Behrendt The Buddhist Architecture of Gandhara PDF
Behrendt The Buddhist Architecture of Gandhara PDF
NA6010.72.G36B44 2003
722'.4-dc22
2003045 119
'
ISSN 0169-9377
ISBN 90 04 13595 2
IQ Coj!)lrig!Lt 2003 by KoninJdijke .Brill NV, L.eiden, V.e Netherlmzds
Copyritted material
CONTENTS
Copyrighted material
VIJJ CONTENTS
Copyrighted mate
i .
.
CONTENTS 1X
Copyrighted material
l
X CONTENTS
Copyrighted mater I
.
CONTENTS XI
8.2.2
T a kbt-i-biibi: Sacred Area XX Loose
Sculpture · .................. .............. .. ... .. ............... ...... 218
8.2.3 Takht-i-babi: T wo-Celled Shrine XXIII:
Loose Sculpture ........ ......... .. ....... ....... .. ............. 219
8.2.4 Takht-i-bahl Small Sacred Area XIV:
Loose and In Situ Sculpture .... ........................ 220
8.3 Loose Sculptura l Finds fro m the Sahrr-Bah.lol
Sites I I I I I I I I I I I I I I I I I 0 I I I 0 I I I I I 0 I I ettttt00 0 , . 0 0 0 0 0 0 0 0 0 0 0 0 •• , ! ! ! 0! 0 0 0 0 0 0 f 0 f f f!!!!!! 221
8.3. 1 Sahri-Balu ol B: Loose Sculpture ...................... 222
8.3.2 Sahrr-Bahl61 C: Loose Sculpture ........ .............. 222
8.3.3 Sahrr-Ba hJ61 D: Loose Sculpture .................... 223
8.4 Thareli: Loose Sculpture .. .. ............. ..... ............ ... .. ....... 224
8.5 M ekhasanda: Loose Sculpture ....... ............................... 226
8.6 Ranigat: Loose Sculpture ...... .. .... ............................. .. ... 226
8. 7 I n Situ Imagery from Sites in the Peshawar
Bas in a nd Swii.t .................... .......................................... 228
8. 7.1 Sikri: In Situ Sculpture and the
Architectural Organization of the Site ..... ..... .. 228
8. 7.2 Sahn-Balllol: Some Possible In Situ
Sculpture .. ......... ....... ................................... .. .. ... 230
8. 7.3 Mekhasanda: In Situ Sculpture ........................ 231
8. 7.4 T hardi: In Situ Sculpture ........ ...................... .. 232
8. 7.5 Sa idu: In Situ Sculpture ............................. ... .. 233
Chapter ine: Buddhist Architecture and Sculpture
of Gandhiira: Conclusions .............. .. .. ........................ .. .. .. .. .. .. 234
9, 1 Phase l ......................... ,,,,,,,,,.,,............................ 23:'>
9.2 Phase n .......................................................................... 237
9.3 Phase m ...... .. .. ............................ .................................. 244
9.4 Late Phase IJI and P hase IV .................................... .. 2.53
Appendix A: The Four-Phase Chronological System ............. . 255
A I Phase I .............................. ...... .................................. .. .. .. 256
A I . I P hase I Dating Evidence from Sirkap ........... . 256
A 1.2 Phase I Character of Masomy in T axila .. .... .. 258
A 1.3 Phase I Structural T es ................................. . 259
A2 Phase II .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . . .. . 259
A2. 1 Phase H Dating Evidence .......................... .. .... 259
A2.2 Phase II Character of Masomy in T axiJa ..... . 260
A2.3 Phase li Struc tural T ypes .......................... ...... 261
A3 Phase m . . ?62
A3. I Phase Ill Dating Evidence ............... .. ... .. .... .... 262
Copyrighted material
'
..
Xll CONTENTS
Copyrighted materi I
CONTENTS XJ]]
D4 Thareli oo o o 0 0 0 o o 0 0 oo o I o I o o o o 0 o o o o o o + o o o o o + o 0 o o o o o o o o o o o o o o o o o o o o + o o o o oo o o o o o o o o o o o o 302
D5 M e khasanda t' t t t' 't I t 0
ID" , ' ' ' ' t I t 0 303
D6 Rani at •••••••••••• 0 ••••••• • ••••••••••••• 0 •••••••••••••• • ••• • •••••••••• • • • • ••• ••• 303
Glossar ...................................................................................... 305
Biblio. 31 1
Index I 0 0 I 01 I I 0 I 0 " 100 I DO I 0 I I I 0 I I I I I I I I I I I I I I I I I I 'P' I I ft I I I I I I I I I I I ft00e0tt0fttt0t00000! f , •• 0 f 0 0 0 0 f 0 323
Copyrighted material
UST OF ILLUSTRATIO NS
Sirkap
10. Plan of part of the city of Sirkap, Ta.xila (modified from Mars hall,
1951, pi. I0).
Copyrighted material
.
XVI LLST OF "lLLUSTRA"riONS
lvfohrii lvfaliii.rii.fi
13. Plan of Mohra Maliarail A temple, Taxila. A- front porch, B-
side rooms, C---rear cella, D-circumambulatory path, P plat-
form (modified from Marshall, 1951 , pi. 120 A).
]wuf.iiil Sites
14. Plan ofj aJ)t;lial C temple, T ax.ila (modified from Marshal!, 1951,
pl. 44). I
15. Plan of j aQt;lial B sacred area, Taxila. Note the plinth and shrine
in court T (modified from Marsh all, 1951 , pl. 91 ).
~
Dlzarmariijikii Complex
16. Plan of D harmarajika monastic area, Taxila (modified from
Marsh all, 1951, pl. 61 ).
I 7. Elevation of Dharmarajika main stii.pa from east, Tax.ila. ote I
•
the east image shrine. Inset shows the tr.ilobe trapezoidal niches I
•
(modified from M arshall, 1951 , pi. 46b, c).
18. K l stii.pa, Dharmarajika complex, Taxila, detail showing a seated
Buddha on the east facade in an axial trilobed niche. The
I
ma~onty of tlus stupa was produced during the transitional period
from diaper to semi-ashlar (Marsha.U, 1916, pl. 12b).
19. Dharmarajika complex, T axila, N 18 shrine. This double-semi-
II
ashlar image shrine contains the feet of a monumental stucco .I
•
Buddha..• as well as an added seated Buddha to the left and an '
•
added standing Buddha to the right (MarshaU, 1918, pL 2).
Kii.lawiin
20. Plan of Kalawan sacred area and monasteries, Truci1a, wid1 phase
ll and phase ill construction indicated (modified from Marshall,
1951 ' pl. 72).
22. Plan of Akhauri B (Chir tope B), sacred area and monastery,
Taxila (modified from Marshall, 195 1, pl. 67b).
23. Small copper seated bodhisattva, found in cell 18 of the Akhaurf
B monastery, Taxila (ASIDGS 1921 - 22 no. 74, courtesy of the
British Library).
24. Plan of Akhauri C (Chir tope C), sacred area and monastel)',
Taxila (modified from Marshal!, 195 1, pl. 68a).
25. Plan of K hader M ohra D l , sacred area and monastery, Tax1Ja
(modified from Marshall 1951, pl. 68b).
26. Plan of Khader Mo~·a D 2, sacred area and monastery, Tax:ila
(modified from Marshall 195 1, pi. 69a).
}auliiiii
27 . J auliaf\ plan of sacred area and monastery, T axila, showing all
period~ of construction (modified from Marshall, 1951 , pl. 10 I).
28. J aulian D I stfipa, Tax:ila, west face showing phase ill stepped
format with rows of stucco figures including atlantes, lions, ele-
phants, and Buddhas exhibiting dharmacakra mudrii and dhyiina
mudrii (photo by K. Behrendt).
29. E l gandhakuti niche in the entranceway to the J auliaii monastery,
Taxila ("Warburg Institute).
30. In situ schist plaque showing the Buddha flanked by worship-
pers, J aulian monastery niche in cell 2, Taxila ('.Varburg Institute).
31. Trapezoidal doorway in thejauliafi monastery, T ax:ila (ASIDGS
1917- 18 no. 158, courtesy of the British Libral)1) .
Mohrii. Moradu
32. Plan of Mohra M oradu sacred area and monastery, Taxila
(modified from Marshall, 195 1, pi. 93).
33. General view of Mohra Moradu main stupa and uihii.ra, T axila
(photo by K. Behrendt).
34. Over-life-size stucco seated Buddha that was added to the plinth
of the Mohra Moradu main stupa, Taxila; note the depiction of
the alms bowl (the Alkazi Collection of Photography).
35. Multiple additive stucco sculptures attached to the side of the stair-
way of the main stupa of Mollpl Moradu, T axila (Warburg Institute).
• 36. Two over-life-size stucco seated Buddha images on pedestal B2
in fi·ont of Mohra Moradu monastery cell 2, Taxila (Warburg
Institute).
37. Gandhaku# image niche S7, from the entranceway of the Mohra
Moradu monastery, Taxila (Warburg Institute).
Pippala
39. Plan of the Pippala sacred area and monastery, Taxila, with .
phase II and m structures indicated (modified from MarshaU, .
1951 , pi. 98a).
40. Small stiipa in cell 31 of the Pippala monaste1y, Taxi la (ASIDGS
1921 - 22, no. 79, courtesy of the British Library).
Kuniila
41. Plan of Kunala stiipa and monastety, TaJ::iJa (modified from
MarshaU, 1951, pi. 86).
Bhamiila
42. Plan of Bhamala sacred area and monastery, Taxila (modi11ed
from Marsh all, 1951, pi. 114).
43. Stucco Paranirvii1.w image found in situ, southeast corner of base .
plinth of Bhamilla main stiipa, Tax.iJa; length is approximately
1.3 m (line drawing by W . Hipsman after Marshall, 1951, pl.
118b).
T akltt-i-biihi
44. Plan of Takht-i-bahr main sacred area and outlying mountain
v£hiiras, Peshawar basin (modified from Hargreaves, 1914, pi.
XVTI and Bulletin qf the Research. Center for Silk Roadology, 2000: 41 ).
45. General view of main sacred area of T a kht-i-bfLhl, Peshawar
basin, seen from southwest. Note the massive foundation of the
extended court and a~sembly hall (H argreaves, 19 14, pi. X VIIIa).
46. Takht-i-bahr, looking east, showing construction in the lower
sacred area (court V). Note the PI main stiipa, stucco decora-
tion of many of the small stiipas, and the reuse of phase m devo-
tional icons (ASIFC 190, courtesy of the British Libra•y).
47. Nlonwnent<ll shrines along the southem edge of lower sacred
area V, Takht-i-bahi, Peshawar basin. On the lt:ft is monumental ·
image shrine M3 (note the visible socket holes for affixing a .
stucco image) and the largest image shrine to the right is M l
(11.2 m tall). The PI main stupa and some small stilpas are vis-
ible in the foreground (photo by K. Behrendt).
48. Small stilpa P37 in court XX, T a.kht-i-bahi, Peshawar basin. Note
the row of in situ stucco images of standing Buddhas under
trapezoidal arches, some of which display the varada mudrii (ASIFC
857, courtesy of the British Library).
49. Two heads and four sets of in situ feet of monumental Buddha
images against the south ""'<tU of court XX; note the sockets in
the wall for attaching the stucco images; T akht-i-bahi, Peshawar
basin (ASIFC 859, courtesy of the British Library).
50. Detail showing heads and in situ feet of three monumental
Buddha images against the south wall of court X..:'C, T akht-i-
bahi, Peshaw;u· basin (ASIFC 860, courtesy of the British Library).
5 1. In situ image base showing nistoric Buddhas and Maitreya from
sluine c in court XIV, Takht-i-bahi, Peshawar basin (ASIFC
960, cowtesy of the British Librmy).
52. In situ image base from shrine d in court XIV, Takht-i-bahr,
Peshaw<u· basin (ASIFC 961, courtesy of the British Libraty ).
53. Three plaques from two-celled shrine T4 in court XXIII, Ta.kht-
i-bahi, Peshawar basin. The first shows the Buddha in dhanna-
ca.Jrm mullrii flanked by haloed worshi.pers, the second shows the
Buddha seated on a lotus throne flanked by bodhisattvas, and
the third shows the Buddha in a pseudo-architectural frame
flanked by kneeling haloed figures , making it akin to Sravasti
composite images. The first two plaques have fill eted frames,
while the third has the same kind of base molding and is of a
similar scale (ASIFC 943, courtesy of the British Library).
Sahri-Baltlol Sites
54. Plan of Sahrf-Bahlol Site A, Peshawar basin (modified from
Spooner, 1909, pi. XXXVI).
55. R ow of in situ images consisting of one Buddha and seven bod-
hisa.ttvas, nonh fa~ade of a rectangular basement that probably
supported a two-celled relic shrine, Sahri-Bahlol Site A, Peshawar
basin (ASIFC 33, courtesy of the British Libraty).
56. Small stii.pa with stucco relief~ at Sahrr-Bahl61 Site B, Peshawar
basin (ASIFC 722, courtesy of the British Libr;uy).
57. Plan of Sahn-Bahlol Site C, Peshawar basin (modified from Stein,
1915, pi. XXXllla).
XX LL'IT OF ILLUSTRATIONS
]amal Garhi
61. Plan of J amal Garhr, Peshawar ba~in, showing sacred areas and
mountain vil!iiras (modified from Cunningham, 1872; Hargreaves,
1921 - 22).
62. Stair risers from J amal Garhl, Peshawm· basin (Warburg Institute).
63. Reliefs from J amal GarbJ, Peshawar basin (ASIM 1015, cour-
tesy of the British Library). A- Stacked relief showing in the
lowest register Buddhas with attendants separated by boxed
pilasters, a pseudo-vedika in the middle register, and standing
figures under arches in the top register. B- Row of seated
Buddhas under arches. C and D- Row of naked putti under
arches; they venerate a central Buddl1a, also under an arch. £-
Relics under arches, including the alms bowl, a cremation flame,
a11d a reliquary under a cloth.
Mekhasandcl
64. General plan of Mekhasanda, Peshawar basin, showing sacred
area and outlying mountain mhiiras (modified from Mizuno, 1969,
plan 1).
65. Plan of Mekhasanda sacred area, Peshawar basin (modified from
Mizuno, 1969, plan 2).
66. Schi~t Buddha image found in situ in shrine 21, Mekhasanda,
Peshawar basin (Mizuno, 1969, pi. 36 no. 2).
67. Standing Buddha from annex room ix, Mekhasanda, Peshawar
basin (Mizuno, 1969, pi. 35 fig. 2).
17zareli
68. General plan of Thareli, Peshawar basin, showing sacred areas C
and D , the outlying mountain uihiiras, and small monastic sacred
areas (modified from Mizuno and Higuchi, 1978, plan 1).
69. Plan of sacred area D , Thareli, Peshawar basin (modified from
Mizuno and Higuchi, 1978, 30).
I
I
I
Material com dir:itos autoraiJ
.
LIST Of' ILLUSTRA'MONS XXl
Ranigat
78. Plan of R anigat sacred areas, Peshawar basin (modified from
Nishikawa, 1994, plans I, 5, 6).
79. Core stripa LOO in east sacred area, R anigat, Peshawar basin
(after Nishikawa et al., 1988, pl. 1).
80. Early small str1pa 134 from pha~e II in east sacred area, Ranigat,
Peshawar basin (after Nish.ikawa et al. , 1988, pi. 4, no. l).
8 1. Pavement stones and bodhisattva at base of stairs of the mrun
stiipa in the east sacred area, Ranigat, Peshawar basin (line draw-
ing by W. Hipsman after Odani, 2000, fig. 6).
82. VedikDs from the double-headed eagle stupa, block F, Sirkap TaxiJa;
the east sacred area of Ranigat, Peshawar basin; and from the
main stiipa of Butkara I (GSt3), Swat (line drawing by W. Hipsman
after Marsh all, 1951, pi. 34c; Nish.ikawa, et al., 1986, pi. 23 nos.
7- l 0, I 3-·14; Faccenna, 1980, vol. 3, no. 5. 1 plate 68a).
..
xxu LIST OF U.LUSTRATIONS
Slziih1'i-kf-rj.heri
83. Shah~jl-kr-<;lherr plan of main swpa and some surrounding small
stUpas, Peshawar basin (modified from Harg1·eaves, 1914, pl. Xill).
Loriyiin Tii(lga:i
84. Loriyan Tir;lgai main stupa, Pes hawar basin (ASIM I 033, com·-
tesy of the British Library).
85. Group photograph showing mostly narrative images, Loriyan ·
Ta1~gai, Peshawa.r basin, taken in the 1890s (Warburg Institute).
86. R econstmcted small stii.pa, 1.4 m high, composed of miscella-
neous architectural elements from the site of Loriya.t1 Ta~Jgai,
Peshawar basin. Photo taken in the 1890s, now in the Calcutta
museum (ASIM 1036, courtesy of the British Library).
8 i. Schist lzarmikii panels showing the Birth and Great Departure,
Loriyan T ar:tgai, Peshawar basin; height is 45 cm (Foucher, 1905,
figs. 154, 182).
88. Group photograph showing iconic images found at Loriyan
Tir;lgai, Peshawar basin, taken in the 1890s (Warburg Institute).
89. Second group photograph, taken in the 1890s, showing iconic
images found at Loriyan T fu)gai , Peshawar basin (Warb1.1rg
institute).
Sikri
90. Plan of the site of Sikri, Peshawar basin (modified from Dean,
1889, plan; Dar, 2000, fig. i ).
91. Sikri, Peshawa.r basin, stii.pa reconstruction circa 1880s (Burgess,
1899, fig. 4).
.Nalhu Sites
92. Plan of upper Nathu sacred area, Peshawar basin (moclifi.ed from
Cole, 1883, pl. 8).
93. Group photo of sculpture recovered from upper Nathou sacred
area, Peshawa.r basin (the Alkazi Collection of Photography).
94. Plan of lower Nathu sacred area, Peshawar basin (modified from
Cole, 1883, pl. i ).
95. Group photo of sculpture from lower Nathou sacred area,
Peshawar basin (the Alkazi C ollection of Photography).
Ali M~id
96. Medium-sized stiipa with phase Ill stucco decoration from Ali
Masjid, Peshawar basin (Warburg Institute).
Butkara I
97. Plan of Burkara I sacred area, Swat (modified from Faccenna,
1980, vol. 3, no. 1, pi. VI).
98. Reconstructions of Butkara I main stiipa (GSt 1- 5) and two-
ceUed shrine (Great Building), Swat (modified from Faccenna,
1980, vol. 3, no. I, pi. 5, figs. 6, 18, 30, 56; vol. 3, no. 3, pi.
XII).
99. Relief depicting a round-based small stupa in a shrine from
Butkara I, Swat (Faccenna, 1995, vol. 2, pl. 267 , l nv. no. 920).
Butlwra Ill
100. Plan of Butkara Ill, Swat (modified from Khan, 1993, pl. 2).
I01. R elief depicting a stiifJa with four pillars surmounted by lion
capitals, Butkara ill, Swat, height 24 cm, width 30 cm (Rahman,
1987, fig. 10).
I 02. Stii,fJas in two-celled shrines D aeft) and C (right), Butkara Ill,
Swat (Rahman 1987, fig. 3).
Plinr
103. Plan of Panr showing all phases of construction and axone-
metric reconstruction of Stiipa 1 and surroundi ng monuments,
Sw~H (modified from Faccenna et al., 1993, figs. 87, 139).
Saidu
I04. Plan of Saidu sacred area and monastety, Sw·a t (modified from
Faccenna, 1995, figs. 22, 23).
I 05. R econstruclion of main stujJa, Saidu, SwfLt (Faccen na, I 995, fig.
282).
Mmjanai
I 06. Plan of Marjanai sacred area, Swat (modified lrom K han, 1995,
pl. 33).
.
:lC'<IV UST OF ILLUSTRATIONS
Ninwgram
109. Plan of Nimogran1 sacred area, Swat (modified from K han,
1993, pi. ll b).
Tokar Dara
112. Plan of Tokar Dara sacred area and monastery, Swat (modified
from Khan, 1993, pi. 7).
113. Possible direct access relic shrine and enclosing wall, Tokar
Dara, Swat (photo by K. Behrendt).
Chakpal
114. In situ garland holders set into the drum of the ashlar Chakpat
stupa, Swat, near the Chakdara fort, no longer extant (ASIM
1158, courtesy of the British Library).
llllllh\lh\1 l J;lnrfllfi
... ....
11
11
11 117. Detail of plan of Tapa-i-kafruiha sacred area, H ac;!<;la, Kabul
basin, Afghanistan (modified from Barthoux, 1930, plan A).
lI
.
I
••
i
•
'
l
I
I
'
'
'
I
l
'
I
I
'
•
'
'
'
f
I
•
k ctpta
11
11
11
,,
ACK!~Ov\'lJEDGME TS
There are many people without whose help tllis book would never
have become a reality. I thank my Ph.D. advisor Robert Brown for
shaping me as a scholar and for his guidance in this project. My
brother Marc Behrendt deserves special thanks for spending a year
with me doing fieldwork in South Asia and for critically debating
many of the issues presented herein. My editor, Faith Rogers, helped
me to present my data and thoughts clearly. Research for and shap-
ing of this work would not have been possible without the support,
patience, and insights of my wife and coUeague Pia Brancaccio.
l am indebted to the giants in the field of Gandharan archaeol-
ogy, Maurizio Taddei, Shoshin K uwayama, Domenico Faccenna,
and Elizabeth Errington, scholars who suggested new tines of inqillry
and whose work provided me with tJ1e foundation for tJus book. I
also e"'J)ress my gratitude to Ronald Bernier, Pierfrancesco CaUieri,
Susan Downey, Anna Filigenzi, Phyllis Granoff; Wayne Hibschman,
Donald McCallum, Michael Meister, David Nelson, Anna Maria
Quagliotti, Koichi Shinohara, Waiter Spink, and Michael Willis.
Scholars all over tJ1e world who share an interest in Gandhara
provided great help to me. My work in Pakistan was facilitated by
many scholars, an1ong whom Abdur R.1.man and Ashraf Khan deserve
special recognition. I also thank tJ1e many people working in the
museums in Lahore, Pesh.awar, Swat, and Ta.xila, as well as the
many chokidars who were so kind to me and who brought to my
attention many features of tJ1ese sites.
1 received institutional support from T emple University, wluch
gTanted me an academic leave, and from the University of Califonua
Los Angeles for two Dickson Support fellowships. The University of
Kyoto (Zinbun Institute) supported me while I completed work in
their outstanding arcluves. Paul Taylor, at the Warburg Institute,
Sophie Gordon, who made the AJkazi Collection of Photography
accessible, the British Libraty's unsut'J)assed holdings, the British
Museum, Istituto Italiano per !'Africa e l'Oriente R oma, lstituto
Universitario O rientale Napoli, tl1e American Institute of Pakistan
Studies, and the American Institute of Indian Studies all were invalu-
able sources. Many other people contributed in various ways to this
venture; I appreciate their help, and I thank tl1em all.
1
I
••
I1
l
l
I
I
I
!
11
11
INT RODUCTION
1
Fa-hian, Si-Tu-/G. Buddhist &cords qf lite Wesum World, x:xxi; Hsiian-tsang, Si-:Yu-Ki.
B11ddhist Records qf the W«rtern World, 97.
2 IJ\'TRODUCTION
basin, in northwest Pakistan (fig. 3), that are bounded by low moun-
tains to the north and west, by the Indus river to the cast, and by
arid land to the south. In this book I refer t:O this specific geograpl1ic
area as the Peshawar basin or as ancient Gandha ra. The tenn Greater.
GandhfLTa is used here for the larger culturaJ sphere that includes.
parts of Afghanistan, Kashmir, and the Swat valley, as well as the ,
Pes hawar basin. H owever, "Gandhira" has been used rather loosely .
in 20th century scholarship to refer to an area that includes the Swftt
valley and Ta.xila as well as to a style of sculpture found in these
regions. To avoid potential confusion, I use aucien.t Gandhlira or the
Peshawar basin in this work to indicate this geographic entity (see sec-
tion 1.2 Geography of Greater Gandhara and the Distribution of
Buddhist Sites).
Greater Gandhara has been of interest historically because it was .
the Buddhism of tlus region that spread to Cllina and other parts
of east Asia. Moreover, only in ancient Gandhara can we observe
an uninterrupted pattern of religious development starting in the 2nd
century B.C.E. and continuing to the 6th centu•1' C .£.; a fragmen- '''
'
tary record extends the pattern to the 8th century C.E. International <
'
trade indirectJy fimded continuous construction of new centers of
worship and expansion of existing sites in ancient Gandhara, whereas
in other parts of south Asia gaps in patronage resulted in an incom-
plete record of BuddJlist arch.itecturaJ and sculptural development-
two key sources for understanding ideologicaJ change. Until recently,
only a few inscriptions and text fragments were known from Greater
Gandhara, so the architectural and sculptural remains were the best
sources of evidence concerning tllis early culture. The recovery of
thousands of manuscript fragments (ea. 1st centu ry-8tJ1 century C.E.)
will w1doubtedly transform the way •ve understand this active Buddhist
community.
This book characterizes the organization and function of the .
Buddhist religious architecture in the public sacred areas as well as .
in the monasteries. While the main focus is on structural features, '
sculpture is also addressed in terms of its integration and use in these :
devotional settings. In this sense, the sculp ture provides evidence for ;
interpreting how various su·uctures functioned, and it provides insight
into how people moved within these sites. R eciprocally, the archi-
tectural evidence contributes greatly to our understanding of how
sculpture was used, and it provides clues as to why these pattems
Copyrighted material
4 lNTRODUCTION
.
Copyrighted mater> l
... .·
INTRODUCTION 5
has led to proposed dates for that ascension ranging from 78 C.E.
to 144 C.E. or later. 4 Without taking a stand on this complex issue,
I accept the date 120 C.E. simply as a convention until new evi-
dence surfaces.
Sir J ohn MarshaU proposed an architectural chronology based on
stratigraphic relationships among the distinct masonry patterns pre-
sent in his excavations at the sites of Taxil.a, which he dated using
coin s.~ The dating system he developed at the beginning of the 20th
century was a major breakthrough. H e recognized four major types
of masomy: rubble, diaper, semi-ashlar, and a variety of late semi-
ashlar (for a full description and discussion of these masomy types,
see Appendix A) (figs. 5, 9). This relative sequence of masonry types
can be applied to at least 15 different sites at Taxila excavated
between 1912 and 193 7 (fig. 6). Because the superimposition of
masomy is still pt·eserved at many of his excavation sites, this sequence
can be corroborated. 6 Although :rvla.rshall's broad sequence is fairly
wel.l determined, his identifications of chronologic sub-units based on
the quality of the fabric in a single masonry type cannot always be
accepted. Different workshops and the amount of money invested
are variables that are not chronologically dependent, yet they alfected
the quality of the masonry fabric. 7 Since Marshall's excavations at
Copyrighted material
6 INTRODUCTION
Noc only was he inconsistent with Utis intetvretacion, but he did not present solid
criteria for his relative dating. Recent study does u1dicate U1at some chronologic
subdivisions can be made within the cliaper sequence; however, for the purposes of
this book, all of the diaper stTuctures are grouped togeth er. See Fitzsimmons, Stupa
Designs at Taxi/a, 5-14, 53--9; Kuwayama, "ln the Time of Late Sirkap and Early
Dbarmarajika: How Ta.xila Introduced Stfijill. Architecture."
8 The Buddhist structures of Sw·at and the 'Peshawar basin were fabticated from
schist, a britrle rock with strong internal nonplanar bedding, which flakes irregu-
larly. These characteri~tics caused the builders tO use less systematized masotU)'
techniques, which varied from site to site, depencling on Ute available J.ocaJ stone.
T he Peshawa.r basin and Swat construction techniques were obviously refined over
time, but a da.tftble pattern does nol emerge. At the Peshawar basin site of Takbt-
i-bllhi, fo r example, walls of image silrines made in the late periods stand as higb
as l3 m. TI1ese walls are lbinner, more flexible, and appitrently more durable than
those erected in eadier ]>eriods a t this same site, }'et a chronologic masonry pro-
gression ClliUlOl be determined.
.
Copyrighted mater
...
lNTRODUG'TJON 7
Copyrighted material
8 INTRODUC'li'ON
9
Early semi-ashJa•· is found at only a few structures at the Dhannan1jika com-
plex (K I, K2, K3, P3, P6, N4, and the benn of the main slfi/Ja). This masonry
subcategory can be atnibur.ed only to the earl)' pan of phase rn, because it seems
to exhibit experimental qualities; thus, it is less useful for dating (:M.arshall, Taxila,
261 - 2).
Copyrighted materi, 11
..
I
•I
INTRODUCTION 9
"' Whether these images represent the Sravasti Miracle is debated , but similar
iconography is found at Aja(•!li, Ka(lheri, and many other ,;tes, indit-ating the pop-
ularity in India of !Jus iconographic program as early as the late 5th century C. E.
and continuing for several centuries. See R. Brown, "The Sravasti Miracles in the
Art of Lndia and Dvaravati," Archives tf A.Mn Art 37 (1984); J. Rhi, "GandJ1aran
Images of r.he Sravasti Mjracle: An Iconographic Reassessment" (Ph.D. dissertation,
University of California, 1991); W. Spink, "Aj<tnta in a Historical and Political
Context," Malun·ashtra Pathik 2, no. I (1990).
11
For a broad discussion of this issue, see M. Taddei, "Chronological Problems
Connected "~th Buddhist Unbaked-Ciay Sculptures from Afghanistan and Surrounding
Areas," in Coins, Art, and C!rratwlotp: Essays 011 Pre-ls!JJmic History tf the Indo-lranum
Borderlands.
11
Sh1ine M2 at Takht-i-bahi is more than 11 m tall. Monumental feet and heads
have been found at many sites, including Takht-i-bahi, 1\llckhasanda, Oharrnarajika
and Jauliiifi. See also M. Tadclei, "Afghanistan, Sculpture, 4th- 9th Century A. D.,"
in Dictionary tf Art (MacmiUan Press, 1996).
Copyrighted material
10 LVfRODUCTJON
in Swat were created, 13 this was generally a time when large parts
of the sacred areas were abandoned. Evidence for the nature of
Buddhist practice is thus sparse, although we can assume that Buddhism
was still being practiced, as indicated by the addition of late coins
to pre-e,.::isting relic deposits of main sltipas.14 The decline in patron-
age is most evident in the Peshawar basin; a limited patronage base ·
in Swat still existed, and extensive construction in Afghanistan and
the Sind is evident.
Previously, it was thought that the invasion of the H ephthalites in .
450 C .E. resulted in the persecution and eventual destruction of the
Buddhist community in ancient Gandhara. However, recent work
has shown that Buddhist pilgrimage sites in the Peshawar basin, such
as the one that housed the alms bowl of d1e Buddha, remained vital
until the middle of the 6th centu ry C.E.; these fmdings suggest d1at
the period of major Buddhist decline coincided with me collapse of
the H ephthalite power. 15 While it is difficult to link mese numismatic
finds with tl1e archaeological data, it is obvious that building of new
structw·es halted suddenly at many sites in the Peshawar basin and
TaxiJa.'6
Extensive reuse of sculpture occurred in the Peshawar basin at
many sites during phase IV (see Appendix C). It appears that moved
sculptures were placed 11ext to me few devotional relic structures that
were still in use at this time of disruption. Patterns of recontextual-
ization can thus help us to underst<md which parts of d1e sacred
areas remained devotionally significant, even if new patronage did
not supply a developing sequence of new architectural fonns. In d1e
Swat valley, where patronage appears to have endured, late images
were installed at some sites, <md many mck-cut figures of Buddhas
" CnJciJonn stupas were built at Shah-jf-ki-t;iher·f and San ri-Bahlol Site G in the '.
Peshawar basin, and .Bhamala in Ta..xila. ln Swat the related sites of Abba Sahib (
China and Tokar Dara probably date r.o phase IV. The domed monastery found
at Tokar Dara io Swat is comparable to a phase rv monastery at the site of
Culdm"R, in the Kabul basin. See C . fussman, Monummts &udlfltiques de la Rigio11 tk
Cahou~ vol. 22, Mbnoires de In Deligotirm Archlolbgiljue Frmzrmu en lljgllmlistan (Paris:
Diffusion de Boccard, 1976), fig. 13, pis. vt-vi.i.
14 Enington, ' um.ismatic Evidence for Dating the BudcU1ist Remains of Candhara,"
199- 200.
'~ S. Kuwayama, "The Buddha's Bowl in C<U1cU1ara and Relevant Problems," in
SoutlzAsian llrciloeolbgy 1987, ed. M. Taddei (Rome: lsl\ltEO, 1990), 964- 71.
6
' For exmnple, at the site of Takht-i-bahi, an expanded sacred area, involving
consu·uction of a massive and cost!)' terrace, was never filled with donated struc-
tures (see section 7.2.4 Takht-i-bahi: Late Phase lU Construction).
Copyrighted materl 11
I. ;
INTRODUC'l10N 11
Copyrighted material
'
CHAPTf:.R ONE
Copyrighted mater 11
.. ·- .
OVERVlEW OF GREATER CANDHARA 13
1
D. MacDowaU and M. Taddei, "The Eady Historic Period: Achaemenids and
Greeks," in 17•e Archaeology '!! Ajglumistan .from Earliest Times to tlte Timurid PerWd, ed.
R . Allchin and N. Hammond (London: Academic Press, 1978), 2 18- 32.
2 Ibid., 204- 12; D. MacDowall and M. Taddei, "The Pre-Muslirn Period," in
17•e llrcluteOlogy '!/ Ajghani.rta11 .from Earliest Times to lite Timurid Period, ed. R. Allchin
and N. Hammond {London: Academic Press, 1978), 233- 4, 245- 8.
' Eni.ngton, " 1umismatic Evidence for Daring the Buddhist Remains of Gandhara,"
194; G. Fussman, "Numismatic and Epigraphic E'~dencc for" iJ1e Chronology of
Early Gancllm-a n Art," in h11Jesti.f50hllg Indian Art,· Etoceedings if a ~ymposium on the
----
.Dwelj)pmu111 of .Eor{y Buddhist mu! Hind11. lconogropl!)',Held ot the Museum of lndi.aJz llrt •
Berlitz ·in Mqy 1986, ed. W. Lobo and M. Yaldiz (Berlin: Museum fu.r Indiscbe
Kuost, 1987). See Chapter Four herein on phase n architectural evidence and
Appendixes A and B for discussions of da.Jing.
• .For a full sunumuy of the issues sun·ouoding Hina)lfliJa and Mahliyana Buddhism
sec W. Zwalf, A Cakdoguc if tile Cmu!Jwm Sculp111J"e in tile British. Museum., 2 vols. {London:
British Museum Pt·ess, I996), 30- 3.
was not dynastic; the great K ushan empire was losing power at this
rime. Only a vague idea of the political landscape of this period can
be formed from coins of the Kushans, Kushano-Sasanians, Sasanians,
and Kidarites. The 5th to 8th century C.E. period is even more
obscure, as Gandhiira feU under the control of the H ephthalites,
Hunas, and Turki Shahis. By the time the Hindu Shahis attained
power in the 9th century C.E., Buddhism in Gandhara was fu.lly
eclipsed. 5
Between the 5th and 7th centuries C.E. , Chinese Buddhist pil-
grims began to visit Greater Gandhara to see famous relics like the
Buddha's alms bowl or his skull bone (u~f~). Short biographies of
many of these Chinese monks survive. The extensive travel accouncs
of f axian (40 I C.E.) and Xuanzang (630 C.£.)6 reveal much about
the late Buddhist tradition of Greater Gandhara. AU of these sources
suggest a dramatic decline in Buddhist patronage in the late 5th cen-
tUJy. The archaeological evidence appears to support the written
records; few post-5th centllly coins have been found in the religious
areas, and new construction was abruptly stopped at most sites.
Sparse evidence of Buddhist activity at some isolated sites in the
Peshawar basin survives, and a few of these centers seem to have
remained active until perhaps the 8th centUJy C.E. The collapse of
patronage was not so severe in the Swat valley, and throughout this
long period Buddhism continued to flourish in A(~hanistan. 7
·' Bnington, ''Num.i;1natk EvidenC'e for Dating tht' Buddhist Remains of Gandhara,"
20 I 3; S. Kuwayama, "Th(' Hephthalites in Tokharistan and Gandhara; Pa11 1:
Gandhara," Lalwre Museum Bulleti11 V, no. I ( 1992); S. Kuwa)•arna, "The H ephthalites
in Tokharistan and Gandhara; Part ll: Tokharist.an " Lal!ort Museum Bulleti11. V, no.
1
2 ( 1992}; A. Rahrnan, Th Los/ Two Dynasties if the ::,ahis: An Analysis if 17wir Hiswry,
Arrhaeology, OJinll!ie and Polorogmf>I!Y (lslamabad: Centre for the Study of the Civilizations
of Ctntral Asia. Quaid-i-Azam University, 1979).
•· Fa-hian, Si-1'u-Ki. Buddhist R=rtls 'if flu 11-~tem World, tnms. S. Beal {London:
Triibner & Co., 1906); H siian-tsang, Si- Yu-IG. Budilhist Records qf tlu Wtstem World,
trans. S. Beal (London: Triibner & Co., 1906}; Kuwayama, "The Buddha's Bowl
in Gandhara and Relevant Problems."
' Eningt.on, "Num.iSll'latir E'~dence for Dating the Buddhist Remains of Gandhar<t,"
199; D. Klirnburg-Saltcr, 77u Kmgdom qf Bam!ywt· Buddhist Art and Cul!ure qf tlte HiJuiJt
Ktuh. (Naples and Rome: lstituto Universitario Orientate Dipart:imento di Studi
Asiat.ici, IsMEO, 1989), 90 -2; Kuwayama, "The Hephthalites in Tokharistan and
Gandhara; Part 1: Gandhara; Part n: Tokharisran"; M. Taddei and G. Verardi,
''Tapa Sa r'((ar: Second Preliminary Report," Ea.,·t and We.rt 28, no. L- 4 ( 1978).
llllllh!'hl1 IJ;lnrf' J
a !
OVERVIEW OF GREATER GANDHA.RA 17
17 A. Cunningham, ':J ~•m!il Garhi," Archaeological Survry '![ India 5 ( 1872- 73);
" ManiltyaJa," Arclweologicai Survry rif India 2 (1863- 64); " Manikya.la," ArchMOlogical
Surory '![India 5 ( 1872- 3}.
u H. BeUew, A Gmeral Report 011 lhe Tust![<:Pis (Llhore: Sang-c-Mcel Publications,
1864; reprint, 1977); F. Wilcher, " Report on the E1tploration of the Buddhist Ruins
at Takht-i-ba.i, .January to Ap1il 1871 , Punjab Government Gazette, Supplement,
6th August 1874, 528- 32," in E. Errington, "1l1e Western Discovery of the Art of
G<llldhara and the F'mds of J arniilt:\"arM" (Ph.D. thesis, London University, 1987).
" D eane's report has been lost; nevertheless, much is known of this site (S. R.
Oar, "The Su:ri Sculptures: Prolegomena on an Exceptional, but U.-w:uc:licd, CoUection
of Gandharan Art in Lhc Lahore Museum," in Silk Road Art and ArciUJeology, Papers
in H01wur rif Fran&int Tissot, eel. E. Errington and 0. Bopearchchi ( 1999- 2000};
H. Dcane, Memorand1m1 011 Excaootions a1 Sikri. Tusqf<Pi (Lahore: 1889; lost). Elizabeth
Errington kindly provided me with two reports: H. H. Cole, Memorandum 011 Ancienl
Afomm1ents i11 Eusof<;ai, witlt a. Description of the &ploTiltimu Undertaken .ftom lite 411t Febmary
to the 16th April 1883: Curator rif Ancient Monuments in India, 1/lwtrated ID• a Map, 8 Pums
if Building.! and 17 P/ats if Rough Sketcltes, June 1882 (Simla: Government Central
Branch Press, 1883); H. H. Cole, Preservation rif National Monuments, bufia: Greeco-
Buddhist Sculptures .ftorn. Tusf<;tzi (Paris: I884 - 5).
18 CHAPTER. ONE
(fig. 83). 15 The prinuuy goal of this work was to gather antiqlllttes
that could be displayed in the major colonial museums in Calcutta,
Bombay, and London. T he recovery of several thousand schist and
stucco sculptures led to the establislunent of tJ1e Pesbawar and Lahore
regional museums to house this material. More than 2,000 unpublished
photographs taken by the Archaeological Survey document the exca-
vations and today represent an invaluable source of information for
reconstructing the work done and the mace1ial recovered at that
time; photographs from this period appe<u· throughout this book. 16
At the peak of British power in south Asia, Sir J otm Marshall
began to dig the urban and Buddhist centers in Taxila in a series
of excavations between 1912 and 1937. 17 Unlike bjs predecessors,
---
" H. Hargrea.ves: "Appendix V: List of Antiqu.ities Recovered During Operations
at J amalgarhi, 1920- 21," in Arcluu!IJiogicnl Sttrol')' of India Fro11Iicr Circle Roporl l 920- J
(Peshawar: Caxton Printing Wor.ks, 1921); "Conservation at Takht-i-bah.i," in ;bmual
Report of tlu: Archacowgiwl Sltrol')' of l11dia, FronJier Circle (Calcutta: 19 11- I2); "Excavations
a t ShlU1-jil<I-<i.herr," in Amwai Report of tile Arclwi!/Jibgical Sttrv!J of India, Fronti11r Circle
(Calcutta: 19 11 - 12}; "Excavations at Shahji-ki-c:lhe!i," in llrdzai!IJ/f)giwl SuTVI'J' of /TUiia
Amwal Rtport (1910- 11) (Calcutta: 19 L4); "Excavations at Ta.kht-i-bahi," in A-rd1D#Jiogiwl
Suro9 of bzdia Armual Repcrt (1910- 1.1) (Calcutta: 1914); D. Spooner: "E.>:.cava.tions
at Sahri-Bahlol," in ArdweobJgicaJ Surory of lndin Annual Report {1906-07) (Calcutta:
Superintendent Government Printing, I909); " Excavations at Sahri-Bahlol," in Ardzaco-
logicai Sunny of lruli11 Annmtl Repcrt (1909 - 10) (Calcutta: Superintendent Government
Priming, 1914}; "Excavations at Shft.h-jiki-Qlleri," in Armltlu Report. of tile !lrcluuo/f)gicai
Su1VI'J' of lrulia, Fronlicr Ci1dc (1907- 08) (Calcutta: Superintendent Covenuncnt Printing,
1908); "Excavations at Shah-ji-kr-cj.bc;ri," in Ardweologicai SwVI})' of /m/Ut Annual Report
(J 908- 09) (Calcutta: Superintendent Government Printing, l 9 13); "E."cavati.o ns at
Takllt-i-bahi," in ArcJtaJ!fJibgical Sv'TVI'J' of India. Amwal Report ( 1907- 08) (Calcutta:
Superintendent Covenuncnt Printing, 191 1); A Stein: ArrJweqlogica/ R=mum!tllzus in
.Nmtll- Wesl.tm India and SouJJ1.-l:."asl.tm lro.n (London: I93 7}; "Excavations at Sah ri-
Bahlol," in AJdwoologietJL SullJ!Y of.li1dia. Annual Rctmt (1911- 12) (Ca.lcutta: Superintendent
Government Printing, 19 15). l! was also at dtis time that Swlit was first swveyed;
see A. Stein, On Alexaruler's Track Ill the !tuhJS: Persmwl .Narralive of ExpllrraMrlS on the
.North-West. Hrmtier of bufia. (London: Phoenix Press, 1929; reprint, 200 I}.
16 A complete archive of Archaeological SuiVe)' of India. photographs is hou.sed
llllllh\lh\1 lJ;Inrfl 1
I
OVERVIEW OF GREATER CANDH.".RA 19
I
~ Marshall used a much more scientific approach in his archaeologi-
cal excavations and kept accurate records of his finds. While Marshall's
archaeological methodology might seem inadequate when compared
to modem practices, his activities mark a turning point for our under-
standing of the Gandharan architectural tradition. His detailed exca-
vation reports of the sites in T axi!a remain i.mpona:nt for understanding
the region.
In the 1950s, after a break because of th e pobtical turmoil that
led to the partition of south Asia into the modern nations of Paki~tan
and India, archaeological activities were resumed. Several foreign teams
began to work in Pakistan. Giuseppi Tucci initiated an Italian archae-
ological mission in the Swat valley; major excavations were carried
out under Domenico Faccenna at Butkara I (see fig. 97), Panr (see
fig. 103), and Saidu (see fig. 104). 18 Pierfrancesco Callieri, who beg-dll
the excavations at the urban site of Barikot in the 1980s, has con-
tinued the work. 19 This research has proved to be fi.mdarnentally
Paki.stm1 Suroeyed in 1960 (Kyoto: K)'Oto University, 1962) and MeldrasmuitJ: Buddlrist
Monastery in Pakistan Suroe.J'ed in 1952-67 (Kyoto: Kyoto University, 1969); S. Mizuno
and T. Higuchi, eels., 17rareli: Buddhist Site in Pakirtan Surot!j>ed in 1963-67 (Kyoto:
Doihoisha, 1978).
11 K. Nishikawa, iUllligal: ;J Buddhist Sile in Gr11Jdhara Pakista11 Surot!j·ed 1983- 92
Editions Recherche sw· les civilisations, 1982}; J. Ba.rthoux, IJJS Fouilles de Hatlda, 4
vols., vol. l, Merrwires de la. .Deligation ArciW1logique Franyaise en Ajglumislm• (Paris: Edi-
tions d 'Art et d'Hist.oire, 1930); .Fussman, Monurmmls Bouddhiques de la Rigion d8 CabouL.
J. H ackin, "Divers R echerches Archeologjques en Afghanistan ( 1933- 1940)," in
T eams from od1er countries were also active. Maurizio T addei led
an Italian excavation of d1e late Buddhist complex of Tapa Sardar
in the 1960s.2'1 The Japanese archaeologists Mizuno and Shoshin
Kuwayama worked at the urban site of Cha.qalaq T epe, the Buddhist
cave sites of Haiba.k, Basawal, Hazar-sum , and Fil-K hana, and the
stupa and sacred area at Lalma, and they conducted a major photo
documentation project at Bamiyan. 25
T he ongoing destruction of evidence of the cultural heritage of
Afghanistan has made the study of d1e Greater Gandharan tradition
increasingly difficuJt. Following the Soviet invasion and the subse-
quent civil war, the Kabul museum was looted mu16ple tinles, as
were many smaller public and ptivate collections. Many archaeo-
logical sites, in par6cular the complex at H a<;l<;la, were deso·oyed dur-
ing the protracted conAict. Political instability since the Soviet
withdrawal in 1989, coupled with extreme poverty, has driven peo-
ple to loot sites; large numbers of artworks of A(~han provenance
con6nue to appear on the international art market. Millions of land
mines placed throughout Afghanistan will prevent any archaeological
"' Fa-hian, Si-Yu-Ki. Buddhist fiLcordJ qf lfze Wt.rtem World, x;"xi; Hsiian-tsang, Si-
Yu-IG. Bwldhist Ruords qf /M f1!tt1/1'1?1 World, 97.
OVERVIEW 01' GREATER CANDH.A.RA 23
?'I For a ftall discussion of r.he ancient geography, see Kuwayama, "'The Hcphthalites
in Tokharistan and Gandhara; .Part [: Gandbara"; "The Hcphthalites in Tokharistan
and Gandhara; Pan U: Tokharistan"; "Kapisi and Candhara According to Chinese
Buddhist Sources," Orient 18 ( 1982): 133-9; "Route Changes and the Decline of
C<uldhara," in Sources rf GlJII(/h/iran Buddhism: Archaeology, A1t and Ta·ts, eel. K.. Behrendt
and P. Brancaccio (VanCOllVet·: University of British Columbia Press, in press).
29
.Because the Taxila sires are beyond the Indus River, they cannot be included
in Candhara p•·opcr, although they share many common features.
suggest that this longer but easier route through ancient Nagarahar~
was favored after 520 C.£. 29
The decline of patronage in the Peshawar basin appears to have
coincided with a period of economic prosperity in the Kabul valley
and the construction of many new Buddhist centers in Afghanistan.~
It seems possible that the shift in txade routes might have been the
cause of the economic and artistic coUapse of the Peshawar area.
2'.1 Kuwa.yanuo, "The Hephthali[CS ill Tokharismn and Gandhara; Part I: Oandhara";
"The Hepbthalites in Tokharistan and Gandhara; Part U: Tokharistan"; "Kapisi
and Gandhara According to Chinese Buddhist Sources," 136.
,., Spooner, "Excavations at Sahri-Bahlol (1906- 07)"; "E.xcav-ations at Sahri-Bahl6J
(I909- 10)"; Stein, "Excavations at Sahri-BabJol"; f. Tissot, "SaJu·i-BabJ6l {Part IV),"
in South Asiatz Archoeology 1993, eel A. Parpola and P. KoskikaJiio (Hclsiol..:i: t994);
F. Tissot, ''The Site of Sahri-BabJol in Gandhara (Part ni)," in Soulh Asian Ardweo/og)>
1987, ed. M. Taddci (Rome: ls.lv!EO, 1990); F. Tissot, "The Site of Sah ri-Bahlol
in Gandhara, Pakistan: Further i nvestigations," in South. 1lritm Archaeology 1985, ed.
K. ftifclt and P. Sorrensco (London: Riverdale: .1989); F. T issot, "T he Site of SaJui-
Balllol in Ga.ndham," in Soul11 Asi1111 A:rdwt10/ogy /983, ed. J. Seho~Srnans and M. Taddci
(Naples: !stituto Univcrsitatio Orientale, 1985). Shah-ji-10-<;lheri was excavated at
the beginning of the 20th century by the Archaeological Survey of India, but it
does not survive today; a modem cemetery has enveloped this site. H argreavcs,
"E.xcav.uions at Sh§hjr-kf-<;lherl( 1910- 1I)"; "ExcaV"ations at Shfth-jT-kf-<lherf ( .1911- 12
AS!FC Circle)" S. Kuwayama, Tile Main Stupa o1 Sltlihji-/(i-rj}un· A ClmmologicaL Owlook
(Kyoto: Institute for Rescat'Ch in Huma nities, Kyor.o University, 1997); "Excavations
at Shah-ji-ki-<;lhcri (J 907- 08)"; " Excavations at Shah:ii-ki-<;lheri ( 1908- 09)." Gaogu
Dher was excavated by tbe Peshawar University Department of Arcl>aeology under
the direction of rarid Khan.
