahiro. “The Transportation of Tax Textiles to the North-West as Part of the Tang-
Dynasty Military Shipment System.” Journal of the Royal Asiatic Society 23, no. 2
(2013): 245–61.
Barbieri-Low, Anthony. Artisans in Early Imperial China. Seattle: University of Washington
Press, 2007.
Beckwith, Christopher I. The Tibetan Empire in Central Asia: A History of the Struggle for
Great Power Among Tibetans, Turks, Arabs, and Chinese During the Early Middle Ages.
Princeton, N.J.: Princeton University Press, 1987.

264
Beer, Robert. The Encyclopedia of Tibetan Symbols and Motifs. Boston and New York: Random
House, 1999.
Bellezza, John Vincent. Death and Beyond in Ancient Tibet: Archaic Concepts and Practices in
a Thousand-Year-Old Illuminated Funerary Manuscript and Old Tibetan Funerary
Documents of Gathang Bumpa and Dunhuang. Wien: Verlag der Österreichischen
Akademie der Wissenschaften, 2013.
Benn, James A. Burning for the Buddha: Self-Immolation in Chinese Buddhism. Honolulu:
University of Hawai’i Press, 2007.
Bentor, Yael. "On the Symbolism of the Mirror in Indo-Tibetan Consecration Rituals." Journal
of Indian Philosophy 23, no. 1 (1995): 57–71.
———. “The Horseback Consecration Ritual,” In Religions of Tibet in Practice, edited by
Donald S. Lopez, 234–54. Princeton: Princeton University Press, 1997.
Bhattacharya-Haesner, Chhaya. Central Asian Temple Banners in the Turfan Collection of the
Museum für Indische Kunst, Berlin Painted Textiles from the Northern Silk Route. Berlin:
Reimer, 2003.
Binyon, Laurence, and Sidney Colvin. “The Late Stanley William Littlejohn.” The Burlington
Magazine for Connoisseurs 32, no. 178 (1918): 16–19.
Bogel, Cynthea J. With a Single Glance: Buddhist Icon and Early Mikkyō Vision. Seattle:
University of Washington Press, 2009.
Brack, Matthew, and Erin Mysak. A Technical Study of Portable Tenth-Century Paintings from
Dunhuang in US Collections. Boston: Harvard Art Museums, 2010.
Buck, David. “The Han Dynasty Tomb at Ma-wang-tui.” World Archaeology 7, no. 1 (1975):
30–45.
Burnham, Dorothy K. Warp and Weft: A Textile Terminology. Toronto: Royal Ontario Museum,
1980.
Buswell, Robert E. Encyclopedia of Buddhism. 2004.
Buswell, Robert E., and Donald S. Lopez. The Princeton Dictionary of Buddhism. Princeton and
Oxford: Princeton University Press, 2014.
Chandra, Lokesh, and Nirmala Sharma. Buddhist Paintings of Tun-Huang in the National
Museum, New Delhi. New Delhi: Niyogi Books, 2012.
Chen, BuYun. Empire of Style: Silk and Fashion in Tang China. Seattle: University of
Washington Press, 2019.
Chen, Shuangyin \]^. “Tubo tongzhi dunhuang shiqi dongku xiujian jingji yuanyin chutan”
_`abcdef‰g(h=Zijkl [The preliminary study on the economic
reason of building caves during the Tibetan occupation period in Dunhuang]. Shijie
zongjiao yanjiu `mnYop 4 (2009): 32–40.
Chen, Zuolong \qQ. “Liu Sahe Yanjiu rksop.” In Dunhuang ziliao kaoxuecdtî

265
uv [Detailed examination of Dunhuang research materials] (Taipei: Shangwu
Yinshuguan, 1979), 212–52.
Ching, Dora C. Y., ed. Visualizing Dunhuang: Seeing, Studying, and Conserving the Caves.
Princeton: Princeton University Press, 2021.
Chou, Po-Kan wxy. “Zhongshi ji zhongguo zaijia pusa zhi chanfa: dui zDatong fangguang
chanhui miezui zhuangyan chengfo jing{ de kaocha ” |`}|~7jkÄˆ6: Å
z9Ù·ıˆ˜¯ÿ˘¿∫={ÇuÉ [An investigation of a confessional
ceremony in medieval Chinese buddhism]. Taida foxue yanjiu 18 (2009): 1–32.
Copp, Paul. The Body Incantatory: Spells and the Ritual Imagination in Medieval Chinese
Buddhism. New York: Columbia University Press, 2014.
Cosmo, Nicola Di and Denis Sinor, eds. Warfare in Inner Asian History (500–1800). Leiden:
Brill, 2002.
Debreczeny, Karl. “Faith and Empire: An Overview,” In Faith and Empire: Art and Politics in
Tibetan Buddhism, edited by Karl Debreczeny, 19–51. New York: Rubin Museum of Art,
2019.
Delbourgo, Suzy R. “Two Far Eastern Artefacts Examined by Scientific Methods.” International
Symposium on the Conservation and Restoration of Cultural Property. Conservation of
Far Eastern Art Objects. (1980): 163–79.
Dien, Albert E. “Turfan Funereal Documents.” Journal of Chinese Religions 30 (2002): 23–48.
Duan, Wenjie, ed. Dunhuang shiku quanji cd!göÑ [Complete works of Dunhuang caves].
28 vols. Hong Kong: Shangwu Yinshuguan, 1999.
Dunhuang wenwu yanjiusuo cd¢TopÖ. Zhongguo shiku: Dunhuang Mogaoku |Ü!g:
cdá>g [Chinese caves: Dunhuang Mogao Caves]. 5 vols. Beijing: Wenwu
chubanshe, 1982–1987.
Dunhuang yanjiuyuan cdopå, ed. Dunhuang shiku neirong zonglu cd!gàâäã
[Catalog of pictorial contents in cave temples of Dunhuang]. Beijing: Wenwu chubanshe,
1996.
———. Dunhuang shiku quanji cd!göÑ. 25 vols. Hong Kong: Shangwu Renshuguan,
2003.
———. “Xin faxian de Beiwei cixiu” ;åçÇéèêë [Newly discovered embroideries from
the Northern Wei]. Wenwu 2 (1972): 54–60.
Dunhuang yanjiuyuan, and Jiangsu meishu chubanshe, eds. Dunhuang shiku yishu: Mogao ku di
112 ku (zhong Tang). Nanjing: Jiangsu meishu chubanshe, 1998.
Eluosi guoli Aiermita shi bowuguan and Shanghai guji chubanshe. Eluosi gu li Ai’ermita shi
bowuguan cang Dunhuang yishu pin íIì~îïñßóSïTò5cdôöõ
[Dunhuang art relics collected in the State Hermitage Museum of Russia] 1. Shanghai:
Shanghai Guji Chubanshe, 1997.