" Saecd-ur-Rehman, cd., Ardweological .RwmnaissmUA in Gandlwra .1996 (Kara.du:
Department of Archaeology & Museums, Ministry of Culture and Spot'tll, Govemmcnt
of Pakismn, 1996).
" G. Erdosy, ''Taxi.la : Political History and Urban Structure," in South Asian
Ardweofbgp 1987, ed. M. Taddei (Rome: IsMEO, .1990), 662- 66.
33 P. Ca.llie,;, "Buddhist Presence in the Urban Settlements of Swat, 2nd Cent.
.B. C. - 4th Cent. A.D.," in Sourus of GandhllraJI Buddhism: Ardweology, Art and Texts, ed.
K. Behrendt and P. Brancaccio (Vancouver: University of Bf'itish Colutnbia Pf'ess,
in press).
Lalchak, Ballar, and many other uncxcavatecl sites (figs. 16, 20, 27,
32, 42, 15, 21, 22, 24, 25, 26). ear Taxila is also the large Mai)ikyala:
stzipa, which has not been fully excavated. The concentrat ed archi-'
tecturaJ evidence from T axila that can be dated between circa 2nd;
century B.C.E. and 7th cenn.uy C.E. ~ fundamentally important for
understand ing the larger Buddhist tradition of Greater Ganclhara.
Although the art and architecture from Taxila arc slightly different
lrom those of the Peshawar basin, the dose proximity of lhe two
areas and the common use of certain types of religious and monas-
tic structures allow for productive comparisons.
l• Fa-hian, Si-Yu -Ki. Buddhist Rtcortfs ~/ tllr Wt.~ttm World, xxx; Hsi~<ll>-t>ang, Si- Yu-
Ki. Buddlti.st &rortir rf lilt I Vtrlml World, I I 9.
" Callieri, Filigenzi, and Stacul, '·Excavations at Bir-Kot-Gh"a nda.i, Swat: 1987."
are several late Buddhist complexes, which include the sacred areas
of Abba Sahib China and Tokar Dara (figs. l l 0, 112).36 In more
remote valleys to the south of the Swat River are the massive stiipas
at the sites of Shnaisha, Shingardar, and AmJukdara.37 The sacred
areas and monasteries at these complexes have yet to be excavated.
Other important sites in Swat located nord1 of the river and at me
west end of it'l valley include Nimogram, Gumbatuna, Ma~anai ,
Andan Dheri, and Chatpat (figs. 109, I06).38 A recent survey of Swat
conducted by the Department of Archaeology in Pakistan has identified
many other sites du-oughout d1e valley.39 The majority of the Swat
Buddhist complexes were constructed along the southern edge of the
valley, indicating that contacts with the religious communities of the
Peshawar basin were important.
96
Khan, Buddhist. Shrines in SWIJt, 59- 62, 64 - 66.
31
Ibid., 53- 55, 62-64; Rahman, "Slmaisha Gumbat: First Preliminary Excavation
Report." An early general survey of Swat, which included the sites of Barikot,
Amlukdara, and Abba Sahib China, was carried out by E. Barger Md P. Wright,
"Excavations in Swat and Explorations in the 0:..-us Terrimries of Afghanistan: A
Detailed Report of the 1938 Expedition," in Memairs qf the Archaeological Swvry qf
bulia (Delhi and Calcutta: M.anager of Publications, 1941).
38 A Dani, "Excavations at Andandheri," Ancient Pakistan IV { I 968-9); A. Dani,
" Excavation at Char.pat." A11cient Pakistan TV ( L968-9); Khan, Buddhist Slni11es in Swat;
Khan, " Preliminary Report of Mrujanai."
l'} Relunan, ed., Archaoological Reco1maissmrct i11 Cmuilwm.
'10 Gera.1'd Fu.ssman suggested, in an open discussion a1 r.be conference " On the
Cusp of an Era: Art ill the Pre-Kusban World" Nov. 8- 12, 2000, Kansas City, that
the tower-like aw.fa was purpose!)• raised on a high base to increase its visibility and
make it a rnore eHective religious monument.
core of every site's public sacred area, and inevitably this stiipa was
one of the first structures erected.
l. 3. 2 Small Stiipas
Smaller stuj)OS in the sacred areas are typicaUy grouped close to the
main stilpa, as in the case of J auliaii (fig. 27). Although most of the
ht:terogeneous subsidiary stftpas were added over time, usually one
of them was constructed at the same time as the main stiipa.4 1 D uring
the phase Ill period of active patTonage at Jau.lian, 28 smal.l stiipas
were built in the sacred area. Some of the subsidiary stiipas, gener-
aUy donated by devotees, contained important relics considered wor-
thy of independent vene1·ation. An example is the small sw.pa D6,
placed in its own enclosure on the east edge of the sacred ar·ea. In
tlus instance, as the site expanded and space for donative offerings
around the main stupa was exhausted, this smaller stiipa was con-
sidered important enough to impinge on the enclosure of the main
stiipa.
The function of these small stiipa.s remains an enigma . Scholars
have suggested tl1at they may represent votive offerings or perhaps
may have hou_~ed the remains of monasl'ic dead. 42 The tradit-ion of
creating min.iatuTe funerruy stii.pas is weU documented at late sites in
other parts of India, where hundreds of stiipas were packed into the
sacred areas (as at R atnagiri), but the situation may have been
different in Gandhara. For instance, the site of Jaulian, like others
in the region, had for about 300 years a multi-storied monastery (it
contain ed at least 56 residential cel.ls), yet only 28 small stiipas were
constructed throughout the life of the site. If the stiipas contained
a~h es of the most important monks, th.ey would have been special
stmctures and not ordina1y burial units; many more monks would
have died at this site in 300 years.
" At Jauliiiii the main stlipa was constJ"U<:ted at about the same time as a small
stilpa to d1e east (Al5 in fig. 27). Kuwayama bas kindly poimed out several other
instances of this practice: Mekhasanda (stilpas 2 and 4), Lalma (sllifia 4}, and Tapa
Shotor (.rttipa 5). See also Kuwayarna, "Tapa Shotor and Lalma: Aspects of the
SWJ!Il Court at Hadda."
42
G. Schopen, "Burial 'Ad Sanctos' and the J>hysical Presence of the Buddha
in Early lndian Buddhism: A Study in d1e Archaeology of Religions," Religio11 17
( 1987).
Copyrighted material
30 CHAP'l'ER ONE
Copyrighted material
OVERVIEW OF GREATER GANDH.ARA 31
Copyrighted material
•
32 CHAPTER ONE
The first image shrines added to this site were small and functionecf
to define and enclose the main stiipa court. In contrast, the latet
phase III image shrines C 14- 16 on the southern edge of the site
were huge and must have contained monumental sculpture. A co~~
sistent cluonological trend can be observed in Greater Gandharj
smaU image shrines gradually give way to larger shrines and ulti~
mately to massive chapels.
Image shrines were usually added one at a time to sacred areas·
apparently by different patrons, and d1erefore they are heterogeneo
in character. At many sites it is still possible to reconstruct the build
ing sequence of these additions, because one shrine overlaps th~
foundation of the next or uses an existing wall for part of its supe~
structure. Generally these U-shaped structures were equipped witli
hinged shutters that could be opened and closed, possibly suggest•
ing resuicted access. 45
Most image shrines appear to have been placed facing the maim
stupa, defming the bolLildaries of tl1e public sacred area. Such clona•
tive structures were also created next to pa~sageways into the sacre4
area (fig. 65) and at prominent locations, to create a dramatic effect
for tlle worshipper approaching the main stii.pa (fig. l, shrine P2 a1
tlle northern entr<mce). However, image shrines were always place4
dose to relic structures,46 and their distribution helps us to undert
stand how a devotee would have moved through dle site and ho"~
this would have ch<mged through time (see Chapters Six and Seven)!
For example, at Jaulifui nine image sluines were added to the past
sageway leading to the sacred enclosure of the subsidiary stilpa D6i
indicating tll.at this particular stupa was important enough to war'~<
• I
rant ongomg patronage.
Small images in stone or in cheaper materials like clay or stucc~ I
., Many of the ir·on hinges sun~ve, and the fi·ont faces of the image shrines hav(,l
a recessed step to accommodate the shutter. ·
46 .Except lor a few late sh:rines placed in monasteries (sec section 6.6 Phase IIJt
Copyrighted material
..
•
OVERVlEW OF GREATER GANDHA.RA 33
1.3.6 Monasteries
A sacred area could not exist by itself; it relied on the care pro-
vided by a resident monastic community. All the religious complexes
in Gandhara invariably have one or more monasteries, or viltiiras;
two main varieties are found in this region. Well known are the
large, multi-storied quadrangular structures, like the one at J aulian,
which have monastic cell~ organized around an open courtyard and
attached rooms for cooking, storage, etc. (fig. 27). When the steep,
mountainous terrain precluded the construction of quadrangular
monasteries, clusters of mountain vihiiras were built. They consi~t of
groups of small, multi-sroried structures set on the hills acljacent to
the sacred area. Each of the independent residential units typically
has two or three rooms per floor, facing out onto an extended foun-
dal'ion serving as a porch (figs. 68, 76 outlying structures).
OJtadran(JI.dar Monasteries
Large mull'i-storied quadrangular monasteries were generally pre-
ferred not only because monks could live together, but also proba-
bly because these large buildings surrounding a courtyat·d offered
more security from passing bandits. The foundations of tower-like
structures can be seen at several sites. A good example is the square
solid foundation attached to the exterior of the northern wall of the
Dharmarajika G monastery in Taxila (fig. 16).4'
41
Also on the northern edge of the Bhamiila monastery. At J amii'tl Garhi, struc-
ture 6 immediately to the east of the main stiipa. coun stands on a massive base,
suggesting 1.hat it too may have functioned as a defensive stmcture of some sort
(fig. 61 ). Buddhism was not exempt from forms of self-defense involving violence;
arrowheads were found at the sites of Kunala, Dhannarajila, and Thru·eli.
Copyrighted material
34 CHAPTER ONE
48 Monasteries were founded with main stupas in Taxila at.Jauliai'l, Mohra Moriid~
Pippala, Kalawlin, Ku.nlila, and the Akhauri-Khaclq· i\lloh.1il sites.
• Salomon, paper presemed at the conference "On the Cusp of an Era: Art id
the Pre-Kushan World," Nov. 8- 12, 2000, Kansas City.
50 Schopen, paper presented at the conference "On the Cusp of an Era: Art id
Copyrighted material
OVERVIEW OF GREATER GANDHARA 35
under the wall of cell 21. Marshall's cross section of this monastery shows lhe slope
of this Aoor, ald1ough he did not discuss d1at feaw rc in his report.
" Although only one early photo documents a trapezoidal doorwiL)' in Swiit, lhis
architectural element is common in the relief sculpture.
$-I Marshal], Taxi/a, 380, 87.
~~ Salomon, Allchin, and .B amard, A11C'itnt Buddhirt SC1'olls Jivm Gmullwl'll: 71~t BriJislz
Librmy h?umJStlli Fmgmm/s, 20- 2. The most complete section of this unprovenanccd
Copyrighted material
•
36 CHAPTER ONE
5th century site of Ajanta (see section 6.6.3 M onastic Images: Late
Phase lli Gandhaku\:i Image Niches).58
Mountain Viharas
Mountain vi/zii.ras, found primarily in the Peshawa1· basin and Swat
(figs. 44, 61 , 68), can generally be dated to phases II- TV.59 They
differ dramatically from the quadrangular monasteries; most were
organized in clusters perched on steep hillsides, the associated sacred
a.rea being located on the mountain crest or valley Aoor.
Each individual mountain vilziira was a two- or three-story build-
ing (two to four rooms per Aoor) having a small front porch (fig.
76). Typically, the upper Aoors were reached via external stairways;
massive fow1dations were constructed to support these structures on
steep hillsides. The foundations are not solid but sit above trabeated
chambers apparently included to increase structural stability, although
they were also probably used for storage. 60
The uihiiras are groups of two or three buildings, depending on
available space, and are linked together by common terraced areas
bounded by low walls, a feature most clearly seen at Thareli (figs.
68, 75 and section 7.3.2 Thareli: Monastic SmaU Sacred Areas).
These terraces have private devotional areas at their junctures, here
referred to as monastic small sacred areas.
T he monastic smaU sacred area, reproducing on a smaUer scale
the plan of larger public sacred areas (fig. 75), had a function sim-
ilar to that of the shrines foW1d in the quadrangular monasteries.
:M ost monastic small sacred areas have a single small stupa. in the
middle surrounded by shtines that would have contained life-sized
in1ages. 61
~ W. Spink, Ajanta: A Briif History 011d G11idt (Ann Arbor: Asian Art Archives of
the University of Michigan}.
"' Little is known about this category of monastery because so few have been
excavated; hw1dreds of these structures survive at sites like Jamal Ga.rhr, Taklu-i-
b5.hi, Thareli, Mekhasanda, Tokar Da.ro, and Abba Sahib China. ·
60
Callie ri, Saidu Sluuij, t!te Moruzsury, 69- 85.
61 The central sttipos in these monastic small sacred areas are similar in form,
size, and probably embellishment to the middle phase m type (see section 6.5 The
Phase 111 Small SllifiCLr; section 7.3.2 Thareli: Monastic Small Sacred Areas}. The
image shrines surrounding the small central stiipa were sometimes constructed as a
set by a single patron (Jamill Garhr area 2 being a possible example), but more
often the image sluines appear to have been added one at a 1:ime by diflercnt
patrons. Thus, rypically a group of heterogeneous shrines will ring the small <:en-
tml stripa (field observa[ion, 1993- 94).
Copyrighted material
38 CHAPTER ONE
62
Mizuno, ed., Mekhas(mda, 87- 8; Mizuno and 1-liguchi, eels., 17umli., 146--50. .
Copyrighted mate! al
CHAPTER TWO
The edicts carved on boulders and pillars at the behest of the north
Indian Mauryan emperor Asoka perhaps helped to consolidate his
empire and to stress some Buddhist values such as dharrna or ahimsii.
More than half a dozen of these rock-cut edicts survive in Afghanistan
and northem Pakistan, but the most famous and pertinent to the
Peshawar basin tradition is rhe one at Shabazgarhi. 1 i\lthough the
edicts do not clirectly address tl1e inr.roduction of Buddhism into
northwest India, most scholars consider them to be evidence of the
beginning of Buddhism in tllis region.
In the legends of Asoka (recorded in the later Afokavadiina), this
emperor is credited witl1 rl1e extraordinary act of constructing 84,000
stzipas containing true relics of the Buddha. 2 It is by means of this
mythic act that relics were "credibly" established throughout the
Indian subcontinent; however, such sti'ipas are very cliflicult to iden-
tify in the archaeological record. It has been suggested that BuddJ1ist
__.... -·
1 For a di scussion of the Kharo~\hi ASokan inscription from Shabaz&>"<~rhi in temlS
of its social and political significance, See R. Thapar, Asoka mu/ lhe Decli.n~ if lhe
Mauryi/S: Wiilz a New A.flerword, Biblwgraplry, and lmlex, rev. eel. (Delhi and New York:
Oxford University Press, 1997), 123- ll. Several M:awyan inscriptions from Afghanistan
have come to light: (I) <ul Aramaic iJ1scription ti·om Laghman {fifth or seventh pil-
lar edict of Noka); (2) a second Aramaic inscription from Laghman {dated to tenth
year of A.!oka; discusses his conversion to Buddhism, expulsion of vanity, a nd pro-
hibition against fishing); (3) a bilinguaJ rock inscription (Greek and Aramaic) found
at Kandah;;u·; (4) a Greek building ulSCI~ption from K andahar {end of ASoka edict
XJI and beb•inni.ng of edict XUl); {5) an lndo-Ara.maic inscription from Kandahar
(pa rt of Asoka piUar edict Vll). See M.acDowall and Taddei, "The Early H iswric
Period: Achaemcn.ids and Greeks," 192- 8}. For· a description of a fragment tl1at
ma)' record an ASokan Brahmr i11Scription /i·om the Buner area of Pakistan, see
B. N. Mukherjee, "A Note on a Briihmi Inscription of Asoka fr-om Pa kistan," in
A~qyanfvf, cd. G. Bhanacharya (Del11i: Sri Sat.guru Publicat:iOilS, 199 1), 5 1- 4.
2 Versions of this text da te to the 5th to 16th centuries C.E., the 5tll cenrwy
Sri Lankan i\IJ:ahavarpsa pi-ovidiJ1g crucial evidence. See J ohn Su·ong, TM Legend If!
IG.ng Asoka: A Study mul TraJislalion if llze A.fokiiuadiino. (Princeton, 1J.: Princeton
University Press, 1983), 18- 2 1, 109.
Copyrighted material
J
'
40 CHAPTER TWO
3 An inscribed ASokan piUar stands true south of Safici stUpa I; see J. Mru·shaiJl,
A Foucher, a.nd N. Majumdar, 17re Ado11ummts of &melt£, 3 vok (London: Probstha~i
1940), 25-9. In 1998, at Satdb!u-a, one of the Bhilsa topes, T was shown a polishe
Chu1~ar sandstone fragment of a bull that was found within the stone encasing
main stii.pa (cir ca lsr--2nd century B.C.E.) adjacent to the brick core. For a discu -
sion of Satdhara, See R. Agrawal, "Stflpas and Monasteries: A R ecent Discove
from Sardhara, India," in Soutlt Asian Ardweology, eel. R. Allchin and B. AUcl"
( 1995); D. Sahru, Guide Ill tlte Buddhist Ruins of SamaJh (Delhi: Antiquarian Boo
H ouse, 1982), 4. .
• MarsbaU (T~ila, 235) made this assertion with respect to the .l:>ha.rma.rajil.ii
stitpa in Ta11:ila.
~ The name Dharmariijika was found at Butkara I painted on a vase associated
with GSt I and also on a schist slab used in the con.st•·uction of the base of th~
stupa (Faccen.na, BaJkora I, vol. 3, no. I, 167).
6 Marsh all, Ta:..-ila, 235, 56.
7 Ibid., 235.
Copyrighted materi 11
ARCHITECTURE AND SCULPTURE FROM PHASE I 41
' Four 2nd century B.C. E. Indo-Greek coins and another 26 coins fi·om the 1st
centwy B.C.E. together Sl.l,gg~oost a founding date no em·lier than the late 2nd cen·
tury B.C.E. (Enington, "Numismatic Evidence for Dating the Buddhisr R emains of
Gandhara," 192). vVhile thi.s provides a general date, some specific strucurrcs also
have associated coin finds. For instance, in the monastic a r·ea of the Dhar-mariijikii,
the kafijrir ashlar sl!ipa 4 had in its relic deposit a coin of Azilises (10 B.C.E.- 6 C.E.)
and another of Augustus (27 B.C.E.- 14 C.E.), which points r.o a 1st cemury C.E.
date for building this strucrure (Marshall, Taxiln, 27 7).
9 !V!arshall did not mention these structw·es in his excavation reports. These
probable stiipas clearly date to phase I; subsequent phase I structures were built
over them; this group was mapped during a field swvey in 1993-94. They consistently
exhibit phase I ea1·Iy rubble and kaiijflr ashla.r masonry rypes (see Appendix A).
10 The D4 stripa. immediately opposite the south gate has in its core a round-
based sttipa similar to the set of I 2 stupas or pillars ringing the main stupa (MarshaU
noted this stiif>a in his 1913 report but did not discuss its significance in d1c final
publication; see MarshaU, "Excavations at TaxiJa," 15 and Taxi/a, 244. Dur·ing
phases 11 and IT!, the northern quadrantS of the prada/cyi!IOpaiha were rebuilt. Therefore,
additional su·ucull"es relar.ed to the Y group, discussed above, could be concealed
under these later monuments .
11
MarshaU, Taxiln, 238.
12
Marshal! dated these .r,rateways to his diaper masonry period, or into my phase
II, but th.is is not in ~~f:,"'ecmcnt with the type of /avJjilr masonry employed and does
not account for the fact that the phase f1 0 9 shrine was constructed over the edge
of the south gate (Marshal~, 1916, 12). By phase III, gates Yl- Y2 and Y7- YB were
incorporated into the fabric of' late shrines.
13 Of the original 12 stilpas or pillars, the bases and the L o•ver drum of 11 of
them a.l'C still extant and one (0 9) is known 1:hrough excavation. lt is possible that
there ' "as a thirteenth stilpa or pillar, framing the nortb &rate, which has gone uJmo-
ticed; it would be buried under the S 10 and BS, 6, 9, and I0 sh.rines. T here are
very few diAerences among these sltipru or pillars in terms of size, construction, a.nd
overall form. The one stiipa or pillar r.bat has a round base (B3) might be an exam-
ple of e1u·ly restoration.
" CommonaJ:iries among the five relic deposits include similar inner gold reli- .
quaries coniaining tiny fragments (identif-Ied by Marshall as bone), in tb.e deposits :
of R4, B3, and SS, and a tiny silvery b l'OilZC reliquruy in BG. Each deposit con- ·
t:a.ined beads and other precious ivory and gold objects. :Fw-ther, both. rhe R4 and :
B3 reliquaries wer·e found 1.52 m (5 feet) below the ground-level of the stupa or .
pillars, an unusual feature that links these structur es (Marsh all, Tll:Xi.kl, 241- 3). See
also Kuwayama, ''In the T im e of Late Sirkap and E~u~)' Dhannari'~ikii.: How Taxila
l.ntroduced Stfipa Architectw·e," paper presented at tbe conference " On the Cusp
of an Era: Att in the .Pre-Kushan World," Nov. 8- J2, 2000, Kansas City.
•~ Faccerma, paper at the conference ''On the C usp of' an Ern: A.t·t in the Pre-
Kushan World," 'ov. 8- 12, 2000, Kansas City.
16
lt is possible tl1at there are additional early su·uctures, now butied under piles
of excavation debris 10 r.be east.
17 Marshall's sug&resrion, but it is open to debate because no fi.ncls were reported
CHAPTER 1WO
stripas. The northern A and B courts at the site were both configured
so that the individual mona~tic cells encircled a large kmrjiir ashla:Ji'
stiipa. 19 The lay population probably would not have used these devo•
tional centers, as they were isolated ITom the main sacred area (by
enclosing waUs in the case of court B). These two courts were sep•
arated from the sacred area by a row of monastic cells (Vl- 5) (figs,
I , 16, northern monastic areas).
In the long EFT monastery on the west edge of the Dharmarajiklil
complex are two Jtiipa shrines, E l and E2, that appear to have been
restricted devotional so·uctu:res, because they are enclosed in cham•
bers that are attached to the residential monastery. Rather than organ•
izing the residential ceU.s around a large central stiipa, as v.ras the
case with courts A and B, in tllis instance one of the stiipas ~
placed in a pre-existing cell and the other was fabricated togethev
with an enclosing shrine structure complete with a circumambula-;
tory path. These shrines face the public area, but they are in. enclosed]
chambers attached to the monastic living quarters, and thus are quita
unlike the small niipas in the public sacred a.rea. 20 T hese monastia
stfljJa shrine~ are a vital testament to the devotional needs and inter-<
ests of the monks, in contrast to those of the public (see section 6.6!
Phase IH Use of Images and Relics in Quadrangular Monasteries).21
19 Borh courrs A and B a.re in a very fr1rgmencary condition, but they minor tho
organizational distribution of main stupa. and monastic cells seen a t Sirkap in, among
olhcrs, the t A and IF complexes. Clearly, Ohannarajika stiifXJS 1 and 4 were impor·
tanl, bt:catL~e they remained in use into phase rv. Later construction and use have
di srupted the phase [ form of these monasteries, especia.Uy court A. See also
Kuwayama, Tilt Main SIItj}(l. at Shiih·ji·ki-tjlt8ri, 88-9.
20 Variations between the masonry of st1ipa Et and the cell walls suggest that the
stiif}(l was inst'<~.lled in a pre-existing monastic cell (field observation, 1993- 94).
21 Although monk's ash es could have been placed in an independeor stilpa shrim:;
suitable for veneration, it seems unlikely that this sllipa. shrine was primarily a fuoer-
aiy monument. Marslmll argued d1at the eady pmctice of placing a stupa in a res-
idential cell commemorated or enshrined an important member of the sang/10. who
had died, thus filling the cell and converting it permanently (MarshaJJ, Taxi/a, 246).
Schopcn has also addressed the issue of the monastic dead (Schopen, "Burial 'Adi
Sanctos' and rhe l)hysical Presence of the Buddha in Early Indian Buddl1ism: A
Study in the Archaeology of Religions"; "Rin1al Rights and Bones of Contention;
More on Monastic Funerals and Relics in the Mulasa:rvastivada-Vi.naya," Jourrud qf
f11diOJl P!Lilosoplry 22 ( L994); "Stilpa and Tutha: Tibetau Mortuary Practices and an.
Unrecogni7..ed Form of Burial Ad Sanctos at Buddhist Sites io India," in 111e BuddJ1isl
Fomm 111 1991- 1993: Papers i11 Honour and Appreciation qf Professor Daoid S!lfort Ruegg't
Calltrilmtion to lndologica~ Buddhist and Ttbeum Studw, ed. P. T. Skorupski (London:
t 994). While these arguments are helpful, they do oot address the presence of stilpm
shrines and later image shrines wid1in Gandl1aran monasteries.
Copyrighted mate 31
...
ARCHTTECTURE Al'iD SCULP'tUR.E l'"ROM PHASE 1 45
22
Of the 12 stiipas or piUars arOtmd the prada/rfiwtpatha, the only one that breaks
the regular pattern is the R4 stupa or pillar, which marked the soud1 edge of the
west gate to the main stiijJa and wh ich was later converted into a direct-access relic
stuine.
23
For a dL~cussion of this evidence, see Appendix A. Signiftcant works on this
topic include: P. Call.ieri , "The North-West of the Indian Subcontinent i.n the Indo-
Greek Period: The Archaeological Evidence," in In the I.muf if" tlze C1Jphons: Papers
on Ce11tral Asimt Archaeology and A1lliqui~, ed. A . lnvernizzi, Monografie di Mesopotamia
(Fir·enze: Casa Editrice le Lettere, 1995), 293- 308; S. Dar, "Dating the Monuments
of Taxila," in Urban Fom1 and Mea11i1v; in SouJh Asia: 17ze Shaping if" Cities f rom. Preltiswric
to Prerolonial 7imes, ed. H. Spodek and D. Srinivasan (Hanover, New Hampshire
and London: University Press of New England: 1993); Erdosy, "Tax.ila: Political
History and Urban Structure"; Marshall, Ta.rila, 30- l, 37- 8, 49- 50, 123, 210- 13.
Copyrighted material
'
46 CHAPTER TWO
2' R. Coningh:m1 and B. F.dwards, •'Space and Society at Sirkap. Ta.xila: A Rt>
E.xamination of Urban form and Meaning.'· Anamt PakiJtm• XIl ( 1997 8): 57 60;
Marshal!, 7a,\1la, 50 I, 54, 58, 63, 67, 73, 83, 91, 142.
1' Callicri, '·Buddhist l'rescncc in the Urban Senlements of Swat, 2nd Cen~
B.C. 4th Crm. i\.D."; Callkri. Filigen1.i, and Stacul, "Excavations at Bir-Kot,
Ghwanda.i, Swa1: 1987.'"
26 Marshall, 7 ttrilt•, 142 44.
ha ran Art," in Souret.r qf Gtmdhamn Buddhism: ArdllUtllofU, Art a11d Texts, ed. K. Bclucnc(.L
and T'. Brancaccio ( l'oronto: University of Toronto Press, in press).
•
ARCHITECTURE AND SCULPTURE r'ROM PI-IASE I 47
"JO Marshal! argued that Sirkap was abandoned at a lime coinciding with the
introduction of' diaper mason•y, at the boundary between phases l and 11. He pos-
tulated that an earthquake occu•·red in 30 C.E., creating a clear horizon or destruc-
tion; repairs were made by using the more stable phase U diaper masonry technique
(i\l[arshall, Taxila, 137). While an earthquake cerlai.nly could have occurred at Tax:i.la,
Marshall's e\~dence is circumstantial, and dtis transition in masomy need not be
linked LO such a dramatic event.
51
For a full review, see Erdosy, "Taxila: Political History and Urban Structure,"
669- 73.
" Late coins fow1d at Butkiu-a l include Varlmm TV, Yazclagird n, Khusra u !!,
Umayyad, Salm b. Ziyad, and Vakkadeva (Erringwn, ". Tuntismatic Evidence for
Dating the Buddhist Remains of Gandhara," 211, 216).
Copyrighted material
'1
48 CHAPT:&R TWO
Copyrighted material
ARCHJn: CTlJIU: AND SCULPTURE FROM PH."\SE I 49
stilpa, in the form of five columns and a single smaU stupa (fig. 97 ,
columns 66, 140, 149, 209, and 214 and stiipa 2 1 8). :~; These columns
were not org<mized to enclose the prada/cyinapatha, nor is there evi-
I
dence of a vedika or other fom1 of demarcation. This somewhat ran-
I dom organization of subsidiary structures is markedly different from
that found in the earliest archaeological evidence from the Dharma-
rajika complex in T axila, where the donative monuments were orga-
nized around the circuman1bulatory path or at entrance points.
The next major renova tion of the main sttipa (GSt3) involved a
total encasement of the earlier dome and plinth.36 The sW.pa was
considerably enlarged , so that four axial stairways led to an upper
prada/cyirwpaJ!za enclosed within a low vedikii railing (fig. 98).
The relative daring of the early Butkara I core stupa (GStl- 2) and
its subsequent modifications and encasement (GSt3) is problematic,
because very few coins were found in association with the earliest
structures. A single Maurya.11 coin hint~ at an early foundal'ion , and
certainly GSt I was in place by the 2nd century B.C.E. (early in
phase I). T he modifications and encasement (GSt2 and 3) can be
placed in the middle and end, respectively, of phase 1.3' In con-
junction with the next encasement (GSt411), 107 coins were found-
mainly of Azes IT and K ujula Kadphises-and one reliquary, which
may have been a later addition , containing coins of Huvi~ka. 38 Tllis
Copyrighted material
50 CHAJ'TE R TWO
Many scholars in the 19th and 20th centuries were preoccupied with
tracing Greco-Roman influences that they believed shaped Gandhara .
This line of inquiry has produced inconclusive results in terms of
elating and characterizing the nature of Ganclharan cultur-e. It did
serve to identify the many foreign groups that came into contact
with the people of northwest India, but the nature of these interac-
tions remains unclear. Perhaps this is because the people in Gandhara
were actively filtering foreign "influences" and selectively appropriating
forms and ideas coming fi·om abroad. Gandha.ra's art, architecture,
and religious tTaditions cannot be seen as a pastiche of Mediterranean,
Parthian, Chinese, and south Asian influences. The view of Gandhara
as a passive recipient obscures and denies independent development
and has led to interpreting tllis culture as one that was polluted-
or ennobled- by the Classical tradition. 40 It is especiaUy tempting to
see the phase I beginnings of Buddhism in Gandhara simply as an
extension of the larger north Indian tradition; without doubt, Buddhism
was introduced to Gandhara from the Ganges basin. Only phases I
and Il in Ganclhara include north Indian architectural forms, all of
which were rapidly changed or modified to serve local Gandharan
Copyrighted materi I
ARCHITECTURE AND SCULP'lURE .FROM PHASE I 51
-··--
" T he Bhilsa stii/JaJ, on ne-drby 1~dges surro~Ulding SaJ'ici, were constn1cted between
the 3rd and I st cenruries B.C.E. See K. Behrendt, "AJ1 Unnoticed Relief fr·o1n IJre
Bhilsa Topes and Its Relationship to the Sculpture of Sai\ci," South Asian Sludies 16
(2000); A. C unningharn, 77ze Bhii.sn. Topes; or, Buddhist Monumen!.J of Cenlral !tulia:
ComprisiJ~g a Brwf Hist.oricai Sketch of du: Rise, Progress mu/ Dedi!u: of Buddhism; witlt llll
Accowrt of dte Operti!tg mui E.wurrirullioll of du: Varimu Groups of Topes arozaul. Bltitsa (London:
S n'l.ith, Elder, and eo., 1854}; Malwbodhi: 77te Great Buddhist Temple w1der lile Bodki Tru
oJ BudJJJw-Cf!YO (D eUri: Munshiram M:anoharlal, L892; reprint, 1998); and 171e Stupa
of BlurrlmL· ll Buddhist Mommrmt Omammted wilil )lftmwrous Sculph.rts, 2nd ed . (Varanasi:
lndological Book H ouse, 1879; reprint, L962); S. D eo and J. J oshi, Pmm:i E~c{IQ(Jtion
(1969- 70) ( agpw~ 1972); M. Willis, Buddltist Rdiquari£sfrom An.UmJ bulia (London:
British Museum, 2000).
' 2 In a very general sense it is reasonable to assume that ma ny of the li·ees1a nding
round-based small slllpas found in sacred areas were created during phases I and
!I when lin.ks to northern lndia were strong. Few slflpos of this type appear a t Late
sites, and some of these ma)' be inswnces of ea1·lier stufJOS being r·estor·ed in Late
masonry types. Small row1d stilJXIJ ue found in phase f masomy at the Obannariijika
complex (Sitlpa. B3 and the core slilpa in D4), among the early finds fi·om Butkara
I (sltipas 66 and 209), and at C hal--pat (ashlar s111pa). They a.lso a1·e found at a vati-
ety of phase 11 sites such iiS at Butkm-a lJJ and Ma1janai in Swlit, at Ranigat in
the Peshawar basin , and in the mon a.'tery of Mohrll Moradu in Ta.'--i.La. A few
rou nd sliipas do appear within the phase UI remains from the Dharmariijika com-
plex, slfijlas J 6 a.nd U 1- 3 (a related group that appear to have been constructed at
Copyrighted material
52 CHAP'l'ER 1WO
the same time). One of these stiipos, U I, which eJthibits phase IT! masonry, has a
very early relic deposit that, on tbe basis of coins, can be dated to phase [ (Ma rshaJJ,
Ta:r:ila, 272- 3). Thus, at least in the case of the U l stupa, its cwTent form may be
a phase ill restoration of an earlier archaic stilpa type. Round stupos can be use-
ful in determining chronological parameters, but only in the presence of other sup-
porting evidence .
., lt appears t.hllt it was stt·ucturally impractical, if not impossible, to construct
a taU, vertical-walled aTJI/.a with stiijm of this size. f or a gcner-d! discussion of south
Asian stiipa construction, see Kuwayama, The A1fain Stupa at Sllfilt·:Ji-ki-<llmi.
H But.ka.r a l stilpa GSt 1- 2. The core of the Ranigat main stripa has a similar
columnar a1'(ia; however, this early phase U example has a square base typical of
the northwest.
•s An intensive su.rvey or the extant main stilpas in Swal, Taxila, and the Peshawa r
basin reveals that it was usually impractical or simply impossible tO circumambu-
late at the level of the drum. By phases IT and ill the base was often quite high;
at a late sl ilpa like Shna.isba in Swlil, the full would have been more than 6 m.
46 While the genesis of such uediklis in nonbem India is unknown, there is a small
monolit!Uc example in a Buddhist context from Siimlith that may date as eady as
the Mnuryan petiod; it is made of polished Cbii~1ar sandstone. However, Asher and
Spink ("Maurya l'lgural Sculpture Reconsidered," Ars Orientalis 19 [ 1989]} ques-
tioned th is polishing teclm.iquc as a dating criteria, because polishing of sandstone
continued into the Kushan period. Small vedikii railings arc also lo und at the circa
1st century B.C.E. Bhilsa sites .
., Facccnn a, Bulhlra I, vol. 3, no. I , 57- 74.
43 Marshall said that the block A vedikii was pieced together \\~th indepcndendy
cut stones and then finished with Lime plaster (Marshall, Taxi111, 143, pi. 23c; for
the block IF and G uedikls, 163, pi. 34c; for block C, 167, pi 29; for' block C,
191).
Copyrighted material
54 CHAPTER TWO
49 The fmd spots of the Rani&rat railin&rs arc poorly reponed , alr.hough the exca-
vators noted that they were probably placed arOlUld the circular upper drums of .
small stilpas, as the "coping stones are curved in shape" (Nishika.wa., Odani, and
amba, Preiimi!lory Report 011 Gandluua Budd1tisl Sites, 8 7). A$ many as five different
uedikas may have OJiginally adomed the early rtiipas surrounding the main stiipa in
the east sacred area, but r.his can be onl)' roughly deduced lrom a close examina-
tion of the published photographs. See Nishikawa, Rtmigat, pis. 168- 70.
Copyrighted mater 31
•
ARCHTTECTURE 1\J'ID SCULPTURE FROM PHASE I 55
I
2.3.2 Use of Columns During Phase I
!>0 J.
Williams, "A Recu t ASokan Capital and tbe Gupta Attitude T owards the
Past," llrtibus llsiae 35 ( l98H32).
$t However, this expla nation does not account for the depiction of such a col-
umn on the balusu·ade of S1il'ic1 sillpa D, a strtpa. that contained relics of monks.
52 Cunningham •·eported finding a column to tbe east side of d1e main sllipa. at
J am!tl Garbr (see fig. 61), but Errington ("The Western Oi covery of the Art of
Gandhara," 229 - 30) noted critical inconsistencies in his reports.
Copyrighted material
56 CHAPTER TWO
" For a fUll discussion of t.be use of columns in Swat, see FaccenM•, Saidu. Sharij.
the StWred Area, 5 1, 481- 502.
M See Ralunan, "Butkara IU." Most of the evidence for the use of colunms in
Gandhltra is fragmentary, because tbey were fabricated in masonry rather than
being monolithic, like those of northern lndia. As a result, commonly only the bases
on tbe corners of stiipa.r survive. However, many depictions of columns apperu· in
relief sCIJ.Iptur~-s and in stiipa-shaped reliquaries done in stone and metal.
" These colunms were built during pbases l- ID. Relic deposits were found in
nos. 33 and 135. For descriptions of individual columns, see Faccenna, Buikm·a I,
vol. 3, no. 2, 263, 265, 296-300, 437; for columns wid1 re[jc deposit~, see Faccenna,
Butkara I, vol. 3, no. I, 57; vol. 3, no. 3, 643.
$6 Faccenna has discussed fully the significance of columns found in sacred areas
in a paper presented at the conference "On the Cusp of an Era: Art in the Pre-
Kushan World," Nov. 8- 12, 2000, Kansas City.
B.C.E. to the lst century C.E. are highly diverse.57 Multiple reli-
r,r:ious traditions are represented, and the objects were produced in
a range of local and foreign styles. Ant1uopomorphic depictions of
non-Buddhist deities clearly appealed to t11e taste of the urbanites of
Sirkap.
The small cosmetic dishes decorated with relief sculpture, com-
monly referred to as toilet trays, have been singled out for study
because they provide a link between the heterogeneous imported lux-
ury objects found in Sirkap and the later Buddhist narrative tradi-
tion. The trays were loca!Jy crafted in schist and exhibit what have
been called lndo-Parthian and Hellenistic artistic styles. Some later
Buddhist narrative panels display these same styles and, like the toi-
let trays, they depict groups of interacting figures in small, framed
compositions. The Indo Parthian style has been connected to some
Buddhist sculpture from Butkara I and the Dharmarajika complex,
in particular to a few complex moldings and figural garland hold-
ers used in the decoration of small stupas. 58 A garland holder from
Sirkap (fig. 12) can be considered Indo-Parthian in style on the basis
of it~ wide-open, bulging eyes and the very fine pleating of the drap-
ery (close parallel lines).~9 It has been noted that this Indo-Parthian
style is also strongly related to "Indian stylistic elaborations, partic-
ularly those of Bharhut,''60 and the narrative content of the north
Indian component is ctitical to understanding its appeal in Gandhara.
This garland holder from Sirkap has a donative inscription that
reads: " Presented by Sruvatrata in the Vihara, in honour of his
57
These objects include many terracotla and stone figurines, a V'ariety of tcrra-
cotta heads, schist toilet trays, a small bronze (which Marshal! identified as depict·
ing the Egyptian god Harpocratc:s), and a variety of h.h~uoy objects bearing um1ges
of Greek deities. Tllis last categ01y includes a silver repousse head of Dionysus, a
gold refX!ussi figure of Aph rodite, and many pieces of jewel!)' and coins sho"~ng
images of deities (see M.arshaU, Taxila, 137- 2 1.3).
58 D. Fabrcgues, "The Inclo-Parthian Begumings of Gandhara Sculpture," Bu.lleti11
l!f ll~e Asia /nstituu 1 ( 1987). Fabr~gucs did, however, argue that her Buddhist mate-
rial is stylistically linked to the van Lobuizen-De Leeuw early Buddha images and
rh at they arc not widely separated in time. Fabregues, "The lndo-Parthian Beginnings
of Gandhara Sculpture," 4l. See J. .E. van Loh uizen-De Leeuw, "New Evidence
with Regard to the Origin of dte Buddha Image," u1 South Asian ArdzaeoiiJgy, ed.
H. Hartcl (Berlin: 1979).
~ Fabrcgut-s, "The Indo-Pardl.ian Beginnings of Gandhara Sculpture," 40.
60 M. Taddci, "Oral Narrative, Visual Narrative, Literary Narrative in Ancient
Buddhist India," in l11dia, Tlhet, Clzi11a: Genesis a11d Aspects l!f TradiliiJnal NarratiJJe, eel.
A. Cadozma (Florence: L999), 80.
61
T his garland holder was found loose but in a stratigrftphic context in block J
of Sirkap (Marshall, "Taxi!a," 118, pi. XXVIII, 9; Taxilo, 700-Q2, pi. 213, no. I l~
62 The A L5 and 0 5 small stUpas were embellished with phase TTI stucco Buddha
images that were individually donated, as indicated by inscriptions (lbid., 374- 75).
"' S. Konow a.nd H. Hargreaves, in Marshall, Taxi/a, 702.
64 T lus sllipa is no longer extant, but Foucher published two photos, ta.ken .in thCI!
l880s, together wid1 a plan. Foucher ml.lSt have seen this stlipa, because he said!
that it was located near the fort of C hakdara in Swat. See A. Foucher, L'Art Grko-
Bmuldltique du CandMra; Elulle sur les Origine.r de l'b!fluence Classique dans i'Art Bouddflique
de l'bule et de l'Extreme-Oriant, 2 vols. (Pa ris: E. Leroux, 1905), 67- 69; figs. IO- l2.
Copyrighted material
ARCHITECTURE AND SCULPTURE FROM PHAS.E I 59
65 f or a discussion of how d1ese stair risers can be reconstr·ucted and the scenes
identified, see Errington, "T he Western Discovery of the A11 of Gandha.ra," 246- 51.
66
Taddei, "Oral Narrative, Visual 1an-ative, Lit.enuy Narrative in Ancient Buddhist
India," 8 l.
6
' rbid., so.
Copyrighted material
60 CHAPTER T\.VO
Copyrighted material
•
CHAPTER THREE
The Chinese pilgrims Faxian <md Xuanzang wrote about relics that
could be seen and touched in Gandhara (direct-access relics). Their
accounts stress the importance of the alms bowl and the Uf?~i$(1 por-
tion of the skull of the Buddha, which appear to have been central
objects of veneration for local p eople and rhose on pilgrimage,
although other signmcant reJjcs, such as the eyebaU, staff, and robe
apparently were also on display. In 630 C.E. Xuanzang visited the
shrine of the sk'1.11l in Ha<;k,la and described it in the foUowmg manner:
Copyrighted material
62 CHAPTER TIIREE
There is here a two-storicd tower: the beams are painted and the
columns colored red. In the second story is a Little slllpa, made of the
seven precious su bstances; it comains the sL:uU-bone of Tathligata; it
is I foo L 2 inches round; th e hair orifices are distinct: its colo ur is
whitish-yellow. It is e nclosed in a prc<;ious receptacle, which is placed
in the middle of the stupa. 2
T his passage tell~ us not only about the shrine that housed tlw skull
bone and its precious receptacle (herein referred to as a stupo/rdi-
quary), but also about the physical nature of the relic itself. The
Chinese pilgrim also described the practices surrounding seeing and
touching the Buddha's si..'UII.
Those who wish to make lucky or unlucky presages (marks} make a
paste of seemed earth, and impress it on the sk-ull-bone; then accord-
ing 10 their merit is th e impression made.... All who wish to see the
skull-bone of Tathiigata havt~ to pay one gold piece; those who wish
to take an impression pay live pieces ... tJ1ough the c harges arc heavy,
the worshippers are num croux. 3
Actually touching the skull-bone was very important, worth the high
price. This relic had spirilllal importance for the d«>votees, pilgrims,
and monastic community, besides being a source of revenue.
Two hundred years earlier, when Faxian was in Gandhara, circa
400 C. E., he visited a shrine in the Peshawar basin containing the
alms bowl of the Buddha, a nd wrote:
The alms-bowl of the Buddha is still in this country ... There are per-
haps iOO priests. At the approach of noo n the priests bring out the
alms-bowl, a nd with the Upasil kas make all kinds of ofterings to it....
lt lthc bowl) is of m ixed rolour, but yet c hiefly black. The lour divi-
sio ns are quite clear. . . . lt is glistening and bright. 4
This account indicates the impo rtance that such relics had fo r the
foreign pilgrim. After Faxian's two companions, Pao Yu n and Sang
King, made offerings to the begging bowl of the Buddha, they
returned to China;l apparently, seeing the bowl was the climactic
goal of the pilgrimage. The writings of Xuanzang, Sung Yun, and
many others, including more than half of the 96 Chinese monks
whose accoun ts are preserved in the Gaosengz/man (The lives qf Eminent
2
Hsilan-tsang, Si-'Yu-Ki. Bluldhist Rl't:MdJ ~l the We.rtern WtrrM, 95 6.
' Ibid., 96 7.
4 Fa-hian, Si-Yu-Ki. Buddhist Ruords ~j'tJu IVutem IJ1qrfd, xxxiii.
' Tbid.
(, y hied me 'e 11
"flU: DEVELOPME!'IT OF RELIC SHRJNES: PHASES I At'IJD ll 63
Monks, edited in 519 C.£ .)6 attest to this also. It appears that Chinese
pilgrimage gready increased in the 4th and 5th centuries C .E. and
that the shrine of the alms bowl was regularly visited. 7
T he importance of direct-access relics is also seen in the relief
sculpture. The mundane origin of the alms bowl is recounted in the
narrative reliefs: the fow· bowls that the Lokapalas offered were fused
into one just before the Buddha broke his fast . Perhaps more
significandy, the alms bowl is shown sitting on a low throne and
being venerated by worshippers. Iotably, these devotional scenes
appear as focal object~ in false gable stapa reliefs (fig. 122, top reg-
ister) and on the bases of iconic images (fig. 123).8 T he many depic-
tions of the Buddha's bowl in the art show the importance it acquired
in Gandhara where it was enshrined.
The sculptural evidence shows that relics, like the alms bowl, were
placed on low platforms, and this method of display was noted in
the Chinese accounts. Faxian, describing the skull bone at Ha<;l<;ia
wrote:
They place it (the skull bone of the Buddha] outside the viM.ra9 [direct-
access relic shr:iJ1e] on a high throne; taking a circular stand of the
seven precious substances, the stand is placed below (the reEc] , and a
t,>iass bell as a cover over it. All the;;se are adorned wi!.h pearls and
gems.10
The Chinese accounts consistendy tell us that direct-access relics were
displayed on thrones or platforms, and this is corroborated by the
sculptural evidence. In addition to the alms bowl, reliquaries simi-
lar to those recovered from Gandharan st!ipas are shown sitting on
low platforms. We even find depictions of relics on platforms, enclosed
in bell-shaped forms, with a clod1 draped over them (fig. 118)--per-
haps the glass bell mentioned by Faxian. These accounts by for-
eigners and the extensive sculptural evidence indicate that shr;nes to
store and display relics were important elements of Ganclharan sacred
areas.
• Kuwayama, "The Buddha's Bowl i.n Gandhara and Relevant Problems," 946.
' Ibid.
3 K. Behrendt, " Relics and Thei.r Representa tion in Gandhara," in Atlmg, ed.
have had this same function: the phase I Ja~:tc,lial C temple near
Sir.kap (fig. 14), probably the pha~e ll Sirkap aps1dal temple in block
D (fig. 10), the phase m or IV Najigram stupa and the pha~e III or
IV T okardara structure (fig. 112). The function and inclividual char-
acteristics of these openable stitpos and specialized shrines are clis-
cussed below in the appropriate chronological sections.
There i~ also clear evidence from China that helps us identity
openable stftpos and shrines used as direct-access relic structures. At
the Chinese site of Famens1 active during the Tang dynasty, four
relics of the Buddha were kept in a chamber under the superstruc-
ture of the stupa, accessible by a pa%age. 13 We know from court
documents that the Famensi relics were removed only for special
occasions, not on a regular basis, as at Ha<;l<;la. 14 Court records show
that this site received imperial patronage, and on at least eight occa-
sions between 625 and 874 C.E., the relics were brought from
Famensi in procession 120 kilometers to be clisplayed in the capital
of China (usually for a period of three days). 15
" Z. Qixin, "Buddhist Treasures from Famensi: The Recent Excavation of Tang
Underground Palace," Orie11Uitio1LS 21 (1990); R. Whit:field, "Esoteric Buddhist Elements
in the Famensi Reliquary Deposit," Asiatische Studien XLIV, no. 2 (1990).
" Qixin, "Buddhist Treasures from Famensi: The Recent Excavadon of Tang
Underground Palace," 77.
I$ Whitfield, "Esoteric Buddhist Element~ in d1e Famensi Reliquary Deposit,"
248.
16
The early date of this stiifJa is also ~1.1pportcd by the recovery of 12 coins of
Soter Megas in the debris arow1d the stttpa:, these coins were buried after the ear-
lier stiipa. was built but before d1e later one was established, accorcting to MarshaU
after this early stiipa had fallen into ruin, a round drum of semi-ash-
lar masonry was built on top of the original base, which retained
the earlier relic chamber. The refurbis ~une nt shows that this monu-
ment continued to be important enough to be repaired long aftar
the nearby urban site of Si.rkap had been abandoned.17 The multi-
ple layers of paint: within the oversized relic chamber suggest that
thjs space was not sealed. It seems likely that it was used to store
relics that were pe•iodjcaiJy taken out, and the fact that people
needed to enter the relic chamber accounts for its large size a nd
interior finish .
Sun-ounding trus possible direct-access slilpa was a phase I mon ~
tic court with rooms that, though irregular, can be compared to the
Sirkap block A sacred area, where the stiipa sat in a vilzara court.
Within trus monastic court is a large slui.ne with a raised platfo•·m
that Marshall suggested was an image shrine (fig. 15: court T and
chapel).18 Because no other image shrines, especially so large, caa
be dated to phase I (most date to phase ill) it seems likely that this
platform was used to display the direct-access reljcs that were housed
in the enterable stupa. The oversized enterable relic chamber would
have provided security for these relics when they were not on clis.-
play. The accounts of the Chinese pilgrims and the many sculptural
depictions or relics being placed and worshipped on thrones cor-
roborate this hypothesis.
Another phase I monument that may have been used to display
direct-access relics is the massive 48.1-rn-long jaJ)c;lial C temple (fig.
14), which, along with J a•:H;lia.l B, is near the main gate into the city
of Sirkap.19 Considerable debate surrounds the function and religious
("Excavations at Taxila," ·~0; Taxilt.z, 356). Only the base of this stiipa survives (circa
lst century B.C.E. Saka pe tiod), but the <-elk chamber can be t-ela ted to this phase
I stage of construction.
11
Marshal!, Tt~.liltz, 354. A small silver relic c.,-u;ket was fou nd in the debris by
an early treasure hunter, who had opened the relic chamber (Marsh.,1.U, "Excava tior!S
at Ta.xila," 40).
18
Jvfarshall, Taxw, 356.
19 Marshal! initially da ted jat;~Qial C to the 1st centtuy B.C .E. because .he believed
in use long after the ci ty of Sirkap had been abandoned, and cenainly imo phase
IJI, if not later.
20 The ]aJ:J~iaJ temple was built of coarse r·ubble with kwijur molclings and was
covered in stucco. There is evidence of repairs, which Marshall took to indicate
th at this building was f.1b.ricated before the 30 C.E. earthquake he hypothesized
(Marshal!, Taxiia, 223). As Marshall's earthquake can only be seen as a Likely pos-
sibility, this suggestion must be treated caULiously . The platform in the inner san c-
tum was added; it overlaps the base molding of the room, and a plaster coating is
vi sible in earl)' photographs, sho''~.ng that this element is not the product of 20th
century restoration. Also, the inner san ctum had doors; fragments of the jambs and
pa.rts of the fi-amc.: were found (Marshal!, Taxi/a., 224). Little significant sculpture
was recovered, with d1e t:xception of a stone image depicting a footprint of the
Buddha (l:v[arshal l, Ta.r:iln, 224). Marshal) concluded that it was not a Buddhist mon-
ument bt:catL~e of a lack of r·elics and images. However, we would not expect tO
6nd images in association with such an early temple, and if this was a direct-access
relic shrine, the relic alrnost certainly would have been taken when d1e site was
abandoned.
21
Marshall, "f:.xcav~ttions at Taxila," 35- 6.
22 M.arshall, Taxiu1, 226, 333. J. Modi argued that the fire altar would have beeo
on dte platform io tbe main sanctum (Marshall, Ta.:ci.la, 226).
113 C . Rapin, " Hinduism in the Indo-Crcek Area: Notes on Some lndia.n Finds
from Bactria and on Two Temples in Taxila," in In the Land qf the GryplronJ: PapetS
011 Cmtral Asian Archaeology in Anliqui!J•, ed. A. lnvemizzi, Mor10grqfie di M=fJolamio.
(FLorence: Casa Editrice le lettcrc, 1995}, 290.
"' A. C unningham, "Shahdheri or Ta.-..-ila," in Ardweological Suro9 qf 1rulin: Repo'fJ
for lhe .rear 1872- 1873, ed. A. Cunningham (Calcutta: Office of the Superintendent
of Government Prirtting, 1875}, 69 - 72. Eni.ngton has attempted to trace Cunningham~
finds; in an analysis of bis excavation, she has pointed out that Cunn.iogham recorded
uncovering the fo undations of only two pillal'S, whereas his plan indicated siX\
Errington ("The Western Discovery of the A•·t of Gandha.ra," 181 - 3) ha.s sugges~.ed
that the portico is a cor~ecrural reconstruction.
2S Cunningha.m, "Shahdheri or Taxila," 70.
Copyrighted material
......
THE DEVELOPMENT OF RELIC SHRINES: PHASES I AND 1l 69
temple and suggested that the two columns in the cella supported a
canopy for an image tJ1at was, perhaps, placed on the back wall.
This edifice clearly is similar to the ] aJ!Qial C temple; I believe that
it was built to house a direct-access relic, which was displayed in
the rear cella. The plaster sculptures reported by Cunningham were
probably donated later, during phase Ill, as important Buddhist relic
structures were common loc1 for gifts of sculpture (see section 6.3.9
Mohra Moradu: Phase III Architecture and Additive Imagery).
Cunningham's description of the sculpture corresponds to the typi-
cal phase Ill stucco production from Taxila (see fig. 34), with which
he was familiar. It leaves no doubt that Mohra Maliaraii, and prob-
ably by analogy J ru:H;lial C, were Buddhist shrines, not temples ded-
icated to Vishnu or Zoroaster.
The apsidal temple D at Sirkap (fig. 10) is a somewhat later struc-
tw-e that appears to be related to the above direct-access relic shrines.
T he entire temple and the outer square enclosure were fabricated
out of phase II diaper masonry. 26 T wo small bases bracketing its
entrance could have supported columns or small slftpas.V The apsi-
dal D temple had an antechamber <u1d rear apse separated by a
doorway, and in place of the stii.pa is a finished pit 9.75 m in dia-
meter and 5.5 m deep. R unning around tJ1e temple is a passage that
is too narrow for circumambulation. The temple was constructed on
a massive scale like J aD<;Iial. C and Mohra Maliarafi, although its
form can be compared to smaller phase II structures such as the I3
apsidal hall in the Dharrnarajikii. complex (fig. 1: western avenue)
and the A 1 two-celled shrine at Kalawan (fig. 20).28 All of these
Gandharan apsidal halls differ from their lndi<u1 counterparts in that
a doorway separated the entrance foyer from where the relics were
kept. In the case of the Sirkap apsidal temple D, the doon vay sep-
arating the cella and antechamber was 4.25 m wide; a massive hinge
for turning the door back against the wall indicates that the chamber
26 This structure may have been reconstructed on the sire of an earlier monu-
ment; if this is rhe case it was ex-panded, because the walled enclosure sits on the
filled-in, leveled remains of earlier buildings (Marshall, Taxila, 150).
v Kuwayama, "L1 the Time of Late Sirkap and Early Dharmarajika: How Taxila
Introduced Stiipa Architecture," paper presented at the conference "O n the Cusp
of an Era: Art in the Pre-Kushan World," Nov. 8-12, 2000, Kansas City.
28 Only the fou ndations of d1e apsidal halls in Gandhara sul\livc (Sirkap block
Copyrighted material
70
did note that inside the sunken apse, at a depth of 45 cm, "'as a horizontal cow11e
of timber with da.do pa.neling (the wood having been replaced by mud) (.Marshall,
Taxi/a, 15 1- 2). He preserved this by filling it with ashlar Limestone blocks.
limestone blocks. This masomy type is atypic,1.l, and the associated sculpture has
been dated to a wide range of periods on the basis of stylistic evidence (Marshall,
Taxila, 154- 55).
" Ibid., 154.
~ Ma r-shall arg·ued on the basis of ha.i rstyle t.hat one of the heads might depict
the Buddha ([bid., 155, pi. 148, no. 6). Susan Huntington has suggested d1ar one
of the heads (see Ibid., pi. 148, no. 5) may be a represema tion of a bodhisattva,
on the basis of the hairstyle, which is a standa.rd f.e ature of later bodhisattva images
(S. Hurrtington and .J. Huntington, 'n~e Art qf Ancient. b1dia: Buddhist, Hiudu, Jain, 1st
eel. [New York: Weathcr·hill, 1985), 117). Also see L. Nchru, Origi11s qf tl1e Gtmdlwran
S!fle: A Stut!y qf Cantribu/Qry lnjlumces (Delhi: Oxford University Press, 1989), 71;
Kuwayama, " lJl the Time of Late Sirkap and Early Dharmarajika: How TaA-ila
Introduced Stltjm Architecrurc," paper presented at the con1e rence "On the Cusp
of an Era: Art in the Pre-Kus.ban World," Nov. 8- 12, 2000, Kansas City.
S9 In fact, the only othe r images larger than tile-size li·om phases I and li ar:e
The fact that the apsidal temple was rebuilt during phase U, after
the city of Sirkap had already begun to decline, indicates that this
religious sanctuary continued to be important to the people of Tax.ila.
The addition of later sculpture to tllis sacred area is an indication
of ongoing patronage and indicates the importance of the relics that
must have been housed in this apsidal hall.
Another Taxila direct-access stilpa from phase ll can be found m
the sacred area of Kalawan. The phase II main st1ipa A4 (fig. 20~
has a circular relic chamber 4.03 m in diameter, which starts t<l
taper inward about a meter above the floor. The interior walls of
this chamber were fulished ; the multiple layers of whitewash indi~
cate that it was regularly entered and that its appearance was significant
enough to necessitate repainting from time to lime.'10 This was one
of the la~t monwnents MarshalJ excavated in Tax.ila; he stated that
he was "quite certain that in this case the chamber could not have
been filled in." 41 This structure was rep<lired throughout the occu~
pation of the site 42 and served as the main monwnent of the sacred
area (see section 4.1.1 Kalawan: Phase II .Architecture). The unfilled
relic chamber of the Kalawan A4 stupa is crucial to an understand-f
ing of the oversized chamber of the j a.t)<;lial B stlipa and the pres~
ence or a finished pit in the cella of the Sirkap apsidal D hall. Aln
three of these monwnents had la..·ge, fmished relic cha..nbers, and
they seem to have shar ed a corrrmon function: to store relics when
they were not on &~play. 45 One puzzling feature is the absence of
doorways leading into the large relic chambers that are at the core
of each of the monuments. It is my opinion that they were all entered
from above via ladders that could be removed, to protect the relics
from the casual thief or passing ma..·auders.+~ Reliquaries are often
in Mathura; examples are the Bala bodhisat.tva or the non-Buddhist Kushan d)rnas~
tic portrai ts from Mat.
"' Marsball, Ta.xila, 323.
"' Marshal], "E.~ploracion at Taxila 1930- 34," 160.
' 2 Mm·shall, Ta.xila, 323-4.
+s 'll•e circa 5th centUiy C.E. main stupa at Mohe•~o Daro had a hollow relic
chamber about 6 m in diameter; painted plaster 6-ag:ments indicate that it 'vas plas-
tered and painted on the interior U. Mm·shall, Mohtrrjo-Daro and lhe buirJJ Cioili.J:pti.fJII.
[London: A Probsthain, 193 1), l I 5). MarshalJ's repo•t questioned the work of his
assistants Banerji and Wartekar, but in his later publication on the Kalawan remains
he reconsidered his skepticism about the finishing of the Mohenjo D aro relic cham-
ber (M.arshaU, Taxila., 324).
44
Marshall (" Exploration at Tax:ila I 930- 3<l," I 72) suggested that the treasuries
in monaste1~cs were scaled cells that were accessed from above.
found stacked along the central access of solid stupas, some of which
were reopened on occasion. T his pattern of relic interment indicates
that regular stilpas were opened, to add or divide relics, by digging
down from above along the vertical axis of the monument. 45
While the phase I and II temples in and around the city of Sirkap
and the openable stiipas fi·om Ja,r:tqial B and Kalawan form an eclec-
tic group, a coherent relic-shrin e type does begin to emerge at non-
urban Buddhist centers near the beginning of phase II. T he
"two-celled" shrine characteristically had a front antechan1ber, some-
times an open porch, that provided access to a restricted cella where
the stiipa or relic was housed; good examples in clude the stiipa shrines
along the edge of Butka.ra Ill (figs. I 00, 102) and the G5- 6 shrine
in the Dharmarajika complex (fig. 1). The codification of the two-
celled shrine occurred in conjunction with its popularization at many
sites in Taxila , Swat, and, to a lesser extent, the Peshawar basin.
While it is difficult to know what the upper superstructure of these
shrines would have looked like, they do appear in the relief sculp-
ture, sometimes as closed structures (fig. 119), sometimes . .vith relics
on display (fig. 118), and there is a relief from Butkara I that shows
an internal stupa on display (fig. 99). By phase ITI, the construction
of new two-celled shrines had dramatically declined; eventually they
died out.
The two-celled relic shrine appears in part to have grown out of
the Gandharan apsidal halls like the D harmarajika 13 shrine and in
part from temples like Mohra Ma!iarai'i or J ai).qial C (fig. 1, west
side; figs. 13, 14). In both of these groups, the rear cella could be
secured behind doors, in contrast to rock-cut ca.iiJ'a halls, like Bhaj a
•~ Stacked relic deposits were found in the Mar:Ukyala and Devnimori main stfi-
pos (R. M ehta and S. Chowdbary, Ercavation at Dll(lni,wri; a Report qf the F..xcavatiorl
C.onductcd .from 1960 to 1963 (M.S., University of Baroda, 1966). ]<or further infor-
mation on the Mal)iJ..-yaJa stUpa, see Cunni.ngham, " Mani.b:yala-1863," 152- 72;
Cwmingharn, " Manikyala-1872," 75-9; S. Oar, "Excavations at Manikyala- 1968,"
Pakistan Archaeology 7 (1970--71); J Fergusson, J. Burgess, and R. Spiers, History qf
Indian wul Eastern Archi18clure (London: J. Munay, 1910); Stein, ArthtuOlogir.t~l Reco>uwissrmu.s
·ill North-Wu tem lrulia cmd South-Eastern lran.
46 Kalawan stupo. sbri.nes AI , A l3, and Al4. Ao extra <ultechambcr was added
late r to the fr-ont of the Kalawan Al4 structure (Marshall, Ta.:rila, 325, pl. 72). The
Akhau ri B monastery has two two-ceUed stftpa slu·ines io the monastic enclosu r·cs,
but this is the only i.nstance of tllis shrine rype appearing "~ chin a monastery that
I arn awm·e of (Marshal!, Taxi/a, pi. 67b). At the Dha nnariijikii complex, lwfiiur frng.
ments of a sttipa were found in the early phase U G4 shrine (.Marshall, Taxiill, 255,
pi. ·~5). Tbere is a stilpa in tl1e .B8 Bhamala two-celled shr·ine (Mar:shall, Taxila., 39~
pi. 1 14). The Bhamala two-celled shrines are slightly difthem; the antechamber is
open on the sides, thus functioning rnore like a porch than a preceding room. A.t
Butkara Ifi, five two-cellecl sttipa sluines were excavated along the edges of the
sacred a rea, wh ich had been cut into tJ1e hil.l;;ide to form cave-Like cha mbel'{l
(Ra.hrnan, " Butkara ID."). At Matja.nai there is a two-celled stiifm shrine (sliipa 31
and the bases of three other str1pas were enclosed in si mple chambers (see fig. IOtf)
(Khan, " Prcli.mi.nary Report of Matjanai," 33).
., T he Dharrnar;ij:ikli two·ceUed shrines incl ude the early H structure; shrineS
GS- 6, L, M2- 2, and probably others originally had the two-celled configuratio~
including C l- 2, 12, and possibly MS and M9 (MarshalJ, Taxi/a, pl. 45). At Lalchal\',
Marshall assumed that tbe shrine to the left of the main stilpa was an image slvin~
but it follows IJ1e two-celled pattem, and, in any case, no images were found. 011
the basis of its shape, l believe the foundation to u1e right was p r·obably a two.
ceUed relic slui.ne (Ma.rshaU, Taxila, 389). The two-celled shrine at Moll.ff; Moriidtl
is not included in Marshall's plan of the site; it is just sou tJ1 of the entrance to tJte
main Jli1Jx• (field observation, 1993- 94}. T he A2 and A5 slu·ines at Kalawan follow
the two-celled format (MarshalJ, Ta.xila, pi. 72). Takht-i-bahi Tl, T2, T3(?}, and
T 4 m·e rwo-celled relic shrines (field observations, 1993- 94). At J amii.l Garhr, immer
diately behind the ma.in stfipa is a sh rine that basically follows the t1vo-celled par,.
tem, altJ1ough it has windows-a unique feature (field observations, 1993-94). The
Butkara I Great Building clearly follows the two-cdled format (Fa.ccenna, .Butkara
I, vol. 3, no 3, pi. A.'Vlii).
Copyrighted materi
•
THE DEVELOPllilEl\"1' OF REUC SHRlNES: PHASES I AND 11 75
shrines.48 It was here that d1e famous silver scroll inscription of the
year 136 was found, together with some relics in ihe G5- 6 two-
celled shrine (fig. I , just nord1 of the western avenue).19 Tllis inscrip-
oon says:
In the year 136 of Azes, on the 15th day of the month of Asha<;lha,
on this day relics of the Holy One [Buddha) were enshrined by Urasaka,
scion of J.riltavhi.ra, a Bactrian, resident of the town of Naocha. By
him these relics of the H oly One were enshrined in his own bod-
hisattva chapel at the Dharmadi.jika stripa at Takshasila, lor the bestowal
of health upon the great king, king of kings, the Son of H eaven, the
Kustmn; in honour of all Buddhas; in honour of the individual Buddhas;
in honour of the Arhats; in honour of all sentient beings; in honour
of (his) parents; i11 honour of (his) fri ends, advisors, kinsmen, and
blood-relations; for the bestowa.l of hea.lth upon himself May this thy
right munificent gift lead to Nirvan a .50
43 Relics were found i.n the C5- 6 shrine and in a.~sociation with the SlO sluine.
Tlus latter structure could have been ust:d either as a stilpn shrine or as a direct-
access relic shrine; two relic caskets wer·e found near the base of the west wall
beneath a block of kaiijur. Fu rther, the basic layout of this fi·agrnentary shrine gen-
erally fo llows the two-celled format, though later reconstruction makes tl1is conclu-
sion rermttive (M arshall, Ta.\'ila, 26 7).
49
Relics were found (apparently purposely buried) in a casket in the G5- 6 rwo-
cellccl shrine nem· the back wall of the a lia. This casket W 'd S abo\rt 30 ern below
tl1e floor <utd contained bon e fragments, along with the silver scroll inscription of
the year 136 (Ibid., 256-7 1).
)() [bid., 256.
Copyrighted material
76 CHAPTER THREE
SI A standing in situ schist sculpmre of the .Buddha was found in Annex Roo1t1;
ix at Mckhas<u1da (see ftgs. 65, 67), which Mizuno, the excavator, identified as a
kitchen. lt seems unlikely that an image of the Buddha would have been placed
in a kitchen. T he ((moat of this chamber is simila r to rl1e assembly rooms pre-
sent a.t other sites; an assembl)' room seems a more likely location for a devotional
icon. See S. Mizu.oo, Meld1asanda: .Buddlrist Monastery in Pokisllm Suruc;red in 1962- 61
(Kyoto), 87.
Copyrighted mater 31
CHAPTER FOUR
In phase ll, the patronage of new and old Buddhist sites grew dra-
matically. T his is the period in which the majority of the extant sites
in Taxila, the Peshawar basin, and Swat were founded (fig. 4). In
Taxila these include: Kalawan (fig. 20), J auliafi (fig. 27), Mohra
Moradu (fig. 32), Pippala (fig. 39), Kunala (fig. 41 ), Giri, Akhaurf
A, B, and C (figs. 21, 22, 24), Khader Moltra D l and D2 (figs. 25,
26), the apsidal temple D at Sirkap (fig. 10), and the maj or addi-
tions to the Dharma•·ajika complex (fig. 1). 1 Although it is harder to
date the Peshawar basin sites, it seems that many important centers
were established in phase Il, including Ranigat (fig. 78), Thareli (figs.
68, 69), and Mekhasanda (fig. 64). While Takht-i-bahi (fig. 2), J amal
Garhi (fig. 6 1), and Shah-jfkf-<;lherr (fig. 83) may all have phase l
foundations , they certainly were active by phase ll; the extant remains,
however are p hase m and IV. 2 In Swat, Butkara I (fig. 97) remained
important, and the sites of Saidu (fig. l 04), Panr (fig. l 03), Butkara
Ill (fig. l 00), M arjanai (fig. l 06), Damkot, Andandheri, and prob-
ably Snaisha were all established.3
' The Taxila si tes can be placed in phase 11 on the basis of d iaper masonry and
numismatic evidence. For a table of coins found, see l?.rrington , "N umismatic
.Evidence for Dating the Buddhist Remains of Gandhara," 212.
1 At Takht-i-bahf a siu.gle T.ndo-Greek coin of ApoUodotus in the fill of the phase
]Jl underground chambers seems t.o have been a stray llnd (Ibid., 194; Ha.rgreaves,
"Excav-ations at Ta.kht-i-bahi," 33-4). An inscription mentioning the I03rd year of
Gondophares is said 1.0 have come from Takht-i-bahi, but it may not be from this
site a nd t11us cannot be used (Errington, "The Western Discovery of the Art of
Gandhara," 115). For coin finds from the various siteS, see £rringtOn, "Numisrn:Jt]c
Evidence for Dating the Buddhist Remains of Gandhara," 213. Kuwayama elated
the early round stiipa. at the core of the Sbiih-ji-kf-Qherf slllpa to roughly the same
time as the Dha rmarajika core stupa- e.g., sometime in phase I or possibly as late
as the beginning of phase II (Kuwayama., The Main Stupa at SM.h·jf-ki-ljlzen~, 23- 26).
Errington has concl tL~ivcly shown that the "Kani~ka reliquary" dates to later than
this monarch's reign, although probabl)' still withi.n our phase Il (Errington,
"Numismatic Evidence lor Dating the 'Kani.shka' Reliquary from Shah-ji-ki-qheri,"
in South Asian Archarolog, 19.99, eel. E. Raven (Leiden: in press).
' f or coin finds from the various siteS, see Errington, "Numismatic E'~dcnce for
Dating the Buclcllust Remains of Gaodhara," 211. For discussions of the elating of
Copyrighted material
'
78 CHAP'I'ER FOUR
Snaisha, see Rahman, "Shnaisha Cumbat: Fi.rst Prelimi nary Excavation Rt:port,,.
16; M. Taddei, "Some Remarks on the Preliminary Repo rts Published on the
Snaisha Excava.tions, Swat," East a11d West 4-8, no. 1- 2 (1998). Some sculpture cer·
tainly can be attributed to phase II , but the major Sll1.1Ct:ures in their present form
date to phase m, with phase IV addi6ons.
Copyrighted mater
THE PHASE 11 SACRED AREA 79
Copyrighted material
80 CHAPTER FOUR
Two more complexes were constructed in phase IV: the poorly doc-
umented Ballar tope in the north and Bhamala to the east. Because
the Dharmarajika drew the major share of available pao·onage at
this time, significant changes were made to the phase I core of the
site; in turn, the phase II structures were built over, repaired, an4
modified in pha~es ill and IV.
When a new phase II Buddhi~t site was created, typically a medium~
sized (7.5 by 9 rn) main stupa was constructed, together with an adja--
cent monastery; examples are the Taxila sites of Jauliiiii or M oh:ra
Moradu (figs. 27, 32).s These two architectural components are con..
sistently found together; public sacred areas were never establishe4
with.out housing for a monastic community. T he reverse is also trud
we never find monasteries that are not connected to public places
of worship. Even though some monasteries are so insignificant or
remote that it seems unlikely that their sacred areas were used by
the lay public, they all had a main stupa. Examples are the minor
sites of Akhauri A <md C, near the Dharmarajika, which have sacred!
areas almost devoid of donative structures. Only a single small stiipa
was added to the sacred area of Akhaun C (probably built togethe:t
with the main sl71pa) (fig. 24}, while Akhauri A appears to have had
no donated structur es at all (fig. 21).
In addition to the public sacred area organized around the main
stiipa, special devotional structures were invariably placed >vithin the
monastic enclosures, apparently for the use of monks in a restricted
and undisturbed setting. During phases I and ll, these took the form
of monastic stiipa. shrines;6 in phase Ill, they were image shrines (see
section 6.6 Phase ill Use of Images and Relics in Quadrangular
Monasteries). Thus, the monks would have used for devotion the
large main stupa adjacent to their monastery, but also other places
that were not readily accessed by the public.
) At both Jauliai'i and Mohra Moriidu, the monastery and main stupa. were ini;.
tially constructed with phase Il late diaper masonry and were later repaired io
phase m with semi-ashlar masonry (Marshal!, T axila, 358, 69). Like the pilasteQl
on the Moh:rfl Moradu stupa, the kaiijur pilasters on the Jauliai\ main stiipa are inte-
grated rather than mot\olithic, and the diaper masonry technically approaches the
semi-ashlar format (field observations, 1993-94). See also Marshall, Emwalions lit
Ttaila: The Sb.pas and Monasteries at ]aulian, 4. All of these features suggest that Jaulian
and Moll:rfl Moradu were founded near the end of phase U.
6 The best known example is the st1ipa placed in cell 9 at Moh.ra Moradu, but
there arc many other inst<mces, including the sWpa in cell 3 1 at Pippala, the sllif;i/1.
at Kalawan in monastery F cell 12, and the stupas in cells El and E2 and in the
courtyards of monasteries G and M at the Dharmarajika.
1 During phase n, a second porch and stairway were added to the front of this
otherwise fairly typical two-ceDed stii.pa shrine. lo phase ill, two small image shrines
were added to its south wall. At the time of excavation, the west wall of this shrine
'vas found in a good state of preservation and thus gives us an idea of phase lJ
pilaster decoration. Although Marshall did not photograph it, he did note IJ1at the
plinth was decorated with "Corinthian pilasters standing on a molded base wilh a
dental cornice above and nor.ched Hindu brackets inserted beneath the architrave"
(MarshaU, "Exploration at Taxila 1930- 34," l 60- 1).
Copyrighted material
82 CHAPTER FOUR
functioned together. 8 Over the course of phase II, these three cen·
tral monuments were enclosed within a court bounded by multiple
relic shrines to the north, east, and west, with the original monastery
(C) to the south (monastery F seems to have been added a bit late11
in phase U).
To the north at K.al.awan stand two empty phase Ill two-celled
shrines, which rest on earlier phase U diaper remains. 9 It is cleat
that during phase II a structure (or several structures) with a 13.1 m
square base was fabricated here in cor~junction witl1 the other shrin~
flanking the east and west of the sacred area. Remains of founda~
tions indicate that this was rebuilt again in phase II and replaced
by the pa.ir of two-ceUed shrines in phase lii. Clearly, the phase I]
northern shrine(s) must have been vital to the site or there wot1ld
not be evidence of reconstruction on three occasions. Further, it
seems likely that this succession of structures to the north probably
fun ctioned in co•~jllllction with the openab1e direct-access stupa A4
that they face. 10 Although we know nothing of the phase Il build·
ing(s), the pair of phase III shrines exacdy COl-responds to the two-
celled shrine format, and it is likely that they were used to display
relics (see Chapter 3).
Large pans of shl"ines A3l - 34·, west of the sacred area, have fal.len
down the hillside, 11 but on the eastern side the large two-ce.l.led st11pa
shrines A l and Al3 were found relatively intact. Next to the reli~
quary, in shrine A 1, a long do native inscription on a copper plate
was found. T his inscription dates the diaper masorny fabrication of
this shrine to the 134tl1 year of Azes (according to MarshaJJ, 86
C.E.; 12 see Appendix A).
The copper plate inscription, found in the donated Al two-celled
stuine A33 is the mirror image of the two-cellcd stripa shrine A I and 01ight have
contained interesting relics (Marshall, "Exploration a t Ta.xiJa 1930- 34·," l 65).
12 Ibid., 162- 3. Tilis dating is based on Ma.rshall's assumption that the V iJmuna
era (of 5B B.C.E.) is tbe same as that of A:4es (:Nfarshall, Taxikl., 53, 257).
Copyrighted materi, tl
• •
nm PHASE 11 SACRED AAEA 83
13
M:arshall, "Exploration at Taxila 1930- 34," 163.
14 Both have been used to date the rnasOnJ)' and structures with which they are
associated (see Chapter T hree).
15 Zwall; Candlurra Sr:ulpture, 30.
Copyrighted material
84 CHAPTER FOUR
Copyrighted mater 11
l
THE PHASE U SACRED AREA 85
monks and lay patrons, the planning of the site must have been a
collective process, involving the lay and monastic communities, occur-
ring through time. The needs <md goals of one generation may have
been quite different fi·om those of the next.
At the end of phase I, the Dharmari:'ijikii complex was made up
of a main stzlpa surrounded by a unified set of stupas or columns.
The main stupa was approached primarily from the west, as indicated
by several large stupas placed to form the beginnings of a processional
avenue, which was greatly expanded during phase li (fig. l; note
the phase I and phase II structures). Rows of monastic cells bounded
the west and north sides of this massive sacred area, and a wall is
visible to the south. To the northeast, two monasteries had been
built (courts A and B), both with large kaiyzlr ashlar stupas at their
cores; these stii.pa courts appear to have been used by a restricted
monastic community (fig. I 6). Lay followers coming from Sirkap
would have entered the site from the northwest (fig. 4), where the
massive two-celled relic shrine H stood; the cella exhibits phase I
masonry.
During phase II, the Dharmarajika complex was ve1y active, as
indicated by the extensive remains surviving from this period. T he
pradalq£1_1.apatha of the main stiipa was augmented witl1 at least 19
shrines, 16 which probably housed relics. These shrines were placed
to emphasize the axial points of access into the prada~1_1.apatha already
marked by gates in phase I. The western gateway was indisputably
the most important entrance into the main stupa's circumambulatory
path. During phase II, the western avenue becan1e fully aJ·ticulated
with the addition of a group of large two-celled shrines 17 and an
apsidal hall, which together augment the earlier pha~e I stii.pas. These
stfipas and shrines fran1e and emphasize the processional avenue in
a way that is analogou$ to the relic shrines that bounded the Kalawan
sacred area (fig. 20). The north gate must also have been important,
judging from the cluster of shrines (poorly preserved) along the pra-
M.k,rirla/Jatha and the large stzlpa Pl 18 opposite this entrance. In general,
16
Much of the phase (U construction in the northern quadrants sits on a com-
plex group of phase [[ structures, making ident:ification difficult.
" The L, G5-6, and M3--2 structures all follow the classic two-celled format
(Marshall, Ta.xila, 25 1- 9, pl. 45).
' 8 It seems likely that this strucr:ure was a relic shrine rather that a •·lfipa (only
the base survives). It can be compared to the Swat core structures at the site of
Nimogram or to the large base at Tokar Dara (figs. l 09, 11 3).
Copyrighted material
86 CHAPTER FOUR
19
Th!' !'arlit'st ulla of the H Slrucrure dates to phase I; this shrine w-d.'l gr~atly
expanded in phase [(. The inner chamber of the H SU'UCture was sealed sometim e
in phase Ill with semi-ashlar masonry; by this time the phase 0 additions were
probably in ruins. Although Man.hall postulated that this addition consisted of a
plinth to support a monumcntnl pc~riJ1irrci1w image, he found no traces of an image,
and ihc plHI>C lll addition is without deco•-ation. Monumental pan'niroli!W images
and assuciatcd structures found at sites like Tapa Sardar [n Afghan istan are quite
different , RO Mm\,hall 's suggcs1ion seems unlikely. See Marshall, 7ilXi/.a, 21·7 8;
Taddei, ·'Tapa Sardar: First Preliminary .Repo n "; Taddei and Vcrardi, "Tapa
Sard1ir: Second PrcliminaJ)' Rc·po rt." The phase I, IT, and lii additions nnd modi-
fica tion~ indica te 1hat tJ1is cdillre was impo rtanl enough to be renovated throughout'
the life of the Ohru·m:uiljikll compl ex.
THE PHASE ll SACRED AREA 87
Copyrighted material
88 CHAPTER FOUR
23 An intact reliquary was recovered about l.S m below the g•·ound-level of the
phase I R4 stfipa (Marshal!, Taxi/a, 253). The somewhat unusual placement of this
relic implies d1'n it may not have been the only one enshrined in this st1lpa. Mulriplc
reliquaries were commonl}' interred along d1e a.xis of d1e stupa, one above d1e other.
The relic finds from the Mal)ikyala, Bin1aran, and Devnimori stftpas are good exam•
pies of this practice (.Errington, 'vfbe Western Discovery of the Art of Gandhara";
Mehta and Chowdhary, ExClWation al Deuninwri). Therefore, the reliquaJy found below
ground kvcl may indicate that it is just one of a group of relics int<.'tTed within
this stupa.
24 "Dle adjacent D3 stupa or pillar, like the R 4, was encased in a shrine during
phase IJ. Because of the poor state of preservation of the 0 3 enclosure, nothin~
conclusive can be said of it as a whole, but its location and the association "~th
an added st.fijJa base Oike the R comple..-x base) suggest that it w·dS related in use tO
the R direct relic complex.
2$ Marshal! argued that after IUs suggested earthquake of circa 30 C.E. a mb-
ble-cored kn;rjur stiipa base (of the Sirkap type) was built as a repair of the R4 stiipa
(Marshal!, Taxila, 253). This possible encasement ex~~ts now only as a base, Oil.
which the R3 slui.ne sits exacdy (and d1us it is not appar·ent on the plan). M.arshall'~
expl<IJlation of the base being a post-30 C .E. repair of the R1· stiipa appears faulty
/or several reasons. .First, it bisects the intact R.4 stupa. exacdy; a repair would have
involved surrounding and encasing d1e pre-existing stupa. Second, a rubble wall (a
masonty type used mainly before Marshall's suggested earthquake) reinforces the
south wall or the base, indicating that it is a pbase I su·ucmre (field observation,
A4 enterable stupa with its oversized relic chamber (see section 4.1.1
Kalawan: Phase Il Architecture). During phase II, a shrine R 1 with
two platforms for relic display was built. T he two platforms, deco-
rated with life events and jatakas, 26 were attached to the west edge
of the base; thu~, for a worshipper using the R 1 sh•;ne, they would
appear as extensions of the stitpa-like baseY
Before entering the prada/q-ir_wpalha of the main stilpa, a devotee
would have used only the R I section of this shrine, where the relics
would have been displayed on the two low platforms. Such thrones,
or plinths, are often seen in sculptural depictions of relics in Gandhara
(fig. 118). Once inside the prada/r.ril.lapalha of the main stii.pa, the devo-
tee would have encountered the R5 stiipa shrine, which abuts the
back of the R3 relic chamber (this shrine survives only in .its late
phase m fi)rm).28
In general, the outlying sacred area of the Dharmarajika is dom-
inated by two-celled relic shrines built to channel the worshippers
along a prescribed path, although the distribution of a few of these
slu-ines does not follow a tightly organized plan. While most of the
additions to the sacred area relate to the western avenue, stii.pa Pl
was fabricated at the main stupa's northern gateway; mirroring this
addition, the large D4 stiljJa just out~ide the south gate was given a
new encasement. Thus, whi.le it is clear that a devotee was meant
to enter the site from the north near the H structure and to approach
the main stiipa using the western processional avenue, the north and
1993- 94}. Third, the style of the base is identical to that of early siUpa bases from
Sirkap. Therc!ore, it seems likely that the base was added, built against the R4
sttipa during phase I, and d1at it~ main function W"'dS to support a relic chamber
(R3), rather than being a repair of the R4 slilpa.
It would seem that the R4 sll1pa was an important element of the R3 relic dum-
be,·, which ;i ts exactly on the edge of the lower base; thus, it is tbe extant R4 stiipa
that constituted the east wall of the R3 relic chamber; that is, tl1e eastern part of
the R3 chamber was the dnm1 of the R4 stitpa. l infer lium this evidence that tl1c
lowe r base was built to support the R 3 shrine not long after tJ1e constJ·ucrion of
the R 4 stupa.
26 The identification of these reliefs is problematic and is complicated by the fact
taU columns like those found at Butkara I. If this was the case, then even during
phase ([ these columns would have been an impressive presence, toweriog over rhe
ci•·cumambularory path.
n Among the many references, see iM arshall 's discussions of the phase I struc-
tures at the site of Jru:u:jiiil and his analysis of' many phase II shrines a.[ Kalawan
(Marshal!, 7lL\ikl, 222- 30, 322- 4 1).
33 The single-celled Thareli shtines include 03, 0 4, 05, and 06, the last of
which contained the base of a small stupa; a range of unrelated srone and stucco
sculpum:s were found in and a round these stntctures (Mi zuno and Higuchi, eds.,
Thareli, 153- 4). For a full discussion of these shrines, see Behrendt, " Relic Shrines
of Ganclhara: A Rcinterpt·etation of the Archaeological Evidence."
,. Marshall, Taxi/a, 249- 50.
)$ These finds include a schist tlie-Le with acanthus leaves and Lhe heads of a
male ~md a female from the 0 5 shrine, a female head f.r orn the 0 8 shrine, and a
f-ragment from the 012 shrine of a narrative panel depicting the cremation of the
Buddha (Ibid. , 249).
The shrines at Thareli also have large fioor areas and low roof&,
and they also have the trabeated dome structure characteristic or
Peshawar and Swat architecture (figs. 69, 70).36 On the basis of the
characteristic C-shaped plan, the Dharmarajika shrines (fig. l , shrines
D5, DS, D6, D9?, D l2, Dl4, B9, and B lO)s 7 can be compa1·ed with
shrines at Kalawan (fig. 20, shrines Al9, A20, and A26),38 Jauliaij
(fig. 27, shrines C 19, B 17),39 Takht-i-bahr (fig. 2, shrines T I, T~
T3, and T4), and Thareli (fig. 69, shrines D 3, D4, D5 and D6), ancl
in Afghanistan to shrines at sites like Tapa-i-kafariha (fig. 117, shrines
20, 22).40
Another factor that hints at the relic association of the Dharmariijiki
prada~?zapmha shrines is that they enclosed the main stiipa. At botb
Ka.lawan and Butkara m, two-celled relic shrines define the perimet
ter of the sacred areas (figs. 20, lOO). Although we do not know the
exact function of the Dharmarajika prada~~wpatlza shrines, the fact
that they were reused, modified, and restored during phase ID4 1 indio~.
cates that they initially housed relics or small stupas.
In his efforts to show that the pradak.ri~wpatlta shrines contained
images, Marshall compared them to the open-faced phase ill imagt~
shrines encircling the main stupa at J amal Garhf (fig. 61).12 Tlris
incorrect suppositionH has led to a general assumption that many o~
the image shrines iJ1 the Peshawar basin could elate to as early as
phase II. T his would mean that image shrines containing large devo"'
tional icons of Buddhas and bocUusattvas were being produced du.r-
B L7 (Ibid., 377, 79) may have been used for tbe display of relics.
.o Barthoux, Les Foui/ks de Hadda., plan A.
•• Shrines like B 13 in d1c northeast were rebuilt in middle and late phase lli
masonl)', retaining pbase n layout. In other instances, large phase lli image shrines
like B4 completely replaced the earlier sr.ructun:.
41 Marsha!J, Taxi/a, 248.
43 The size and s.hape of the J amal Ga:rhJ image shrines arc completely different
More important, the n:tai.n stupa (ru1d s1.uwundi.ng COttrt of image shrines) at Jamlil
Garhr sits on a high platform built over earlier structures. This plat1orm is only
about one-third the diameter of the Dharrnarajikli main siUpa com·t. The ooJy sim-
ilarities between the rwo courrs are that lhe sWpas are round and d1M bod1 are
e11circled with subsidia.r)' S[I"Uctures, a d ra.ractcril."tic found in one fom1 or another
at most Peshawar basin sites.
ing the time of Kani~ka I, but that dating is not supported by the
archaeological record at other sites in Taxila, nor by other significant
data from Swat or the Peshawar basin (see section 6.3.1 The D harma-
rajika Complex: Possible Late Phase II Image Shrines).
.., £ . £rrington, "In Search of Palusha, a City of the Cenu·aJ Gandhara Plain,"
Bulletin of lhe Asia fnstitule 7 ( 1993).
Like the adjacent site of J auliai'i (fig. 27), the main stupa and monasteo/
at Mo~·a Moradu (figs. 32, 33) appear to have been established ne~
the end of phase Il.46 Although the pha.se ill sculptural embellisli-
•s lt seems very Likely that the ftve J UJmar sites were satellites of a major center
located near the present town. It is not coincidental that .Bhaja, Bedsa, and Kaili
are so closely grouped, and the same could be said of the ca l"ly Buddhist cemers
around Gi.rnar in Gl~arat.
46 Both main stiipa and monastery were built in diaper ma.~onry; repairs in sen:tl-
ashlar suggest that they were initially fabrica ted near the time of LTansicion in build-
ing types (Marshal!, Ta.xila, 358, 62}.
., Ibid., 365.
96 CHAPTER fOUR
sme that this w-as the main stitpa of the site because of the presence
of other smaU stiipas. Stupa K, on the other hand, stood by itself out-
side the mona~tic enclosure. This configuration with the main stupa
placed within the monastery is unusual; the only other phase lti
example is found in T axila at Kbader Mohra D2 (fig. 26). If th:e
main stilpa was accessible only to the monastic population, then tire
lay community wouJd have been unlikely to make donations to th:'e
site. In fact, when this Pippala monastery was reconstructed in phase
Ill , the builders went to great lengths to obliterate the phase n
monastic enclosure, leveling the waUs and ftlling the old monastery,
so that this stupa could be a public one. It is not coincidental thctt
Khader Mohra D 2 shows no signs of expansion or of occupatiotil
after its phase IT.
---··-
40
Cole, Memorandum on Anc-ient Monuments in EuS11f;:a~ 2.
49 Trench TI is not direcll>' associated wirJ1 the main stii.f){J; rlHJS, rJ1cse coins at
be;1 give us an idea about when use of the site began. See Errington, "Numismatic
Evidence for Dating the.: Buddhist Remains of Gandhara," 194; Nishikawa et al.,
Gandllara 2, 52, J09.
.10 The actual spot where the eight Vima Kadphlses coins were fotmd is unclear.
The excavation report says, 'The base of the 90 cm-side green schist is placed on
the plastered surface of the plinth [of the core sttipa]. Also over the torus ;trch-
shaped green schist covers the clamps, wi1J1 eight copper coins benea1J1'' (Nishikawa
et al., Gandhara 2, 85). Later in this same report it is stated that the Vima Kadphiscs
coins were found in the core stiipa {Ibid., 89). En·ington accepted this evidence to
date dlis sliifia to 110- 120 C.E. (ErringtOn, "Numismatic Evidence for Dating the
98 CHAPTER FOUR
the site was founded before or during his reign (circa 110- 120 C.E.).
At the beginning of phase Ill the main stii.pa was encased; 12 smalJ
stftpas in tlus sacred area were partially engulfed in the fabric of this
addition (stttpas 105, 106, 108, 109, 115, 116, 124, 129, 132, 133,
134, and 135; stti.p{IS 102 and 122 probably also date to this time)
(fig. 80). As these small stiip{IS predate the encasement, we can plaoe
them in phase II, the addition to the main stujJa being broadly datcil
to the beginning of phase Ill, again on numismatic grounds. At till:
base of the stairs leading up to the main stzifJa, severaJ paveme11t
scones were found in situ that had holes bored in them for the donar
cion of coins. Of the more than 180 holes, 14 contained in situ coin:s
(fig. 81 )Y T his coin evidence and the recovery of an inscription mell'>-
tioning the Kushan king Vasudeva indicate that the encasemeJl.t
occurred around 200 C.E., or near the beginning of phase l11.~2
The excavation of the northeast corner of the phase m encasement
of the main stii.pa exposed one-fourth of the early core stii.pa. Becausf!
it was pr·otected by the encasement, t11e core st!tpa was fotmd intact
(fig. 79). Its square base is roughly comparable to the phase I court:
A and B stlip{IS in the D hannarajika monasteries or t11e early Pl
main stzi.pa in the lower court at Takht-i-bahi (figs. 16, 2, 46). Perhaps
mor·e sigrllficant is the fact that the high circular drum of the R arugat
core stii{Ja is very similar to the phase I form of the Butkara I main
stiipa (fig. 98). The link to early Swat material is also evident when
comparing the small, presumably phase Il stup{IS from Ranigat (fig,
80), which are quite sinlllar to those found in stiipa shrines at Pippal~
(fig. 40), Butkara lli (fig. 102), and Mmjanai in Swat (fig. 107).
Another indicator of an eal"ly dare of the R anigat eastem sacred
area, one that might even suggest phase I, is the presence of sev-
eral set~ of freestanding vedika raiJings (I 31 vedikii pieces were recov-
ered). These fences appear to have stood on the edges of the square
bases surrounding the drums of some of these smal.l stilpas (fig. 82)
(see section 2.3.1 The Phase I Stiipa and Appendi.'X D). When taken
together, all of the architectural and numismatic evidence demon-
strates that the core stiipa and surrounding smaiJ stiipas at R anigat
are among the earliest finds from the Peshawar basin.
$' The subsequent expansion of the main sl!if;a occul'l'ed in five stages (GSt4/ 1- 5).
In relic deposirs placed along the pe1irn eter of' the GSt4/ I encasemenr, 107 coin s
were fow1d, most being of Azes ll and Kttiula Kadpltises, and a few of Hmri.~ka
{Facccnna, Bulkara 1, vol. 3, no. I, 115- 16, 169). lf one assumes that Ka.n i ~ka l's
era started in 120 C.E., then Huvi~ka would have reigned fi·om 14·6 to 184 C .E.;
thus, the GSt4/l period must date to the end of the 2nd century C.E., putting this
stage of consu-uction in phase !I. GSt4/ 2 was constructed approximately in phase
ill; the enca.sement GSt4/3, '" hich can be tied to an earthquake or some other
natural disaster, probably dates to the second half of the 4th centtuy C. E. (faccenna,
Bulkm·tJ. I, vol. 3, no. 3, 109 fl; 635; Faccenna, Gob!, and KJ1an, " A Report on the
Recent Discovery of' a Deposit of Coins u1 the Sacred Area of Bur.kara I (Swat,
Pak-ista n),'' 113). The date for GSt4/ 3 is based on the thin evidence of a single
Kidarite coin (middle of phase ID). However, the date of this earthquake or pe1iod
of restora tion can be determined becatL~c coins of Huv~ka , Vasudeva, and, notably,
Kav-dd were found in associa tion with niches Ql, Q3, and Q!0- 12, but not in the
origil1al reijquaries placed behind these niches (Faccenna, Butko.ra I, 1 13- 14). GSt/4
and GSt/5 can be dated only roughly, in that they p•·ecede a general collapse of
the monument, apparently in the 7th cernury C.E. Thus, these five stages would
sp<m, roughly, the end of ph ase m through phase IV. [n gcne.-al, less evidence is
available for the dating of Butkara I, so its chronology is less secure than that of
the Dharrnarllji.ll<, . However the data from But.kam l, togethe•· with evidence fi'om
Saidu and Pa nr, provide a good foundation for dete rmining the chronoiOf,')' of the
monuments of Swat.
100 CHAPTE R FOUR
columns were absorbed into the main pradak!iJ.zapa~,ha (figs. 97, 98).54
Only at this stage did the main stilpris circumambulatory pad1 become
endosed. Tllis integration of the SLu·rounding structures is a feature
that Butkara I shares with the contempora.ty Dharmarajika stupa in
Taxila (fig. L).
One of the largest two-celled shrines in all of Greater Gandhara
was constructed near the end of phase I,;,; and it is referred to iD.
the excavation reports as the Great Building (G B) (fig. 97, northern
edge of site; fig. 98). Trus dating is solidly supported by comparisons
to the many two-celled shrines elating to phases I and 11 from Taxila
(see section 3.3 Two-Celled Stupa Shrines and Direct-Access Shrines:
Phases I and 11). The massive Butkara I Great Building, with its
enclosing circumambulatory path, is quite sinlllar to the Dharmariijikft
H structure or the earlier Mohra Maliariiii two-celled temple (fig. L3).
Like d1ese other large relic shrines, the Great Building sat on a raised
plinth, and until the end of phase Ill, access was facilitated by a
paved path from the north gate of the main stiipa. T races of ancie11t
chisel marks along the l'as;ade base of the Great Building indicate
that sculptw·e was placed there at one time, as well as along the
sides of the steps.SG The walls of the Great Building were pla.~tered,
and there is evidence of successive modification and restoration, incli-
cations that tllis monument was used for a considerable time.57
Additional relic shrines bracketing the Great Building form a wan
of sh1ines along the northern boundaly of the Butkara I sacred area.
The east wing shrine (the FA complex to the right of the G B in fig.
97) probably was also a direct-access relic shrine. This shrine was
modified so many times that the original form of the building remains
unclear. However, we know that it initially housed a long rectan-
gular podium accessed by steps and that over time it was expanded
into a complex of three room s built on a high plinth.~ Directly
opposite t11e cast win g shrine is a phase I stii.pa, which by phase li
had an attached chamber with a stucco floor (built in two phases),
M Faccenna, 811tkam I, vol. 3, no. I, 77. The eat·lier CSt3 lLpper circumambu-
latory path was completely covered and integrated into the GSt4/ I. encasement.