266
Fan, Jinshi úqù, and Mei Lin ûü. “Mogaoku di 112 ku tuxiang zakao” á>g† 112g°
Y¢u [Miscellaneous studies on iconography of cave 112, Mogao Caves]. Dunhuang
yanjiu cdop 4 (1996): 5–7.
Feugère, Laure. “Introduction.” In Textiles from Dunhuang in French Collections, edited by
Zhao Feng, 14–23. Shanghai: Donghua University, 2011.
Fo Guang Shan International Translation Center, The Lotus Sutra’s Universal Gate Chapter on
Avalokiteśvara Bodhisattva. Hacienda Heights, Fo Guang Shan International Translation
Center, 2016.
Fraser, Sarah E. “Formulas of Creativity: Artist’s Sketches and Techniques of Copying at
Dunhuang.” Artibus Asiae 59, no. 3/4 (2000): 189–224.
———. Performing the Visual: The Practice of Buddhist Wall Painting in China and Central
Asia, 618–960. Stanford: Stanford University Press, 2004.
Fujieda, Akira. “Sutain Tonkō shūshū e-iri Kannongyō sasshi (S.6983): Tonkō ni okeru
mokuhitsu no shiyō.” Bokubi 177 (1968): 3–45.
Galambos, Imre. Dunhuang Manuscript Culture. Boston: De Gruyter, 2020.
Garcia, Violaine. “‘La quête des mérites’, conservation–restauration d’une banniere votive peinte
du Xe siècle provenant de Dunhuang (grottes de Mogao), conservée au Musée national
des arts asiatiques-Guitmet,” diss., Institut national de patrimoine, 2015.
Gasparini, Mariachiara. Transcending Patterns: Silk Road Cultural and Artistic Interactions
Through Central Asian Textile Images. Honolulu: University of Hawai’i Press, 2020.
Gell, Alfred. Art and Agency: An Anthropological Theory. New York: Oxford University Press,
1998.
Gerhart, Karen M. The Material Culture of Death in Medieval Japan. Honolulu: University of
Hawai’i Press, 2009.
Giès, Jacques, ed. Les arts de l’Asie centrale: La collection Paul Pelliot du musée national des
arts asiatiques-Guimet. Paris: Réunion des musées nationaux, 1995.
———, ed. The Arts of Central Asia: The Pelliot Collection in the Musée Guimet. London:
Serindia Publications, 1996.
Giès, Jacques, Hero Friessen, and Roderick Whitfield. Les arts de l’Asie centrale: la collection
Paul Pelliot du Musée national des arts asiatique-Guimet. 3 vols. London: Serindia,
1994.
Gropp, Gerd. Archäologische Funde aus Khotan Chinesisch-Ostturkestan: die Trinkler-
Sammlung im Ubersee-Museum, Bremen. Bremen: Röver, 1974.
Guan, Youhui. Dunhuang zhuangshi tu’an cd£§23 [Dunhuang decorative patterns].
Shanghai: Huadong shifan daxue chubanshe, 2010.
Gulik, Robert Hans van. Chinese Pictorial Art as Viewed by the Connoisseur: Notes on the
Means and Methods of Traditional Chinese Connoisseurship of Pictorial Art, Based upon