~~ The dat.c is based on an Azcs II coin found in the foundation pit of the wall
in room A and on two coins, one of Kuju la Kadph.ises and another of Azcs U.
found in a shrine (FA and FAU) immediately to the east of the Great Building
(Ibid., vol. 3, no. I, 1.68).
56
Ibid., vol. 3, no. I, 15 7- 8.
11
Ibid., vol. 3, no. I, 158.
53 Ibid., vol. 3, no. I, 160- 3.
59 This chamber was built at the same time as the fourth encasement of d1e
main stiipa (GSt<~/l }. Traces of the jambs and hinge sockets survive (Ibid., vol. 3,
no. I, 160- 3).
60 Over rime, offerings would have been given to the rclic (notably sculpture}, so
d1at ultimately a room sucb as this would have contained much more rhan the reli-
quary and the actual relic.
6 ' Facccnna, Khan , and Nadiem, Panr I (Swat, Pakistan), I 05- 6.
62 TI1e coins include one Huvi~a, two Huvi~ka type, and two Vasudeva I (Ibid.,
129, 131- 2).
Copyrighted material
102 CHAPTER FOUR
., The excavators argu~;;d that the foundation of Pfu1!' corresponds to the pe riod
between Butkara I GSt 3 and GSt4/ I, on the basis of architectural similarities aqd
masonry; this is approximate!)' phase [[. They divided COJ'lSil'uccion of the s.ite into
tlu·ec periods: Period l, from the middle of tl1e Ist centmy to the 2nd ceotuey
C.E.; Period 2, the 3rd and 4dl centuries; and Per·iod 3, the 4r.h and 5th centuries
(fbid., 65, J29- 30).
64 £bid., 4·0, 1!>4- 62, 182-90.
65
Main sWpa !6 and surrounding monuments 13, 24·, 20, and 25. The excava-
tors suggested that the presen ce of 31 and 30 would complete this symrn etridll
organization (Ibid., 49).
66
See fbid., I 05- 6.
6' For an analysis of the Sa.idu ma.iJ1 st1ipa. reconstruction, sec Fa.ccen.na, Sai(/J;
Slwrif, the Sacred A·rea, Appendix A, 43 1- 5 76.
68 Fc)r period 1, one coin of Azcs Il, one coin of Huvi;.ka, and an inscriptiO):!
(S 23 14) were found . Fussman believed that the inscription was written between 5:0
B.C.£. and 50 C.E., but he also accepted a 78 C .E. date for Kan~ka's era; th(ls
Copyrighted materi I
THE PHASE U SAC RE.D AREA 103
this dare would be pushed fotward about 40 years, following the 120 C.E. acces-
sion date accepted here. Therefore, the monastery was founded during phase 11
(Callieri, Saidu Sharij llu Mo11astny, 118- 9).
69
T he following coins were fotmd in the sacred ~trea: one Vima Kadphises, one
Huvi$ka, tluee Huvi$ka -type, two Vasudeva I, two Vasudeva-typc, and two that a re
worn and unidentifiable. All but one, a H uvi;;ka coin from pe riod 11, come from
the alluvial layers and collapse material in period rv (Facce nna, Sltidu Sharif, the
Sacred Area, 158- 63; see also the discussion of numismatic evidence fi-om the monastery
in Callieri, Saidu Shmif, /he Mollaslt~)', 118- 9).
70
Soapstone ashla r and schist diaper masonry teclmiques are found at Saidu,
Pa nr, a nd Butkara l. Schist diaper is the excl usive mason ry technique a t Butkara
I in the later periods H and nr (Faccenna, Said11 Slunif, the Socrtd Area, l 00- 1).
" Ibid., 9 1- 100.
72 [bid., 5 1- 55.
13 [bid., 5 1; D. Faccen.na et al., Il .Frcgio Figurato dello Stiipa Aincipale neli'hea Sac:m
Buddhista di Saidu. Shmif I (SwaJ, Pakistan), vol. XXVffi, /slA 0 Reports a11d Mcmoin;
(Rome: TslAO, 200 1).
Copyrighted material
104 CHAPTER FOUR
The earliest structures placed in this sacred area were built in front
of the main stiipa, and they stand on this early schist floor (F3).
Several of these are symmetrically organized pillars, one pair being
24 and 29; the structures 69 and 75 probably also had this fun<>
cion, although their identification is less clear. TI1e first small stiipas
arc also paired, 21 mirroring stiipa 31; if we add stiipa 32, it still i.s
aligned with 31. 14 This kind of organization would have had a dra-
matic visual effect, and it is quite different from the less organized
sacred areas of Trudla and the Peshawar basin of the same period.
situ and others within the sluines at the bases of stiipas, providi ng
•
considerable evidence for accurate reconstruction. 17
T hese two-ceLJed shrines all have front porch areas v.tith low plinths
on either side for the donation of sculpture. The inner chambers
may have been sealed with wooden doors. 78 T he stupas in these
shrines were fasruoned out of extremely refined diaper masonry, sug-
gesting that they were expensive commissions. The shrine housing
stilpa 13 is atypical: it has pillars bracketing its entranceway, tlu·ee very
small stupas ( l 0, I l , and 12) stand in the front part of the cham-
ber, and the primary stflpa is in the back, framed by a small niche.
T he relationship of these two-celled shrines to the small center
stii.pas is unclear, because it seems probable that the most important
structures in this sacred area are on the periphery, not in the cen-
ter, unlike the configuration at Ka.J.awan (fig. 20) or the phase ll
Dharmarajika, although both also have two-celled shrines framing
parts of their sacred areas.
77
Ibid., 693, 697. Several panels were found in situ '"hile others were found on
the floor near the :rllipas, where sockets lor their placement were evident, suggest-
ing that the narratives were attached at the "springing point of the dome" (Rahma.o,
" Butkara Dl," 702- 5). However, the preliminary report does not provide the specific
find spots nor photographs of this extensive body of material. The excavators noted
that the site was inhabited after the Buddhist period, a factor that they suggested
considerably disrupted the site . ln particular they noted that the tops of some of
the stiipos were leveled (this is tr·ue for some stiipos in the centcr cow·t and some
in the t:wo-celled sluines) (Rahman, " Butkara ill," 706).
18
Nails were recovered that the excavators suggested were used to lix these doors
in ~lace (Rahman, ''Butka.ra ill," 694).
Khan, "Preliminary Report ofMllljanai," 12- 13. See also Enington, "Numismatic
Evidence for Daring the Buddhist Remains of Gandhara," 2 1I.
80
The excavator suggested two phases, on d1e basis that ashlar masonry was
used on the eastem side of the site and diaper on the west, but no specific details
were given in the report.
Copyrighted material
106 CHAPTE R FOUR
81
An o1J1er example is Baligram, where three lar<.,:e stiipas face an even Larger
fourth (Khan, " Preliminary R eport of l'vla1janai," pi. VI).
112 Ibid., 10.
"' Ibid.
&I T he panels an: w·Ufon·n in scale, a.re made of the same ma terial, and are
curved appropriately {fbid., LO, 13- 20). For· a discussior1 of these reJjefs and their
relation to other sets of Candhararl na r-rative, see K . Behrendt, " arrative Sequences
in the Buddhist Reliefs fi·om Gan dha ra," in SouJ/1 Asiat1 Archaec/Qgy 2001 (P<u·is: in
press); M. Taddei, " R ece nt Archaeological Researches in Gandh ara: T he New
Evidence," in Source.r if Ga11dhiira11 Buddhism; Archaeology, A•·f ar1d Texts, ed. K. Behrendt
and P. Bra ncaccio (Va ncouver: University of British Columbia Press, in p ress).
Copyrighted mater 31
THE PHASE Il SACRED AREA 107
8.1Even though the G5- 6 and L sr.ructures are slightly ofr the acrual processional
path, they must have been used in conjunction with this approach.
Copyrighted material
1
eG TI1e circa 5th cenru.ty C.E. main sllipa. a[ Mohenjo Daro in the Sind has IUl
enterable relic chamber (see Chapter 3). After the 5th cennu-y C.E. a few stOjJIU
were constructed in north and west India the ma>' have had entemble relic chanv
bers, the sites of Sopara and Kush.inagara being of particular interest.
Copyrighted materi I
•
CHAPTER FIVE
1
There are many instances of srnaU traces of gold stiU adhering t.o schist imagery.
2
T housands of phase IT schist sculptures and architccturdl fragments were found
at sites like Thareli, Ranjgat, and Takht-i-bahi alone (see Appendix D).
Copyrighted material
110 CHAPTER FlVE
tllis means that the Taxila evidence is less helpful for understanclr
ing the development of schist sculptural production than the Peshawat.
basin sites, which of course are not as well documented. Althougl'l.
stucco and lime were used to finish masonry structures throughoul
the northwest, stucco imagery did not become common until phase
Ill. There are, however, a few stucco sculptures from T axila an~
the Peshawar basin that can be directly compared to the phase N
schist narrative reliefs. 3
The narratives would have been placed in a band around the
drums of these small stapas so that devotees could read them in thet
process of circumambulation (figs. 8, 121 ). Additional muTative ele--
ments were placed in false gable reliefs and on four sides of the
hannikii (figs. 122, 87, 95B). T hese figurative panels appeared Ol)
str.ipas together with supporting elements such as atlantes, lions, anc;l
elephants; decorative bands that include fit:,tt•res cany ing garlandS1
pseudo-vedikiis shown in relief; checkerboard and floral motifs; dom~
embeUished with lotus patterns; and chnttriiva/i (see section 5.4 Tl~
Phase II Small Stiipa: Sculptural Embellishment and a Proposed
Reconstruction and ftgs. 85, 93, 95). \IVhile some of these stil.pa embelb-
ishments, such as lions supporting stflpa bases, false gables, and hannikii
reliefs, remained popular during phase Ill, a raclical transformation
sometime near the end of phase IT made sliif;as •-inged w.ith narra.
lives obsolete. When iconic images became popular and began t0
appear within the sacred areas, in independent image shrines ot-
a.fflXed to preexisting structures, the practice of illustrating the Buddha~
actions in narrative panels became superfluous. Well-preserved phas~
Ill sttipas with in situ sculptural embellishment such as tl1ose in the
sacred areas of J aulian (fig. 28), T akht-i-bah1 (fig. 46), and Ali Masjio
(fig. 96) are adorned with row upon row of iconic Buddhas and bod-
hi.sattvas. Na1Tative references, when present at all, have been reduced
to a few hand gestures (rnudl'ii.S) that refer to the major events suJ.'I-
rounding the Buddha's enlightenment and role as a teacher (see sea.
ti.on 6.5 The Phase m Small Sttlpa).
arrative sculpture that recounted the life and actions of the
Buddha appears to have been created as an attestation of ind.ivid~
3 A small group of stucco reliefs depicting }ilt1kas were fow1d in sitll in the Rl
shrine attached to pedesrals at tJte Dharmarajika complex; a few smaU stucco na.t-
rative scenes under Little arches smvive from Takht-i-bah.f; and a un:ique depiction
of the first sermon can be low1d at Tharcu {see Appendix B).
Copyrighted materi, I
•
THE PHASE n DISTRJ.Btr.riON AND FUNCTION Of SCULPTURE 11 [
ual faith. In the hope of gathe1ing merit tor oneself or family, Buddhist
d evotees in Gandhara would sponsor the construction of small stu-
pas and shrines. A handful of inscripl"ions, such as the copper plate
record fi·om .Kalawan (see section 4 .1.1 KalavV<ln: Phase JI Architecture)
speak of this practice. A rare inscription on the base of a narrative
relief from Begram, showing the Buddha sitting in abh0;ya. mudrii and
flanked by devotees, reads " ... gift of Antarr; through this root of
bliss [may it be] for the worship of (her] father." 4 While such a
donative statement is open to a range of interpretations, it does seem
that Antarf paid tor a single element of a sti1pa embellishment and
that the offering was made for the spiritual benefit of her father.
The architectural evidence demonstrates that devotion was cen-
tered on rel.ics during phase 11, as is illustrated by the diffusion of
stupas and relic shrines.5 In a small panel lrom an unknown site in
Buner, we get a sense of the nature of this kind of devotional activ-
ity from an illustration of a group of monks and lay devotees cir-
cuman1bulating a small stzipa (fig. 120). This emphasis on relics is
also seen in the stories depicted; commonly, a sequence includes the
Pminirviirza, his coffin, the cremal"ion, the dlo;play of the relics (a bell-
shaped obj ect covered by a cloth sitting on a throne), the division
of relics, the transportation of relics, and finally a stilpa under wor-
ship, Like the Buner example (fig. L08). In these pictorial accounts,
a scene is commonly devoted to the veneration of the Buddha's
relics, thus mirroring actual practices that we know took place in
the pha<;e II sacred area (see section 3.1 Chinese Pilgrims' Account<;
or D irect-Access R elic Shrines in Gandhii..-a).
• " ... y(e) A(rh)t.ariye danrunmt•he ime~a kusamule•:ta pituno pt(iae Q)havatu)."
S. Konow, "Khamshti Inscription on a Begram Bas-Relief," Epigrap!tir,a lndica x--x.n
(1933- 34): ll - 15. While the narrative an from Afghanistan is broadly related to
that fTom Pakistan, far less is known about its development or chronology. Some
Afghan narrative panels were being produced as late as phase Ill, judging from
their iconographic sophistication.
j Relics have been recovered from many small sltipas in GaJ1CU1ara, but they are
not always present. While relics were looted by treasure hunters who destroyed
srnaU stiipas looking for reliquaries where coins could be found, it seems that some
smaU sllipas actuaUy were built without any relics at tl1e core. Nonethcless, they
would have been understood as relic structures by the religious community.
11 2 CHAPTER FIVE
5.1 Tize Issue of Pro11enance: Late 19th and Early 20th Century
Photo Documenl.atUm if Gmzdhiiran Sculpture
6 For an extensive discussion of 19th <md early 20th century documentation o:11
Gandhiira, see Errington, "The Western Discovery of the Art of Gandbara."
1
Gandharan sculpture also went to museums in .Bombay, Madras, Luckno<~,
Allahabad, and, following Partition, Chandigarh.
8 The British colonial musewns were tmderstaffed at a time when an enormous
M. Senatt between 1882 and the 1890s.10 Together with other rel-
evant images, these photographs were assembled as the Archaeological
Survey of India Indian Museum series. 11 Several group photographs
from dus period show the assembled sculptures from a given site,
providing an essential tool for understanding the regional schools.
T ypically, only well-preserved sculptures were photographed, along
with major architectural elements. Group photos exist for the sites
of Koi Tangai, Loriyan Tarigai (figs. 85, 88, 89), Mala Tarigai, Mian
Khan, Mir J an, upper and lower Nathu (figs. 93 , 95), Sanghao
Rhode, Sanghao C hinlai, and Sanghao Nullal1, and individual sculp-
tures were photographed from ilie sites of Sikri and Kafirkot. Major
H. H. Cole, who excavated some of these sites, published two reports,
one including plans (figs. 92, 94) and the other providing photo doc-
umentation and descriptions of some sculpture 12 (see sections 5.3.1
Upper Nathou Sacred Area: Original Sculptural Placement and 5.3.2
Lower Nathou Sacred Area: Original Sculptural Placement).
The first two important collections of Gandharan art sent to Europe
were lost. One, consi~ting of pieces from J amal Garh1 assembled by
Sir Clive Bayley, was destroyed when the Crystal Palace in London
burned down irt 1866. In 1885 Cunnirtgham sent to England a large
collection that must have contained material from J amal Garhr; it
wa~ lo. t at sea when the steamer Indus sank off the coast of Sri
Lanka. 13 It seems probable that some of the sculpture photographed
by Caddy and Senart •-v-as on the Indus; thus, these early photograp!L~
m·e aU the more valuable.
The group photos provide us with specific information about sacred
areas that are no longer extant. Some of d1ese photos depict nearly
1
° Caddy's name appears on many of the Archaeological Su1vey of India IJtdian
tvfu.qeum series photographs held in the India Office of the British Libr:uy, but
much earlier H. H. Cole mentioned that he employed M. Scnan to take 32 pho·
tographs of sculptures of Sanghao, l\llian Khan, and Nathu for his publications
{Cole, A1femorandum 011 A7ltie1l/. Mommre11ls in Eusoj~ai, 11 ; and Greet·o-Buddl!ist Sculph1res
.from ru~ai). The photographs taken by Senart were 1·ttcr atu-ibutcd to Caddy.
11
'Dte Indian Museum series of Gandharan finds survives as a labeled, intact
set of about 250 photos, now housed io the India OfFice of the British Library.
Many of these photographs arc also in the Warburg Institutes photo archive, the
British Museum 's holdings, and the Alkazi collection's Fluke Album.
12
I thank Elizabeth Enington for providing me with copies of these two reports,
which have otherwise not survived in the scholarly record: Cole, .Memqrandw11 on
Ancienl Monummu in Et~sqfttli; and Greet»-Buddhist Sculptures .from Yusjz:ai.
11 J. Burgess, The Cwrdlwra. SculpliJrn: A Selectio11 of .11/Rstratimu i11 twmtyfive P!nle.!
•
THE PHASE Il DISTRIBUTION AND I'UNC'IlON m· SCULPTURE 115
16
BeUew, A Gmt'tal Report 0 11 the Yusuj<.ttis, I 31; Enington, "The Weste m Discove•y
of the Att of Ganclhar.1," I l l.
11
Faccenna, Sttidtt Sharif, the Sacred Arll(t, 535; D. Faccenna et al., 11 Fregio FiguraJo
dello Stupa Pri11cipa/e ne/J'Area Sacra Buddkista di Saidu Slurrif I (Swal, PaJ..ista11), vol.
XXVIII, IslAO RL/>or!J and /otfemoirs (Rome: lslAO, 2001). At Butkara !, 16 niches
(15 are extant) ri.ng rhe drum of the Butka.ra I GSt4/l encasement of the main
stiipa, each containing a large green schist figu.ral panel (14 of the 16 survive), and
relic deposits were found behind each of these relief.$. These panels were not equally
spaced and are of different sizes, suggestin g that they were reused {sec Appendix
C) (Faccenna, Butkara I, vol. 3, no. I, 83). Unfortunately, the sculpnu-al represen-
tations are so splintered that the original image can on.ly be made out in one
instance, niche Ql. This relief has been identified as a depiction of the "adoration
of the Buddha," and on the basis of this relief it has been suggested that the set
of 16 panels also would have related to the life of the Buddha (Faccenna , Butkara
I, vol. 3, no. I, 83. lt is my opinion that this relief post-dates the phase Il narrative
] 16 CHAPTER FIVE
tradition, because it shows many similarities with later phase I]] iconic depictions.
Beyond the f.1.ct that no specific na.rrative event is indicated, the hierarchicaiJ>' large
central Buddha is worshipped in the top row by three Buddhas (identifiable by their
lll!~'i{OS and halos) (Behrendt, "Narrative Sequences in the Buddhist Reliefs from
Gandhara'').
18
A survey of this area conducted by K yoto U niversity w1der the direction of
S. Kuwayama located a group of remains that could be upper and lower Nath ou,
but these data need to be cross-checked with Cole's plans.
19
Cole, NlerfllJra11dum on Ancio1l tlfonumoiiJ ;,. Eusrif<;a~ 15.
20 ASu\1 nos. 11 05- 111 5; Cote, Gree«J-Buddl!ist Sculptures fiom Tusj<;ai, pl. l 0.
21 Ibid., 54- S, pi. l Q-19. 'Dtree scu.lprures appea1i.ng in plates 17 and 18 actu-
aUy were e.xca,~ated at Sanghao Rhode at tllis time by Cole.
n This set of reliefs is addressed in detail in Behrendt, "Narrative Sequences in
the .Buddhist Reliefii from Gandhara."
Copyrighted mater 31
THE PHASE IT DISTRIBUTION AND FUNCTION OF SCULPTURE 1) 7
depicting atlantes (F); five lions (G, not shown in fig. 93); one false
gable (H ); one harmikii relief (I); a set of four yalqi images (J); one
very small seated Buddha (K); three small seated bodhisattvas (L,
two shown in fig. 93); and three small to medium-sized standing
bodhisattvas (M , one shown in fig. 93).
The sculptures are primarily from phase IT. O nly five devotional
icons were found (K- L), all small, and no late architectUt·al features
appear in Cole's plan. Further, the recognizable sets of images indi-
cate that this sculpture must have formed the original embellishment
of at least two and more probably three small to medium-sized
stzipas?:J (see section 5.4 The Phase J1 Small Srupa: Sculptural Embellish-
ment and a Proposed Reconstruction). This is in agreement with th.e
plan, which seems to indicate that the schist imagety was found
around three stiipas.
23The existence of at least two slupas is indicated by two sets of narrative reliefs
(a group of ten related narrative panels and a two-panel set. The three b;mds of
standing figures and figUres under arches appear to have come fi·om three different
stiipas.
2<>Cole, M111rwrandum 0 11 A11cienl Mo11umeT1ts ·in Eusqf:t;ai, 15- 16, pl:m 7; Greeco-Buddhist
Sculptures .from r.ISj;:.ai, 53- 4, 56-7, pi. 6, 14, 2D-23. ASIM photos 1100- 1104.
Copyrighted material
118 CHAPTER FIVE
shrine was recovered. The other two chapels have small, projecting
front walls, ,.vjth a C-shaped plan that is more characteristic of the
single-celled relic shrines (see Chapter Three and section 7.3 Tharel~
Phase fi and Ill Architecture). This hypothesis is corroborated by
the type of sculpture recovered. About 25% of this marerial is ;i
mixn1re of phase ll and Ill sculpture that could well have bee~
reused p1ior to its deposition in these shrines, a feature characteris~
tic of sculptural assemblages from relic shrines (see Appendix C)~
T he remaining 75% consists of sets of related reliefs that originall~
em belli shed at least two small stzipas and include (fig. 95 ): a set of
three naiTative panels,~ along with tJ1e associated pilasters (A); a sec
of two hannikii relief~, one showing a figure supporting tliratna wheel$
and the other a depiction of the Pminiruiirw (B); a band made up of
nine panels showing figures supporting a gar·land (E); a set of I 1
panels with figures under arches (F); a set of two panels Mth figure~
under arches (D); a set of two panels with rows of figures (G); and!
a set of four pseudo-vedildi rel.iefs (H). All of these reliefS would have
adorned small to medium-sized stUpas. Also significant is a large falsa
&rable relief made up of multiple schist panels (C), which must hav~
originally been attached to a lar ge stii.pa (fig. 95; four of the five
known pieces are visible in this photograph).26 Note that structure Si
could not be the base of such a stiipa, because it lacks projections
adjacent to the stairs and has a rectangular plan.
2.1 Two panels show rl1e Buddha and worshipers, and the third depicts the first
m"ditation of' SiddM.rtha.
26 The top of the false gable appears in ASIJ\1 11 02. Four of the na rrative pan-
els depict scen~-s in the palace, and the fourth shows the Great Depanure.
Copyrighted materi I
•
THE PHASE 11 DISTRIDUTION AND FUNCTION OF SCULPTURE I 19
27
Deane, .Memmmufum. 011. ExUil/()tions ol Sikr~ Yusqftai. For the restored plan and
an extended discussion of this site and its sculpture, see Dar, "The Sikri Scu.lptures."
~ It has been assumed that Deane found d1e Sikli stlipa intact, although in other
instances where narrative reliefs have been found in stupa shrines (Marjanai and
Butkarlt III), dH;y have been lying m·ound the stripa in haphazard arrangements,
obscw·ing the original order. The Sikri slii.pa was first reconstructed near the site
at the Mardan Session Court in the Garden, and later outside the Lahore Museum
(Dar, "The Sikri Sculptures," 2 1- 23. The tlm:e photographs in the India Office of
the British Library {photos 449/61, 62, 66) record the reconstruction in one of
d1ese two early outside locations. One of d1ese photos appears in A. Foucher, "Lcs
Bas-Reliefs du Strtpa de Si hi (Gandhara)," Jmtmol Asiatique, no. 2 {1903): pi. l, and
a second in Burgess, Tize Candlzam SculjJtures: A Selectitm rf 1/JustrotimiS in auentyftve Plales
.from !he British and !Ahor Museums, pi. 4.
29 Two photos now in the Warburg Institute show that the relief order was main-
tained when the Sikri .!tupa was subsequently rebuilt inside the Lahore Museum.
For a more complete discussion of this set, including the ordering of specific reliefS,
see Behre ndt, "Narra6ve Sequences in the Buddhist Reli efs from Gand.ha ra."
30 R. Skene Grant, "Report on the Exploration of the Buddhist Ruins near Ha.rkai
Yusafzai, During d1e Months of March and Aptil 1874, by tJ1e 9th Company
Sappers and Miners Under the Command of Lieutenant Skene Grant, R. E., Punjab
Public Works Depttrlment Proceedings, Civil ·works, No. 17- A: Appendix a, l- 4," in
E. Errington, "The Western Discovery of the Art of Ganclhara and r.he Finds of
Jamalgarhf," (J>h.D. d1esis, London University, 1987). This report and other early
accounts related to d1e site of K.arkai are addressed in Errington's th esis, 138-•~6.
Copyrighted material
120 CHAPTER FIVE
st Elsewhere in the Karkai report the arches arc described as being "built evi:-
deotly in the ordinary Buddhist style, each row of stones projecting slightly above
that beneath it until the opening was spanned" (Skene Grant, "Exploration of the
Buddhist Ruins at Takht-i-bai," 440).
31 Ibid.
Copyrighted material
THE PHASE D DISTRIDUTION Al'iD FUNCTION OF SCULPTIJR.E 121
the interpretation of the horoscope) (fig. I 08), the birth, first bath,
return to Kapilavastu from Lumbini, Siddhartha in school, offering
of grass, the Parini1viir.za and the cremation, and finally the division
of relics. These stflpa panels would have been read in the course of
circumambulation, from right to left. The story that unfold~ within
this group of narrative sculptures is a familiar one, empha~i zi ng events
surrounding the Buddha's birth and death. It directs attention to the
corporal presence of the T eacher and the circumst.:1.11ces ultimately
leading to the placement of the relic within the stiipa.38
In the case of Marjanai, we can reconstruct the appearance of
the stflpa. It has a high base surmow1ted by a projecting molding,
tlu::n a vertical drum section, another molding, and then the dome,
which is poorly preserved at the top. It is unclear how exactly the
narrative images were attached, but several loose (reused?) late stucco
fragments were found in situ on the first molding, a position where
we could imagine the schist narrative ringed the stfipa in an earlier
period. 39
Copyrighted material
122 CHAPTER FIVE
have round bases, multiple horizontal moldings, and high domes that
were crowned with umbrellas, giving us a vivid idea of how taU and
imposing even the smaller stupas must have been.
Because no pha'!e li stiipa survives with in situ reliefs, the mo[t
accurate sources for reconstruction are d1e depictions that exist wi.th_in
the phase II narrative relief sculpture. Considerable progress has
been made in this direction by D. Faccenna in his discussion of thfl
reconstruction of the main stii.pa at Saidu (fig. 105).40 Although t.he
small sW.pas follow a different developmental pattern, this foundatioo
is invaluable. sw.pas are shown in 3 1 reliefs, and structural infor.
mation can also be gleaned from more than 15 model stUpas an<il
stiipa-shaped reliquaries.41 The Buner relief (fig. 120) is fairly typi.cat
although better preserved than most. This relief is onJy 13 cm tall
but it is possible to distinguish the elevated parts of the stti.pa, corni'"
monly lost in tJ1e archaeological record. Above the square base are
multiple bands of decorative sculpture interspersed with plain area,ss
crowning the st.-ucture are a lotus dome, decorated harmikii, and cha"'
trli.valf.. The stli.pa in this tiny relief provides a summary view of what
a structural sliipa would have looked like; the narrative reliefs and
the false gable were not indicated because of the small scale of tJ1e
panel. Nevertheless, this kind of evidence is most useful for under'-
standing the vertical ordering of sculptural elements. Even more help-..
ful are the many relief sculpture, themselves, which offer numerous
clues to their placement.
Many pieces of carved schist would have been required for a
complete sculptural assemblage and would have made these small
phase Il stiipas quite expensive (see fig. 8). From tl1e sets of sculp•
ture preserved in the early photographic evidence compiled by the
Archaeological Survey of India, a small to medium-sized st"Upa might
have required the following. Lions, elephants, or atlantes would have
adorned the base (a set of eight atlantes [fig. 93, F) was found at
upper athou, and a large set of 14 was reported fi'om J amal. Garlu"). 4~
A set of narratives would have encircled the lower part of the drum
Copyrighted material
. '
THE PHASE 11 DISTRIBUTION AND FUNCTION OF SCULPTURE 1.23
Copyrighted material
124 CHAPTER FtVE
+~ Because the curvature of d1ese relie£~ can be measured, the diameter of 1.he
domes of the stiipas can be extrapolated. Domes ranged in diameter from as little
as a half meter to perhaps as large as 3 meters. Beyond a diameter of approxlt
mately 3 meters, the curva ture of the relicts becomes difficult to detect. If 1.be dianll-
eter was suflkicntly la.rge, it would have been unnecessaty for the sculptor to cuovt
the relief at all
·~ Altl1ough theoretically they could also adorn large-diameter stiipas, the small
scale of the reliefs often precludes this conclusion.
46
At 1\vujanai a relid" of figures under arches was found •vith the nail still i(l
the hole (Khan, "Preliminary Report of Matjanai," I0, pl. 22b. Stein mentioned
that a nail for afl'ixing sculplUre was found a t Takht-i-biihl in court XIV (Steu:l)
" E..xcavacions at Sahri-Bahlol," l 08).
•• For a summary of constouction evidence, see Zwalf, Callllhara $culfJJt11'8, 50-63.
Copyrighted material
• J
THE PHASE li OTSTRffiUTJON AND FUNCTION OF SCULPTURE 125
along the top of the base or around the drum (see section 2.3.1 T he
Phase I Stiipa). This distinction appears to be a defining factor for
reconstructing the phase II stupa base. Three categories of phase II
Ga.ndharan stupas can be recognized, all of which have different
bases: (1) main stV.pas on high bases accessed by stairways attached
to a front projection; many of these stupas also have columns on the
fom corners of tl1e base (figs. 101, 104, 105); (2) small stupas, found
in sacred areas, usually on square ba~es, but witJ1out attached stair-
ways (figs. 2, 46, 78, 80, 120); and (3) small stupas placed in relic
shrines, which in phases I and II never had square bases;48 in these
instances the enclosing shrine appears to define the sacred space,
rendering the base superfluous. For example, altho ugh the stiipa
depicted i11 the Buner relief (fig. 120) has a square base, it~ upper
superstructure is nearly identical to me round-base stupas preserved
in shrines from Pippala, Marjanai, and Butkara ill (figs. 39, 40, l 06,
107, 100, 102). Except for ilie square base in the Buner example,
all the stilpas share the lower stepp ed register and similar sets of
upper moldings and recessed bands. If tJus hypothesis regarding round
stupas is correct, then the base is not truly part of the stupa; rather,
the upper circular superstructure is the relic m onument. In contrast,
ilie unenclosed phase II stupas from R anigat's east sacred area (figs.
78, 80) have square bases like the Buner relief
Other facto1-s also suggest that ilie base functioned to define sacred
space, in tJus sense as a separate structm al element. This may explain
why relic chambers often were placed at the height of the base. It
could also be ilie underlying reason iliat pilasters commonly appear
on sti£pa bases from all periods in Gandhara; they may be making
an architectural reference to defining tl1e sacred space of th e stupa.
It has even been suggested that tJ1e base was fabricated to elevate
the stiipa so that it would be visible from a greater distance. 49
Sculptural depictions nearly always show stiljHIS wiili square ba~es,
tl1e only exception being a single example from Butkara I that shows
a small stupa with a round base, but in this instance it is housed
within a shrine (fig. 99). In six out of the 31 relief examples, stairways
8
' For examp.le, middle to late phase Ill squ<tre-based .rtiipas are found in shrines
at Dhannarajika (shrines R5, N 16) and at Kalawan (shrine A 16).
9
' G. Fussman suggested this function during a panel discussion of a paper at
the conference "On the Cusp of an Era: Art in the Pre-Kushan World," Nov. 8- 12,
2000, Kamas City.
Copyrighted material
126 CHAP'l'ER FlVE
are shown, Lndicating that a main stiipa is being depicted. This group
can be ex-panded to 20 examples if we include representations <if
.rtfipas with pillars on the corners of the base. Tlus leaves only 11
relief.~, som e of which almost certainly show small stilpas, the Btmer
example being among them.
The stUpa in the Buner relief has a rnolding at the very bottom,
then a plain section topp ed by a cornice that imitates wooden archir
tectural beams (termed herein a pseudo-beam cornice). This arrange-
ment seems very like that of the many extant stilpa bases found in
the Peshawar basin and Swat; typically in the real str~.pas, the plaio
base was fabricated out of masonry and finished in plaster, wherea:s
the pseudo-beam cornice was commonly constructed of multiple smal'J
pieces of schist and also finished in plaster. The square bases arce
undecorated in all but two of the reliefs; in these instances, flowers
fill tllis section.so The phase II stilpa H in shrine 31 at Pippala has
lar-ge stucco lotus flowers set between pilasters encircling its base (fig&.
39, 40).5 1
T h e relief sculpture showing stfljJas does not help us to understancl
how schist sculptures were integrated into the bases of small stiiJHISo.
The obvious categoties of sculpture that almost certainly were attached
to tlus part of the stftpa include lions, elephantS, and adantes. These
figures must have supported the base in some fashion, while at the
same time differentiating the stiipa from the sphere of the mundane
world. T he set of eight small adantes (fig. 93, group F)s2 and the
set of five lions found in the upper sacred area of fatl1ou probably
were affLxed to one of the st1ipas indicated in Cole's plan of the site
(fig. 92).
It is difficult to determine the location of the adantes, lions, and
elephants because extant phase 11 stiipas such as those at R anigat
(fig. 80) do not provide hin ts as to their placement. H owever, con-
siderable phase III stucco evidence shows that Lions (fig. 96) or a
combination of lions and atlantes (fig. 28) or elephants and adantes
served as transition elements between stories within the base. T ypicaU)'I,
phase nr stilpas are stepped; in essence, these stftfJas have multiple
so Faccemta, Saidu Sllilrij, the Sacred Area., pl. 266c, 77a; see also 66a.
so Similar loll•ses set between colum ns but executed in schist were found in th.e
Peshawar basin at Sanghao RJ1ode (ASfl\ll photo nos. 1137, 1138) <md at Sanghao
Chinlai (ASlM photo no. l 032). For a discussion of the dating of the J>.ippala B
st•iPa and shrine, see MarshaU, Taxila, 366.
sz At J amal Garhi 14 schist atlantes were recovered in 1873 and another 15 i:il
1920 (Errington, "The Western Discovery of the Art of Gandhara," 228, fig. 8.22.
bases. The lowest and largest base provides a sanctified platform for
the second base, which in turn provides a base for the more sacred
area of the dome. The transitional zones between stii.jJa bases are
clearly marked; the figures are shown lifting the stii.pa up and sepa-
rating it from the mundane world. In a similar fashion, immediately
above the step of the first base are, in some cases, rows of lions sup-
porting the next level, as can be seen in the Ali Masjid example
(fig. 96).~3 vVhile it is clear that the stepped base is a phase Ill devel-
opment (see section 6.5 The Phase III Small Stz1pa), leaving the phase
II stf!.pa form in question, the atlantes, lions, and elephants support-
ing bases probably originated in the earlier period. In the small phase
II stii.pa recon truction proposed here (fig. 8), atlantes, lions, and ele-
phants are shown in register A, but, as noted above, the relation-
ship to the lower moldings is not understood.
S$ G runwedel also rightly pointed out tbat the second and third bands of the
dnun should be reversed, so that the checkerboard pattern is in the center re&-is-
ter (A. Grlinwedel, and J. :Burgess, BuddhiJt. Art i11 l11di.a. translated by A. Gibson,
London: Bernat·d Qaritch, 1901, 155), an element that corresponds to the pseudo-
uedikii in many stacked reliefs. While the stepped base is to m1ally con·ect, both tb.e
row of Buddhas and the row of putti are variants of the figure-under-arches care-
gory (see below} and thus should occupy the top register. These incongruities sug-
gest that these sculptural disks were borrowed fi·om multiple small sltipas.
!<\ The reconstructions of both the Loriyfm Tat1&>ai and Sikri sliipas (fig. 9 I) su~
gest t.bat narrative •vas placed at the level of the base (see Faccenna, Saidu SllaJjf,
/he Sacred Area, 537), but in both iJlStances tl1e narrative could equally have been
placed on a step above the base, or in the case of Sikri, on the drum itself. The
fact that pseudo-be;un cornices sometimes crown narrative panels, as in the Sikri
example, is misleading; it has led some to conclude that the reliefs belong to the
upper part of the base. However, this motif occurs in many contexts; for example,
in the Buner relief it appears at four different levels, topping the base, the stepped
platform, and the pseudo-uedikii, and above d1e harmiM.
~' Sometimes figures such as putti or women standing on pots are enclosed in
these li·am es. ln the larger narratives sculpted on separate pieces, commonly d1e
pilasters were sculpted separately, as at upper Nathou (fig. 93, A). Some of the
pilasters were adorned with characters such as monks, lay devotees, or Buddhas. A
few instances of continuous narrative exist in the Peshawar basin, an ideal exam-
ple being the J a111L'il Garhi depiction of the white dog barking and several other
panels that form a set.
Lay followers, monks, putti, and so forth. 59 Some of the arches con-
tain a row of relics (figs. 63, £; 95, D); in others the Buddha is
under an arch and flanking devotees are under adjacent arches (fig.
63, C).
4. A row of niches, each containing a Buddha (fig. 63, B), or a
row of cells bracketed by pillars, each of which contains a Buddha
and flanking devotees. T hese Last types appear to be a Late phase II
or early phase III development. Dwing phase Ill, after narrative
had been abandoned, the stucco production commonly included
related rows of Buddhas and bodhisatt:vas, either in niches or brack-
eted by pillars (figs. 96, 28).
Thus, a continuation of a pattern seen in phase II is observed in
the phase Ill stii.pas, where bands of Buddhas and bodhisatt:vas under
arches are depicted. The main difference is that the pseudo-vedika
band and the naJTatjve depictions have been replaced with rows of
small devotional icons. I have argued elsewhere that a chronologi-
cal transformation can be observed in the Gandharan narrative tra-
dition m1d that a late narrative gro1.1 p e..'Xists that is typified by the
Sikri stupa muTative reliefs (fig. 91 ), although related narrative sets
would include the one consisting of ten panels fi-om upper Nathou
(fig. 93, A) and three sets, six panels, eight panels, and l 0 panels,
from Sanghao Rhode. These late narrative examples display a large,
centrally placed Buddha, shown as a teacher, while the unfolding
ncuTative is marginalized. Thi~ format seems to be a precursor to
the repeated non-naJTative Buddha images that became so common
in phase Ill Gandharan stupa decoration.
A conclusive link between the na rrative tradition and the phase
Ill stucco evidence is provided by a stacked relief from Jamal Garhi,
photographed circa 1880- 90 (fig. 63, A). T he J amal Garhr stacked
relief follows the conventional pattern discussed above: the middle
register has a pseudo-vedikci, and the top register has figures under
arches; however, the bottom register, instead of being occupied "'~ th
narrative reliefs, has a row of cells sepcu·ated by capitals, each of
which has a Buddha flanked by worshippers. This is an explicit
9
$ P. Brancaccio ("Gateways to the Buddha: Figures under .1\rchcs in Early
Gandbaran Art," in Sources qf Ga11dlumm Buddhisnt· Arcltaeolagy, A·rl tmd T exls, eel. K.
Bebrendt and .P. Brancaccio [Vancouver: Univemty of British Columbia Pres , in
press]) has <u·gued on the basis of matetial fi·om Swat that t11e figures, each in turn,
face a cemraJ Buddha and th us can be read as devotees or· possibl)' as donors.
60 For depictions of false gables, see AilS, negarive no. 436.47a; FaCCCJula, Saidu
Sharij, Jha Sacred Area, pl. 274b, 75a, 75c, 76c, 77a, 81a, 8lc. For architectural frag-
men~q showing the position of false g-<~.bles, see H. lngholt and !. Lyons, Gandlwrar•
Art in Palcistan (New York: Pantheon Books, 1957}, pl. 471; Mizuno, ed., Meklwsmzrkl,
pls. 48- 2, 64- l 1; Nishikawa, Ranigat, pis. 185- L, L85- 16.
Copyrighted mater al
THE PHASE U O!STRIDUT!ON AND FUNCTION OF SCULPTURE l .'33
lobe extended above the band, which typically contained the pseudo-
vedikli (fig. 8, zone G).61
For small stupas the false gable was carved out of a si11gle piece
of schist (fig. 122), but we also find some vety large multi-piece com-
plex false gables (fig. 95, C). A particula rly well preserved example
from Loriyan T angai is more than 2 m tall and was composed of
ten separate panels; it is large enough to have been affixed on the
face of a main stiipa.. As noted elsewhere, some of the earliest devo-
tional icons that can be dated on the basis of masonry were fow1d
within trilobe frames (fig. 18) (see section 6.3.2 The Dharmarajika
Complex: Early Phase III Stiipas with Axial Image Shrines).
In Afghanistan a phase IV main stupa at T op-Darrah has a mas-
sive in situ trilobe niche on the upper part of the drum (fig. I 17),
positioned somewhat higher than was the case with the phase 11
small slUpas. This niche contained stucco devotional icons, as indi-
cated by the sockets for their support and traces of a halo in the
central ruche. 62 Its presence provides solid evidence for the use of
large trilobed panels on main stupas, and it shows that this element
was still in use during phase TV.
61
For the Loriyan Tangai reconstructed .rtiipa, the lhlse gable, which does not
appear in the early photo of this reconstruction, should ex-tend down one more
band and its crest should rise to the height of the sholtlder of the dome; i.e., it has
been placed too high and its scale is inaccurate.
~ S. Mizuno, ed., Basuwal tllld Jelalabad-Kabul: 13uddlzist. Cau-Temples and Tapes in
South-East. ilfglra11ist<m SulTifled Mai11£y i11 1965 (Kyoto: Kyoto University, 1970), 126,
pi. 48-9.
'
stiipas (fig. 87, 95B, two pieces shown). It is easy to identify the nar-
rative sculptures used to adorn ltannikiis because some were carved
on the four faces of a single block of stone and in other cases fouT
interlocking square slabs were used. 63 The content of the four sccnc,s
almost always relates to the lour great events in the .Buddha's life-
birth, enlightenment, the first sermon, and death.
The surge in patronage that characterized pha~e ITI left a dear trace
in the archaeological record and can be corroborated with numis-
matic and inscriptional evidence. So·uctures built in phase ill masonry
dominate many of the sites in Taxila, and typologically similar pat-
terns of construction are present in Swat and especially in the Peshawar
basin. Although masonry cannot be used to date structures in the
Peshawar ba~in and Swat, the majority of coins recovered fl-om
Buddhist centers date to phase lll; tlus is true for Taxila as well (see
Appendix A). 1 Along the Karakoram trade route, Kharo~~hr ancl
Brahmf graffiti inscriptions that record dates also appear to show
activity in the 2nd and 3rd centuries C.E. 2 and thus corroborate the
architectural and numismatic evidence.
At the Dharmarajika complex at the end of phase IT, only a few
monks seem to have resided in the small phase I viharas. A few lived
along d1.e west edge of the sacred area in the E, F, and T monas-
teries, and a small number lived to the north in the loosely organ...
izecl vihiiras (courts A and B), and some more in cells V 1- 5 (figs. l,
16); only 3 7 ground-level rooms survive from these monasteries.3 The
majority of those using the Dharmarajika on a daily basis must have
lived in the nearby, larger phase II monasteries of Akhauri A, B,
and C and Khader Mohra D I and D2 (figs. 21, 22, 24, 25, 26) (see
section 4.1.3 Dharmarajika Satellite Monastic Complexes: Phase U
Architecture). The Akhauri and K.hader Mohra viM.ras have 119
e>-.'tant ground-level rooms, and another 34 can be postuJatecl on the
basis of reguhu·ity of the uihiira plan, for a total of 153 rooms. At
Kftlawan another housing area comprised about 90 phase ll monas·
tic cells (45 cells in monasteries known to have at least second st~
ries).'1 Additional unexcavated small phase ll monasteries appear to
have existed to ilie norili, but the total population of monks seems
to have been linlited and dispersed.
These monasteries were apparently allowed to fhll into ruin dm,.
ing phase III, exceptions being Kalawan and the Dharmarajika com-
plex, where stiipas I and 4 in the court A and B viha.ras were
maintained and augmented with imagery into phase IV (fig. I 6,
courts A and B). During phase Ill, four new monasteries were added
monastery at the end of phast: Ill. It is even possible that it dates to phase IV o r
to a date similar to that of the Taxila site Bhamala. See Ibid., 35 1- 53.
1
The chronology of Peshawar basin mOtmtain vihliras is poorly understood, bur
it is certain that these sites as a whole were built primarily during phase Ill.
8 Under the phase III quadrangular monastery, remains of an early structure arc
visible along the northcm edge, the only level ground suitable for building a significant
phase I or U monastery (field observation, 1993- 94).
9
For the dating of the SITuctures in the sacred area of J3utkara I, see Faccennrt:,
Bullwra I, vol. 3, nos. I, 57, 115- 18, 130- 33, 160, 167- 74.
10
Rather than attempting to make Butkara l conlonn to the phase chronolo!m
I use here the struct'ures built in Faccenna's petiod of GSt3. This group faUs withio.
phase 11. Faccenna has establi:;hed a relative dating for many of the secondal}'
structures at Butka.1-a 1, b<Jsed on sll~•tigraphic relationships of multiple paving lay.
ers preserved in the sacred area; this provides a broad seq uence for the fabricatiol)
of subsid.i3.1y structures. While it might not be possible to assign an absolute dat:e
to a givco monument, structural typologies that can be extended to Panr and Saicl.u
corroborate this relative devcloprnemal pattern.
11 Structures built during Faccemta's period GSt4/l- 4/ 4.
structu res in the form of additional stilpas still dominated the phase
lU and IV production at Butkara I.
What are the implications of the phase lii increase in production
of sacred stiuctures, especially of image shrines? From a cluonolog-
ical perspective, the increase suggests that many of the surviving large
schist stucco iconic images from Gandhara must date to phase Ill
(see Appendix D), because almost no life-size image shrines date to
phase II (see section 6.3.1 The Dharmarajika Complex: Possible Late
Phase 11 Image Shrines). An upper chronological Limit for this kind
of image production also seems clear; by phase TV, monumental
sculptures and images with complex iconography had supplanted the
phase fil tradition. D ating of the schist images is difficult, but exca-
vations at J auliafi, Mohra Moradu, the D harmarajika complex, and
Butkara I show that stucco and clay images were common during
phase Ill (see specific discussions of these sites and Appendix B).12
The source of the increased phase ill patronage remains a mystery,
but the physical evidence at these sites provides a picture of its impli-
cations for the expansion of the Buddhist tradition. At some point
in the 5th century C.E. (the beginning of phase IV), this wave of
new construction halted suddenly (see section 7.6 Phase IV: The
D ecline of Buddhism).
12
Mru-shall's excavations at J aulian show that most of the structures were built
near the transition between phases Il and Ill and that most repairs are in phase
m mason ry (Marshal], Taxi/a, 369, 71). Exceptionally well prese•ved St1.1CCO images
at this site, along with evidence from other sites, such as Mohra Moradu and Ha~lga,
provide strong evidence that srucco imagery Aourished in phases IT! and IV. See
M arshal!, Ta.,ila, 259, 66.
13 Marsball, Taxilo.
11
Marsball's predisposition for identifying structures as image shrines is illustrated
in his discussions of the phase 1 strucnu·es a t the sites of j at;lc;li al B and C and in
his analysis of many of l11e phase 11 shrines at Kalawan (Ibid., 222- 30, 322- 41 ).
15 Because image sl11ines, like small stiipas, were usually added to a site one by
one by single patrons, it is often possible to determine tJ1 e order of their con-
struction. Image sluines commonly were constructed in rows, one next to another,
so an added shrine " pdS often built in part on the base rnolding of an aqjacent pre-
existing shrine. I was able to determine micro-chronologies of shrin e construction
at tl1e sites of Takht-i-biH11, J a ul.iii.n, and Tha reli. On tl1e basis of such observa-
tions, it seems clear that tha t the earliest sh1i.nes were small and had a narrow
recess suitable for an under-life-size image (field observations, 1993- 94). Sec also
detailed discussions of these sites.
16
Ma.rshall, Taxila, 248.
17
Th e relalive stratigraphic superposition shows that shrines B I 0, B9, B7, D I I ,
0 7, and D3a postdate the unified group of sliipas or pillars (58, 59, 8 20, Dl 3,
0 tO, D I, D2, 0 3, R4, and the sllipas nex-t to B7 a ad B 16). They also postdate th e
la rge phase ll unified group of shrines (D5, 0 8, 0 6, D9, D 12, and D 14) (field
obse1vations, 1993-94).
in the middle or late part of phase II. These live shrines are significant
because they are the only possible linage shrines from all of Taxila
fabricated from diaper masonry; their presence may mark a poirJ!t
of transition, the hu·ge munbers of phase m semi-ashlar image shrul($
indicating the degree to which this trend became important.