267
a Study of the Art of Mounting Scrolls in China and Japan. Rome: Istituto italiano per il
Medio ed Estremo Oriente, 1958.
Haesner, Chhaya. “Banners from Dunhuang and Turfan: A Comparative Study and Their Indian
Counterparts.” South Asian Archaeology 1995. Proceedings of the 13th Conference of the
European Association of South Asian Archaeologists 2 (1995): 715–30.
Hansen, Valerie. The Silk Road: A New History. New York: Oxford University Press, 2015.
Hansen, Valerie, and Helen Wang. “Introduction.” Journal of the Royal Asiatic Society 23, no. 2
(2013): 155–63.
Hao, Chunwen •¶¢. Dunhuang Manuscripts: An Introduction to Texts from the Silk Road.
Translated by Stephen F. Teiser. Diamond Bar: Portico Publishing Company, 2020.
———. Tang houqi Wudai Songchu Dunhuang sengni de shehui shenghuo ß®f©w©kc
d™UÇ´EÖ¨ [The social life of monks and nuns at Dunhuang in the late Tang,
Five-Dynasties period, and early Song]. Beijing: Zhongguo shehui kexueyuan, 1998.
Hickman, Money L. “Notes on Buddhist Banners.” Boston Museum Bulletin 71, no. 363 (1973):
4–20.
Ho, Puay-peng. “Building on Hope: Monastic Sponsors and Merit in Sixth- to Tenth-Century
China.” Asia Major 17, no. 1 (2004): 35–57.
Huang, Shih-shan Susan. “Daoist Uses of Color in Visualization and Ritual Practices,” In Color
in Ancient and Medieval East Asia, edited by Mary M. Dusenbury and Monica Bethe,
223–33. Lawrence: Spencer Museum of Art, the University of Kansas, 2015.
———. Picturing the True Form: Daoist Visual Culture in Traditional China. Cambridge:
Harvard University Press, 2012.
Hunan Sheng Bowuguan ≠ÏÆïTò [Hunan provincial museum]. Changsha Mawangdui
yihao Han mu œ%§F$Ø∞±≤. Beijing: Wenwu, 1973.
Huntington, John C. “The Technique of Tibetan Paintings.” Studies in Conservation 15, no. 2
(1970): 122–33.
Huo, Xiliang. “Mogaoku di 72 ku jiqi nanbi Liu Sahe yu Liangzhou Shengrongfo ruixiang shiji
Bian” á>g† 72g≥¥Ïµrks∂∑∏óπ∫òY•∫u [Cave 72 in the
Mogao grottoes and the mural on its south wall depicting Liu Sahe and the history of the
miraculous image of the holy-visage Buddha of Liangzhou]. Wenwu 2 (1993): 32–47.
Jackson, David, and Janice A. Jackson. Tibetan Thangka Painting: Methods and Materials.
Boulder: Shambhala, 1984.
Jera-Bezard, Robert, and Monique Maillard. “Origine et montage des bannières peintes de
Dunhuang.” Arts asiatiques 40 (1985): 83–91.
———. “Le rôle des bannières et des peintures mobiles dans les rituels du bouddhisme d’Asie
centrale.” Arts asiatiques 44, no. 1 (1989): 57–67.
Kapstein, Matthew. “Weaving the World: The Ritual Art of the Pata in Pala Buddhism and Its

268
Legacy in Tibet.” History of Religions 34 (1995): 241–62.
———. “New Light in an Old Friend: PT 849 Revisited,” In Tibetan Buddhist Literature and
Praxis: Studies in Its Formative Period, 900-1400, edited by C. Wedermeyer and R.
Davidson, 9–30. Leiden: Brill, 2006.
———, ed. Buddhism Between Tibet and China. Boston: Wisdom, 2009.
Kieschnick, John. “Blood Writing in Chinese Buddhism.” Journal of the International
Association of Buddhist Studies 23, no. 2 (2000): 177–94.
Kim, Haewon. “A History of the Central Asia Collection at the National Museum of Korea.”
International Journal of Korean Art and Archaeology 4 (2010): 159–72.
———, ed. Kungnip chungang bangmulgwan sojang cungang ashia chonggyo
hoehwa국립중앙박물관 소장 중앙아시아 종교 회화 [Central Asian religious paintings in the
National Museum of Korea]. Seoul: Kungnip chungang bangmulgwan, 2013.

Klimburg-Salter, Deborah. “The Tibetan Himalayan Style: The Art of the Western Domains, 8
th
–
11
th
Centuries.” Cultural Flows Across the Western Himalaya (2015): 427–75.
Klimburg-Salter, Deborah E., and Christian Luczanits. Tabo: A Lamp for the Kingdom. New
York: Thames & Hudson, 1997.
Kosek, Joanna, Jin Xian Qiu Qiu, Keisuke Sugiyama, Carol Weiss, David Saunders, Catherine
Higgitt, Janet Ambers, and Caroline Cartwright. “The Admonitions Scroll: Condition,
Treatment and Housing 1903–2014.” The British Museum Technical Research Bulletin 9
(2015): 25–43.
Kuo, Li-ying. Confession et contrition dans le bouddhisme chinois du Ve au Xe siècle. Paris:
Publications de l’École française d’Extrême-Orient, 1994.
Kyan, Winston. “Family Space: Buddhist Materiality and Ancestral Fashioning in Mogao Cave
231.” The Art Bulletin 92, no. 1–2 (2010): 61–82.
Lai, Guolong. “Colors and Color Symbolism in Early Chinese Ritual Art,” In Color in Ancient
and Medieval East Asia, edited by Mary M. Dusenbury, 25–43. New Haven and London:
Yale University Press, 2015.
Lee, Sonya S. “An Interview with Roderick Whitfield on the Stein Collection in the British
Museum.” The Silk Road 17 (2019): 10–25.
———. “Repository of Ingenuity: Cave 61 and Artistic Appropriation in Tenth-Century
Dunhuang.” Art Bulletin 94 (2012): 199–225.
Lehnert, Martin. "Tantric Threads Between India and China," In The Spread of Buddhism, edited
by Ann Heirman and Stephan Peter Bumbacher, 247–273. Leiden: Brill, 2007.
Li, Fuhua }ªº, and Jiang Dezhi ΩÓb, eds. Dunhuang renwu zhi cdæTø [Biographies
of people from Dunhuang]. Lanzhou: Gansu renmin chubanshe, 2009.
Li, Ling }¿. ““Yinlu pusa” yu “lianhua shou”: Han cang chi lianhua guanyin xiang bijiao ““∞

269
±jk”∂ “J)¡”: ±5¬J)_aY√ƒ” [“The soul-guiding bodhisattva” and
“Padmapāṇi”]. Xizang yanjiu ≈5op 3 (2006): 59–62.
Liu, Xinru. Silk and Religion: An Exploration of Material Life and the Thought of People, AD
600–1200. Oxford University Press, 1999.
Lojda, Linda, Deborah Klimburg-Salter, and Monica Strinu. “The Tibetan Himalayan Style:
Considering the Central Asian Connection,” In Transfer of Buddhism Across Central
Asian Networks (7th to 13th Centuries), edited by Carmen Meinert, 121–46. Leiden:
Brill, 2016.
Luczanits, Christian. “Ritual, Instruction and Experiment: Esoteric Drawings from Dunhuang,”
In The Art of Central Asia and the Indian Subcontinent in Cross Culture Perspective,
edited by Anupa Pande and Mandira Sharma, 140–49. New Delhi: National Museum
Institute-Aryan Books International, 2009.
Ma, De ∆Ó. Dunhuang gudai gongjiang yanjiu cd«wÄÅop [Dunhuang ancient
craftsman research] Beijing: Wenwu chubanshe, 2018.
Mair, Victor H. “The Origins of an Iconographical Form of the Pilgrim Hsüan-tsang.” Tang
Studies 4 (1986): 29–41.
———. T’ang Transformation Texts: A Study of the Buddhist Contributions to the Rise of
Vernacular Fictions and Drama in China. Cambridge: Harvard University Press, 1989.
Matsumoto Eiichi »8…Ø. Tonko-ga no kenkyu cd Àop [A study of Dunhuang wall
paintings]. Tokyo: Toho bunka gakuin, 1937.
McNair, Amy. Xuanhe Catalogue of Paintings: An Annotated Translation with Introduction.
Ithaca: Cornell East Asia Series, 2019.
Meinert, Carmen, ed. Transfer of Buddhism Across Central Asian Networks (7th to 13th
Centuries). Leiden: Brill, 2016.
Miller, Daniel. “Materiality: An Introduction,” In Materiality, edited by Daniel Miller, 1–50.
Durham: Duke University Press, 2005.
Mollier, Christine. Buddhism and Taoism Face to Face: Scripture, Ritual, and Iconographic
Exchange in Medieval China. Honolulu: University of Hawai’i Press, 2008.
Murase, Miyeko. “Kuan-Yin as Savior of Men: Illustration of the Twenty-Fifth Chapter of the
Lotus Sūtra in Chinese Painting.” Artibus Asiae 33, no. 1/2 (1971): 39–74.
Ng, Zhiru. The Making of a Savior Bodhisattva: Dizang in Medieval China. Honolulu:
University of Hawaii Press, 2007.
Ning, Qiang. Art, Religion, and Politics in Medieval China: the Dunhuang Cave of the Zhai
Family. Honolulu: University of Hawai’i Press, 2004.
Pal, Pratapaditya. The Art of Tibet: The Meditation and the Rituals. New York: The Asia
Society, 1969.
Papa-Kalantari, Christiane. “The Ceiling Paintings of the Alchi gsum brtsegs: Problems of

270
Style,” In Buddhist Art and Tibetan Patronage, edited by Deborah Klimburg-Salter and
Eva Alinger, 85–113. Leiden and Boston: Brill, 2002.
Reynolds, Valrae. “Luxury Textiles in Tibet,” In Tibetan Art: Towards a Definition of Style,
edited by Jane C. Singer and Philip Denwood, 118–32. London: Laurence King, 1997.
Rhie, Marylin M. “Tibetan Painting Styles, Sources, and Schools,” In Worlds of Transformation:
Tibetan Art of Wisdom and Compassion, edited by Marylin M. Rhie and Robert A.
Thurman, 45–74. New York: Tibet House, 1999.
Robson, James. “Signs of Power: Talismanic Writing in Chinese Buddhism.” History of
Religions 48, no. 2 (2008): 130–69.
Rong, Xinjiang. “The Nature of the Dunhuang Library Cave and the Reasons for its Sealing.”
Cahiers d’Extrême-Asie 11, no. 1 (1999): 247–75.
Rong, Xinjiang Ã;Õ, and Zhu Lishuang Œœ]. Yutian yu Dunhuang –—∂cd [Khotan
and Dunhuang]. Lanzhou: Gansu jiaoyu chubanshe, 2013.
Seung, Jae Hee. “T’urŭp’an Asŭt’ana, K’arahoja kobun ch’ult’o pogŭi yŏwa do tosang yŏn’gu”
[Iconography of Fuxi and Nüwa paintings from Astana, Turfan, and Karahoja].
Chungang ashia yŏn’gu 8 (2003): 131–49.
Schäfer, Dagmar, Giorgio Riello, and Luca Molà, eds. Threads of Global Desire: Silk in the Pre-
Modern World. Woodbridge: The Boydell Press, 2018.
Sha, Wutian %“”. Dunhuang huagao yanjiu cd]‘op [Studies on the sketches of
Dunhuang]. Beijing: Minzu chubanshe, 2006.
———. “Dunhuang shiku Yutian guowang huaxiang yanjiu” cd!g–—F]Yop
[Studies in the image of the king of Khotan at the Dunhuang caves]. XInjiang shifan
daxue xuebao ;’ù÷9◊◊ÿ 27/4 (2006): 22–30.
———. Tubo tongzhi shiqi Dunhuang shiku yanjiu _`abŸfcd!gop [Research on
the caves of Dunhuang during the Tibetan occupation period]. Beijing: Zhongguo Shehui
Kexue chubanshe, 2011.
Shaftel, Ann. “Conservation Treatment of Tibetan Thangkas.” Journal of the American Institute
for Conservation 30, no. 1 (1991): 3–11.
———. “Notes on the Technique of Tibetan Thangkas.” Journal of the American Institute for
Conservation 25 (1986): 97–103.
Sharf, Robert H. “Art in the Dark: The Ritual Context of Buddhist Caves in Western China,” In
Art of Merit: Studies in Buddhist Art and Its Conservation, edited by David Park, Kuenga
Wangmo, and Sharon Cather, 38–65. London: Archetype Publications, 2013.
Sheng, Angela. “Determining the Value of Textiles in the Tang Dynasty in Memory of Professor
Denis Twitchett (1925–2006).” Journal of the Royal Asiatic Society 23, no. 2 (2013):
175–95.
Shimono, Akiko ¡⁄¤‹. “Tonkō Bakukōkutsu dai 217 kutsu nanheki kyōhen no shin