6.3.2 The DhamwTiijikii Complex: Ea.Tf:y Phase IIl Sntpas with Axial
image Shrines
At the beginning of phase ill, most people using the Dharmarajikti
complex would have entered from the north. This is especially true
for the lay community, who would have lived in Sirsukh, and any
possible remnant population from Sirkap. T he phase II circulation
pattern would sti11 have been used, directing \·vorshippers to the lit
two-celled shrine and then to the western processional avenue lined
with relic shrines (fig. l ). 18 Some of the first phase Ill additions
appear in the northern quadrant of the sacred area; they are easily
recognized because an early fonn of semi-ashlar masonry was employed
that relied on extensive use of kafijur blocks. 19 The northern gateway
into the prada~irwpatha of the main stii.pa took on greater importandl
at this time; the large stupas Kl, N4·, and 5, together with th(t
smaller stupas K2, K3, P3, and P6, were all built along the newlr
forming northern avenue. 20
D uring th.is ea.rly part of phase IIT, datable stupas were constructel(i
that have axial image shrmes mcorporated into the facades of their
bases. Stupas K I and N4, both built of early phase Ill proto-senU..
ashlar masonry, have axial trilobe niches in the center of each face
of the bases. In the north niche of the Kl stzipa, an in situ stuce0
18
The phase Ill importance of the western avenue is attested to by the rcpaii;s
carried out on its structures (some as late as phase lV), even after th e western ga.(i:
was sealed in. the late part of phase m or the beginning of phase rv. Late (possi>-
bly phase JV) \\~.ills buttressed the C5-6 and RI shrines, and many scmi-ashlaJ·
and double semi-ashlar repairs were made to the 12, C3, C4, and CIO relic StntC-
tures (field observations, 1993-94).
19 Marshall, Toxila, 261- 64.
20 Stiipas K2 and K3 extended over an earlier water tank, which as a COJlSe-
quence was filled in. Within this !rulk coins of Soter Megas and Kadphises were
found. Therefore, these sWpt;.s clearly postdate these 1st centu.ty C.E. coins (Ibid.,
261). Further, sll1pa K3 had three coitlS of Kan~ka, and slii.fJa P6 had 10 coins of
Huv~a and seven of Vasudeva (Marshal!, Taxila, 263). The coins of Huvi~ka, espe-
cially, indicate that probably much of this phase of' construction must have occu•Tci:l
sometime not long alter 200 C.E. (see Appendix A).
.
Copyrighted mate1 31
'
PHASE m ARCHITECTURE AA'D SCULPTURE FROM TA..XJLA 145
Copyrighted material
146 CHAPTER SIX
G monas te ry (fig. 16) had about 50 cells and probably was multi-
storied. A smaU stiipa in the courtyard and on axis with the entrance
perhaps i~ related to the later north I ndian practice of placing sUZ-
pas and image shrines in similar locations, as can be seen at sites
like alanda . T he M5 villiira (fig. 1) at the D hannar~jika complex
also had a stiipa in the center of the courtyard. These two monas~
teries appear to be the latest Greater Gandhara n examples to con-
tain a central stiipa in the court. I n phase 11 most monasteries
contained inte rnal stilpa shrines for monastic worship; onJy the phase
Jl K hacler Mohra D2 (fig. 26) contained a central court stiipa, whereas
this was more common in phase I (see section 2.2.1 Dharmm·ajika
Complex in Taxila: Phase I and section 2.2.2 Buddhist Stwctures
in the Taxila City of Sirkap: Phase I). Evidence !i-om man y other
sites in Taxila show that by phase lii stupas placed in monastic cells
were being augmented or replaced by iconic images (figs. 29, 36)
(see section 6.6 Phase liT Use of Images and Relics in Quadrangular
Monaste ries).
6.3.4 77ze Dharmarajikii Comple.x: Phase Ill Image Shrines Along 1/ze
Northern Avenue
The bu.iJding of new m on a~ tic residences made the northern ena·ance
to the sacred area in1po1.tant for the resident community, and it
seems no coincidence that considerable additions were made to the
northern avenue at this time (fig. I). The sa·uctures in th.is m·ea can
be placed in early middle phase Ill, in a time before shrines N27
and 28 blocked the eastem end or
this corridor, cutting off circ u-
lation and making this a less desirable location for donations. Sl1rines
one of Kani!)ka found in a pot in ceU 16. Another sm.aJJ coppe r casket comain.ing
five coins of Vasudeva was found in some debris outside the north wall. These
hoards certainly were deposited at a time when Vasudeva's coins were the prinlal'f
currency in circula tion, providing an approximate date for the consu·uction of this
monaste ry. The presence of foUl· coins of the Kidara, 11 of the late •· Kidll.ra, ant'!
two of Chandragupra D suggests d1at this uilziira w·dS occupied throughout phase lli
and inro phase JV (Marshall, Taxila, 282- 83).
A wide range of coins found in the MS monastery include: one lvl:eoander, one
1-lennaeus, five Azes Il, one Kadphi ses I, one K.adphises II, one Kani~a, 19
Vasudeva, ruJd dwee Sasanian (Nlarshall, Ta>.ila, 290--9 1). lt is difficult to account
for r.hc p•·esence of some of the vety early coins, other than ro suggest that r.hey
remained in circulation or that the monas tery was built on ea rlier structu res, but
the presence of 19 coins of Vasudeva in corytmction with semi-ashlar masonry
would seem to date this monastery ro near the bcguuling of phase Ill
Copyrighted matem l
Pl-lA.SE lll ARCHITECTURE AND SCULPTURE FROM TAJOLA 147
N27 and 128 probably date to the end of phase Ill or even to
phase IV, as they bracketed the eastern avenue together with image
shrines N 17 and 18; these later shrines both housed monumental
images and were built in late double and triple semi-ashlar masonry
(fig. 19) (see section 6.3.6 The Dha•·marajika Complex: Late Phase
m AJ·chitecture).
A considerable number of smaU stupas were built along the northern
avenue, augmenting those added early in phase Ill (see above); they
include stiipas M1 3, Ml4, . :ll 5, P9, Pl O, and P12. 24 An early relic
deposit from the U l round stiipa25 may indicate that this stiipa and
U2, U3, and possibly K 7 date to earlier periods, and that with the
expansion of this part of the sacred area they were refurbished in
phase ID masoruy. The first large group of image shrines also appears
in this part of tl1e sacred area: shrines 29- 32, P7a, P7b, Pll , P I2a,
P l2b, and P1 2c. 26 Image shrines P12a, Pl 2b, and P l 2c were built
against the earlier P 12 stiipa; their presence indicates that worship-
pers walked past d:Us corner to venerate stftjJas J4 and N5. It also
seems possible that the phase Ill small stupa N26 had some special
importance, perhaps similar in function to d1e subsidia•y sacred area
d1at formed around the small stii.pa D6 at Jauliai'i, which included
an addition of •line image shrines (fig. 27; eastern edge of site).
The phase m image shrines along the northern avenue are rela-
tively small and therefore probably contained images of modest size.
As noted above, in Taxila <U1d the Peshawar basin, where the sequence
of shrine construction is known, relatively smaiJ shrines were con-
sistently constructed early in the site's development and were grad-
ually supplemented by larger and larger image shrines as the idea
of f~tbricating monumental images took hold. This trend is also seen
at the Dharmar~jik.'i complex, where all of the monumental image
1
' ·n,ere a re other small stitpa.r created in single-course semi-ashlar outside of the
east gate of the main swpa, but in rny opinion they date to the c•·ca6on of the
eastem avenue, and thus to late phase m.. They a re associated \\~ th the massive
double and u·iple semi-ashlar image shrines built on rJ1 e nonhern end of the east-
em avenue.
2
~ The st1tpa contained an e<u·ly •·elic deposit, including phase [ coins (M arshal.! ,
Taxiln, 272). Tllis stflpa is discussed at greater length in a loomote in Appendix A,
ph ase ]]] evidence.
26
All of these shrines were fabricated of single-COlU"Se semi-ashlar masonry; walls
of 3 1 and N32 rest on the base of the early phase ill K I sl!l.pa (field observa-
1
tions L993-94; Marshall, Taxila, 27 L). Marsh all identified P LI as a stiipa, but because
this structure has a rectangular base and because of its relation to the P 12a- c shrines,
t11ere scents little dou bt 1.hat this is the base of an image shrine, not a stupa.
Copyrighted material
148 CHAPTER Stx
shrines were constructed of late phase Ill double and triple semi..
ashlar, and the smaller liie-size shrines in diaper and single-course
semi-ashlar.
Why was the northern avenue created? It does not provide access
to the main stilpa, and it does not seem to have focused attentio1'}
on the H two-ceUed relic shrine (the fa_s:ade of this slu-ine was eve~
tually sealed late in phase ITI).21 Yet t11e northern avenue is where
the first coherent group of structures housing devotional icons wa$
built. O nly with the addition of these image slu·.ines is the northern
avenue a fully formed processional corridor. O ne could argue th~
the circumambuJatory passageway arotmd the main stiipa was alread~
full by this time and that space was available in this outlying are4
for donative structures. It would not have been easy to displace th~
already existing and significant relic shrines; few irnage shrines were
ever built in this area, and those that do exist, such a~ B l and B4!,
must date to the end of phase ill or to phase IV.
Image shrines could have been relegated to the northern avenue
because of religious conservatism. T he Dharmarajikii complex was
an important regional center, and perhaps new religious practice$
involving the use of images were only hesitantly accepted; image
sh•·ines are also sparse at Butkara I, the only other regional centeJ.
that ha~ been excavated. If this hypothesis is correct, then perhaps
the construction of image shrines was purposely pushed to the edg<:
of the sacred area.
The bank of images would have provided a dramatic point df
entrance into the sacred area. O ne can postulate that devotees, after
entering at the northern gate, would have turned east and walked
down the length of the northern avenue, proceeding past the N4
stiljJa and tl1en turning down the not yet fully formed eastern avenue-
in essence, circumambulating the perimeter of the sacred area. T he
idea of an outer pradak#rw is not improbable; tllis route would have
brought the worshipper in contact in the sout11 with the large J l
and D4 stiipas and would have led ultimately to the western avenue,
which was still a significant point of access to the main stilpa.
The monastic population would have accessed the sacred area
along the northern avenue, so it is pos~-ible that this addition of new
image shrines might have been built to serve their ritual needs rather
27
Marshal], Taxi/a, 248.
Copyrighted materi 11
•
PHASE TU ARCillTECTUR.E Al'iD SCULPTURE FROM TAXILA 149
28
T races of diaper masonry may indicate that the semi-ashlar facing does not
mark an expansion of the main sliipa, only a restoration. See Kuwayama, 17ze Main
S~a at S/!iil!-y'i-ki-t/fleri, 25- 6, fig. 4.
Some section.s of the dome show a partial second cou rse of ashlar stones,
although this might be d1e product of modern restoration (field observations, 1993- 94).
so Probably these axial image plinths were not const:ructed at the same time,
becatL~e they have different dimensions. The core of the north plind1 appears to
predate the rdiocing of the stapa, when it was modified and paired with th e south
plinth. The west plind1 is poorly preserved, but it appears to postdate the refacing
of the stupa, as does the extant east shrine (field observations, 1993- 94). These
shrines can be loosely dated to the middle of phase Ill. They did not contain mon-
umental images typical of late phase Ill and phase IV. Late phase Ill or phase IV
shrines of double semi-ashlar masonry seal the north and west emrances, suggest-
ing that the axial shrines are earlier.
31
In the case of shrine B 13, the umer walls are single-course semi-ashlar and
the outer are double semi-ashlar. This practice of rebuilding early sti1Jctures in late
masonry is also apparent at Klilawan (two-celled shrine A2).
Copyrighted material
150 CHAPTER SIX
32
Facccnna, Saidu Shnrif, the Sacred Area, 22Q-2 L, 223- 24, 264- 65, 269- 76; my
field obse1vations, L993-94.
33 I obse1ved this pattern wheo I first visited Jamal Garhi in 199 1 and again i1t
L993, but when I returned to take detailed notes at the end of 1993, stiijJa court
3 had been completely rebuilt, in the interest of conservation.
Copyrighted materi I
•
PHASE lU ARCHlTECTlJR.E AND SCULPllJRE FROM TA.XlLA 151
s-o The ori&oinal gateway, constructed mostly in kaiifrtr, still exists. !t is clear that
this structure was simply modified, and then a large double semi-ashlar shrine was
added to the original gate (field observations, 1993-94).
!$ The R6 shrine verges on double semi-ashlar in places.
36
[mcrscitial chips are lL~eful in determining the relative order of construction.
For ins~ance, if wall A were constructed first, then large blocks would have been
used to finish the corner. At a later time, when wall B was built up against the
corner of wall A, instead of blocks being used to finish the edge of the new wall,
small chips of stone would have been used to fill this g-ap, as larger finished blocks
were required only on exterior surfaces. Because the edge of the second wall is
made up of chips, the order of construction .is simple to detem1int:. It appears that
Copyrighted material
152 CHAPTER SIX
with relic shrine 83. 37 Then shrine 84 was added (using one of the
walls of 83).38 Thus, tllis group of shrines (R7, R6, 83, and S4) all
can be securely placed in the late part of phase UI; tlus group cart
be extended to include R5, because double senli-ashlar masonry was
also used in its fabrication. The closed facades of these shrines indi~
cate that they contained relics rather than images, as is clear in the
case of the R5 smpa shrine. In tl1e southwest part of the S3 shrin~
a sizable raised platform of mud contained a large number of earl1
Gupra period clay seals with the Buddhist creed.39 Little remains tc:t
give these seals a proper context, but it seems likely that they werd
housed in a stii.pa, as is the case in at least one instance in Afghanist:an. 4~
In any case, it is unlikely that tlus base supported an image, as bas~
for images were fabricated using masonry in Taxila at this time,
Although the massive rectangular S3 shrine is structurally atypicaJ,
because it had an entrance portal with a cent'fal pillar and flanking
pilasters, it seems in line •vith single-celled relic shrines. Related ta
tllis continuing interest in relic structures are the 25-30 small stupasJ
dating to the middle to late parts of phase m, that are scattered
about the D harmaraji.ka sacred area, esp ecially along the eastem
the circwnambulatory path was partially filled at this time, because shrines .R6 andl
R 7 si t on a layer of din that is stratigraphically higher than tbe base of the west~
ern stairs. Approximately 45 cm of debris accumulated in the western gate priol;'
to the consn·uction of the R6 and R 7 shrines; in fact, the fo undations of the Rii
shrine have settled because of tllis lack of a finn footing (field observations, 1993-94);
Marshal! also noted this lack of foundation on his site plan, but he made no ref.
erence to it in IUs text.
" Shrines R6 and S3 share a common wall that extends ltcross the fu.cades of
both structures. Corroborating dlis dating is the (a.c t that the ba.ck wall of S3 wall
done in double semi-ashlar.
sa The S4 shrine appea.rs 10 follow the plan of an earlier phase .ll su·ucru.re.
Supporting this cont.ention is the retention of a small amoum of phase D' masonr)'
in its northwest com er.
" The dating of the seals to the early Gupta period is based on the form of the
Bralunr characters (Marshal!, TtiXilll, 266- 67). A large number of these seals were
found in a raised platform; they were stamped with the Buddhist creed "ye dharma
Hetu prabhav...." Although M.arshall noted that such seals are found in the COr e$
of stllfNlS, he identified this as an image shri.ne, suggesting that the seals were pla.cect
in the cores of Buddhist statues. While it is a·ue that seals are found in stitpar, they
are not, 10 my knowledge, associated with sta tues in Greater Gandhara. Therefore,
it seems very likely that this shrine was instead used ro display or contain relics.
10
' M. Taddei, "Inscribed Clay Tablets and rvfiniau•re Stl7j>as from Ga1JJi," EastJ
mu/ West 20, no. 1- 2 (1970); M. Taddei and G. Verardi, "Clay Smpas and Thrones,
at Tapa Sardar, Gazni," in ;;jnbun: A1mwirs qf tlze Resmrclt Institute for Hummtistic Studie3-
(Kyoto: Kyoto University, 1985).
Copyrighted materi I
..
PHASE ll1 ARCHITECTURE AND SCULPTURE FROM TAXU.A 153
avenue. Late phase ill repairs were also made to relic shrines along
the western avenue, indicating that they also were being used at this
time.4'
At the end of phase Ill, two points of access into the sacred area
can be recognized: the eastern avenue leading into the mona~tic area
and the presumably public northern entrance opposite the monu-
mental P2 image shrine . The northern entrance was important
throughou t phases I, II, and Ill, as ind icated by the periodic
modjfications made to the H two-ceiJed shrine, as weU as the for-
mation of the northern avenue in the early to rruddle part of phase
Ill. It is clear that at the end of phase Ill tills entrance was still
being used, because the colossal P2 double serru-ashlar image shrine
was built here. The colossal scale of tills shrine suggests possibly even
phase IV construction (see section 7.5 R.arllgat: Phase II to Phase
IV Architecture). Even before this shrine was built, sculptures were
being commissioned for the entranceway; in mjd-phase Ill, two over-
life-size seated images of the Buddha, nearly identical to those found
in the monastery of Mohra Moradu (fig. 36) (see section 6.6.2 Monastic
Images: Phase Ill Pedestal Images and Image Shrines built in Pre-
existing R esidential Cells), had been attached to the back of the P l
stii.pa to face this northern entrance. Shrine P5 and the base of an-
other colossal image shrine, P4, augment the late phase ill P2 image
shrine. Although it is unclear how the public would have entered
the site, as the densely packed northern aven ue has no clear exit
except toward the H structure, these commissions show that this area
remained an important component of the D harrnarajika complex.
In contrast, the eastern avenue was organized to create a clear
processional path that charu1eled the monastic population directly to
what had become the most important entrance into the prada~~apatha
of the main stupa. It would seem that at this time, the lay and monas-
tic communities used and moved through the sacred area in different
ways. With the blocking of the northern avenue, the lay population
lost convenient access, via the eastern avenue to the prada~r:tapatlza.
Perhaps the northern avenue was purposely blocked, to restrict traffic
and segregate the secular devotees.
" Single-cow·se semi-ashlar repairs and modifications were made to the C4, C3,
and 12 shrines, while shrine G LO was fabricated out of double semi-ashlar masomy,
as were repairs to the outer wall of shrine G4. These repairs probably pre-date the
closure of the western prada/qi!wpali1a gateway.
Copyrighted material
•
42 The 16 sttlpa slu·iuc can be elated by comparisons ''~th similar sltlfJiiS f.-om
.Jaulilh1_. which ar(~. much better preserved. MarshaU was able to _solidly doo.unctll
at .Jatilifli\ a sltljm lonnat tJ1at he used as a datutg mdtcator at Ta.,.ila. See Mar~~hall,
Excl1lJ(I/wns at TaxiIn: ·n, Stupas and Mont1steries aJ ]auJian, 7- l 0; Marsh all, Taxila,
372 73; Fouchl'r, "The Decor;uion of the Stuccoed Stupas."
<l Marshall, Taxikt, 267 68.
PHASE OI ARCHJTECTURE AND SCULPTURE f"ROM TA.'ULA 155
By the end of phase II, the sacred area wa~ organized around a
main stilpa, A4, ""ith an oversized relic chamber, stupa shrine A 14,
and normal-size stii.pa Al2. These core monuments were enclosed in
a court ringed by multiple one- and two-celled relic and stii.pa shrines;
the monastic population lived in monasteries C and F and the out-
lying H site (see section 4.1.1 Kalawan: Phase II Architecture). ln
phase rn the monastic population increased significantly with the
addition of monastery B; stairs indicate that at least a second Aoor
existed, so this structure could have housed 80 or more residents.'14
During phase Ill, new relic structures were added to the now-con-
strained space within the sacred area, such as the single-celled stii.pa
shrine A 16, which sits between two other empty single-ceUed shrines,
A IS and A l 7, that also probably hou~ed relics. Single-celled shrine
A 15 contained the kind of heterogeneous do native sculpture that is
often found in relic shrines at other sites45 (see Chapter Three; sec-
tion 8.2.3 Takht-i-bahr: Two-Gelled Shrine XXJII: Loose Sculpture;
and Appendix C). The shape of shrine A26, near the entrance of
the monastery, also conforms to the single-celled typology on the
basis or its plan. Also indicative of the importance of relic structures
to the phase Ill community of K ftlawan is the fact that the main
A4 stupa (with its oversized relic chamber) was repaired, along with
two-celled stupa shrine A 13, while two-celled shrines AS and A2 were
completely rebuilt in phase Ill semi-ashlar masonry.<H>
The in1age shrines that were incorporated into the sacred area at
this time are mainly affiliated with relic structures: image shrines
were attached to the facades of the A5 and A2 two-ceUed shrines,
and two shrines, A27 and A28, were affixed to the rear wall of the
Al4 two-celled stupa sluine. Severa.l more were built near the entrance
to rnonaste.r y B (A24 - 25 and A30), and an iconographically com-
plex image group was housed in a slu·ine placed within the monaster)'
itself (see section 6.6.3 Monastic Images: Late Phase Ill Gmuihakuti
Image Niches).
Thus, relic shrines were used to frame and delimit the Kalawan
sacred area and the zone around the main stiipa at the Dharmarajika
complex. In contrast, at many phase Ill sacred areas, including the
Ta>:ila site of J auliaii (fig. 27), most of the Peshawar basin centers,
Copyrighted material
156 CHAP'l'ER SlX
and some sites in Swat, image shrines were being used as boundary
structures. Although image shrines and relic shrines must have had
radicaUy different ritual value, they both were used almost like the
northern Indian vedikii to create a bounded sacred space for the relic
stiipa invariably found at the center. In this sense, certain aspect<; of
the earlier relic shrines were apparently partially transferred to image
shrines in phase ill. In effect, while the relics in the main stiipas of
all the Greater Gandharan sites remained the primary o~jects of
power and the main focus for devotion, it was the large iconic images
that gave the relics a manifest form. At Kalawan, it might have beet!
that the earlier relic structures were only complete after the attach•
ment of phase Ill image shrines.
47
These coins comprise four local Taxila.n, one Apollodoms ll, one Gondophares,
one Kadphises, seven Kani~ka I, six Huvi~ka, 37 Vllsudeva (including late imiia·
tions), LI mudentified Kushan, fom Sbapur !!, two Shapur ill, three Honnazd 11,;
one Varaharan 11, 24 lndo-Sasanian and Sa.~anian, and 17 late lndo-Sasaniatl;
(Marsha.IJ, Taxila, 385- 86). See also Erri.ngton, "Numismatic £,~dence for Dating
the Buddhist Remains of Gandhara," 212.
48 Here d1e late diape r masonry starts to technically resemble the semi-ashlar fo r-
ma t (Marshal!, Exc(W(I/wns at Taxilll: 77u Stupas aJUf Morummes at Jaulian, 4).
49 Seams in the masomy, both in the upper and lower courts, show that the
shrines were donated individually and in groups (field observations, 1993-94).
50 Behrendt, " Relic Shrines of Gandhara: A Reinterpretation of d1e Archaeological
Evidence"; M.ar1>hall, 1iv:ila, 377, 379.
Copyrighted material
158 CHAPTER SIX
~· This foUows l'r·om d1e scale of th ese shrines and their double semi-ashlar
masonry.
j
2 !Vlany of these sculptures are now in the Pcshawar M'useum, Lahore l\ll usewn.
Karachi Museum , Indian Museum i.n Calcutta, and Chandigarh Museum. The most
complete record of Peshawar basin finds is the Archaeological Survey of India
Frontier Circle {ASLFC) photographs and the Archaeological Survey oflndia 11tdiau
M.useum photographs (ASI:NI), in the India Office of the Brir.i.s.h LibrlHY in London.
1' TI1is shrine was built against the hill immediate!)' to the south of the main
stiipa's stairway. It is enti rely composed of sem.i-ash.la.r masonry and has an amecham-
ber typical of other two-celled shrines (field observations, 1993- 94).
34
The ma.i.n sllipa "~dS constructed of diaper masonry that verges on semi-ashlar.
This base has pilasters made from integrated blocks of lwiJur, Like the early phase
IJl pilasters of the proro-semi-ashlar K l sttipa at the Oh1u·marf~ikl\ complex and
unlike the insc1ted single-piece pilasters typica.l of phase .II monuments in Taxila
(field observations, 1993-94; Marshall, 7lLnu•, 358, 62).
~~ These coins comprise one punch marked, three local T axilan, one Azes l, two
Azes n, one Kadph.ises I, one Kadph.ises D, two Soter Megas, 12 Kani~ka, one
Huvi~ka, 4·9 Vasudeva, one Kidara Kushan, four Sasa11ian, one Toramana, one
Samanradeva, and four late lndo-Sasanian (Ibid., 363- 64). See also Errington,
"Nwnismatic Evidence for Dating the Buddhist Remains of Gandhara," 21 1.
~ Traces of the original red paint that coated the main .rtiipa can be detected
behind some of the images, especially the seated image on the dome at the top of
Copyrighted material
160 CHAPTER SIX
site was deeply buried in debris, especially the images placed against
the south wall of the main stUpa plinth.
The sacred area was not enclosed; however, the large individual!¥
commissioned devotional icons encircling the main stilpa might have
had this function, perhaps similar to the icons housed in shrines, ctS
at Jaulian (fig. 27). AJthough these attached devotional icons tool-
out from the main slflpa, they do encircle it mucl1 like banks of
image shrines would. The pattern of image distribution obseJVed ~
Mohra Moradu seems to differ from that of other sites, but this may
be simply because image shrines more often survive in tbe archac:--
ological record while exposed images do not. Images placed outside
of image shrines are easily confused with reused sculptures; this is
especially problematic in the case of schist devotional icons (see see}-
tion 8. 7.2 Sahri-Bahl61: Some Possible In Situ Sculpture and Appe11r
clix C). There is less ambiguity about the stucco work that is typical~}'
attached, and even when the actual sculptures have been lost, soc:Je..
ets for their attachment sometimes survive (see below).
Additive sculptures are best preserved at the sites of Mohra Morad~
J auliaii, K.alawan, the Dharmarajika complex, and T akht-i-baht,
although it seems likely that this pattern of image placement was
common throughout Greater Gandhar a. At J auJian, the main stftjm
base was embellished with a variety of heterogeneous, individually
commissioned Buddha and bodhisattva images, which appear to date
to the end of phase m or to phase rv.s
7
At Kalawan, main stujfP.
A4 and stupa Al 2 were embeJijshed with additive images of which.
only fl:·agments sUJVive.~8 In the Dharmarajika complex, two lati!
phase Ill, larger than life-size images are attached to the back fa~
of the Pl stii.pa.YJ At Takht-i-bahi, four late phase lii or IV adcli-
tive unenshrined monumental images were placed against the soutb
wall of court XX; the surviving heads and feet show that the sock-
tbe stairway. Many of the images added to the plinth of the main sl:ilpa projeet
beyond its base molding (field observations, 1993-94). Most of the Mohfii Mora®
additive images, some of which are larger than life-size, a.re now in the Tax.ila
museum.
$? Field observations, 1993-94; Marshal!, Toxik;, 371. Their large scale argu~~S
for a late date, as does their style, which is comparable to, for example, in sitll
sculprut·cs in the Oharmarlijikl'i Nl 8 shrine (fig. 19).
38 Bodhisattva heads and other fragments were fmmd at the base of the A4 slupa,
and around the A12 sltipa. many sculptutes of lay worshippers were found among
the debris (Ma.rshall, Taxila, 324- 25).
M £bid., 276.
Copyrighted materi I
...
PHASE UI ARCHITECTURE A.ND SCULPTURE FROM TAXlLA 161
ets in the wall above were originally used to support these sculp-
tures (figs. 2, 49, 50) (see section 8.2.2 Takht-i-bahi: Sacred Area
XX: Loose Sculpture).60 There is also evidence to suggest that adcti-
tive images were at times placed in pre-existing shrines, a good exam-
ple being the added images in the monumental image shrine N18
at the Dharmar~j ika complex (figs. 1, 19). In my opinion, many of
the disembodied stucco heads found at sites throughout Greater
Gandhara were originally parts of adctitive images either placed in
shrines or left exposed.
For the donor, an additive image would have been expectient, and
it avoided the expen e of building an image shrine. Commissioning
an adctitive image might have been a patron's on.ly option if all the
available space for image shrines had been exhausted. Thus, the
additive images provide anoth.er valuable strand of evidence for tmder-
standing the growth and popularization of iconic imagery in Gandhara.
60
Note the sockets in the wall in figs. 49 and 50; these holes must have ori.gi-
naUy contained wooden posts to support the monumental images. Such sockets can
be found in many shrines in the Peshawar basin (see fig. 4 7) and are useful indi-
cators for determining the scale of the images (field observations, 1993--94).
6' K. Behrenclt, " Relics and Their Repn.>sentation in Gandhara," in Marg (Mumbai:
Marg Publications, 1993).
Copyrighted material
162 CHAPTER SIX
62 1\ few large image shrines were also constructed at the Taxila site of Giri.
&l The apparen t lack of image shrines at this site may be a case of incomplete
excavauon.
&t A single-image platform was added to the monastery courtya•·d.
M T hree small two-n·lh·d sh1incs were built in the sacred area of Bbamiila. Ouc
shri1w houst'd a stripa, bu1 Ji·agmcmed rew;cd images were found in all three. · n~e
only over-lifi·-sizc image was found in a shrine auached to the gate st1·ucturc se~
arating the sacred area from the monastery. Unexcavated stnJctures to the we>t of
the main jflifJa, as M;u-shal suggested, may have suppo11:ed image shrines (Marsha'tl,
Taxik1, 39 1 97).
66 Shoshin Kuwayama, personal communication, 2001.
•
PHASE Ill ARCHITECTURE ANO SCULl"TT.JRE FROM TAXlLA 163
tional icons on plinths along the edges of the interior courty ard;
(3) small niches, set into the walls of the courtyard, containing icono-
graphicaUy sophisticated image groups (exhibiting what I c.:'lll gand-
hakuti iconography); and (4) small images, suitable for personal
devotional practices, in monastic cells.
In the mountain. uilzii.ras of the Peshawar basin, images were not
incorporated in the rnonaste•·ies; rather, a somewhat different kind
of pattern is observed. At the junction of several of these residential
structures are monastic small sacred areas, typically consisting of a
small stiipa surrounded by a walled court and commonly having peri-
meter image shrines (figs. 68, 75) (see section 7.3.2 Thareli: Monastic
Small Sacred Areas).
Thus, it appears that patterns of in1age worship in monastic contexts
differed from those in the public sacred areas. It seems likely that
in the monasteries more complex image use developed (see section
6.6 Phase ill Use of Jmages and Relics in Quadrangular Monasteries).
67
Ali Masj id ca n be placed in phase m on the basis of numismatic evidence of
two coins of Vasiska (242- 260 C.E.). See Errington, "Numismatic Evidence (or Dating
the Buddhist Remains of Gand hara," 197.
68 For a detailed discussion of stzipa development in Taxila, see Fit:'t.Simmons, Stupo
Designs at Taxila, 30- 47.
69 Mlu·shall said that a mold for the face of a Buddha(?) was among the terra-
cottas !ound at the Dharnuuiijika and J aJ:Jc.l.ial (Marshall, Taxila, 441, pi. L36k).
were easy. Some of the best documented small st11pas with stucco
decoration are those from the site of J auliafi in Taxila (fig. 27). \Nidi
the exception of small stripa A 15/ 0 all had rubble cores and were
faced with phase III semi-ashlar masonry 71 (fig. 28).
With the popularization of iconic images of the Buddha, strifJP.
embellishment showing his Life events seems to have waned; only a
few narrative reliefs are extant in the stucco medium (see section
5.4.2 The Drwn: Developments in the Na rrative T radition anti
Appendix B). Still, the phase Ill stuccowork is part of a continuous
tradition that has its foundations in the earlier schist phase II .rtrip:a
adornment. Architectural motifs such as base molclings, pilasters,
atlantes, and lions all come directly out of this earlier tradition. The
phase n motif of figures-under-arches developed in such a way that
Buddha and bodhi.sattva images now occupied these registers (figs.
95, F; 63, A), which appear as stacked bands on the smfaces of
phase ill stupas (figs. 8, 28, 46).72 It is noteworthy that similar trape;.
zoidal and trilobed niches encircled the phase Ill drum of the
Dharma.rajika main stilpa (fig. 17) a nd that related axial trilobed
niches embellished the ba~es of stripas Kl and N4 (fig. 18). In some
cases the niche motif is not present; often, the encircling seated or
standing Buddhas and bodhisattvas are simply bra cketed by pilasters,
a~ can be seen at Ali Masjid, Takht-i-bahr, and Sahrr-Bahlol Site B
(figs. 96, 48, 56). At J auliai'i even the pilasters were omitted between
the Buddhas in stilpas A2 and A 15. In general, the niche and Buddha
motif is used on the upper parts of the ba~es and on the drums.
while the Buddha and pilaster pattern occurs on the lower parts of
tJ1e base, suggesting that the arch may have hierarchic significance.73
0
' A 15 was originally constn1cted in diaper but was later repaired in serni-ashla:r
mason.ry, and the imagery appears to have been " restored" as well during phase m
(MarshaU, Excavatiolls at Taxiia: The Stuf!as a11d Monasteries aJ Jaulum, 8- 10).
71
T he .Jauliai'i stiipa format can be used to date otb er small stupas in the TaKila
area. See Ibid., 7- 10; Foucher, "The Deco1-ation of rhe S[Uccoed Stuptd'; Marshal\
Taxila, 372- 73.
n P. Bra.ncaccio, "Gateways to the Buddha: F~res lUlder Arch es in Early Gandha.
ran Art," in Sources of Cmuiharm• Buddhism: ll rclzaeology, Art 1111d Texls, ed. K. Belu·endt
and .P. Brancaccio (V:mcouver: Universi ty of Bri tish Columbia Press, in press); see
section 5.4 .2 The Drum: Developments in the Narrative T radition .
., T his is 011ly a general trend ; for example, Buddhas in niches are found a t the
level of the base on one of d1 e Ali Masjid stiipas.
Copyrighted material
166 CHAPTER SlX
76
When d1e lat~ r monastery was built, this stfipa was integrated into t.he phase
l1I viltiira, but the original, lower phase U Aoor level was preserved. Ibid., 365.
11
The date of this stilpa's crea tion is difficult to determine, but. d1e presence of
two phase ill terracona heads, which Marshal! SUggel>1Cd were images of the patrons,
seems to indicate dmt. this stilpa was a later addition (Marshal!, 'Taxila, 355, pis.
73a, 79a, and 137, nos. 77, 78).
111 Ibid., 36 1.
19
Marshal!, Taxil.a, 316-J 7, pi. 186g. Sec also ASIDG 192 1- 22, no. 74.
;
worshippers and five intact small Sravasri plaques were found in the T4 .x.xtJt!
two-celled shrine a.rea. At d1e sites of SallTi-Bahlol C and D another l6 of th~
plaques were recovered, but in these instances no architectural context is providet3.
However, the Sahri-Bahlol plaques were found together ,.,;th considerable sculpture
that can be attributed to the late part of phase m (sec Appendixes B, C, .Q).
Part:icula.dy helpful for dating is the fact that many of these plaques exhibit Sravastj-
type iconography.
8< rbid., 34·0, 717- 18, pi. 220, no. ll9.
PHASE lll ARCHITECTURE AND SCULPnJRE FROM TAXU..A 169
dus image was almost certainly reused, judging from the early date
of production and .its worn condition, it most likely functioned in
phase Ill like the other devotional images addressed above.
6.6.2 .Monastic Images: Phase Ill Pedestal Images and Image Shrines built
in Pre-existing Residential Cells
In the quadrangular monasteries of Taxila are many instances of
large images added to the interior walls of the courtyards. Often
only the bases of these more than life-size, presumably late phase
ID, stucco devotional icons swvive, as in the Mo~-a Moradu monastety,
which has traces of seven pedestals of these additive figures (B l- B7).85
Fortunately, in this monastery six intact monumental in situ Buddha
images are extant (bases B 1, B2, and B3 each support two Buddhas)
(fig. 36); depictions of lay donors on the base of B2 provide a clue
about who offered these sculptures. At other Tax.iJa monasteries,
traces of pedestals have been excavated, one in the J auliaii monastety
in front of cell 25 (fig. 27), two from K una.Ja in front of cells E9
and N2 (fig. 41 ), and possibly one at Pippala in front of st1ipa shrine
31 (fig. 39).
It seems quite likely that diflerent donors constructed these images
over time, and thus we should expect them to be placed in devo-
tionally significant locations. This is clearly the case at Moh.ra Moradu,
where seven pedestal. images were placed along the eastern side of
the monastet)', four of d1em (B4-B7) grouped around ilie cell 9 st11pa
shrine (fig. 32). This cluster is augmented by the conversion of cell
8 into an image shrine, where a nearly life-size bodhisattva (fig. 38),
a small bodhisattva, and a false gable relief were found. 86 All three
images are in remarkably good condition, suggesting that even if
these images were .reused, they must have been placed in the shrine
near the time of their fabrication, probably in the first half of phase
Ill. Their placement in cell 8 clearly reflects a desire to donate
images in association with the relic stupa standing in the adj acent
cell. Together these schist devotional icons, the additive pedestal
images, and the cell 9 stzi.pa transformed tllis section of the monastery
into a fuJJy functioning monastic smaJJ sacred area. It should be
81 A broken relief wid1 two Buddhas was found in cell 19 (no. 188) together wi(h
a st:ucco head (no. 362). A seated bodhisattva (no. 19 1) was fowtd in cell 17 (lbi<f,,
362).
88
Ibid., 381 - 82.
99 The entrance via cell 2 was later blocked in phase Ill semi-ashlar masonry.
90 At dte phase f11 site of .Ktmala, cell S5 rnight be a shrine, became it breal\s
92 The gaudlwlmti or perfumed chamber where the Buddha resided is known from
Buddhist Literature and has been addressed by Srrong ("Gandhaku!i: The Perfumed
Cham ber of the Buddha," HistOI)' qf Religions 16, n.s. •I· (1977), 390- 406). It seems
approp ri ate to use d1is tenn to describe a monastic image shrine, as Schopen ("The
Buddha as Owner of Propeny and Permanent Resident in Medieval Indian Monas-
teries," JoumaJ. qf Jndia!l Phiiosoplty 18 (1990], 181 - 21 7) has shown that the Buddha
was conceptually resident in early monasteries, at least in western India. Gandhaku!i
iconography as used here refers to the sculptural assemblage of iconic images found
in middle to hue phase ill monastic image shrine.~. Typically, tllis iconographic pat-
tern had a central Buddha I'Lanked by a complex assemblage of secondary divine
fi,aures, bodhisattvas, donors, and, in seve ral cases, attendant Buddhas (fig. 29).
93 The cell L (SI ) gmufltakuti niche has a Buddha image on a throne, supported
by atlantes and flanked by two attendantS, one male and one female (Marshall,
Exctwations aL Taxila: 17u Stupas and Monasteries at Jtmlian, pi. X VIa}. The gmu/Jwkuti
niche in front of cell 2 (S2) has a Buddha image in dl!)•lina. mndra, on a throne.
Multiple attendant figures include probable donors, a monk, a small, seated Buddha,
and a figure holding a fly whisk and wearing a j eweled headdress (possibly lndra
or a bodhisattva) (Marshal], Excaua.tilJns at. Taxila: The Stupas mzd Monaste~ies at Jaulian,
pl. XVlb). The cell l 7 (S3) gandhaku[i contains a seated Buddha in dlzarmocakra mudrii;
to the lefi a.r·e two standing figures, and behind d1em are the remains of several
smaller figures (Marshall, Excauatwns at Taxi/a: 17ze i:>rupas and Monasteries at Jaulirm,
17- 18).
~ MarshaU, Ttm:tLt, 382.
172 CHAPTER SIX
~ Ibid., 360- 6 1.
96 Ibid., 360, pi. 97b.
9' MarshaU, Excavations al. Taxiln: The Stupas mul Monasteries ai Jnu{i(m, 12.
98 Ibid.
PHASE Ill ARC'..JIITECTURE AND SCULPTURE FROM TAXJT..A 173
s At TaxiJa sometimes only foundations were made of stone, a nd d\e upper walls
were made of earth, as is the case at J>.ippala (Marshall, Ta>.iW., 365- 66). This m~
explain tbe incomplete foundations observed at sites like Sahn-BahJol.
• This interesting site is currently under excavation by Far-id Kha.n; reports have
not yet been published.
For the small, relatively simple site of Mekhasanda, in the hills above
the tow11 of Shabaz Garhi (fig. 64),5 the numismatic evidence sug-
gestS a foundation in phase II and occupation well into phase IlL
Of the nine recovered coins, one of Karu~ka I and one of Hu vi~ka
relate to phase II, while six issues of Vasudeva and one Kushan-
Sasanian coin suggest that the main period of construction occurred
during phase IIJ.6 The architectural and sculptural evidence appears
1
Ibid., 96. The eru·ly main .rtilpa was 7.6 by 6. 1 m (Mizun o, ed., Mekhasanda,
83). SrnaiJ. stiiptJS 3 and 5 probably also have an early date.
8 Miz uno, ed. , i\tleklwsa11da, 66, pl. 48, nos. 1- 3.
9 Ibid., pis . 40- 43. In addition to the narrdtive reliefs, semi-lunette fragments,
a.tla.ott-s, lions, garland bearers, rows of seated Buddhas in niches, decorative motili;,
moldinb'S, and multiple dmttra were found (Miztmo, ed., /ltMdlasaJu/a, 90, pis. 39- 49).
See also Appendix D.
10 Earlier, someone had already opened r.hc main sttpa and dug a pit in r.he cen-
ter to bed rock. It was in the process of clearing tll.is debtis that the four narrative
reliefS were recovered (Mizuno, ed., M'ekhtiSallda, 90, pis. 39- 49).
11 A likely candidate fo r such refurbishment would be sl!ipo. 20 because dw·ing
phase lll, image shrines were built ru·ound this sliipa at the entrance to the main
Copyrighted materi I
PHASE Ill AND IV ARCHITECTURE IN THE PESHAWAR BASIN 179
sttipa's processional passageway; th us, the stiijla pre-dates the shrines. Pe rhaps when
tllis a pparendy importam stilJM was redecorated with phase Ill stucco reliefs, its
original em bellislunents were reused.
2
' See Mizmto, ed., iiJekJwsamla.
" The platform along the east side of the court has rnoldin1,rs and traces of plas-
ter indicating that these were origi.naUy image shJ·ine bases and that d1ey were con·
structed all at the same time. On tl1e west side of the court, on d1e basis of similar
data as well as the presence of shrines 28 and 30, we can assume that these pla t-
forms originally supported image shrines. Ibid., 84.; my field obse rvations, 1993-94.
180 CHAPTER SEVEN
life-size images, like the mid-phase ill shrines in the upper stupa
court at J auliai'L This is conftrmed by the presence of a significant
number of in situ schist and stucco sculptures found in association
with these shrines (see section 8. 7.3 Mekhasanda: In Situ Sculpture).
The total sculptural production at Mekh a~anda may help us to
understand the relative proportion of schist to stucco devotional icons
that once stood in these shrines. About 22 devotional icons executed
in schist can be estimated (counting heads and bases as full images),
while more than 180 stucco heads and body fragments survive (many
of these can be discounted, as they clearly belonged to images attached
to phase ill small stiipas). Even given these pronounced Limitations
and the looting that occurred, it seem~ likely that the majority of
image shrines at this site would have originally housed stucco figures
(see Appendix D). It would appear that the widespread construction
of image shrines must have occurred in conjunction with sculptural
production in stucco.
Late phase Ill structures at Mekhasanda are limited to two rnon•
umental image shrines (shrines ii, xv), although shrine 30 probably
can also be so identified. 14 The excavator did not designate these
structures as monumental image shrines, I$ but compelling compar·
isons can be made to other sites; for example, at Takht-i-bahi (figs.
2, 47) the Ml shrine in the lower east stupa court has interior dimen·
sions of 3 by 2.25 m, compared to 3. 75 by 2.5 m for shrine xv at
Mekhasanda. In any case, the recovery of a monumental Buddh~
head and fragments of a second monumental head leave little room
for doubt. Regardless of the placement of these monumental images,
their presence shows that Mekhasanda was still occupied and receiv-
ing patronage in the late part of phase Ill, at a time when con·
struction of monumental images at other sites was common.
14
Shrine 30 began as a stupa in the main slilpa court; later it was modified and
transformed into a shrine, making its acu·ibution as a monumental image shrine
tentative, as no upper walls survive (field observation, 1993-94). In the case of
shrines ii and xv, the masonry and superimposition of strucmres indicate that the
moownental image shrine.~ at Mekhasanda were added alter the smaller shrines
were already in place. Even a cursory examination of the plan reveals that these
momunental sh rines were added late in the site's occupation where space permitted.
1
~ Nlizuno identified probable .rnornunental shrines ii and xv as rooms for guards
or watchmen. However, there is no precedent for non-sacred structures being
attached to stfipa courts (i\llizuno, ed., MekllliSallda, 85- 86).
PHASE ID AND IV ARCHITECTURE IN HIE PESHAWAR .BAStN 18 1
16
Bellew described the nature of the remains on the three main spurs, noring
the five or six quadrangular monasteries along the highest east-west ridge that sep-
arates Takht-i-bahi from Salui Bahlol (BeUew, A General Report on the .rus'!ft.ais, 126-
27); only one monastery was visible at the time I surveyed this ridge.
17 The 60 rnountai.n uilziira.s are mostly multi-storied, as are the two quadrangu-
the remainder of the lower sacred area (IV), in I910- 11 and the
trabeated rooms and passageways along the western edge (VIII, IX,
and the tunnel under XX). T o his surprise, he also uncovered the
late phase III sacred area XX. 22 In 1911- 12, A. Stein cleared the
two-celled shrine T4 (XXIII) to the soutl1west and the small auxil-
iary sacred area XIV to the southeast. 23 Even though plans were
made and the fmds photographed, no report was ever produced (see
also section 8.2.2 T akht-i-bahl: Sacred Area XX: Loose Sculpture;
section 8.2.3 Takht-i-bahr: Two-Gelled Shrine XXIII: Loose Sculpture;
section 8.2.4 Takht-i-bahi Small Sacred Area XIV: Loose and In
Situ Sculpture; and Appendix D).
soutJ1east small sacred area XIV (see AppendL'< D and section 8.2.4
Takht-i-bahr Small Sacred Area XIV: Loose and In Situ Sculpture).
The layout of tJ1e sacred area surrounding the Pl stupa is compa-
rable to other sites known to have phase II foundations. Characteristic
is the presence of a small main stupa with tighdy packed stftpas around
it, as can be observed at R anigat, Mekhasanda, and Thareti D (figs.
78, 65, 69), all sites that can be assigned a phase II founding date
on the basis of nurnismati.c evidence.
AnotJ1er way to determine the presence of small phase Il stupas
in this sacred area is to consider what impact they might have had
on the placement of later phase Ill image shrines. In the case of
small stftpas P2l- P24 near the entrance of the later monastery, it is
clem· that they must have been present when ilie small image shrines
10- 14 were squeezed into ilie remaining space, though tJus does not
necessarily mean iliat the stzipas date to pha~e H. While thi~ kind of
argument can provide only a relative order for construction railier
ilian criteria for dating, it does appem· tJ1at over tinle donative struc-
tures were built on the western edge of the lower sacred m·ea as
space was used up.
Structural superimposition and the fact that 1l1e image shrines abut coun X
20;
and monastcty 11 provide criteria to determine the order or const.rucrlon.
27
Food preparation srrucrures were added as separate structures at monasteries
such as those at Saidu, J aulian, and Mohra Mon'idu. In contrast, they were pad
of d·•e original phase ID su·uctures of the later Pippala monastery or the phase £V
Bh~m~tery. .
18 The Ta.kht-i-babi shrine appears to be part of tbe original snlJcture, but mocl).
em restoration Leaves this an open question. TheJaulifui and Mohra Morlldu mon~
tic image sh rines date to the end of phase lll, which does not correspond to tl'tt
time of the foundation of the Takht-i-babr monastery.
~ All 15 image shrines are nearl)' identical in spacing (the south shrines are sq>-
arated by about a 60 cm gap, and the east and west shrines are separated by aboul
85 cm), internal hei&rltt (they range li·om 210 to 230 cm from the Aoor to the b~
of the dome}, and i_ntemal width (ranging from 141 to 152 cm for the east and
west shri.nes and 130 to 136 cm for the south shrines). ~ntis level of consistency
indicates that this assemblage was fabricated as a single unit.
30
Masonry seams clearly indicate that image shrines 18- 24 were constructed
after the raised south .rtOpa court was in place. Image shrines 25- 3 7 postdate the
consu·ucrion of the quadrangular monastery, because they all evenly abut the wall
of th.is monastc•y .
" Marshall, Ttuoi/a, pi. 82b.
32 Kaiijiir uiJobed niches similar to those in d1e T2 Tak.ht-i-bahr relic shrine are
found in the early phase rn K I , N4, and J l small stiipas at the Oharmaraj i.kli com-
plex. Vruiations of this niche type also occur in coryw1ction with rn.icldle phase IU
srructu res .in T ax:ila.
a phase lii date for their establishment (the identification, fun ction,
and dating of these two-celled shrines is adcb·essed in Chapter 3 and
section 4.1.2 The Dharmarajika Complex: Phase H Architecture and
the Relic Shrine). The presence of smaiJ independent image shrines
on the north em edge of the T l terrace (39- 4 1) suggest that this
tem-ple was constructed at around the same time as the small image
shrines in the lower sacred area (1- 37): the middle part of phase ID.3a
All four of these relic shrines were located near enb·ances into the
sacred area. Shrine Tl sat on its own terrace facing the sacred area
main entrance, a location that would have made it the frrst devo-
tional structure encountered. This position is analogous to the H
two-celled shrine at the Dhannarajika complex (fig. 1). Shrine T 4,
in court XXIII, also is located along a major path leading to the
sacred area; in this instance, however, the path would have served
the monastic popula tion of the west ridge mountain viharas rather
than the general public. As court XX and the western extensions of
the sacred area had not yet been built, two-celled shti nes T 2 and
T 3 would have been encountered at the only remaining point of
access into the main sacred area. The monastic population living
on the central ridge would have used tlli.s last enb·<m ce. It appears
that these two rdic sh•·ines also functioned as componentS of the
sacred area .
n Only I l small image niches were set imo the interior wall of the T I shrine,
and their frames are less sophistica ted. Thus, T l may well be slightly older· man
T2- 4, which are consistent in design.
square assembly haU (I) (figs. 2, 45). All of them project out over a
steep hillside, requiring the largest set of foundation structures known
in aU of Greater Gandhara.
Late phase Ill prosperity is also apparent at many other Peshawar
basin and Taxi.Ja sites. Numerous monumental image shrines are
extant at R anigat, Thareli, Jamal. Garhi, and Mekhasanda; in Taxi.Ja
they are seen at Jaulian and in the Dharmar~ika complex. The list
of siteS e-"<hibil'ing late phase ill patTonage can be e-x-panded tO include
Sahr1-BahJ6l centers, Mohamed Nari, and Loriyan Tangai, if sculp-
tural evidence is taken into account (see Appendix D).
The western extensions of Takht-i-bahf probably began with the
construction of platform VIII, which is supported by ten below-floor-
level tnbeatecl chambers. These rooms were first noticed by Bellew,
who gained access through a hole in the roof of a cell; apparently
at tllis time court IX was filled witl1 deb1·is and the souiliern entrance
and stairway had not been uncovered. 34 When this part of the site
was finally excavated in 1909- 10, considerable confusion surrounded
the function of these chambers.35 Similar trabeated chambers at the
Saidu monastery in Swat led P . Callieri to concluded that these
underground trabeated rooms were built to increase stability by reduc-
ing the weight of the loose fill that pushed against retaining waUs. 36
Because trabeated vaults are held in place by weight from above,
they were ideally suited for terrace extension. It appears that the
Takht-i-bahr trabeated cells were built so as to provide a stable exten-
sion of the lower acred a rea, which probably was secondarily used
for storageY After the trabeated terrace was in place, it appears that
assembly hall P 8 and sacred area XX39 were built. The last addi.-
tion was the 8-m-thick court IX retaining wall. It seems that the
lower sacred area had been expanded by filling court IX with debris
Q1ence the need for a massive retaining wall). A false impression of
a lower cowt was created when the Archaeological Survey of L1dia
removed this fill to expose the court VIII trabeated ceUs. 40
Although some small stupas at T akht-i-bahi, such as slfipas P31
and P38 and the edges of P33- P35, were built over trabeated vaults;
it seems that the builders were hesitant to place many heavy masonry
buildings of tlus type on the court VI II terrace.41 Because the space
for further donations had been fully used in the lower sacred are~
first court XX was built, and then court IX was added. T hat so
much effort and money were directed toward expanding the sacre4
area argues for the late part of phase 111 being the period of great•
est use. Tlus is vitally important, as it may indicate the time whert.
the 60 mountain viM.ras dotting the ridges around Takht-i-bili were
fully occupied (see section 7 .3.1 Thareli: MoLmtain Vil!aras).
sculptural finds (sec section 8.2.2 T akht-i-bahf: Sacred Area XX Loose Sculpturo
and Appendix D). Sehrai's chronology places the court of three st•ipas (court XX}
at a time equivalent to our late phase Ill period, attributing the lower court and
rv
the extended west stiifJa court to dlc period loll owing the HUllS 01' phase (Fidaullab:
Schrai, A Guide w Tald!t·i-bahi [Peshawar: ew F'mc Printing Press, 1982], 59).
1
40
.lt seenlS that a level floor existed in fi·ont of court VIll prior to G'lbrication
of the IX sacred area extension.
•• For a discussion of the strength of this vaulting, see Hargreaves, "Excavations
at T akht-i-balti," 35. There is a tedmical limit for the schist trabeated arch, as ulti·
ma.tely the stone ,,~]] shear. The largest known extant example is Thareli structure
C98, which has a 4.3-m-wide trabeated chamber (Mizuno and Higuchi, eds., 1hareli,
plan I 0).
<: In earl>' ASl photos, very high extant walls suggest tha t even the second story
was fabricated i.o stone, Lmlike Ta:xila monasteries, which must have had wooden
upper floors.
l n this light, the th.ree main stilpas present at T akht-i-bahr are nQt
easily explained.
I suggest tha t patronage was the underlying reason for the mul.
tiplication of sacred areas. Using Takht-i-bahr a~ an example, it ill
evident that the Pl stitpa in the lower sacr ed area was the earlieSJ
main stapa and the one that remained t11e center for donative ofierin~
throughout occupation of tlus site. Small stilpos were buiJt here clu~tt
ing phase II, small image shrines and two-celled relic shrines in phast
m, and monumentaJ image shrines in late phase III. In contras\
individual.ly donated structures were not placed in upper stfljJa count
X. As noted above, this independent sacred area, together with tl~
quadranguJar monastery, was probably donated by a single bene.-
factor. These two structures form the backbone of the site, and theit
importance to th e BudcllUst community of T akht-i-bah'i sh ouJd no.t
be underestimated. Our linlited understanding of scuJptural fmc~
from court X suggests that minor patrons were erecting images ther6.
Between each of the original 15 image shrines (S l- Sl 5), small imag$t
niches were later built. Some of the I 10 images fo und in this coulft
by \1\lilcher must have been independen t cornnlissions, but issues df
later scuJptural reuse cloud indicati ons of pan onage in tins instance
(see Appendix D).
Late in phase Ill, another a uxiliary main stii.jJa was built in coUJ)t
XX (the largest stupa at the site). T his extra main st11pa appears te
have been constructed rogetl1er with the four evenJy spaced rnonu,-
mental Buddhas found against the south wall (known from in sitU
feet and attachment socket.~) . Thus, like upper sacred area X, th~
donation o f this auxiliary main stiipa was undertaken in conjunctioo
with the creation of an approp.-iate sacred precinct; this probably
also explains the presence of small stz~pas P37 and P38. Although aJil
of the individualJy donated monumental image shrines of l.ate phase
Ill were placed in the lower sacred ar·ea near the Pl main stflpa,
two heterogeneous sma.Jl stftpas were erected in court XX (P39 and
P40), suggesting that other donors we•·e interested in making offerings
to tills part of me sacred area. Far more evidence for individual
donations appears within the sculptural remains found in court X~;
multiple patrons donated devotionaJ images (see section 8.2.2 Takht-~
bahi: Sacred Area XX: Loose Sculpture and Appendi.'< D). T hus,
from the extant evidence at Takht-i-bahr, it seems mat a particu-
larly wealthy inclividual couJd donate a complete sacred ar ea, as U1
court X , or {ound a new sacred area open for public offe1ings, like
XX. At Thareli and Ranigat the auxiliary sacred areas were filled
with the .kinds of heterogeneous donations that typify a regular dona-
tion pattern; suggesting that an extra main stupa was built and then
multiple individual patrons donated the smaller structures. ln other
words, suitable extra space was created at the site for donati.ons after
the original sacred area had been fiUed.
Thareli (figs. 68, 69, 70, 74) is a large complex o n the extreme
northern edge of the Peshawar basin near the sites of Siloi a nd Jamal
Garhr (see also section 8.4 Thareli: Loose Sculpture; section 8. 7.4
T hareli: In Situ Sculpture; and Appendix D). It was first surveyed
in 1881 by H . Garrick, but it was not excavated until the 1960s,
by a J apanese team under the direction of S. Mizuno and T. Higuchi.'f3
Its surroundings are spectacular, especially area C, which is on a
rocky spur surrounded on three sides by steep cliffs; however, the
spring below area D must have been the deciding factor for loca-
tion of the site. Tl1is complex had two m<~or sacred areas, C and
D, and no fewer than five monastic small sacred areas. While some
of the monastic population lived adjacent to sacred area C in a small
quadrangular monastery, the majority resided in more than 50 moun-
tain uihiims dotting the hills above. About 135 ground-Aoor monastic
cells are extant, suggesting a peak resident population well over 200.
The numismatic dating for T hareli is somewhat in question because
of inconsistencies in the excavation report, but 38 recovered coins
do allow us to draw some broad conclusions.44 Earliest is a single
'3 H. Garrick, "Tour through Beha m1, Ce111nl India, Peshawa r and YusuiSai,"
in Ardweowgical Suroq f![ bulia ( 188 1- 82}; Mizuno and Higuchi, cds., 'Dwreli.
"" The reporting of the numismatic evidence from Thareli is inconsistent. First
it is recorded that 39 coins were found (M.izuno and H iguchi, 16 1), and then later
in the same volume that 37 were found (Ibid., 162). From a st:para te table repon-
ing the combined coin finds from Me.khasanda and Thareli, if we subtract the
Mekhasanda finds that arc accurately reponed, the number of coins from Tha reli
amoun ts tO 38 (£bid., I08}. The photographs document only 26 coins (Ibid., pi.
144}. The excavators seemed to Clu·elully identify the issues of Azes, Kani~ka, and
Huvi5ka, but coins of Vasudeva and those called middle Kushan dating from circa
250- 350 C .E. (Shiva and bull l)1Je and Ardokhosho type) are less well identified
(£bid., 108, 161 - 2). A very genera l Kushano-Sasanian period that a.lso includes
Gazni period coins was identified as spanning dates between 350 and 750 C.E.
(Ibid., 108, l62).
coin of Azes, found near the spring below area D , but this water
source could have been in use before the site was founded. 45 The
founding of upper sacred area C seems to be attributable to ph~
II on the basis of one coin of Virna K adphises, one of Kani~ka t
and one of Huvi~ka, while later Sasanian coins 46 reflect occupat'io(l.
in p hase ill. Sacred area D also was founded during phase IT, af;
indicated by one coin of Kani5ka I, one Huvi~ka coin, and a Hu~ka;
type coin, along with three " K ushano-Sasanian" coinsY T he pre~
ence of late coins and montunental image shrines in both sacrea
areas indicates that Thareli was occupied throughout phase Ill, btd:
sculptural remains show that the main period of donative activit?
began in phase II and had concluded by middle phase ID.
Sevtral strands of evidence suggest that sacred area D (fig. 69)
was the first part of the site to be established; the earliest structLu·~
would include the main stupa and sw-rounding smaU stupas. This p
of the site would have been most reacliJy accessible to the publi
and it is located near rhe spring. In contrast, the configuration o
the earliest remains from sacred area C consists of three small stuptq
(SSt l- SSt3) subsequently enveloped by a later main stiipa (St 1) (fi~
74).i8 Thus, sacred area C originally W'd.S a small monastic sacred
area that was transformed into an auxiliary public sacred area latet
in the site's history, probably after much of the space in sacred are~
D had been exhausted. Originally sacred area C must have servec3
the monastic community who lived in the adjacent quadranguJat
mona~tery and in scattered mountain viM.ras on the ridges above (sec
section 7.3.2 Thareli: Monastic Small Sacred Areas).
Lower sacred area D has a small main slil.pa comparable to od1e:t
phase II main stUpas, like those of Mekhasanda (fig. 67), Takht-i~
ba.hr (fig. 2, P l main stilpa), or Ranigat (fig. 79, east court). Othet
typologically characteristic structures that could date to phase It
include the row of single-celled relic shrines D 3--D 6 and a two-celle4
shrine 025'19 (for a detailed discussion of stii.pa sh.rine D 6 see Chapte:i?
3 and section 4.1.2 The Dharmarajika Complex: Phase II Architecture
and the R el.ic Shrine). Extensive sculptura.l and architectural rema.in§
45
Ibid., 16 1.
46
Ibid., 162- 3.
41
Ibid.
48 Ibid., 1•1·9.
¥-l Ibid., 153- 4, 56, pi. 65.
Copyrighted materi )I
•
PHASE Ill AND IV ARCHITECTURE IN THE PESHAWAR BASIN 193
found in tllis sacred area show that multiple small phase Il stilpas
once stood here (see Appendix D and section 8.4 Thareli: Loose
Sculpture).50 On the basis of patterns of small stupa disu-ibution
observed at otl1er sites and the displacement of later phase Ill image
shrines, it seems reasonable to suggest that at least stilpas St8, St9,
St2, and D26 have phase II foundations.
During the early part of phase Ill, many small to medium-sized
in1age slu·ines were built (D 9- D 19 and others). Most of them had
u·abeated domed roofs like those at T akht-i-bahf, although D ll had
a vaulted ceiling structure.~ ' In image shrine D 8, small holes for
attaching stone sculpture and for mountings for wooden doors are
extant.~2 Large, presumably middle to late phase Ill image shrines
include D 7, D 8, D 20, D37 , and D 38, but only D21 can be con-
sidered monumental .s.1 An unusual characteristic of Thareli sacred
area D is the fact that the south side was never enclosed with image
shrines, even though a portico and a terrace structure provided ample
room.$<1 Only at the nearby site of Sikri is this same spatial organi-
zati.on observed (fig. 90, west side of site facing ravine).
U pper sacred area C is perched on a dramatic promontory; as
noted above, area C was probably founded in phase ll to serve as
a monastic small sacred area (fig. 74). The central structures in this
sacred area are three west-facing strtpas, the largest of which (Stl)
encompasses three small stUpas. This main strtpa was encased on sev-
eral occasions,55 and small stupas and image shrines were added to
the court; however, the relative sequence of construction in this court
is difficult to determine because of the poor state of preservation.
Little more than bases of presumably medium-sized to large image
shrines survive to the west (viii- xii). M ore of the eastern structures
are extant; shrine ii has the C-shaped plan of a single-celled relic
shrine, while, on the basis of scale, shrines i and iii- v seem to have
housed monumental images. Unfortunately, sculptural finds from
this sacred area provide evidence only for the existence of phase II
small stiipas (probably those inside of main stii.pa 1). A few small to
30Sculptures found in 1964 mainly c<une fi-om area D (£bid. , 159, pls. 89- 136).
51
[bid., 153.
sz fbid.
S$ Monumental image shrine D21 had a trabeated domed ,·oof (Ibid., 156).
54 fbid., 153.
ss Ibid. , 149.
Copyrighted material
194 CHAPTER SEVEN
61 Por a technical discussion of trabeated chambers, see Callieii, Soid.u Slmrif, tJte
MonMtery, 69 -ss.
62 Miztmo and Higuchi, eels., 17wreli, 158. Elsewhere in the repon this image is
ro;porwd to be from mowltain uihtira 062, but its appearan ce in the illustrations
and pl ans show that 0 60 is the correct location (Mizuno and Higuchi, eds., 17weli.,
160, pi. 78, nos. I, 4; plan 30).
65 Mizuno and H iguchi, eels., 17wreli., 148, pi. 15, nos. 2- 4, and plan 7.
sacred areas include 061 , 056, A5, Cl09, and CI06 (figs. 68, 75).64
Clusters of mountain vihiiras surrounding monastic small sacred areas
also define the distribution of residences at J amal Garhr (fig. 61),
and the same pattern is evident at Takht-i-bahf (fig. 44).
T he largest of the Thareli monastic small sac•·ed areas, C 106 (fig.
75), can be dated approximately, and a relative sequence of con-
struction is recognizable. The area was founded with the construO.
tion of small stUpa i, then a pavement was laid down and small stilpQs
ii-vi were built on it, and finally small stupas vii and viii were added
beyond the paving.65 The finds from this court are fully documented,G6
and there is considerable phase II sculptural evidence to show that
the area was founded in phase II and that many of the added sta-
pas were also built at tllis time. The last period of constructioP
involved the addition of image shrines 1- 7, wllich occurred in phase
m, as indicated by structural typologies and the presence of dev(i)-
tional scu1ptures. 67 It is certain that the surviving image shrines (th:e
northern edge of the area has been lost to erosion) were construct®
after all eight of the small stUpas were in place, because the shrines
neatly encircle the last of the small strJ.pas to be added (vii and vii~.
Thus, from the evidence preserved in this cow't, we know that monas-
tic small sacred areas were being constructed in phase Il. This has
broad implications for dating monastic small sacred areas through-
out the Peshawar basin as well as the mountain viltiiras they served.
It appears that sacred area C (fig. 74) wa~ also originally founded
in phase II as a monastic small sacred area. O riginally the court
consisted of three small stilpas, SSt l- SSt3; these structures were
encased in the core of the later main stupa Stl. Thus, it appears
that what started as a monastic small sacred area was later coo.-
verted into a public sacred area as space ran out in sacred area El.
The numismatic and sculptw·aJ evidence allows for a fairly secute
dating of sacred area C (see above).
The monastic small sacred areas lound at Thareb (figs. 68, 7:1),
J amal Garhi (fig. 61), and Takht-i-bahi (fig. 44) are all similar in
layout 2md apparent fi.mction. At the juncture of multiple mountain
vilwra terraces, a small monastic sacred area was established that
6ol Ibid., 14·6, 150- 5 1' 160, pis. 30, 33, 37- 40, plMS 13, l •L
~ Ibid., ISO.
66 Ibid., pis. 137- 4·0. See also Appendix D.
61
Ibid., pi. 137.
always had a small stupa at its core. On the basis of evidence from
Thareli, this appears to be the phase II form of the monastic small
sacred area (a feature that is consistent with phase II stfijJa shrines
found in quaru·anguJar monasteries; see section 7.3.2 Thardi: Monastic
Small Sacred Area~) . H owever, at all of these sites during phase Ill,
image shrines were commonly added to the small stftpa court~. In
these instances a low wall defines the perimeter of the monastic small
sacred area, where slwines were not donated, a feature that is espe-
cially clear in Jamal Garhi monastic small sacred area 9 (fig. 61 ).
Thus, d1e phase HI monastic small sacred area reproduced, on a
smaller scale, the plan of larger public sacred areas.
68
A total of 47 mona.scic cells are extant, but stairways suggest that most of the
mountain viluiras were at least two-storied, so there would have been considerably
more cells.
..., Erringt.on, 1987.
[fig. 81]; see below). contained one in situ coin of Vasudeva, and
seven more coins of this ruler are listed in Cunningham's 1873
report. 7° Four sih-er coins of K.idara £1 (circa 4th cenrury C.E.) from
the vicinity of the main stiipa complex were recovered in the 1920 21
excavations. 71 Thus, the numismatic e\-idence confirms a phase m
occupation of J amal Garhi, which is readily apparent from the archi-
tectural remains.
The compact structure of J amal Garhf's phase HI sacred area is
unusual, especially given the fact that ample space was available. 1t
is divided into three adjacent strlpa courts (l, 3, +), one providing
access to the next. The phase m devotee would have passed through
the large coun 7, where assembly halls I 0 and 11 stood, to reach
the lowest of the sacred areas (4). Tllis lowest court has no apparr
em focus; it appears that tJus group of heterogeneous small sfftjJIM
and image shrines was the overflow from the more significant uppC1r
sacred areas (3 and I).
Court 3 was approached by a flight of stairs, which seen:ts to havji:
faced image shrine 3a for dramatic effect. It is dear that this couct
was designed to accommodate circumambulation, a process th<\t
would ultimately ha,·e brought the devotee to the ba~c of the staio;
leading up to tht: main stiipa in coun I . The interiot· of court 3 is
filled to capacity with smaJJ slli.pas and image shrines; some of these
shrines were built on the remains of earlier stiipas. 12 As considerab!J:
phase 11 culpture was found in this court/3 it seems clear that some
of the small slupm were built at this time. ome single-cellcd relic
shrines in this court probably also date to this period. Court 3 is
surrounded by image shrines, a pattern sinlllar to that of Tharcli
court D (fig. 69) or Takht-i-bahi lower sacred area IV- V (fig. 2). t\o
single focal monument is present, so this court was built to hold
donarive offerings, but not as an alternative focus for devotion (see
1'' Errington hlL' fulty examined the J amal Garhr numismatic evidence (Ening10n,
1987, 2 19, 234 5; 2000, 197). See also Cunningham, 1872- 73, Appendix A.
11 Er1ingt011, I987, 23.'i; 1-'1. Hargrc·avcs, "Appendix V: List of Antiquities Rccow rad
During Op("-ations atjamltlg<lrhi, 1920 2 1," Arclta8ological SrmNy qf India Frontier Cirtb
Report I 920- 1, Pcshawar: Caxton Printing Wor·ks, I92 I: 23, 28.
'~ When T visit ed this site in I 99 I and t993, [ had the oppot1"unity to exami rlc
tl1is block of structures. When I returned in 1994, 1he site had been reston:d, ~~
process that grca lly rnndilicd t.his gruup of donative monw·nents and obscw-cd tltc
imemal configu ra tiou.
" Hargreaws, "Appendix V."
PliASE lll AN D IV ARCHIT ECTUR.E. IN THE PESHAWAR BA.SIN 199
11
Cunningham, ':Ja mal. Ga.rhi," 48; Eni ngcon, "The Western 'Discovety of the
Art of Gandh ara," 230.
75
AS FC no. 173. Schist atlantes were probably aL~o part of the main sllipa's
embellishments (Errington, "The Western Discovery of the Art of Gandhara," 228).
76
Ib id., 246- 51.
17
In I 994 J auempt~d to map and sott out the successive laye rs upon which
the main .rllipa is built, but I was unable to trace a clear patte rn .
78 ishikawa et al., Candhara 2, 47, 52, llS- 22; Nishikawa, Odani, and Namba,
Preliminary R«port on Gmuiltam Buddhist Sites, 90- 4. Identifiable finds include two loose
coins of Azes (either Azes I, 5 7- 10 B.C.E. or Azes IT, 6- 17 C.E.) found a short
distance from the eastern sacred area in trench ll (Enington, "Numismatic Evidence
for Dating 1.he Buddhist Remains of Candhara," 194; Nishikawa et al., Garullwra 2,
52, 109). In direct association "~th the core of the main stilpa., St lOL, eight coi.os
of Virna Kadphises were found, although t.he precise location spot is unclear
(Nishikawa et al., Garu/hara 2, 85, 89}. ln conj tmction with the encasement of main
stilpa. StlO I, several pavement stones were fotmd which had holes bored in them
for the donation of coins. Fourteen of the holes contained in situ coins, including
one Azes II coin, one Kttit•la Kadpbises coin, three Huvi~ka coins, and eight coins
of Vasudeva (Nishikawa, Odaoi, and Namba, Prelimi11ary Rcpm~ on Gmuihara Buddhist
Sites, 92-93). Coin finds from other parts of the site include one of Azes, one of
Soter Megas, one Vim a Kadphises, ten of Kani~, two H uvi~ka, and one Huvi~a
imitation, four of Vasudeva, two of later Ardokhosho type, two la~.er Vasudeva imi-
tations, six KtL~hano-Sasanian issues, and one Ephthalite. Because of inconsistencies
in the excavation reports, tJ1ese numbers probably should be w1derstood in terms
of sketching a broad chronology. See also Enington, "Numismatic Evidence for
Dating the Buddhist Remains of Gandhara," 213. A Kharo~~hi inscription men-
tioning Vasudeva was also found at Ranigat (Nishikawa, Odani, and Namba,
Pl'etiminal)>Report 011 Gm1dlwra Buddhist Sites, 95}.
1'i The excavators noted that some of d1e stucco stii{itlS must have originaUy been
embeUished "~th schist re(jefs. See Nishikawa, Odani, and Namba, Preliminary Report
tm Gamllmm Buddhi!t Siles, 85.
80
Ni:lttikawa, Ranigal, pis. 26-31.
8
' Ibid., pi. 26, no. 2.
82 See above footnote listing coin fmds from Ranigat.
83
Nish.ikawa, Oclani, <utd 1amba, Preli:minao• Repa.rt on G1mdham Buddltist S1:tes,
92- 94.
84
Ibid., pis. 53- 87.
*-' See especia.l.ly Ibid. , pi. 53, no. 2.
86 Ibid. , pi. 71, no. 4.
87
h is possible that is was built to acconunodate Jay or monas tic dead; see
Schopen, " Burial 'Ad Sanctos' and the Physical Presence of the Buddha in Early
Tnctian Buddhism: A Study in the Archaeology of Religions."
88 Errinf:,"lOn, "Nwnismatic Evidence for Dating the Buddl-.ist Remains of Gandhara,"
the end of phase Ill. The southwest sacred area was built on a mas-
sive platform that had a large trabeated tunnel running through it.89
T hjs foundation platform <md entrance turmel are comparable to
sacred area XX at T a.kht-i-bahr (figs. 2 , 45), which can securely be
dated to the latter part of phase J1I (see section 7.2.4 Takht-i-bahr:
Late Phase III Construction). The tunnel appears to have been <m
impo1·tant entranceway into the Ranigat complex; perhaps it !i.mc-
tioned to better fortify the site.
The southwest sacred area is basically composed of two stmctures,
a medium-sized st71pa (St30 l) (addressed in section 6.5 T he Phase Ill
Small Stilpa) and the biggest image shrine (R30 1) thus far uncovered
i.n all of the Peshawar basin . T he sh1ine is 10.3 m, wide and 9.1 m
deep and has 2.l-m-d1.ick walls; while there is no way to gauge how
high it stood originally, remains of walls in me southwest corner are
5.8-m-high. 90 Such large image shrines are found at only t\v6 other
sites. At Abba Sahib Cl1ina in Swat, there is a comparable massive
image shrine (fig. 11 0, shrine D); socket holes on the rear wall indi-
cate d1at it original1)' contained a monumemal stucco u·iad. 91 This
Swat shrine was supported by an 8.2-m-square base and has ex-
tant walls 9.5-m-high (fig. 111 ).92 At me Dharmarajika complex in
Taxila, image shrines Nl8, P2, and P4 are all quite large; the P4
base is 8.5 by 6. 1 m (figs. l , 19). T he presence of this massive image
shrine at Ranigat can also be loosely compared to the late colossal
Buddha images being fabricated in Afghanistan at sites like Bami-
yan (fig. 115), but iliis Afghan example seems to be an even later
phenomenon.
The configuration of me southwest court is completely different
from mat of the east or west sacred areas. Side by side, facing the
place of the devotee, are the massive image shrine R30 l and stftpa
St30 l . It is clear that these structures were not circumambulated or
entered, as the entrance to me trabeated passageway would have
blocked access. A related configuration of monuments can be seen
at the Swat site of Abba Sahib China (fig. 11 0). There, monumen-
tal image slwine D (fig. I 11) is in a row with two large sttipas B and
93 Various names have been !:,>iven to these foreign invaders, induding Ephthalitc:.t
Wh.ite Hw'Ls, and Hw:oas. See Kuwayama, "The Hephthalites in Tokbaristan an(!
Ga.ndhara; Part 1: Gandhara"; "The Hephthalites in Tokharistan and GandJ1ara<;
Part 11: Tokharistan"; Rahman, 77ze Last Two P,nastW I![ 1/ze $tutis: A11 ArUI/)'.ris I![
77zeir Hirt()ry, ArcluzcologF, Orinage and Palaeography .
94 Sung-yun, Si-1~u.-Ki. Buddhist Rf!C()rds I![ lfle W<t~/81?1 World, trans. S. Beat (Londoo:
Triibncr & Co., 1906), c.
~ Hsua.n-tsang, Si-Tu-JG. Buddhist R.ecords I![ 11,. W«SI.ern World, 98.
96 MarshaU, Taxila, i 6.
'» Ibid.
'l6 There is kss evidence of a decline in patronage in Sw·at: at the site of Ni.mogr.un
(fig. l 09), tht: sacred area ringed by image shrines was expanded to the west (the
F3 area), but as was the case with the sites from the Peshawar basin and Swat,
this new area was never fully enclosed nor filled with strtpas (field observ.u:ions,
1993-94).
Thareli (fig. 70), and probably Sikri (fig. 90), only a few late phase
III monumental image shrines were built, and they were squeezed
into sacred areas constructed in the early and middle parts or phase
Ill; no attempt was made to expand the sacred a reas to accommo•
date these structures. Supporting this contention is the fact that sculp:-
rures with sophisti.cated late iconography were not found at thest
sites (see Appendix D). All of this evidence indicat es a decline i.o
patronage late in phase ill; perhaps these sites are some of Xuanzang~
" I ,000 smighii:riimos, which are deserted and in ruins."
Although the exact relationship between the Buddhist communi•
ties of the Peshawar basin and the new H un dynasty is not entirely
clear, there is considerable evidence to suggest that Buddhism co~
tinued under H un rule, though perhaps this incursion marked the
end of maj or prosperity . K uwayarna was the first to challenge tl1c
old assumption that BuddhistS were persecuted under the Huns; he
was able to find textual evidence to show that Chinese Buddhist pih
grirns continued to visit Gandha.ran sites in the Peshawar basin into
the early 6th centur)' C.E.; many made this trip to visit the shrine
of the Buddha's alms bowl. 99 Errington also has been able to shed
light on this issue, using numismatic data that provide compelling
evidence for continued earl)' phase rv occupation. Late Sassanian
and Hun coins from the mid-5th century to the early 6th century
were found in Taxila at the Dha.rmarajika complex, K unala, Gir~
Bhamala, and L'ilchak, in IJ1e Peshawar basin at Takht-i-bahi, Ranit,rat:,
Sahd-Bahlol B and F, J amal Garhr, and Shah-j1-kr-c;lheri, and in Swfll
at Butkara I and Shnaisha. 100 FoUowing this dating there is a numi~
matic gap that spans the 6th centwy to the beginning of the 7th
century C.E., corresponding approximate!)' to the time when the
H uns ruled Gandhara. Errington has suggested that coins were not
being minted by these rulers, who instead relied on earlier coins
already in circulation, and thus the lack of coins from Buddhist cen;.
ters ccumot be seen as indicating their abandonment. 10 1 This sup--
position is supported by the facts that the numismatic gap is not
99
These ideas first appeared in Kuwayarna, "The Buddha's Bowl in Gandhar~
and Relevant Problems," 968. See also Kuwaya ma, ·ne Mont. Swpa t1i Sho.h11-kf-l/hcfi!.
96- 7; and "Route Cha nges and the Decl ine of Gandhara."
100
Errington, "Numisma tic .E:videnc~ t:o r Dating the Buddhist Remains of Gan•
dhara," 21 1- 13.
101 fb.id., 20 I.
restricted to Buddhist sites and that coins postdating the gap were
found in relic deposits in Taxila at tJ1e Dh armar~ika complex and
in the nearby Ma.Ifikyala stii.pa. 102 T hese late coins found in relic
deposits leave no doubt that these isolated sites were still receiving
patTonage. Clearly, Buddhism was in decline, but this process appears
to have occurred gradually.
There is slill the issue of why the Chinese pilgrims Sung-yun and
Xuanzang described ilie Peshawar Buddhist sites as being in ruin.
An eariliquake is a tempting explanation, given the seismic instabil-
ity of the region, but there is no documentation to support this
hypothesis. Faccenna and Marshall suggested iliat earthquakes cre-
ated early phase I and II horizons of destntction at Tax.ila and
Butkara I, respectively. 103 Earthquakes would account for the facts
that essential.ly no phase II sculptu re has been found in situ and that
so many early stftfHIS were refurbished during phase m. A circa 6th
century eartJ1quake in Swat has been suggested by CaUieti to accoun t
for an observed level of destruction. 11>4 An early phase IV earthquake
might e>q,la.in why so many arguably late phase Ill sculptures m·e
found in reused contexts at sites like Sahrr-BahJol and T akht-i-bahl
(see Appendix C).
Our understanding of phase IV Buddhist a rchaeology is still in its
inlancy. Although there is compelJing evidence to suggest p hase IV
activity at the Dharmarajika complex, Bham aJa, Kunala, BhaUar
tope, Matfikyala, Shah-jl-ki-<;lherr, Sahri-BahJ61, Ranigat, Butkara I,
Abba Sahib China, Tokm· Dara, and Shnaisha, this question is not
easily resolved without fw-tJ1er research. However, som e late archi-
tectural features at these sites deserve special note.
The Taxila site of Bhamala has characteristics that appear to post-
date phase Ill , but on tl1e basis of numismatic evidence it cannot
be uncontrovertibly dated to later tll<m ilie end of phase Ill or the
vety beginning of phase IV (fig. 42). O n the basis of 11 9 coins found
in tJ1e core of tJ1e main smpa' 05 and 160 small copper coins in small
---·--
12
" Ibid. , 21 1- 12.
103
Ibid. , 202; Facce1ma, Bulkara I, part I, 134- 5; M;u'Shall, 7iu1/a, 118.
11
" Oral presentation b)' P. Callieri at the South Asian Archaeology Conference,
Paris, 2000.
•M At a depth of 3.35 m , six coins were fowtd, one or which Marshal I idemified
as being Sasanian. At a depth of 4.4 m, 113 coins (Marshal!, Taxila, 393) were
found that appeared to be middle 4th century late Kushan issues (ErTington, "Numis-
matic Evidence for Dating th e Buddhist Remains of Ga.ndhara," 212).
stftpa A 15, 106 it would seem that the site was founded in the middle
of the 4th century, or late in phase Ill. However 21 Alchon Hun
coins (circa 500 C.E.) found p1imarily in d1e monastery suggest a
somewhat later date of occupation. 107 All of these coins could have
been deposited long after their production, if we consider the numis--
matic gap noted above, although the fact that 93% of them were
minted before the end of phase Ill cannot be ignored.
The strongest reason to place Bhamala in phase IV is the fact
that its main stupa is cruciform in plan. As noted by Kuwayama and
Fitzsimmons, the Bhamala cruciform stii.pa can be compared to 7dl
to 8th century cruciform stUpas in Kashmir, Afghanistan, and othe:r
parts of Central Asia. 108 Certainly, the plan of tile Bhamala mail!
stiipa is unique in Taxila, ald10ugh cruciform stiipas were constructed.
in tile Peshawar basin at Shah-jf-ki-<;lherf (fig. 83)109 and Sahri-BahJ&
Site G (fig. 60), 110 sites that have compelling evidence to sugge~
phase IV construction.
The Bh<unala monastery also had an tmusual plan; most readily
noticed is the complete lack of internal image or relic shrines 111 and
two large chrunbers that were built in its courtyard. The fact that
this monaste1y was fabricated of double semi-ashhu· rnasoruy rnigllt
indicate a phase IV date.' 12 Thus, for Bhan1ala there is confticti~
evidence, though it seems to be phase IV in elate.
106 E1-rington, "Numi smatic Evidence fo r Dating the Buddhist Rem a ins of
Gandhara," 2 12; Marshall, Ta:rila, 394.
107 ln addition to the coins found in the main stftpa a nd in small .rtilpa A 15, 32
coin s were recovered at .B hamala : one gold coin of " Bachara.na," one of Huvi~a,
five of Vasudeva, one lndo-Sa.<;anian, and 21 silver Alchon Hwt coins (20 from tbe
doo<>V<t)' of monastic ceU 13 and one from ceU 8) (Marsha.U, T tJJ.ila, 396).
108 Kuwayama, Tlll Main Stupa at ShiiJz-ji-J..'i -t/herf, 49- 50, 93; Fitzsunmons, Swpa
119
In situ monwnental feet were found in huge axial njches set into the base
of the main stripa- a fcattu'C that certainly cannot be dated to before the end of
phase m.
114
Fussman, Monuments Bouddlriques dt la l?igifm dt Caboul.
Copyrighted materi •
•
CHAPTER EIGHT
The Peshawar basin evidence for the diso·ibution and use of devo-
tional icons is quite different from that in Taxila. Because schist was
much more widely used for sculpting devo tio nal im ages in the
Peshawar basin, a different body of data is available for study (for
counts of sculpture recovered at many of the Peshawar sites, see
Appendix D). In any case, it appears that much of the schist imagery
from Taxila was irnpo•ted from the Peshawar basin and to a lesser
extent from Swat. Because schist is more durable than stucco, a
greater proportion of the total sculptural production in the Peshawar
basin survives in the archaeological record. The durability of schist
sculpture also meant that it was available for reuse in ancient times,
while the more friable stucco sculpture was only ra rely recontextu-
alized (see Appenclix C). Thus, it is sometimes possible to recon-
struct the assemblage of schist sculpture that stood in the Peshawar
basin sacred m·eas, even though later reuse obscures the original
placement. At the sites of Loriyan Tang-ai, Takht-i-bahr, Sa1ui-Bahl61,
T hareli, and .M ekhasanda, l9th and 20th century photographic doc-
umentation allows us to discuss the total recovered schist sculpture.
Crucial to our understanding of sculptural use in the sacred areas
are the rare instances ...vhere images were actually recovered in situ.
In the case of Sik1i, a.lthough only an incomplete record of sculp-
tural finds is documented, there is a plan indicating th e in situ loca-
tions of some iconic imagery (fig. 90). Further in situ evidence comes
from a handful of oth er sites, nota bly Thareli, 'M ekhasanda, and
probably Sahri-Bahl6l A. At Takht-i-bahi, three sections, the last
area excavated, were individually documented, providing an invalu-
able set of evidence (courts XX, XIV, and two-celled shrine T 4 in
court XXIII) (fig. 2).
T he loose sculpture recovered at a given site provides hard evi-
dence that can be interpreted in a range of ways. Most immediately
it tells us a great deal abo ut late practices of sculptural reuse (see
Appendix C). As a point of departure, the loose sculpture was divided
into readily recognizable categories. O ften specific categories of images
Copyrighted material
212 Cti.APTER EIGHT
Copyrighted material
PHASE Ill SCULPTURE IN THE PES HAWAR BASIN 213
upper story; a long projecting foundation for the front stairs is also
visible. To my knowledge, the only solidly comparable structure is
the Kunala main stzi,pa in T axila, which was constructed in phase
Ul semi-ashlar masonry (fig. 41).3 Another related structure is Ranigat
stiijJa St30 I (11g. 78), which is much smaller but does have a similar
plan and a stepped base, although tlris stii.pa dates to the end of
phase III or phase IV (see section 7.5 Rani gat: Phase II to Pha~e
rv Architecture).
In contrast, a great deal is known about the schist sculptural
imagery found at this site, which had probably remained undisturbed
until its excavation in the 1890s. Its remote location appears to have
protected it from 19th century collectors, an assumption borne out
by the considerable quantities of sculpture that were recovered (see
Appendix D). The 39 phase II narrative panels appearing in Caddy's
photographs can be subdivided into five setS (multiple panels from
a given stupa) (fig. 85 shovvs some of these paneL~) . A good example
of one of these sets is the panels on the base of the reconstmcted
Loriyan T angai stii.pa now in the Indian Musewn, Calcutta (fig. 86).4
It appears that these narrative sets originally adorned four small
stii.pas. T lus seems in agreement with the fact that four small schist
stiipa domes appear in these photos; i.e., the number of early stii.pas
can be estimated. Surprisingly, this site, which produced one of the
most intact bodies of phase II narrative sculpture in the Peshawa.r
basin, seems to have had only about four phase Il stfljJas.
One of the narrative sets, consisting of two reliefs, most likely were
attached to the harmika of the main stii.pa (fig. 87).s The 45-cm-ta.ll
panels contain depictions of the birth and great departure, common
su~jects for four-faced hannikiis. Such groups typically show various
combinations of the birth, great departure, attack of Mara, first ser-
mon, a trimtna, parinirvti.~w, cremation, and display of the relics (another
example where two ham1ikii reliefs survive is shown in fig. 95 B, from
lower Nathou).
3 Marshall, Taxikt, 348-50. l'l tzsimmons (Stupa Designs at. Taxi/a, 73- 4, 107) grouped
th is stlifia wil11 d1at at the siu: of Balla.r, and gave them a much later date.
• .In addition to the eight reliefs on the base of the reconstructed Loriyan Tangai
stiipa, there are sets with two, two, five, three, and 13 reliels, for a total of 32 of
the 39 narrative pan els found. The set of 13 is illustrated in Behrendt, " amuJVe
Sequences in rhe Buddhist Reliefs from Gandhara."
5 ASJM 1053- 4.
Copyrighted material
214 CHAPTER .EICI-rr
Two large multi-piece false gables ar1d one small single-piece false
gable (fig. 85, upper right) are also preserved in this corpus. The
largest of these is over 2 m call, a scale that would seem suitable
only for a main stupa. Although it is impossible to know if these
reliefs were attached to the late phase rrr
encasement of the main
stftfJa, photographed by Caddy, reuse patterns from Swat suggest that
important panels on main sta.pas were sometimes retained .into latet
periods. 6 It would be a mistake to assume that either the false gable
or harmikii reliefs were restricted to phase II, because mudrtis cha.r--
acteristic of later phase Ill production (see Appendix B), are sho"vn
in some hannikii reliefs. T he false gable reliefs a.lso appear to have
had a long life span; they were popular during phase IT, but some
contain motifs Like the lotus throne, suggesting phase lll productima
(see Appendix B). The false gable is sometimes integrated in Sravasli
composite panels (fig. 127), suggesting that this form remained
significant to the latest Buddhist communities. Reciprocally, its pres.
ence in the Sravasti composite panels suggests that this latter relief
type might have originally embellished stiipa domes. In the case of
the Loriyan T arigai false gable and hmmika reliefs, no late m:udriis or
motifs like the lotus throne are present.
Phase Ill devotional icons from Loriyan T arigai survive in con~
siderable numbers. Two group photos are especially helpful, as they
provide a sense of the overall body of devotional sculpture found at
tlus site, as well as the relative scale of the images (figs. 88, 89).
Many of these schist images are smaU, certainly too small to waf'.
rant the construction of an image shrine. D onors commissioning
these small devotional icons at Loriyan T arigai clearly preferreci
seated images and commissioned Buddhas n.vice as much as bod..
hisattvas.7 The small icons c<umot be easily dated, although it is clear
that the earlier image shrines tended to be small. Economic con-
siderations and lack of space would have encouraged the production
of small images, regardless of their date. In general, it seems rea·
sonable to place this group in phase III, perhaps with greater pro-
duction in the earlier part.
6
The reuse of Lhe QJ panel of.' 1.he GSt4/ 1 encasement can be roughly dated.
to phase Ill on the bru;is of nurn.ismatic evidence. See Faccenna, Bulkara I, vol. 3t
no. I, 1 LS- 16; Faccetula, Cobl, and Khan, "A Report on the Recent Discovery of
a Deposit of Coins in the Sacred Area of Butkara l (Swat, Pakistan)," l J 3.
7
The diminutive Buddha images include 21 seated and si.x standing, while the
small bocUusattvas consist of tune seated and two standing im ages.
Copyrighted materidl
PHASE Ill SCULP'JUR!'. IN THE PESHAWAR 8AS£N 215
Copyrighted material
216 CHAPTER EICHT
8.2.1 Taklzt-i-biihi: The Upper and Lower Sacred Areas: Loose Sculpture
Various parties collected sculpture from this site th1·oughout the 1 9~
century. In 1850, Bellew reported that treasure hunters had d~
aged the site, and Wilcher reiterated this in 1871.8 Wilcher, w~'
conducted the. first systemati~ exca~ation of the upper stii.!a c.ourt 1-
rep orted fincling 110 devononal 1cons, though none m situ (sot
Appendix D).9 It seems likely that the disrupted placement or imag<is
that he reported is a reflection of later reuse, the sculpture being
moved closer to this ~port:mt stupa. vVlu~e. it is tempting suggqt :o
that these 110 devot10nal 1cons were ongmally housed m the 1~
meclium-sized image shrines and 16 niches that encircle this cour9,
it seems possible that sculptures from other parts of the site wer~
taken there during this late period of reuse. Similar patterns of pr~
sumably phase IV image concentration characterize the distribution
of imagery at the Sahri-Bahl61 sites (fig. 58) (see below and Appendixe.s
C and D).
At the beginning of the 20th century, the Archaeological Surver
of India conducte.d a series of excavations at T~kht-i-bahf in an e~o1
to conserve the stte and to recover the extet1Slve sculptural remam~
T he excavation reports that resulted from this initial work ar~
superficial, but e>.."tensive photo documentation produced 367 archiveeJ,
photographs. 10 On this basi.s, it is possible to reconstruct a broa9
scheme for finds in the lower main sacred area (courts IV and .;
the greatest number of sculptures were found in this section}11 (fig~
2, 46) (see Appendix D). .
More than 42 significant phase Il sculptural panels were found iQ.
the IV- V lower sacred area. T he presence of this phase II scul~
tun: in tltis part or the sacred area, which we know was one of th~
oldest parts of the site, indicates that a group of e<u·ly small stupa;r
once stood here. Especially significant in terms of forming an accur
rate impression of the number of phase 11 stupas are the remains df
8
.Bellcw, A GfJIIt'l'al Report 011 tire Tusuj~ir, 92; Wilcher, "Exploration of d1e Buddh.i$
Ruins a.t Takht-i-ba.i," 435.
9 Wilcher, "&:ploration ol' d1c Buddhist Ruin s at Takht-i-bai," 435, 37.
10
Ardweological Surory if India Photographs: Jf1de.~ of LocatifJn, l11din Qf!ice Collections fit
the British Lihrary (London: Emmett Publishing, 1993).
11
A meal of 472 sculptures were found in court V, including 1J1e corridor betwecf~
court X and the quadrangular rnooasrery (Spooner, "Excavations at Takht-i-biihi,"
J 35).
Copyrighted materi I
PHASE lli SCULPTURE IN THE PESHAWAR BASIN 217
12
Errington, '"TI•e "Vestern Discovery of the Art of Gandhara," 113.
Copyrighted material
21.8 CHAPTER ElGH!'
Copyrighted material
•
220 CHAPTER EIGHT
and worshipper plaques. The reasons for this concentration are not
readily apparent, but the plaque reliefs are all in good conctition, a
clear i.ndic.:<tion that they were specifically manufactured for deposi-
tion in this relic shrine and thus are not reused images. The presence
of these late plaques is especially interesting because the only othet
known context for this type of i.n1agery is in the Jaulian monastery
(fig. 30) (see section 6.6.1 Phase Ill Use of Personal D evotional
Images in Monasteries). The body of sculpture in n·vo-celled sl1rine
T 4 gives us a sense of the kinds of donations that were made t()
reljc shrines in the late part of phase Ill, and it expands ow· knowl•
edge of the deposition of images in reused contexts (see Appendixes
C and D).
8. 2. 4 Taklzt-i-biilli Small Sacred Area XIV Loose and !11 Situ Sculf;tu:re
The southeast small sacred area XIV (fig. 2) is particularly impor~
rant because it was found intact and was photographed. 17 It is per..
haps the only sacred area in all of the Peshawar basin where wo
can determine reasonably accurately the original locations of the
recovered sculptures.
Phase II sculptural finds from this court all belong to a single
small stftpa (M): one false gable, a narrative set composed of three
panels showing six scenes, four other pieces of this stiipa's sculptural
embellishment, and nvo clwtt:riivali. Because all of these stftpa remains
are related to each other and only a single stzipa was built here, it
seems reasonable to conclude that this phase 11 stupa stood through-
out the occupation of the site. It appears to have been the founda-
tion structure around which the rest of this small sac1·ed area was
Ol'gallized.
Within tllis small sacred area there are ll extant small image
shrines, and platform K probably supported two or three more. The
devotional icons include five mectium-sized and five smal.l ones, as
well as ten miscellaneous stucco heads (see Appendix D); thus, there
are almost equal numbers of image shrines and devotional iconic
images. In this instance it seems quite possible that each of these
image shrines contained a single devotional icon. Supporting this
contention a•-e two detached i.n1age bases found in situ in in1age
shrines C and D (figs. 51, 52). One shows a row of five seated
Buddhas (two with lotus bases) Ranked by worshippers, and the other
shows a row of standing Buddhas, presumably past Buddhas. These
reliefs are positioned halfway back in the shrine, forming the front
face of a step; the devotional icon presumably stood on top of this
elevated section and was affixed to rhe rear wall. As such steps are
found in many of the shrines at Takht-i-baru and at many other
sites in Greater Gandhara, this category of sculpture can be given
an architectural context. 18
This small sacred area can be dated to the early to middle part
of phase Ill for several reasons. A row of in situ stucco Buddhas
along the bases of the image shrines in the southwest corner can be
directly compared to what is seen at Thareli (fig. 73) and Mekhasanda.
Of all these devotional icons, only a single Buddha i in dha1macakra
m.udrii. Further, most of the image shrines are very small, vvith the
exception of shrine L, which is medium-sized. Although srna.LJ shrines
were certainly built in later periods, the consistent sma.LJ scale of the
image shrines in this court suggests an early date of production, and
there is no indication that monumenta.l images were part of dus
assemblage.
18
J.mage bases a re discussed in G iovanni Vcrardi and Istittn.o Orientate di apoli,
Honw a11d Other Fire Rituals i11 Cmuilwra (Naples: btituto Univcrsitmio 0 1i entale, I994).
Separate in situ schist image bases were fow1d at Shotorak in Afghanistan in shrines
DS and F6 (see J Mcunie, Slwtorak, fif,IS. J l, 28).
19
E. Errington, "In Search of PaJu.sha, a City of the Central Gandhara Plain,"
Bulletin of lite 1L-ia Institute 7 ( 1993).
20 F. Tb:.ot. "Sahri-Hahlol Pan r\ '),'' in St.tuJh. Asum Archom~· 1993, ed. A. Parpola
and P. Koskilaillio t H cl,inki: 1994); F. TbSOt, "lltc Site of Sahri-Bahlol in Gandh:t.ra
Part 111 )," in South AJurn Ardwo/ogr 1987, ed. :\1. Taddei Rome: ls~IEO. 1990);
·'The . itc of Sahri-Bahlol in Candhara, Pakistan: furtlter Investigations.'' in Soulh
Asian Ardwolbf!J' 1985, ed. K. Ftifeh and P. Sorrensen London. Ri\'erdalc: 1989: ' 'The
Site of ahri-Bah161 in Gandhara," in Soldh Astan Archoeo/Qg'; 1983, ed. J. Schotsman.s
and ~I. T addei 1aples: l stituto Univcrsitario Orientale, 1985 .
11
ASlfC 623 727; D. Spooner, Ardu:ol~cal ~of India Frontin Circle.for 1909 /0
Pcshawar: Gm·et1tment Press, Nonhw<:st Frontier Province, 1910; Spooner, '·Excava-
tions at Sahri-Hahlol 1909 I 0\."
11 Spooncr, " Excavations at Sahri-Bahl61
1906-07),'' 47- 8.
n Of the 67 coin~ found at Sahri-Bahlol n, only a few could be idemificd tu
the time of tht·ir rx<·avation , hut with the exception of a silver issue of Sotcr Mcg-<lS,
the coins appt•ar to lw latt- Kusltan , Sasan ian, and ]{jdi\:ra-H un , supporting a l:tlc
plutsc Il l dating for the site (Errington , "N\tm isma tic Evidence for Dating tile
Buddhist Remains of Gandhara ," 2 13; Spooner, "Excavations at Salu·i-Duhlt\1
( 1906 07)," 50).