271
kaishaku” cdá>g †›Ø«gÏµ=uÀ;fifl [New interpretation on south
wall of Cave 217 of Mogao Caves, Dunhuang]. Bijutsushi ‡ö• 54, no. 1 (2004): 96–
115.
Shinohara, Koichi. “The Moment of Death in Daoxuan’s Vinaya Commentary,” In The Buddhist
Dead: Practices, Discourses, Representations, edited by Bryan J. Cuevas and Jacqueline
I. Stone, 105–33. Honolulu: University of Hawai’i Press, 2007.
Silbergeld, Jerome. Chinese Painting Style: Media, Methods, and Principles of Form. Seattle and
London: University of Washington Press, 1982.
———. “Mawangdui, Excavated Materials, and Transmitted Texts: A Cautionary Note.” Early
China 8 (1982): 79–92.
Soper, Alexander C. “Literary Evidence for Early Buddhist Art in China.” Artibus Asiae.
Supplementum 19 (1959): 1–296.
Sørensen, Henrik H. “Donors and Image in Dunhuang: A Case Study of OA 1919,0101,0.54.”
Buddhistroad Paper 4, no. 1 (2019): 3–25.
———. “Giving and the Creation of Merit: Buddhist Donors and Donor Dedications from 10th
Century Dunhuang.” BuddhistRoad Paper 4, no. 3 (2020): 3–40.
———. “Perspectives on Buddhism in Dunhuang During the Tang and Five Dynasties Period,”
In The Silk Roads: Highways of Culture and Commerce, edited by Vadime Elisseeff, 27–
48. Paris: UNESCO Publishing, Berghahn Books, 2000.
Stein, Aurel. Innermost Asia: Detailed Report of Explorations in Central Asia, Kan-su and
Eastern Īrān. Oxford: The Clarendon Press, 1928.
———. Serindia: Detailed Report of Explorations in Central Asia and Westernmost China. 5
vols. Oxford: The Clarendon Press, 1921.
Stoddard, Heather. Early Sino-Tibetan Art. Bangkok: Orchid Press, 2008.
Tasaki, Chiharu ”·‚¶. Tōrufan, Tonkō no bukkyō kaiga no kiteīzai ni tsuite„‰ÂÊÁË
・ cdÀÈYÍ ÀÎÏÌÓÔÒ [On the base material of Turfan and Dunhuang
Buddhist paintings]. Tokyo: Fuji Xerox Kobayashi Setsutarō Commemoration Fund,
2012.
Tong, Tao and Patrick Wertmann. “The Coffin Paintings of the Tubo Period from the Northern
Tibetan Plateau,” In Bridging Eurasia, edited by M. Wagner and Wei Wang, 187–213.
Mainz: German Archaeological Institute, 2010.
Tsukamoto Zenryū, “Hikiji Bosatsu shinkō ni tsuite” ∞±jkHÚÓÛÒ [A study of Yinlu
Bodhisattva belief], Tōhō gakuhō Ù·◊ÿ 1 (1931).
van Schaik, Sam. “The Naming of Tibetan Religion: Bon and Chos in the Tibetan Imperial
Period.” Journal of the International Association for Bon Research 1 (2013): 227–57.
———. “The Tibetan Avalokiteśvara Cult in the Tenth Century: Evidence from the Dunhuang
Manuscripts,” In Tibetan Buddhist Literature and Praxis: Studies in Its Formative