1' C. Bautzc-Picron, "'l1tc
imbus in India up to the Gupta Period ," Silk Rolui
ll rt and Arc/weo/Qgy I (1990); K . Hehrcndt, "A Dateable Halo Type from Gandham"
(M .A. thesis. Univmity of California, 199 1).
n ASlFC 1079 1185; Stein, llrdWJiogicoJ. Sumry of India Frrmli" Cirdt for 191J 12;
Stein, " Excavations at Sahri-B:thlol."
PHASE Ill SCULPTUR E (N THE PESHAWAR BASh'l 223
26 ASIFC l l86- 1254; Stein, Ardzeowgical Sumy of India Frontier Circk fqr J9 J 1- 12;
Stein, "Excavations a t Sahri-B~,hJol."
T hareli was excavated in the 1960s (fig. 68); the photo docwnenta-
tion of recovered sculpture is exhaustive.27 Thus, much more is kno"'tl
about the body of imagery recovered at this site than was the ca~'e
with sites excavated by the Archaeological Survey of India at th;e
beginning of the century. T he report provides documentation of all
of the sculptural and architectural material, even quite ITagrnentaqy
objects. T he finds from T hareli are remark.:tbly different from otht.rr
Peshawa.r ba~in sites because they are executed primarily in schis~
the excavators suggested a 4: l schist to stucco sculpture ratio (s~
Appendix. D).28 The predominant use of schist as a meditun migijt
also suggest that the site dates to phase IT and the beginning ~f
phase m. This contention is born out by the types of sculpture
recovered; it is mostly composed of phase Il material (narrative parr
els, associated sculpture for stiipa embellishment, and many phase n
architectural fragments). Relatively few phase ill devotional icorts
were found, and virtually no late iconographic forms. Aside fro~
the presence of one monumental shrine in area D (D21), no evi-
dence of larger than life-size sculptun: is evident at the site. There
is also no tnce of sculptures with Sravastf iconohrr-aphy, and evei'l
the dlza.nnacakra rnudrii appears in only four instances. 29
More than I ,000 architectural fragments, largely from phase lJ
stii.pas, warrant further study, because tllis category of structure ill
not extant in the sacred areas. 30 This site must have been ve11' actiVl<:
during phase rr, as indicated by the 52 intact narrative panels, 37
significant fragments, six. false gable reliefs, and another 34 large
.
p•eces.
Because the remains of devotional icons appear to be representar
tive of the earlier part of phase Ill , they deserve special attention.
21
Mizuno and Higuchi, eels., Tlwreli., pis. 8 1.- 14 .1.
28 Ibid., 158.
29
Ibid., pi. 81, no. 20; pl. 97, nos. 95, 97; pl. 137, no. 5. Two of these instances
are known only from fragments of hands, though one very small image with a dott•
ble halo and the dltanrwcakra mudrii must date to the late phase m period. In thf.s
last instance, the scale suggests that this could have been the personal possessiO!l
of a monk (see section 6.6.1 Phase lii Use of Personal Devotional Images in
Monasteries).
30
These architectural pieces are mostly moldings, pilasters, cltattriiualf., etc., tbot®
some sculptural fragments are also included, sucl1 as figures under arches and tJil:
more fr:agmenta.ry p.ieccs of narrative and false gable reliefs.
31 Unfo•tunatd)', no scale was provided with the photographs, so it was not pos-
sible to differentiate the major devorionaJ icons li·om minor ones.
'II
II Bahan dengan hak cipta
228 CHAPTER EIG HT
Copyrighted material
PHASE IJ( SCULPTURE IN THE PESHAWAR BASIN 231
found exactly as they had been left, a common chiragh pamp]; one
of them was blackened at the tip by tJ1e wick that had burnt out at
the socket." 44 In 1872 Cunningham reported finding a basement at
Sahri-Bahlol that once supported several separate chapeh. T en in
situ Buddha figures were found, some of them still upright; two were
colossal, and the others were life-size or smaller. 45 Because reuse was
so common at the kno·wn Sahri-Balllol sites, the placement of these
images may not reflect their original context.
Copyrighted material
C HAPTER Nll'SE
Copyrighted material
•
BUDDI-UST ARCHITECTURE AND SCULPTURE OF GA."'DHl\..RA 235
9. 1 Phase I
As is commonly the case when origins are traced, the evidence gets
increasingly scarce as one approaches the point of inception. Un-
doubtedly, no record exists of the first Buddhists to visit Ganclhara,
and the earliest remains that can be securely identified provide tan-
talizingly little evidence for understanding this nascent tradition. A
possible Mauryan-period presence of Buddhism in Greater Ganclhara
is suggested by several ASokan inscriptions, but their content does
not directly aqdress the introduction of Buddlusm. The earliest coins
found at archaeological sites also date to this time, but they have
been found as isolated, worn examples and can only hint at possible
early activity. However, by the end of the 2nd century B.C .£., th.e
Dharmarajika complex in Taxila (fig. l) and at Butkara I in Swat
(figs. 97, 98) were occupied, as indicated by a statistically significant
number of coins.
At these early sites the distribution of donated structures provides
us with clear evidence for determining which parts of the sacred
area were devotionally significant, and their placement hints at how
Copyrighted material
BUDDHIST ARCHJTECTURE AND SCULPTURE 01' CAl'll)HARA 237
9.2 Phase If
Copyrighted material
BUDDHlST ARCHITECTURE AI'ID SCULPTURE OF OAJ\'DRARA 239
..
Elements sous droits d'autc 1r
Bl.lODJoiiS'T ARCHlTEC'fURE Al'\'0 SCULPTURE OF' GANDHARA 24 1
(fig. 20) has an oversized cell with front pillar bases and an inter-
nal plinth around the rear. However, because this shrine and monastery
were constructed in phase 11, it seems likely that they originally
housed a stUpa. This pattern of monastic construction with a shrin~
opposite the entranceway is seen at the Gujarat site of Devnimori,
which probably dates to circa 400 C.E. Thus, a developmental trend
can be traced in the northwest that must have been the architeo-
tural prototype for .l ater northern and western Indian sites such as
Aja.J:l~a, Aurangabad, Ellora, Nalanda, Ratnagiri, and others.
In the quadrangular monasteries of Taxila there is one more type
of internal image shrine, smalJ in situ image niches fabricated m
phase Ill semi-ashlar masonry and containing sophisticated late ga11dl101'
ku.Ji iconography (figs. 29, 27, 32, 20). AtJaulian and Mohra Moradu.
these niches bracket the entrance of the monastery or are within thq
actual entrance passageway. In this context it is interesting that at
Mohra Moradu several large, phase III pedestal images occupied
positions that otherwise would have been ideally suited for the con-
struction of garulhaku~i niches; pedestal images B l and B2 sit adja.~
cent to th e entrance. This interrelationship suggests that the pedestal
images were already in place when the construction of gandha.ku:/4
niches commenced.
Because intact gandlwku~i assemblages are known only from three
Taxila monaste•-ies, which happen to have been carefully excavated,
it is impossible to assess how widespread this kind of image use was
in the Greater Gandharan area, but an empty in1age niche found
in a comparable location in the quadrangular monastery at Takht-i-
bah1 (fig. 2) could well have housed such imagery. It also seems pos+
sible that some of the image shrines in the public sacred areas
contained groups of sculptures exhibiting gandhak:u# iconography.
While early phase IIl in1age shrines clearly housed single Buddha or
bodhisattva in1ages, it is not inconceivable that complex groups of
icons were placed in shrines, especially toward the end of phase IlL
The bases of some schist bodhisattva sculptures were left unfinished
on one side, suggesting that they were or.iginaUy installed as atten.
dant figures in triads.
FinaUy, a relationship appears to ex':ist among the monastic g(JJ'ldl!a·
ku.fi assemblages, Sravasti' composite images, and S•-avastr triad image5t
especially in terms of the integration of bodhisattva attendants and
a general tre nd toward complex groupings of figures. At the 5th cen..
tury western Indian site of r\iam.a, similar monastic g(IJUUtalw.# image
Copyrighted material
254 CHAPTER NINE
Copyrighted material
• '
•
APPE ·orx A
THE FOUR-PHASE C H RONOLOGICAL SYSTEM
1
Errington, " Nurnimmtic Evidence for Daling the Buddhist Remains of Ganclha.ra,"
197.
Copyrighted material
256 APPEl'<DIX A
Phase II: Circa middle to late lst century C.E. to circa 200 C .E.
Coins of the Great Kushans, including kings Vima Takto, Vima
K adph:ises, Ka.ni~ka I, a nd H uvi$ka, date this phase.
AI Phase I
1 Enington (Ibid., l92), has reviewed this issue. See also section 2.2.1 Dhmmariijilia
Complex in T axila: Phase J and section 2.2.3 Butkara I in Swa[: Phase I.
3
R. Con.ingbam, and B. Edwat·ds, "Space and Society at Sirkap, Ta.xila: A Re-
Exam ination of UrbaJ'l Fomt a nd Meaning," Anrienl. Pakistan, Vol. XD., l997- 98:
..
Copyrighted materlll
APPENDIX A 257
and II together contain 1817 coins of Azes ll (6- 17 C.E.) and 630
coins of Gondophares (20- 45 C.E.). Stratum Il contains I 780 coins
of the K ushan Kujula K adphises (30- 78 C.E.); this evidence indi-
cates that Sirkap was occupied by the Kushans toward the end of
its use.4 It was during this time that rubble (fig. 9) and km1Jii.r a~h
lar masonry (fig. ll , note masonry) technjques were employed to
constn1ct the various Buddhist centers in Sirkap. It has been sug-
gested that the shift to diaper masonry (fig. 9) and hence the begin-
ning of phase II may have coincided with the occupation of the
Kushans under Kujula K adphises in the mid-1st century C.£.5
Although it is difficult to know exactly when this shift in masonty
types occurred, the coin evidence indicates tl1at there was still a
significant population living in Sirkap in the late lst century C.E.
and into the 2nd cennuy C.E., people who used this new type of
masonry to repair and construct buildings; the archaeological remains
suggest that diaper masoruy was in use for a short period before the
abandonment of Sirkap.
D etenuining the end of the occupation of Sirkap is diA1cult because
fo•· the pe riod after the rule of Kt0ula Kadphises, very few coins of
kings Kani~ka I or Huvifika are found in all of Taxila; thu , their
relative paucity in Sirkap is not surprising. 6 H owever, Sirkap seems
to have been abandoned by the time of Vasudeva (184- 220 C.E.);
only 12 of his coins were recovered, compared to 1900 from tl1e
larger area of Taxila. 7 H owever, the end of Sirkap habitation could
have occurred significantly earlier.
D atable diaper masomy st1·uctures at other sites show that this
phase Il technique was being used in the late part of the I st cen-
tury C.E. Rubble and kaiyiir structures can be associated with coins
of Maues, Azes, Azilises, Azes II, and Kujula Kadphises, all of whom
mled dwing phase J.8 'DUs analysis also reveals that in three instances,
47- 75; G. Erdosy, "Taxila: Political Histor)' and Urban Struc!llre," Soutl1 Asian
Artluuo/ogf 1987, vol. 2, Rome: Istitur.o ltaliano per il Medio ed Esr.n:mo Odcnr.e,
1990, 65 7- 62
• Ibid., 668- 670.
~ R. Allchin, ''Archacolog}' <md the Date of Kan~ka: The Taxila Evidence."
Papers 011 tl~ Date if Karriska, cd. L. Basham, Leiden, 1968, 13.
6 Ibid., 670.
1
Ibid.
8 Kuwayama (''In the Time of Late Sirkap and Early Dhannarajika: How Taxi.la
lno·oduced Stupa Architecll.u·e," paper presented at the conlerence "On the Cusp
of an Era: Art in the Pre-Kusha.n Wodd," Nov. 8- 12, 2000, Kansas City) traces
Copyrighted material
258 APPE-1\'D IX A
__ ___,.
the appearan ce of coins in specific mbble, lwiijrir, and arguably early diaper struc-
tures. Using this evidence, he proposes a chronological sequence for these masotJ.Jy
types, noting the following relationships: Mohfll. Maliarail (n1bble masoruy), 12 co.in.s
of Azes; Dharmarajikii stiipa S8 (rubble}, four coins of J\1.aues Md four coins of
Azes; Dhann ariijika sllipo. R4 (kaiyur), one coin of Azes and one coin of Azilises;
Sidcap sacred a rea lA (rubble and koiijur) th.ree coins of Azes ll; Sirkap sacred area
IG, eight coins of Azes IT; Dhamtarajik.a relic deposir stUpa r7 (diaper), one cGin
of Azes; Dharmarajika slllfJll T 12 (kai"ijiir), one coin of Kujula Kadphises; Sirkap
sacred area l D (diaper), seven coins of Azes IJ and one coin of Kujula Kadphises;
Sirkap I E (rubble/diaper), one coin of Azes II M d two coins of KujuJa K.adph.ises;
Sirkap area l C (diaper), one coin of Azes 11 and two coins of Kttiula Kadphises.
9 J\<
larshall, Taxi/a, I l 8-9, 260.
'" Ibid., 137, 237.
11 A soft calcium carbonate rock more commonly known as travertinc.
A2 Phase !I
12
Kiziijiir ashlar masonry was covered with a thin coat of lime plaster; the finished
strucntre would presumably have been painted.
" Marshal!, Ta.>.'ila, 327.
" Fi rzsimmon.s, Stupa Designs at Taxila, 4 1.; MarshaU, Taxila, 48, 53, 256- 7, 327- 8.
Two coins of' Vasudeva found in this strucmrc conoborate this dating.
rs At Ka.Jawan the A l stiipa shrine can be direcdy related to the A l 3 and Al4
slft.j!a shrines; the main A4 stii.pa; •-elic shrines A3l, AS2, ASS, and AS4; small s/11-
pas A6, A7, AS, A9, A I 0, A l 2; and monasteries C and F (fig. 20). In the Dharmariijika
sacred a.rea, many phase JI diaper so·ucn.u'es were built, notably along the western
avenue, G4, G t-G2, G5- G6, 13, 12, and L, and par ts of the R complex_ Ma.oy
di aper masOluy shrines encir·cle the circumambulatol)' path of the main slti.pa ns
well as the expansion of the H. two-celled shri.ne. Noteworthy is the ~m ph asis on
constructing stupa stuines and rwo-ceJJed relic shrines.
16 See Mar-shall, Taxi/a, pi. 55, no. 5 for 1.he !3 apsidal hall and pi. 55, no. 6
11
Ibid., 198- 99; my field obse1vations, 1993-94. At many or these sites the
masonry is clearly evident in photographs.
A3 Phase Ill
18
A reassessment of all of the extant numismatic evidence provides a date that
is reasonably close to Marshall's initial suggestion of the mid-2nd century C.E.
(lV[arshaU, Taxi/a, 261 ).
19
MacDowall, "The Clu-onological Evidence of Coins in StfijXt Deposits," 730- 3:i
20
A good example is the multiple relic deposits from the Mat~iltyala sl1ipa. In
conjunction "~th successive enlargements of the main stii.pa, new relic deposits were
established (sec Errington, "Numismatic Evidence for Dating tbe .Buddhist RemainJ
of Ga.ndh:u-a," 196- 7; MacDowall, "The Chronological Evidence of Coins in Stiip~
Copyrighted mater! )I
I APPENDIX A 263
Semi-ashlar stupas from T axila that had coins in their relic deposits
include the Dharmarajika K 3 stiipa (three coins of Kani~ka I, taken
to be 120- 146 C.E.); the Dharmarajika P6 slilpa (three coins of
Huv~ka, 148- 84 C.E. and another seven of Vasudeva, 184- 220
, C.E.); and the J auliai'\ A 16 stUpa (two coins of Vasudeva, 184- 220
C.£.).21 T aken together, this numismatic evidence from semi-ashlar
stzi.pa relic deposits suggests that this new masonry type had come
into use by 200 C.E. The appearance of coins of Kani~ka I and
H uvil.lka might be taken to indicate that tllis transition took place
slightly earlier, possibly as early as circa I 50 C.E. In the semi-ash-
lar G monastery at Dharmarajika, a hoard of 53 1 coins of Vasudeva
and one of Kani~ka I were found in cell 16,22 another indication
that by the time of Vasudeva, semi-ashlar was in use and phase Ill
bad begun. The continuity of this late masonry type is confirmed
by the recovery of 15 Shapur ll coins (309- 79 C.E.) in the Dharma-
rajika N 11 stiipa and five Kidara Qate 4th century C.E.) coins in the
P7 stupa. 23 The latest numismatic evidence that can be associated
with the use of semi-ashlar masonry is the extensive coin finds from
the Taxila site of Bhamala (fig. 42). T his site can be dated on numis-
matic grounds to at least the late 5th centmy C.E.; 285 coins were
found as relic deposits in the main stzi.pa, small stiipa A5, and small
stzi.pa A l 52' (see Bharnala discussion in section 7.6 Phase IV: The
Decline of Buddhism). The cruciform shape of the main stiipa at
DepositS," 730). In a few rare insta nces, such as stilpa. U I i.n the Dharmarajika com-
plex, the .stiipa was restored in a late period and an ea rly relic deposit was re-
interred. In this stltpa, the relic casket contained a coin of Apol.lodotus U (80- 65
B.C.E.), another of M aues (90- 80 B.C.E.), one of Vonones (75- 65 B. C. E.), and a
defaced coin; the Jtiipa itself appears to be a r·estoration comtructccl i.n phase l1I
semi-ashlar masonry. See MaJ·shall , Taxila, 272.
71
Mar-shall, TaxiW., 263; Ku,vayam a, The Main Stupa at SM!z-ji-kf-tjhen., 23- 26;
Marshall (ExctiiJations at Ta:cila: The Stupru and Monaste1ies at ] aulia11, 8) also discu.'ISecl
the implications of this numisma tic evidence from d1e Dharmarajika complex.
72 MarshaU, TtL~ila, 282 .
., Ibid., 267, 270; see aL~o Kuwayama, " In the lime of l.alC Sirkap and Early
Dharm arajika: How Taxila Inrroduced St1ipa Architectu re."
14
Marsha.U (Taxi/a, 393- 4) recorded two main stilp a coin deposits, the first 3.4
m below the top of the mound, where six coins were found, and the second down
'~. 4 m, where 113 coins were located. He reported that small sllipa AS contained
a Varhan ll coin (276--93 C.E.) and six more coins like those found in the main
slilpa. Sllipa A 15 contained an additional 160 of these small copper coins (1vl arshaJ.l ,
Taxila, 393- 4). ErringtOn ("Numismatic Evidence for Dating the Buddh ist Remains
of G anclbara," 212, 216) has identified these copper coins as late 4th century C.E.
Kushan imitations.
2!1 FiF.lSimmons, Stupa D8sigrzs at Taxila, 46- 7; Kuwayama, The Main Stupa at Shlifr.
j"i-lci·'(iht!li, 92- 3.
26 See l\l.larsba ll, Ta.•ilo, pl. 48b. Note the masonry of the D barmarajika mai:il
' V. Zwalf's catalogue of the British Museum broadly reflects the current acad-
emic consensus on the chronologic development of Gandltaran sculpture, a.nd it
n:views man)' of the a.rguJnents that have been put forth here (Zwalf, G111ulJW'Il
Sculpture, 69 - 72}.
2 Only a few narrative relidi> have been found in Afghanistan, and they seetD.
to follo·w a diflerem development. As so little is known about this topic, it is impos-
sible ai this time to make any assertions about chronology.
Copyrighted mater 31
APPENDIX B 271
6
For a full analysis of this discussion, see Callieri, "BuddJ1ist Presence in the
Urban SettlementS of Swat, 2nd Cent. B.C.- 4th Cent. A.D." and SmLf and Smlings
.from. the .NortJz- West rif the l11dian Subcot~tinent and AJCilm•irtatz (411z t·et~tury B. C.- 11th century
A.D.): lJJcal, illdian, SasaniOJL, Graao-Persian, Sogdian, Roman (Naples: [sJAO, 1997).
1
Kuway-ama, "ll1e Buddha's Bowl in Gandhara <md Relevant Problems''; Marshall,
TaAila, 76, 86.
While contact with the Roman world via this network of trade is
certain, the issue of foreign influence remaim unresolved. 8 As fol'-
eign objects and possibly even people arrived in Gandhara, forms
and motifs were selectively appropriated when they served local needs.
In this way, Roman, Parthian, Chinese, and Indian objects and peo·
ple had an impact on the established artistic traditions of Gandhara. 9
Although usiog motif analysis to sequence Gandharan art is difficult,
there is one particularly useful datable example. Characteristic drink-
ing cups appear in a limited number of Gandharan panels, espe-
cially among supposed stair risers. A similar vessel, which had been
used as a reliquary and was therefore inscribed, appeared on the art
market in the l990s. 10 TI1e names of the royal donors mentioned
on this vessel can be r·elated to inscriptions from Mathurii and to
numismatic evidence, allowing it to be dated to the I st century B.C.E.
A similar drinking vessel excavated in the city of Sirkap can be
roughly dated to the 1st century B.C.E. on the basis of its archae-
ological context; 11 other copper and terracotta examples were also
recovered at Taxila. The use of the cup as a datable artistic motif
is effective because it appears only in a few reliefs, which were prob-
ably made around the time when such vessels were actually used
(phase I).
Another factor that has complicated establishing an early chronol-
ogy for Gandhara is the 20th century debate over the origin of the
first anthropomorphic images of the Buddha, an issue having inlpli-
cations for the dating of Gandharan narrative sculpture. 12
8 T hese issues are reviewed in R. Allch.in et al. , eds., Gat~dhoran Art m Cot~Jat:
Ji.lJJt· Wt<St Excllangt<S at lht Crossroads of Asia (New Delhi: Regency J>ublicatioos, 1997).
9 Scholars finding Roman influence also argue tbr Parthian contaCt, which can
Kha r'llosta and Prince Indravannan," J ournal of IJze American Orienlai Sodq, 116, no.
3 (1996).
11 Marshall, Taxi(f;, 612.
' 2 For a full summa•)' of dlis debate, see A. Coomaraswamy, Atl~ditt!ual Sinltalese
Art (Broad Campden: Essex House Press, 1908}; A. Foucber, "Les Debuts de !'Art
Copyrighted matet 31
274 APPEND IX B
" Behrendt, " Narr·ative Seq uences in the Buddhist Relje(S from Gandhar11."
Copyrighted materi<:
•
APPENDIX B 275
The architectural eyjdence shows that only a few sacred areas were
founded in phase I, many more sites were established in phase 11,
and phase Ill is the p eriod when most of the extant remains were
constructed. Because image shrines suitable for housing iconic images
do not appear in the phase II architectural record of T axi! a, it seems
reasonable to use this shift as a defining indicator of the phase IT-
phase ill boundary-the time when image shrines became popular.
Further, the large devotional images of Buddhas and bodhisattvas
(figs. 88, 89) that would have been housed in these shrines appear
in conjunction with the decline of narrative imagery; this transition
is most readily apparent in the typ es of small stupas that were being
constructed (see section 5.4 The Phase II Small Stilpa: Sculptural
Embellishment and a Proposed Reconstruction and section 6.5 The
Phase Ill Small Stilpa).
Tt is striking that the phase Ill iconic images of Buddhas and bod-
hisattvas are devoid of narrative content aside from hand gestures
(mudras). These gestures indicate enlightenment (bhumisparsa mudrii)
and the first sermon (dharrnacal..7a mudrii), as weU as being less specific,
such as d1e ablzaya rnudrii and the uarada mudrii. With a few notable
exceptions, 18 these devotional icons have no narrative elements; typ-
ically, their bases have small depictions of relics, devotees, Buddhas,
and bodhisattvas (figs. 88, 89, 123). A clear separation in rime bet>veen
the narrative cycles and the devotional icons is suggested by the
absence of narrative elements from this body of phase lll iconic
sculpture.
T he presence of certain rnudriis and the absence of others in the
narrative sculpture are also teWng. The ablzaya mudrii commonly
appears in narrative scenes (fig. 93, A); it is the most common hand
gesture for the iconic images (fig. 93, L) and thus represen ts a con-
tinuous tradition. The bhumisparsa mudrii appears only in conj unction
with the attack of Mara, but it became a frequently depicted hand
gesture in the later devotional images (especially outside of Gandhara)-
18
1 am awa re of embedded narrative scenes on the bases of only three devo-
tional icons out of the thousands that a re exram: the piowing scene on the base of
a first meditation of Siddh~iltha icon, fi·orn Sahrr-Bahlol site C (Stein, " Excavations
at Sahri-Bahlol," pi. XL, fig. 12}, now .in the Peshawar Museum; the scene of mer-
chant~ breaking the Buddha's fast on the base of the sta rving Buddha, from Sikri
(Dar, "The Sikri Sculptures," 23, 25), now in the Lahore Museum; and a depic-
tion of the Pariniroli110. on the base of a Budd11a image from Junal G;lfhl, now in
the La hore Museum.
i.e., another iconographic element that stems from the narrative tra-
dition.'9 H owever, the only narrative depictions to show the bhitm:is-
parsa rnudrii are found in lzmmikfl panels of phase H- Ill. It is not
present in the narrative panels used to encircle the drums of phase-
II small stttpas, and thus it probably postdates phase II (see section
5.4.4 The Dome, Harmikii, and Clzattriivali). T he dlw:rmacakra mudrii,
associated with the first sennon, never appears in narrative contexts.
In narrative contexts the Buddha is shown turning a wheel (an lmusu-
a11y late exan1ple in stucco is extant at Thareli; see fig. 72) or some-
times simply sitting in abhaya mudrii with his first five disciples (fig.
93, A at upper right).20 In contrast, the dhamzacakra mudra is com-
mon in late phase ill icons (fig. 88, seated Buddha on the left end;
fig. 126). The late dating of devotional icons with this mudra can be
archaeologically corroborated , because at Takht-i-bahr late in phase
UJ , sacred area XX was built (fig. 2), and many late sculptures were
donated there (see section 7 .2 .4 T akht-i-bahl: Late Ph ase IH
Construction). Six of the eight excavated schist Buddha images fo1md
in this court djsplay the dltannacakra mudrii, while the other sculptures
found there include late phase ill monumental Buddha images and
fragments of Sravastr composite panels (sec section 8.2.2 T akht-i-
bahf: Sacred Area X)(: Loose Sculptut'e). Fur tJ1er confirrning the la te
dating for the dhamzacakra rnudrii is it~ consistent presence an1ong
groups of late sculpture such as those from the Sahn-Bahl61 sites.
T he varada mudrii is very rarely found in the Peshawar basin, Taxila,
and Swat, with one exception that I am aware of In sacred area
XX at T akht-i-bahl, in the upper register of stu.pa P37, several in
situ stucco Buddhas exhibit the oarada mudrii (figs. 2, 48).21 Otherwise,
this late mudrii is known only from a few late examples from
Afghanistan. C ertainJy the varada mudrii. postdates the narrative tra-
dition. Thus, a general iconographic development can be used to
separate phase II, phase ill , and late phase lll production.
19
The bhiimisparJa mudrti often appea.rs on the embei.Lislum:m of hanniklis, which
could easily belong to phase m.
20 T he dlwmzacakra mudrii is present in a few MmUkii reliefs, which show the fo ur
great life even111; in these i.nstanc~-s, it suggests a phase Ill date for the im age.
21 Hargreaves, "Excavations at Takht-i-blihf," 36.
While many sculptures in the Peshawar basin and Swat were carved
in schist and soapstone, the majority of sculpture must have been
done in clay and stucco (and occasionally terracotta). Signiftcant
quantities of stucco and some terracotta sculptures survive fwm sites
in Taxila (figs. 18, 19, 28, 29, 34, 35, 37, 43) and from Butkara I,
but the majority of excavated stucco sculptures have been lost. 22
Excavation photographs preserve additional evidence from the Peshawar
basin sites of Ali Masjid (a site on the Khyber Pass) (fig. 96), J amal
Garhi, Mekbasanda, Thareli (figs. 72, 77), Sahri-Bahlol, Shah-ji-kf-
Qher1, and T akht-i-bahi (figs. 46, 48, 50) and from the Afghanistan
sites of Bamiyan, Fondukistan, Hat;L<;Ia, Shotorak, and Tapa Sardar.
We know that by phase Ill, such images were being produced at
other sites in the Peshawar basin and Swat, because many extant
image shtines and sacred-area walls contain small sockets for a:ffi:cing
and supporting the bodies and heads of clay and stucco images (figs.
47, 49, 50). By phase IV, in Alghanistan, unfired clay was the medium
used for most images;23 because such clay is fragile and SU$Ceptible
to water damage, little has smvived. It appears that commonly the
body was made of clay, and the head, and sometimes hands and
feet, of stucco, a fabrication technique that may help to explain the
thousands of disembodied stucco head$ that survive from Greater
Gandhara. 2'1
On the basis of evidence from Taxila sites like the D harmarajika
complex, J aulian, and Mohra Moraclu (see Chapter Six), a strong
correlation is found ben-veen the use of phase ill masonry and the
advent or popularization of large devotional sculptures in clay and
stucco. Had these media been in use during phase II, we would
22
Many of the 19th and early 20th century excavations ignored this type of
sculprure, because it would have required expensive and time-consuming conserva-
tion. The most extensive body of stucco sculpture survives at Taxila, where sheds
were built to protect the imagery from rain.
23 For a summary of the m~or issues, see Taddei, "Chronological Problems
Copyrighted material
278 APPEl'lOIX B
z:> These media can be rapidly worked and molds can be used, one of which
cao be seen in the Tax~la museum.
26 These reliefs mainly appear in the Archaeological Survey of India Frontier
Circle photographs housed in the Briti.sh Librat)'· They include the birth of the
Buddha, from Takht-i-bflhf (ASF photo 350); a second depiction of the birth and
the seven steps, both probably from Takht-i-bahl and now in the Peshawar Museum
(ASF 14·0 I}; Siddhlirtha leaving the palace, lrom Takht-i-bahl (ASF 348); a depar-
ture scene from Sahri-Bahlol (ASF 1265); two other life scenes from Sahrf-Bahlat
(ASF 1271 ); and a relief from Tharcli (Mi:wno and Higuchi, eds., 77umli, pl. 69;
no. 3).
21 The RI plinths were fabricated wung diaper mason.ry and were added to dtt
west {ace of the phase I kafij1ir ashlar base supporting the R3 chamber. The fact
that dtesc rcliei.S were attached to diaper masonry plinths, coupled with their style,
led Marshall to attribute them to his diaper period (phase U}, though they could
be later (Manhall, Taxilo, 253). His identifications cannot be corroborated, because
the reliefs in question were not photographed and toda)' are in a fragmenta.y cor-..
clition. At the Dharrnarlijika complex there is also a.1 in situ depiction of Trapussa
and Bhallika offering food to the Buddha, adjacent to the steps leading into court
A (Marshall, Taxi/a, 277, pi. 20, no. 1L4}.
Copyrighted materi I
280 APP£1\'0IX B
solid early phase m horizon for this architectural element (see sea.
tion 7.5 Ranigat: Phase IT to Phase IV Architecture).33 While th~s
evidence tells us only that this specific image postdates the deposi,.
tion of the coins, it nonetheless is a datable benchmark for the study
of stucco production.
Along -.vith the phase m and IV stucco p roduction, some schist
devotional icons were still being ca1"Ved. .Even though this use of
these two material types overlaps chronologically, there is good rea--
son to believe that schist was a more popular medium early in phase
Ill and that clay and stucco were preferred late in phase m and in
phase IV. For example, relics are commonly depicted on the bases
of schist Buddhas and bodhisattvas (fig. 123), but 1 am aware of only
one instance where a relic appears at the base of a stucco image: ~t
stucco seated Buddha attached to the ba~e of the main. stupa at
Mohra Moradu (fig. 34·). The schist in1age is typical of the less thatJJ
life-size mid-phase Ill devotional icons that exhibit conventiooa11
iconography (no relic depictions appear on schist Buddha images
that can be attributed to the later parts of phase ID). Thus, the
slightly over life-size stucco Buddha at Mohra Moradu probably dates
to about this time, or near the beginning of the common use of the
stucco medium.
A more telling chronological factor for dating stucco production
is the trend toward monumentality; life-size and smaller icons gen-
erally relate to early phase m and the very large icons to the late
part of phase Ill and phase IV (see section 6.3.6 The D harmarii.jika
Complex: Late Phase ill Architecture; section 6.3.8 Jauliafi: Phase
fll Architecture and the Image Shrine; section 7.2.4 T akht-i-bahf:
Late Phase m Construction; and section 7.5 Ranigat: Phase 11 to
Phase IV Architecrure). Monumental images were fabri cated using
stucco and clay for technical reasons, cost, and the difficulty in finding
large pieces of schist. In contrast, only a few monumental schist
images have been excavated (see below), whereas monumental stucco
heads (some more th<m 50 cm from chin to top of forehead) have
been recovered fi·om the Dharrnariijika complex, Jauliai'i, Mekhasanda,
Sahrr-Bahlol, and Takht-i-bahf (fig. 50), and sockets suitable to sup-
port monumental stucco images are extant at Abba Sahib China,
Ranigat, Takht-i-bahi (figs. 4 7, 49), and Thareli (see specific di~cus-
33The 14 in ;itu coins include one of Azes II, one of Kujula Kadpnises, three
of Hu~ka, and eight of Vasudeva.
Copyrighted material
..
APP£1\'DIX B 281
sions of these sites). Using material types to order phase Ill and IV
sculptural production must be approached with caution, as it is clear
that schist remained in use late in phase Ill and into phase IV.
Five Peshawar basin schist images have inscribed dates, but the eras
of these dates are unknown. After considerable debate., the follow-
ing possible eras have been suggested: the O ld Saka, 3 ~ the Vikrama,
Copyrighted material
,
282 APPF..NDLX 8
36 K. Dobbins, "Gandhara Buddha Images with Inscribed Dates," &st and West
18, no. 3- 4 (1968); G. Fussman, "Documents Epigraphiques Kouchans, Buddha de
l'An 5," .Bulletin de l'Ecole Franfaise d'Ettdme-Orimt. LXI (1987); A. Harle, "A Hitherto
Unknown Dated Sculpture from Gandhara: A Preliminary Report," in South Asia11
Archaeology, ed. J. E. van Lohuizen-De Leeuw (Leiden: E.J. BriU, 1974); K. Kha.nda-
lavala, "1l1e Five Dated Gandhara School Sculpt1rres and Their Stylistic Implications,"
in l11dian Epigraf>lty: Its &ari11g on tl~ His/m)• qf Art, ed. F. Asher, G. Gai, and Americnn
Institute of Indian Studies ( ew Delhi: Oxford, 1985}; G . Mitterwallner, "The
Brussel$ Buddha li·om Gandbara of the Year S," in l11vestigati11g bufiatl Art., ed.
lV!. Yalcliz and W. Lobo (Berlin: Staatliche M.useen Preussischer Kultur·besitz, 1987};
Rosenfield, ""flze f)ynflSti' Arts qf tlze Kuslums. The issue of dropped hundreds was pr·o-
posed for Mathu111 inscriptions by J E. V<Ul Lohuizen-De Leeuw, "The Second
Century of the Kaniska Era" (paper presented at a Seminar on .Late Ktc~han
Nurnismarjcs, Loodon, 198 1). This theory has been critiqued O· Williams, "The
Case of the Omitted Hundreds: Stylistic Development in Mathura Sculpture of the
Kusana Period," in Mathura: 111e Cullural Herililge, ed. D. Srinivasan [New Delhi:
American lrostitute of Indian Studies, 1989)), a.nd new evidence corroborating itS
existence outside of Gandhlira has appeared in R. Salomon, "Three Dated Kha.r0${hl
In scriptions," Bulletin tf tlte Asia /nstiJule. no. 9, 1995: 127- 141.
Copyrighted material
APP.ENDDC B 283
ered too late, as they would place these sculptures, which seem to
relate to the main period of schist devotional imagery, well into
phase IV. Tllis kind of late dating does not agree '>vith the archi-
tectural evidence indicating that life-size image shrines came into
vogue in the early to nliddle part of phase III, or sometime after
200 C.E., a nd that by phase IV monumental imagery was more
common. The indirect architectural evidence of image shrine scale
is most in accord with the dating that results from applying the
Vikrama era.
Another image with an inscribed date, but lacking an archaeo-
logical context, is a Sravastf triad with an inscription mentioning the
year 5 of an unknown era (sinlihu in iconography and style to fig.
125). For it too, a range of eras has been suggested; some argue for
a date as early as ~ka l's era,37 or Kan~a II's era, 38 while others
advocate later eras, notably the Gupta or Sita-Hu~a King Khingila
eras. 39 The Buddha displays the dharmaetd.7a 1m1drii, sits on a lotus
throne, and is attended by two bodllisattvas-all elements that would
seem to place it later than the two inscribed standing Buddhas dis-
cussed above. This irnage is most readily compared to other Peshawar
basin Sravasti triad and Sravasti composite devotional icons. This
body of sculptures with Sravastf iconography appears to be related
to 5th century C. E. images outside of Gandl1ara, at western Indian
sites like Aj~~· Therefore, it seems unlikely that the year 5 inscrip-
tion refers to the Vikranla (58 B.C.E.), Saka (78 C.E.), Kani~ka (120
C.E.), or Ka1li~ka Il (dropped hundred) eras, because all of those
would place the relief too early.
The suggestion that the year 5 refers to the Gupta era or to 325
C.E. 40 seems viable, though we have no evidence for the use of this
era designation so fm· from the Gupta seat of power. An alternative
late dating would put this image in the 5th year or the Sita-Hurya
37
G. Fussman, " Documenls Epigraphiques Kouchans, Budcllm de l'An 5," Brdlai11
de l'&ole Frarlfaise d'Extreme-Orie11t (1974); Harle, " A Hitherto Unknown Dated Sculplure
from GnndJ1ar-a: A Pl'CtirnuUll)' Ro:port"; P. Pal e t Ill. , ligllt qf Asia: Buddlw SoJ.ymmmi
i11 .r'lsUm Art (Los Angeles: Los Angeles Cow1ty Museum of Al1, 1984), 19 1.
38
J. C•ibb, "The Origin of the Buddha Image: The Nwnismacic Evidence," in
Soutlt Asi'm Ardraeoln~·, ed. B. Allchin {Cambridge: Cambridge University Press, 1984},
124; S. Czuma and R. Monis, KushaJI. Sculpture: Images ftom Ear(y Irzdia (Cleveland:
Cleveland Museum of Art in cooperation with Indiana University Press, 1985}, 35,
79, 198; Lohuizen-De Leeuw, "Th e Second Cemwy of' the Kaniska Era."
39
KhandalavaJa, 'The Five Da ted G-dlldha.ra School Sculptures and Their Stylistic
Implications"; Mittcnv-aUner, "The Brussels Buddha from Gandham of the Year 5."
«l Khandal:•v:tla, 'The Five Dated Gandhara School Sculptures and Their Stylistic
implications," 69.
Copyrighted material
284 APPENDIX B
41
Mjnerwallner, "The .Brussels Buddha from Gandhara of the Year 5," 222.
42
A good example is a relief from Sanghol Rhode lhar is part of a larger oar>-
rative set of eight panel~; it shows the Buddha with flames S\UTOtutding his balo
and water issuing from his feet (see ASIM no. I I 39).
4
$ Brown, "Tbe Srllvastl Mirades in the Art of lodia and Dvarav:ui"; Foucher,
L'Art Crico-Boudtflrique du. CmuiMra; Etude szu les Origitres tk l'lnjlu111u:e Classique dons l'Afl
Bouddhiq11e dt L'l11dt et da fExlrimB-Orient; J. Huotington, "A Gandha.ran Image of
Amjtayus Sukhavati," Arwali de/Nstituto U11iiJcrJitario Omm.Ulk di .Napoli 40 (1980)i
Lohuizen-De Leeuw, The Scythian Period; Rbi, "Gandha.ra.n Images of the Srliva.stf
Miracle: An Tconogmphic Reassessment."
Copyrighted materi I
..
APP&"'DIX B 285
Copyrighted material
286 APPENDIX B
the Sravastf plaques are in such good condition that it appears they
were commissioned specificaUy to be deposited here and thus must
postdate the shrine (see section 8.2.3 T akht-i-bahi: Two-Celled Shrine
XXill: Loose Sculpture and Appendix C). T he greatest concentra>
tions of sculptures \--vith Sravasti iconography were found at Sahr£..
Bahl61 sites C and D, where no fewer than six Sravastf composite
panels and 16 Sravasti plaques, one base, and two fragments were
excavated. These find~ from Sahri-Bahlol were in reused contextS';
but most of the associated sculpture and the sites themselves can be
dated to the middle and late parts of phase m (see section 8 . 3.~
Salu·I-Bahlol C: Loose Sculpture; section 8.3.3 Sahri-Bahlol D: Loose
Sculpture; and Appendix C). Also significant is the absence of SravastJ
sculpture at Thareli and Mekhasanda, both sites that were more
active elUt-ing phase Jl to the midcUe part of phase Ill (though a few
late phase Ill monumental image shrines can be found at both ceru.
ters) (see section 8.4 Thareli: Loose Sculpture; and section 8.1
Mekhasanda : Loose Sculptu re).
Several other strands of evidence point to a late phase m
or phase
IV dating for the Sravastl plaques, triads, and composite devoliona:l.
icons. The phase IV rock-cut Buddhist images from Swat include
depictions closely related to the St·avasti triad images of the Peshawat
basin, showing that tlus format for depiction remained significant i ~
the northwest in later centuries.4 1 Similar relief~ also appear in many
5th and 6th century C.E. cave complexes in western India, sug-
gesting that the Gandharan forms, though perhaps earlier, are not
widely separated in time from the northwest Sravastf sculptures. T ho
earliest western Indian examples, directly related to the Gandharaa
Sravasti triad and composite images, are the multiple reliefs found
in the late 5th century C.E. uilziiras and cail:)'as at Ajru:•~a. T h e westt
ern Indian examples are stylistically different from those of Gandhar~
but they do share many iconographic features, such as the lotus
throne with nagiis at the base, devotees, and complex groups o.f
attendants (e.g., boclhisattvas), and they are relatively large icons,
Significantly, the western Indian examples follow a reguJar codified
47
Filigenzi ("Mru·ginal Notes on the Buddhist Rock Sculptures of Swat," in South
Asum ArcJzaeolog)• 199i, ed. M. Tadclei and G . de Mru·co (Rome: IslAO, 2000]) has
addressed th.is body of imagety and r.raced many of tbe iconographic variMts. ThiS
is also the best source for a full bibliography on this topic.
Copyrighted materi I
•
APPfu'ID lX D 287
Copyrighted material
APPENDIX C
' Crompton noted that many of the sculptures atjamlil Garhi had been removed
from their original positions (A. Crompton, " Report on the E.xploration of the
Buddhist Ruins at J amal Garhi Du.ring the Momhs of March and April 1873, by
the 8th Company Sappers and Miners, in Pwyt1b Publi~ Works Deportmmt Proceedit~g~,
Local Funds .Branch, December 1873, C ivil Works: Building no. 1-a, Appendix a,
1- 7; Pwyab Coummumt CazetU, Supplement, 12 February 1874, 1- 7," in E. Errington
"The Western Discovery of the Art of Gandhara and 1.he Finds of Jamalgarhi"
(Ph.D. thesL~. London University, 1987), 446. Wilcher went so far as to suggest that
al Takht-i-bahi d1e scuJpt,u·e was intentionally destroyed and muti11ted, as none of
it was found in situ (Wilcher, "Exploration of the Buddhist Ruins a1. Takht-i-bai,"
435).
2 Several phase m stucco Buddha figures were reused at the site of £~11ma in
Afghanistan , as Shoshin Kuwayarna, one of the excavators, pointed out to me. See
also Mizuno, ed., Dun>1an Tepe and Lalma: Buddllu t Sites in iljglu111ulon Suroeyed ill
1963- 65, pl. 22, fig. 2. A careful review of the archaeological evidence shows that
stucco imagery was also reused at Sahri-Bahlol and at the Dhannarl\jika complex,
where many heads were placed in two-cellcd shrines. It seems likely that many of
the loose stucco heads recovered throughout greater Gandhara were also reused.
' I have discussed tlus at length in Behrendt, " Relics and Their Representarjou
in Gandhara."
Copyrighted material
APPENDIX C 289
the image itseU' had taken on sacred qualities and was not easily dis-
carded; this might account for imagery being incorporated into the
masonry of stiipas or generally used as building material in the sacred
area, in a location where it could not be seen or venerated. Occasion-
ally, reused images were integrated into a new stUpa at the time of
its initial construction. During the latest period of reuse (phase IV),
a practice developed of gathering all of the extant devotional imagery
from the sacred area and assembling it around relic structures that
were apparently still being worshiped. It is significant that during all
periods of reuse the sculpture was moved to the holiest points in the
sacred area--almost without exception, relic shrines and stiipas. T his
pattern indicates that such reuse was the result of a donative or
devotional practice; many of these nodal structures were repaired
and maintained, and in some instances it appears that the sculpture
was deposited over a long span of time.
In Greater Gandhara, imagery was being reused continuously, as
structures fell out of use or were refurbished. However, it is possi-
ble to recognize two major chronological ranges when reuse wa.s
more prominent. The first period probably occurred near the end
of phase II and into phase ID, when many narrative sculptures were
moved, most often to two-celled shrines in the sacred area. The sec-
ond period of reuse appears to correspond to a time in phase IV
when patronage had collapsed and Buddhist population was declin-
ing. It is characterized by the reuse primarily of phase III devotional
images. Although a few narrative images appear in such deposits,
they are typically worn and broken and do not form coherent sequen-
tial sets, suggesting that they had been in circulation for a long period
of time before their deposition. This late period of reuse can be
roughly dated, especially at the sites of Sahrr-Bahlol, as the major-
ity of recontextualized imagery found here consists of large devo-
tional images of Buddhas, bodhisattvas, Sravastr triads, Sravastf
composites, and Sravastf plaques, along '1-vith a few monumental
stucco heads (see discussions of Sahrr-Bahlol loose sculptural finds
and Appendix D). It seems that the last Buddhist communities did
not have the economic resources or perhaps even the ability to cre-
ate new images. The patterns of late reuse provide a valuable means
to better understand this period of Gandharan Buddhism.
The recycling of Buddhist imagery is to some extent a site-specific
occurrence, though certain patterns are shared. Here, I discuss sculp-
tural reuse in the Dharmarajika complex and Kalawan in Taxila,
Copyrighted material
APPEJ'IDIX C 29 1
Copyrighted material
292 APPF..NDIX 0
cr eated dUJing phase 11, were found scattered around the exterior
of the building (fig. l , southwest of the main stilpa). In this instance,
out of 23 sculptural fragm ents, only three images belong to phse
m,12 suggesting that the imagery was probably deposited soon aft(lr
the narrative relic·& lost favor in the sacred area; otherwise, we wouid
e>..'J)eCt to see more phase m: imagery intermixed. Another compa-
rable instance of early reuse is seen at Kalawan in the A 1 two-celled
stupa slu·ine (ftg. 20). In this shrine, in addition to a reliquary, cop>-
per plate inscription, and coin of Vasudeva, 18 miscellaneous fi·ag-
ments of phase II schist sculpture and a stucco head were recovered.13
T he evidence for dating the sruine to phase 11 is overwhelming: a
datable phase n inscription, a phase II coin, phase n: masOfll)', and
n
a body of phase sculpture that mu~t have been deposited not long
aiter these images lost their or·iginaJ context.
Copyrighted material
APPENDIX C 293
•~ The T 4 two-cellcd stiipa shrine (colu·t XXHI) and aqjacent stiipa were exca-
VlJted by the Archaeological Survey of India, but lUll descriptions were not pub-
Lished; see AS!FC 938- 50; Stein, Archeological SutuO' qf India Frontier Circle fo r 1911- 12,
i-iii, 2- 3, plan 30- 1. Many of the 133 pieces found in and around this shrine were
fragmentary and difficult to identify, but the following rough count can be made:
2 1 fragments of phase l.I schist narrative sculpture, 40 fragments of phase l.I stii.pa.
embellishment, five fragments of fa lse gable reliefs (one quite large), four fragments
of phase UI icons, five phase ID heads, three small panels with a seated Buddha,
seven mall panels showing a seated Buddha with some elements of Sravastf devo-
riooaJ iconography, one large unfinished Sriivaso composite relief, and lour frag-
ments of Sravasti composite relief.~.
16 111e s:i.ngle-celled Tha.reli shrines include D3, D4, D5, and D6 (which con-
tained the base of a stiipa). In D4, the excavator noted, "Inside this room were
found stone figures of the Buddha, of Vishnu riding on a garuda ... and srucco
figures of the Buddha, among other things." DS is said to contain niches for votive
offerings, and D6 has two niches on r.he inside back wall. l.n front of this shrine
and "between building 0 5 and 0 6, numerous stone sculptures, stuccos and other
objects were lo und" (Mizuno a.nd Higuchi, eels., n1areli, L53- 4).
17
Khan , " Prel.Lmi.oa.ey Repon of l\II•u:janai.," 13; see also plates. [n the S 3 two-
cclled stiipa shrine, at least 1.2 reu sed phase U a.ud Ill reliefs were excavated. Many
of these reliefs appea.r to have Oliginally embellished the strifia (see section 5.3.5
Marjanai: Original Sculptural Placement), but. at least 12 were donated a.licr the
shrine had been established.
18 At .Butkara Ill, several complete sets of narrative reliefS were reported from
individual sh•ines; th ey were apparently piled a t the bases of stiipas isohued in
enclosed shri nes. We caJU10t know if this reflects reuse or if these were part of the
original stiipa's embellishment, as the exact constituents of these assemblages have
yer ro be published (Rahma n, "Burka ra ID," 703).