272
Period, 900–1400, edited by C. Wedermeyer and R. Davidson, 55–72. Leiden: Brill,
2006.
———. “Towards a Tibetan Paleography: Developing a Typology of Writing Styles in Early
Tibet,” In Manuscript Cultures: Mapping the Field, edited by Jörg Quenzer, Dmitry
Bondarev, and Jan-Ulrich Sobisch, 299–340. Berlin: De Gruyter, 2014.
van Schaik, Sam, and Imre Galambos. Manuscripts and Travelers: The Sino-Tibetan Documents
of a Tenth-Century Buddhist Pilgrim. Boston: De Gruyter, 2011.
van Schaik, Sam, Agnieszka Helman-Ważny, and Renate Nöller. “Writing, Painting and
Sketching at Dunhuang: Assessing the Materiality and Function of Early Tibetan
Manuscripts and Ritual Items.” Journal of Archaeological Science 53 (2015): 110–32.
Vitali, Roberto. Early Temples of Central Tibet. London: Serindia, 1990.
Von Glahn, Richard. The Economic History of China: From Antiquity to the Nineteenth Century.
Cambridge: Cambridge University Press, 2016.
Waley, Arthur. A Catalogue of Paintings Recovered from Tun-huang by Sir Aurel Stein, K.C.I.E.
London: British Museum and the Government of India, 1931.
Wang, Binghua. “A Study of the Tang Dynasty Tax Textiles (Yongdiao Bu) from Turfan.”
Journal of the Royal Asiatic Society 23, no. 2 (2013): 263–80.
Wang, Helen. Money on the Silk Road: The Evidence from Eastern Central Asia to c. AD 800.
London: British Museum Press, 2004.
———. “Textiles as Money on the Silk Road?” Journal of the Royal Asiatic Society 23, no. 2
(2013): 165–74.
Wang, Helen, Helen Persson, and Frances Wood. “Dunhuang Textiles in London,” In Textiles
from Dunhuang in UK Collection, edited by Zhao Feng, 12–21. Shanghai: Donghua
University Press, 2007.
———. “Dunhuang Textiles in London: A History of the Collection,” In Sir Aurel Stein:
Colleagues and Collections, edited by Helen Wang, 1–8. London: British Museum
Research Publication, 2012.
Wang, Le Fı, and Zhao Feng ˆ˜. “Dunhuang fan de shiwu yu fenxi ” cdVÇ¯T@˘˙
op [Study and analysis of existing fabric of the Dunhuang banners]. Dunhuang yanjiu
cdop 1 (2008): 1–9.
Wang, Michelle C. “From Dhāraṇī to Maṇḍala: A Study of Mogao Cave 14 and Esoteric
Buddhist Art of the Tang Dynasty (618–907),” PhD diss., Harvard University, 2008.
———. Maṇḍalas in the Making: The Visual Culture of Esoteric Buddhism at Dunhuang.
Leiden: Brill, 2018.
———. “The Thousand-armed Mañjuśrī at Dunhuang and Paired Images in Buddhist Visual
Culture.” Archives of Asian Art 66, no. 1 (2016): 81–105.
Wang, Michelle C., Xin Wen, and Susan Whitfield. “Buddhism and Silk: Reassessing a Painted

273
Banner from Medieval Central Asia in The Met.” The Metropolitan Museum Journal 55
(2020): 8–25.
Wang, Ming F˚. “Pusa yinlu: Tang Song shiqi sangzang yishi zhong de yinhun fan” jk∞
±: ß©ef¸˝V˛|Ç∞ˇV [Road-guiding bodhisattva: Soul-guiding flags in
funeral rituals in the Tang and Song dynasties]. Dunhuang yanjiu cdop 1 (2014):
37–45.
Waring, Luke. “Writing and Materiality in the Three Han Dynasty Tombs at Mawangdui,” PhD
diss., Princeton University, 2019.
Whitfield, Roderick. The Art of Central Asia: The Stein Collection in the British Museum. 3 vols.
Tokyo: Kodansha International, 1982–5.
———. “Four Unpublished Paintings From Dunhuang in the Oriental Collections of the British
Library.” The British Library Journal 24 (1998): 90–97.
Whitfield, Susan. The Silk Road: Trade, Travel, War and Faith. London: British Library, 2004.
Winter, John. East Asian Paintings: Materials, Structures and Deterioration Mechanisms.
London: Archetype Publications, 2008.
———. “‘Lead White’ in Japanese Paintings.” Studies in Conservation 26, no. 3 (1981): 89–
101.
Wong, Dorothy C. “Guanyin Images in Medieval China, Fifth to Eighth Centuries.” Bodhisattva
Avalokiteśvara (Guanyin) and Modern Society (2007): 254–302.
Wong, Lori, and Neville Agnew, eds. The Conservation of Cave 85 at the Mogao Grottoes,
Dunhuang: A Collaborative Project of the Getty Conservation Institute and the
Dunhuang Academy. Los Angeles: Getty Conservation Institute, 2014.
Wu, Hung. “Art in a Ritual Context: Rethinking Mawangdui.” Early China 17 (1992): 111–44.
———. “Rethinking Liu Sahe: The Creation of a Buddhist Saint and the Invention of a
‘Miraculous Image’” Orientations 27, no. 10 (1996): 32–43.
———. “What is Bianxiang?: On the Relationship Between Dunhuang Art and Dunhuang
Literature.” Harvard Journal of Asiatic Studies 52, no. 1 (1992): 111–92.
Xie, Jisheng !"#, and Zhao Yuan ˆ$. “Mogaoku tubo yangshi bihua yu juanhua de chubu
fenxi" á>g_`%˛µ]∂[]Çk&@˘ [A preliminary analysis of Tibetan style
murals and silk paintings in Mogao Caves]. Xibei minzu daxue xuebao ≈é'(9◊◊
ÿ 4 (2010): 65–73.
Xie, Shengbao !Ö), and Xie Jing !*. “Dunhuang wenxian yu Shuilu fahui: Dunhuang
Tang Wudai shiqi Shuilu fahui yanjiu” cd¢+∂,o6E: cdß©wef,o6
Eop [Dunhuang manuscripts and the study of the ceremony of saving lives of the
Water and the Land during the Tang Dynasty to the Five Dynasties]. Dunhuang yanjiu c
dop 2 (2006): 40–48.
———. “Dunhuang yihua yu Shuilu hua: Dunhuang Tang Wudai shiqi Shuilu fahui yanjiu zhi

274
er” cd-]∂,o]: cdß©wef,o6EopÄ› [Dunhuang manuscripts
and the study of the ceremony of saving lives of the Water and the Land during the Tang
Dynasty to the Five Dynasties, second part]. Dunhuang yanjiu cdop 4 (2006): 1–7.
Xu, Ziqiang ./0, Zhang Yongqiang zO0, and Chen Jing \1. Dunhuang Mogaoku ti ji
hui bian cdá>g2y34 [Dunhuang Mogao Caves inscription compilation].
Beijing: Wenwu chubanshe, 2014.
Yim, Yŏng-ae. “Chungguk kodae pulgyobŏnŭi yangshik pyŏnch’ŏn’go” «w |~ ∫5V의 %
˛u6u [Stylistic changes in Buddhist banners from the anceint China]. Misul sahak
yŏn’gu 189, no. 3 (1991): 69–109.
Yiyuan duoying 789: 9 (1980): 6.
Yu, Jimmy. Sanctity and Self-Inflicted Violence in Chinese Religions, 1500–1700. New York;
Oxford: Oxford University Press, 2012.
Yuan, Ting ;<. Dunhuang zangjingdong chutu huihuanpin yanjiushi cd5=‰CM=]õ
op• [Historiography of paintings from the library cave of Dunhuang]. Lanzhou:
Gansu jiaoyu chubanshe, 2016.
Zhang, Xiaogang zn∂. “Dunhuang ruxiang tu zhong de Yutian huguo shen wang” cdòY
2|Ç–—>~“F [Protection gods of Khotan found in auspicious image paintings
of Dunhuang]. Dunhuang yanjiu 1 (2005): 50–56.

Zhang, Yuanlin zªü, and Xia Shengping ?Ö@. “‘Guanyin jiunan’ de xingxiang tushi:
Mogaoku di 359 ku xibi kan nei pingfeng hua neirong shidu” ‘_a¨A’Ç.˚2B: á
>g† 359 g≈µCàDE]àâflF [Illustration to Guanyin’s story of saving life
from sufferings: An interpretation of the screen paintings inside the niche of Cave 359
Mogao Grottoes]. Dunhuang yanjiu cdop 5 (2010): 36–46.
Zhao, Feng ˆ˜. Dunhuang sichou yu sichou zhi lu cd0G∂0GÄ± [Dunhuang silk and
Silk Road]. Beijing: Zhonghua shuju, 2009.
———, ed. Textiles from Dunhuang in French Collections. Shanghai: Donghua University
Press, 2011.
———, ed. Textiles from Dunhuang in Russian Collections. Shanghai: Donghua University
Press, 2014.
———, ed. Textiles from Dunhuang in UK Collections. Shanghai: Donghua University Press,
2007.
Zhao, Feng, and Le Wang. “Glossary of Textile Terminology (Based on the Documents from
Dunhuang and Turfan).” Journal of the Royal Asiatic Society 23, no. 2 (2013): 349–87.
———. “Reconciling Excavated Textiles with Contemporary Documentary Evidence: a Closer
Look at the Finds from a Sixth-Century Tomb at Astana.” Journal of the Royal Asiatic
Society 23, no. 02 (2013): 197–221.

275
Zhao, Feng, Le Wang, and Xu Zheng. “A Typological and Chronological Study of Dunhuang
Textiles,” In Textiles from Dunhuang in UK Collections, edited by Zhao Feng, 22–27.
Shanghai: Donghua University Press, 2007.

Appendix: A List of Long Banners from the Library Cave (Cave 17), Mogao Caves of Dunhuang

No. Museum Number Title Dimension Material Date Institution
1 1919,0101,0.121
Bodhisattva of the
Sun
H. 213.0 x W. 25.5 cm Colors on silk 10
th
century CE British Museum
2 1919,0101,0.195
Long Banner of
Seated Buddhas
H. 538.0 x W. 18.0 cm Ink on silk
Five Dynasties,
mid-10th century
CE
British Museum
3 1919,0101,0.196
Long Banner of
Bodhisattvas
H. 341.0 x W. 28.5 cm Ink and colors on silk 956 (?) British Museum
4 1919,0101,0.205
Long Banner of Nine
Bodhisattva
H. 835.0 x W. 28 cm
Ink on yellow silk, red
silk on the top edge
926–975 British Museum
5 1919,0101,0.214.1
Long Banner of
Bodhisattvas
H. 339.0 x W. 30.0 cm
Color on dark blue silk
(painted in silver)
Five Dynasties,
mid-10th century
CE
British Museum
6 1919,0101,0.214.2
Long Banner of
Bodhisattvas
H. 347.5 x W. 28.5 cm
Color (yellow, not
gold) on dark blue silk
Five Dynasties,
mid-10th century
CE
British Museum
276

7 1919,0101,0.215
Long Banner of
Bodhisattvas
H. 409.0 x W. 30.0 cm
Ink and colors on blue
silk
10
th
century CE British Museum
8 1919,0101,0.216
Long Banner of Nine
Bodhisattvas
H. 396.0 x W. 59.0 cm Ink and colors on silk 956 British Museum
9 1919,0101,0.217 Standing Bodhisattva H. 386.0 x W. 68.1 cm
Silver on red damask
silk
Late 9
th
–early 10
th

century CE
British Museum
10 EO. 1137 Bodhisattva H. 183.2 x W. 58.0 cm
Ink and colors on
vermilion silk
Tang Dynasty
(second half of the
9th century)
Musée Guimet
11
EO 1418 Bodhisattva H. 196.0 x W. 57.0 cm Ink and colors on silk 9
th
–10
th
century CE Musée Guimet
12 EO 1398 Bodhisattva H. 43.7 x W. 31.9 cm Ink and colors on paper 9
th
–10
th
century CE Musée Guimet
13 EO. 3647 Bodhisattvas H. 580 x W. 32.9 cm
Ink and colors on navy
silk
9
th
–10
th
century CE Musée Guimet
14 EO. 3648 Bodhisattva H. 915.0 x W. 31.0 cm
Ink and colors on dark
blue silk
9
th
–10
th
century CE Musée Guimet
15 EO 3657/1 Bodhisattva H. 185.0 x W. 54.4 cm
Ink and colors on
vermilion silk
9
th
–10
th
century CE Musée Guimet
16 EO 3657/2 Bodhisattva H. 172.0 x W. 56.0 cm Ink and colors on silk 9
th
–10
th
century CE Musée Guimet
17 EO 3657/3bis Bodhisattva H. 161.0 x W. 59.0 cm Ink and colors on silk 9
th
–10
th
century CE Musée Guimet
277