Copyrighted material
294 APPENDIX C
o? 1\llarsha!J indicated the presence of these late masomy o·epairs and buu.o·esses
on his map. His arguments for a nondescript phase IV masonry type are supported
by late modifications made to the Dharrnarajika comt H and J monasteries (Marshall;
Taxila, 286-7). This late irregular masonoy cannot be accurately dated, but it does
postdate the original construction of these monasteries.
20
An extensive discussion of these sites, along with information regarding the
specific location of sculptures, is found in Tissot, "Salu-i-Bahlol (Part IV)"; ''The
Site of Sahri-Bahlol in Gandhara (Part lli)"; "The Site of Sahri-Bahlol in Gandhar""
Pakistan: Further Investigations"; "The Site of Sahri-Bahlol in Gandhara."
21
Stein said that small stiipa ili, which had a 2 m square base and is decorated
with a stucco frieze, was used for the deposit of miscellaneous sculptures at a later
time. He al~o noted a late pavemenl that showed this stapa was still in use whco
the rest of the site had fallen into ruin (Stein, "Excavations at Sahri-Bahlol," I07- 8.
ASFC photos 1100- 1105 show S<.;ulpture found around stiipa base iii. The exposed
stiipa base is shown in ASFC photos l 105- 6. Stein noted that the stucco sculpture
of this sti~opa is somewhat narrative in content and suggests an early foundation
(Stein, "Excavations at Sahri-Bahlol," I07--8}.
27 Stein, "Excavations at Sa.hri-Bahlol," I 04. Photos record the recovery of 19
narrative fragments, I .I false gable fragments, and two llamrikii panels. However,
these narratives cannot be grouped into sets, and the fragmentary nature of this
earl)' sculpture shows that it has completely lost it original context, probably because
it h;~d been i.n circulation for considerable time before its deposition.
23 Ibid., 11D-·11. Imagery from around the base of stiifx• i can be seen in ASIFC
photos 1214- 17. It incl udes nine large sch.ist sculptures of seated and standing
Buddhas and bodhisattvas, along with a considerable number of sculptural frag-
mentS from devotional images. A third possible .sh1pa is platfonn iv (where reused
sculpmre was deposited), which has a 2.3 m square base.
Copyrighted mater I
'
APPF...J\'DlX D
DI Lorryan Tangai
I AS IM 1O:i2 69.
APPENDIX D 297
D2 Takltt-i-b1iltf 2
2
ASfM nos. l I 73- 74; ASIJ'C 5- 7, 181- 302, 337- 366, 436, 444, 446, 62 1.- 622,
767- 790, 798, 825, 856- 970, 1403- 1410, 1477- 1 •~93, 1845- 48. See also Arclzaeolngical
Suroay qf lntiW. Photographs: !tu/ex qf Location, IndW. Office Colkctums in 1/ze British Library.
' Bellew, A G1!7Zeral Report on the Yusufi;ais, 92; Wilcher, "Exploration of d1e Buddhist
Ruins at Taklu-i-bai," 435.
4
Ibid., 437.
Copyrighted material
300 APPENDIX D
D3 Salzri-Bahliil Sites
D3.1 Sahri-Bah/.Ol A
When the site of Saht-r-Bah16J A 10 was excavated in 1906- 07 morQ
than 300 sculpttw<Ll f"r-<~.gments were recovered (fig. 54), but unlike in
the later Sahrr-Bahl61 excavations, a complete photographic record
was not made. The excavator did mention the recovery of 83 frag•
ments of phase ll stupas and 28 pieces of narrative sculpture.'' Thus,
while ~-peci.fic sculptures of interest are noted and discussed, little cal1i
be said about the total body of recovered sculpttu·e that was deposited
in the Peshawar Museum. Noteworthy are eight devotional icons
found in situ lianking the entrance of a poorly preserved structure
(lig. 55), which appears to have been a two-celled relic shrine, on:
the basis of structural typology (see section 8. 7.2 Sahrf-Bahlol: Some
Possible In Situ Sculpture). 12
D3.2 Sah1i-Balzlol B
T he 1909- 10 excavation of Sahrr-Bahl61 B yielded considerably quan-
tities of schist sculpture. 13 One of the excavated stujJas14 was embell-
ished with in situ stucco sculpture (fig. 56). This site has a moderate
number of phase IJ remains, which include 12 nm-rative panels, a:
single false gable, and two hamlika pieces. Phase Ill medium to life..
size sculptures include eight standing Buddhas, 12 seated Buddhas,
three standing bodhisattvas, and five seated bodhisattvas (one of which
has the late crossed-ankles motif). otable late phase Ill material!
10 ASO'C 2 1.- 87; .D. Spooner, Arcltoologi.cal SuTVI!)' iflndi.a Frorllier Cinlefor 1906-07
(Peshawar: Govemment Press, Northwest Frontier Province, 1907); Spooner, ~Excava
tions at Sahri-Bahlol (1906- 07)."
11 fbid., I 07, 109.
12 [bid., l 05- 06.
13 AS1FC 623- 727; Spooner, ArcJ@/.ogical Survry if buli.a FrorllWr Circle for 1909- 10;
Copyrighted materi I
J
APPENDlX D 301
I
T he numbers presented here are approximate, because large quanti-
ties of often very fragmentary scu1ptw·e were found at Sahrr-BahJ61 C.
The phase II material i~ composed of 19 narrative fragments, 11
I false gable fragments , and two hamzika panels. These narratives can-
not be grouped into sets, and the fragmentary natu re of all of this
early sculpture means that it has lost its original context. Phase Ill
I
schist sculpture includes four medium-sized standing and 15 seated
Buddhas (three of these seated Buddhas are in tUwmwcakra mudrii), as
• well as two small standing Buddhas and five small seated Buddhas.
T he medium-sized bodhisattvas include 15 standing and ten seated
(one of the seated examples has a lotus base, two have crossed ankles,
and two are in dJ1amwcaJcra mudra), six bases, and eight small, stand-
ing figures. There are also three Sravastf composite panels and ten
Sravasti plaques.
The stucco finds include one Buddha, one monumental Buddha
head, and 166 heads of Buddhas and bodhisattvas ranging in scale
from very small to life-size. Tllis high number of heads suggests that
conditions were not favorable for preservation of the more fragile
bodies of tl1e stucco sculpture at this site; this kind of selective sur-
vival of stucco heads can be seen at many other sites, notably
M ekhasanda, som e Taxila sites, and H ac;lc;la in Afghanistan.
D3.4 Sahri-Bahl6t D
At Sahri-Bahlol D great quanttbes of sculpture were recovered in
reuse contexts 16 (see Appendix C) (fig. 59). The approximate count
of phase Ill sclllst scuJpmre comprises one medium-sized standing
Buddha, I 2 medium-sized seated Buddhas, three small standing
Buddhas, and four seated Buddhas. T he few bodhisattva.s include
two medium-sized standing images and one that is seated, along with
15
ASIFC 1079- 11 85; Stein, Archeological Survry of f11dia Fmaticr Circle for 1.91 1- 12;
Stein, "Excavations at Sahri-Bahlol."
16
ASIFC 1186-1254; Stein, Archeolngical Sw~' of b1dia Frontier Circle for 1911- 12;
Stein, "Excavations at Sahri-BahlOI."
Copyrighted material
APPENDIX D 303
0 5 Meklwsanda22
0 6 Ranigat 23
22
Mizuno, ed., Mek11asanda, pls. 35- 66.
21 To avoid the possibility of recou nting sculptures, the on.ly material considered
is that which appears in photographs in Nishikawa, Rtmigat.
Copyrighted material
•
304 APPENDIX D
Dating to phase Il (and to some extent phase Ifl) are 173 small
schist heads and no fewer than 448 fragments of small schist figures,
many of which were parts of narrative panels. Corresponding to
these figures are 1212 schist architectural fragments and 131 elements
of vtdikiis. Also found were six large fragments of pillar capitals.
The schist phase LU devotional icons consist of 29 standing Buddhas;
seven bases for standing Buddhas, 17 seated Buddhas (five of which
are in dhannacalcra mudrii), seven heads, nine standing bodhisattvas,
four bases of standing bodhisattvas, 12 seated bodhisatrvas (one in
dhannacakra mudrii, three with crossed ankles, and one seated on a
stool with a pendant leg), and three bodhisartva heads. Also found
were two large donor figures and I 31 fragm ents of schist devotional
icons. Although no Sravasti figures were fOtmd, 35 fragments attest
that this late phase lll category of sculpture was present at the site.
Stucco finds are limited to 15 medium-sized heads, 121 small heads,
and 228 small fragments of figures. Also documented in these pho~
tographs arc 58 large coins and 74 small coins.
GLOSSARY
additive image
An image that was placed, where space permitted, against a wall
in a sacred area or vihiira courtyard that had no shrine-like enclo-
sures (figs. 35, 36).
assembly hall
A square hall found at most Gandharan sites; many were large
enough to require a set of four internal pillars to support the roof.
In many cases, the hall is attached to the monasteries, but in some
instances it is by itself. The exact function of these chambers is
not known; their association with monasteries suggests that monks
may have assembled in th em.
Bactria
Ancient name for a reg10n in Afghanistan.
chattriivali
Stacked umbrellas that sheltered the vertica.l a-xis of a stilpa. Their
presence is also linked to creating a proper auspicious environ-
ment for housing a relic.
diaper masonry
A masonry type used in T axila, the Peshawar ba~in, Swat, and
Afghanistan in which large, roughly shaped stones were set into
beds and small chips of stone filled the interstices (fig. 9). In the
Taxila area, this masonry type can be used to date a monument
to phase n.
direct-access relic
A relic that was openly displayed rather than being placed in the
inaccessible core of a stupa; e.g., the skull bone of the Buddha
that was on display at Ha<;l<;la.
direct-access relic shrine
A shrine that displayed a relic that could be viewed and some-
times hanclled (figs. 118, 119).
Copyrighted material
•
316 BffiLlOGRAPHY
Khandalavala, K. "The Five Dated Gandhara School Sculptures and Their Stylistic
Implications." In Indian Epigraphy: Its Bearing 011 li1e Histrny of Art, edited by Frederid
M. Asher ~1t1d G. S. Gai. New Delhi: O,.-ford & ffiH , 1985.
Klimburg-Salter, D. "Front an Art Historical Perspective: Problems of Chronology
in the Ku~~a Period." In Coins, Art, mui Clumwlogy: Essays 011 the Pre-Islamic His!IYIJ
of the lndo-lramim Bord.erlmuls, edited by M. Alram and D. Kl.imburg-Sa.lter. Viennl(l
Vedag der Osten·cichischcn Akademic der Wissenschaften, 1999. ,
- -. 77w Kingdom of Bamiyan: Buddhist Att and Culture qf the HilutJ., Kush. Naples and
Rome: lUO, 1989.
Knox, R. Amaravati: Buddhist Sculpture.fivm the Grent Stupa. London: British Museunj;
Press, 1992.
Konow, S. " Kharoshti lnscription on a Begr-am Bas-Re.lief. " Epigraphica Indica XXJJ.
( 1933- 34): ll-1 5.
Krishan, Y. 171t Buddha image: lt.s Origin and Develi>pment. New Delhi: Munshiram
ManoharlaJ Publishers, 1996.
Kuwayama, S. "Route Changes and the Decline of Gandhara." ln Sourw <rfGandhariJI)
Buddhism: Archaeology, Art and Texts, edited by K. Beh.rendr and P. BrancacciO',
Vancouver: Univer..-ity of British Columbia Press, in press.
- -. " Km1jur Ashlm· and Diaper Masomy: Two Early Building Phases in Taxila,.
I11 111e Cusp qf an Era, edited by D. Sri.nivasan, in press.
- -. 17UI Main. Stupa at Slliilz-fi-l.i-t/lwi: A Chr01wlogical Outlook. Kyoto: Kyoto Universit)4
[nstitute fo r Research in Humanities, 1997.
- -. "The Hephthalites in Tokarirum and Gandha1-a; Part 1: Ganclhara." Lahore
Museum Bulletin V, no. I ( 1992): l- 24.
- -. "The H ephthalit~-s in Tokharistan and Gandhara; Part D: Tokharistan." lA/wfl
Museum Bulletin V, oo. 2 ( 1992): 1- 18.
- -. "The Buddha's Bowl in Gandhara and Relewmt Problerns." In South Asia~
Arch"'wlogy 1987, edited by M.. Taddei, 945--78. Rome: WvlEO, 1990.
- -. "Ti•pa Shotor and Lalma: Aspects of the Stupa Court at Hadda." Arma1i
dell'lstitulo Uniuersitaria Oriet11ale di Napoli 47 (1987): 153- 76.
- -. " Kapisi and Gandhara Acco•·ding to C hinese Buddhist Sources." Orient lll
(1982): 133- 39.
Larson, E., and P. Birkeland. Pl111111m!s Geology. 4th ed. New York: Oxford University
.Press, 1982.
Liu, X. AncienJ h1dia. and Ancii!Til Clli11t1: Trade a111i Religious Exclwnges ;1.D. 1- 600. Delhi!
Oxford University Press, 1988.
Lohuizen-De Leeuw, J. E. van. "New Evidence with Regard to the Origin of r.b.e
Buddha Image." ln South Asian Archaeology, edited by H. Hanel, 377- 40 l. Berlil'l;l
19 79.
- -. "The Second Century of the Kaniska Era." Paper presented at the Seminar
on late Kushana Numismatics, London 1981.
- -. 111e "Scytllia11" Period: A11 Apt1roadt to tire Hiswry, Art, Epigmpi!J mul PalneograP'r/
of North India .fi7Jm Lite 1st Cenhl')' B. C. to the 3rd Century A.D. Leiden: E. J. B~
1949.
M:acDowaJJ, D. "The Chronological Evidence of Coins in Stitj>a DepositS." ln Soul/t
Asum Archaeolo,{!' 1987, edited by M. Taddei, 727- 35. Rome: IsMEO, 1990. '
MacDowaiJ, D., atld M. Taddei. "The Early Historic Period: Achaemenids an4
Greeks." [n 111t Archaeology qf Afghanistan .fiY>m Earliest Tvrws w tJu 71murid Periodj
edited by R. Allchin and N. Hanunond. London: Academic Press, 1.978.
- -. "The Pre-Muslim Period." lo 17It ArcJweolOg)' qf A.fi,hanistan from Earliest Tvnts
to lire Ttmurid Period, edited by R. Allchin and N. Harmnoud. London: Academic
Press, 1978.
MarshaJJ, J. Archaeological Guide 1o Ta~:ila. 4th ed. J odhpur: Scientific Publishers, 19601
Copyrighted mater 31
l
•
LNDEX
Abba Sahib China, 27, 38. 203- 204. Archaeological Su rvey of India (ASI),
207, 2 10, 255; colossal image shrine I , 17- 18, 113- 11 4. I lB. 188.
at, 137. 203. 245, 253- 254. 267, 2 16-217. 2N, 269. 296- 297. 302
280, fig. 111 ASokan edicts, 13, 21., 39. 235:
abhaya mudrii, 168, 172, 251; as columns, 55; sttipas erected by,
chronologic indicator, 274- 276 39- 40
Adashir, 2.i6 Asokavadana, 39.
additive sculpture, 32- 33, 69, 71, 110, assembly hall, 35, 76. 198: in situ
159- 161, 165, 2 17. 229. 246; typical sculpture found in, 23 1. 241, 249,
distribution of, 161; placement in fig. 6.1
pre-existing shrines, 161 , 233; sockets a tlantes, 117, 126- 127, 165, 243, 247;
for anachrnem of, 2 18, 247; schist depictions of on phase lll stupas,
iconic sculpture, 232, 247; figs. 28. 9fi
monwnental iconic sculpture, 190, Aurangabad caves, l.ll.. 251
218 Azes, 68, 87, 97, 192, 25 7- 259
Afghanistan, 2, 4, 10- 12, 14- 16, Azes 0 , 49. 68, 2iZ
20- 25, ;g, 39- 40, 79, 92- 93, I 04, Azes Dheri (see Cangu Dher)
152, 203, 208- 2 10, 218. 229, Azilises, 2iZ
254- 255. 261- 262. 264, 269, 271,
276-277, 30.1 Bactria, 2..2., 1.5.
A.i Khanoum, !1.,. 271 Bagh, lfi1
Aja~o!Ji., !...!., 37, l36, 171 , 174, 252, Baligram, g.§, !ill
283, 2Bfi Ballar rope, g.§, 80, 162, 207, 255
Akhaurf, Khacler Mollfii sites, 25, 34, Bantiylin, u., 14, 20- 22, 2 1o. 255,
36. ?1., 80, 93- 95, 138, 142 277; monumell!al devotional icons
Akhauri B, 74, 167, 251 at, 203- 204, 253, 267; pass at 2.3
Alchon Htm, 202, 208, 2'16 Barikot (ancient city of), 19, 26, 46
Alexander the Great, l2 Barthoux, J., 279
Ali M~jid, 277; phase UI stucco Basa,va.l, 2.1
stupas at, 110, 127, 163-165, 20 1, Bayley, Sir Clive, ill
265, f·ig. 9fi Bazira (see Barikot)
alms bowl of the Buddha, 10. 15, 30. Begram (ancient city of), I l l
61- 64, 175, 236, 238; 6th century Bellew, H. W. , 17, 11 5. 18 1, 187,
C.E. pilgrimage to, 206; depictions 216-2 17, 230, 291
of, 238, figs. 122, 123 Bhaja, 239
Amaravati, 130. 273 Bhamala, 74, 80, 15 1, 162, 206.
Amlukdara, 21 207- 208, 2 10, 254, 263, 264,
awfa, 28, 52- 54. 95 266-267, 279
Anda.n Dheri, 77. 21i.l Bharhut, !i, 15, 53, 55, 57- 59, 124,
Andhra, 8, 14, 55, 130, 273 273
aniconic period, 273 BhiJsa topes, 5 L 53. 55, 58. 93, 236,
apsidal haU (see cairya hall) 241
Copyrighted material
324 INDEX
Copyrighted material
INDEX 325
326 INDEX
monastery, 33, 9~. 139, 145 146, Gaoseogzhuan (The Lives of Eminent
167, 263; J quadrangular monastery, ~Ionks, eclited 519 C.E.), 62 63
9~. 139; H quadrangular monastery, garland holder, 57. fig. 12; io situ
94, 139, 170: 1\15 quadrangular placement of, 58, 60, 237. fig. 114
monastery, 1-!5 146, 165; satellite Garrick, H., 19
monastic complexes, 93- 9-1, 250; Gautama. 59
monastic population through time, Ghazni, 194
93 9+, 138 139 Giri. 77. 79, 206
Dipankara Jtitaka, 129 130 gold leaf, 109, 172
Dir valley, 23 Gondophares. 182, 25 7
Gujarat, 11 , 170, 252
eanhquakcs, 207, 242, 290 Guldara, 20, 209, 261
Ellora, 11 , 171, 252 Gupta period, 55, 152, 282 283
Ephthalitcs (sec Hcphthalitcs) Grant Skeoe, 120
Errington, Elizabeth, 197, 206 Grcco-Roman style, 7, 16, 50
Greece, 271
Facccn na, Domcnico, 19, 47, 122, Grlinwedel, 128
207. 279 Guides mess, Mardan, 16
liusc dormer (sec false gabll') Gumbatuna, 20, 27, 56
l>tlsc gable, :ll , I I 0 I I I, I I 7,
122- 123, 128 132-133, 179, 217, Haq(la, 20- 21, 61- 65, 277, 279, 30 I;
220, 224, 227. 243- 244, 253, satellite monastic complexes of, 93,
297 304; lilc sJJan of motif; 110, 24 1
133, 174, 214, 244; placement on Haibak, 2 1
sltipru, 214; depictions of; 132; halo, 133; double halo, 222
multi-piece examples, 11 8, 123, 133, Hargreaves, H., 181
2 14, 244, 297; iconic images in harmikD., 28, 54, 110, 117, 122,
tri-lobed niches, 144 145; placement 133- 13 4, 165, 179, 298, 300 30 1,
in monastk cell, 169; incorporation 303; life span of motif, 110, 214;
in Srnvastf composite reliefs, 174, setS, 118, 123, 133. 213. figs. 87.
214, fig. 127 9S B; iconographic developments s<:<'ll
F;uncnsi, 65 in. 276
Faxian. I. 15, 22, 61 65. 175, 239 Hashmagiir Buddha image with
figure carrying garland motif, 110, inscribed date, 282
118, 123, 127' 1.10, fig. 95£ Hazar-sum, 21
figure under arch motif, I 16, I 18, Hellenistic traclirion, 7. 12. 57
120. 123. 127, 129, 132, 243, 298, Hephthalites, I 0, 15, 23. 205; role in
ftg. 121 supposed destruction of Buddhist
Fil-khana, 21 community, 204- 205
Fit.GSimmons, T., 208 Higuchi. T .. 191 , 194
Fondukistan. 20, 255, 277 Himalayan mountain range, 22, 271
forgeries of Gandharan sculpture, 269 Hinayrma Buddhism, 14
Hindu Kush mountain range, 22 23
Gandhlira, ancient (sec also Pcshawar Hindu Shahis, 15, 256
basin), 1, 10 11, 16, 25 Hindu temple, 68
Gandhara, Grcatco·, 2, 6, fl, 12 13, Huo:tas (see Hephthalites)
15 16, 22, 26, 65, 78, 95, I 09, 142, Hun, 206- 207, 256, 271
146, 149, 157, 163, 166, 174, 187, Hunza valley, 23, 25
194, 235, 253, 255 ·256, 267, 27 1, Huv~a, 13, 49, 103, 177, 192, 237,
27 7, 295; histooy or, 12 13; 255- 25 7, 263
geography ut; '.!2 '.!4·
Gand hari language, 35 image shrine[s), 3, 9, 14, 27, 31 ·33,
Ganges basin, 40, 50 37, 66, 103- 104, 11 0, 11 7, 135,
Gangu Dht:r, 2<l, 176 139- 143, 147, 156-160, 183, 189,
arep11
•
328 J.:NOEX
82; single- and double-cellcd shrines, fig. 86; phase m sculpture, !...!.±.
!lQ, ~ 166; phase lll emphasis on 136. 177. 217; late phase m and
relic shrines, 15-1- 155, 157, 162; phase IV sculpture, I 7i. 187, 248,
organization of sacred area, 285; multi-piece false gable, 133;
105 106. 1-}2, 240; additive imagery, Buddha wirh inscribed date, 282:
160; rcUSI'd sculpture, 289. 291 292: loose sculprure, ?96- ?97
monasteries, 167 168, 170. 251; Lowari Pass, 21
monastic population, 138; monastic Lumbini. l l
personal devotional image, Ui8
K~herr, l1 55, 1.61 ~ Iahii)'iiml Buddhism. 14, 284, 2B1
K~ka. 13, ~ 2iz 93, I 05. 177. ~[ala Tangai, l! 3
192, 2H. 237, 256 257. 263. 273, ~ lalakand Pass 21
282 283 , 285; era, elate of, -} 5; ~lal,likydla. stupa.
17, 26, 255;
Kani.5ka 11, c:ra of, 282 283; phase IV relic deposits in, 207.
Kaniljka 's st u/J<I, ll.5. ru:
kai\jur (sec also masonry); use in cores manuscript fragments, 2, 15 '16
of phase lll stucco sculptures, 163 Mara, 275
K;u:~ ~haka, 218 Mardan , 228
Kal'llkoram Pass, 23, 25, 1.311 Matjanai, 20, 27, 11 74, 77, 79, 98,
Karkai, 97, I 15, I UJ 120 11 5, 120- 121; phase rr, 105 106,
Karli , LJ I 08; two-cellecl sWpa slui.ne S3, I06,
Kashmir, 2, 16, 22 23. 2011, 25.6 125, 229; reused sculpture, 290 291, ·
Kashmir-Smast, 20 2.9.3. '
Khadcr Mnhra 0 2, 96, 108, lAli. Marshall , S ir J olm, 5 6, 18 19, ~ 58,
Kham51.h i, 35, l.3ll 60. 66-68, 70- 72, 75. 82- 83, 2.Q,
Khusrau TI, 25.6 92. 114, 143, 173, 205. 207' 256,
K hybcr Pass, ll.. ill 278, 279
Kicrn rn. 2fi3. masonry, 5-6. I 76. 244, figs. ;!.. lt
Kida ra 11, I..!UI phase I, 258-259: phase 1 rubble
K ida rites, 15 masonry from T axila, ±z. 25 7. 260.
Koi Tangai. 96, ill 267; phase I kaiijur ashlar masonry
Kujula Kaclphises, ~ 255. 25.1 from Ta.xila, 42. 44, 46, §Q, ~ ~
K uoiila, ?.L 142, 162, 206 207. 213; 88. 25 7- 259; kaiijur ashlar masonry
phase Ill foundation, 139; as a reflection of patronage rather
monastery, 167, 169, 265 266; as than chronology, 259: phase IJ.
satellite monastic complex, 9-1-, 139 260 261: phase li diaper masonry
K ushan. 13, ~ 107. 237. 256 257. from Ta.xila, 28, 47, 69. ~ 82, 2.!.,
271' 273 ~ 142, I 44, 159. 238. 257- 261,
Kushano.Sasanian•, 15, 105. 177. 192, 262, 264-. 278: diaper masonry i11
245, 25fi the Peshawar basin. 6-7, 96, 141,
Kuwayama. Shoshin, 21. 206. 208, I & 260- 261; diaper masonry in
2lD Swat, 103. 105, 261; soapstone
ashlar masonry in Swat. I 03; early
Lahore M useum, 1, 18. 2.1 112. 118. phase m proto semi-ashlar masonry
228, 239 from Taxila, 141 , 144 145, 264;
La lchak, 26. 74-, 162, 2!lfi phase m semi-ashlar masonry from
Lalma, 2.J Taxila, 8. 2.!., 136 137, 141 142,
Lauriya Nandan~'llrh, !i!i /-14, 148- 149, 157, 164, !1..!., 245,
Locban r I, I!, Jll , ..9.9. 26 1, 262- 265, 277; late phase Ill
Loriyiul 'l'l!J)gai, 2 11 , 212 215; phase double and triple semi-ashlar
fl sculpture, 96, 11 3, 123, 177, 2 /J. ma.sonry from Taxila, 13 7, 14 1, 145,
227; phase TT stupas, 179, 2 /.~; small 147- 150, 153, 158, 208, 245 N6,
stii.pa reconstruction now in the 262, 264, 266, 267, 277; phase IV
lndi<Ul Museum , 124, 2 13, 218, masonry, 266- 267; interstitial chips,
INDEX 329
261 , fig. 9; incorporation of 196- 197, 249, 262, 266-267, fig. 61;
sculptw·c as building material, 2.89 images associated with rnoumain
Masson, 16 uil1iiras, 195. 233, fig. Ifi
Mathura, 272- 273 mona~tic area, 27. 80, 11 9. 1..35
Maurynn period, 13, 24·, 39, <Ul. 49. monaslic image sh rin es, 36, 80, 149.
235. 2.55 162- 163, 166-174, lTh 251, 253;
Maus, 2.'i.2 monastic personal devotional images,
Mediterranc<tn, !1, 22. 50, 56, 36. 163, 167- 169. J95, 251' figs. 23.
270- 271 27: groups of icons placed in
Mckhasanda, 20, 2•1·, 35, 38. 77. pre-existing monastic cells, 163,
136- 137. 143, 232, 261: phase n, 169- 171. 251; monastic pedestal
96. 177- 180. 183, 192; phase rn, images, 36. 163, 169- 171. 251 - 252;
150. 156- 158, 177- 180, 221, 229: monastic gandltaku!i image niches,
late phase m and phase rv, 178, 36-37, 136,155. 163, 167,171- 174.
187. 205: relative chronology or, 252, figs. 27. 29: chronologic
178- 179; in situ sculpture, 180, 21 I , relationship between pedestal images
231. figs, 66. 67: loose sculpture, and gandllal111ti image niches, 172
137. 2 11- 212. 226, 277. 280. 286. monastic populations at given sites, 29,
30 I. 30.~: reused sculpture, 290- 291 43, 93- 94, I 0 I , 138- 140. 181, 188,
1vl.ian Khan, 17, 96, ll3, ill 191, 197, 24:9
J\l.lingora, 1..02 monastic relic shrines, 61. 66. 166-167;
Mir Jan , ll.3 stiipa shrines in monastic ccUs, 36.
Mi.rpurkhas, ll 44, 95, 14·6, 166- 167, 169, 24·0;
Mizuno, Sciichi, 20 191, 194 two-cclled relic shrines in
Mohammed ari, Sr:ivasti composite monasteries, 36. 80, 166; sllipas in
reliefs from, 172, 187. figs. 126, 127 monastic courtyards, 95-96, I 08,
Mohenjo Daro, di rect access sltipa, 64, 146, 166-167, 240, 250
239 monastic smaU sacred area, 27. 37,
Mohra Ma.lia.ra.ii, 64- 65. 68-69, 136. 163, 169. 193- 197, 249- 250;
73- 74·, ~ 100, 153, 236 private devotional needs of monastic
Moh ra Moradu , 33. 36. 69, 74, 77. comm unity, 166- 167
79. 80, 94, 136. 159- 161' 233; Mora Pass, 212
phase IJ, 94- 95; phase m, 142, mountain viJUira (see monastery)
156, 162- 163, 245, 250, 255: StuCCO mudrii, 1 I0; relative chronological
imagery, 141, 277. 280: additive development or,
274- 276, 27ll
imagery, 159- 161, 229: rwo-ceUed naga, 2.86
relic shrine, 94-95, 159;
quadrangular monastery, 159, 183, agarahara , 2 3-2 4
262; stiipa added to pre-existing Nag-a:rj unakor:rc)a, 27 3
monastic cell, 165, 167. 169: Naj~>ram, 65
devotional icons added to alandii, 146. L71. 251 - 252
pre-existing monastic cell, 169. 1alhou complex, 17, 24, 96. lid,
fig. 38; monastic gandlwkuti shrines, 115- 118. 122- 123, 177: Lower
171- 174. 184, 252, 287: monastic alhou, 116- Ul, 123, 213; upper
pedestal images, 169, I 7'2 N:.thou, W - 118. 122- 123, 126,
monastery, 27- 28, 33- 38, 76, 80, 82, w
84-85, 93-94, 136. 195: phase (, imogram, 20, 27. .'>4. 106. 135. 261:
43-44, 46, 21.. ~ 55, §§_, 84- 85. phase III, 150. ill
138. 166, 240, 259; phase ID, niroii11a, 8.3.
135- 137, 140, 162- 163, 166-171, Nishikawa, Koji, 20
250- 251' 265- 266; late phase m, numismatic evidence, 3- 4, 10- 12, 15,
1 11- 114; phase IV , 209, 267; 16, 41, 45, 49. ~ 68, m,
2§,
mountain oihiim, 33, 37- 38, I '16. 97- 98, 101-103, 105. 111, 137, 145,
139- 140, 163, 188- 189. 194- 195, 156, 176. 177, 183, 191, 194,
Copyrighted material
330 INDEX
1..9.6- 198. 200, 202. 207, 222. 227, l LO, I 12, I 19, 129, 135-137. 139,
235. 238, 244, 248, 253- 256. 259, 147, 155-156, 158, 163, 165-166,
262- 263, 269. 27 1- 272, 279, 292. 175-2.15. 24-1- 242, 244·, 246,
304. fig. z 248- 249, 253-255. 261' 265. 267,
269. 273, 276-277, 28 I . 284- 286.
Palai , 2.8..1 290
Panr, 19, ~ 42, 77, 94, 99. 103, 105. Peshawar Musetml, I , 18, 112. 300
270; phase 11, 101- 102, 108 Peshawar University, 20
Parinitvfu:ta, I I 8. 121, 123, 2 I 3; phase phase system (see cllronology, phase
m and IV iconic representations of, system)
264, 279, fig. 43 pilgrimage, 10. 27. 6 1. 62- 63, 64. 94
Parthim-t, 13, 23. 45, 50, 272 (see also Chinese pilgrims}
patronage; periods of abwldanCc, 7, pillar (see column)
14, 77, 107, 135-136, 235, 273- 274; Pippala, 25, 36. ll. 79; phase 11,
phase m increase in, 137- .141. 142, 95-96. LOB; ph a.se Ul, 136. 142,
175, 187- 188, 245, 250, 253. 255; 156, 162, J69. 245; monastic slupa
phase IV decline in, 8, I 0, 15, 141 , shrine and phase I] stfipa, 98, 12 l,
175, 189, 204- 205. 245, 253, 267, 12:-r-126, 165, 167, fig. 1(1
295; clu·onologic gaps in, 2, 205; prada}qitta (see circumambulation)
Auctuations in at given sites, 162; for provenance of sculpture, 11 2- 1 14
drama tic effect, l 04. i'l·8; by la)'
community, 83, 96; monastic conb·ol Old $a.ka era 2BJ
over, 84- 85; of main sliipas and oew '
sacred areas, 14, 28. 34, 81, 96. Ranig-dt, 20, 21. 53- 54, 77' 96. l 97.
107, I 84, 190; of secondary sacred 199, 213, 261 :. phase l and II,
a.reas, 189- 19 1; of srrucrures added 97- 99, 108, 183, 192, 200- 204:
w sacred areas, 59. 78. I 03. ph ase 11 stripa.r, 121, 125, 126,
.188- 191. 23.5- 236, 250; paLteJ'tlS of f'S. 80; phase m, 136. 150. 157.
clustered offerings, 90. 146-14 7. I 75, 200- 204, 245, 255; late phase
153, 189; of sma.ll stupas, 28- 29, 41 , lll and phase IV, 137. 166, 187,
49, 2.L 58. 84. !...!.L 122- 123, 127, 200- 210, 248; phase lV colossal
157, 184, l9Q-19l, 242, 247; of image shrine, 137, 245, 254. 26 7,
relic strucmres, 30. 7.5. 79. 83. 84. 280: multiple sacred areas, 189, 191,
I 90, 274; of sculpture in relic 200, 202- 203; in situ sculpture, 279;
shrines, 69, 72, 105- 106, 118, loose sculpture, 137, I 77. 226-227,
2!9- 220, 230; of columns, :2.L 59; .10FI04
of image shrines, 32. 132, 15 7, 162, Ratnagiri, 29, !l.L 25 2
184- 185, 246, 274; of' rnonurnental r·cfurbishment during phase lU of
image shrines, I 86. 189- I 90; of earlier st:ructures, 147, 149, 175.
sculpture, iU., 58- 59, 190, 217, 235, 178, 182, 242, 248, 289- 290
274; of iconic images, lli- 2 15; of relic, 7- 8, JJ- 14, 28, 30, 43, 60- 61,
additive sculprur•e, 33, 16 1, 2•n ; of 72- 73, 25-76. 108, 135, 189, 238,
reused sculpture, 87, l1.L 106, 288; 274- 275, 291: rel igious emphasL~ on
of pavement stones with embedded during phase IT, 107- I 08; emphasis
coins, 98. I 97. 20 I: of monaste ries, on during phase []], l...'ill
28, 34, J84, 265: of satellite relic, deposits in main Jtupas, 4. LO. 28.
monastic centers, 94; of images in 30, ~ 4:6, 147, 207, 254, 256, 259,
monasteries, 168- 169 262- 263; in small stfipas, •1·2
Patuli, 5 1, 273 relic, direct-access, 15, 6.1- 70, 75, BB.
paving stones with holes for coin 100- 10 1, 143, 238
donation, 98. 197, 202. 279, fig. 8 1 relic, securit)• for, 6 1, 64. 66- 67. 72,
Peshawar basin , 6, 9 - 10, I 5- 16, 20, 76, 88, l 05; restricted tu:ccss in relk
22- 23. 24- 27, 33, 37. 39.. 53. 54, shrines, 67, 69 - 70, 73- 74
62. 73, 77- 79. 92-94, llii-99. 107, relk shri ne, 8, ~ 14, 27, 32, 6.1- 76.
'
I
I
INDEX 331
I
•
79. 143, 149, 245, 247; sculptural 189, 230, 238- 239, 241; phase III,
depictions of, 56, 61, 63- 64, 73, 89. 1.15- 137. 156, 161 , 163, 189, 230,
!.!.L 165, figs. ~ I08, 118, Ltl, 240, 25l; phase nr ex,ans ion of,
I 122, 123; direct-access relic shrine, 187- 191. 253; phase IV, 203- 201.
30, 73- 76, 236; stupa shrine, 30. !ill, 209- 210, 246, 253; evidence of
11 5, 125, 139, 154- 155, 229; phase IV collapse in patronage of,
two-ccUcd stiipa shrine, 30- 31. 205- 206, 253: core r·elic su·ucrur·es
•
73- 76, 78, 81, ill!, 104- 105, 109, in, l 06, 136; multiple or subsidiary,
120. 155. I 85, 239, 259. 26 1, 29. ~ 101, 140, 147. 159,
fig. 107; two-ceUed shrine, 73-76, 189- 191, 192, 199, 220- 221, 249;
W, 82, 85. 87. 95. I00, I 09, distribution of structures in, 78,
139- 140, 155, 162, 185- 186, 192, 84- 85, 92, I 04- 105, 136, 162, 179,
228- 229, 236, 239- 24·0, 242, 261. 235; organiUltion for dramatic effect,
285, 294, 300; two-cclled shrine wi1J1 I 04. l 07- 108, l 36, 179, 185, 198,
loose relics, 74- 75; single-celled stiipo. 204; movement through (see also
shrine, 91, 109, 1. 20. 152. 155, Dharmarajika complex and Butkara
228-229, 239; single-cclled shrine, I), 32. 42, ZQ,!!:!, 198- 199. 236,
90, 92- 93, 109, 157' J85, 192- 193, 240; with relic shrines at entrances,
198: importance of in phase m 186; with perimeter image shrines,
sacred areas, 154- 156, 185; 158, 165, 179, 247; unenclosed, 160,
establishing a relic in, 75 . 83: image 193, 229: assemblages of iconic
shrines added to pre-existing relic sculpture found in, ill
shrines, 161, 185, 247; in situ Sahri-Bahl61, 17, 24. 26, 94, 114. 176.
imagery associated with, 228, figs. 187, 207, 255; in situ sculpture 21 I,
55. 90: additive sculpture placed in, 229, 230- 231, figs. 54·, ~ additive
161, 219-220; reused sculpture imagery, 160; loose sculpture, 21 1,
associated wid1, 289. 29?-?94 216, 221- 223, 227, 278, 280,
rcliquaty, 16, 70, 72- 73, 106. 11 2. 284- 285. 300- 302: late sculprure,
262, 29 1- 292; stupo.!reliquary, 62, 177, 248, 276- 277, 281 ; sculptural
75- 76, llii reuse, 189, 229, 253, 289- 290;
reused sculpture (sec also appendix C), satellite monasric complexes, 93, 181,
!..Q, 87. 91 . !.J..:!i 118, 120, 178, 189, 22 l. 241; Sah t·i-Bahl61 A, 137. 21 1,
190, 211, 2 14, 216, 219- 220, 230, 229, 294, 300; Sahrr-Bahlol B,
249, 28 L 298, 30 L fu-st period of /63-164, 206, 222. 300- 301:
reuse in phase m, 199. 242, 289. Sahri-BahJol C, 222- 22.'1, 286,
290- 292; second period of reuse in 294- 295, 301, fig. 58; Sahri-Bahlol
phase IV, 205, 210. 22 1. 223, 228, D, 223, 286. 294- 295. 301- 302;
242, 253, 286, 289, 292- 295; as a Sahri-Bahlol F, 206; Sahri-Bahlol G,
reflection of phase IV Buddhist 208, 254-. 267, 295, fJg. fill
praclicc, 253; sites not subjected to Saidu, 19, ~ 55, 77. 94, 99, 115,
reuse, 225- 226, 229; placed in relic 135; phase U, 102- 104; phase III,
shrines, 87, 161, 219- 220, 244, 285. 150, 189, 255; main stiipa,
288, 292- 293; placed around sllipa 102- 103, 109, 115. 122, fig. 105;
bases, 223, 288, fig. 58; confusion monastery, I 0 1; trabeated
with addilive images, 160, 246-247; cl1~.rnbe rs, 187; in situ sculpture,
possibly damaged in earthquakes, 233; sculpture, 270
207, 243; confusion with Saka era, 282- 283
iconoclasm, 2.0.1 Saka-Parthian, 2.3
rock-cut sculpture, 2 10. 255, 2ftfi Saka period, 45
Rome, 271 - 272 Sakyamuni, 14, ·10, I 72
Sai'icr 8, 40, 52- 53, 55, 58- 59. 1 2·~.
sacred area, public, 27, 32- 33, 37, 42, 236, 273; axial image sh rines on
44, •1·5, §.!., 83. 102- 104, 135. !1!., main stiipa, 145; temple I 7
185, 197, 275; phase n, 77- 108, identification as relic shrine, 239;
sateiEte monastic complexes of, 93, 247, 279, 287. 300, f.gs. 28. 43, 46,
24· 1 48, 49, :ill, 56, 72, 73, 8 1, 9fi
Sanghao sites (Sanghao Rhode, sculpture, monumental, 8- LL 14, 7./,
Sanghao Cbinlai, Sanghao Nullah), 137, 153. 174. 180, 186. 212, 2 15.
I 7, I 13; narrative sets from, l11 2 18- 219, 226, 23 1, 253. 266, 269,
SankMya, 2 73 276-277, 279, 280- 281. 283. 289,
Sarnath, 40. 5.3 295, 298, 30 1- 303; triad images,
Satv.islavada, Buddhist sect, B3 203; popularity of stucco media for,
Sasanian, 15, 23, 192, 206. 245, 251i 2H£l
schist, ~ 18, 57. I 16; used to r sculpture, narrative, 7- 9. 14, 31, 57,
sculpture, I 10, 158, 163, 213. 224, 59- 60, 64, 74, 96. 107 - 108,
227. 277- 278, 279- 280, 298, 109- 134, 176-177, 212. 223-224,
3illl- 304; durability of sculptures 227, 233, 237- 238. 262, 273- 275,
executed in, 21 I , 249, 290; 278- 2 79. 284- 285, 21l9- 290, 292.
suitability for reused sculpture, 21 1, 294, 296-304: phase n distribution
288; as building ma terial in and religious function of, 107,
Peshawa r basin, 109, 176. 251; as I 09- 111, 121, 216 ; in situ and
building material in Swllt, l 09, 261 ; original placement of, 1J 4- 121,
see also masomy 127- 132. 228- 229, 242; sequence of
sculptu re, phase 1, J9- 50, 199. 237; narrative events, 129: phase ll
phase n, 109- 134, 176. 196. narrative sets, 106. 114, 116, 120.
LillH 99. 201, 2 12- 213, 216. 219, 128, 131. 213, 220. 228, 242, figs,
224- 225, 242- 243, 245; phase m, 85, 86; phase Il narrative sets in
176. 178, 211- 233. 245; late phase stilpa slu-ines, 106 .. 115, 119.
m, 176. 186- 188. 2 1s . 218- 224, 120- 12 1, fi!:,'S. 90. 9 I); phase Il
227; phase rv, 176, 216, 2 19 (see narrative sets on main stiipas 103,
also chronology, phase system 109, 115, 242; scale of, 109, 115,
applied to sculpture, reused 242; impo,·ta6on of, I 09; •lle.'Uls of
sculpture, rock-cut sculpture) attachment of, 124; exmnples
sculpture, iconic, 3, 8-9, 14, 31- 32, executed in stucco, 1 I. 0, 233, 24 7,
I 10, I I 7, 135-137, 142, 156, 161, 278. fig. 72; iconographic
164- 165. 196, 2 12. 214, 216, 220, development in, 274- 276, 279; lack
226-227' 24<~ -245, 249, 265. of in situ sculptural evidence tor
275-276, 278, 2S..1, 289, 292, 29 7. phase Il .!liipas, 122, 242
300; iconographic developmeors of, sculpmre workshops, 112 . 173, 270, 281
171- 174. 274 - 276, 279, 287; scale Scylhians, 13
in relation to chronology, 137, 214; semi-lunette (see false gable)
occurrence as means of dating Senan , M., 113, 242
arclutecture, I 76-177· depictions of Shiibazgarlll, 24, 39. I 77 182
relics on bases as dating indicator, Shah-ji-ki-cJh eri, 17, 24, 77, 114. 168,
280. fig. 123; with Late icono!." -aphic 176. 277: phase II, 96.. 158; phase
characteristics, 206. 2 15. 21 8; rv. 2illi-20S, 21 o, 254- 255, 267;
appearance in tri-lobed frames, 133, identification as Kan~ka's stUfia, ill
ftg. 18; triads, 2 17: placement io Shaikban-dheri, 1l.fi
relation to relic structures 161- I 62; Shapur II, 263
placement in monasteries, 162-169: Shin~:,rardar, 21.
fig. 3fi Slmaisha, 20, 2 7, 56, ?2 201>--207
sculpture, in situ evidence, 268- 269, Shotorak, 277, 279
288; schist in1ag~-s, 35- 36, 58. 60. Siddhartha, 21.8
104- 105. 109, 168. 211, 221 . Sikri, 17, l.9.. 113. 115. 136, 19 1, 19 7.
228- 229, 232, figs. il, g 71, 91; 225, 228--229; phase 11 sculpture,
stucco images, 110, 14 1, 144- 145, 96, 1/8-1/9, 177, fi!:,>"S- 90, 9 I;
169- 174. 194, 218. 222, 232- 233, phase ill, 157; in s:in• sculpture,
ill 228- 229; similarity of sacred executed in, 133, 160- 16 1, 2 18,
area to Tbarcli, 193; decline in 277; placement of additive images
patronage o( 206: single-cdled sl!ipa executed in, 159- 161, 2 18;
shrine and stilpa now in Lahore embellishmen t of phase [[( stripas,
Museum, lU., 119. 123, 131. !1L I 63-·166, 199, 233, 290, 300;
228-229, 239 on the bases of image shrines, 179,
Sind, 10- l l, 64, 239 fig. 73; narrative depictions, 110,
Sirkap, 7, 25. ~ 64- 66. 73, 85--86, 233, 2.J.B.
88, 166, 239; phase I, 43, 45-47, structu ral typology, 6, 10, 27- 38, 109,
53- 58, 60, 237, 255, 257, 259, 268, I 76, fig. 5, appendix A; phase I, 46,
271; lA sllipa and complex, 43, 66, 2.!., 60, 259; of phase I monasteries,
84, C stiipa and complex, 0 . aps.idal 259: of phase I small stilpas, 259; of
hall, 47, §1, 69 - 71, 72, 74, 77, 236; phase I main stiipas, 98, I 36, 259;
fig. 1?)53; F st.Upa (Double Eagle) phase U, 78- 79, 109, 261; of phase
and complex, 43, 46, 52- 53 11 main stilpas, 97 - 98, I02- 103, I09,
Sirs u.kh , 1.!1:!1: 192, figs. I!!, ~ l 05: of phase II
Sita-Hii1~a king Khingila, 282. 2.8:! small siUpa, 95- 96 , 98. 119. 262,
Soter Megas, I 03 104 fig. !t phase m, 135- l 37. 265; of
Spooner, D., 181 phase Ul main suipa, 136, I58, 179,
$ravasti sculpture, 2 12, 217. 2 19, 226, 183; of phase III small st.Upa,
269, 293. 304: daring arguments, 126- 127, 165-167, 179, 265, figs.
268- 269. 284- 287; Sravasrf Miracle 28, 96; of phase Ill sacred ar ea,
narrative images, 284; Sravastl .156, 158, I 83; of phase TV
plaques, 168, 2.19, 223, 285- 286, cruciform stilpas, 208, 26 7, fig. 83:
289. 292. 299, 301- 302, fig. 53; of phase TT and UI qt•adrangular
Sravasrf triad images, 168, l72- 174, monasteries, 183, 265-266: of
253, 284, 289, 292, 294- 295, mountain uihiiras, 194- 195, 262,
297- 298, 30 1, fig. 125; Sravastl 265- 267; of phase IV domed
triad with inscribed date of 5, 283, quadrangular monasteries, 209, 267,
287: Sravasu composite imagcs;l72, fig. 1I2: of siJlgle-celled shrines, 91.
174, 2 15. 227, 253, 276, 283- 286, 11 7- 118, 142- 143, 192, 265: of
289, 292. 297- 299, 301-302. lig. two-celled shrines, 265- 266: of
I 26, I 27: relation to f.'llse gable image shrines, 9 1, I I 7- 118
sculptures and their placement, 174. 142-143, 177 265: of monumental
214, 223, 244; Mohammed N:ui image shrines, 177 180, 266; of
Sravasti compo;ite image, 287, colossal image shri nes, 203, 267,
frg. I 26 fig. I 11; of phase Jl and III
Sri Lanka, 113. quadrangular monasteries, 183,
stacked narrative reliefs, 124, 129, 265- 266; of mountain uilziiras,
fig. 121 194- 195, 262, 265- 267: of monastic
Stein, A., I 82 small sacred a reas, 19.5=1.97
stucco sculpture, 9, !..1 18, 32, 58, shipa, direct access ma.in, 31, 64,
68- 69, 71' 177' 2 17' 223, 226- 227' 70- 72, 78, 81- 82, 88- 89, 108, 155,
242, 296, 299- 300, 302- 304; phase 236, 239
Jn and IV dating for majo•·ity c•f stiij>a, maill, ~ I 3, l6,
21- 28, 32- 3<1·1
production, 110, 180, 201- 202, 219, 39-40, 42- 43, 48, 60, 80, 85- 87'
233, 266, 274·, 277- 281. construction ~ ~ 95-98, 102- 104, 125,
techniques, J63- 164; regional tL~e for 148- 150, 161- 162, 192- 193,
sculpture, 110, 249; survi_,.al in 198- 199. 229: phase I, 4·1, '~5-52,
archaeo]()gical record, 21 L 223. 30 1; §Q, I 36, 236-237, 259; phase II,
scarcity of at some sites, 2 15; 108, 109, 136. 200- 201, 240;
suitability for reuse, i l l 288; phase ill, 136, 158, 2 12, 2 18, 25 l ;
socketS for the support of images relic placement in, 72- 73, 106;
Copyrighted material
334 lNDEX
Copyrighted materi l
INDEX 335