18 MG 17675
Fragment of a Long
Banner of Seated
Buddhas
H. 37.6 x W. 19.5 c, Ink on silk
Second half of the
10
th
c.
Musée Guimet
19 MG 17791
Fragment of a Long
Banner of Seated
Buddhas
H. 66.0 x W. 19.0 cm Ink on silk
Second half of the
10th c.
Musée Guimet
20 EO 1165
Six Buddhas Seated
on Lotus Blossom
H. 303.0 x W. 19.2 cm Ink on silk 9
th
–10
th
century CE Musée Guimet
21 EO 1166
Long Banner with
Five Apsaras
H. 301.0 cm x W. 19.6
cm
Ink and colors on silk 9
th
–10
th
century CE Musée Guimet
22 Bon 4025
Banner of
Bodhisattvas
H. 180.4 x W. 28.2 cm Color on silk 10
th
century CE
National Museum of
Korea
23 Bon 4020
Banner of
Bodhisattva
H. 323.0 x W. 27.0 cm Color on navy silk 10
th
century CE
National Museum of
Korea
24 Bon 4019
Banner of
Bodhisattva
H. 165.0 x W. 36.8 cm Color on navy silk 10
th
century CE
National Museum of
Korea
25 Bon 4043
Banner of
Bodhisattvas
H. 371.0 x W. 28.5 cm Color on navy silk 10
th
century CE
National Museum of
Korea
26 Bon 4022
Banner of
Bodhisattvas
H. 191.5 x W. 25.2 cm Ink on yellow silk 10
th
century CE
National Museum of
Korea
27 Bon 4023
Banner of
Bodhisattvas
H. 205.0 x W. 34.5 cm Ink on yellow silk 10
th
century CE
National Museum of
Korea
278

28 Д x 58 Bodhisattva H. 114.0 x W. 26.5 cm
Blue silk with line
drawing
mid-10th century
CE
State Hermitage
Museum of Russia
27 Д x 59, 115, 39 Bodhisattva
H. 45.0 x W. 31.0 cm
H. 20.0 x W. 31.0 cm
Blue silk with silver
line drawing
mid-10th century
CE
State Hermitage
Museum of Russia
29 Д x 291
Streamer with
Avalokiteśvara
H. 48.0 x W. 37.0 cm
Red silk with line
drawing
Five Dynasties,
mid-10th century
CE
State Hermitage
Museum of Russia

279

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A grain of sand: Yingzao fashi and the miniaturization of Chinese architecture

PDF

Collaboration in color: master printers, dealer-publishers, and avant-garde lithography in late nineteenth-century France

PDF

Art into everyday life: department stores as purveyors of culture in modern Japan

PDF

Making sense of sacrifice: Sensory experience in Greco-Roman cult

PDF

How to be a social media influencer: the work process of the entrepreneurial labor force

PDF

Interpreting historic murals: a case study approach and best practices used in Fengguo Temple

PDF

Computational materials design for organic optoelectronics

PDF

Effects of cationic species on the structure and properties of hybrid organic-inorganic materials

PDF

Walking out of ground zero: art and the aftereffects of the atomic bombs in postwar Japan

PDF

Dollplay: narrative rituals in nineteenth-century Britain

PDF

Expanding the synthesis space of 3D nano- and micro-architected lattice materials

PDF

Vibrational spectroscopy and molecular orientation at interfaces: correlating the structure of a material's surface to its properties

PDF

Making Manhattan: sovereignty, material culture, and the politics of place, 1614-1664

PDF

The French Romanesque portals of Moissac, Souillac and Beaulieu: a response to the papal reform movement and popular heresy

PDF

The role of Rho-GEF signaling in synapse function and autism-related disorders

PDF

The impact of minimum wage on labor market dynamics in Germany

PDF

The origin and development of continuous narrative in Roman art, 300 B.C. - A.D. 200

PDF

Columella and the cultivation of empire

PDF

Sheila Hicks: weaving to the world

PDF

Self on the move: lyrical journeys in the twentieth-century Russian poetry

Asset Metadata

Creator Hwang, Yoon Ah (author)

Core Title Projecting wishes onto flying banners: materiality and painted banners from Cave 17 in the Mogao Caves of Dunhuang, China, in the ninth and tenth centuries

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Art History

Degree Conferral Date 2021-08

Publication Date 07/16/2022

Defense Date 05/06/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Cave 17,Dunhuang,Library Cave,materiality,Mogao Caves,OAI-PMH Harvest,painted banner,Silk,textile

Format application/pdf (imt)

Language English

Advisor Lee, Sonya S. (committee chair), Lai, Rongdao (committee member), Yasin, Ann Marie (committee member)

Creator Email hwang.yoonah@gmail.com,yoonahhw@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15611966

Unique identifier UC15611966

Legacy Identifier etd-HwangYoonA-9784

Document Type Dissertation

Format application/pdf (imt)

Rights Hwang, Yoon Ah

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu