CORINNA WESSELS-MEVISSEN

The Gods of the Directions in Ancient India

Origin and Early Development

in Art and Literature
(until c. l000 A.D.)

DIETRICH REIMER
BERLIN 200 1
 Contents

Acknowledgements .................................... ....................................................................................................................................... IX

List of Tables .......................................................................................................................................................................................
XI
List of Illustrations .............................................................................................................................................................................
XI1
List of Figures ......................................................................................................................................................................................
XI11

I 1 Literary Evidence .........................................................................................................................................................................

4
The Vedic Literature .......................................................................................................................................................
Early Vedic Literature .........................................................................................................................................................
Origin and Sylnbolism of the Direcdons .............................................................................................................................
The Classification of the Directions ....................................................................................................................................
The Early Assignment of the Directions ................................. . . .......................................................................................
Evidence from the Br%hmatps,Aranyakas, Upanisads, and Siitras ....................................................................................
The Tripartite World (triloka) and the Directions ..............................................................................................................
The Classes of Beings and the var!las ................................................................................................................................
The Sacrificial Groulld ........................................................................................................................................................
The Senses ...........................................................................................................................................................................
Double Assignment of Deities ..........................................................................................................................................
The bali offering ..................................................................................................................................................................
The Inclusion of the Intermediate Directions and the Incipient lokapila Concept ...........................................................
Directional Assignment in the Later Vedic Period ................................ ........ .....................................................................
The Directional Guardians in the Epics (REn~iiyn!znand Mahibhirata) ................................... .. ................................ 12
Evidence from the Puriiyas ..............................................................................................................................................
13
Other Early Texts ..............................................................................................................................................................
16

Predecessors and Earliest Evidence of the Depiction of Directional Deities ............................................................. 18

Early Sculptures of the Guardians of the Directions from Buddhist Colltexts ........................... . . .............................. 18
Foul~dationStones and Deposits .........................................................................................................................................
19
The Early Depiction of Assetnblies of Deities ............................................................................................................... 20
Panels of Four Figures with Identical Features ...................................................................................................................
21
Solne Early Sculptures of Indra and Agni .......................... . . ........................................................................................22
2.2 Centres of Early Develop~nent(C .400-650 A.D.) ............................................................................................................ 23
2.2.1 The Deccan .........................................................................................................................................................................23
Ba&llzi ................................................................................................................................................................................
23
~ l ~ ~ ............................................................................................................................................................................
' l ~ ~ ~ ~ t ~ 26
Ellori, Caves 14, 21 29 .............................................................................................................................................. 26
Alanll~llr,K~rriiirnBrnhri~iitenll1le .................................................................................................................................... 28
2.2.2 Central India ....................................................................................................................
28
pau, iiyi .....................................................................................................................................
29
Bhillzarii ..............................................................................................................................................................................
29
2.2.3 Eastern India ........................................................................................................................................................................
31
NfilOnr/i ...............................................................................................................................................................................
31
M~~!~cleSvnriHill (Kirizgadh) ................................................................................................................................................ 31
Bhubarzeswar ......................................................................................................................................................................33
2.2.4 Other Early Evidence .......................................................................................................................................................... 34
Terracotta Sorna froiiz Uttar Prallesh ................................................................................................................................34
2.3 Further Dcvelopnlcnt in tile Different Regions ............................................................................................................. 34
2.3.1 Early Phase ( c. 651-750 AD.) ............................................................................................................................................. 35
2.3.1.1 The Deccan ........................................................................................................................................................................
35
BErIirili. Jariz.!x/lirigeSvara Teiilple; Nigarai, Nigaizitlza terill~le;Mnhiki!a, Mallikirjurln Ten~l)le................................ 35
Aihoie, Huccappayyn Tenlple .............................................................................................................................................. 37
 Pa f~aclakal.Pdpatzdtlza Tenzple ...........................................................................................................................................
Alninp~lr,Navabralzmii Ternples ..........................................................................................................................................
K~~<laveli,Snriganzeivara Tenlple ........................................................................................................................................
Satyavol~i,REmalinges'vara Temple ....................................................................................................................................
Central India ........................................................................................................................................................................
Paliiri, SiddIzeSvara Temnple ................................................................................................................................................
Anzrol, RiinzeSvara Malzndeva and Ddnebiibd Tenzples ....................................................................................................
Gwdlior, Tell-kE-nzandir .....................................................................................................................................................
Western India .......................................................................................................................................................................
Cittaudga<llz,Kiilikdnziitii and K~~mbhaSyrTma Tenzples ....................................................................................................
Bithli, Mahiideva Tenzple ...................... . . ..........................................................................................................................
Pipad, Pipldd~lzntiiTenzple ..................................................................................................................................................
Osinfi, "Slirya" Tenzple rzo. I and "Harihnra " Temi~pleno. I ...........................................................................................
Liinzbd, Old Tenzple .............................................................................................................................................................
Eastern India and Bangladesh .................... . ...................................................................................................................
Dev BaratzErak (Deo Barzi!zdrk) ........................................................................................................................................
Later Phase (c.75 1-1000 A.D.) ...........................................................................................................................................
The Deccan ........................................................................................................................................................................
Central and Northern India ..................................................................................................................................................
Western India ................................... ..............................................................................................................................
.
Eastern India ............................. . ......................................................................................................................................
Southern India .......................... . ..................................................................................................................................
Concepts of dikpiila Representation ..............................................................................................................................
Group Representation ..........................................................................................................................................................
Vertical Panels .....................................................................................................................................................................

Iconography of the dikpsl;ls .............................................................................................................................................

Indra .....................................................................................................................................................................................
Agni .....................................................................................................................................................................................
.
Yama .................................................................................................................................................................................

Conclusion .....................
..... .................................................................................................................................................................
111
German Summary (Zusarnmenfassung in deutscher Sprache) ............................................................................................................... 113
Bibliography .................................................................................................................................................................... 117
Abbreviations ....................................................................................................................................................................................... 127
General Index .........................................................................................................................................................................................
128
Figures ............................................................... ..............................................................................................................................
133
 Acknowledgements

The subject of the present study, which is a slightly modified ver-

sion of my Pl1.D. thesis titled The Fori~zatiorzarzd Early Develop-
inerzt of tlze Bralzinaizical Group of Directiorzal Guardiarz Deities irz
bzdiaiz Art (urztil c. 1000 A.D.), has been chosen under the guidance
of Prof. Dr. Adalbert J. Gail of the Institut fur Indische Pliilologie
und Ki~nstgeschichte,Freie Universitat Berlin. I am grateful to him
for his support extended throughout my work on the subject. I am
also indebted to Prof. Dr. Marianne Yaldiz, Director of the Museum
fur Indische Kunst Berlin, and the Council of the Ernst Waldsch~nidt
Foundation, for considering the publication of my dissertation under
the auspices of this renowned institution.

I would like to thank those who most kindly assisted lne during my
two visits to India, in 1992 and 1995: Prof. M.A. Dhaky, Dr. U.S.
Moorti and Shri J.S. Yadav of the Center for Art and Archaeology,
American Instilute of Indian Studies (A.I.I.S.), then at Ramnagar,
Dr. T.K. Biswas, Joint Director, Bharat Kala Bhavan, Varanasi,
Dr. R. Nagaswamy, former Director, Department of Archaeology of
Tamilnadu State, and the late Dr. Franqoise L'Hernault, fonner
Head of the Centre d'Histoire et d'Arche'ologie, Ecole Fran~aise
d'Extr61ne-Orient. I also thankfully acknowledge the kind technical
and other assistance rendered by the archaeological officers at
Alampur and the temple authorities at AIampur and Dharmapuri.

The following institutions most kindly provided me with photo-

graphs and the permission for publication, the Center for Art and
Archaeology, A.I.I.S., Gurgaon, the Arcllaeological Survey of
India, New Delhi, the French Institute at Pondicherry, the Rijks-
museum voor Volkenkunde, Leiden, the Linden-Museum, Stuttgart,
the Museum fur Indische Kunst Berlin, and Sotheby's. While it is
both my duty and pleasure to acknowledge the assistance also ren-
dered in this respect, by friends as well as teachers and colleagues:
Prof. Dr. Michael Meister, University of Pennsylvania, Philadel-
phia, Prof. Dr. Karel R. van Kooij, Rijksuniversiteit Leiden, Prof.
Dr. Klaus Bruhn, Freie Universitat Berlin, Dr. Gerd Kreisel,
Linden-Museum Stuttgart, Privatdozent Dr. Joachim K. Bautze,
Sudasien-Institut, Universitat Heidelberg, Dr. Gouriswar Bhat-
tacharya, Ms. Martina Stoye, M.A., Ms. Petra Haubold, Dr. Falk
Reitz, Ms. Britta Schneider, M.A., Ms. Gudrun Melzer, Mr. Jurgen
NeuB, M.A., Mr. Otto Schorr and Mr. Gerd Mevissen, M.A. Pho-
tographs without any specification of copyright are by the autl~or.
Further, I am beholden to Ms. Vandana Nadkarni, M.A., Dr. Louis
Nebelsick, Mr. Mansour Mansour and, again, to Mr. Gerd Mevissen
for kindly assisting me in various ways, while I am alone responsi-
ble for all the mistakes which might have crept in.
List of Tables

I Directional Guardians mentioned in the Atlzarvaverkl ....................................................................................................................6

I1 Directional Guardians mentioned in the Yajurverlcl ........................................................................................................................ 6
I11 Directional Guardians mentioned in various Br5hmanas ................................................................................................................ 9
IV Directional Guardians mentioned in various Aranyakas ................................................................................................................ 9
V Directional Guardians mentioned in various Upanisads ................................................................................................................. 9
V1 Directional Guardians mentioned in various Siitras ..................................................................................................................... 10
V11 Directional Guardians mentioned in the Epics ............................................................................................................................. 13
VIII Lokapiilas and Associates mentioned in Mahiiblziiratn 2 ............................................................................................................. 13
IX Lokaplilas and Associates mentioned in Malziiblziirrrta 5 ........................................................................................................... 13
X Iconographic specifications of dikl~iilasfrom various Puranas .................................................................................................... 15
XI Iconographic specifications of dikpiilas from various early texts ................................................................................................
17
XI1 Sculptural typology of dikl~iilnsat AIampur ................................................................................................................................. 40
XI11 .-- .................................................................................................................................... 68
Sculptural typology of clikplilas at Os~an
XIV Iconographic Chart of Indra ..........................................................................................................................................................
96
XV Iconographic Chart of Agni ..........................................................................................................................................................
97
XVI Iconographic Chart of Yalna .................... . ................................................................................................................................
99
XVII Iconographic Chart of Nairrta .....................................................................................................................................................
100
XVIII Iconographic Chart of Varuna ......................................................................................................................................................
101
XIX Iconographic Chart of V5yu .........................................................................................................................................................
103
XX Iconographic Chart of Kubera ......................................................................................................................................................
104
XXI Iconographic Chart of So~na/Candra............................................................................................................................................ 106
XXII Iconographic Chart of T&na ..........................................................................................................................................................
107
List of Illustrations

Sigrzs Used in Illustrarions:

Arrows .d~rectionin which the figure moves on its vehicle
Bold bar .bottom of a panel
P . false portico
(...) .later inserted image
[ ...] - hypothetical placement

1 Biidiimi. Cave 3. location of ceiling panels ........................................................................................................................................... 24

2 MundeSvari Hill. ground plan and sketch of the reconstructed arrangement of lokapiilas ............................................................... 33
3 Bhubaneswar, ParaSuriimeSvara Temple: sketch of arrangement of deities on the south-western portion
of the jaga111011a11a.................................................................................................................................................................................
33
4 BidBmi, Jambulinga Temple, ceiling panel of ardliamandapa in front of northern sanctum ............................................................ 35
5 Nagara!, Niganiitha Temple, central ceiling panel of nza!iqlapa ....................................................................................................... 36
6 MahZkuta, Mallikirjuna Temple, central ceiling panel of m a ! l ~ a p a .................................................................................................. 36
7 Aihoje, Huccappayya Temple, lokapiilas on beams of central bay .................................. . . ........................................................ 37
8 Pattadakal, Papanatha Temple, eastern nza!zdapa, central ceiling ......................................................................................................... 38
9 Paftadakal, Pipan5tha Temple, western portion, outer wall .................................................................................................................. 38
10 AIampur, B5la Brahmi Temple, exterior wall niches ........................................................................................................................... 41
11 Alampur, Svarga Brahma Temple, exterior wall niches ....................................................................................................................... 42
12 Alampur, ViSva Brahmi Temple, exterior wall niches ................................ . . .................................................................................... 43
13 Satyavolu, RnmaliligeSvara Temple, exterior wall niches .................................. . . ............................................................................ 44
14 Amrol, RameSvara Mahadeva Temple, plan ........................................................................................................................................ . 46
15 Amrol, Dgnebaba Temple, plan ...................................................................................................................................................... ...... 47
16 Gwglior, Teli-ki-mandir, plan (after DEVA1991: fig. 7) ...................................................................................................................... 47
17 Cittaudgadh, Kalikam5ta Temple, garbllngr/za plan ............................................................................................................................. 49
18 Cittaudgadh, KumbhaSy$ma Temple, gar.!d1agr]za plan ....................................................................................................................... 50
19 Cittaudgadh, KumbhaSyama Temple, exterior wall, plan .................................................................................................................. ... 51
20 Pipid, Piplidmat2 Temple, plan ..................... ..................................................................................................................................
..... 53
21 osiifi, snrya~~~~l~ 1, plan ................................................................................................................................................................. 54
22 Osiaii, Harihara Temple l , plan ..................... ........................................................................................................................................
55
23 ceiling panels of the D~~~~~region, c . gth-loth centuries (sketches): a) Alur, ArkeSvara Temple (Karnataka) ...............................58
b) Arajaguppe, KalleSvara Temple (Karnataka) ............................................................................................................................... ........ 58
c) ~ h ~Saligam, 1a i ~ i Temple (A.P.) ......................................................................................................
~ ~~ f i ~~ gSangameSvara ..I......... 58
d) Dharmapuri, Mallikirjuna Temple (T.N.) ........................................................................................................................................... 58
e) Kambadahalli, Siintinathabasti (Karnataka) ........................................................................................................................................ 58
f) Kambadahalli, Adinatha basti (Karnataka) .................................................................................................................................. ......... 58
g) ~ ~ ~R ~~ Temple ~ ~(Karnataka) ~ S ~ ~ n ~ ~ ~ ~ ~ l ~ ~ ..............
~ ~ ............................................................................................................... , 58
h) pfipangsanam, group g (A.P.) ............................................................................................................................................................... 58
i) Pratakota, NggeSvara Temple (A.P.) ................................................................................................................................ ................... 58
24 pattadakal, K ~ ~ i v i S v a n ~Temple,
tha ceiling panel ........................................................................................................... ..................... 59
25 ~ 1 1 16~ ( ~~ 3 ~ i~ l a~ ~ gopura,
~~ ~ dikpiila l ~ niches ........ 62
) , .....................................................................................................................
26 p i n g , Siva Temple, plan .......................................................................................................................................................................
63
27 Indor, siva Temple, plan (after M E ~ ~ T 1982: E R fig. A) .................................................................................................................... ..... 64
28 Gwalior, CaturbhujaTemple, plan sketch of sanctum ................................................................................................. ......................... 66
29 Pihgrpur, Somapura Vihara, plan .......................................................................................................................................... ................ 72
30 GallLvalli, KgmeSvara Temple, plan .................................................................................................................................... ................. 75
31 Representations of Manu Lokapalas (8th through loth centuries) ............................................................................................... ......... 89
List of Figures 16 Skanda (photo: Joachim K. Bautze)
Bldlmi, Cave 3, southern ceiling medallion,
Abbreviatiorls Used irl Ca(~tior~s: main hall, 578 A.D.
A.I.I.S. American Institute of Indian Studies, Center for 17 Varuna (photo: Joachim K. Bautze)
Art & Archaeology, Gurgaon Biidiimi, Cave 3, western ceiling medallion,
IFP Institut Francais de PondichtrylEcole Francaise d'ExtrCme-Orient main hall, 578 A.D.
18 Kubera (photo: Joachim K. Bautze)
1 Incised foundation bricks (after D. MITRA1973-74, Fig. 2) Bgdiimi, Cave 3, northern ceiling medallion,
Sagrahawa (Nepal), Structure 8, "mediaeval" main hall, 578 A.D.
2 Kubera (photo: A.I.I.S.) 19 Ardhanlriivara Siva panel (photo: Otto Schorr)
Gosna Khera, c. 1st century A.D., Elephanta, Cave 1, c. mid-6th century
buff sandstone, H: 27 cm, W: 12.5 cm 20 NateSa panel (photo: Gerd Mevissen)
Governmellt Museum, Mathura, no. 39.2856 Elephanta, Cave 1, western annex, c. mid-6th century
3 Viiyu on AnantaSayana ViSnu panel 21 NafeSa panel (photo: A.I.I.S.)
(photo: Karel R. van Kooij) Ellorii, Cave 29, west wall, near corridor of northern annex,
Deogadh, Gupta Temple, southern exterior wall, C. 600-650 A.D.
C. 500-550 A.D. 22 Nafeia panel (detail) (photo: Gerd Mevissen)
4 Fragment of a panel with four IokapZlas (?) Ellorl, Cave 21, east wall of southern annex, late
(photo: Linden-Museum, Stuttgart) 6th century
North India, c. 2nd century A.D. 23 Yama
terracotta, H: 31 cm AIampur, Kumiira Brahmii Temple, near top of Siklznm,
Linden-Museum, Stuttgart, no. SA37028L south, c. 630 A.D.
5 Inscribed, fragmented panel with four lokapalas (?) 24 Kubera (?)
(photo: Rijksmuseum voor Volkenkunde, Leiden) Alampur, Kumka Brahmii Temple, near top of Siklznrn,
probably Mathurii region, c. 1st-2nd centuries A.D. north, c. 630 A.D.
mottled red sandstone, H: 28 cm, W: 27 cm, inscribed: 25 Yama (after Progress Report of the Archaeological Survey
1. bnlnvnsul)utr[e!z~,2. [pi]tLipriynr?ztn Dhn... of India, Western Circle, Bombay 1922, pl. XXVa)
Rijks~nuseu~n voor Volkenkunde, Leiden, no. 2498 BhUmarl, ~ i v Temple,
a loose gnvEk.y (medium size),
6 Indra (photo: A.I.I.S.) C. 520-530 A.D.

Sitaliighiiti, c. 4th century H: 46 cm, W: 37 cm

Government Museum, Mathura, no. 46.3226 Allahabad Museum, no. 153
7 Agni (photo: A.I.I.S.) 26 Kubera (after R.D. BANERJI 1924: pl. XIIIa)
Mathurii District, c. 3rd century BhUmarii, Siva Temple, loose gnviikp (medium size),
H: 20 cm, W: 9 cm C. 520-530 A.D.

Government Museum, Mathura, no. 42-43.3048 27 Kubera (extant fragment, same as in fig. 26)
8 Skanda (left) and Agni (photo: A.I.I.S.) Bhiirat Kall Bhavan, Varanasi, provisional no. T 261
Kapi KedaradevdMathurii, c. 3rd century (in 1992)
mottled sandstone, H: 51 cm, W: 58 cm 28 DandapPyi (after BANERJI 1924: pl. XIV c)
Government Museum, Mathura, no. 2883 BhU~narii,Siva Temple, loose gnvLik;a (large size),
9 'Churning of the Ocean' scene (lower tier) C. 520-530 A.D.
(after M.B. GARDE1924-25: pl. XLIIId) H: 60 cm, W: 79 cm
Pawiiyii (M.P.), torn!ln fragment, c. 400410 A.D. Indian Museum, Calcutta
H: 67 cm, W: 195 c111 29 Agni
Central Archaeological Museum, Gwalior NiilandB, Stone Temple no. 2, base frieze, east,
10 ViSnu surrounded by dikpiilas (photo: A.I.I.S.) Panel 7,7th century
Biidiimi, Karnataka, Cave 3, central ceiling panel of 30 Kubera (photo: A.I.I.S.)
vestibule, 578 A.D. Nllandl, Stone Temple no. 2, base frieze, east,
11 Indra surrounded by dancers and n~usicians Panel 14, 7111 century
(photo: A.I.I.S.) 31 Viiyu
Biidiimi, Cave 3, second ceiling panel from the east, Nllandii, Stone Temple no. 2, base frieze, east,
vestibule, 578 A.D. Panel 19,7th century
12 Varuna surrounded by couples (photo: A.I.I.S.) 32 Indra and his elephant?
Biidiimi, Cave 3, second ceiling panel from the west, Nglandii, Stone Temple no. 2, base frieze, east,
vestibule, 578 A.D. Panels 22-23,7th century
13 Brahmii surrounded by four lokap~3lasand g i s 33 Kubera
(photo: A.I.I.S.) Nllandfi, Stone Temple no. 2, base frieze, south,
Biidlmi, Cave 3, third ceiling panel from the west, Panel 200,7th century
vestibule, 578 A.D. 34 Agni (photo: Gerd Mevissen)
14 B r a h m l (photo: Joachi~nK. Bautze) MundeSvari Hill, ruined temple, 6th-7th centuries
Biidlmi, Cave 3, central ceiling panel, main hall, 578 A.D. H: 43 cm, W: 33 cm
15 Indra (photo: Joachim K. Bautze) State Archaeological Museum, Patna, no. 601 1
Biidlmi, Cave 3, eastern ceiling medallion, main hall, 578 A.D.
 Daqdapiiqi (photo: Gerd Mevissen)
Mundeivari Hill, ruined temple, 6th-7th centuries
H: 42.5 cm
State Archaeological Museum, Patna, no. 6004
Soma (photo: A.I.I.S.)
Mundeivari Hill, ruined Temple, cornerstone,
6th-7th centuries
H: 30 cm
State Archaeological Museum, Patna
Siirya (same as in fig. 36) (photo: A.I.I.S.)
Agni (photo: A.I.I.S.)
MundeSvari Hill, ruined Temple, cornerstone,
6th-7th centuries
H: 30 cm; State Archaeological Museum, Patna
Indra (same as in fig. 38) (photo: A.I.I.S.)
Kubera (photo: A.I.I.S.)
MundeSvari Hill, ruined Temple, cornerstone,
6th-7th centuries
H: 30 cm; State Archaeological Museum, Patna
Yama (same as in fig. 40) (photo: A.I.I.S.)
Varuqa, Yama and Daqdapiiqi (from the left)
Bhubaneswar, ParaSurHmeivara Temple, jaganzokarza,
south wall, niches between window and door,
c. early 7th century
Agni or Soma? (left) and VHyu
Bhubaneswar, ParaSuriimeSvara Temple, jagamoha~zn,
south wall, second and third niches from the western corner,
c. early 7th century
Doorkeeper (left) and Kubera?
Bhubaneswar, ParaSurImeSvara Temple, jnganzolzarza,
west wall, southern corner, c. early 7th century
Soma (photo: Sotheby's, after Sales Catalogue,
New York, 28.10.1991, Lot 302)
Uttar Pradesh, c. 6th-7th centuries
terracotta
Panel with four lokap5las (photo: A.I.I.S.)
54 PSiina (photo: Joachim K. Bautze)
Aihoje, Huccappayya Temple, central bay, eastern beam,
left side (north-east), c. 722-730 A.D.
55 Indra (photo: Joachim K. Bautze)
Pa!!adakal, Piipantitha Temple, eastern rna!z<lalm,
ceiling panel, east, c. 735-750 A.D.
56 Agni (photo: Joachim K. Bautze)
Patiadakal, Piipaniitha Temple, eastern rna!zdapa,
ceiling panel, south-east, c. 735-750 A.D.
57 Yama (photo: Joachim K. Bautze)
Pattadakal, PBpanHtha Temple, eastern mn!7dal~a,
ceiling panel, south, c. 735-750 A.D.
58 Nairrta (photo: Joachiln K. Bautze)
Pattadakal, PipanrItha Temple, eastern ~lzn!i&r~~a,
ceiling panel, south-west, c. 735-750 A.D.
59 Viiyu (photo: Joachim K. Bautze)
Pattadakal, PBpanHtha Temple, eastern rna!~<lnpa,
ceiling panel, north-west, c. 735-750 A.D.
60 Kubera (photo: Joachim K. Bautze)
Pattadakal, PHpantitha Temple, eastcrn mn!iclnpn,
ceiling pancl, north, c. 735-750 A.D.
61 ISHna (photo: Joachim K. Bautze)
Pattadakal, PiipanHtha Temple, eastern ma!z~lnpn,
ceiling panel, north-east, c. 735-750 A.D.
62 Varuqa (photo: IFP)
Paitadakal, PiipanHtlia Temple, exterior, western 1iza!z</a1?~,
north(-west), late 7th or early 8th century
63 Kubera (photo: IFP)
Pattadakal, Piipaniitha Temple, exterior, western l ? ~ a ! z < l n ~ ~ ,
north(-east), late 7th or early 8th century
64 Indra
AIampur, Btila Brahmii Temple, east wall,
c. 650-670 A.D.
65 Agni
AIampur, BBla BrahmH Temple, south wall,
C. 650-670 A.D.

BiidHmi, Jambulingeivara Temple, antarda of northern 66 Yama

shrine, ceiling, 699 A.D. AIampur, Biila BrahmH Temple, south wall,
Panel with four lokap5las and Brahmii in the centre c. 650-670 A.D.
(photo: A.I.I.S.) 67 Varuqa
NrIgara!, NiiganBtha Temple, nza!z&pa, ceiling, AIampur, BHla Brahmii Temple, west wall,
late 7th century c. 650-670 A.D.
Panel with four lokapiilas and Brahmii in the centre 68 Viiyu
(photo: A.I.I.S.) AIampur, Biila Brahmii Temple, west wall,
Mahiiku!a, Mallikarjuna Temple, riza!z~la/~n, ceiling, c. 650-670 A.D.
late 7th century 69 I<ubera
Indra, Aihole (photo: Joachim K. Bautze) Alampur, Bila Brahmii Temple, north wall,
Huccappayya Temple, central bay, eastern beam, centre, c. 650-670 A.D.
c. 722-730 A.D. 70 Indra
Agni (?) (photo: Joachim K. Bautze) AIampur, Svarga BrahmrI Temple, east wall,
Aihole, Huccappayya Temple, central bay, eastern beam, c. 680-700 A.D.
right side (south-east), c. 722-730 A.D. 71 Agni
Yama (photo: Joachim K. Bautze) Alamp~~r, Svarga BrahmH Temple, south wall,
Aihoje, Huccappayya Temple, central bay, southern beam, c. 680-700 A.D.
centre, c. 722-730 A.D. 72 Yama
VHyu (photo: Joachim K. Bautze) Alampur, Svarga Brahmii Temple, south wall,
Aihole, Huccappayya Temple, central bay, western beam, c. 680-700 A.D.
right side (north-west), c. 722-730 A.D. 73 Nairrta
Kubera (photo: Joachim K. Bautze) Alampur, Svarga Brahmii Temple, south wall,
Aihole, Huccappayya Temple, central bay, northern beam, c. 680-700 A.D.
centre, c. 722-730 A.D. 74 Varuqa
AIampur, Svarga Brahmii Temple, west wall,
c. 680-700 A.D.

XIV
viyu 95 Siirya (photo: A.I.I.S.)
AIampur, Svarga Brahmii Temple, west wall, Amrol, Diinebiibii Temple, (north-)eastern kar!za niche,
c. 680-700 A.D. c. mid-8th century
Kubera 96 Indra (photo: A.I.I.S.)
AIampur, Svarga Brahml Temple, north wall, Amrol, Diinebiibii Temple, (south-)eastern knrtza niche,
c. 680-700 A.D. c. mid-8th century
CandraISoma 97 Agni (photo: A.I.I.S.)
AIampur, Svarga Brahml Temple, north wall, Amrol, Diinebiibl Temple, south(-eastern) kar!za niche,
C. 680-700 A.D. c. mid-8th century
ISiina 98 Agni
AIampur, Svarga Bral~mlTemple, north wall, GwLlior, Teli-kii-mandir, base frieze, Panel 19,
C. 680-700 A.D. c. mid-8th century
Siirya 99 Varupa
AIampur, Svarga Brahmii Temple, east wall, Gwiilior, Tell-kii-mandir, base frieze, Panel 69,
C. 680-700 A.D. c. mid-8th century
Yama 100 Viiy11
AIampur, ViSva Brahma Temple, south wall, Gwllior, Teli-kii-mandir, base frieze, Panel 70,
late 7th century c. mid-8th century
Kubera 101 Kubera (standing)
AIampur, ViSva Brahmii Temple, north wall, Gwiilior, Teli-kii-mandir, base frieze, Panel 75,
late 7th century c. mid-8th century
CandraISoma 102 Kubcra (seated)
AIampur, ViSva Brahmii Temple, north wall, Gwiilior, Teli-kii-mandir, base frieze, Panel 80,
late 7th century c. mid-8th century
Nairgta 103 CandraISoma? (photo: Klaus Bruhn)
AIampur, loose image in the Arka Brahlnii Temple, Gwiilior, Teli-kii-mandir, base frieze, Panel 113,
probably from the Garuda Brahmii Temple, c. 700 A.D. c, mid-8th century
CandraISoma? (photo: IFP) 104 Agni (photo: A.I.I.S.)
AIampur, Garuda BrahmL Temple, north wall, second Cittaudgadh, Kllikiimlti Temple, garbhabhitti, south(-east),
niche from the east, c. 700 A.D. c. late 7th century
Yama 105 Yama (photo: Michael W. Meister)
Kudaveli, Sanga~neSvaraTemple (reconstructed at AIampur), Cittaudgadh, KiilikiimiitiiTemple, garblznbhitti, south(-west),
south wall, c. late 7th century c. late 7th century
Kubera (photo: A.I.I.S.) 106 Nairgta (photo: Michael W. Meister)
Kudaveli (?), c. late 7th century Cittaudgadh, Kiiliklmiitii Temple, garbhabhitti, (south-)west,
H: 68 cm, W: 45 cm c. late 7th century
Alampur Museum, no. 5 1 107 Varupa (photo: A.I.I.S.)
Kubera Cittaudgadh, Kiilikiimltii Temple, garblzablzitti, (north-)west,
Kudaveli, SangameSvara Temple (reconstructed at Alampur), c . late 7th century
priikdra wall, northern side, western corner, c. late 7th century 108 VByu (photo: A.I.I.S.)
Agni (photo: A.I.I.S.) Cittaudgadh, Kiilikiimiitii Temple, garbhabhitti, north(-west),
Satyavolu, RiimalidgeSvara Temple, r~~a!z~lapn, south(-east), c. late 7th century
c. early 8th century 109 Soma (photo: A.I.I.S.)
G a r b l ~ a g ~overdoor
l~a with dikpilas Cittaudgadh, Kiilikiimfitii Temple, garbhabhitti, north(-east),
(photo: A.I.I.S.) c. late 7th century
Paliiri, Siddheivara Temple, c. late 7th or early 8th century 110 ISiua (photo: A.I.I.S.)
Agni (photo: A.J. Gail) Cittaudgadh, Kiilikiimiitii Temple, garblznbhitti,
Amrol, RiimeSvara Mahldeva Temple, southern kapili (north-)east, c. late 7th century
niche, c. early 8th century 111 ISBna type (photo: Jiirgen NeuB)
Yama (photo: Joachim I<. Bautze) Cittaudgadh, KurnbhaSySma Temple, garbhabhitti,
Amrol, Rii~neSvaraMahiideva Temple, south(-westenl) south(-east), c. early 8th century
kar!la niche, c. early 8th century 112 JSina type (photo: Jiirgen NeuB)
Kubera (photo: A.I.I.S.) Cittaudgadh, KumbhaSyiima Temple, garbhabhitti,
Amrol, DlnebiibI Temple, (north-)western lnr!zcza niche, south(-west), c. early 8th century
c. mid-8th century 113 Nairrta? (photo: A.I.I.S.)
VByu (photo: A.I.I.S.) Cittaudgadh, Ku~nbhaSyiimaTemple, garbhablzitti,
Amrol, Diinebiibl Temple, north(-western) kartza niche, (south-)west, c. early 8th century
c. mid-8th century 114 Varuya? (photo: Jiirgen NeuB)
IS51iaor Soma (?) (photo: A.I.I.S.) Cittaudgadh, KurnbhaSyfima Temple, garbhabhitti,
Amrol, Dlnebiibii Temple, north(-eastern) kart10 niche, (north-)west, c. early 8th century
c. mid-8th century
il 115 Indra (photo: A.I.I.S.)
l
Cittaudgadh, KumbhaSyHma Temple, gl<llzarna!z,$apn,
exterior, (south-)east, c. early 8th century
116 Agni (photo: A.I.I.S.)
Cittaudgadh, KumbhaSyiima Temple, gl,$l~aman<lapa,
exterior, south(-west), c. early 8th century
117 Skanda (photo: Jurgen NeuR)
Cittaudgadh, KumbhaSyiima Temple, inllal~rC?siirla,
exterior, south(-west), c . early 8th century
:
l 118 Nairgta (photo: Jurgen Neul3)
Cittaudgadh, KumbhaSyZma Temple, mllaprijsijda,
exterior, (south-)west, c. early 8th century
l
119 Varuqa (photo: Jurgen NeuR)
Cittaudgadh, KumbhaSyiima Temple, rni,ilapriisijda,
exterior, (north-)west, c. early 8th century
120 Yama (photo: Jiirgen Neul3)
Cittaudgadh, KumbhaSyiima Temple, nzllapr~isiirla,
exterior, north(-west), c . early 8th century
121 Kubera (photo: Jurgen NeuR)
Cittaudgadh, KumbhaSyiima Temple, gi,i,$hariza!~&pa,
exterior, north(-east), c. early 8th century
122 ISiina (photo: Jiirgen Neufi)
Cittaudgadh, KumbhaSyiima Temple, gC<llzarnn!z<lapn,
exterior, (north-)east, c. early 8th century
123 Siirya (photo: A.I.I.S.)
Pipiid, Pipliidmiitii Temple, (south-)east,
c. second quarter of the 8th century
124 Kubera (photo: A.I.I.S.)
Pipiid, PipliidmHtii Temple, south(-west),
c. second quarter of the 8th century
125 Unidentified deity (photo: A.I.I.S.)
PipHd, Pipliidmiitii Temple, (south-)west,
c. second quarter of the 8th century
126 Varuna? (photo: A.I.I.S.)
Pipiid, Pipliidmiitii Temple, (north-)west,
c, second quarter of the 8th century
127 Vlyu (photo: A.I.I.S.)
Pipiid, PipliidmHtii Temple, north(-west),
c. second quarter of the 8th century
128 &a? (photo: A.1.I.S.)
Pipiid, Pipliidmiitii Temple, north(-east),
c. second quarter of the 8th century
129 AziadikpZla panel with Somaskanda in the centre
(photo: IFP)
Pattadakal, KiiSiviSvaniitha Temple, ceiling,
c. mid-8th century
130 Brahml (photo: Joachim K. Bautze)
Aihole, Kunti Group, north-east Temple, ceiling, centre,
c. third quarter of the 9th century
131 Indra (photo: Joachim K. Bautze)
Aihoje, Kunti Group, North-east Temple, ceiling, east,
C. third quarter of the 9th century
132 Agni (photo: Joachim K. Bautze)
Aihoje, Kunti Group, North-east Temple, ceiling,
south-east, c. third quarter of the 9th century
133 Yama (photo: Joachim K. Bautze)
Aihoje, Kunti Group, North-east Temple, ceiling, south,
c, third quarter of the 9th century
134 Kubera (photo: Joachim K. Bautze)
AihoJe, Kunti Group, North-east Temple, ceiling, north,
c. third quarter of the 9th century
XVI
135 i S ~ n a(photo: Joachim K. Bautze)
Aiho!e, Kunti Group, North-east Temple, ceiling,
north-east, c. third quarter of the 9th century
136 A~fadikpZlapanel with NafeSa in the centre
(after I.K. SARMA 1992: pl. 45)
Arajaguppe, KalleSvara Temple, ceiling,
c. early 10th century
137 A,$adilp;Zla panel with NateSa in the centre
(photo: A.I.I.S.)
Pathasivarama, MailareSvara Temple, ceiling,
c. mid-10th century
138 AsfadikpZla panel with UmBmaheSvara in the centre
(after B. DAGENS1984: ph. 805; photo: IFP)
BhavanHsi Sangam, Rnpiilii SaiigameSvara Temple,
ceiling, c. second quarter of the 9th century
139 Exterior view with TSlna on the right side
(photo: A.I.I.S.)
Aihoje, Temple no. 52, east wall,
c. second quarter of the 10th century
140 Agni (photo: Gerd Mevissen)
HemBvati, fragment of ceiling panel,
c. second quarter of the 10th century
Government Museum, Madras
141 Indra
AIampur, fragment of ceiling panel ('no. 1'1,
c. 9th century
H: 50 cm
AIalnp~~r Museum
142 Agni
Alampur, fragment of ceiling panel ('no. l'),
c. 9th century
H: 50 cm
AIampur Museum
143 Agni
AIampur, fragment of ceiling panel, ('no. 2'),
c. 9th century
H: 57 cm
AIampur Museum
144 ISlna
Alampur, fragment of ceiling panel, ('no. 2'1,
c. 9th century
H: 57 cm
Alampur Museum
145 Siirya
DBlig, Siva Temple, (south-)east,
c. third quarter of the 8th century
146 Agni
Diing, Siva Temple, south(-east),
c. third quarter of the 8th century
147 Yan~a
?%rig, Siva Temple, south(-west),
c. third quarter of the 8th century
148 Nairgta
Diing, Siva Temple, (south-)west,
c. third quarter of the 8th century
149 vayu
Diing, Siva Temple, (north-)west),
c. third quarter of the 8th century
150 Varuna
DHrig, Siva Temple, north(-east),
c. third quarter of the 8th century
 ! 151 Goddess (lunar symbolisni?) 171 Agni
Diilig, ~ i v Temple,
a (north-)east, Gwiilior, Caturbhuja Temple, south(-east), c. 875 A.D.
c. third quarter of the 8th century 172 Yatna
I 152 Indra (photo: A.I.I.S.) Gwiilior, Caturbhuja Temple, south(-west), c. 875 A.D.
I Indor, Gargaja MahCdeva Temple, south kapili, 173 Nairrta
c. third quarter of the 8th century Gwiilior, Caturbhuja Temple, (south-)west, c. 875 A.D.
l 153 Agni (photo: A.I.I.S.) 174 Varuqa
l Indor, Gargaja Mahiideva Temple, blzarlra, south-south-east, Gwiilior, Caturbhuja Temple, (north-)west, c. 875 A.D.
l
I c. third quarter of the 8th century 175 VPyu
iI 154 Yama (photo: A.I.I.S.) Gwiilior, Caturbhuja Temple, north(-west), c. 875 A.D.
Indor, Gargaja Mahiideva Temple, bhadra, south-south-west, 176 Kubera
c. third quarter of the 8th century Gwiilior, Caturbhuja Temple north(-east), c. 875 A.D.
155 Nairrta (photo: A.I.I.S.) 177 Varuqa (after A. DE LIPPE1962: no. 33)
Indor, Gargaja MahZdeva Temple, bhadra, west-south-west, probably Mathurii region, c. early 9th century
1
, c. third quarter of the 8th century stone slab, H: 3 1.75 cm
iI 156 Varuqa (photo: A.I.I.S.)
Indor, Gargaja Mahiideva Temple, bhadra, west-north-west,
Metropolitan Museum of Art, New York, no. 19.90
178 ISGna (plioto: Gerd Kreisel)
c. third quarter of the 8th century Mathurii, from the river Yainuni, c. early 9th century
1 157 VGyu (photo: A.I.I.S.) stone slab, H: 30 cm
I Indor, Gargaja Mahiideva Temple, bl~aclra,north-north-west, Government Museum, Mathura, no. 1570
1
! c. third quarter of the 8th century 179 Garbl~ag~ha overdoor with dikpzla frieze
158 Kubera (photo: A.I.I.S.) (photo: A.I.I.S.)
Indor, Gargaja Mahiideva Temple, bhadra, north-north-east, Barwiisiigar, Jariii-kii-math, c. 900 A.D.
c. third quarter of the 8th century 180 Loose dikpBla frieze
159 ISGna (photo: A.1.I.S.) Badoh, c. 9th-10th centuries
Indor, Gargaja Mahiideva Temple, north kapili, H: c. 50 cm, W: 215 cm
c. third quarter of the 8th century Central Archaeological Museum, Gwalior, no. 93
I 160 Yama (photo: A.I.I.S.) 181 VSyu (photo: A.J. Gail)
l Kora (M.P.), c, mid-8th century Sesai, Siirya Temple, north(-west),
l
I sandstone, H: 72 cm, W: 40 cm c. second quarter of the loth century
l, Central Archaeological Museum, Gwalior, no. 63
161 Temple exterior with Kubera (left)and Varuqa in the
182 Kubera (photo: Petra Haubold)
Osiiiii, Siirya Temple 1, (south-)east,
I karya niches (photo: A.I.I.S.) c. early 8th century
Batesarii, Temple no. 4, north, c. late 8th century 183 Agni (photo: Petra Haubold)
1 162 Indra (plloto: A.J. Gail) Osiiiii, Si~ryaTemple 1, south(-east),
l
I Umri, Siirya Temple, (south-)east, c. early 8th century
I c. second quarter of the 9th century 184 Yama (photo: Petra Haubold)
1l 163 Agni (photo: A.J. Gail) Osiiiii, Siirya Temple 1, south(-west),
l
Umri, Stirya Temple, south(-east), c , early 8th century
1 c . second quarter of the 9th century 185 CaiidrdSonia? (photo: Petra Haubold)
164 Yama (photo: A.J. Gail) Osiiiii, Siirya Temple 1, (south-)west,
l Umri, Si~ryaTemple, south(-west),
c . second quarter of the 9th century
c. early 8th century
186 Varuqa (photo: Michael W . Meister)
165 Nairrta (photo: A.J. Gail) Osiiiii, Siirya Temple 1, northern kapili,
Umri, Siirya Temple, (south-)west, c, early 8th century
1 c. second quarter of the 9th century 187 VPyu (photo: Petra Haubold)
1 l66 Varuqa (photo: A.J. Gail)
Umri, SOrya Temple, (north-)west,
Osiiiii, Siirya Temple 1, (north-)west,
c. early 8th century
ll c. second quarter of the 9th century 188 Indra (photo: Petra Haubold)

1 167 VPyu (photo: A.J. Gail)

Umri, Siirya Temple, north(-west),
c. second quarter of the 9th century
Osiiiii, Harihara Temple 1, (south-)east,
c. mid-8th century
189 Agni (photo: Petra Haubold)
I 168 Kubera (photo: A.J. Gail) Osiiiii, Harihara Temple 1, south(-east),
I Umri, Siirya Temple, north(-east), c. mid-8th century
c, second quarter of the 9th century 190 Yama (photo: Petra Haubold)
l
:
169 ISGna (photo: A.J. Gail) Osiiiii, Harihara Temple 1, south(-west),
Umri, Siirya Temple, (north-)east, c. mid-8th century
c . second quarter of the 9th century 191 Nairrta (after D. HANDA1984: pl. 109)
170 Varuya (photo: A.I.I.S.) Osiiifi, Harihara Temple 1, (south-)west,
Nacnii, Caturmukha Mahiideva Temple?, c. mid-8th century
I
I 9th-l 0th centuries
l
I Bhiirat Kalii Bhavan, Varanasi, no. 20768

l
XVII
 192 Varuya (photo: Petra Haubold)
OsiBii, Harihara Temple 1, north(-west),
c. mid-8th century
193 Viiyu (photo: Petra Haubold)
OsiBfi, Harihara Temple I, (north-)west,
c. mid-8th century
194 Kubera (photo: Petra Haubold)
OsiBA, Harihara Temple 1, north(-east),
c. mid-8th century
195 PSiina (photo: Petra Haubold)
OsiBii, Harihara Temple 1, (north-)east,
c. mid-8th century
196 Nairrta (photo: Petra Haubold)
OsiBfi, Snrya Temple 2, (south-)west,
c. third quarter of the 8th century
197 Varuya (photo: Petra Haubold)
Osiiiii, SUrya Temple 2, (north-)west,
c. third quarter of the 8th century
198 Kubera (photo: Petra Haubold)
OsiBfi, Sarya Temple 2, south(-west),
c. third quarter of the 8th century
199 Loose image of Kubera (photo: Petra Haubold)
OsiBfi, in sanctum of the Pipalridevi Temple,
c. first quarter of the 9th century
200 Exterior view with Kubera in karca niche (photo: A.I.I.S.)
Osiafi, Sfirya-Visnu Temple, north(-eastern) corner,
c. last quarter of the 8th century
201 Kubera (photo: A.I.I.S.)
Osiaii, Visgu Temple no. 1, north(-east),
C. third quarter of the 8th century
202 Kubera (photo: A.I.I.S.)
OsiBii, Visnu Temple 2, north(-east),
c. third quarter of the 9th century
203 Varuya in pratiratha (left) and Nairrta in karca niche
(photo: A.I.I.S.)
Osi~fi,Pipaladevi Temple, west wall, southern portion,
C. first quarter of the 9th century
204 V B Yin~ karpa (left) and Kubera in pratiratlla niche
(photo: A.I.I.S.)
Osi~fi,Pipalidevi Temple, west wall, northern portion,
C. first quarter of the 9th century
205 Indra (photo: Petra Haubold)
Osi~ii,Harihara Temple 2, (south-)east,
c. third quarter of the 8th century
206 Agni (photo: Petra Haubold)
Osi~fi,Harihara Temple 2, south(-east).
c. third quarter of the 8th century
207 Yama (photo: Petra Haubold)
OsiBfi, Harihara Temple 2, south(-west),
c. third quarter of the 8th century
208 Nairgta (photo: Petra Haubold)
OsiBii, Harihara Temple 2, (south-)west,
c. third quarter of the 8th century
209 V a r u ~ a(photo: Petra Haubold)
OsiBfi, Harihara Temple 2, (north-)west,
c. third quarter of the 8th century
210 Viiyu (photo: Petra Haubold)
OsiiiA, Harihara Temple 2, north(-west),
c. third quarter of the 8th century
211 Kubera (photo: Petra Haubold)
Osiiiii, Harihara Temple 2, north(-east),
c. third quarter of the 8th century
XVIII
, / / \ / \
212 PSiina (photo: Petra Haubold)
OsiLii, Harihara Temple 2, (north-)east,
c. third quarter of the 8th century
213 Indra (photo: Petra Haubold)
OsiBii, Siva Temple 1, (south-)east,
c. third quarter of the 8th century
214 Agni (photo: Petra Haubold)
Osiiii, Siva Temple 1, south(-east),
c. third quarter of the 8th century
215 Yama (photo: Petra Haubold)
OsiBii, Siva Temple 1, south(-west),
c. third quarter of the 8th century
216 Nairrta (photo: Petra Haubold)
Osiiiii, Siva Temple 1, (south-)west,
c. third quarter of the 8th century
217 Varuya (photo: Petra Haubold)
OsiBB, Siva Temple 1, (north-)west,
c. third quarter of the 8th century
218 Viiyu (photo: Petra Haubold)
Osiilii, Siva Temple 1 , north(-west),
c. third quarter of the 8th century
219 Kubera (photo: Petra Haubold)
OsiHii, Siva Temple 1, north(-east),
c. third quarter of the 8th century
220 ISiina (photo: Petra Haubold)
OsiBii, Siva Temple 1, (north-)east,
c. third quarter of the 8th century
221 Varuya (photo: Petra Haubold)
OsiBfi, SUrya-Visnu Temple, (north-)west,
c. last quarter of the 8th century
222 Varuya (photo: Petra Haubold)
OsiBii, Vis~!u Temple 1, (north-)west,
c. third quarter of the 8th century
223 Varuya (photo: Britta Schneider)
OsiBii, Harihara Temple 3, (north-)west,
c. last quarter of the 8th century
224 Indra (photo: Falk Reitz)
Abiineri, Har,satmiitB Temple, garbhabhitti (secondary
position), c. first quarter of the 9th century
225 Agni (photo: Falk Reitz)
AbBneri, HarsatmBtri Temple, gnrblzablzitti, south(-east),
c. first quarter of the 9th century
226 Indra (photo: Petra Haubold)
Khed, Ranchodji Temple, gnrblzabhitti, ( ~ o u t h - ) ~ ~ ~ t ,
c. mid-9th century
227 Agni (photo: Petra Haubold)
Khed, Ranchodji Temple, gnrblzabhitti, south(-east),
c. mid-9th century
228 Yama (photo: Petra Haubold)
Khed, Ranchodji Temple, garblzablzitti, south(-west),
c. mid-9th century
229 Nairrta (photo: Petra Haubold)
Khe?, Ranchodji Temple, gnrbliabhitti, (south-)west,
c. mid-9th century
230 Varlya (photo: Petra Haubold)
Khed, Ragchodji Temple, garbhabhitti, (north-)west,
c. mid-9th century
231 Vlyu (photo: Petra Haubold)
Khed, Ranchodji Temple, garbhabkitti, north(-west),
c. mid-9th century
 232 Icubera (photo: Petra Haubold) 248 %na
Khed, Ranchodji Temple, garblzablzitti, north(-east), (after J. VIJAYALAKSHMI/M. KRISHNA KUMARI 1991: pl. 19)
c. mid-9th century Galliivalli, Kiimeivara Temple, cleul, north(-east),
233 i 8 ~ n a(photo: Petra Haubold) c. first half of the 10th century
Khed, Ranchodji Temple, garbhablzitti, (north-)east, 249 Indra (photo: Falk Reitz)
c. mid-9th century Ganeiwarpur, Paficapiingava Temple, south-eastern
234 Indra (photo: Gerd Mevissen) subsidiary shrine, (north-)east, c. 970 A.D.
Piihiirpur (Bangladesh), Somapura Vihiira, base frieze, 250 Agni (photo: Falk Reitz)
Panel 29, c. 8th century Ga~!eiwarpur, Paiicapiindava Temple, south-eastern
H: 76 cm, W: 56 cm subsidiary shrine, (south-)east, c. 970 A.D.
Piihiirpur Site Museum 251 Yama (photo: Falk Reitz)
235 Agni (after K.N. DIKSHIT 1938: pl. XXXII b) GaneSwarpur, Paiicapiindava Temple, south-eastern
Piihiirpur (Bangladesh), Somapurn Vihiira, base frieze, subsidiary shrine, south(-east), c. 970 A.D.
Panel 34, c. 8th century 252 Nairhta (photo: Falk Reitz)
H : 84 cm, W: 33 cm Ga~>eiwarpur, Paiicapiindava Temple, south-easter11
236 Viiyu (photo: Gerd Mevissen) subsidiary shrine, south(-west), c. 970 A.D.
Piihiirpur (Bangladesh), Somapura Vihiira, base frieze, 253 Varuna (photo: Falk Reitz)
Panel 39, c. 8th century Ga~!eiwarpur, Paiicapiindava Temple, south-eastern
H: 82 cm, W: 41 cm subsidiary shrine, (south-)west, c. 970 A.D.
Piihiirpur Site Museum 254 V i y u (photo: Falk Reitz)
237 Candra (photo: Gouriswar Bhattacharya) Ganeiwarpur, Paiicapiindava Temple, south-eastern
Piihiirpur (Bangladesh), Somapura Vihiira, base frieze, subsidiary shrine, (north-)west, c. 970 A.D.
Panel 60, c. 8th century 255 Kubera (photo: Falk Reitz)
H: 84 cm, W: 36 cm Ga~!eiwarpur, Paficapii~!dava Temple, south-eastern
Piihiirpur Site Museum subsidiary shrine, north(-west), c. 970 A.D.
238 Agni (centre) (photo: Gerd Mevissen) 256 k n a (photo: Falk Reitz)
Piihiirpur (Bangladesh), Somapura Vihiira, base frieze, in situ, Ga~!eiwarpur, Paiicapiindava Temple, south-eastern
c. 8th century terracotta panel subsidiary shrine, north(-east), c. 970 A.D.
239 Viiyu (photo: Gerd Mevissen) 257 Indra (photo: Falk Reitz)
Piiharpur (Bangladesh), Somapura Vihiira, Ganeiwarpur, Paiicapi~davaTemple, south-western
c. 8th century terracotta panel subsidiary shrine, (south-)east, c. 970 A.D.
Piihiirpur Site Museum 258 Agni (photo: Falk Reitz)
240 Frieze with GayeSa and dikpiilas Ga~!eSwarpur, Paiicapiindava Temple, south-western
(photo: Gouriswar Bhattacharya) subsidiary shrine, south(-east), c. 970 A.D.
Gaur (Malda Dist.lWest Bengal)?, 9th century 259 Y a m a (photo: Falk Reitz)
H: 80 cm, W: 274.4 cm Ganeiwarpiir, Paiicapiindava Temple, south-western
British Museum, London, no. 1956, P.T. Brooke Sewell subsidiary shrine, south, blzadra, cornice, c. 970 A.D.
241 Agni (photo: Falk Reitz) 260 Yama (photo: Falk Reitz)
Biirambii, Simhaniitha Temple, jaga~~zolzana, south(-east), Ganeiwarpur, Paiicapiindava Temple, south-western
c. third quarter of the 9th century subsidiary shrine, south(-west), c. 970 A.D.
242 Yama (photo: Falk Reitz) (photo: Falk Reitz)
Biirambii, Simhaniitha Temple, ga!~Gi,south, cn~zdras'nln, 261 Nairrta (photo: Falk Reitz)
c. third quarter of the 9th century Ganeiwarpur, Paiicapiindava Temple, south-western
243 Agni subsidiary shrine, (south-)west, c. 970 A.D.
(after J. VIJAYA LAKSHMI/M. KRISHNA KUMARI 1991: pl. 14) 262 Varuya (photo: Falk Reitz)
Galliivalli, KiimeSvara Temple, cleul, south(-east), Ganeiwarpur, Paiicapiindava Temple, south-western
c. first half of the 10th century subsidiary shrine, (north-)west, c. 970 A.D.
244 Nairgta 263 Viiyli (photo: Falk Reitz)
(after J. VIJAYA LAKSHMIM. KRISMNA KUMARI 1991: pl. 15) Ganeiwarpur, Paiicapindava Temple, south-wester11
Galliivalli, Kiimeivara Temple, clecrl, south(-west), subsidiary shrine, north(-west), c. 970 A.D.
c. first hall of the 10th ccntury 264 Icubera (photo: Falk Reitz)
245 Varuna Ganeiwarpur, Paiicapiindava Temple, south-western
(after J. VIJAYA LAKSHMI/M. KRISHNA KUMARI 1991: pl. 16) siibsidiary shrine, north(-east), c. 970 A.D.
Galliivalli, Kiimeivara Temple, deleul, (south-)west, 265 h a (photo: Falk Reitz)
c, first half of the loth century Ganeiwarpur, Paiicapiindava Temple, south-western
246 Viiyu subsidiary shrine, (north-)east, c. 970 A.D.
(after J. VIJAYA LAKSHMI/M. KRISHNA KUMARI 1991: pl. 17) 266 NateSa
Galliivalli, KiimeSvara Temple, cleul, (north-)west, Dharmapuri, Mallikiirjuna Temple, ceiling panel, centre,
c. first half of the loth century c. 9th century
247 Kubera 267 Indra
(after J. VIJAYA LAKSHMI/M. KRISHNA KUMARI 1991: pl. 18) Dharmapuri, Mallikiirjuna Temple, ceiling panel, east,
Galliivalli, Kiimeivara Temple, cleul, north(-west), c. 9th century
c. first half of the loth century
XIX
 268 Agni 284 VByu (photo:.IFP)
Dharmapuri, Mallikiirjuna Temple, Bikkavolu, GoliligeSvara Temple,
ceiling panel, south-east, south, jnrigl?E, on the right of Indra,
c. 9th century c. 9th century
269 Yama 285 Soma? with consort
Dharmapuri, Mallikiirjuna Temple, Kiiiicipuram, Kailiisaniitha Temple,
ceiling panel, south, priikcra wall, between Cells 43 and 44,
C . 9th century c. first quarter of the 8th century
270 Nairrta 286 Unidentified deity with a club
Dharmapuri, Mallikiirjuna Temple, Kaficipuram, Kailiisaniitha Temple, priilciirn wall,
&ling panel, south-west, c. 9th century between Cell 52 and the north-eastern corner,
271 Varuya c. first quarter of the 8th century
Dharlnapuri, Mallikirjuna Temple, 287 Two "ISBna type" riders with consorts
ceiling panel, west, Kiificipuram, Kailiisaniitha Temple, priikdra wall,
c. 9th century Cell 46, upper portion of Viniidharamirti panel,
272 VBYU c. first quarter of the 8th century
Dharmapuri, Mallikiirjuna Temple, 288 Two directional guardians?
ceiling panel, north-west, South India, c. 9th century
c. 9th century Government Museum, Madras
273 Kubera 289 Agni
Dharmapuri, Mallikarjuna Temple, South India, c. 9th century
ceiling panel, north, H: 112 cm
c. 9th century National Museum, New Delhi, no. 69.45
274 jSana 290 Agni (side view) (same as in fig. 289)
Dharmapuri, Mallikarjuna Temple, 291 Agni
ceiling panel, north-east, Tirunelveli, 9th-10th centuries
C . 9th century H: 78 cm, W: 45.5 cm
275 Nairgta (photo: IFP) Government Museum, Madras, no. 2675
Pratakota, NageSvara Temple, 292 Agni (side view) (same as in fig. 291)
vimtinn, griv; level (?l, 293 VHylt
c. 9th-10th centuries Tirunelveli, 9th-loth centuries
276 Varuya (photo: IFP) H: 79.5 cm
Pratakota, NiigeSvara Temple, loose image, Government Museum, Madras
c. 9th-10th centuries 294 Agni (photo: Gerd Mevissen)
H: 64 cm Tafijiivir, RiijarijeSvara Temple,
277 VByu (photo: IFP) rlevnkulikii cell, priikiira wall, south-east,
pratakota, NiigeSvara Temple, loose image, c. 1000 A.D.
c. 9th-loth centuries 295 Varuya (photo: IFP)
H: 66 cm Tafijiivfir,RiijariijeSvara Temple,
278 Djkpila frieze (photo: Gudrun Melzer) devnkulikd cell, prEkEra wall, west,
Ellora, KailLsa Temple, western gopura, c. 1000 A.D.
western face, north of entrance, Panels 8-10, 296 ISBna (photo: Gerd Mevisscn)
c. third quarter of the 8th century Taiijiivir, RiijariijeBvara Temple, clevnlclllikti cell
279 ~ ~ h i i s s u r a m a r d i panel
n i (photo: A.1.I.S.) in the north-eastern corner of theprEkEm wall,
Ellorz, Kailiisa Temple, inner (east) side c. 1000 A.D.
of the gop~lra,facing north, 297 ~ i l h a r aof devalrulika of Nairrta
c. third quarter of the 8th century (photo: Gerd Mevissen)
280 Siva YogiSvara panel (photo: Gerd Mevissen) TahjSvfir, Riijariijeivara Temple, c. 1000 A.D.
Ellorii, KailFtsa Temple, 298 ~ i k ~ ~ of a rdevakuliki
a of Varuya
base of the main temple, facing west, (photo: Gerd Mevissen)
c. third quarter of the 8th century Tafijiivfir,Riijariijebvara Temple, c. 1000 A.D.
281 Siva YogiSvara panel (left side) 299 ~ i k h a r aof devalrulilri of Vsyu
(photo: Gerd Mevissen) (photo: Gerd Mevissen)
Ellorii, Kailiisa Temple Tafijiivfir,Riijariijeivara Temple, C. 1000 A.D.
282 Siva Yogiivara panel (right side) 300 Silrhara of devaku~likiof Soma
(photo: Gerd Mevissen) (photo: Gerd Mevissen)
Ellorii, KailSsa Temple Tafijrivor, RiijarfijeSvara Temple, c. l000 A.D.
283 Indra (photo: A.I.I.S.) 301 Siva Vaivlhikatniirti
Bikkavolu, GolingeSvara Temple, (after A. GOSWAMI 1956: pl. 15)
south, jtuiglzii, centre, Kanauj, c. 8th century
c. 9th century Collection Ram Narayan Kapur (formerly?), Kanalj

XX

-,
l,-.\ /'
302 Siva Vaiviihikamiirti
Kiimiin (Bharatpur Dist.), c. 8th-9th centuries
H: 98.5 cm, W: 68.5 cm
National Museum, New Delhi, no. 69.134
303 Siva Vaiviihikamiirti
(detail of upper left portion, same as fig. 302)
Kii~niin(Bharatpur Dist.),
c. 8th-9th centuries
304 ~ i v Vaiviihikamiirti
a (photo: Martina Stoye)
Manasigaiig;l/Goverdhan,
9th-10th centuries
Government Museum, Mathura, no. 87.8
305 Siva Vaiviihiltainfirti (photo: Martina Stoye)
MathurB, 9th-10th centuries
Indian Museum, Calcutta
306 Siva Vaivlhiltamiirti
Etah (U.P.), c. 10th century
Bharat Kalii Bhavan, Varanasi, acc. no. 175
307 ViSvariipa (photo: Linden-Museum, Stuttgart)
South-eastern Rajasthan,
c. 8th century
H: 47.7 cm, W: 45.7 cm
Linden-Museum, Stuttgart, no. SA 01582
308 ViSvariipa (right portion, same as in fig. 307)
(photo: Linden-Museum, Stuttgart)
309 ViSvariipa
Tumain, c. 8th-9th centuries
H: 58 cm, W: 61 cm
National Museum, New Delhi, no. 78.998
310 ViSvariipa (right side, same as in fig. 309)
311 Manu Loltapiila panel
(photo: Museum fiir Indische Kunst, Berlin)
South Bihar, c. 8th century
H: 15 cm, W: 54.5 cm
Museurn fur Indische Kunst, Berlin, no. I 672
312 Manu Lokapiila panel (photo: Gerd Mevissen)
Bihar, c. 8th century
State Archaeological Museum, Patna, no. Arch. 11324
313 Manu Lokaplla panel (incomplete?)
(photo: Archaeological Survey of India)
Guneri, c. 9th century
Bodhgaya Site Museum, no. 208
314 Agni
Samiith, c. 6th-7th centuries
estimated H: 120 cm
Archaeological Museum, Sarnath, no. 535
315 Agni (photo: Museum fiir Illdische Kunst, Berlin)
Rajasthan, c. 10th century
H: 59.5 cm, W: 29.5 cm
Museum fiir Indische Kunst, Berlin, no. I 10072
316 Varuna
Madhya Pradesh, c. mid-8th celltury
H: 39 cm
National Museum, New Delhi, no. 67.37
317 Varuva (photo: A.J. Gail)
Rajasthan, c. 9th century
H: c. 80 cm
Jhalawar Museum, no. 64
318 Vlyu (photo: Museum fiir Indische Kunst, Berlin)
Gandhiira region, c. 2nd-3rd centuries
H: 30 cm
Museum fiir Indische Kunst, Berlin, no. I 4 3
319 Vlyu (photo: A.J. Gail)
Rajasthan, c. 9th century
H: c. 80 cm
Jhalawar Museum, 110.96
320 Viiyu (photo: Sotheby's, after Sales Catalogue,
New York, 19.09.1996, Lot 108)
Rajasthan (?), c. 8th century
H: 48 cm
321 Icubcra
unknowtl findspot, c. 6th-7th ce~lturies
Bharat KalL Bhavan, Varanasi, no. 23981
322 Kubera (?) (photo: A.I.I.S.)
Naresar, Temple no. 17, west,
c. early 8th century
323 Dandapiini
u~ik~lown findspot, c. 5th-6th centuries
Bharat Kala Bhavan, Varanasi, provisional no. T 467
(in 1992)
324 Daijdapiiqi (photo: A.I.I.S.)
Hiiigliijgadh (M.P.), c. 10th century
H: 50.5 cm, W: 40.5 cm
Central Museum, Indore
325 Harihara-Dandapiini (after V.C. OHRI1991: p]. 3.41)
Baijniith (Kangra Dist./H.P.), ~ i v Temple,
a
c. 13th century
326 NafeSa group wit11 four lokapiilas
(after P. BANERJI 1985, ill. 12 - mirror-image, rectified here)
Pattadakal, Mallikiirjuna Temple,
hall, central ceiling panel
c. second decade of the 8th century
327 Visnu AnantaSayana panel
(photo: Museum fiir Indische Kunst, Berlin)
probably eastern Rajasthan,
c. 9th century
H: 38 cm, W: 60 cm
Museum fiir Indische Kunst, Berlin, no. I 9972
 Introduction The allotment of specific deities to these categories of space was
gradually modified, eventually resulting in a system of subtle dif-
ferentiation of elemental realms, characters and functions among
The common Sanskrit expressions referring to a directional the directional guardians (tables 1-11). However, several parallel
guardian deity are "lokop6la" (guardian of the world in general, or versions of this system have always existed and continued to be
of a particular world or sphere) or "dikpiiln" (guardian of a specific described in the later Vedic texts (tables 111-VI).
direction), as well as a few related words. The ca~lonicalgroup of The development in the representation of dikpdlas, which is first
directional guardians consists of eight or ten members. The group traceable from the two labelled guardian figures on the Buddhist
of ten guardians consists of the following members, listed below stipa at Bhiirhut, c. 2nd century B.C. (section 2.1.1), was not a
with their common vehicles and attributes. In the group of eight linear evolution, but a very complex process, consisting of several
(n;fadikpilns), the last two members have been omitted, and for distinct main lines of development, a few of which ended abruptly
the group of four lokaj~dlas,popular in the early period, only the (sections 2.1-2.3). Particularly during the early period of develop-
guardians of the four cardinal directions are represented: ment, various concepts regarding the composition of the group inust
have existed side-by-side.
While emphasis is laid on the visual representation, a study of
Directional Guardian Direction Vel~icle Attributes the literary concepts of the dikjxilas from the time of their earliest
Slmpl~ficdC l i d ~ ~ ~ c t e ~ t ~ n t ~ o ~ ~
pictorial representation (sections 1.3-1.4) is included here, thus
Indra KIIIPof cod\ east elenhant thunderbolt
enabling a comparative study of the literary and visual tradition.
Agni cod or Ftrc rosary and waterpot Although the number of texts including detailed iconographical
Yanla L O Iof~ De.1tI1 descriptions of the dikpdln images is relatively small and the corre-
sword spondence between text and image remains largely unclear, the tex-
also rcfcrrcd to a \ N I I I ~ I (dnthropo~d)
tual study has assisted in the clarification of several problems and
Lord of the Waters
has afforded considerable information. It has not only facilitated in
an~mall an interpretation of the sculptures, but has also highlighted the reli-
antelope billowing clolh gious background and aided in the understanding of the contempo-
God or Ihc Wlnd 01 Air or standard rary artistic freedom.
Kubcra varying, The visual form of the directional guardians is composed of the
mostly !lam usual features of Indian deities: their bodies exhibit signs of excel-
(anthropold)
or ram; lence, which are enhanced by jewellery; the headdress is elaborate;
varying rosary and waterpot one or more attributes, usually carried in the hands, indicate the
w ~ t hCandra, tlic Moon God function and character of the deity; the vehicle of the god, or in its
ISHna "LOICV. north-east bull trident stead, a particular device below the god, indicates another aspect of
outw'lrdly ~denucalw ~ t hSlva his inherent nature. Less frequently, a consort or attendants accom-
Brahmi zenith goose book pany the deity.
I Crcstor I I I I
The time limit of 1000 A.D. has been chosen here, since this is
Ananta nadir tortoise wheel considered the tentative date for the introduction of the complete
Serpent vcli~cleof V ~ \ n u
dikl76la set in the extreme south of India. Shortly before this date, in
North India, the general introduction of four-armed dikpdlns reveals
These gods, some of whom were also worshipped independently, a marked change of concept with new iconographical aspects, which
were given the task of guarding particular parts of a temple. This is not included here. Thus, although the development differed great-
protection against external evil forces functions within an intricate ly in the north and (extreme) south of the country, the date of 1000
system of religious beliefs and ritual practices. During the early A.D. has a special relevance in both regions.
development of stone temple architecture in India, c. 5th through Documentation and investigation of dikpdla representation is
8th centuries A.D.,the probably already existing invisible, ritually still little developed. In some publications, the clikpblns are treated
invoked presence of directional guardians at building sites (section in the same way as in an encyclopaedic work of the last century,
2.1.2), was complemented by the visual representation of the deities J. DOWSON'S Classical Dictiorlary of Hirzdu Mjltl~ology& c. (1879:
in the form of stone panels in high relief. Apart from the depiction 180). This historically rather late version, most probably following
of clikpdlas on temple exteriors, which is largely restricted to North the MZtarigalil6, prescribes elephants as vehicles for all the clikpiilas
India, the directional guardians can also be found on ceilings and as well as for their co~lsorts.This concept, however, played no role
other parts of the temple structure. in art.
Theologically, the dikpdlas represent a group of minor gods, In Indian art historical research during the 20th century, very
subordinate to the main gods of Brrihmanism or Hinduism, with few studies have been devoted to the subject of dikpGlas. A
extensions into Buddhism and Jainism also. Their presence sym- profound article written by J.E. VAN LOHUIZEN-DE LEEUW:Tlze
bolizes space, and thus, their subordillation signifies the main Dikj~tilakasirz Ar~cierztJava (1955), lists some basic facts about
deity's supremacy over the universe. dikj~iilasbut is also rather speculative regarding a hypothetical
In the Indian subcontinent, the awareness of space was highly early concept of dikpiilas that existed in South India. Since then,
developed at an early period. The Vedic ritual practices (the early however, no particular evidence has come up to prove this hypo-
Vedic texts date from about 1500-1000 B.c.),in which the whole thesis.
world was involved in ritual action, first led to the absolute orienta- During the last decades, a number of monographs have ap-
tion of these actions. The necessity to clearly differentiate between peared, which contain sections on diklkilas. A limited number of
the directions resulted in the elaboration of their particular charac- articles has been devoted exclusively to the subject of dikpiilas.
teristics and their descriptions occurred in most of the earlier texts. However, these articles often deal with images only at particular
sites (H.C. DAS1974; G. MJCHELL 1973; Bh. SAHAI1994; L.K. TRI- ual treatment of the deities, with an emphasis on the particular types
PATH1 1965-66; D.N. VARMA1988) or have a limited scope, e.g. not
referring to any of the previous research works (S.B. SRIVASTAVA
1993-95). The rather large number of articles and other works on
single members of the dikpala group contain some interesting
observations and other preliminary results (e.g., K.D. BAJPAI1997; (i.e. Central Indian versus Western Indian vehicles) has been par-
L. VAN DEN BOSCH1982; B.Ch. CHHABRA 1935; D. KLIMBURG- ticularly fascinating. Cases of regional development could be
SALTER1981; M.R. MAJMUDAR 1943; K.P. MERH1996; P. PAL observed as well as cases of transfer or influence from one region
1979; D.M. SRINIVASAN 1990). Another important step has been
made by G. BHATTACHARYA (1987; 1990), who first distinguished
between the usual group of directional guardians and a related group
of deities, for which I have chosen the term "Manu Lokapiilas".
Regarding the basic concept of space, a dynamic aspect of the
generally static set-up of the regions has been recently traced by
Hilde K. LINK(1999: 150-166).
The development of dikpala representation in art has been
briefly sketched by the authors of monographs (T. E. DONALDSON way:
1987; R.D. TRIVEDJ 1990). Their attempts form the true beginnings
of systematic research, as it has been undertaken in the present study.
An unpublished Ph. D. dissertation by R. P. GOSWAMI
A~fadikpfilasirz Literature arid Art (1979) forms an admirable
collection of literary and art historical facts on the Briihmanical,
Buddhist and Jaina versions of the dikpiila traditions, but lacks a
comparative, analytical approach in its art section. It is tempting in
Indian art history, to follow the traditional Indian approach by
regarding the literary concepts as absolutely valid and therefore
only evaluate, literally from case to case, inhowfar these have been
realized in art. However, a truly analytical approach should critic-
ally highlight the relationship between text and image, without
taking an unquestionable authority of the texts for granted.
In the present work, both in literature and art, data has been col-
lected from a broad material basis. It was carefully analyzed and
studied from different aspects. In the process of describing the art
objects and evaluating the data, an unbiased position has been
taken, e.g. by keeping literature and art apart and only treating them
as parallel expressions of a specific tradition. Emphasis will be laid
on the art tradition as a largely independent system, with its own
potential for development. Even when a literary passage is men-
tioned during the discussion, it is treated only as kind of parallel
evidence. In this connection, I refer to a passage from a study of the
relationship between treatises on architecture and sculpture
(Silpaidstras) and works of art:
If we rely exclusively upon surviving shilpashastric texts to inter-
pret Indian art, we shall never understand it, since these texts are
manifestly those of a small group not composed of artists but of their
social superiors, the priestly guardians of cultural tradition. [...l It is
a culturally contextualising vehicle for a field of creativity which
ran the risk of individualisation (how acutely perceived we do not
know) and hence also the risk of a loss of cultural identity on the
part of the individual artist [...l It supports, rather than censors art
by not permitting it to be seen losing sight of its cultural vision. (T.S.
MAXWELL 1989: 15)
The main section of the present study is formed by the chronologi-
cally arranged, analytical descriptions of the art works of specific
regions of modern India. At the same time, subjects leading to a
better understanding of the development and also cross-references
will be included in order to form a coherent picture of the complex
processes which were involved.
In the final, iconographic sections, the general modes of repre-
sentation of the rlikpdlas will be described, followed by an individ-
2
of their depiction, without categorizing too much.
Among the newly gathered information on the material, the consid-
erable regional variation in the depiction of directional guardians
to the other. To investigate whether any political systems were
involved in the latter progress would lead too far in the present
context. This aspect could form a subject for further study. How-
ever, a basic connection between the dikpiilas and political power
already exists, since a true sovereign has to hold sway over the
directions.
In the present work, the regions of India are treated the following
Northern India Jammu and Kashmir, Himachal Pradesh,
Uttar Pradesh (U.P.)
In a sinzl>lifiednzanner, "Nortlz Iizdia " rlzay
also refer to all the regiorzs rzortlz of tlze
Deccarz.
Central India Madhya Pradesh (M.P.)
Eastern India Bihar, Bengal, Orissa, northern Andhra
Pradesh
Deccan Karnataka, southern part of Andhra Pradesh
Southern India Tamilnadu, Kerala
In a siri~l~lified
nzanrzer, "Soutlz Irzdia " 111ay
also refer to the Deccarz, Tarnilnadu
Kerala.
Western India Rajasthan, Gujarat
Spellirzg: For the transcription of site names, diacritical marks have
been employed here to be as adequate as possible, although being
aware of the shortcomings of this method. For the rendering of place
names other than art historical sites, the diacritical marks have been
omitted.
Figures: For the photographs of artworks illustrated here, previous
publications, if any, of the respective piece have not been strictly
referred to.
Directiorzal rcfererzce: The direction which a dikl>iila image faces
is mentioned without brackets, while the direction implied, by the
placement of the image near the respective corner of the wall facing
that direction, is referred to in brackets. Thus, 'north(-west)' means
'placed on the north wall, close to the northwestern corner.'
Map of India with Major Temple Sites
1 Literary Evidence
1.1 The Vedic Literature
Almost all categories of Vedic texts contain references to the
guardians of the directions in space. When it comes to the descrip-
tion of rituals, e.g. in the $atnpathabrdlziizana, these are particularly
numerous. Regarding the contexts in which an invocation of the
directions and their overlords is crucial, the sphere of royal power
and royal consecration has been prominent from the earliest times
onwards. However, a consistent development towards the later
directional lokapdla concept has not taken place in that sphere, and
has therefore been largely left aside here.
1.1.1 Early Vedic Literatore
In the Rgverla, the Atlznwavedn and the Yaj~iwerln,there is ample
evidence for the number of directions. The latter two moreover con-
tain many passages naming the presiding deity or deities. While cer-
tain tendencies and predilections can be discerned, the classifica-
tions often vary. Particularly noteworthy are the Atlzarvavecla hymn,
later referred to as 6iZpdliyanz (AV I.31), which is addressed to the
four directional guardians, and the two so-called cligyukte hymns of
the same text (AV 111.26-27). The guardians' protection is also
expressly asked for in AVXIX.17. In the Rgveda there are possible
traces of a rudimentary development in the idea of overlords
(n'dlzipatis) of the directions. Apart from rather abstract references
to the directional guardians, an interesting visual concept is pre-
sented in TS 1.8.7 c, where the various groups of gods, together with
their leaders, have occupied their seats in the four main directions
and in the zenith.1
1. l . 1.l Origin and Symbolism of the Directions
In a somewhat later part of the Rgverla, the well-known Purusasukta
hymn (RVX.90.14), the directions are described as arising from the
ears of t h e p u r ~ i ~In
a . another passage (RVX.121.4; cf. AV IV.2.5),
they form both arms of a superior being. In later texts, there are ref-
erences to both these notions.2 In the same chapter, the directions
are regarded as being born from the goddess Aditi (RV X.72.3),
immediately after existence itself came forth. This stresses the ele-
mental nature of the regions in the world system for which space is
a natural precondition.
A highly interesting ritual aspect which throws light on their
conceptual importance is the ritual "mounting of the regions"
(digvy6st/z6pananz),which is done with the desire to attain the heav-
enly state. Thus the directions are conceived as forming the entrance
to the celestial region, from which the whole world can be mas-
tered.3 In connection with this, it has been maintained that, on a
microcosmic level, "the regions (points of the compass) correspond
to the covers of the embryo."4
1.1.1.2 The Classification of the Directions
The Vedic texts offer an interesting diversity in the numbers and
positions of the classified directions, starting with the commonly
mentioned fours or sometimes five6 regions (rliS-, pradii-, or BS6)
of the Rgverla and the Atharvave~fa.It is obvious that the four
regions, which are rarely specified (except, e.g. KV VII.72.5), but
without doubt represent the four main directions, predominate in
the Rgvecln, where the earth is also said to be four-cornered (RV
X.58.3). This number further corresponds to the number of heaven-
ly rivers or oceans which are envisaged as originating in the highest
heaven7. In later Vedic scriptures, the number four as referring to
the total number or directions appears less frequently.
Regarding the number five, which includes the central axis as a
point or direction, "It has long been observed that the predominant
role of the number five in the Veda (cf. priiica jn'nn!z, pn'i'ica krgi!z,
prii'ica carp~f!z, priiica ksitf!z) is due to the fact that the five points
of the compass (ycii'ica pradiia!z [RV] IX.86.29) 'represent in the
Vedic scriptures the entire world.' When five points are occasional-
ly mentioned, that in the middle, where the speaker stands (nzadlzy-
atn'!z RS.[RVj X.42.1 l), denotes the fifth."8 Five coordinates of
space continue to appear in later Vedic texts (e.g. SB IX.4.3.10), the
fifth point being the zenith, however. This concept is also rooted in
the Yajurvedic tradition.',
The Atliarvaverla, besides mentioning four or five directions,
quite often refers to six directions10 (including a "fixed region" or
centre - cllzr~ivddik,lI and the zenith - fircihvd rlik), which are gen-
erally governed by certain n'dlzipntis (see below).
I Cf. also JB 11.141 and other passages mentioned by KUIPER 1979: 57, referring
to the ckrivi sabhd or clcrivi srrr!lsdd.
E.g. concerning the senses in tlie Vedinta philosophy, the directions preside
over the ear (MONIEIC-WILLIAMS 1899: 167).
TS I.8.13,I-2; MS 11.6,lO; VS X.lO-14 (cf. SB V.4.1,3-7). and other passages
quotcd by HEESTERMAN 1957: 103. This rite is essential to the royal consecration, and
is completed by a ritual which symbolizes the stepping down (ibid.: 196199). In tlie
SB, an explanation for tlie ascent of the regions lias been provided: "it is the seasons,
the year, that he thereby makes him ascend; and having ascended the seasons, the year,
he is high, high above cverything here" ($B V.4.1.8, after EGGELING). Most revealing
is tlie statement in MS 4.4.4: d i b vdi svcrrgu' lokd!~- "tlie heaven is Lhc quarters of
space" (cf. SBVIII.6.1,4). In ChU 111.13,l-5 tlie five openings of the heart, situated in
the five directions and equated with particular deities and senses, represent the "door-
keepers of heaven" (.sl~argu.synlokasyn dviircrpd!~).
HEESTERMAN 1957: 38 (cf. 105). He continues: "The gods presiding over the
points of the colnpass and the rskyases driven off by them 'from the regions' seem 10
coincitle with the embryonic covers, whereof they then represent between thcm the
protecting and the opposing dangerous aspccts."
RVI.164,42;VII.35,8;72,5;X.19,8;51,9;58,4; 128,l.AVI.l 1,2;31.1;11.10,3;
X1.2,lO; XII. 1,4&55.
h RVII.13,lO; IX.86,29.AVI.30,4; 111.4.2; 20,5; V111.9,15; XIII.3.6. VSXIV.13;
XVI.64-66; XX11.24.
LSIUERS1959: 285-288. KUIPER(1975: 110) rather regards the111as co~iling
down from the central mountain which represents the crsi.s 111rrr7di.
R KUIPBR 1962: 148. RVX.42,Il ("B~haspatisol1 uns von hinten und van oben,
von unten vor dem ijbelwollcnden scliiitzen, Indra sol1 uns van vorn und in dcr Mitte
l...]cinen Ausweg schaffen."[after Geldner]) quoted by him does not enunlcrate all the
directions but refers only to two coordinates on the horizontal plane and three coordi-
nates on the central axis (!). Interestingly, B[haspali reniains associated with the zenith
until Brah~niis assigned to this position. If Indra's position "in front" is interpreted as
an castern one, this Inay be a very early hint at the direction which is later regarded as
undisputably liis (at first quite rarely, as in AV XV111.3,25). The centre has also some-
times been attributed to him in later texts, as in ~ 6 d / t $1.61 ~ (CALAND 1927: 173).
".g. TS V.5,8; V11.1,15; MS 11.8.9; III.12,8; VS X.IO-14; IX.32; XIV.13;
XV.10-14. "The normal term for tlie zenith in the Yajurveda is byhatf dik - 'the high
region', perhaps with assonance of B~haspati."(KUII'ER1979: 57) Here also, KUIPER
expresses liis opinion that ~I/I~III~CI (Ilk (TS: i)rir!i dik) denotes the nadir (see bclow).
l[' AV 1V.l 1,l; 20,2; X.7,35; X111.3,l. This division of space lias pcrllaps bcen
anticipated in RV VI.47,3.
l ' Cf. KUIPEIZ 1979: 57. Imagining the "fixcd", solid earth, a reference to the nadir
seems to be implied. However. the expression avcici clik (MS II.I3,2I) which definitely
points to the nadir, occurs only rarely.
 Of particular interest are the cases where seven directions12have
been described. While one would naturally be inclined to count the
cardinal directions, the centre, the nadir and the zenith, a "trackless"
or probably "midway" (vyadlzvci) quarter has been added to the fixed
and the upward quarter in AV IV.40.1-7, which is also implied
in AV IV. 14.7-8. In the latter passage, most interestingly, the
directions are described as being analogous to the bodily parts of a
sacrificial goatl3. The "trackless" quarter refers to antn'rik.~a(air
space), which can be deduced from the fact that Vjyu presides
over both. 14
The classification of seven directions in AV XV.G.1-9 tran-
scends our present-day notions about space and again shows that the
directions were in ancient times closely related to the lokas, the par-
ticular spheres in space. The directions, called dhr~lvddik, iircllzvd
dik, ~ittamiidik, Orlznti dik, pcircliizii dik, 1zddi;tlzii dik, and n'iitivrttti
dik, all seem to be lined up on the central axis and extend far into
space, which is practically inconceivable. The last-mentioned
region is associated with Aditi, a goddess, who is elsewhere (RV
X.5.7) described as abiding in the highest heaven.
Further examples of different classification concepts include a
mention of eight directions in the Atlzarvaveh-PariSisfn.15 Accord-
ing to LUDWIG (1888: 229,231), RV 1.52.1 1 and VIII.90.13 refer to
ten directions. This cannot be proven, however. It is still interesting
to note that ten directions, which already correspond to the latest
development, are quite explicitly enumerated in AV V.10.1-7. The
concept becomes perfectly clear in SB VI.2.2.34, VI.8.2.12,16 and
Sa@B IV.4.
1.1.1.3 The Early Assignment of the Directions
Here again, F.B.J. KU~PER (1979) has done intensive research and
has covered allmostevery aspect of early classification. ~ ~ ~
to his view, "In Vedic cosmogony [...] cidlzil7ati is the specific tern1
for a god or group of gods who, in the system of classification, have
the function of urotector(s)
. . of one of the points of the compass. As
such they exist as soon as the Cos111os arises from Chaos, for the
world order is inconceivable without a system of classification."l7
In his opinion, the assignment of particular groups of gods, the
vasus, r ~ r h s and
, a~fityas,to particular directions in the Atlrnr-
vaverh and other slightly later Vedic texts is closely related to this
sequence, in which they have already been referred to in the Rgve-
rln.18 Thus, "The Vasus are always associated with the East, the
Rudras with the South and, with rare exceptions, the Adityas with
the West" (ibid.), which implies a clockwise order. For the north,
for the zenith, and, if at all mentioned, for the centre or nadir, the
classification differs.
In tracing the antecedents of the later dikpiila group, however,
it is of particular interest to focus on the single deities associated
with the directions. With rare exceptions, e.g., where all the direc-
tions are allocated to only one god19 or to the months of the
different individual gods or groups of gods generally preside over
the directions. Often they are accompanied by a "defender"
(rak~it!), an "arrester of arrows" (Izetindr?~pratidltartf), or just
"arrows" (i~n'va!z). The position of a defender is never held by a
main god, but is mostly assigned to serpents.21 The terminology
already indicates that a magical protection was being sought.
While some general tendencies in the assignment of the direc-
lions are obvious, such as the allotment of the east to Agni, the south
to Indra, the west to Varuna, the north to Soma, and the zenith to
Byhaspati, there is no consistency in the respective texts. Exceptions
are ubiquitous. In the distributional tables, the evidence from the
Atlzarvaveda and from some Yajurvedic texts has been separately
listed. For the sake of clarity, the mythical names of the directions
as well as those of the defenders, listed by KUIPER(1979: 56-57),
have been omitted here.
Single gods are quite often combined with a group of deities
with which their relationship is not apparent at first glance. Some of
the conventional combinations are Agni with the vasus, Indra with
the nldras22, Yama, originally the first man who died, together with
the pips, the deceased, and Byhaspati with the viive rlevd!z.23 Since
Varuna is regarded as the first-born dditya (KUIPER1979: 39), his
association with this group is only natural. However, the exchange
of groups and other alterations in the classification system point to
the fact that the assignment of the directions in the Vedic texts was
subjected to priestly speculation and hardly reflected a living tradi-
tion. This is in particular indicated by the fact that Rudra, who is
12 A well-known i~ista~ice is RV IX.114.3, starting with saptci diSo 11dn6slir)~i!i
sol116 hdr6r.a rtvija!~"Seven are the directions, each with its own sun. Sever1are thc cer-
e~noniantsand Iiotars." (BKEKETON 1981: 4) Tlie first part is repeated in TA 1.7.4, and
an equation of seven regions with seven harp has been made in SB V11.4.1.20 and
X.2.6.2-3. A mention of sevcn regions further occurs in 9.8 IX.5.2.8. Anotliere~iig~iiatic
case is RV L22.16: cito devd avar~trrr ~ ycito
o vi?r,iurvicnkrurrlk, prrhi~yd!~ saptci cll~dr~iab-
hi! "Von dort aus sollen uris die GBtter helfen, von wo aus Visnu ausgescliritten ist
durch die sieben StBLien der Erde." (FALK1987: 128) Although ihc tern1 dlzGtlia~tdoes
not mean "direction" (GONDA1967: ~~as.siin, p. 42 com~nentingon tliis passage), some
kind of solar sy~nbolis~ii is possibly shared by both passages, as the second passage may
refer to tlie seven different monthly positions of tlic ~iiiddaysun during a year (FALK
ibi(1.).
13 In tliis case it beco~nesobvious that (II~rui~(i dik denotes tlie centre.
l 4 VVriyu governs the "midway" quarter in AVIV.40,6. He is described as l~ll~i/>nti
of arltdrik~ainAVV.24,8 and TSIII.4,5, cf. also J.R. JOSIII1973: 234,239,241, In
VI.3,6 Viyu shares this position with vr;!i (rain). This passage refers to tlie same
division into seven regions, the intermediate one explicitly being called anmrik;n, the
~ ~ d i ~ ~ ~
lower one pflhivi (earth).
15 As in the following Atharvave(1a passage, the intermediate directions are only
treatedcollectively. In this case however, reference to the fear ofparticular perils in the
different rcgions lnay have led to the omission of tlie fixed and the upper quarter.
'Wegarding this passagc, doubts about the general validity of the statement may
arise as two verses before, tlie number ofrcgions is only nine ($B VI.8.2.10).
1' KUIPEK 1979: 27. Sce also: 52-57. Tlie term cidltil~oti,generally translated as
"overlord", is also used for deities presiding over certain realms, groups of beings or
ever1 ab5tnct eriiities (particularly AVV.24; TSIIL4,5). It is inicresting to find sclch ref-
erences co~iibinedwith descriptions of the cidlriparis of the regions as in AV XV.14.
18 "the order in which the three groups were enumerated was fixed ever since tlie
Rigveda. [...l Tlie first question that arises is whether these fixed associations [with the
directions] werc due to tlie fixed ordcr oftlie three groups [...l or if, inversely, tlie asso-
ciation was from the beginning a given fact and thus led to enumerating the groups in
the ordcr of tlie classificatiorr systeni. As far as I can see, [he sccond alternative is cor-
rcct" (KUIPER 1979: 47).
19 In TS 11.4.14 (cf. AV VI.98.3), different aspccts of Indra govcrn ilie four main
directions, while in TSV.5,7, only Rudra is addressed in fivc regions. The lattcr instance
is similar to AV XV.5,l-7, where six directio~isplus tlie i~itcr~ncdiate directions arc
assigned to scveli different names of Rudra.
AV4.1-6 lnelitio~lstwo months respectively as guardians (goplf) forcacl~of the
six directions. The only deity referrcd to is "Aditya" for the zenith.
21 E.g. AVIII.27; TS V.5.10.
22 This seelns to be confined to the Yajrrnle[Io (see table 11) and can be explained
on the one hand by Indrn's close association with the tllor?rt.s (HAZKA 1981: 115-140),
whicli are also called tlie rr~drcrs,and on tlie other hand by the rather liabitual enumer-
ation of gods, in whicli the r?c(lr.c~sfollowed the ilasrrs and Indra followed Agni. For the
rr~dms,cf. AKBMAN 1922: 156-7: "Docli kennt [...l die spitvedisclie Literatur einen
Begriff r~~drci!i,dcr sic11 niclit niit Rudrns Scliaren identifizicren IXsst, niirnlicli jeiie
Rudras, die in1 Ritual allei~ioder zusalnliien niit dcn Vasus [...l i ~ n ~ nin e rganzlich far-
blosen, scliablonenm8ssigenZusam~iicnstellungeliersclieinen. Diese Rudras liaben mit
Rudra nichts gemein, lassen sich aber ebenso wenig nit den Maruts identifizieren."
23 "The nature and the function of tlie vis've dci~ri!i are still obscurc, altliough
they were clearly connected with the idea of totality" states Kulren 1979: 50. For
a rccent discussion of tlie encompassing nature of this group see BODEWITZ 1973:
87-89, n. 26.
 Table I - Directional Guardians mentioned in the Atl~arvaveda

nntcirikp Zdityas ipn, d;acl/~i!z Indra, r~iaruts

Table I1 -Directional Guardians mentioned in the Ynjurverla

/

- MS
Yajurveds
11.8.9;
East
vasus,
South
nrdras,
West
idltyas,
North
ntnnrts,
Zenitli
viSve clevi!~,
CentreINadir
-
KS XVII.8; Agni Indra Sorna Varu~!a Byhaspati
KapS XXVI.7
MSII.13.21 Agni Indra Sorna Varuna Brhaspat~ V i ~ n u(nvici dik)
/

TS 1.8.7 Agni Yarna Savitr Varuna Brhaspati -

/

TS 1.8.13 Agni Indra r~iaruts Mitra, Varuna Brhaspati -

- TS IV.3.3
TS V.3.4
Agni
Agni,
Indra
DhBtr,
viSve devc?!~
Mitra,
Mitra, Varuna
Indra, Savitr
Brhaspati
-
-
-
vnsus iclityns Aditi
P-
---

-TSTS V.5.8
/

Agni Indra Savitr Mitm, Varuna Brhaspati -

V.5.9 vnsus, nrdrns Yarna, pi+ Hdityas, viive Dyutina MBruta, "the gods, whose chief is Indra, [...l below and [...l
clevfi!~ IIICZ~LL~S above"

-TSTS VI.1.5
V.5.10 Agni
Pathyi Svasti
Indra
Agni
Soina
Soma
Varuna
Savitr
Brhaspati
Aditi
Yalna (iyom dik)

VS IX.35 Agni Yarna viSve devE!z Mitra, Varuna, Soma -

- VSXV.10-14; VNSUS, nrdrcls, Zelityas,

or lllclnlts
~~tar~its, viive de~lc?!t, -
(So VIII.G.1.5-9) Agni Indra Varuna Sorna Brhaspati
4

traditionally associated with the north, never functions as a guardian
of that direction in early classification^.^^
The tables show that the traditions of the Atlzarvaverla and the
yajurveda differ to a certain extent. In both the Vedas, Agni and
~ n d r aquite regularly appear, on the east and on the south respec-
tively. However, Soma and Varuna generally occupy the west and
the north in the Atharvnveda, but tend to have exchanged positions
in the Yajurvecla, the latter often being conjoined with Mitra. Gods
tyhich are less regularly included, such as Savitr (west, north, or
Zenith) and Dhiitr (south or north)25, often vary in their positions.
T h e goddess Aditi seems to have a certain connection with the
central parts of the coordinate system (rlhrcivd dik and zenith), but
she is also found on the west. The few cases in which deities are
to one region are Yama and the pitrs26 for the south and
prhaspati for the zenith.27
24 Admittedly, the numerous references to the north being Rudra's direction gen-
erally come from later phases of thc Vedic literature: TS 11.6.6.5; 5% 1.7.3,20;
I1.6.2,5&7; V.4.2,10; IX.I.I,IO;Ap~hSII.2.4,6;rip~~XV.1 l,7; K'III~B11.3. However,
it can be assutned that this traditional association is older. The only mcntion of Rudra
as the sole guardian of tlic north is ~nndcin a passage of the fragments of t11c viditSs
(CALAND1927: 173).
25 Tlie latter cvidencc comes from SOXIV. 1.3,22.
KUII'ER 1079: 56, n. 183 proposes a taboo as "tlie main reason why the I'itcirtrs
are but seldom mentioned in the system of classification in connection with the region
that is characteristically theirs." However, the observation already made in the casc of
Rudra may also be applied lierc, viz. the general exclusion of folk beliefs from tlie rit-
ualistic connotations of the directions. Regarding the latter, it is highly significant that
hc is treated separately in systems in which all the regions are exclusively rulcd by him
(cf. note 19).
27 SCHMIDT 1968: 87-89 lias argued for ~ h cpriority of Savitr as guardian of the
zenith (as inAVXV111.3,29).

,.
 Regarding the whole situation from the point of view of the later
development, it is interesting to note that in the Atlznrvnvedn, with
the rather regular assignment of the west to Varuna and the north to
Soma, the later convention was already known. Yama has made his
appearance in both the Athnrvnvedn and Ynjurvecin as being associ-
ated with the south. In the following development, he must have
ousted Indra, who also quite regularly occupied this direction. The
latter's shift to the east, which here occurs only once (AV XVIII.
3.25), has not been initiated as yet. V i ~ n ufor the centre or the nadir2s
still keeps some mythological connection with Ananta, who later
guards the latter, and Brhaspati can be regarded as the precursor of
BrahmI as guardian of the zenith. The intermediate regions have so
far only been treated collectively. However, their mention shows
that they had not escaped the attention of the early priests.
1.1.2 Evidence from the Briihmagas, &anyakas, Upanisads
and Siitras
The literary stratum consisting of ritual and philosophic texts of
c, first millenium B.c.,viz. the BrIhmanas, Aranyakas, Upanisads
and Sfitras, witnesses the advent of the loknl~nlaconcept, although it
appears only in a later phase and even then sporadically. The main
emphasis in the treatment of the directions is still placed on the con-
tinuation of the earlier traditions, however (see tables 111-VI). In
this period, the assigninent of the directions to particular deities
becomes much more lucid than before, which is due to the theolog-
ical development reflected in the scriptures. The number of classi-
fied directions generally remains five to six, with the rare inclusion
of the intermediate directions, to which groups of deities are often
assigned29. Interestingly, these aviirztnrndiks are, in $B XI.1.6.27,
associated with the pntnisnrpynjns, i.e. offerings to the wives of the
gods, e.g. Indriini, RudrBni, ~arviini,and BhavIni (PGS 111.8.10).
Thus it becomes clear that the substantially different treatment of the
intermediate directions has not been resolved in the majority of the
later Vedic scriptures. A few novel traits of directional classification
will be introduced here. With all the emerging objects of classifica-
tion, it is often extremely difficult to judge, when they were newly
introduced, and secondly, whether they could have been the cause
for a change in the traditional systems of directional assignment.
1.1.2.1 The Tripartite World (triloka) and the Directions
Apart from the classification of the "powers of the sacrifice (metres,
stomas, sImans) and the forces which constitute between them the
social cosmos (brahman, ksattra, viS, supplemented by force and
lustre)"3o, the fundamental divisions of space, viz. the earth, the
intermediate region, heaven31 and "the quarters"", are for their part
assigned to the directions in a clockwise order, starting from the
east. The divisions of the tripartite world and the directions, which
- at least partly - penetrate these divisions and possibly even tran-
scend them33, are thus invoked, always with a view to mastering
them (GONDA1966: 51-53). This highly complex system of spatial
reference, which might already have been implied in the earlier
Vedic tradition, largely accounts for the distribution of directional
guardians. Thus, the most colnnlon order of Agni (east), Il~dra
(south), Varuna (west) and Soma (north) can be easily explained by
the fact that Agni is classified as a terrestrial deity, Indra as govern-
ing nrztnriksn (together with Viiyu), and V a r u ~ ~asa an dditjln, thus
belonging to heaven by definition33 (MACDONELL 1897: 43,54,88).
Soma has ancient connections with the directions, a fact which
qualifies him as their leader.35 A conscious application of this dis-
tributional system, however, is only found in SB VIII. 1.1.4-1.2.7
and 6.1.16-20, where Agni, Viiyu, Aditya and dis'alz - the directions
(respectively, yajiin - sacrifice) are assigned to the directions in a
clockwise order. The contrast between this and other, often varying
assignments further shows that this "multidimensional" system of
spatial reference must not necessarily be presupposed in every sin-
gle case. However, at this stage, the sequence implies a progress
towards the concept of totality.36
28 When SCRMIDT 1968: 87 states this as a general rule, lie is criticized by K U I P ~ R
(1971: 283). According to tlie latter's view, Varu~pis the supporter of the world axis
par excellettcc (as in RV VIII.41,IO). Kuipcr continues: "Nowhere, however, is Visnu
said to support the skati~bhd-.Me rather is the pillar, and his natural place would be in
ils middle, as the connecting link between underworld and upperworld."
2 " ~ 1.8.1.40 assigns tlic pifrs to tlie i~itcrlnediatedirections, Scr~lB IV.4 lists the
cidityas, urdrcts, riiaruts and vtrsus as presiding over tlie same.
30 k I 1957: 104.~Cf. VS X.10-14.
~ ~ ~ ~ ~ ~
31 E.g., as referred to by DANDEKAR 1979: 300, the fundamental threefold divison
with tlic respective presiding deities Agni (earth), Viyu (attmrik~n)and StirydAditya
(heaven), lias been expressed in RV X.158.1. For tlie "three worlds" Ohtir (= prfhivi),
bh~li>as (= atttarik~a),and simr (= dyalrs) sec WAYMAN 1986.
32 These four entities are often mentioned as being gained, in this succcssion, by
thc four steps of Visnu, e.g. TS IV.2.1 a: SBVL7.2.13- 1 6.
33 Here we touch on a cotnplex problem which has to be tackled in a two ways.
Firstly, it must be asked whether all known directions or only four are to be counted,
and secondly, the real relationship between tlie lokas and the diias must be established.
Another basic qucstion concerns the principles of classification. Regarding the classi-
fication of four items, it lias been stated: "The fourth element often represents some-
thing which on the one hand is added to a threefold totality and on tlie other hand
includes the three preceding items" (BODEWITZ 1982: 47, cf. HEESTERMAN 1957: 34-
37). and, "The fourth can, by its inherent ~iumericalvalue, be heterogeneous, represent
another concept and connect tliis in some way or other with tlie three" (GONDA1976:
I 17). Although in most cases the riurnber of directions lias not bee11specified, accotd-
ing to KLAUS(1986: 27.48-50) and GONDA(1976: 117) tlie (four) directions denote
only the horizontal extension of tlie universe, without includilig the lokas. That the rcla-
tionship between the divisiolis of the tripartite world and the directions is highly com-
plex and somewhat paradoxical, has been summarized by GONUA1966: 110, n. 17:
"The quarters are said to be both inside and outside these 'worlds' (SB. 6,5,2,7), but
elsewhere (7.3,1,13) tliey are in tliese 'worlds' (cf. 6,5,2,6), or between them [viz, heav-
en and earth] (7,3,1,27); 8,1,2,4 tliey are said to be 'above everything here' and at the
same time to be identical with the svarga loka [heaven]; according to SB 6,5,2,22;
13,6,1,3 all this (universe) is as great as tlie lokas and the regions (difa!~)."
34 E.g. GONUA1974: 220. However, Varuna never presides over heaven. An
attempt to replace this god in tlie west by Stirya, the ruler of tlie heavenly realm, liad
already been made in AVXIX. 17.5, and Aditya takes tliis position in SB VIII.1.2.1 and
VIII.6.1.18.
35 RV 1X.86.29: "Dein sind diese finf Weltgegenden in ganzer Ausdelinung"
(after Geldner). AilB I. 14 relates how the gods and dclnons liad fights in tlie four dircc-
tions, the latter always being victorious. Then, "They contended for the north-eastern
quarter; tliey were not conquered thence. This is tlie unconquered quarter [...l 'Through
our lack of a king they conquer us; [...l' They ~nadeSo~iiaking; with Soilia as king they
conquered all tlie quartcrs" (after Keith). According to BODEWITZ1982: 50-51, the
quarters and SomalCandra~nasl~lioon, together with otlicr members, sharc the fourth
position in tlie cosmic classification. A more direct connection betwcen Illern has been
described by GONDA (1967: 74-75): "By pcrforniing tlie ritual act called ttigrcibka [...l
the gods made Soma 'consort' (copulate) with tlic regions as his 'matc' (tttbhrrr~etta),
his 'loved resort' (EGGELING: $5. 3,9,4,20). Here the qucstion arises as to how the quar-
ters of the univcrse could at tlie same time be spouse and dhi~iian[abodc] of Soma."
3 V h i s scclns to be implied by tlie assignment ofthe viSve devcib ("all tlie gods")
and tlie dual deity Mitrivarunau (cf. note 54) to tlie west or to the north. The zenith, by
its very position, represents the totality. An intercsting succession of heavenly strata is
described in Kou?U 1.2-3, where tlie deceased person has to pass successively through
the worlds of Candramas, Agni, Viyu, Varuna, Indra and PrajHpati, to reach tlie world
of Brahman. Witli some modifications, it is possible to apply this succession to tlie
system of directional classification, starting with the north, proceeding clockwise, and
again touch the north (on a higher level) with Indra. Either both Prajipati and Brahman
have to be located in tlie zenith, or the latter has to be imagined above the other, or Brah-
man's world is here conceived of as transcending space. The cosmograpliical inconsis-
tency lying in tlie fact that Agni is enu~neratedafter Candramas has been noted by
TIIIEMC 1951-52: 33. Our interpretation wouldexplain this succession. Curiously, com-
paring this succession with the known types of classification, it is also possible to start
with Ca~~dranlaslSoma on the south and proceed anti-clockwise, wit11 Agni on tlie east,
VHyu on thc north, and so on
 1.1.2.2 The Classes of Beings and the vnrlzas
The classes of beings are not entirely new objects of classification37
and their assignment still varies to a certain extent. The Satn-
pnthabrrllzma!zn alone offers three different versions of their direc-
tional distribution. In the usual clockwise order, the gods, pitrs and
men are allocated to the east, south and west, respectively (SB
1.2.5.17). The second version substitutes the serpents (snrpn) for the
human beings on the west,38 while the latter have been shifted to the
north (SB 111.1.1.6-7). In the third version, the gods and pitrs again
keep their positions, while "cattle" (paiu) comes on the west, "off-
spring" @rajcl) on the north; the sacrificer (ynjnr?zHan) himself is
referred to in connection with the zenith (SB XIV 2.2.28). From
these concepts, which may well have been popularly accepted, it is
not difficult to arrive at a valuation of particular directions. The
north and the cast were generally regarded as "positive", while the
south and the west were regarded as "negative" (cf. MOODY1980:
80). The most prominent positive valuation often occurs for the
combination of the north and the east, the intermediate direction of
the north-east which has been characterized as the direction of gods
and men.39 The unrivalled positive quality of a direction which was
practically excluded from the classificatory scheme (and comple-
mented by the extremely negative association of the south-west,
Niryti's direction) may have formed one of the reasons for the even-
tual break-up of the four-plus-two scheme of spatial reference.40
The vartzas are also not entirely new objects of classification.
The VS X.lO-12 refers to the three superior social classes starting
with the highest one in the east. Although the directions are not
expressly mentioned in SB X.4.1.9 and JB 1.68-69, Agni is recog-
nized as the deity of the briilznza!za, Indra as the lord of the k~atriyn
and the viive &vH!z as governing the vni;ya, while no deity is asso-
ciated with the iiidra.41
1.1.2.3 The Sacrificial Ground
Considering the overwhelming importance of the layout of the
sacrificial ground" and the positions of the priests on the one hand
and the directional classifications on the other, a close connection
between the two should be expected. A high degree of directional
determination in the ritual actions moreover provides innumerable
hints as to the directional symbolism involved. However, the dif-
ferent levels of sylnbolism and the differing traditions both make the
whole sphere of ritualism a difficult field of enquiry. Thus the lay-
out of the sacrificial ground offers a safe starting point.
The three fires, viz. gdrlzapntyn (west-circular), dhavarztyn (east
- square) and r1nk;iydgni (south - semi-circular) through their dis-
tinct forms reflect connections with the divisions of the tripartite
world (discussed under 1.1.2.1). The most common assignment is
gfirhnpatya -earth, 6havantya - heaven, clnk;i!zrigni - intermediate
region, while the fire-less space on the north may stand for chaos or
the "fourth world" ($B 1.2.4.12-14).43 During an interesting ritual,
thepaficerlhriliya, which is performed at the close of the crlturnzdsya
year, the Hhavarliyn fire is divided and pushed out to the four points
of the compass, leaving a fifth fire in the middle.44This practice was
meant to drive off the demons from the main directions.
Regarding the priests' positions, it is generally accepted that
they are symbolically connected with particular directions." The
assignment propagated in SBXI11 (5.4.24; 6.2.18; 7.1.13) is: hot: -
east; brahmarz - south; arllzvaryu - west; udgrltr- north. Some other
instances of spatio-directional significance can only be speculated
about: the patni's (wife's) seat on the south-west, which has been
discussed in connection with Aditi's position," the rubbish (~itkarn)
heap on the north, which might bear an allusion to Rudra's posi-
tion.47 Curiously, the north-east is characterized by aquatic symbol-
ism, as the prn!zitrl!z water-pot is located there.48
A relationship between the directional guardians (goptgs) and
particular areas in the elaborate sacrificial ground is established in
the ViiclhlhLllnsiitrn (table VI). Interestingly, the enumeration is anti-
clockwise, and the three lokas are equated with the vecli, uttarnvecli,
and uttarnrzrlblzi. Thus they are excluded from any directional impli-
cation.49 In descriptions of the agrlicnynr~nritual, during which an
elaborate altar of cosmographic significance is constructed, the
3' An earlier stage oTthis classification is mct with in TS 1.6.5 d: Gods and pricsts
(rfvijn!~)arc allocated to tlic east, tnonths and thcpit!.s to thc south, houscs and cattle to
the west, waters, plants ancl trccs to the north and the sacrifice, the year, and the lord of
sacrifice (ycrjfinl~nli)to thc zcnith. Another passage fro111this text distributes tlie gods,
pit!& men and rrrdrcrs to tlie east, south, west, and north respectively (TS VI.1.1.I).
A close relationship bctwecn tlic serpcnts and Varuna has been postulated by
SHENDE 1948: 281 and JOHNSEN 1965-66: 260-261.
" E.g. SB VI.6.2.3. It is adtlcd tliat thc gatcway to llcavcn is also located here.
Exceptionally, according to TA 1.23, the IIICI~:111dtlie gods arc both assernbletl in the
zenith, while in MtritrU VII.6 tlie men sham thc nntlir with rclativcly inferior beings
(see table V).
40 This is also corroboratcd by the cxceptio~ialassignment of particular deitics or
groups of dcitics to the north-east (table VI). On the material level, tlic importance of
the summer solstice may havc increased during this pcriod, as advocatcd by FALK
(1997: 87). The sunrise on the north-east annouliccs tlie advent of the rainy season.
4 1 The tlirectional order "Agni - Indra - vihe cle\~ci!~"occurs in $B V.5.1, whilc
the devci-lesscharacter of the north is also known from $B V111.1.2.4.
42 Sketches are available in EGGELING 1885: 475; SnN 1978: plans 3-6; STAAL
1983: inside of the book cover.
4 " ~ 0 1980:~ ~ 83-87. The differcnt order is highly significant with regard to the
matters discussed under section 1.1.2.1, where the enumeration starts with "tlic earth"
on the east (this is still maintained by TIIIEME 1966: 89). Then the sequcnce of thc three
loltns would start with the west and proceed anti-clockwise (see below). Exceptionally
howcver, antcrriksn is represented by the - generally fire-less - space on the north: In
CI1U111.13.1-5 (table VI), Candra is allotted to the south, terrestrial and celestial gods
to thc wcst and cast respcctively, and a typical nrrtarik~ndeity, Parjanya, to thc north.
However, a linear arrangement of "worlds" is favourcd by SB VI.1.2.23, c111tcrrik.yr
being represented by the cllri.~!~ycrfircs wliicli are situated on the central axis, bctwcen
the gcirlicrpntycr (representing earth) and thc Bhavaniya (representing heaven). I11 any
case, diffcrcnt interpretations may have existed sitle by side. For furthcr relleclions
on thc spatial symbolism of tlic sacrificial fires see S~~ARREBOOM/HEESTERMAN 1989:
119-123.
44 HEESTERMAN 1957: 31-32,37-39. It is likely tliat the objcct of this fivefold par-
tition, the cihtrvcrr~iycr,sy~nbolizesthe idcntity of thc directions with the heaven.
45 In SB X11.8.2.36 it is statcd that tlie officiating priests arc [analogous with] tlie
quarters. JB 1.72 expresses tlie fact that particular priests arc conncctcd with particular
deities, whose quarter they face. See Booewrrz 1984: 43-45 on tlie southern position
of the b,uhrl~brzpriest. A relationship between tlie pricsts and tlie directional guardians
is established in the r1i.fci111ove.gcryer!l (HEESTERMAN 1957: 196), wlicrc it is striking to
find "Mitra and Varuna" related to "both adhvalyus".
" SB 1.3.1.17.: "Let lier not sit to the west of thc sacrifice, with her face towartls
the cast. For Aditi [...l indeetl sits on tlie west [...l and this lady would, thercfore. rnisc
herself to lier [...l. For this reason let her [thepnrrli] sit somewliat to tlie south." (aftcr
EGGELING) At tlic same time there is a dcfinitc - though mthcr co~nplcx- connectio~l
of the patni with Aditi and Varu~y(5% 1.3.1.12-16). Considering thc signific~uiccof tlie
south-west, it is a well-known fact that this direction bclongs to Nir!.ti ($U V11.2.1.8).
The strongly ncgative bias of this goddcss in that period, liowevcr, forbids any specu-
lation about a possible sy~nbolicconnection with Nirrli.
47 Rudra, who is known to inhabit tlie north, is "most closely linkcd in ~nytliand
ritual to tlic remainder. He is tlic godpcrr exceller~ceknown as uccheser!rohhcigu" (SIIINI-
VASAN 1983: 555, cf. 552-554). Elsewhere, the rrrkscrs arc invoked at n domcstic rub-
bish licap (GGS 1.4.1 1).
4m The north-east is clearly connected with watcr, which is also hinted at by tlic
rite of sctting up a watcr barrel in that position during the construction of a house (AWE
1954: 142). The symbolic character is described as decidedly fenlale in SB I. 1.1.18-21.
Accordingly, the fcmale water has to be placcd north of thc ~nalccihcrvctr~iycrfire on the
cast, "sincc thc woman lies on the lcft (or 11ort1i)side of the man" (aficr EGGCLING).
4"cfercnces to a practicc of sacrificing in three differcnt altitudes (ARBMAN
1922: 120) point to the salnc idca.
 Table I11 - Directional Guardians mentioned in various Briihmanas

BrHhma~!a East South West North Zenith Nadir Centre Intern~ediate

AitB 11.1-2 Aditi Agni Soma Savitr (Pathyii) - - -
AitB VIII. 14 VNSLLS lsirlras iditjas viivr deviilz mnnits, srirllzyns, iptyas - -
arigirasas
JB 11.141 V~SL~F r~rdras iidityns ~nar~rts viSve devii!~ sidlzyas, Gptyas Indra -
SaclB IV.4 Agni Yania Varu~~a Sorna siitllzyas pitrs - iirlit)ias, rudms,
177awts. vnsus
SB 111.2.3.15-18 ( Agni ( ~oma ( Savitr ( Pathyii Svasti ( Aditi - - -
SBV.2.4.5 Agni Yama I vis've devci!~ I Mitra, Varuna, I Soma - - -

SB V.4.1.3-8 Agni Indra 117crruts Mitra, V a r u ~ p Byliaspati - - -

SB v.5.1 Aeni Indra or Soma viive deviill Mitra, Varuna Brhasoati - - -
I I 1 I I-
SBVIII. 1.1.4-2.7
SBVIII.6.1.5-9 I
Agni
Agni, vcislrs
Viyu Aditya diicr!t
I Indra, nrdrns I Varuna, iidityas I Soma, nrarirts I Byhaspati,
Candra~nas

I vifve devi!~
(
I
-
-
-
-
- 1
SBVIII.6.1.16-20 Agni VHyu 1 Aditya )lojficr I Pnrjanya - - -
SB ~ 1 ~ . 1 . 3 . 1 9 - 2(3 Agni ( Indra I Savit~ 1 DhStr 1 Brhaspati 1- -
TB 111.1 1.5 I Agni Indra I ~oma I Mitra, Varuna I Brhaspati ( - l Aditi I -

Table IV - Directional Guardians mentioned in various &anyakas

Aranya~ta East South West North Zenith Centre

JUB 111.5.2.2 Indra ISHna Varu~!a Sonia Prajiipati -
TA 1.23 Aditya Agni vayu Indra men, gods, pirys, Ptisan
garrcll~arvcrs,a/>sarcrses

Table V - Directional Guardians mentioned in various Upanisads

Upaniyad East South West North Zenith Nadir

B 11919-25 Silrya Yama Var~n!a Sonla - Agni
(ccrky~s) (ycjfin ) (il~as) (clik,~fi) - (vic)
CIzUIII.1-I0 vnsirs, Agni rirdrcrs, lndra n'/if)as, Varuna rrraruts, Soma sc7dlzycls, Brahman -
(mhita) (brkltr) (kr:~qti) (parn!ikr,y,iu) -
CIzU 111.13.1-5 Stirya Candra Agni Parjanya Viiyu -
MtlitrU VII. 1-6 Agni, vasus Indm, Sorna, r~rtlras nranrts, idifyns vis've devi!~, Varutp, Mitrivarunau, ~ a n iRahu,
, Ketu,
sdtll~yas trrigirasns, Candranias uragns (serpents),
rnksas, yak.ys,
mcn, etc.

regions are adequately referred to. Here it is significant that the
"trilo1cn"-symbolism, which seems to have originally been separat-
ed from the directional symbolism,s0 has to a certain extent been
intermingled with it in SB VIII (table III).SI At the same time, a
transference of the directions from nrltariksn to heavens2 seems to
have taken place, but diSyG bricks are still foulld in the third,
nrztnriksa layer (SB VIII.3.I. 1 1-14). Another version of the corn-
plex symbolisln of the ucelestiay layer consists of the tenprfi!lns on
the east' four gods and six cosmic On the
south, the ten directions on the west, and the ten months on the north
($B VIII.5.2.13). The altar, which often assumes the form of a bird
(iyenn or slrpnr!~a)oriented towards the east, clearly refers to the
east as the region of the gods.
5" "~nthe first, third, and Tiftli layers Ile tlie three naturally-perfolated brick$,
representing thc three worlds through which he [the sacrificer]will have ro pass on his
way to the fourth, inv~siblc,wotld, the realln of i~n~nortal Ilfe." (EGGCLING 1897:XXII)
Intcre\lingly, rlie directions arc symbolized by diSyZ bricks in tlie layer which cor-
responds to nrltnriLgn (TS V.7.5 c; SB VL2.3.4).
51 Hcre, $8 VIII.1.1.4-2.7 belongs to the first layer of the ag,ticcr)ar~n, tlie two
belollg to layer correspondingto heaven,
52 Cf. note 44, $B VIII.6.1.4.
 Table V1 -Directional Guardians mentioned in various Sfitras

Siitra East South West North Zenith Nadir Other Coordinates

ASVGS1.2.5-6 [B] Indra Yama; pitys Varuna Sorna; r a k ~ a s Brahman - -
BkGS 111.13-14 [B] IndraIAgni Yama Varuna Soma - - Centre: Brahman, Prajiipati;
NE: saptay~i,sarvoblzlitd!~
GGS IV.7.41 [B] Indra Yama Varuna Soma Brahman Viisuki SE: Viiyu; SW: pitys;
(crit. cd : 1V.7.36) NW: Mahiiriija; NE: Mahendra
Kau;GS 111.10.6-8[B] Indra Yama Varuna Soma Brhaspati - NE (opordjitd rliSd):
11raruts,ns'vir~s,chandor,~is;
SE: Agni; SW: Niryti (fein.);
NW: VByu; NE: Rudra
MGS 11.12.12-16 [B] Indra Yama Varuna Soma - - Centre: Brahman
(cf. KdthGS 44.1 1-16;
ViDlzS 67.15-20)
MSS~1.7.1.7 Indra Yama Varuna Soma, Kubera Brahman Ananta SE: Agni; SW: Nirrti (fern.);
NW: Viiyu; NE: k n a
PGS 111.4.14-17 Agni, Aditya alzar, rdtri nrzna, prd!?a Candramas,Viiyu - - -
$CS 11.14.6-7 [B] Indra Yama Vamna Soma Brhaspati - Direction of the sun (?):
Aditi, ddityas
$$S IV.21.8-12 Agni, vasus Yama, pitrs Varuva, ddityas Indra, r u d r u ~ Prajiipati, - -
viSve devdlz
$$S VI.3.1-7 Stirya, Candra Yama, Mytyu Mitra, Varuna Soma, Rudra Byhaspati, - antarik~a:Viiyu, vygi (rain);
Indra Pythivi: Agni, arzrzn (food)
vddlz$S 26c vusus r~~dras ddityas viSve devdh sddlzyas arigirusas -
(CALANO 1927: 130)

vadhSs 6 1 Aditya Yama Aja Ekapiid Rudra - Centre: Indra (hotriyn)

(CALAND1927: 173) (dlzavaniyn) (nidrjdliya) (gdrlzapatya) (r7grzirll1ri)la)
ViirGS XVII.6 [B] Agni Yama Soma Varuna - - Centre: Varuna,
Aryaman, Brahman

1.1.2.4 The Senses ~ r a u t a s ~ r($SS

Another level of meaning, which is attached to the directional
guardians in the $atnpathabriihma!za and in some of the Upanisads,
is the microcosmic sphere of the senses or sense organs and other
bodily faculties. It seems as if this kind of classification has - like
the one mentioned under 1.1.2.1 - promoted the position of Vayu,
who was only sporadically included in the earlier systems. Here,
vgyu is the representative ofprcipa. Aditya or Siirya is always asso-
ciated with the eye, as Agni is with speech. The directions (diia!z)
are related to the ear (Srotram). However, the classification of the
senses does not seem to imply any strict order of enumeration and
therefore exerts no influence On the spatial distribution of the
deities.
1.1.2.5 Double Assignment of Deities
the literary stratum under discussion, the assignment of two
deities for each direction5) occurs more frequently than in the early
Vedic texts. This phenomenon may be regarded as harbin-
ger of the subsequent break-up of the four-pl~~-two scheme of spa-
tial reference. Whereas Mitra and Varuna, as the dual deity
Mitrsvarunau, almost regularly occupy the north in the yajurvedic
tradition and appear as a complementary coup1e9a comprehensive
"dualit6 unit6",s4 one Gyhyasfitra (PGS 111.4.14-17) and one
a VI.3.1-7)55 offer a distribution of the deities, in
which those gods who may be regarded as similar in some respects
have been paired.56 It may therefore be postulated that it was felt
necessary to accom~nodateall the gods with claims to a particular
position, the old classification systems proving too narrow for this
task.
53 Rare ~nqlancesof double awgnment can be observedcven ~neclrl~erVcd~c texts
(cf. tables 1-11), and a regular double asslgn~nentforthe four m a ~ nreg~onsha\ occurred
already IIIAVXIX 17 1-10.
54 C ~ N I 1974.208
)A "That t h ~ paw
s of gods was bel~evedto encompcl%thc whole
twofold unlvelse, 1s In vlew of the~rcharacter and the lmportancc of the Rta principle
not SLllpllslng " (rbrd ,207). Regarding their p o ~ t ~ othe
n , author (rbrd.: 192) adm~tsthat
he 1s unable to find a reason for the northern posltlon of Ih~sdual de~ty
55 GONDA 1974. 6 0 4 1 analyzcs t h ~ spasage and ~eflcctsupon it: "one 1s undel
[he ImPresslon that the author, wlllle coupllng together powers wh~chale more or lew
tradltlonally connected w~ththe seven dlrect~ons,consc~ouslydeviated from the Inore
usual ploccdure whlch 1s characterized by the allocat~onof one single deity to every
d~~ection [ ..] there can be no doubt that In h15 eyes these double invocat~on\suppl~ed
a need, that they made the prayer more effective."
Solne of thew couples are sull palrs of opposites (ahrrr - day, rcitrr - n~ght)
Or tend (0 be complementary (Sfirya, Candra) However, Agnl and Adltya, often
compared for thelr bnlllance, Soma and Rudm, wlio can both clal~nto rule the north,
and m0St obviouqly, Yama and Mrtyu, both connected with death, clearly conclst of
~elateddeltles.

10

/" \
 1.1.2.6 The bali offering
In some passages of the ~rauta-,the Gyhya- and the Dharmasiltras
(marked with [B] on table VI) bali offerings to the guardians of the
directions are prescribed. This popular rite was executed in the pri-
vate sphere of the household and basically served the worship of
deities in connection with particular loci in the house. An invoca-
tion of the directions was not always included,s' but if it was, the
"hosts of Rudra"s8 were another common recipient of oblations "in
all the directions". Interestingly, individual hosts (aizugii!z) have
also been assigned to the traditionally known guardians of the direc-
tions (cf. Marz~lIII.87), the almost invariable order being Indra -
east, Yama - south, Varuna - west, Soma - north, Brahman - zenith.
It is a somewhat misleading fact that the recipients of bali offerings
are generally regarded as "divine powers of a lower order".sV~nfact,
particular deities or groups of deities (most prominently the vis've
cleviilz) are worshipped with non-homa oblations, before offerings to
other, lower classes of beings (bhfita) are made. Different traditions
must have existed side by side, a fact which becomes obvious when
the worship of ten guardian deities at a house construction (GGS
IV.7.41) is compared with the expressly "silent" preparations of the
ground in other texls.60
With regard to the development of the directional guardians,
two important innovations can be observed in this popular rite, viz.
the expansion of the number of guardians to ten, which occurs in a
few cases, and the rather regular appointment of Indra as the
guardian of the east.61
1.1.2.7 The Inclusion of the Intermediate Directions
and the Incipient lokapiila Concept
The term lokapiln had hitherto been used as an epithet for single
deities (e.g. $B XIV.7.2.24) and seems to have become a colnlnon
term for the group of guardians of the regions or directions only
subsequently. A rare instance of the latter meaning in this literary
stratum is ASvGpar 11.7. This passage describes the lokapiila-
iivii/mi~nnz,an invitation to the lokalxilns, without mentioning any
directions. However, the sequence Indra, Agni, Yama, Niryti,
Varuna, Kubera, TSiina implies the order know~lhereafter, with
the exception of VByu being mentioned separately from the other
guardians, together with Ksetrapala and Aka5a. This passage also
provides one of the earliest known iconographic descriptions of the
group. There are no guardians for the zenith and the nadir,62 but a
substantial increase in the number of single guardians has taken
place.
An inclusion of the intermediate directions can be observed in
some of the prescriptions of balicliii~a(KIiGS 111.2.10; IV.2.21-22).
Reference to particular connected deities" or an offering to such
deities is still quite rare, however. One of these cases is GGSIV.7.41
(table VI), where the eight guardians of the quarters, VBsuki on the
nadir, and Brahman on the zenith are all offered balis on the occa-
sion of the construction of a house.64 It speaks in favour of the
chronological priority of the group described here that the assign-
ment of the intermediate directions differs from the later known
order." Ten guardians are named once more in MSSXI.7.1.7. This
passage, taken from the Miinava-Grlzya-PnriSi;~l1a,forms a later
addition to this text and seems to be one of the earliest references to
the canonical group of eight directional guardians. The only devia-
tions from the later system are the fact that Nirrti is regarded as
female, S o ~ n aand Kubera are sharing the north66 and Rudra gov-
erning the north-east.
It Inay not be accidental that A.N. PANDE'S statements on the
general nature of the deities in the Grhyasiltras, "the functional divi-
sion of gods in the Grhyastitras is of a thoroughly spiritualised and
specialised nature" and "the general characteristics of gods [...l is
that they pervade the different quarters, wearing armour, wielding
strong bows and arrows, not winking the eyes and vigilant about the
affairs of men" (PANDE 1958-59: 92; 131) seem to be most appro-
priate for the lolcal~iilas.Their roots belong to this period.
57 Eg. &DhS 11.2.3-4. A few directional references arc made here (south: pit!s;
north: Rudm), but tlicre is no special worship of the dircctions. PGS 11.9.3-10 may be
regarded as a trnnsitio~ialform witli a quite elaborate, but not systematic treatment of
tlic directions. Interestingly, Vlyu is tlie only god referred to in connectio~iwith the
outer regions; Bmhnian, arttariksn and Silrya come in the centre, all gods and all beings
on the north, Usas and the Lord of beings (Rudra?) further north, the pir!s on the soutli,
and Yaksma on the north-west. Incidentally, the Iattcr, under the name of Pspayaksman,
is also posilioncd near the north-west in the vr?.strt~~m!~~lnlo grid (ByS LIII.45).
58 E.g. ASvGS IV.8.22; PGS III.8.I 1. Tlie close relationship of Rudra with all the
directions is corroborated by his epithet dis'dlp pati in the Satari~dri)~ci (e.g. TS IV.5.2
a). It has been stated in ASvGS IV.8.24 that tlic worship of the dircctio~is(diidr!~
ril>asrlrcinarr~) has to be performed at all sacrifices to Rudra.
59 GONDA1977: 563. MDhs'II.70 and TA 11.10.4 list the bali offerings among thc
five types of sacrifices and name the bhritns as their recipients.
611 ApesVII.17, cf. HGS 1.8.27. Significantly, A/IGS V11.17.1 prescribes that the
ground should slope to the south-west (!), while GGSIV.7.3, like many other texts, calls
for a plot of land inclining towards the north or the east. The bali offerings in ten direc-
tions without any reference to deities could be regarded as a possible intermediary stage
(KlrGS IV.2.21-22).
" Exceptions appear in the Vcirciha-Gfipasritru (table VI) and in Kalifs VI11.3;
L.3-5 and other passages, where AV IIL26 and 27 (table I) are prescribed For bali
worship. All these passages assign tlie east to Agni.
a As in the earlier types of classification of four, five, six, seven, err., directions,
the authors wcre able to choose whichever system suited the context. Regarding tlie
lokapdlos, eight was tlie favourite number. It is highly significant that one of the earliest
references to lokapcilas as a group of regional guardians in AitU 1.1.4 lists eight mem-
bers (Agni, VByu, Aditya, diici!~, o~adhi-i~nrrn.sl~atci~~n!t, Candramas, Mytyu, cil,crs),
which are different from the group under discussion. The first four of them, however,
correspond to tlie guardians of the four main directions in $B VIII.1.1.4- 2.7 (table 111).
They are also included in a list of eight cosmic deitieslelements given in MIP 161.14
(Arka [Aditya], Soma, VByu, HutlSana [Agni], salila, antarikp, r~nkptni!li, cliio drrSo),
and the lint three (Agni, Vlyu, Aditya; supplemented by Pythivi, Anta~iksa,Dyaus, VBc,
PrBna) recur in JUB 1.1.l, where the eight mentioned - principles
. are stated to contain every-
thing in this world: metdns!oir clevoti. ercivad idot?ts n r t ~ a(JUB
r L 1.7. l, cited by K L A ; ~
1986: 28). It is again striking to find many ~ne~iibers of this group (particularly of the two
latter versions) shared by tlic eight vasus, as they are listed in SB X1.6.3.6 and B ~ U
111.9.3: Agni, Pythivi, Vlyu, Antariksa, Adilya, Dyaus, Candramas, NaksatrB~!i. Tliesc
deities have been regxded by KIRPEL (1920: 5) as corresponding to tlie eight lokos.
The casc of BhGS 111.13-14 is rather strange. In referring to tlie boli offerings
to the direclions, the interiiiediate ~.egionshave been labellcd will1 Ihe naliies of thc lat-
er known guardians (respectively, Agni, Niryti, VByu, iSBn:l), while in enumerating tlic
directional guardians during twoprnduksi!?d rounds reference is made only to those of
tlie mail1 directions and to the north-cast (hcre: irttclralrrirvcl). Therefore, this passagc
seems to be corrupt. KausGS 111.10.6-8 also sceliis to be in disorder, as the north- east
is counted twice (first reference: apardjitci diici, cf. AirB 1.14). The emendation must
have been ~iiadeat an early date, as Nir!.ti is still feminine, and Rudra takes thc placc of
1Slna.
M This passage has been quoted by BANER~EA 1956: 521.
6s Significantly, Viyu (A~iila)on tlie south-east and the pitys on the south-west
arc found in the same position in the vci.strtrr~a!r~rrlagrid (ByS LIII.43-44). Maliirlja on
the north-west and Maliendm on tlie north-east is somewhat n~ysterious.The fomier of
these names Inay refer to Kubera (TA 1.31.6: K~rberci)~a Vais'rol~o!rd)1ci Mohcircijdja
rtnnta!~). Tlie grammarian Piinini, of about thc 5th celltury KC., according to V.S.
ACRAWALA (1963: 361) mentions a deity Maliirlja without implying its identification
witli Kubera.
" It is interesting to find "king" So11iaand "king" VaiSravana juxtaposed in HGS
11.8.19.1.
 1.1.2.8 Directional Assignment in the Later Vedic Period
The colours67 and the trees68, two other new objects of classifica-
tion, will not be discussed here, as in their cases the assignment
varies considerably.
TO summarize briefly, it has become apparent that the period
under survey witnessed a considerable development in the concep-
tions of the directions, their guardians and particular associated
components of the material world. It has also become obvious that
the rigid system of ten directional guardians lies only one step ahead
from here.
The diversity of the systems makes it quite difficult to sum up
all the new traits of classification. The most significant development
is the replacement of the old concept, still prevalent in the
Briihmanas, in which the main cosmic divisions are assigned to the
five directions, starting with the earth on the east (continued: atmos-
phere, heaven, fourth world, and totality), by a new system which
starts with heaven (Aditya, Indra69) on the east. This tendency is not
entirely new however, as the Bdityas have been assigned to the east
before, in a few Atknrvaveda passages (cf. table I). This distribu-
tion has now been adopted by some of the Aranyakas (table IV),
some of the Upanisads (table V), and most of the Gyhya- and
~rautasitras(table VI). In the case of Aditya (or Aditi) ruling the
east, Agni has sometimes been shifted to the south (AitB 11.1-2; TA
1.23). In the case of Indra ruling the east, however, Agni never occu-
pies the south, which is instead, at least in the Sitras, governed
exclusively by Yama. Indra can be also found in the centre (JB
11.141; vE&$s 61). The variability in the positions of Soma70 and
Varuna is still considerable; this seems to come to an end in the rigid
order prevailing in the sphere of the bnli offering, however. Here
again, there is a surprisingly high variability, this time concerning
the intermediate directions. It has been pointed out before that the
north-east, being of outstanding importance and regarded as invin-
cible (aparc?jita71), was probably the first direction which called for
an individual assignment. The south-west, due to an already exist-
ing tradition, was easily taken over by Nirrti, who was still regard-
ed as female. Rudra, who traditionally belongs to the north but is
also known as diiiirlz pati (master of the directions), justifiably
deserves the most important direction, the north-east. The south-
east, for which an occupation by Viiyu was not unknown,72 seems
to have quickly become the stable seat of Agni, who had been oust-
ed from his place on the east. Stirya/Aditya maintains the eastern
position, which he had gained in some texts other than the
BrBhmanas, only in the v~7stunza!z&ln grid (B$ LIII.43). The
zenith, which was almost invariably ruled by Byhaspati in the earlier
Vedic texts and in the Briihmanas, has now been handed over to
"Brahman", with only a few exceptions. Prajiipati also makes his
appearance here (JUB 111.5.22; $hIV.21.8-12).
It is fascinating to observe how a new eight-plus-two system of
spatial reference has been created, in which some of the "old"
guardians (e.g. Yama, Varuna, Soma) retain their positions, but
which also makes room for some traditional assignments of deities
(Niryti, Kubera73, Rudra) which had hitherto been excluded from the
directional classification. Although the new system is known only
from the later parts of the Ritual Sitras, the intricate layout of the
sacrificial ground and the complex nature of the ritual actions must
have paved the way for an extension of the old coordinate system.
The latter, which was not much more than an abstract succession of
cosmic layers projected into space, was thus partly modified and
supplemented by new members relating to the popularly existing tra-
ditions of directional symbolism. However, the dichotomy of direc-
tion (clik) versus cosmic region (loka) seems to have been retained.74
Now we are in a position to characterize in brief the diverse
sources of origin of the ancient Indian group of eight or ten
guardians of the directions in space: their ancestors are the cidl~ipatis
of the directions in the Yajur- and the Atharvaverln, the increase of
their number and their fixed assignment is owed to the Ritual Sitras,
while a tradition of eight loknlxilas has been referred to by some of
the Upanisads. This complex genesis of the group might have led to
the fact that some irregularities regarding its composition continue
to occur.
1.2 The Directional G ~ ~ a r d i a in
n sthe Epics
(Riirizc?ya!za and Mnhiiblziirata)
Contrasting with the rather speculative and schematic treatment of
the directional guardians in the Vedic texts and later in the
~ i l ~ a ~ i i s t rthe
a s ,epics and some of the Puriinas contain the wealth
of popular belief with all its mytho-cos~nological,figurative and
narrative aspects. The guardians' treatment in the epics is discussed
separately here, because a basic difference in the conception of the
directional guardians prevails: their number is generally four and
never eight.75 The conception of four outer directions corresponds
to the earlier Vedic texts. Similarly, the positioning of some of the
guardians has not yet been completely fixed, so that a certain degree
of variation occurs. Regarding this variation, it will suffice to refer
to the table of "epic LokapHlas" published by E.W. HOPKINS(1915:
152).76
" In Cl1UI11.l-I0 the rays of tlie sun in the four main directions are described as
having particular colours (cf. DAMAIS 1969: 77), which thc respective deities enter. It
is also interesting to find the mention of ants of scven different colours related to tlie
scven directions known from AV IV.40.1-8 in Karr.fS CXVI.3: Svern - cast; kr!a -
south; rajatc~- west; ,ulrirc~- north; boblirrr - centrc; hrrrirrr - v)ra~lIil~tr ("trackless");
~)?1!1(1 - zcnith.
" It is stated that the nhjanha, plokscr, nyngrodlin, ancl 11(1~rriiDn,n
trccs should bc
avoided on thc east, south, west and north respectively of a building sitc, as they are
sacred to Aditya, Yama, Varuna, and Prajripati respcctively, ancl seem to induce (unwcl-
come) manifestations of thesc gods' powcrs (GGS IV.7.22-24 [crit. cd.: 20-221).
A directional significance of particular species of trees call bc fi~rtherinferrcd from
particular ritual prescriptions mentioned by HECSIERMAN (1957: 31; 114). The only
recurring assign~ncntis that of rrdu~iil~arcr - north.
"' In tlie old clasification, Indra is never assigned to heaven, but the ncw tendency
of regarding him as Lord of tlie heavcnly gods, has led to a shift in his position. For
Indra - heaven see Jl3 1.105.
It is conspicuous that, in thc BrBhmanas and Upanisads. Soma/Candra occurs
scveral times on thc south, a position which is also found in thc Alhorvavefln
(XIX.17.1-10). This move might be related to tlic sy~iibolisniexpressed in thc layout
of the sacrificial ground, where the southern fire (tIak;i!rclgrii) assumes the shape of a
half-moon and is somctimcs stated to signify tlic fourth world (CIIU 1V.12.l , cited by
BOIIEWITZ 1982: 50).
7 1 Cf. note 63. Also, e.g. KorrSS XV11.9, 22; XXXV111.17,18,30.
72 GGS 1V.7.41, cf. also his position as "Anila" in tlic ~~cistrr~iia!r<hln grid (&S
LIII.43). In the former text, the close association with Indra may be due to their com-
Inon rulership of crnrrrriksa, which, in a deified form, is his neighbour in tlic latter
instance.
73 Dlianapati (Kubera) should be individually worsliipped on the north according
to Kau~GSIII.10.16a n d $ ~ ~ 1 1 . 1 4 . 1 7 .
74 Cf. notc 62 for the classification of the guardians of tlie lokas without any dircc-
tional refcrencc.
75 Cf. E.W. HOPKINS 1915: 149.
7 V o r references sce ibid.: 149-152, but unfortunately, the author does not refer
to or give a list of tlie rcspcctive passages.
 Table V11 - Directional Guardians mentioned in the Epics Table VIII - Lokn11iilns and Associates mentioned in
Mnhfiblziirntn 2
Direction Groups of lokc1[>c7lnsmentioned in the Epics
(~,,eatoncdor
~mpl~cil)
I Mbh l East I South l West l North I
2.8-11
East I Acni l Kubera l Aeni l Kubera l Indra Lokcrlx7ln l Sakra l Yama I Varu~!a I VaiSravana
I (Indra) I I (Kubera)
Wcst I varuna ( Varuna I Varu~!a I Varuna l Varuna Attribute l vcrjrn I - l pdin -
North I Itidra I Indra I So~na I So~na I Kubera Consort l Saci - I Vliruni I 1000 women
Residents rriaruts, pitp dclitytrs, rrnrcrs,
(selection) sddlzyds cIcrit)~c~s,
four grrlr)~~kcrs,
Most conspicuous are the unrivalled positions of Yama in the south oceans, rivers, kirzrinrtrs,
and V a r u p in the west. Matters are somewhat obscured by the fact directions (rli.<n!l) kirii/~rrrlr;crs
that the lokn~>dlnsare often mentioned as a group, without any ref-
erence to their status and without any directional assignment.
In some epic and PurZnic passages, the guardians of the direc- Table IX - Loknpnlns and Associates mentioned in Mnhiibhiirntcl5
tions are literally brought to life. Detailed descriptions of their

r
abodes belong to the sphere of mythic geography in the epics. Mb115. East South West North
Besides these vivid illustrations, the members of the loknl~iilngroup 54.105-109
also act as clrnt17ntispersonae, though in a restricted way. Loltnl?dln Sakra Yama Varuna Dhanada
The composition of the group is not very different from the (Indra) "gopnti" (Kubera)
Vedic schemes (tables 1-11). The positioning as expressed in Residents ~dclli)da pitrs, nrrir~lcrs, Vriyu, Brahm3,
AVIV.40.1-8 for the outer directions (table I), viz. Agni - east, (selection) vike devas Viq~~u rdkstrsas,
Yama - south, V a r u ~ p- west, Soma - north, is found in the epics, j~trksas,
as are other versions. A remarkable change, however, is the fre- gtrrrtllrnrvns,
quently occurring inclusion of Kubera who is described - as far as riaiytas
can be observed - only in the MSS (XI.7.1.7) and probably as
"MahLrZja" in the GGS (IV.7.41; table VI). He appears as guardian
of the east or the north. Both Kubera's installation in this position lokn-/rlikpiilnsare mentioned. Perhaps due to the variability of par-
(attainment of lolcn~~iilatvn- the status of a lokal~iiln)and his attain- allel traditions and concepts, Inoreover representing different stages
ment of immortality are referred to in the Riimiiyntzn (7.3.17) and of development, their descriptions are to a certain extent incongru-
the Mnhiibhiiratn (3.258.15). Kubera is a rather ambivalent deity of ent. It is essential to the present study to observe the probably exist-
a rather enigmatic origin. In the epics, he is closely connected with ing relationship between this variability and the diversity found in
the riik;nsn demons. In a well-known passage in SBXIII.4.3.10, he early dikl>iilnrepresentations in art.
appears as the lord of thieves and criminals. Subsequently, in the Taking information in particular fro111an exemplary survey of
Purlnas, his ynk~n(superhuman creature) aspect becomes his most the Mntsynp~lrE!ln,which contains numerous relevant passages, but
prominent characteristic. Kubera's inclusion can be probably ex- also from other P u r l ~ p s the
, different contexts of dikpiiln activity
plained by the infiltration of a prevalent popular tradition,77 accord- and relevance call be listed as follows:
ing to w l ~ i che
l ~ was a powerful king of immeasulnble wealth in the
north. In this position he is associatcd with Mount KailZsa and the 1 The directional guardians act as warriors, who fight to uphold
god h a , who resides there. The descriptions of his well-defined, the moral order of the cosmos (e.g. MtP 133.64; 135.71;
but ambivalent characteristics are symptomatic of the clear, vivid 135.77-78; 148.81-85; several passages in chapter 150;
concept of the world guardians and their domains contained in the 153.208-219). Their presence as armed warriors and atten-
epics. dants of ~ i v causes
a offence to Viraka, ~ i v a ' swatchman, in
The case of Kubera shows how a gradual evolution towards the MtP 154.570-572.
later fixed order of directional guardians took place and elements 2 Their character as (minor) gods with limited power is empha-
from diverse cultural layers became involved in this process. sized in passages where they are described in a poor, defeated
It is important to note that the residences of the four loknl~iilas state (MtP 154.18-26; "ndrcis" are nientioned instead of
are not yet conceived as being located on the world's central moun- ISBna). Hiranyakasipu and other mighty demons are able to
tain, but they are described as situatcd in their respective directions usurp the positions of the clikl~iilnsand other deities (MtP
(tables VIII-IX).78 This, with respect to their function quite logical 161.14-15; 177.58). Such a dangerous situation is finally
setting of mythical geography has in tlie Purlnas been superseded resolved by the guardians' return to their respective quarters:
by the concept that the lokn~~iilas' abodes lie on Mount Meru, i.e. 1oknl~iile;usarve;Li d i k p sar?lyiir~nvarti;u(MtP 176.41b).
at the centre of the world.

l .3 Evidence from the Pl~rByas
hi^ large body of religious didactic texts contains, in our opinion,
the most valuable clues to the genesis of the guardians
as a popular SroL1P of cult deities. There is a surprisi1lgly large
1 degree of variation in the aspects and contexts in which the
77 This liypothesis is supported by tlie Buddhist tradition, in wllich thc protcctor
or the north (Kubera-VaiSravana) is rhc only one of tlic directio~lalguardians who
co~iformstothe Brih11ia"caI version (cf. section 2.1.1).
7"s an example, Varuna is repeatedly refcrred to as residing on llie "sunset
i ~ ~ ~ M,,,~ 3 . 1 6 0 . 1 ~I).
mountainn( ~ f 4.41.38-39; 1 Tllis is pal.licularly interesting as
elscwhcrc, tliis sovcrcign of tlie waters inhabits a sublerrancan palace (cf. KUIPEII
1975: 115).
 3 They are objects of worship, e.g. in the ritual of bathing at the
time of a solar or lunar eclipse (MtP 67.9-15; Nairrta is miss-
ing). During the consecration of divine images the lokal~dlns
are worshipped and their banners hung in a iiza&apa in the
appropriate directions (MtP 264.21). Among several impor-
tant ritual contexts are the sixteen nznkdrliinas, "Great Offer-
ing" ceremonies, in which the loknpiilns are regularly invited
(loke~iivdlzananz,e.g. MtP 275.3), and the rituals in which
particular substances are offered. In the latter ritual context it
is prescribed that images of the eight loknpiilns, made from
precious metals, should be put in the appropriate places of a
miniature model of the cosmos (e.g. MtP 83.18; 84.4). The
description of the modes of worshipping the loknpdlas (lokn-
pdldblzip~ijarza)is expressly mentioned in the summary of the
contents of the Mntsynp~~rii!zn (291.3b).
4 In the chapters on cosmography, the lokapiilas appear to-
gether with their particular cities on the mountains surround-
ing Mount Meru (MtP 124.19-24). In this case, the number
of directional guardians is only four (Indra, Yama, Varuna,
Soma). The Matsyap~lrii!zndoes not mention the cities of the
eight loknpdlns situated on top of Mount Meru, which most
Puriinas refer
5 Some of the dikpiilns make presents to the god ~ i v on a the
occasion of his wedding (MtP 153.208-219; 154.487-490).
Interestingly, the group of Manu LokapBlas (Indra, Agni,
Yama, Varuna, Viiy11,Kubera, Siirya, Candrago; cf. following
section, p. 41)81 has been referred to in this connection.
G A group consisting mostly of dikpElns (i.e. the Manu
Lokapiilas with Prthivi substituting Kubera) with the particu-
lar functions of its members forms a model for the behaviour
of an ideal king in MtP 226.2-1 1. Taking into account other
versions of this group from other texts, the only passage in
which the complete group of rlikpiilas is mentioned in the
same context82 is contained in Lirig~zp~~rLi!zn 1.35.5-6. It is a
remarkable fact that Soma and Kubera appear here side by
side, thereby raising the number of divine regents to nine.
7 The guardians of the zenith and the nadir, BrahmB and Anan-
ta/Sesa, are included in the list of ten "dikpdns" in MtP 93.52
(cf. 266.27-28; 274.50-5 l). Ten is the usual number of direc-
tions referred to in the Puriinas (e.g. MtP 161.14). It is some-
what mysterious that an entirely different group of four direc-
tional guardians, also variously referred to as "lokE~~iilns",
"dikpiilns" or "diSiinzpiilns", occurs in the Mcrtsya and in a
number of other Puriinas (cf. KIRFEL1927: 229). The names
of these guardians differ slightly from text to text. In the Mat-
synpurii!~n (124.94-96), they are called Sudhiimii Vairiija,
Kardama PrajBpati, Hiranyaroma Parjanya and Ketumiina
Riijasa. In a list of pious kings in MtP 143.38-39, some of
their names are mentioned. Their installation as overlords of
particular directions is described in most Puriinas (e.g. MtP
8.9-11).83 The abodes of this second lokapdln group are
located in the centre of the Lokiiloka mountain.84 The
Lokiiloka mountain range is generally regarded as forming an
outer ring around the inhabited world, and "in the centre" in
this case could mean on top of the mountain ridge. Thus, these
guardians may be the protectors of the very periphery of the
world according to the ancient Indian tradition. No particular
relationship can be traced between this group of directional
guardians and the usual divine regents.
It has become obvious from the close study of the Mntsynpurii!~n
that the latter group of directional guardians, which was still varied
in its composition (particularly referring to the interchange of
Kubera and Soma), was either already very popular, or it was
intended by the authors of the work to increase its popularity anlong
the worshippers. The variations occurring with respect to the attrib-
utes and the vehicles (table X)86 can bc interpreted as being indica-
tive of a primary stage in the formation of the finally canonic group,
still bearing testimony to the parallel existing traditions which must
have contributed to its genesis.
Among the Puriinic descriptions of rituals, the "offering of the
loknpiilns" in Lirign~~itrE!~n 11.43, is particularly worthy of note. It
belongs to a different version of the sixteen rlznlzddErla offerings
than those enumerated in the Mntsyp~tra!in (cf. Lip 2.28-44; MtP
274-298). In this case however, the regents are not symbolized by
images but are represented by eight priests. The statement that the
ritual was performed by those desirous of becoming an emperor
indicates that it must have been more apt for the already more
wealthy and powerful strata of the society.
Among the rare irregularities in the composition and the con-
figuration of the a.~;ndikpiilngroup is the occurrence of VirUpiiksa
as regent of the south-west in some texts (KIRFEL 1920: 95; cf. VcNzP
on table X) as well as an exceptional enumeration of deities in
Viirnnrzal~urii!zn 32.22-23. Starting from the east, the deities are
here named as (clockwise, with intermediate directions): Vajri
(Indra), Vahni (Agni), DandanByaka, Kubera, PBSi (Varuna), Viiyu,
LaksmiimSu (Soma), ~ i v (iiana).
a Kubera's position in the south-
west, the direction of the riik~nsndemons, is unusual. Dandaniiyaka
may be Yama, as that particular name refers to Yama's weapon, the
rod (rla!zcln). It remains possible, however, that this name might
describe a separate deity with a horizontal staff, here referred to as
Dandapii~!i (see section 2.5.10), who appears in a number of images.
Another irregularity occurs in the mention of both Kubera and Soma
for the northern position in AgP 56 (table X), a discrepancy which
is symptomatic of a lingering uncertainty regarding the protector of
the north, something which is also expressed in the sculptural art of
the time (Biidiimi 3, see section 2.2.1).
Various attempts have been made to assign dates to the
Purii?a~.~7 However, hardly any chronological indications are
contained in them. As a general time bracket for the well-known
"C
'C KIRFEI.1920: 95. It is quite rensonctblc to assutnc that the schemc of four
lokcr~~cilas is thc older version. I11 some Purit>as, both versions cxist sidc by sitlc
(Blr~~rvcrtcr~~rrrri~to5.16.29 and 5.21.7). The o,s!crdik/~iltrschc~ticmust have developetl
from the oldcr modcl, as the north is, in the older version, generally presided over by
Solna (Kubera exccptio~iallyoccurs in the Vurcih0/1rrrc7!1cr, see K I R ~ Eibid.). L:
Agni, SUrya and Candra are tiien~ionedas being inherent in the eyes of Sivit
(MtP 154.441).
" One Inay speculate that both Nair!.ta arid iS.ina, who are missing, are regarded
as identical with Siva himself. In thc casc of ISlna this is clcar, as he is alternatively
callcd Siva or Sahkara. For Nir!.ti/Nair[t;i it may bc relcvant to note that he is listed as
the first of the elcven rrrdrrrs (a group of dcitics closely conncctcd with Siva) in MrP
171.38.
H2 The supposed analogy bctwcen particular deities atid the king is an ofien re-
curring subject in thc early tcxts undcr survey.
H V tappears quitc logical that in this recurring passage which describes the coro-
nation of Prthu and olher divine kings, thc dircctions arc not distributed among the usual
lok(r/~ci/os,but am Iierc entrusted with their pri~iiarydomain of regency (Indra with the
rrrcrrrr/s; Yatna with the pitrs; Varuna with the waters; SomdCandra with mcdicinal
herbs, ere.).
MfP 124.94a: lokcr/)B/ii!~srhiriistcrrrcr lokrilokcrsya rirocl/~ycrtcr!r.
X5 (Sce tablc next page).
XVooie further, particularly interesting instances of variation in thc vehicles
found in thc Mnl.s)~cr/~rtrci!rcrare the occurrences of ~>yrilcr (snake) for Kubera ( M P
133.64) and horse for Varu~a(MtP 174.13).
n7 The dates for all the known Purinic scriptures proposcd by earlier authors have
been cornpiled by L. Rocllen 1986: 133-254.
Table X - Iconographic specifications of clikpiilns from various Puriinas

I'urlnas East85 South-East South South-West West North-West North North-East

AgP51.14-15 Indra Agni Yania Nairrta Varu~p Viyu Kubera I<ina
gaja (elepl~aet) c11Lga(be-goat) malrifa - ~nakara(n~ythical nrga (deer) mefa (ram) v r p (bull)
(buffalo) aquatic animal)
vajra (bolt) h k t i (spear) da!~<ln(rod) kho<lgn (swo~d) p6Sc1 (noose) dltv~rja(standard) gaclri (mace) -
AgP 56.17-3 1 Ilitlra Agni Yama Nairyta Varuna Viyu So~na IGna
also: Kubera!
gaja (elephant) c11Bga(he-gnat) alahiya (buffalo) vHhana ~nekara viihana vHhana V ~ ~ (bull)
L I
-unspecified - (n~ytbicnl -unspecified - -unspecified -
[96.28: kl~ara aquatic animal)
(donkey)l
vrrlrci (bolt) .icrkti (spear) dcr!i& (rod) khndgc~(swold) p6Ea (noosc) ill~vr~jci
(stsriddrd) gad6 (mace) Scilo
(spearllrldcnt)

MrP 148.81 ff. Devarija (Indra) HutiSana Ya~na Rdkgaseh JaleSa (Vart~na) Pavana (Viyu) Dhanddhyaksa -
(bdule scenc) (Agni (Nairyta) (Kubera)
rathe (chariot)/ clliga fie-go;~l) olahip (buffalo) narayuktaratlla bhujagendr;~ - ratha (chariot)
gaja (elephant) ("n;~ls-chariot") (king of snakes)
- .fakti (spear) clrr!r~ln(rod) klia~lga(sword) p i i a (noose) orikufrl (goad) gaclri (mace)
nrirrlgarc~(hammer) god6 (mace) Sakti (spear)
MtP 260 65- Surarija (Indrd) Vahni (Agni) Yama Natr! ta Varuna VByu Kubera fSi ( h a )
26 1.23
n?attnv.?ra!?a :I$ (ramWe-goat) nlahi~a(buffalo) nara (aathropoid)/ jl?ap (fish) n ? g a (deer) nnrayukta~inlHna v g a (boll)
(infuriated narayukta~~imBna ("nar;~-chariot")
elephant) ("nara-chariot")
nksnscitr nkn
vnjrn (bolt) (rosary) r. c / a ~ ~ (rod)
{/a khn~lgn(swo~d) p i i n (noose) patdka (flag) gad6 (mace) tr ifcilrr (tr~dent)
icll~nln knnra!~rIrrlu p6io (noose) r//rvflja (standard)
(bluc lotus) (w,~te~pot) 1.
VCIIIP2.104. ~ a k r (lndra)
a Valini (Agni) - (Yama) Viriipiksa Varuna ~ i y u Kubera ISina
42b-50a
gaja (elephant) Sukay'Tna nrallifa (bufk~lo) oypa (camel) harpsa (goose) - vyoo~ayina vb'p (b1111)
(parrot-chariot) (sky-vehicle)
vnjr U (bolt) - cln!rcla (~od) khndgn (sword) p r i h (noose) ( I I I I I I I -~ ~ garl6 (mace)
c~rcleof a ~ r ) -

VdIrP 3.50 fr. ~ a k r (Indra)

a Valln~(Agni) Yania VirilpHkss Varu~!a Viyu Dhanada Gauri(vara
(Kubera)
gaja (elephant) SrlkayH~ln rnahi~a(bulfalo) u~!m (camel) saptahaqlsa - nara (anthropoid)/ -
(parrot-cbariot) ratha SibikH (p;dooqoin)
(goose chariot)
l i a r.
l ~ r r ~ l ~(lotus) jvrild (Ilamc) r. kha(lga (swo~d)r. cln!r~lcr(~od) I I O U )r vci)~vilil,ciritovnstr.ngcrcli (mace) r, nk~nri~dlri
crrikuEcr (goad) r. tr~iirlo(tr~dent)r. cla!r{ln (rod) r. p 6 b (noose) r. (111flaledcloth) inkti (spca~)r. (rosa~y)r.
vujrn (bolt) I. nkl;nrrdIi carr~ia(sli~eld)1. hriklru (conch) 1. triiciltr
( ~ o s a ~1.y ) I atnol)dtm (tr~dcnt)r.
(gem-vcsscl) 1. ~lc~rl~a!~o
(millor) I.
rndi~~arcr
(blue lolu\) 1.

major PurAnas, the first ~llilleniumA.D. is quite probable, but it is
possible that subsequent changes have crept in. R.C. HAZRA (1940:
176-177) has tentatively assigned dates to most chapters of the
Matsyapurri!za. One interesting result of the attempts to date Puriina
passages, which has some bearing on the developlnent of thc
astnclikpfiln group, has been secured by HENSCEN(1958). The
author, having noticed reference to some intricate details in the
cosrnic layout in Kalidiisa's works, postulates that the coslllological
passages of the Brnhmri!~~ln,Viiyu, and Mntsynpurii!in predate the
poet Kiilidiisa (c. early 5th cent. A.D.).Incidentally, the passages
referred to by H E N S G E Nform
~ ~ a kind of textual bracket for the
description of the abodes of the four loltnl>iilnsIndra, Yama, Varu~!a
and Kubera on the Miinasottara mountain in the Mntsynpirrfi!lcl
(see above, no. 4). Thus, the concept of only four loknpiilns, which
is already known as the older one, also seems to belong to an early
Puriinic stratum, prior to the 5th century. Significantly, one century
later, the earliest dated literary evidence for the as!mfikl)filas occurs
in Variihamihira's Yognyfitrz, as will be shown in the following
section.
85 Tlie d~rection\are not ment~onedIn thc texts, but they are iniplied by the
scquence of the deities.
Km See HLNSGCN 1958: 162-163, where lie refers to KIRTCL1954: 135, (1. 74 TT.;
230, il.29 ff.
 From the passages utilized in table X, those from the VdlzP
probably also belong to a particularly early stratum. The dates pro-
posed by some authors vary between "A.D.400-500", "A.D. 450-
650", and "A.D.600-1000" (ROCHER1986: 252). Some interesting
hints of a chronoIogical priority lie also in the mention of an ancient
(Kusiina period) form of Viiyu89 and in the reference to unusual
deities (VirnpBksa instead of Naiqta, GaudSvara instead of h a )
and unique vehicles of some of the guardians (Agni, VirUpiiksa,
Varuna). A passage in the Skanclnpirrii!~a(2.9.68-85, see A.B.L.
AWASTHI1976: 209-213) draws much material from the V&P. Its
posteriority can be deduced from the fact that the configuration of
each deity, portrayed so elaborately in the V&P, has been abridged
in the Skanr1~11>urii!za,so that, e.g. the nooses, formerly held by
attendants of Yama and VirUpBksa, respectively, arc, in the Skari-
cl~il~~rrh!za,assigned to these deities tl1emselves.90 In the case of the
VrlliP, it is somewhat unexpected to find that very elaborate descrip-
tions of the directional guardians' images91 are contained in such an
early text. Keeping in mind the aspect of "propagation" contained
in the Mntsynl>~rrE!zn,however, this need not be surprising. In any
case, it will be seen below that many of the features mentioned here
did not succeed in gaining acceptance in art.
Finally, it is worth mentioning a chronological assessment of
the late 6th century for chapter 51 of the Agrzipi~rh!za (table X),
which was recently provided by an iconographical study of another
group of deities, viz. the rzavngrnlins, which are described in the
same chapter92. Except for the omission of Nairyta's vehicle and
Iiiina's attribute in this passage, the characteristics of the clikl~ilas
are standard and already conform to the canonical version.
1.4 Other Early Texts
The designation "early" is here applied to texts which, as with the
selected Puriinic passages, can be rather safely dated before 1000
~ . ~ . 9Only
3 a few of these contain iconographic descriptions
(table XI).
The texts surveyed belong to various categories, and thus in-
formation about the directional guardians can be expected to be
diverse. The political treaty Kautilya Arthaihstra, which probably
dates from the pre-Christian era, mentions four deities as the
guardians of the city gates (Artlz 2.4.17): Brahmii, Indra, Yama and
Seniipati (Skanda). The directions are not specified, however. The
well-known NiinBghiit inscription from c. the late first century B.C.
includes the "four lolca~~iilns" Yama, Varuna, Kubera and Viisava
(Indra). It has been noted by D.C. SIRCAR (1971: 239) that Indra and
Dharma, the latter akin to Yama, have been invoked separately,
indicating that "the identification of Dharma and Yama and of Indra
and Viisava may belong to a later date". The sequence of deities
seems to start with Yama in the south and proceeds in the usual
clockwise order.
In this connection, however, it is relevant to note the occurrence
of the relatively later group of a~facfikl>Elas in the particular texts,
as well as to trace iconographic passages. The most important
reference to a group of eight is contained in the Mnrzusnzrti, which
has been assigned to the period of B.C. 200 to A.D. 200. While in this
text, the number of classified directions amounts to eight (e.g. 1.13),
a group of eight deities which is not explicitly linked to them has
been variously called "lokrrl?hlns" (Mar~~i 5.96), "1okeSas" (basical-
ly synonymous with the former; Mnrzu 5.97), and "sirrerzdrns"
(kings of the gods; Mnrzu 7.4). The composition of this group does
not exactly tally with the later established a,g0clikl7hla set (Mnrl~~
5.96):
Soma, Agni, Arka (Snrya), Anila (Viiyu), Indra, Vitta
(Kubera), Appati (Varuna) and Yama.
In this group, which is referred to as "Manu LokapBlas" (see section
2.4.1.3) in the present study, Nair~taand iiiina are replaced by Si~rya
and Soma, the Sun and Moon Gods.95 The group is referred to again
in Marzrr 7.4 and 7.7, but in different orders. In Marz~r9.303-31 1,
Kubera is substituted by the Earth Goddess Plthivi. In other texts,
the analogy between the king and certain deities, which is the under-
lying theme here, recurs,Y6 but the group of Manu Lokapiilas as men-
tioned in Mnrz~r5.96 and 7.4 has not, according to our knowledge,
been referred to anywhere else. In this context, there is only one
interesting reference to a group or eight or nine "loknpdlas", corre-
sponding to the directional guardians.97
An early version of the clikl?iila scheme, contained in the
renowned treaty on dance, the NiityaSEstrn, preserves the principle
of allocating a number of deities and groups of beings to the eight
directions, with Brah~uaat the centre.98 This order may be general-
ly related to the versions contained in the GyhyasUtras (table VI).
Regarding iconographic passages, tlie two works of the author
Variihamihira (until 587 A.D.),the Brlzatsnr!zlzitii and the Yogaydtrd,
contain the earliest known rules for making images of the dikphlns
(see table XI). While the YognydtrE significantly describes the
dilc/~iilnsin their respective regions, thcir images being carried on
military standards (cf. A.M. SHASTRI 1964), it includes only few
details about their appearance. The Brlzatsnr?llzitiialso refers to the
X V h i stlcscription clearly conforms with the "running type" representation of
VByu with dislicvellcd hair and an open mouth, holding an inflated cloth, wliicli is
known from early rcprcsentations (see section 2.5.6). The omission oftlie mention of a
vehicle also does not seem to be accidental.
""n thc casc of Yama, the attendant KBla holds the noosc in V(//?P3.5 1.7; in the
case of VirkpBk~a,his consort Nirlti carries tlie noose, according to VdIrP 3.57.3. For
the SkP, see 2.9.27.72-73.
" Nowhere else has such a Sully detailcd and evcn interprelativc account of [lie
directional guardians been madc. Tlic mostly four-armed tlikpcilas arc minutely
described, together witli their vehicles atid consorts, and in some cases, witli their atten-
dants. Thc sccond part of the account explains thc symbolic mc:~tiingof tlie attributes.
the vehicles, etc. Tlic regent himself is always idcntificd witli an aspcct of Vist!u, the
consort with his wife Lakstni.
S . MAIZKEL 1995: 165. The author observcs "thc SLIC~that the vcrscs ( 5 1: 1 ,l l-
12) ascribe to Candra, B:liaspnti ant1 ~ u k r the ~ i 'carly' iconographic features of the
rosary and tlie watcr vessel, yct accord the 'late' attributes oftlie lunar crcscent to RBhu
and a sword and, uniquely, n lamp to Ketu." This description of Kctu's latiip has bcen
assigned to "L1 porlie orrciotrre" by M.T. DE MALLMANN (1963: 4). dating to c. 6th ccn-
tury (ibid.: 10), while MARKEI. prefcrs a late-6th ccntury date.
" This time bracket roughly corresponds with tlie carly phase of representation of
the dircctional guardians.
" The directions arc expressly mentioned only in tlic Yogfi)~)'cTlrci.
'"oolne authors (e.g. 1. D o w s o ~1879: 180) havc tricd to Sorcc tlicsc dcities. thc
scqi~enccof wliicli changes from pussagc to passage, into a dircctional schcme. In doing
so they have :~ssignedSilrya to tlic south-wcst ancl Soma to tlic north-cost. No corrob-
oration whatsoever can be found in other tcxts or in tlic artwork of tlic pcriotl.
"On the tlivitic fonctions and tlieir royal analogies sce also A.S. ALTCKAIZ 1949:
61-63.
" LLirigcr/~rird!rrr 1.35.5-6; in this passage, a king cxplains tliat lic himself holds the
pllysical body of the eight, but in ;~ctitalfact probably ninc /okcr/~ci/as(if iSvarn is undcr-
stood as ik~na):Indra, Agni, Yama, Nairlta, Varuno, Viyu, Soma, Kubcra, iSBna (?) -
(l.~!~rifir!t/oku/~G/)'cTrt~71!1 ~~(r/>~~rc//~circry(rte
rrr/~fr!r / t~r.srr~cidirt(/ro
/~yfr/rcrr~i
~~~i/r~riryorrter.<c(i
rrilrl!ti.smrhci ~~~rr~i!rcricoi~~o vci)~rr.<crrsor~tod/rcrrru~/rewr cccr / Uvoro 'lmr!t ria .sortdclro
Ill7 'l~flllrcrlrt~lll)~~i
Cl,fI cc1 //,
" 111Nci! 3.23-31, corrcspondcnce witli the usual ortler lics in the positioning of
Mahcndra on tlic east, Valini (Agni) on tlic south-east. Yama on 11ic south. V a r u ~ pon
thc west, ancl Dlianada 011 the north. Unusually, [lie south-wcst has only a group OS
II(I~~!.~NS,and tllc north-west the scvcn winds (s(r/~to~~ci)~crvcr!t), wliilc ~ i v ais installed
11ot on tlie north-east, but on tlie east.
Table XI - Iconograpl~icspecifications of diklxilas from various early texts

Early texts East94 South-East south I ~outli-west North-East

Brllcit- Malicndra (Indra) Y ama -
sol!~hiId
58.42; 57 dvilm (elephant) nmhiya (buffalo) havisa (goo~e) nara (anthropoid)
11crjrn (bolt)
Indra Valini (Agni) Vaivasvata Raksad~n Varuqa Viyu
(Yama) Pari (Nairyla) (Kubera)
Airlvata avi (sheep) makarn (n~ythical 111rga (deer) 11ara (anthropoid) v ~ y a(bull)
(personal elephant) aquatic animal)
vnjra (bolt) nk~asitlrc~ pdSn (nooae) dlrl~aja gods (mace) Sila (spearltrident) r.
(rosary) r. (standard) ialiklra (co~icll) v~r~clarllrrdra
krr!i~lika(pot) I. (bestowing gesture) r.

t-
kol~rilr(skull-cup) I.
11rigc11~7.fn
(snake-

Malicndra (Indlx) Agni Varuna Viiyu Dlianada

(Kubera)
Airlva~y -

(personal eluphant) 11n.icr (noose)

vnjra (bolt) - cla!r<lo (rod) grr(1i (mace)

as!adik/~lila group?" but only mentions the iconography of four of
them. It is difficult to decide whether this incomplete infor~nationis
exclusively due to the fact that the works reflect an initial stage
in the description of directional guardians. In the Amarakoia, a
lexicographical work which probably also dates from the 5th cen-
tury,l00 similarly only few of their characteristics are revealed.101
The Hayaiir~nl~aiicarlitra is a religious text which probably
dates from c. 800 A.D. It includes a complete description of the
images of the "lokeias", as the directional guardians are called here.
It does not mention the respective directions, as is generally the case
in the early texts. While there are few exceptions to the descriptions
known from the Puranas (table X), the text is extremely important
in that it depicts 1 6 t h as four-armed, while the other members of
the group are only two-armed. The text also contains a rare version
of Nairyta's vehicle, which is a bear. This unique reference could
explain some peculiar forms of Nairrta's vehicle in Central India
(see section 2.5.4).
In summing up, perhaps also anticipating a comparison or the
actual representations, it is i~nporta~ltto note that the variation in the
iconographic prescriptions of the first lnillennium A.D. seems to cor-
respond more or less to the variation in the artwork of the period. It
would be futile to try to link a particular text to a particular image,
but a few plausible connections and general tendencies can be rec-
ognized, as the number of relevant texts is relatively small. There
remains no doubt that specialized ritual functions must have pro-
moted the popularity of the directional guardians; their features
were not generally very elaborately described. No mention will be
made here of the colours of their skin and clothes, which have been
referred to in a number of texts but are not relevant in the present
context.
" 54.4 and 86.75; it is a significant aberration that, contrary to the evidence
listcd on table XI where the respective guardian is Kubera, rlic north is allotted to Somu
(called Indu) in both passages.
1") C. VOGEL 1979: 309-310.
'0' The group of eight is enumerated in dig~jarga,lines 2b and 3a. A list of the
guardians' names, including also their wives, vehicles, etc. (elaborate only in Ihc case
of Indra), appears in the section s~~c~rgcri,nrga,
lines 44-75a.
 2 The Images
2.1 Predecessors and Earliest Evidence of the Depiction
of Directional Deities
According to the results of archaeological investigation, the earliest
instances of the installation of directional guardians are found in the
Buddhist stllpas. While the guardianship of Buddhist structures was
later entrusted to the Brahmanical group of directional deities,loz the
earliest known guardians represent the Buddhist tradition of the four
great kings (caturnzahiiriijas). Unfortunately, no evidence for the
complete group of guardians has survived.
Practices of the foundation of a temple are possibly of greater
antiquity than could be proven so far; they include the interment of
caskets as foundation deposits and the engraving of symbols on the
foundation stones of religious structures. Here, only those cases
in which the a~fadiklkilashave been clearly referred to will be
selected for discussion.
Possible predecessors of the representations of the rlikpiila
group in the early reliefs depicting assemblies of gods will also be
considered. The well-known panels at Udayagiri and Deogadh offer
early instances of the depiction of particular directional guardians in
a more or less well-definable context.
A particular type of Kusiina panels shows four personages of
identical outward appearance who may be tentatively identified as
directional guardians. In the discussion of groups of identical or
nearly identical-looking deities the question arises, as to how far the
development of individual visual types for the loka~~iilas-to-be pro-
ceeded in the early centuries A.D. Representations of Indra, Agni,
Vayu and Kubera were fairly popular during this period. As there is
clear evidence for the continuation of earlier visual formulas, Indra
and Agni have been chosen as exemplary cases. Early figures of
Vgyu, particularly this deity 'S predecessors from outside India, have
been analysed recently in a comprehensive article by K. TANABE
(1990). The iconography of Kubera, both in early and in later phas-
es, is quite complex and will therefore be analysed in the section
dealing with iconography (2.5.7).
2.1.1 Early Sculptures of the Guardians of the Directions from
Buddhist Contexts
While there is fairly frequent mention in Buddhist literature of the
catiirnznhiiriijika group of directional guardians,103 only very few
examples in sculpture still exist. The members of this group are:
Dhytariistra (Piili: Dhattarattha) - east
Vifidhaka (Piili: Virtiihaka) - south
- west
Viriipiiksa (Pali: Viriipakkha)
VaiSravanaIKubera (Piili: Vassavana) - north
Well-known early depictions are the images of Kubera
(inscribed "Kupiro yakho") and Virtidhaka ("Virudako yakho") on
the corner pillars on the north and south side respectively of the raiI-
ing of the Bhiirhut stllpa from the 2nd century B.C. Only from the
inscriptions is it known that these guardians were regarded as y a k ~ a s
(class of semi-divine beings). Possibly, guarding the directions was
one of the earlier functions of this class of tutelary deities.104 At
Bhkhut, the directional guardians Kubera and Virtidhaka are both
shown frontally in relief on a pillar face; they are holding their palms
together in aiijalinzudrii and wearing turbans. Kubera (BARUA 1937:
pl. LV, 60) is supported on the shoulders of a gnomish figure which
has been identified as a lzara (anthropoid), Kubera's usual vehi-
cle.105 VirQdhaka (ibid.: pl. LV, 58) is standing on rocks.106 L.A.
WAD DELL^^^ has attempted to identify two of the other figures on the
southern corner pillar as the missing lokal~dlas,but this highly spec-
ulative and has not gained acceptance (BARUA1937: 6). Similarly,
the pairs of guardians on the pillars of the gateways of Sttipa no. 1
at Siifici have been regarded as guardian yak~asllokal?dlasby Man-
jushree RAO (1994: 101; pls. 27-29). But, as she also concedes,
"They do not show the characteristic features or attributes given in
the texts" (ibid.), such an identification is highly hypothetical and
therefore not of much value.
Other evidence for the depiction of the early Buddhist direc-
tional guardians comes from Sri Lanka. As the directional symbol-
ism is amply testified by the presence of the so-called directional
animals108 at the cardinal points, which are further emphasized by
so-called vdlzalka~lastructures, some of the relief figures located
there may be identified as lokapiilas. VONSCHROEDER (1990: 58-
59) has summarized the results. Here it is only KuberaIVaiSravana
who can be safely identified in several instances.l09 He seems to
have been one of the most popular gods of early India and Sri Lan-
ka and certainly was not only invoked in his capacity as the guardian
of the north. VONSCMROEDER'S attempts to identify an apparently
thunderbolt-bearing figure, which is found at the northern end of the
eastern viihnlkacla of the Abhayagiri Thtipa at A n ~ r i i d h a p u r aas
,~~~
Dhytariistra or Indra cannot be accepted however, due to some com-
ponents which would not be appropriate in such a case.111 The
figure has so far been regarded as a Cakravartin (emperor), which
seems to be more likely than VON SCHROEDER'S suggestion.
I n 2 For Sri Lanka sec A.J. GAIL1980, U. VON SCIIROEOER 1990: 239-240 (sculp-
tural evidence) and the Mtr,ijuSril~l~h~i/n-Viist~~vidy~s'Hs~r~~, e.g. 11.158-16% (textual
evidence).
In? For references see P.K. ACRAWALA 1989: 74 and notes; sce also G. GRONDOLO
1984: 364-365, R.E. FISIIER 1995, atid LESSING 1942: 3 8 4 4 . Tlle expression lokrrl~cikci!~
has bcen used for this group as well, e.g. in the Dhnn~tnsa~~lgrahn (VII).
IoJ S.S. PARUI (1976: 60-65) mentions fourpksas who, as dvcirnpilas (guardians
of doors or gateways) of Kurukyetra, guard this place from the intcrtnediatc directions.
'OS COOMARASWAMY (1928: 8) writes: "the Naras here in qucstio~iare not men, but
mythological beings variously describcd". RAVEN1988 elaborates on this topic and
gives thc Bhirliut Kubera as an examplc for Kubera as Naravihana (125; pl. 55).
Ii'".K. AGRAWALA (1989: 69-70) wonders why Viri~<lhakadoes not cxliibit any
peculiar traits of the Kumbliinda iconography, the pot-bellied dwarfs over which lic
presides. The same author (1987) has tried to recognize in some of tlie figures dc
pitted on the toro!lcrs of thc SdAci s/Lipns Viri~dhaka(ibid.: fig. 2) and the Kumbhinh
folk.
Io7 WADDELL1912: 137-144. His candidates are Cakaviko Ndgarija for the west
and Garpgito yakho for the east. He argues that the conception of Dl~rtards!ra and
Virfiplksa had not yet developed. For an objcctivc asscsstne~itof tlic labellcd figurcs at
Bhirliut sce LODERS1941: 10-20.
'OR Recently, G.V. VAJRACIIARYA (1999: 69) has pointcd to the supposcd origin of
the animals of the four quartcrs "from the atmospheric appearance of cloud animals".
Also cf. notc 201.
In9 E.g. VoN S c ~ ~ a o e n 1990:
e ~ < pl. 16D. The figure is shown in a shallow rcliefon
a pillar from thc norther11viihnlko<!nof the Dakkhinn ThOpa, Anuridhapulx (2nd- 3rd
centuries A.D.). Tlic deity stands in con~rnppostoon a circular platform with onc hantl
in abha)~amirdrciand the othcr rcsting on his hip. Below him is a small figurc shown
pouring out a shower of coins to the left, corresponding to Kubcra's linc of vision. A
figure holding a huge club apparently resting on tlie floor (ibid.: pl. 8B) could be
Kubera, who also carries this weapon in Mathurd art (fig. 2). However, he is not shown
holding a club of a man's height, as is tlie case with the fri~gmclitfrom tlic Abliayagiri
Thbpa, Anurtidliapura.
' l n VONSCIIROEOER 1990: 58-59; pl. l ID.
The thunderbolt (vnjrcr) seclns to be Iicld in the right hand, which is gracefully
raised above the head. On the propcr right of thc deity the heads of an elephant and a
horse (?) are visible. Two other, either animal or human heads, pccp into the scene from
the left-hand sideof the figure. Floating above the figure's hcad are a squarc and around
objcct, the latter of which occurs elsewhere (similar figurc, but without "~lojrn", VON
SCI(ROEDER 1990: pl. 12D). Thcse objccts havc bcen previously convincingly identified
as the gem and wlicel of a Cakravartin (cited ibid.: 82).
 There is some - though meagre - evidence for early predeces-
sors of the warrior-like lokapdlas which have been popular in Vajray-
Lna Buddhism.112 Those early examples have appeared in the area
of GandhiiraIPakistan. Several isolated images of standing male fig-
ures wearing chain-mail skirts have been found there (M. TADDEI
1962: fig. 2 1;P. PAL1997: cat. 191, 192). A similar figure in bronze
has come to light in present-day Afghanistan (TADDEI1970: Abb.
161). One of the warriors (PAL 1997: cat. 191), wielding a mighty
club with a roundish top and standing on the back of areclining lion,
has been identified as Kubera or VaiSrLvana, the guardian of the
north. The stucco relief, measuring 7 1 . 2 ~ 3 4 . 3 ~ 1cm,
4 belongs to the
4th century A.D.The tiny bronze image (height: 9.5 cm) could have
originally held a spear, which is preserved in another case (TADDEI
1962: fig. 21). The bronze figure has been assigned to the 5th cen-
tury. A set of four lokapdlas each holding a spear has been published
by R.E. FISHER(1995: fig. 9). The figures are arranged facing the
four directions atop the dome of a votive stupn made of bronze,
dated c. 5th-6th centuries, also from the GandhLra area.
A textual mention of warrior-like directional guardians has been
made in VdlzP (111.84.12). Here, the four lokapdlas appear on a
peculiar symbolic structure called nidukn. They are referred to by
their common Buddhist names (see above), carry spears and wear
"dresses like the Sun" (slrya-vein-dlzard!z, i.e. in armour).
2.1.2 Foundation Stones and Deposits
These significant ritual objects surely deserve their own specialized
study. Much work on the subject has been carried out by Debala
M I T R A Due . ~ ~ to~ the inaccessibility of deposits buried below archi-
tectural monuments at varying depths, e.g. twenty-three metres
below the sanctuary floor in the case of Ankor VLt in Cambodia
(O'CONNOR1966: 53), there is only very sporadic evidence, partic-
ularly within the borders of modern India. What is relevant to our
study is the appearance of the attributes or insignia of the dikpiilas
in miniature form or as engravings, and also the rare finding of small
bronze statues of the eight dikcpdlns in a foundation deposit. There
are literary prescriptions for some of the practices met with, but as
might be expected, these rarely tally completely with the archaeo-
logical findings.
Among the cases discussed by O'CONNOR and MITRA,only very
few offer a complete set of clikpdln symbols or representations of the
directional deities. Most revealing, however, is the wealth of evi-
dence from Sagrahawa/Nepal, where the remains of "early
medieval" brick structures were excavated in the last century.114
Some of the foundation bricks,ll5 which were usually found placed
above small copper vases, were crudely carved with the pictograms
of weapons, the most elaborate layout being a central brick deco-
rated with an eight-petalled lotus surrounded by eight bricks show-
ing the clikpdln-iiy~ylrdk7ns(fig. 1). Indra's thunderbolt (vnjra) should
have been placed on the east. Its asymmetric number of prongs is
unusual: four prongs are pointing northward and only three are
pointing southward. This is followed by the upper part of a spear
(iakti),ll6 often described as Agni's weapon, on the south-east, a
staff (~ln!~~ln) for Yama on the south, a sturdy sword (khaclgn) for
Nairytn on the south-west, a loop with a circle below representing
Varuna's noose @Lis'n) on the west, a flag (~lkzvnja)for VLyu on the
north-west, a club (gaclci) on the north for Kubera or Soma,ll7 and a
trident ( t r i i i h ) with prongs joining at the top for TSiina on the north-
east. The layout of the bricks does however vary from structure to
structure, the only comparable distribution of pictograms being
found in Structure 6. The other structures showed eight partly dif-
fering symbols (no. 9),"8 only lotuses (nos. 3-5, etc.), lotuses and
svastikns (nos. 1 and 2), and a somewhat corrupt arrangement of
four (no. 13) or seven (no. 10) ciyudlzas.
A unique carved stone slab found at Candi Ngrimbi, Jom-
bang/East Java combines the characteristics of a kcirnza-s'ild (tor-
toise stone) with an elaborate lotus pattern showing on its outer
petals a complete set of dikpdla symbols.ll9 The stone slab is pecu-
liar for the fact that the symbols could not have pointed exactly in
the correct directions, as the dividing lines between the petals must
have corresponded to the directions. Thus, Indra's vajra is shown
south of the eastern line, followed by some kind of fire vessel
(MITRA:agni-pdtra), from which smoke seems to emerge, for Agni.
The other attributes are staff (dalzda) for Yama, unusually what
seems to be a double axe for Nairyta, a curved snake as noose (rziiga-
pds'a) for Varuna, an elephant goad (aiikus'n) for Viiyu, a four-
spoked wheel (cakra) with emerging flames for Visnu (generally
not in the directional scheme), or probably for Kubera as a great
king, and the expected trident (tris'lla) for TSiina in the north-east.
Two other pieces of evidence from western India have been col-
lected by M I T R A These
. ~ ~ ~ two squarish stone slabs, hailing from
different places in Gujarat, bear crudely carved symbols, a trident
for h a and a sword for Nairyta, respectively. The former slab was
found at the corner of a jagati (platform) in a ruined brick structure
of the Chaulukya period, with the trident pointing towards the north-
east.
The recovery of miniature weapons in gold and silver foil has
been reported from Site no. 16 in the Kedah area/Malaysia by
H.G.Q. WALES(1940: 34-36). Among other precious objects, the
circular bronze relic casket contained small metal figurines of the
directional animals: elephant, bull, horse and lion.l2l The other
objects were: a bow, two arrows, a sword, a dagger, a long piece of
wire probably symbolizing a noose, a staff, a shield, a clnr~zar~i (hour
glass-shaped drum), a rectangular sheet tentatively identified as a
~ ~ ~ G R ~ N(1984: R O 364):
L D "spatestens zu Beginn des 8. Jh. n. Chr, wurden sie zu
Kriegertypen mit ersclireckendem AuOeren". A Japanese set assignable to c. 12th-14th
centuries is published in FISIILIZ 1995: fig. 1.
113 MITRA1972: 231-247; 1973-74 and n.d. are detailcd studies of foundational
practices. The former two present interesting and so far unique archaeological evidence
from Sagraliawa in the Nepalese Tarai. Tlie second study ~iiainlyconcentrates on find-
ings from South-East Asia, which can be closely linked to the Indian theory and prac-
tice however (cf. also O'CONNOR 1966).
114 Tlie seventeen brick structures were at first considered to bc Buddhist stfipas,
but MITRA(1973-74: 2) states that they "did not yield any specifically Buddhist antiq-
uities."
115 Tlie laying of tlie (first) layer of bricks, i~!nki-ni)~ivcisn,
has been described in a
number of texts. With regard to the treatment of the dircctions, KRAMRISCII 1946: 105-
107 summarizes the evidence: "Five or nine 'bricks' are thus laid down, one in the cen-
tre of a square, the others in the cardinal and intermediate dircctions following the
course of thc sun (pradaksinj) beginning from tlie east."
116 Cf. MITRA 1973-74: 4. She offers the alternative interpretation of a flame, but
a .iakti, wliicli is recognized as Agni's weapon in some of tlie earlier texts (e.g. M P
261.1 1; cf. table X), is even more clearly shown on the respective brick of Structure 6
(i/)i~l.:fig. 4,l).
117 For the Moon God Sorna, tlie club is not tlie usual attribute, but it occurs in a
few cases, in texts as well as in images (e.g. in the Agrtiprrrci!?a, cf. table X).
110 This is the only example of the bricks being arranged in thrce regular rows of
three. The symbol on the east looks like a trident (rriSlila)without a handle, resting on
a socle, while tlie one on the south-east may represent either a stylized bolt (i~crjra)sur-
mounted by a flame (for IndrBgni?) or a firc-altar. On tlie south is a staff (dai,lcla), on
tlie south-west an axe @araSu), on the west a noose O)ciSa), a standard (dl~vcrjn)on the
north-west, a mace (gadi) on the north and a flattened ring on tlie north-east.
IIWITRA 11.d.:47 and pl. 28. For kirrrr~a-Sildsee ibid.: 45-46.
Ibid.: 48; pls. 30-3 1.
121 Tlie distributio~iis generally: elephant - east; bull - south; horse - west: lion -
north (cf. VON SCHROIZDER 1990: pl. ISA-D, KanLaka CetiydSri Lanka, 1st-2nd cen-
turies A.D.).For the animals which liave already bcen depicted in rclief on tlie well-
known ASoka capital at SBrnBtIi, see also V.S. AGRAWALA 1964: 60-63.
 book, a bell or seal, a lotus, a miniature bowl, a ploughshare and a
yoke.122 While the presence of a noose would suggest a connection
with Varuna, as does the sword with Nairrta and the staff with
Yama, it is not as easy to assign the other attributes to particular
guardian deities. As the thunderbolt and the trident are conspicu- rogative for their actual depiction on the walls of the sanctuary.
ously absent, a full set of dikpnlas was probably not not intended
here.
Rare evidence for a set of more than four123 dikl~iilasin the form
of bronze figures comes from the Tdpavava Drigaba, Polonna-
ruva,124 which is dated c. 9th century. The brick container, which is Two important reliefs dating back to the early 5th and early 6th cen-
divided up into twenty-five chambers125 inside, was recovered from
the dome of the stapa and appeared to have been partly plundered.
However, it still contained figures of Indra, Agni, Yama, Naircta,
Varuna and Brahma, which were identified by their respective posi-
tions in the ynrztra-gala. The figure from the northern chamber has
so far been regarded as VisnuIAnanta, but, having a fierce expres-
sion, a demonic hairstyle, and a sword and shield in his natural right
and left hands, it conforms rather more to the image of a fierce pro-
tective deity, e.g. KsetrapBla.l26
Textual support for the arrangement of the guardians of the
directions at the bases of religious structures has been cited by
M I T R A However,
.~~~ the texts she refers to cannot be dated before
the 12th century A.D. She also mentions the comment by P.O. SOM-
PURA (1960: 177) that the practice of inscribing the dik/Iiila symbols
on foundation-stones has survived until the present day among the
Sompura architects. Regarding references in earlier texts, a proba-
ble indication of such practice is contained in the Vi;!z~irNzarrlzot-
tarapuriina, which may date between 400 and 650 A.D. (ROCHER or Scid,zr, as prescribed by thc Mu)~omntcr:'The deposit proper to ;I .fcidro is equally suit-
1986: 252). This text describes how an elaborate riza!idala is estab-
lished as preparation for an ablution rite connected with the direc-
tional guardians (dikl~iila-sniirza). In the rna!z&la, the dikpElasl28
are installed on the eight petals of a lotus, the centre of which is
occupied by the gods Visnu, Brahmii and Rudra (VdlzP II.104.39-
50). The four oceans should be arrayed around this arrangement,
and around them again a number of symbols should be "drawn"
(liklzet). These attributes are enumerated in VcIIzP 11.104.51-52 as:
lotus (parlrlza), wheel (cakra), staff (da!z<ln), thunderbolt (vnjrn),
q u a t i c creature (nzakara), spear (Sakti), standard (dl~vaja),trident
(triifila). Although the number of these iiy~lcllzasis eight, they do not
fully correspond to the attributes of the a~;arlikpiilagroup. The lat-
ter six symbols obviously stand for Yama, Indra, Varunal2" Agni,
VLYU and ISBna, respectively. It is tempting to hypothesize that, as
the symbols of Nairrta and Kubera are absent, parlrlla and cakrri
denote Brahmii and Visnu/Ananta.l30 Moreover, the example of
V~YU who, is here represented by his banner (dlzvqja), but who else-
where in the same text carries a billowing garment (V&P III.58.1),
shows that the dik/?dla symbols apparently follow a tradition differ-
ent from that followed in the iconographic passages on the rlik/?iilns
( v ~ / / f 11.104.42-50;
p 111.50-53; 55-58). In any case, a cloth would
have been difficult to depict. Regarding cakra, it is also found on
the slab from Canqli Ngrimbi.
The above passage does not refer to the construction of a build-
ing but a ritual shed with an enclosure Q7riikdrn) and doors. How-
ever, a nlan&~lnin which deities are installed (devatE-rzyZsa)~3~ is bearing flag (ket~o!tnicrko,ns)~cr)is assigned to him in 111.52.3.
for the laying of the foundation stone (Silii-nynsa; VrllzP
11.29.71-72). Although there is no actual reference to the engraving /~(ldfllfland thc rnkra for the protectors of tlle nildir and the ccntrc, respectively (VAN
of the symbols of the directional guardians on the foundation stone,
two preconditions for such a practice call be discerned: the reference
to a particular set of diklx7la symbols, and the establishment of a
cle~at~-ma!z~lala at the foundation rite. It is quite plausible that
were found to be more easily carved than complete figures
of the guardians. They may signify the clevatd-ma!z~lalaand even
comprise in them all the other deities ("/?ark-rlevntns") which
are traditionally located on the foundation grid of a building
(viist~~l~~ir~i,c.a~~~a~~~IaIn). This outwardly invisible symbolism of the
clikl~dlnsin the foundation of a temple could have constituted a pre-
2.1.3 The Early Depiction of Assemblies of Deities
turies A.D., respectively, are found at Udayagiri and Deogaqlh, both
in Madhya Pradesh. A key article by Debala MITRA(1963) has dealt
with the former relief; the identification of some figures in the latter
one is proposed in a footnote. Both reliefs, together with some oth-
er rclevant examples, have been discussed by K. VAN KOOIJ(1985).
At Udayagiri, the upper row of largely identical figures on the
proper left side of the colossal Varaha image in Cave 5 includes rep-
resentations of Indra and Varuna followed by ten other ndityas, and
Agni and VByu followed by six other vas~ls,thus coinpleting the two
groups of twelve and eight deities, respectively. In each case, only
the first two gods are characterized by individual features. Below
this row of twenty gods is another row or twenty figures which lack
individual traits, but exhibit certain group characteristics. The first
eleven figures from the left have been identified as r~ldms(MITRA
The latlcr tllrcc itc~nsquite clearly itle~itifythc donor of the temple as a 1,ai.fya
able for a voi.f)vr [...] (12.82b); it co~nprisesa seed bowl, a gold plough and a coppcr
yoke" (12.80b). Reference to the directions is made thus: "there is a silver cow at each
of the cardinal points and, in the centre, a bull" (12.81a; aficr B. Dagens).
l*' VONSCIIIIOEDEI 1990: 239-240 discusses the "Lokapilns of the Bralimanical
System" of the Late Anuridliapura Period (c. 300-1000 A.D.).He introduces only sets
consisting of the four guardians Indra, Yama, Varuna and Kubera, while a larger set
from the T8paviiva DigBba, Polonnaruva (sce below) is not mentioned.
' 2 " ~ ~1914:
~ 28-30; pl. LXI, etc.; GAIL1980: 263; p]. XX/5 (frotn lcft to right:
Varuna, Y a m , Agni); MITIIAn.d.: 44.
Iz5 For the litcrary descriptions of foundation deposit containers (garbha-pcitrcr;
Sinhalese ycr111rc1-gcrln)sce KRAMRISCII 1946: 126- 128. "Inside, tlie casket is divided
into compartments, likc the VBsto~na~!<ala, and of varying number. [...l the divinities
of the Visturnnndnla rcsitle in its compartments. The latter Illoreover are replete with
thc wealth of thc eanh". Attention has bee11drawn by VAN LOHUIZEN-I)B LEEUW1955:
376 to an ilnprcssive stonc gnrl)hcr-l~ritrcrfrom Jalatu~~<a/Jnva (latcr ~nentionedby
O'CONNOII1966: 58 and MITRA11.d.:49; pl. 39). Dating fro~rlthe end of thc 10th ccn-
tury, thc box contained, among ollicr trcasurcs, picces of gold lcafinscribcd with invo-
cations to iSdnn and Agni.
[*"Like Bmhmi, thc guardian of the zenith or centrc. who was acti~allypositioned
near the centre of the container, Ananta should be cxpectetl in a ccntral position. Con-
cerning the attributes of the figure, only sword and sl~ieldsurvive intact. Thcrefore, it
could well bc idcntificd as Ksetrapila, wlio is eight-armcd in his particularly fierce
"tcirirosn" aspect (e.g. Mcryorlrotn 36.178-1 82).
MITRA1973-74: 5; iderem n.d.: 45.48.
I Z X As clsewhcre in the VdlzP (111.57.1-6), the south-west is occupied by a deity
referrcd In as Virilplk~a:lnd not by N;~irta.It is only here, in VLIIIP11.104.39-50, that
Viricpiksa is attributed thc sword (klro~lg'~). As this is also Nair!.ta's main attribute. both
gods seem to be closcly rclatcd, tlicir main difference consisting in Virfipiksa's ~jcihrrrzn
camcl (rrgrtr). For !$inn, wlio is tlcscribetl as antlrogynous in 111.55, his rcgular nppear-
ance and name (wrongly "Bhilacandra" in P. Shah's translation) are given hcrc.
l*" Although Varu~!a has a goosc (Ircrr!rscr) as his vcl~iclein this text, n rrinkartr-
l'" It is an intriguing fact that thc IINI,CISN~I~(Igroup of directional deities, popular
in ancient Java and Bali, which shares the elnble~nswith the (r~~rrr/ik/~ci/cls, also adds the
LOIIUIZCN-D!! LEEUW1955: 372-373). However, Bralimi and Visnu have been appoint-
ed as the guardians of the soutl~and the north, rcspcctively. Thc author suggests that
this group, tlie other tne~nbersof which arc all forms of Siva, "undoubtedly [...l gocs
back to a group which must have cxistcd in India long before that." (ibid.: 382). It is
also significant that the r~n~~crsatigcr group is generally rcprcsented only by its attributes.
Most probably tllcse are thc dcitics of thc site, vris~rr-clevtr-gcr!~~,
wllicll are men-
tioned in 11.29.73.
(D. SRINIVASAN 1990: fig. 5). If we stretch the available data, the
panels of four directional guardians could be regarded either as a
"late" version of directional y a k ~ a s(as at Bhiirhut) or an "early"
group of br3hmanical gods (as at Udayagiri, mentioned above,
respectively).
2.1.5 Some Early Sculptures of Indra and Agni
Both Indra and Agni were fairly popular in the period under survey
in this special inquiry (c. 1st century B.c.-5th century A.D.).There
is hardly any doubt that the early sculptures treated here do not rep-
resent either of these gods in the capacity of a directional guardian.
In order to gauge in how far these deities were expected to assume
a particular appearance, i.e. how far their images were typologically
determined, some early images of Indra and Agni will be classified
according to their prominent features.
Indra can usually be recognized from his impressive cylindrical
crown and a horizontally placed third eye on his forehead. There are
some beautiful examples of detached heads of this typel39. Alto-
gether, there are many more examples among Buddhist reliefsl40,
which could not be included here. The following list refers only to
broad iconographic types:
I - INDRA
A Without thunderbolt (vajra)
In cifijali-~rz~frlrZ:
1 Jamiilpur/Mathurii, State Museum, Lucknow, no. B 208, c. 2nd cen-
tury, publ.: J.M. ROSENFIELD 1967: fig. 40 b. 1967.
2 Mathurl Museum, 1st century, publ.: P. PAL1979: fig. 14.
3 Niigiirjunakonda, 3 r d 4 t h centuries, publ.: A. RAY1983: pl. 202.
With waterpot:
4 Near Charsadda, Samuel Eilenberg, New York, 1st-2nd centuries,
inscribed "Imrlra!~ rlevariji", reconstructed height: 46 cm, publ.:
PAL1979: fig. 1 (commented on by D.C. SIRCAR 1981); CZUMA/
MORRIS 1985: no. 57.
On elephant:
5 Nlglrjunakonda, 3rd-4th centuries, publ.: S.K. GUPTA1983: pl. 17.
6 Bhiijii (after GYANI1951-52: king Mindhiit3 in the guise of Indra),
2nd-1st centuries B.c.,ZIMMER 1960: pl. 41 (with turban).
7-9 Three terracottas from Candraketugadh, 1st centuries B.c./A.D.,
heights: 12.5; 12.7; 18.5 cm, publ.: S.S. BISWAS 1981: pl. XIIa, b;
Christie's Amsterdam, Sales Catalogue, 23.10.91, lot 57 (toy
cart!).I4'
Jj With thunderbolt
With waterpot, 1.:
9 Siifici, Sttipa no. 1, northern torci!in, inner side, lowest beam (ViSvan-
tara Jiitaka), 1st century, publ.: MARSHALLJFOUCIIER 1940: pl. XXIX,
3c.
Torsi, vajm in 1.:
1 0 Mathurii region, Virginia Museum of Fine Arts, Richmond, 1st-2nd
centuries, publ.: PAL1979: fig. 20.
]I Mathud region, c. 2nd century, publ.: I~zdiarzArclzaeology - A
Review 1981-82: pl. XXXIV C (with garland of leaves - varzn11ziIc7).
12 Acquired at Tarsi, Govt. Museum Mathurii, no. E 24, c. 2nd centu-
ry, height: 53.3 cm, publ.: J.Ph. VOGEL
1909-10: pl. XXVIIIb (with
varzamil7).
With small elephant vehicle (attribute broken, most probably vajrcl):
l 3 Sitaliighiiti, Govt. Museum Mathurii, no. 46.3226, c. 4th century,
publ.: R.C. SHARMA 1976: fig. 68 (fig. 6).
I1 - AGNI
A Surrounded by flames, gesture of fearlessness (left) and waterpot
(right)
14 Bharna Kalan, Govt. Museum Mathurii, 3rd quarter of 1st century,
inscribed "[ ...l i[gn]isa pra[f]i[nz](Z) [...l", height: 196 cm, publ.:
D. SRIN~VASAN 1990: fig. 3 N B (right arm broken, ogee-shaped
nimbus arising from the head).
15 Kankiili TildMathurii, State Museum Lucknow, no. J 123 or G 280,
1st-2nd centuries, height: 83 cm (incomplete), publ.: N.P. J ~ S H I
1972: fig. 43 (right arm broken, flames only behind the upper part
of the body).
1 6 KatrUMathurii, Govt. Museum Mathur?, no. 40.228, 2nd-3rd cen-
turies, publ.: G. KREISEL1986: Abb. A 21 (right arm broken, flames
extend down to the elbows).
1 7 Probably from Mathurii region, Bharat Kalii Bhavan, Varanasi, no.
23171, 2nd-3rd centuries, publ.: D. SRIN~VASAN 1981: figs. 294-
296 (stylized flames, same as above, covering the whole back of the
figurc).
18 Mathurii District, Govt. Museum Mathurii, no. 42-43. 3048, 3rd
century, publ.: R.N. M I ~ R 1981:
A fig. 5 (fig. 7; flames behind the
whole body).
19 KatrdMathur3, Govt. Museum Mathurii, no. 40.2883, c. 3rd centu-
ry, publ.: M.M. NAGAR 1942: pl. 11, fig. 3 (fig. 8; only a few isolat-
ed flames arise from behind his head and shoulders).
20 Bhankari, c. 430-460 A.D.,publ.: J. WILLIAMS 1982: pl. 74 (in the
populated nimbus of ViSvartipa Visnu [for a detail photograph see
MAXWELL 1988, pl. 541; round flaming aureole behind his head).
21 RiijghiiT/Varanasi,Gupta period, publ.: V.S. AGRAWALA 1984: no.
385 (sealing; flames arising from two aureoles, behind the lower
and the upper part of the body, respectively; attendant with spear or
trident on each side).
22 Ahicchatrii, National Museum, no. 62.262, 5th century, publ.: C.
S~VARAMAMURTI 1976: fig. 16 (terracotta; separate nimbus behind
the head).
B (doubtful) On a cart drawn by two rams, with or without attendants:
23,24 Two terracottas from Candraketugadh, Asutosh Museum Cal-
cutta, c. 1st and 2nd centuries, respectively, heights: 18.4 and
18.8 cm, publ.: S.S. Brsw~s1981: pl. XI11 and XIVa.
The variation in the depiction of Indra may be partly due to his asso-
ciation with the Buddhist tradition. In general, however, both Indra
and Agni are characterized by a rather high degree of conventional-
ity during this early period. Indra has been shown accompanied by
his elephant from an early period onwards, but a small-size version
of his vehicle standing behind him, which corresponds with the lat-
er convention, is only known from the early Gupta image from
Sitaligh3ti (fig. 6). It seems to be significant that in this image, his
usual cylindrical crown has been replaced by a crown of a different,
less individual type. The vajral42 has always been Indra's undisput-
ed attribute, from SiiRci (9) onwards, but there are also a number of
exceptions to this combination, particularly in narrative scenes,
where the identity of the god was understood from the context.
However, in the later Kusiina period, "Indra with vajra" seems to
prevail.
139 Sec K.M. SRIVASTAVA 1986: pl. L111 A, ofc. 1st ceniury A.D.;P. PAL1979: fig.
10, of c. 2nd century A.D.
'4W.g., the popular tlie~neof Indra's visit to the Buddha, depictions of which have
been studied by T. BLOCII 1898 and A.K. COOMARASWAMY l928a.
141 The only faint hint at such an identification is the fact that Sarya with a quodri-
ga also appears at that site (BISWAS 1981: pl. XI a) and thus Inay complete the pair of
Si~ryaand Indra which is known from Bh5j3 (see abovc).
142 For a typological study of the vojm see A. BANIZRJI 1981.
143 Both sculptures have been tentalively regarded as a pair of doorkeepers, which
would not speak against this view. Thc analogues for this figure collected by D. SRINI-
VASAN (1990: pls. 2; 15; v o SCIIROEDER
~ 1990: pl. 1I E) secln to suggest that the g o k p
aspect is quite pronounced.
 In the case of Agni, the yukp-like build and the enormous size
of the earliest known image from Bharna Kalan (14) are particular-
ly striking. Its close association in situ with a sword-bearing y a k ~ a
of the same size143 makes it even more likely that the image of Agni
descended from the y n k p typel". Such a yak~n-aspectcan less
easily be applied to the other images of the Kusiina period which all
show more modest proportions. However, there is another large,
impressive statue of Agni, found in Siirnath, which can be dated to
the 6th or 7th century (fig. 314). Although a number of individual
traits are always present, representations of Agni did not vary
greatly even in this early, supposedly formative period. One of the
few subjects of variation is the way of rendering the flames, but this
variation generally persists in later periodsl45. Defining the shape
and functioll of the flames, "aureole" (radiating light behind the
body) would be an appropriate expression, except in a few cases
where only the head is encircled by flames (nimbus). That the flames
were not generally perceived as the nimbus of the deity call be also
inferred from the fact that a separate nimbus is indicated at Ahiccha-
tri (22). There is no evidence of an animal vehicle being associated
with Agni in the early centuries A.D, if we do not accept the identifi-
cation of the Candraketugadh terracottas (23, 24) as Agni.
Evaluating the evidence gathered on Indra and Agni and con-
sidering the iconographical development in the later periods (see
sections 2.5.1-2), it becomes apparent that earlier concepts of
these deities were drawn upon by those artists who created the
nsfadikpdln set.
2.2 Centres of Early Development (c. 400-650 A.D. )
The beginning of this formative period is marked by the panel from
Pawiiyii/M.P., which provides, to our present knowledge, one of the
earliest pieces of evidence for the depiction of a group of direction-
al guardians (fig. 9). Subsequently, there are no surviving records
for more than a century, until about the second quarter of the 6th
century, when the fine gaviikp-framed reliefs of Yama and Kubera
from BhOmarii were crafted.
The period under study is characterized by the fact that the com-
plete set of eight directional guardians (Indra, Agni, Yama, Nairyta,
Varuna, Vayu, Kubera, iSiina) is not known to have been depicted-
generally, the older concept of four directional deities or an incom-
plete group was represented. An interesting set comes from
MundeSvari. This group of now only six, but originally eight deities,
which includes Si~ryaand Candra, has been referred to as "Manu
Lokapiilas" here (see section 2.4.1.3). Progress towards the canoni-
cal group is most advanced at Ellorii, Cave 29. However, 110 final
agreement on the date of this cave temple, which closely resembles
the well-known MaheSa shrine at Elephanta (Cave l), has yet been
reached. Although it is generally considered to date before 650 A.D.,
a wider range of probable dates has been suggested by the authorsl46.
A rather ~nysteriousensemble is found at Biidiimi, Cave 3 (fig.
10), where the central medallion of the ceiling of the porch consists
of eight deities arranged according to the directions, without any
distinctions in size. Although ~ i v (south-west)
a and Brahmii (south-
east) appear among these, the other deities seem to belong to an
early stage of the agaclik~~Liln scheme. There are also two complete
ensembles of the - smaller - group of four directional guardians in
the same cave temple, indicating that the development towards the
representation of the group of eight had not yet been completed.
It is important to note that even the earliest images of the direc-
tional guardians are found on diverse architectural members (lintel,
ceiling, wall, cornerstone, iikhnra), in different arrangements
(frieze, rosette, singly) and functions (as main or attendant figures),
with variations from region to region.
2.2.1 The Deccan
The Deccan plateau has proved to be one of the cradles of dikpiiln
representation. In this region, the guardians' distinctive vehicles
have always been rendered prominently. Thus, it seems that the
animals or other carriers of the directional guardians formed an
essential part of their syn~bolism.
In the heartland of the Calukyas, the guardians generally occu-
pied a subordinate, peripheral position, as they remained restricted
to the ceiling panels. This treatment started at Biidiimi during the
early phase. While the rlik-llokapiilns also appear in a subordinate
position on the great relief panels at Elephanta and Ellori, their
introduction on the temple tower (s'ikhnrn) at Alampur (Andhra
Pradesh) marks their first occurrence on the outer shell of a temple,
probably indicating an increase in their importance. The same site
subsequently produced the earliest known complete sets of direc-
tional guardians on the temple exterior.
Caves l and 2: These cave temples, which are less elaborate in struc-
ture and decoration than Cave 3, are adjoining and are reached in
approaching the rock face of the Southern Fort. They have been
regarded as the earliest of the Biidiimi cave temples, dating to 560
and 570 A.D. according to TARR(1970: 184). Friezes of intricately
carved narrative reliefs embellish the entabulature of these two
caves, as in Cave 3. Groups of directional guardians seem to have
been included in two cases, in Cave 1 among the guests at the
a Parvati (R.D. BANERJI1928: pl. IV a), and in
wedding of ~ i v and
Cave 2 among the deputation of gods sent to Devi (ibid.: pl. XI c).
Both these panels are situated in the porch, on the north side of the
westernmost (right) bay and on the north side of the central bay,
respectively.
As the crowded panels are not easy to read, it cannot be decid-
ed with certainty how many individual lokapcilns are shown. In the
former case, four deities follow on the right side of Indra, who is
unmistakably characterized by his thunderbolt (vajrn) in front of his
trunk, depicted on the right portion of the panel. Some of these
deities may have once been provided with vehicles. With regard to
this, it should be recalled that in the MnhiiblzLirnta and other texts,
Indra p l u s the four loknpilns are sometimes referred to (E.W.
HOPI<INS 1915: 150).
In the scene showing the gods approaching Gaja-Lakyni, a
number of deities are depicted on the left side, with and without
14.' Tlie early develop~nentof the Agni image has bee11studied by R.C. AGRAWALA
1965 and B. SAIIAI 1967. Even before the discovery of the Bharna Kalan image, both
authors were of the opinion that "Ancient sculptures of Agni [...l are pretty colossal and
follow the early tradition of two-ar~nedYaksha statues." (AGRAWALA 1965: 15 1; simi-
larly SAIIAI 1967: 146).
145 Agni's hair appears in the shape of flames in South India (see section 2.5.2).
An early piece with a similar features (D. SRINIVASAN 1981: fig. 297; G. KREISEL 1986:
Abb. A 22) most probably does not portray the Fire God, contrary to the views of the
authors. Here, the features of the face are prollouncedly grotesque, which is never the
case with Agni.
. " '1 BEINSON(1992: 198) in her stylistic treatriicnt of the Ellor-a caves, com-
ments on Cave 29: "All the sculpture is in a rapid state of decline", and proposes a
chronological position of"some time in the late seventh or early eighth century". The
lowcr limit is marked by a date short after the completion of Cave no. l at Elephanta.
 B Brahmii I Indra K Kubera Kii Khrttikeya S Soma S Siva V Varuna Vi Visnu Y Y a m

Ill. 1: Biidiimi, Cave 3, location of ceiling panels

viihanas, moving towards the centre of the panel. On the right side,
there might also be one more deity belonging to this scene. S.L.
BUCHANAN (1985: 56) identifies the group on the left as SCirya (?),
Soma (?)147, Yarna, Varuna, Indra, Khrttikeya, Visnu, Siva and
Brahma. It is quite clear that, apart from a small figure on the
extreme left, two directional guardians riding their vehicles
(Varuna, Yarna) are superimposed over the other two (Soma [?l,
Indra), the left figure representing Varuna on his awe-inspiring
rizakara (mythical aquatic animal). Some other scenes contain
figures of single lokapiilas, e.g. in "Indra entering the court of Bali"
(BANERJI 1928: pl. XI b).
As other gods accompany the four directional guardians, it is
obvious that the latter cannot be representing the gods in general.
They simply form a particular group of deities which is always iden-
tifiable by their distinctive ~h~ presence of directional
guardians in in Caves and makes it 'lear that although
they were known at this stage, they were not yet chosen as subjects
for ceiling reliefs.
Cave 3: The earliest sets of four directional guardians
located in their respective regions are extant here, on the ceilings
of the porch (tnukhnnza!z~lapn) and the hall (mal~iinza!z~apa)of
this Vaisnava shrine~48.~h~~~is also an enlarged version of at least
six zokflpfilas located in the porch' The absolute date of A'D' 578'
referred to in an enhances the this cave
temple.
The oblong nzukhariza!z<laj~ahas on its ceiling six square panels
bearing representations of particular deities in their circular centres.
Read from the west to the east, these are: -none-, V a r u ~ ~BrahmLi,
a,
Visyu, ~ i v aIndra
, and an u~lidentifiedfigure149 (ill. 1). The ceiling
of the first bay from the west is devoid of any relief, but bears a
painted concentric design. If we count Indra (fig. 11) and V a r u p
(fig. 12) in this arrangement as belonging to a particular system of
spatial reference, there are three more systems in this cave temple,
14' Ag.il11, tlicrc IS a d~fftculty In ~dentlfytligthe antmal. but its statulc ~ c ~ c m b l c s
the lton vriho~~rr c the Bralilni pdnel In Cavc 3 (lig. 13)
on tlic n o ~ t l ~ es ~ nd of
148 The flrat dct.llled deacr~ptlonof the cclllng panels tllu\tratcd by tllawtngs of
the ~nrr/</lnn~cr!~cIc,,~npanels was publ~slicdby B U R G I I.877.360-362.~~ De\pltc the quite
exdct ~endcllngof the outl~nes,tnany of Ill\ ~tlcnt~ficat~on\ are e ~ r o n e o uPhotograph\
~
wcle 111cIudcdby BANERJI (1928 30-57, cell~ngrcl~cf\ 55-57) 111 h15 docutncntat~on
on tlic cavc tcmpIc\ of Biidltn~ HI\ refercncc numbcl fol the glcat V.uqnava cave 15
"IV" In thc tcxt, but "III" ~ntlie plate \cctton Only tlic Vl$no p ~ n c ha\ l becn rcp~oduccd
by TARR(1970 I 66-167, fig 24) T I ~p~ncl? C tllc I I I ~ I ~ ~ ~ I I ~ ~ I I ~except
~ / ~ I , I for
[ I , t11cccn-
tral Bralilni (K V SOUNDARA RAJAN198 1 p1 XIIB), have, to my knowletlge, rc~iiatnecl
unpubllshcd
1" Many dtlferent ~dcntlf~cat~on\ li,ive been p~opo\cdfor l h ~ ?figurc, wli~chcdr-
rles a club ( r ) and a cup (I , dftcr BAN~IIJI) and ?lt? on a \cat endtng ~n two r~~ol~trro
IIC,~, BURGLSY (1877.361) ~dcnt~ftcd the figure a\ Pradyumna WIIIICTARR(1970
166) only mcntlon\ tt a\ an "uti~dentlf~cd god", SOUNDARA RAJAN (1981 63) m~stakcr
11 for VarunaIt I S \[range thdt both llic attr~butc\could belong to Kubcra, however not
In 1111scolnb~liat~on A tea\onable but not provable propo\al ha? recently bcen made by
by BUCIIANAN (l985 58-59) She regard\ the figure a? a represetitatloll of "Pulake61 I,
fathcr of both Cavc 3's patron Mdngale<a and K ~ n gK i r t t ~ v a ~ m cnv~sioned
~, a\ ~csldtng
among thc god\ "

24

--. /
each one bearing individual characteristics: Brahmii of the
iizuklzaiiza!z~ln~~n
is surrounded by four directional guardians
(fig,. 13); Visnu, the main deity of the shrine, is attended by six
lokapiilns plus Brahmii and ~ivalso(fig. 10); and Brahmii inside
the nln!zdn/?n appears in the midst of four guardians, Kiirttikeya
being substituted for Yama on the southl5I (figs. 14-18). In the
iizukhniiia!z&pn the symmetry is maintained, as all the central
figures are surrounded by compartments which are, in the absence
of directional guardians, filled with celestial couples, sages and
the like. In the nzalzEiiza!z~n~~n,
the medallions "float" in squarish
sunken panels. The outer frame of the ceiling consists of a row of
lotus petals.
In addition to this, some of the IoknpCilas are included in a
battle scene on the iiz~lkhmiza!$n/?n entabulature, on the south side
of the second bay from the east152.
Iconography: Apart from the vehicles, which are impressive for
being rendered lifelike, relatively few iconographic features of the
two-armed loknl~iilnscan be recognized. This makes it difficult to
identify them in some cases. Beginning with Indra, he tends to be
particularly diminutive in comparison to his mighty elephant
Airiivata, who dominates the composition. This is most obvious on
the eastern panel of the ii~nhCiiiln!z~lnpn(fig. IS), which has an
impressive counterpart at Aihoje, Rr?va~!aPhadi cave's,. Here, the
animal is also shown in motion, in a leaping or flying attitude. There
is, however, a better balance in the size of the vehicle on the one
hand and the deity and his attendants on the other hand. Indra is in
all cases accompanied by one or two attendants carrying an umbrel-
la or a standard (fig. 15). Twice, he seems to be holding his thun-
derbolt, which is barely discernible. In one depiction, Yama sits
with ease on his galloping vehicle, carrying his staff (clrr!l&) and
attended by a flying unlbrella bearer (fig. 13). He appears devoid of
his attribute in another depiction (fig. 10). V a r u ~ pis also rendered
once without any prominent attribute (fig. 13), twice with his usual
noose154 (figs. 10, 17), and once with a flower in his right hand,
which emerges from a long, curved stalk (fig. 12). In two cases, it
seems that Varuna is holding a flower or shell in his lert hand, close
to his lap (figs. 13, 17). The ram-rider in the north-eastern position
of tlie elaborate Visnu panel (fig. 10)155, as well as in the northern
position in the iiinhciriln!i~lryn (fig. 18), most probably Kubera,
unfortunately has no attribute. The deity governing the north has a
lion for its vehicle in two cases (figs. 10, 13) but no particular
device, except perhaps a lotus bud in its right hand in the former
example. In the same panel, this deity stands out for appearing not
on, but beside his animal. Although there is no extant analogue, it is
quite probable that this deity is Soma, who is mentioned repeatedly
in the texts as guarding the north. As an argument for such an iden-
tification, some authors have suggested that CandramiiISoma is
assigned a lion banner (sii~ihatNzvnj(~) in VdlzP 3.68.15156. The ref-
erence to a lion banner for Sorna is not very significant however, as
the same text also assigns lion banners to Kubera (3.53.6a) and
Sbrya (3.67.8a).
A personage riding a horse on the north-west in the elaborate
eight deity-scheme (fig. 10) could be Viiyu, who later in Central
India is characterized by this animal and not by the more usual ante-
lope (e.g. figs. 157, 301). As he is not carrying an attribute appro-
priate for a wind-god, but probably a fruit in his hand, this remains
a hypothesis.
Brahmii, the guardian of the zenith, occupies the central position
in two instances and the south-east in the other case. He holds a
waterpot (kaiiln!idnlu) by its neck in his rear left (figs. 10, 13) or
front left hand (fig. 14), and a staff-like implement which is curved
below in the rear right hand (figs. 10, 13, 14). In another instance,
his rear left hand carries a rosary (fig. 14), while the front left hands
in the other images rest on his thigh. There is no variation in the front
right hand, which is held in the gesture of "fearlessness" (abhnyn).
His goose vehicle occurs in two cases as a support (figs. 13,14), and
a double-lotus in one case (fig. 10). Kiirttikeya, who is included in
the lokqiila-scheme of the iizalziimn!zdnpn (fig. 16), appears with
three heads astride his peacock vehicle. The left hand rests on his
thigh, while the right holds a trident (tris'dn), which shows a simi-
lar composite construction as the weapon of a loose b n a image
from the vicinity of Mathurii (see fig. 178). Two pot-bellied dwarfs
with flywhisks accompany him.
The dresses of the directional guardians vary slightly in the de-
pictions. They wear the briihmanical cord, not too lavish ornaments
and high crowns in most cases, although the ram-rider (figs. 10,18)
forms a conspicuous exception with a low, helmet-like crown.
Completing the description of the world guardians in Cave 3,
two further images of V a r u ~ phave to be mentioned. One of these is
set in a medallion, carved on the western face of the eastern front
pillar of the bay of the iil~lkhaiizn!l~ln/~n,which has V a r u ~ pon its
ceiling (BANERJI 1928: pl. XXI c). This image resembles very much
ceiling panel (fig. 17), with the excep-
the one on the iiinhiiii?n!i~la/~n
tion that the snout of the iilnkarn (mythical aquatic animal) is shorter
and the noose either badly damaged or originally substituted by a
lotus or something similar. In the other, likewise majestic image,
Varuna is shown guarding the nectar obtained from the mythical
"Churning of the Ocean", surrounded by other deities (ibid.: pl.
XXIII a). Here, the composition of god and vehicle is again similar
to the former image, a very realistically rendered noose being held
150 Tlie depiction of BralimB and Siva, who also flank Vi$~!u in their own panels
whicll neighbour his, has led BUCIIANAN (1985: 56-59) to reject tlie opinion exprcsscd
in earlier works (e.g. BURGESS1877: 361; TAIIK 1970: 167;SOUNDARA RAJAN1981: 63)
tliat tlic eight peripheral deities arc basically identical with tlie tlikpcilns. I have else-
where (Wcsse~s-MEVISSEN 1997) argued that her identifications, which only recognize
four of tlie figures as loknl>8los,cannot be accepted. My main argument is tliat at least
onc dircctional guardian, Kubera, occurs on thc north-east, which shows that tlic
Iokt1p8lns' presencc was not rcstrictcd to the four ~iiaindirections.
151 BUCHANAN (1985:59) has proposed thnl Yama's prese~iccin frotit of thc sanc-
tum would have bcen regarded as inauspicious. She reflccts funher on this: "Possibly
KBrttikeya, war god and patron oftlic Calukya dynasty, here was rcpresentcd in tlie role
each Calukya king envisioned for IiimselE lord of the southern quartcr. An inscription
[...l refers to PulakcSi 11 as the lord of the earth's southern regioti ((lnksi~rcil,o~ho-
(lord) which is used to refcr to Ksrttikeya"
employing the tenii s~~cii~~iir
p!~rhiiy'i!r-,sl~ci~i~i)
(ibid.:91, n. 156).
15' This is tlic battlc between (Iclel~crsand trsrrrns which ensucd from the churning
of thc occan (BANERJI1928:pl. XXlll c; cf. BUIIGESS1877:platc 0111). p. 360). Varul!a
is the best recognizable 011 the extreliie right of tlic battle scene in Cave 2. Althougli tlie
diminutive size of tlie figures makes it extre~iielydifficult to identify the details, it is
quite apparcnt that not more than three loko~~cilos appear on tlie right behind a massivc
elephant, which, however, does not carry lndra but an anonymous warrior. They are
(from the left): Kubera (?), Yama and Varuna.
153 Tlie ceiling of tlie porch shows a centml lotus flankcd by niedallions contain-
ing itnages of Visnu and Indru (R.S. GIJlTe 1967:pl. 138), respectively. Indni does not
hold an castern position Iicre.
154 It is inleresti~igtliat tlie reprcsentation of the noosc varies. In fig. 10, it appears
to be staff-like with a stiiall round top. la fig. 17, the loop is extremely large and circu-
lar, while tlie central "hole" seenis to bc hardly rcccssed.
15s Since BURGESS'S(1877: 361) timcs this deity has bcctl - tuost probably
erroneously - identified as Agni. BUIIGESS co~iipletelydisregardcd tlie proper direclions
and lier re fore understood this figure to belong to the south-east.Only BUCIIANAN (1985:
58) came forward with ncw ideas, suggesting tlie god Kubera or a king as possible iden-
tifications. One single text, wliich locatcs Agni as "Sikhin" on the north-east in a foun-
B!.S 53.43).is not enough to support an
dation grid of a building (o8.srrcl1rrr.ri~n111n~1~InI(1;
identification as Agni.
15V.A.G.RAO(1916: 319 and App. C: 93) erroneously labelled a passage from
tlie Vi~!irrt/ltnr?~~ottornp~irci~rn (3.68.15)as "r2lnr.syol)ro.8!,cr".This was citcd by DE
MALLMANN (1963: 136). who first suggested tlic indentification as Sonia for tlie figure.
 in Varuna's right hand. This scene is situated on the south side
of the third bay from the east, on the entabulature of the
niukhanza~ldapa.
It was probably not accidental that the unusually strong refer-
ence to spatial symbolism at BBdiimi, Cave 3, was found in an
exceptionally spacious cave temple. It is the only cave at BBdiimi,
in which the central part of the nzahiinia!zdal7a is not supported by
coloumns but has been conceived as an open hall corresponding to
nine bays in size. It is somewhat surprising that these nine compart-
ments have not been used to represent more than four lokapdlas
grouped around a central divinity, as it seems improbable that the
ceiling reliefs of the large hall were carved prior to those of the
rn~rklzama~~&zpa. In any case, the enlarged set of directional
guardians in the porch must be regarded as the early apex of a devel-
opment, which finally led to the representation of the canonical
group of agadikpiilas.
In the Jaina Cave 4, an isolated image of Kubera is reported to
be located on the eave. The figure is pot-bellied and is shown squat-
ting, without a vehicle.
The most elaborate cave temple (no. 1) on the well-known island of
Elephanta (or GhBrBpuri), situated in the Mumbai harbour, provides
the earliest known images of directional guardians incorporated into
monumental relief panels. These carefully composed panels are all
centred on ~ivanziirtisls7:
I ArdhanBriSvara (southern wall of main hall, left of MaheSa;
fig. 19)
2 GangBdharamCirti (ditto, right of MaheSa; O'FLAHERTY et al.
1983: pl. 45)
3 RBvaniinugrahamCrti (northern wall of main hall, eastern-
most panel; W.D. O'FLAHERTY et al. 1983: pl. 70; C.D.
COLLIN~ 1988: fig. 3)
4 NateSa (western annex, south of the linga shrine; fig. 20)
The complex shrine has been assigned to c. middle of the 6th cen-
tury 158. It is notable that only some of the directional guardians have
been represented here, and that they have been arranged, at least in
the main hall, according to the actual directions.
The huge central icon of MaheSa or SadiiSiva on the southern
wall is flanked by the panels 1 (fig. 19) and 2. Each of these north-
facing reliefs has on its left (eastern) side the god Indra riding a
mighty elephant. The ArdhanlriSvara panel ( l ) bears an image of
Varuna on niakara in a corresponding position on the opposite
(western) side. It is somewhat peculiar that the Galigiidhara panel (2)
Varuna, seemingly for want of space, although his presence as
a representative of the waters might have been appropriate here.
The ram-riding figure on the right of RBvanBnugrahamCirti (3)is
most likely Kubera, although this figure has hitherto been identified
as Agnils9. While the latter would be the only guardian of an inter-
mediate direction at Elephanta, Kubera's presence would allude to
the northern position of the Kailiisa mountain, the mythological
scene of the theme.
The image of the dancing ~ i v in a the western annex of the cave
(4; fig. 20) again appears to be flanked by two lokapiilas, Kubera
or Agni (?; south) and Indra (north). The head of the vehicle of the
former is somewhat damaged, but there remains little doubt that
the depiction represents a ram. While the correct orientation of the
directional guardians (Indra - east; Varuna - west) is clearly de-
picted on panel I, here we are confronted with a strange distribution
of the directional guardians. There is no plausible explanation for
the choice and placement of the lokapnlas on this panel, which faces
east (slightly turned northwards).
While the lokal~lilasare depicted on only a few of the relief pan-
els at Elephanta, the recurring arrangement of the attendant deities
Brahma and Visnu provides a good example of a fixed "frame"
for the central theme. With the exception of a particular group of
panels in the main cave160, Brahmii is always represented on his
goose-vehicle on the left (proper right) side and Visnu on his vehicle
Garuda on the right (proper left) side. In those cases where the direc-
tional guardians occur together with Brah~niiand Visnu, they appear
inside the frame formed by these attendant deities.
Clearly at Elephanta, the directional guardians were not con-
ceived as a visually inseparable group, but the presence of a few
members of the group seems to have conveyed a particular symbol-
ism referring to actual or mythological space.
Iconography: Only few iconographic traits are visible on the less
damaged panels I and 2. Indra and V a r u ~ pwear the ~ ~ s uornaments
al
and a wig-like curly hairstyle. While Varuna's noose, hanging down
from his left hand, is clearly discernible, Indra's attribute in his right
can hardly be identified. It may be a thunderbolt or a lotus. He is,
however, characterized by an impressive, intricately decorated
cylindrical crown.
Ellorii, Caves 14, 21 ar~rl29
There is a general consensus among scholars that these early Hindu
cave temples at the well-known Deccan site Ellor2 date from the
6th century and that the relative chronology runs thus: Caves 29,21,
14I6l.It is supposed that Cave 29, which is both in layout and sculp-
tural programme largely a replica of Elephanta (Cave l), is only
slightly posterior to the latter, which belongs to c. middle of the 6th
century. This assumed posteriority, which seems to be somewhat
contradicted by the inferiority in the quality of the sculpture, is
nevertheless confirmed by the introduction of "new" members of
the 1okal)iila group at Ellorii. It is still possible that Cave 29 post-
dates A.D. 600, as there exist no other similarly developed loka~~nla
sets at such an early date. However, it is thrilling to find that an
otherwise unidentifiable figure among the lokqnlas grouped
around NateSa in Cave 29 (fig. 21) seems to have been modelled
after the guardian of the north in the eight deity-scheme at Biidiimi,
Cave 3 (fig. 10). Therefore, although EllorB 29 appears to be truly
innovative in its depictions of directional guardians, it contains
references to the traditions of both BBdiimi and Elephanta.
For the ground plan of the cavc temple see O'FLAIIERTY et al. 1983: fig. I
(p. 18). HARLE'S (1986: ill. 96) indication of the north should bc rotatcd by 180 tlegrees.
'5Ulthough the cavc had earlier been assigned to the 8th century, tlie mid-6th
century has become the generally ncccpted date of origin ( O ' F L A ~ I E ~etTal.
Y 1983: 5).
A closc association with the Knlacuri king Kysnarija (c. 535-575) was revealed by the
find of a coin hoard on the island (S. GOKIIALE 1976). S.V. GORAKSIIKAR (1981: 250-
251) and K. KIIANDALAVALA (1990: 19) give a similar chronological assessment, but
they prefer the Kolikan Mauryas as the authors of the cave temple.
159 O'FLAI-IERTY er NI. 1983: XVII. C.D.COLLINS (1988: 43) suggests as a mytho-
logical rcnson the fact that Agni was present on Mount Kailisu tluring tlie birth of
Skanda.
1" On the pancls of AndliakisuravadliamUrti (O'FLAII~RTY er al. 1983: pl. 591,
P~rvati-prrri!icr)~a/Vaiv8hikami1rti (ibid.: pl. 51), and or the gambling match between
Siva and Pcirvati (ibid.: pl. 27), Brahmi ancl Visnu do not occur. Interestingly, all these
panels belong to tlie "inner circle" of imagcs, as they are situated at the extreme ends
of the inncr rectangular space containing the sanctum.
1" Recent rescarch by G.H. MALANDRA (1993: 25) furtherco~~firrns this. She sug-
gests a time bracket of 575-600 A.D.for these three caves.
 Regarding the dikl~iilas'function at Ellorii, D.N. VARMA (1988:
3 10) has made an observation which holds true for most of their rep-
resentations, including even the later ones, at this site:
l
1 ... the Dikpiilas at Ellora have beell portrayed as clrarizatis
/-'m-sorzae for heightening the intensity of the event portrayed in
l
the sculptural panels and not for guarding individually any
~
l
particular quarter.
Varuna at Elephanta (fig. 19). The most exceptional feature of this
panel is the fact that the two directional guardians form a visually
coherent group with Bra111n%and Visnu.
Cave 29: The Nateia panel on the extreme north-west (facing
east)l6? of the main hall is attended by probably eight directional
guardians, with one associated personage below (fig. 21). The relief
is executed in a strangely rough style, partly unfinished, and replete

1
I with empty space -thus, the background of the composition is not
The depiction of the directional guardians as a group may have been crowded with figures. The upper sphere seems to be occupied ex-

~
l
I
the primary concern of the artists, their directional symbolism being
explicit only in two cases in the Kailiisa temple (Ellorii IG), which
belongs to a later period.
clusively by the supposed directional guardians. On the left (proper
right) side of NateSa, a homogenous group of four deities is de-
picted, while the right side shows five strangely scattered figures.
The following identifications are proposed herel63:

1 Cave 14: The directional guardians on the left (proper right) side of
the NateSa image (C. SIVARAMAMURTI 1974: 173, fig. 10) resemble
those at Elephants. The Na!eia panel is situated in the centre of the
five panels on the southern wall of the squarish cave facing the
south-west. Four figures occupy a spandrel-like section which is
demarcated by the handled object (lance?) held by the central deity.
left side: I
2
3
4
Vehicle
man-like
ramlgoat
ram
elephant
Attribute
lotus?
lotus
lotus
thunderbolt?
Deity
Nairgta
Agni or Kubera
Kubera or Agni
Indra
l The first figure in the upper row, situated right in the corner, is right side: 5 (lotus) bowl? Soma (?)
shown frontally, without any vehicle, holding its hands folded in 6 i~iakara noose? Varuna
lI arijalinzuclrSi. To its right is Indra on his elephant, and besides him 7 deer standard Viiyu

i
l
Varuna on his rilakara, the head of which is shown in profile. Below
this group of three one more deity, who must be Kubera, is depict-
ed in half-profile, apparently on a ram. Having identified three of
below:
8
9 -
buffalo staff (short)
staff (long)
Yama
Da~~dapiini?

l the four figures on the basis of their vehicles, the first figure in the
l corner could be identified as Yama, because he is the only deity left
~ to complete the guardians of the four directions. On the right (prop-
er left) side of the dancing Lord, only Brahmii and Visnu occupy the
celestial region.
Iconography: It is impossible to recognize more than the coni-
cal crowns worn by all the deities except Indra, and the lively depic-
tion of the vehicles. Regarding the representations of loka1x7las, two
differences between Elephanta and Ellorii 14 can be observed: at
Ellorii, the guardians appear irrespective of their canonical direc-
tion. The complete group of four has been depicted, aesthetically
counterbalancing the attendant gods Brahmii and Visnu, which are
here both situated on the other side.
Cave 21: Here, a similar stage in the developnlent of the lolcapiila
group seems to be reflected. The eastern wall of the southern annex
bears a well-known NateSa panel of sublime expression. On his
right (proper left) side, amidst clouds, four deities form a phalanx,
all of them shown in half-profile (fig. 22). From the left, there are:
Brahmii on his goose (har?zsn), Visnu on Garuda, Indra on his
elephant Airiivata, and, what seenls to be Kubera, on a ram. The fur
of the ram is signified by a deep hatched decoration of the body. An
identification of the rider as Agni is somewhat less probable, as a
goat has only straight hair. The depiction of Indra and Kubera may
reveal a conceptual parallel with Elephanta (fig. 20), where appar-
ently the sarne group of regents are depicted on panel 4 showing the
sarne theme in the centre. The directional guardians are clearly por-
trayed as subordinate deities, as they hold their right hands raised
high in salutation near the brim of their crowns. There seems to be
no conceptual connection of the clikl>rilaswith the four flying celes-
tials on the left side, which include Sfirya, the Sun God, and Candra,
the Moon God.
hi^ panel provides only few iconograp~licdetails. T I crowns~ ~
of both Indra and (?) are as and broad as those of
other deities of the celestial region. They also War similar mna-
ments. The curly hairstyle closely resembles that of Indra and
Surprisingly, Iiiina is missing among the clik,~Silas,while the likely
occurrence of both Kubera and Soma (5) recalls the set at Biidiimi
3 (fig. 10). The presence of what seems to be the rarely associated
deity Dandapii~!i (9; see section 2.5.10) among the lokaj>iilas is
particularly striking, however. If one doubts the identification of
its attribute as a staff, rather regarding it as a string instrument, this
can be easily conlpared with the one held by the musician on the
extreme left below. There, the sound-box at the upper end of the
instrument is clearly visible. Apart from Ellorii 29, images of the
guardian of the site Dandapiini are not found in the Deccan (cf. sec-
tion 2.5.10).
In conlparison to the eight deity scheme at Biidiimi (fig. 10),
some substantial differences can be noted. Both Brahmi (south-
east) and ~ i v (south-west),
a who are not directional guardians, have
been omitted. Agni (2 or 3) and Nairgta (1) have been included in
their place, as they represent those directions which the former
occupy at BBdimi. The horse-rider on the north-east seems to have
been replaced by Viyu (7), while the "double representation" of
Kubera (2 or 3) and Soma (? 5), both representing the northenl
direction, appears to persist.
The second relief showing lokaj~filasis situated across the
rita!t~apa, at the south-east corner (facing west). It depicts the
wedding of ~ i v aand Piirvati and is compact and well-composed
(VARMA1988: figs. 144-148). The directional guardians share the
celestial sphere with other beings, mainly pairs of divine nlusicians
(gnndlzarvas). Their arrangement is as follows:
162 For the g~ouudplan see W.M. SPINK 1983: pl. 25. Tlic givcn datc oTuc. 465-
475 be a fol ..c. 565-575 A,D,= (cf. 253: wdlsruption of tile
cave's pauoliage late in tlie sixtli centuryn).
An attempt at ide~it~fy~~ig the cltlpcilo\ has bce~i~iiadeby VAICLIA (1988: 31 I).
However, lie i~ico~rcctly describes I as Kubera, 2 a\ iLi~la,and 5 a\ Bralinii I)
Kube~a'shuman veliiclc 1s uncommon in southern Ind~a,and he quite clea~lyappcals
on one of the two gontfram veli~cles;2) tlic vehicle IS o goatfram and not a bull, 3) tllc~e
arc no ~ndicat~ons q The last figure (9) har
that t h ~ \is Bralimi except for the l o t ~ ~scat.
not been ment~onedby him.

27

l
 Vehicle Attribute Deity
left side above: l * buffalo staff Yama
below: 2* antelope flag Viiyu
3" goat rosary (?) Agni
4:': ram - ? - Kubera
right side (2nd row): 5Gelephant thunderbolt Indra
6" riznknrn noose Varuna
Iconography: Some significant differences among the figures
2 * 4 * of the guardians have been detected by M.A. DHAKY
(1988:
441). He observes:
The Deccani Dikpllas [ l g 5*,
, 6*] are in higher relief, wear broad, elab-
orate crowns, each figure looking as though only 'symbolically seated',
more on a throne than over the mount, with legs widely spread out. Their
right hand is raised straight up to the lip-level, whether just wielding an
attribute or si~nultaneouslyhinting at tlie gesture of praise in the north-
ern fashion. Their serene godly faces are squarishly round. The three
Dikpilas in lower series on the left [2 "-4*],on the other hand, exhibit
very different mannerisms. Their legs travel down, tightly clasping as
they do the mount's body, as is done in real life. Their left hand is raised
in the gesture of praise with spread out palm in tlie typically southern
fashion; their faceted, or whorled, somewhat narrow and tapering coni-
cal crowns likewise are Calukya in mode and mould, sharply contrast-
ing with the broad and elaborately bejewelled crowns of their opposite
companions. While the rest of the figures are rendered in very high
relief, in some instances seeming almost freed from the rock-matrix,
these three Dikpilas are in medium relief and seem rock-bound in the
typical southern fashion. Ethnically, too, they diverse widely from the
general type seen here with the majority of the minor figures. For they
clearly look early Karnitan.
DHAKY'S arguments are absolutely convincing. In looking for simi-
lar traits on the NateSa panel (fig. 21), it is interesting to find both
types largely merged, e.g., the riding posture is rendered realistical-
ly in most cases, but, in the salutation gesture, the palm is not shown
as separated from the thumb, although - this gesture seems to be
carried out in the lower, southern fashionl6" Considering this dis-
crepancy, it may be suggested that the Deccani sculptor has used
sketches of the "southern" type or has worked on the basis of
southern models.
AInmpur, K~inziirnBrnlzrnii temple
AIampur in Andhra Pradesh (cf. section 2.3.1.1), situated on the
river Tungabhadrii, was the eastern outpost of the Western Calukyan
dynasty. It must be regarded as a important discovery that
BUCIIANAN (1985: 122) has noticed small relief images of Yarna
(fig. 23) on the south and Kubera (?)l65 (fig. 24) on the north close
to the top of the tower (northern type iiklznrn) of the Kumiira
Brahmii temple. The images are framed by gavc7k;n arches, which
crown the top of the tower's middle portion (nzarllzynlntii). Accord-
ing to her, the construction of this temple took place during the reign
of PulakeSin 11, probably in the period 61 8-630 ~0.166.The lack of
a Suknniisn (antefix above the roof of the vestibule) is a clear sign
of its antiquity.
The figures of Yama and Kubera have counterparts on the east
and west side of the tower. However, these are not directional
guardians. It should also be noted that images rarely occur on tem-
ple towers of the northern type. BUCHANAN (ibid.) gives the follow-
ing, accurate description of this situation:
The Nafeia on the east side of the Kumira Brahma temple tower is
found within a multiple nZsiknltrefoil frame. The god is only four-
armed, but the pose of his legs and the front left hand thrown across the
chest in gnjn-hnsm recall the Nafeia of Cave 1 at Bidinii. In a compa-
rable frame and position atop the tower's south side is Yama, dikpZla of
the south, riding on his buffalo; on tlie nortli side is a seated male fig-
urc. S ~ n c ethree later Calukya temples at Alampur feature an image of
seated Kubera on the north exterior, this figure probably represents the
same deity, dikpZln of the north. The comparable trefoil on the west side
contains an unidentified sariinbknrign standing figure; it secms to hold
[its] right hand at its waist, to grasp a staff-handled attribute (axe?) on
the left, and to be flanked by two crouching creatures. Certainly it does
not appear to be Varuna, clikpaln of tlie west, but even its sex is uiicer-
tain. NapLa's position at tlie Kumrira Brahma temple was not repcated
in Calukya art; later he typically appeared on a SrrkanZra at tlie base,
rather than at the apex, of a s'ikharcl's front side, and without figures of
comparable importance on the north, south, or rear.
Among the few iconographic details which can be recognized are
the prominent, broad crowns of the deities. Those of NateSa and
Kubera taper slightly towards the top, while those of Yama and the
western image broaden at the top. The heads are also relatively
large, which may be due to considerations of perspective and the
angle from which thc figures would have to be viewed.
The unique ensemble of NafeSa, Yama, Kubera and the western
- most probably female - figure is open to different interpretations.
However, it seems logical that ~ i v as a the presiding god and his
spouse should together occupy the main axis of the temple's orien-
tation. Yama and Kubera just serve to symbolize the guardianship
over the four main directions, two of which they represent. Icono-
graphically, it is important to note that the image of Yama, squat-
ting above his couchant vehicle, has a probably contemporaneous
parallel at Bhubaneswar, on the ParaSurii~neSvaratemple (fig. 42).
The importance of the Kumiira Brahmii evidence can hardly be
overestimated, as it shows the first appearance of the lokal~iilason
the exterior of a temple in this part of India (contemporanous with
MundeSvariIBihar in the north) and because it provides the earliest
example of directional deities on a temple tower.
2.2.2 Central India
In Central India there are two examples of lokn/~iilnrepresentation
from the Gupta period. The earlier evidence167 (c. 400-410 A.D.)
from Pawiiya includes the depiction of a group of three or four
lokapiilas witnessing the mythological "Churning of the Ocean" (in
Sanskrit mostly referred to as anz~tnri~mztlznrzn- churning for the
elixir of immortality). At Bhtimarii, only three single representa-
tions of directional guardians (c. 520-530 A.D.) have survived. Due
to this lack of evidence, no conclusions can be drawn as to whether
the directional guardians were conceived as a complete group at
this site.
l" An exceptionally high hand position occurs in the casc of Soma (?). But cven
here, the arm is hcld at an angle from thc body, as is the casc with nlost of thc other
depicted deities. Moreover, as has already bcen argued, he is allnost a rcplicu of the
guardian of thc north in the eight deity schcmc, B i d h i 3 (fig. IQ),Iiowevcr with the
addition of a lotus seat and without thc protome (forcpari)of a lion.
1" The possibility rc~nainsthat thc figure represents Soma, considering the prob-
able prcscnce of this deity on thc northern sidc of thc cenlral medallion at BBdimi
(fig. 10).
l(fi BUCIIANAN 1985: 119. This opinion had already bccn expressed by 0.
DIVAKARAN (1971: 59; 73). B.R. P R A ~ A(1988:
D pl. 635) dates it towards the cnd of
his reign, at c. 640, and C.E. RADCLII:I:E (1981: 567) later than this, in the pcriod of
c. 654-660 A.D.
The datcs are citcd aficr J.G. WLLIAMS 1982.
 Although both these modes of representation had no direct suc-
cessors, they offer extremely valuable evidence for the early iconog-
raphy, context and probable placement of the dikprllns.
At PawiiyYM.P., the ancient town Padmiivati, a pyramid-shaped
heap of debris is what remains of a brick structure of the Gupta
periodl6" One of the most important relics found here was a fiag-
ment of a gateway (torn!zn) lintel, on both sides of which mytho-
logical scenes are depicted in relief. The better preserved side
(WILLIAMS 1982: pl. 50) is now displayed in the Archaeological
Museum Gwalior, where the reverse side remains al~iiostinacces-
sible (fig. 9). The lintel is broken just above the outstretched legs of
the deity Trivikrama (obverse) at its centre. The adjoining scene on
the left, which is horizontally divided in two registers, also belongs
to the Viimana-Trivikrama episode. The reverse, which shows at
least two of the lokcn/xilns, was centred around the kirrizn (tortoise)
nvntiirn of the god V i s ~ y Unfortunately,
. the portion which once
showed this is now lost, and even the central figure in the upper
register has been almost completely effaced. From the left of the
lower register, four deities are extant. They are engaged in pulling a
rope, which is actually the hind part of the serpent Viisuki or ~ e ~ a l 6 9 . the plot are, however, quite arbitrary.
Below the rope, a tiny female figure carrying a bottle and a four-
legged ani~nalcan be seen proceeding towards the right. They rep-
resent some of the treasures gained through the "Churning of the
Ocean". The animal must be either Uccail!Sravas, a divine horse, or Bhfimari, Dt. Satna/M.P., is one of the few sites of partially pre-
0 , the woman could be the goddess Sri, served Gupta temples. Nowadays, only the plain walls of the cella,
Surabhi, a divine ~ 0 ~ 1 7while
a nymph or the goddess Viiruni. The latter is the most likely identi-
fication, as the vessel she carries is probably meant to contain surii
(liquor). To the right of the last of the four gods in action follows a
god sitting on a low throne171 wit11 his legs splayed. His right liand study are three reliefs framed by gaviikp ("cow's eye") arches
is held in the protection-bestowing gesture nbhayarlzlidrrl, his left
hand rests on his thigh and holds a noose which emerges from a
stiff handle. Judging from the noose, this deity must be V a r u p The
next figure is rather enigmatic, as only the upper part of its body is
visible between the two throning deities. He wears an elaborate
crown. The third figure, whose throne is not in line with the first one,
sits in the same position as Varuna and holds a mighty staff in its
left hand. It seems niost likcly that Yarna is represented here. This
would make it the earliest portrayal of the God of Death. Above
him are two designs in low relief which might represent heavenly
bodies. The thmnc of the fourth figure is partly hidden behind
Yama's seat. This figure is seated in the same fashion as the others, Surablii does liot appear in tlic epic vcrsions, but shc is mentioncd in the Vi!!rrr/~~rrci!tcc
but the arms are held differently. His left hand at shoulder level
holds an object that seems to be a rather large pot, while the right
hand is kept in front of the breast as if pointing towards the attrib-
ule. A possible identification for this figure is Kubera, although his
attitude WOLIICI be quite unl~sual(cf. section 2.5.7). Alternatively, the
figurc could be meant to represent Dhanvantari, carrier of thc pot of
nnzytn (elixir of immortality)172.
In the upper register of the panel, supposedly for~ilinga part of
the same scene, a typical Gupta representation of the divine bird
Garuda can be seen on the extrenie right. To its left, four more thron-
ing figures are lined up before the central, fmgmented one. The first
two of these are haloed and equipped with prominent ~rl?nvlt(zs (Mbh version, ibid.: 9).
(briihmanical cords worn across the trunk), wJ~ilethe one placed
slightly off-centre appears to be characterized by a snake-hood.
Tlie general concept behind the "Chunling of the Ocean" is to
demonstrate the common effort of the gods and the demons. Spec- cases, Varuna is the first god to be approached for lielp (MtP 249.14).
tators are generally not depicted. The upper row of celestials could
represent an assembly of gods preceeding the churning process, an
exhibition of the treasures obtained through the common effort173or
a group of subordinate deities.
Although, to our knowledge, the directional guardians are not
mentioned in any of the descriptions of the "Churning of the
Ocean", their presence on the relief may have served to enhance the
cosmic significance of the churning act. Tlie central mountain which
is used ,?s a churning stick is established as the axis riilrlitfi during
the processl7" A reference to the four directions in space may
therefore be regarded as appropriate. Although presented in a
crowded manner, the group of four loknpiilas is most probably
sliown here. It is unlikely that more members had been been depict-
cd on the missing left side of the relief, since it must have repre-
sented the realm of the nscrrn demons, the antagonistic party in the
cosmic act.
Besides his loknl>iilnfunction, Varu~!a, as lord of the waters, has
a particular connection with the "Churning of the Ocean"l7s. There
is explicit mention of Yama in the Mntsynplirrl!~n (250.46), where
he is described as being dismayed by the emerging poison Kiilaki~ta.
Indra has an important role in the myth, as his elephant Airiivata is
produced from the ocean, and it is an original aspect of the myth that
it confirrns Indra's supremacy over the world (RUPING 1970: 8).
These particular connections of the individual lokn,~?iiln deities with
with an impressively decorated doorway however, and the renlains
of a pillared hall (1iin!1dn17n), a modest superstructure and the niould-
ings of the plinth survivel76. The images relevant to the present
depicting Yama (fig. 25), Kubera (figs. 26, 27), and Dandapiini
(fig. 28). These three images resemble those from Pawiiyii in that
M M .GARDE1924-25; WII.LIAMS1982: 52.
l" The diffcrelit versio~isincluded in thc epics Mnlrciblzcirata and RZrrrci)a!rcr and
sotiic of the later tlcvclopments of the crn~ytcrrrrtrr~tI~a,,o story liave bcen outlined by
K. ROI'ING 1970.
I7O The short ncck points to die latter idc~llification.It is i~ltercstiligto note (hat
and in the S!ysiklzcr!~~/cr of thc P(rcInlcr/)crrci!zn as 111~first of tlic I ~ C B S L I ~ C and
S , Vlrupi as
tlie sccond (ROI'ING 1970: 31).
17' For the simplc thronc scc the rc~iiarksbelow (under B111i1ricrr.Z).
172 This ~iiotif,which is often dcpicted in sculpture, appears in the MalrcihhBrrrrcr
(1.16.37) and, eg.in the M~~/.g~npcl,d!cn (25 1.6).
'73 This is tlie cnsc wit11 the upper register on the much later pancl from Badoh
(K.S. DI~SAI 1973: fig. 57). Curiously, Cakrapurup (personification of the whccl)
sccms to be inclutlctl, as the sccond figure from thc left. Cakrapuruy is also thc most
likcly idcntification for the figure to tlie left of Garuda at PawByi. However, he is not
rcgaldcd as a product of the churning in any of the versions of !lie myth known to U$.
Garuda also does not bclong to the c~ncrgingtrcilsurcs.
Tllc "uprooti~ig"oftlie mountain is always included in tlie story ofnrrr~crrr~crrr-
th~rrla(ROPING1970: 6, 8, 9, etc.). It is probably another rcfercncc to tlie spatial sym-
bolis~lithat Ananta, guardian of tlic nadir, is givcn the rolc of liftilig mount Mandarn
175 Tllc ocea11(also referred to as ~'crrrr!~Z/n).a- seat of Varu~y) is the sccne of thc
thcmc in earlier versions of the myth (Hari~~aq~Sa and Rri~rrci)~cr!~fl versions, ROI)ING
1970: 6-8). In liltcr vcrsions, it is replaccd by the milk oceali (ksirorla), but cvcn in these
I 7 V e e R.D. BANERJI1933 (plan: pl. XVII) for an early docu~nentation, ~und
WILLIAMS 1982: 117-122 (plan: p. 119) and DCVA1988: 4 0 4 4 for a rccent assessment.
WILLIAMS ilssigns it to C. 520-530, while DIZVA prcfers a date close to 500 A.D.
 they are seated on low, simple thrones, of the type which often
occurs with tutelary deitiesl77.
Some misconceptions about the identification of these deities
still persist, particularly regarding Yamat78 and DandapCnil79.
These have only arisen as a result of the unique concept which is
expressed here at BhumarC. To understand this concept, it is rele-
vant to discern the different categories of gnv8k;a images found at
the site. The following list has been compiled by WILLIAMS (1982:
120), with nos. 4 and 8 being emended here, as well as highlighting:
Large180 I KumBra, holding spear, seated on peacock
2 Mahisamardini, four-armed, grasping the buf-
falo's tail
3 Silrya, arms raised to hold lotuses, with Dandin
and Pirigala
4 Daqdapsni, seated on a throne with legs pen-
dant, holding a staff across his knees
Medium size 5 GaneSa, four-armed with axe and rosary
6 BrahmC, four-headed and four-armed, seated
with ascetic's band, with lotus and curved stick
7 Kobera, corpulent, legs pendant, with three
pots, holding a lotus [proper r.]
8 Yama, two-armed with club (1.) and fruit (r.),
wearing a cylindrical crown
Small 9 Siva dancing, six-armed
10 Dancing figure, possibly a ga!zn [gnome, associ-
ated with Siva]
Regarding the original placement of the framed reliefs on the tein-
ple, the author remarks (WILLIAMS 1982: 118):
... the superstructure of the shrine proper can be reconstructed with more
security on the basis of the similar temple at Marhis. Bhumara's long
friezes of gcr!irrs alternating with ornament would seem to have formed
Dl~ir~iiisor upper stories resting on the flat roof. It is possible that ccrrz-
d,-rr.i~IZs[= gavZksc~frames] decorated each side, for these have been
found in three sizes, corresponding to the three types of roof frieze.
On this in fn. 57: Since none of the roof friezes indicates a break in the
design, it is possible that the ccrrzclrcriZ1iswere attached to the top of tlie
lower level and overlapped the frieze.
This interpretation seems reasonable, as it explains the different
types of ornamented slabs and gnv8ksns. A piece decorated with
lotuses and framed by gnvh1c;crs was probably a corner element of
the original superstructure (DEVA1988: pl. 75, cf. BANERJI1924:
p]. VIb).
Yama (fig. 25; 8) is shown broad-shouldered and with mighty arms.
Short locks of hair stick out from underneath his tall cylindrical
crown. He wears the usual ornaments and an illxzvita across his
breast, which has a chain-like appearance. A large halo and his
serene face account for a highly dignified expression. A part of his
lower garment hangs down from the throne, which, together with
the high crown and the mighty staff (cI~!I</~)I~I,strengthens the ver-
tical aspect of the composition. The emphasis of the vertical axis is
enhanced by the symmetrical arrangement of two relatively tall
female flywhisk bearers. Stout lathe-turned legs support Yama's
throne, which slopes to the right in the same manner as those of the
two other deities under discussion. The legs of the throne continue
above the seat level, forming knobs at its corners. The conspicuous
sloping of the throne, which violates the symmetry of the compo-
sition, may be explained by the intention of the artist to create a per-
spective view.
Kubera (figs. 23,24; 7) at BhGmarC is corpulent, but not pot-
bellied, as he is often depicted. His head with long, curly hair is
adorned by a diadem. His lips are fleshy, his expression calm. His
ornaments are similar to those of Yama, with the exception of the
vnlnyas (upper arm ornaments), which are spiral-shaped. His right
hand holds the gracef~~lly curved thick stem of a lotus bud. While
the vertical axis dominates in the depiction of Yama, there is here
particular emphasis on the horizontal axis, which is enhanced by
two globular pots placed on the throne on both sides of the god.
Even the throne and its legs are especially broad. A halo is visible
behind Kubera's head. A third pot appears to be squeezed in under-
neath the throne, which again slopes to the right. The left hand, the
left leg and the portion above the head have all been destroyed.
Daqdapsqi's (fig. 28; 4) head, from which rows of curls reach
down to his shoulders, is adorned by a simple headband. His fea-
tures and those of his two male attendants have been blurred by
water erosion. Of the ornaments they wear, only the necklace and a
cross-ribbon are still discernible. The throne slopes in the same
manner as the other two described previously. This feature is com-
plemented here by a silnilar sloping of the attribute, a staff which is
held close to his knees. Both hands clasp it on either side froin
above. Dan?apii~!i's attendants have wig-like hairstyles and stand in
different attitudes, the one on the right holding a bowl in front of his
chest. The heights of the attendants actually echo the sloping of the
throne and the staff, which once more reinforces the hypothesis that
a perspective was intended to be shown.
Without any clue as to the original position of the framed images, it
is impossible even to propose a hypothetic distribution of the
images. One would naturally expect Kubera to have been located on
the northern side and Yama on the southern side of the temple. HOW-
ever, the examples from MundeSvari Hill and from Bhubaneswar
(see below) indicate that matters were not always that predictable
during the period under discussion. At MundeSvari, Yama and
Kubera appear on two sides of a cornerstone (figs. 40, 41), while
177 VONMIITIIRWALI~NI:R (1989: 373), in commenting on this type of thronc with
"baluster-like Sect", refers to similar ones upon which tlic early 5th century 111r71,bA god-
desses at Besnagar are seatcd (CLJ.C. HAIILE1986: ill. 73). In this connection, she dis-
cusses a figure, callcd a "yokyr" by an inscription, supposedly from KahkBli TilB (K.D.
BAJI'AI1947: from Malioli!) of tlie KusB~y-Guptatransitional period, which, in licr
opinion, squats in front of his seat. There are four morc interesting a~lalogucsfor tlcitics
squatting on a thronc, two of which certainly represent Kubera, sincc they liold a purse
in the propcr left liand (fig. 321; F.M. ASIIER1980: pl. 187, from Rolloi). The olhcr two
deitics hold n fruit in tlicir right hand, wliilc thc lelt hand rests on the thigh (PAL1978:
fig. 12, from Bhoblincswar) or holds :I tlamagcd, indistinct objccl (Sotheby's, New
York, catalogue, 17.06.93, lot 113). In our opinion, thcsc four cxatnples dale from tlie
6th or early 7th century. With tlic exception of the onc illitst~~atctl licrc (fig. 321,
hcltnet-likc liairstylc), the hair of the deities is curly end rcaclics down to the sliouldcrs.
If any crowns or diadems are tlcpictcd at all, they arc of a rather modcst type.
178 Altlio~~gh tliis i~ilagehas ofie~ibccn correctly idcntificd as Yania (e.g., MEIS-
I'KRIDIIAKY/D~:VA 1988: pl. 74). WILI.IAMS (1982: 120, no. 8) calls it "Vi~gu".This was
probably because slic intendcd to identify no. 4 as Ynma. LOBO(1992: cat. 121) leaves
thc question of its identity unresolved.
17" For the reccnt definition of Dat!?apBqi by tiiysclf and others see section 2.5.10.
So far, "lodra (?)", inlroduccd by BANBRJI (1924: pl. XIVc), and the abovc-mentioned
itlentificalion as Ya~iiawere the only attempts to interpret this ilnage.
1X" One of tlie largc go~~ci/(,yrs, now in ii highly fragmented state (BANEIIII 1924:
pl. Xlllc), has been oniittcd in the lisl. O~ilyn crouching bull is prcservcd in the lower
scclion.
wliich is prescribed for Yanin in iconographic texts, has liere
' H I The staff((/r!~~/~r),
assutncd the appearance o f a thin club (grrrlci). It is faceted, tapers bclow and is topped
by an onion-sliapcd clemcnt. A similar feature can be found at Bhubaneswar, in the
ParaSurameSvara temple, where tlic identity of Y i ~ ~ nisadetcr~iiincdby thc prescnce of
his buffalo 1~r7/1or1(1
(fig. 42, cenlrc).
DangapSni is only represented by a larger image which belongs to
different set of figures (fig. 35). At Bhubaneswar, DandapSni is
shown in a line together with an uncertain number of directional
guardians on the south side of the vestibule of the ParaSurSmeSvara
temple (fig. 42).
2.2.3 Eastern India
Some of the ancient roots of the canonical a~farlikj~iila group can be
traced in Eastern India. The evidence for two differently con~posed
sets of directional guardians froin Mu~>deSvari Hill in Bihar is par-
ticularly striking and is discussed here for the first time.
The te~nplebase seems to have been orten used for the depiction
of clikl~iilnsin this part of the country. Their introduction on the
"frieze" (jarighii) portion of the wall, which represents an important
innovation, also probably occurred for the first time - as far can be
inferred from the remains - at MundeSvari Hill.
3 Viiyu Panel 19 (fig. 31), east (features eroded): the figure is
shown in the "flying" posture with the proper right leg raised
besides the trunk; he wears a helmet-like hairdo, the proper
left hand holding a rope-like cloth which encirles the head;
the right hand is extended below, obviously showing in the
gesture of bestowing boons (vnmclniizrrrlrii).
4 Indra? Panels 22-23 (fig. 32), east: he is seated in an awk-
ward position almost like squatting, has a wig-like hairdo,
large earrings, holds a thunderbolt184 (only hall visible) or
bud (I.); his right hand is resting 011 the lower leg; Panel 22 to
the right of no. 23 is a half-panel (proper right half), but
recessed from it, showing the head of an elephant which pro-
ceeds to the (proper) right.
5 I<ubcra Panel 200 (fig. 33), south(-east): hc is half-squat-
tiilg, pot-bellied, has large earrings and a hdra, wears a hel-
met-like hairdo with a central diadem, holds a purse (1.) near
the left ankle and a fruit near the lower leg on the right; two
globular pots are shown floating in the space on both sides,
just below the angles of the framing horseshoe arch.

The second image of Kubera (fig. 33; 5) is relevant insofar as Pan-

Although this well-known site in Pat113 District does not provide el 200 interestingly shows an arrangement of pots which bears some
much indisputable evidence on lokapiila representation, some resemblance with the one at BhCmad (fig. 26). It is quite u~llikely
aspects revealed by the Brlhmanical Stone teinple no. 2 are relevant that Kubera was regarded as a directional guardian at NrTlandS.
to the present study. There are 17 Inore images of corpulent males of the "Kubera" or
At this site, which is situated near the remains of Buddhist "yak~n"type (Panels 27, 35, 38, 42,48,49, 80, 88, 100, 109, 124,
monasteries, the roughly square plinth of a teinple now lost has been 129, 140, 183, 195, 198,205), so that the role of Kubera as guardian
well preserved. A remarkable series of 220 sunken stone panelsl82, of the north is not clear.
separated by pilasters, serves to decorate the plinth. Their average At NSlandii, Stone temple no. 2, a few members of the rlikc~~iila
size is c. 30 X 30 cm. On stylistic grounds they have been assigned group are depicted, however in an irregular sequence. Regarding the
to the 7th century, "c.650 A.D." being the date recently favoured by positions, only Agni (east, south of the entrance) and Indra (east,
Krishna DEVA(1988: 112). There is a wide range of subjects depict- north of the entrance) have been placed according to their directions,
ed, summarily characterized by DEVA(ibid.: l l l ) thus: with the identity of the latter being somewhat uncertain. On the
extreme ends of the eastern side of the teinple are images of the god
The themes depicted in the niches comprise gods and goddess- ~ i v aIt. is possible that the one at the north-eastern corner represents
es, vidysdharas, gandharvas, a kinnara and suparips [celestial IS5na. However, this cannot be conclusively proven.
beings], scenes from the Riimiiya!~n,~niscellaneoussecular sub-
jects like animal fables, cock-fight, musicians and dancers, Mci!7des'vari Hill (Rdnzga<llz)
acrobats, ascetics, teacher and disciple, mother and child, and
amatory couples. Ainong gods and goddesses (represented with The MundeSvari temple185 is a well-known ruined structure in the
two-arms and showing simple iconographic features) are ~ i v a ,Rohtas district (formerly Shahabad) of western Bihar, which is sur-
Gajalakgni, KubEra, Balariima, KSrttikEya, Agni, and probably rounded by a large number of architectural and sculptural remains,
HSriti. now partly removed to the Patna Museum. The site must have been
an important pilgri~nagecentre, as it is situated on a hill (c. 180 m
Those panels, which belong to an incompletel" set of lokn~~ilas, high) the rocky ascent of which is studded with pilgrinl-records.
will now be described. The panels are refeued to by the numbers Two parts of a loose foundation inscription, dated in the year 30 of
introduced by DEVA(1950: 202-212), which start at the (south-)east an unspecified era, have been recovered from the ruins. It records:
corner and proceed anticlockwise:

I Agni Pancl 7 (fig. 29), east, southern portion: he is half-

squatting, pot-bellied, with a coiffure resembling flames, has 1x2 A descsiption oftlie panels is given in K. DEVAIV.S.AGI<AWAI.A 1950; cf. also
flames behind head and shoulders, earrings, a simple neck- DEVA1980 ancl 1988: 110-1 12.
lace (hiirn), wears a broad, ornamented brShma~!ical cord 18.' Although tbe number of directional guardians rcmnincd inco~nplclclicrc and

(~rl~czvitn),
holding a bottle in his proper left hand and a rosary
with the right hand in cirzri~uclri(the thumb touches the index
finger) in the corner below.
2 Kubera Panel 14 (fig. 30), east: he is half-squatting, pot-bel-
at other carly sitcs, it should not bc i~nplicdthat tllc full o.gac/ik~)rilrsct was u ~ ~ k ~ i o w n
during tlic period of construction. Lilernry cvide~lcc113s bccn reported from thc 6th ccn-
rury A.D.On thc particular fcatures of tli$cilci sets i~icorporalcdin friczcs scc section
2.4.1.3.
184 It liiay be rclcvant to note that the MundcSvari Indra carrics a similar "llcxi-

lied, with hair reaching to the shoulders (head damagecl), with
a purse held near the left thigh (I.); his right arm is supported
by his right knee, the right hand probably holding a rosary;
the god is on the left side separated by a plant fro111a couchant
sheep, described as a bull by DEVA(1950: 200, 203).
blc" tliunderbolt. Thcrc il is co~npletelyshown liowcver. The matter cnnnot be finally
resolvcd at NSlandS, as the typical crown of Indra is also missing.
1x5 Thcrc arc a ~iu~nbcr of rccent publications, some of which contain summaries
of earlier contributions: WII.I.IAMS1982: 166-168; G.S. GAI1985; DEVA1085; 1988:
118-121; P.R. SRINIVASAN 1991. A stylistic analysis of the sculptures has bccn ~nade
by ASIIER 1980: 38-42.
 ... there was a temple of Visnu or Stiryaniiriiyana (NrTrrTya!7c~-devnk~11a)
called Mn~7~krledvnr~svrT11zirzwhich wa5 an old establishment [...]. T o
that temple was appended a Slva temple (with monayter~?) called
Virlite.fvarc~-riin!hn to wli~ch the foundation inscription specifically
refers. (DEVA 1985: 126)
It is difficult to connect these facts with the extant structures:
The hilltop has not space enough to accolnmodate two temples, unless
it is assumed that the little shrine measuring about 2 m 43 c m square
with a Sivalinga, immediately to the southwest of the main temple was
the ViniteSvara temple. But this shrine, besides being insignificant, is
definitely a later votive shrine (ibirl.).
Therefore, questions arise concerning the era of the inscription, its
palaeographic analysis, and finally, the date of the temple remains.
The era has generally been regarded as the Harsa era starting in 606
A.D. A recent palaeographical analysis, however, does not corrobo-
rate the resulting date of 636 A.D., but points to a 6th century datel86.
The sculpture obtained from the site has been so far dated to the 7th
century. The stylistic differences have been chronologically inter-
preted by ASHER(1980: 4 0 4 2 ) , while DEVA(1988: 120) prefers to
state "several idioms and levels of workmanship [...l all stylistical-
ly of the early seventh century A.D."
The octagonal shape is an exceptional feature of the ground plan
of the MundeSvari temple (ill. 2). The four doors are situated on the
-narrower - walls facing the cardinal directionsl87, while the walls
facing the intermediate directions each bear a large, central niche
flanked by two smaller niches. The niches are now empty. The walls
are extremely massive and were apparently designed to support a
towering Siklzarn. Regarding the first set of directional guardians, it
most probably occupied the eight lateral niches (clevakn~flzns),as
has been suggested by DEVA(1988: 120). Although their squarish
shape is hardly compatible with the oblong shape of the niches (the
width being adequate, however), two sculptures in the Patna Muse-
um most likely belong to this series:
1 Agni (fig. 34): seated on a low throne encircled by a mandorla
of flames, the (proper) right leg drawn towards the torso, the
left one pendant with the foot resting on the socle; he is wear-
ing a flattened hair-crown (jn[dnzukuta), a double-row neck-
lace and a shawl partly covering the corpulent torso, support-
ing an oblong vessel (with lid?) with the left hand resting on
the left knee and a rosary in the right one resting on the right
lower leg. The god exhibits clear-cut features with a solemn
expression.
2 DandapCni (fig. 35): seated on a throne like Agni, but with
both legs pendant, a halo behind the head, corpulent, with a
serene expression, wearing a headband, having a curly hair-
style, a double-row necklace, cross-ribbon, valnyns (upper
arm ornaments) and a long cllzoti covering the legs. He holds
a staff with both hands, apparently from below, horizontally
over his knees.
What is striking at first sight is the similarity of the second figure
with the Dandapini image from Bhtimari, common characteristics
being the open hair and the cross-ribbon (charzrzavira). Furthermore,
he is shown in a perfectly symmetrical manner. With Agni repre-
sented here, the novel situation arises that a loknpdla of an interme-
diate direction has been included. The occurrence of Dindapiini,
However, undoubtedly one of his counterparts, indicates that the
later canonical group of directional guardians had not yet been
formed. Unfortunately, there are no remains of the other - probably
six -members of the group.
The second set of loknl~dlasis represented on loose corner-
stones, which have remained almost unnoticed hitherto (ill. 2). The
arrangement in pairs on the adjoining faces of tlle stone blocks
reveals a very ullusual directional assignment. ~h~ ollly lilllitation
is the fact that their original orientation is unclear. It is also unclear
to which building they once belonged. Three cornerstones are
extant, showing SomaICandra (fig. 36) and SLirya (fig. 37), Agni
(fig. 38) and Indra (fig. 39), and Kubera (fig. 40) and Yama (fig. 41),
respectively. Although the fourth one is missing, there can be hard-
ly any doubt that it showed Varuna and Vgyu (cf. section 2.4.1.3).
The most plausible reconstruction of the directional arrangement is
given here (ill. 2). It is also probable that they had been employed
in the interior of a structure, which would mean that the orientation
of the depicted deities differed from thc reconstruction suggested
here. With the temple base remaining intact, the only probable posi-
tions for the second set, according to our present knowledge, could
have been the cornice or the superstructure of the temple or its now
lost hall.
The six guardians of the second set are carved in shallow nich-
es framcd by pilasters. They do not have thrones but cushions to sit
on and are shown in varying half-squatting poses like the Nilandi
images. Only Stirya (fig. 37) is depicted on a lotus throne on which
he is seated in the ~~pright lotus-posture O~rrrlrllZsann).His sym-
metrical posture stands out in contrast to the fluid body shapes of
the other guardians. The latter along with the serenity of the faces,
at first glance, recalls the spirit of the Gupta period. According to
ASHER'S(ibid.) arguments, however, the extremely sturdy figures
with delicately modelled features, are posterior to the boldly con-
ceived, vivid images, to which the first set belongs. The fact that the
folds of the cll7oti have been indicated by incised wavy linesIg8also
supports this interpretation. Thus, there may have been a chrono-
logical gap of a few decades between the first set (in niches) and the
second set of directional guardians18" The height of the corner-
stones is 30 cm.
1" Soma (fig. 36) has a coiffure resembling a knitted beret, and
a crescent behind his head; he wears earrings and a simple
necklace (shared by all the others). His legs are covered by
a pleated c/lzoti (common to all of them, except for Sfirya),
and he holds a waterpot covered by a lid in his left hand just
below the knee and a rosary in his right hand held at shoul-
der level.
2:VSrya (fig. 37) wears a cylindrical crown (relatively low as
compared to Indra), below which the hairline is visible. He
is seated in the lotus-posture on a lotus throne, holding the
curved stalks of l o t ~ ~flowers
s in both hands, the flowers
being almost covered by the palms.
IXWAI (1985: 130) proposes thc first half or middle of tllc 6th century, a date
basctl on inscriptions oftlie Maukhari dynasty. According to D e v (1985: ~ 126). the lal-
ter has to be assignctl to tllc later half of the 6th century A.D."
1x7 It is a notcwos~liyfact tint tlle sculptural progmnmes of thc doors arc clearly
diffcrcntiatcd: cast - Gang3 and Yamuna; soulh - Dandi and Piilgala; west - Scri~a
eIvcirci~~i1ers;north - fcmnlc dooskcepcss.
188 ASIIER (1980: 42) mentions this characteristic forthc group of standing images
(ibirl.: pls. 57-59) which hc regards as latcr, probably dating from the latc 7111 ccntory.
18') If this is true, it bccomcs cvcn mow likcly that the second set bclongcrl to tlic
pillarcd hall (~ii~!~~/cipc~). Tlie latter has bcen regardcd as postcrior to tlic octagon (ASII-
BR 1980: 39). as its con~tsi~ctio~i obscured some of tlic nichcs of tlic sanctum. Aslrel~
already suggcsted that the riiei!~~/cr~~ci was addcd "pcrhaps when thc latter [postcrior]
group of sculptures [...l was installed" (ibicl.).
 I * Solna and Dangapiini (3). To the left of Varuna's niche, a window inter-

1
rupts the row of seated images. Two figures and a small fragment of
(Varuna) 2.': Sfirya a third one are extant to the left of the window. Here, however, their
vehicles are absent. It cannot be decided with certainty if these three
deities were originally conceived to be a part of the lokap6la group.
A T.E. DONALDSON (1985: 55-56) in his co~nprehensivework on Oris-
san temples has argued that an alignment of the original eight
6* Yama 3:' Agni clikpiilas was disturbed by the later insertion of the window. The
width of the grilled window nearly corresponds to the breadth of two
I I
lokapdla niches. Furthermore, he suggests
-- that the southernmost
5* Kubera 4" Indra image on the western wall of the jngai~zolzaiza represents one of the
two relocated figures of the set (fig. 44, right).
Ill. 2: MundeSvari Hill, reconstructed arrangement of cornerstone
loknlxilns
7 Kubcru? I A
6 drlprilri? 5 Somd 4 Viyu? l wlndow 3 Va~una2 Yama 1 Ddnda-
34' Ag~ii(fig. 38) wears a hair-crown (~nfiimukufn) with boldly or ~ g n l ? I
I
P~IJI

carved strands, flames behind head and shoulders; he is I

corpulent, with the usual ornaments, a briihmanical cord

(Ilpavita) or shawl across the breast, supporting a vessel 111.3: Bhubaneswar, ParaSuriimeSvara temple, south-west corner of
(with lid?) with his left hand held above the knee and a jagai~~ohaila
rosary at shoulder level in his right hand.
4'"1ndra (fig. 39) wears a high cylindrical crown. He has a hor-
izontal eye on the forehead (?) alld holds a slightly curved DONALDSON'S version Seems to be quite convincing, but he disre-
thunderbolt ("ajra) in the left hand at shoulder level, his the pecllliar sequence and composition of the group. His iden-
right arm and right leg being damaged. tification of the deity on the extreme right (I) as Indra is definitely
5 * I(ubera (fig. 40): with a hair-crow11 or a helmet, the head incorrect1". However, the better preserved figures on the left side
slightly tilted to the (proper) left, he exhibits a paunch. He of the southern wall could be Solna or Agni (5)192and V S Y (4; ~ fig.
llolds an elongated rnoney-bag in his left halld at shoulder 43)'". Based on the facts described SO far, it is unlikely that an
level, alld a vessel with a slightly constricted nsfadil~hlaSet was originally depicted. It could have been a rudi-
in his
right resting on the knee. lnentary or an intentionally incomplete set, perhaps with Dandapii~!i
6" Yams (fig, 41) his hair arranged ill the raised, demonicreplacing one of the other guardians. There is also the possibility
fashion (ircl]lVnke,fa). He is and lIo]ds a staffthat the original concept was that of a Manu Lokapiila set, as neither
topped by a human head in the left halld and an indistinct Nairrta nor PSiina of the as[aclikl~rilasCan be identified, who would,
object (fruit or pot?) in the right, above the knee. in that case, have been replaced by Strya and Candra. Images of
these deities are nowadays located in the north-east section of the
The composition of this group is remarkable, as it represents the ear- ja~alllolzfllla.
liest known combination of directional guardians with the Sun and With the present sequence of images, no particular system of
cods, which is refe,.,.ed to here as u ~ a n ~u ~ kset ~(see s reference
l ~ ~to the directions can be discerned nor hypothesized. These
section 2.4.1.3). ~tis difficultto decide ifany directional symbo]isln are the particular features of the seven surviving figures or figure
was connected with this group. fragments:
It is quite certain that the first set did not include Si~ryaand
SomaICandra, because these deities belong to the well-defined I Da?'Sla~Hl!i (fig. 42, right): largely damaged, sitting with
group of Manu Lokapiilas, of which the staff-bearer DandapL~jiis pendant legs on a throne like Varuna's, with an elaborate
not a member. hairdo of cascading curls, holds a staff in both hands which
With Agni belonging to both sets of clikpiilns or diklx7la-related rest on his knees (right portion completely damaged).
deities, one can assume that a conscious experimenting with this Yams (fig- 42, middle): squatting on a reclining bllffalo, -
group was in process at Mundeivari. crown damaged -, wearing a small earring in his right ear and
a large stud in the left, two strands of beads around the neck,
BIz~~Oarzeswar key~irasand a short sacred thread similar to Varul!als, and a

The ParaSuriimeSvara temple at Bhubaneswar in Orissa, a remark-

able, well-preserved piece of early architecture, shows a row of T.E. DONALDSON 1985: 51; MITRA1988. 256. A d c s c ~ ~ p t ofthe
~ o n tcmple Sol-
male deities on the souther~lside of its vestibule, the so-called jaga- low' In both
l" 111that he scems to follow BANI:RJI (1924: pl. XIVc), who tentatively suggert-
llzoknlza'The lavishly decorated t e r n ~ llnost
e likely dates from the [llls Iden[lfica[lonfor [llc ilnage of D,lndnpjcl BhilllarP
first decades of the 7th century190. 192 Ncai the no~th(-east)collier of the jagrrmoha~in, besides a standing Surya,
Images of three deities, belonging to the group of directional CandrdSoma i~ show11111 very sl~lillarway (DONALDSON 1985: fig. 79). A g n ~who
, IS

guardians or associated it, can be easily ~h~~ are chaaclerlzed by the s.111ie attributes, at Bhubdneswar has always flames behind the
situated between the door and the window, in neighbouring niches body 1976' 'lgs' 9-10)'
The figu~ecarries a stall whlcli Inay have forined a part of a banner. In t h ~ s
pilasters and crowned pediments (Lldgan7ns) low part of India, however, and particularly at th19 early date, Vjyci should be expected to
relief. They are, from the left (ill. 3; fig. 42): V a r u ~ p(I), Yama (2), carry an lnflatcd scarf (CLsectloll 2.5.6).
 girdle made of metal (?) sheets with a square buckle (hardly it is inscribed. Its height of 27 cm conforms largely with the usual
visible in the case of Varuna), holding a fruit (?) in the right size of c. 30 cm of other early lol~al~(ila representations (NBlandii,
hand and a long faceted staff or sceptre in the left hand; the MungeSvari cornerstones, the New York Varu~p,fig. 177, and the
buffalo has a bell around its neck; its back is covered with a Mathurg iiana, fig. 178). Regarding its date, the 5th or 6th century
blanket. has been proposed. It appears to be a predecessor both in size and
3 Varuqa (fig. 42, left): sitting cross-legged (legs damaged) on concept to the New YorWMathurii set, which seems to date from
a low throne with lathe-turned supports, wearing a diadem, the 7th-8th centuries, leaving a rather big time gap. It is therefore
pendent earrings, a simple necklace, pot-bellied, holding a possible that such loose panels of directional guardians had a more
fruit in his right hand and the end of a large, circular noose in continuous history than is known at present. The image has the
the left hand; below the throne is his vehicle, the goose, stand- following characteristics:
ing in profile. This figure was identified and first discussed
Soma (fig. 45): orange-coloured terracotta; the figure sits with
by MITRA(1960)19J.
splayed legs and crossed feet197 on the lower part of a raised
4 Viiyu? (fig. 43, right) is seated with his feet on the ground.
frame, which is horseshoe-shaped above and profiled on all
The coiffure seems to have been a simple topknot, the ear-
sides by a simple groove; he wears his hair tied in a topknot
rings are asymmetrical like those of Yama (2). The left hand
above layers of short spiral curls, is adorned with pendent ear-
holds a rod, which seems to extend to the top of the panel. It
rings (not discernible on the left) and a simple necklace (only
might have been part of a standard (cllzvaja) or goad (aiikuia).
three central jewels or pendants are visible); the face is dani-
The right hand probably holding a fruit rests on his right
aged; the torso with a slight paunch is well modelled; a sash
thigh.
crosses the torso like a briihmanical cord; the legs are covered
5 Soma o r Agni? (fig. 43, left). The figure is seated in a man-
by a dlzoti extending below the knees, which is held below the
ner rescmbling his neighbour, but is even more obese. He
navel by a knotted cord; he holds a rosary in his right hand at
wears his matted hair in a simple hairstyle tied on the head
shoulder level and a bottle held at the neck in the left one rest-
with a few strands hanging down. He holds a waterpot with
ing on the left knee. The small horizontal inscription is located
incised decoration in his left hand and a rosary in his right
on the frame to the (proper) left side of the figure. An interest-
hand which is unproportionally large. The god's ascetic
ing feature is that the figure projects out of its frame, as if emerg-
aspect has been especially emphasized since he wears no
ing from it.
ornaments, has elongated earlobes and wears a particularly
long sacred thread. A similar figure of the Moon God Somd
If this Soma figure originally belonged to a set of lolca/~dlas,it must
Candra, but executed in a somewhat less refined manner, is
have represented the guardian of the north, as Soma is mentioned in
found on the north(-east) of the jngai~zohana(DONALDSON
this capacity in some of the Puriinic texts. It most likely formed part
1985: fig. 79).
of a temple, perhaps situated at the base or roof. Such a position
6 Dikpsla? (ibid.: fig. 75). Only a tiny portion remains, con-
would conform to that of other early pieces, mostly from the 6th and
sisting of a part of one leg (knee) and a part of one arm or hand
7th centuries.
above this, probably holding an attribute. This is sufficient to
let one surmise that the figure is seated in the same cross-
legged posture as the neighbouring deity. It is quite likely that
2.3 Further Development in tlie Different Regions
the fragment has been preserved in its original place.
7 I(ubera?~95(fig. 44) wears a hair-crown and asymmetrical
Towards the middle of the 7th century, a threshold seems to be
earrings like Yama, but interchanged. His ornaments resem-
crossed, which leads to a phase that fully knows and employs the
ble those of the clikpnlas Yama and Varuna, but the sacred
complete set of eight directional guardians. AIampur in Andhra
thread is absent. An ~~~larabailcllza ("belly band") is worn
Pradesh is the site where this transition is attested most clearly. For
around the slightly obese belly. Like Yama and Varuna, this
the KiilikamIta temple at CittaudgaghIRajastl~anwith another com-
figure also holds a fruit in his proper right hand. Only a small
plete set, a late-7th century date has been proposed, which may be
portion of the left arm survives, as also part of the lower right
questioned on the basis of its advanced architectural and icono-
leg which is drawn up towards the torso. In the latter feature
graphic features, however. Thus, AIampur can still be regarded as
he resembles Yama.
the main site for the beginning of the theme's period of early
maturity. It must be noted that during this period the scts are some-
On the jaganzolzaiza of the ParaSurameSvara temple, a set of more
than four directional guardians doubtlessly existed. It cannot be
decided whether the dikpnlas or the Manu Lokapalas were depict- '" Varuna is not rcgarded as il tlircctional guardian by her, but as a "deity othcr
than a clikpriln" (MITRA1960: 2 ) .
ed, and whether it was a complete set at all. In any case, a Dangapiini Iq5 The other remaining deitics Indra, Agni, Nairrta and iiina arc lcss likely
guardian figure was an associated member. candidates for an identification. In the case of Indra, a cylindrical crown would be
expected, as Agni, lie would have to hold a rosary in his proper right hand, Nairrta
wields his sword exclusively in the right hand, and as iiina, he should be carrying onc
of Siva's attributes in his right hand. The identification as Nairrta has, only 011 the basis
2.2.4 Other Early Evidence of the -actually sccondary - (south-)westcm position of the deity, bcen proposed by
K.C.PANIGRAIII (1961: 70).
Terrncottn Sonla fro111 Uttar Pr~z~leslz l" G. BIIAITACIIARYA (1993: fig. I). I should like to thank the author for kindly
pulling the original photograph at my disposal.
l" G. BIIATTACIIAIIYA has co~nparedthis pcculiar sitting posturc with that of
An interesting piece of early evidence is the recently published
Brah~nifrom Blinmara (BANERJI 1924: pl. Xllb). Apart from sonic minor dil'fcrcnces,
terracotta image of Soma (thus inscribed), reportedly from Uttar howcver, the lattcr wears a yogcrl,cr!!n to keep up this strenuous pose. Othcr dctails of
pmdeshl96. The image is noteworthy both for its aesthetic quality thc rendering also indicate that Soma's posturc is somewhat formalized and is not
and the uniqueness of several of its features, particularly the fact that meant to be realistically porlri~yed.
times not yet purely canonical in their composition. A peculiar fea-
ture in this connection is the loose association of the Sun atid the
Moon God with the clikpdlas.
For the later phase of this development, much evidence comes
from Osiiiii in Rajasthan, where a cluster of temples, as at Alampur,
shows both unifor~nityand diversity in the treatment of the theme
and the shaping of the individual sculptures of the directional
guardians. At other Gurjara-Pratihsra sites also, dikpnla images
were part and parcel of the sculptural programme of the temples.
Their popularity apparently spread over large parts of North-west
and Central India and the Deccan during this phase.
2.3.1 Early Phasc (c. 651-750 A.D.)
It is a striking fact that the Calukya region of thc western Deccan,
which already provided quite early evidence for the portrayal of four
lokapdas, conservatively clung to this restricted number198 until
c. 730 A.D. The more colnlnon placement on the temple ceiling was
only temporarily supplemcnted by the placement on the temple wall
in the western region. Further east, at AIampur, lokapnla ceilings
were first unknown, but the wall niches contained directional
guardians from c. 660 A.D. onwards (Bala BrahmB temple). Whether
the concept of four or eight lokapdlas was depicted probably did not
depend on the founder of the temple, as may be inferred from the
fact that Vinayiiditya's queen seems to be inscriptionally connected
both with the Svarga Brah~niitemple at Alampur featuring eight
lokapcllas and with the Jambulinga temple at Biidiimi presenting
four world guardiansl99. Another significant aspect concerns the
often close architectural affinity of particular temples or temple
groups, which are linked by similarly designed lokapiila sets.
Many of the sets from this phase are at present incomplete. In
these cases, it has to be decided if the extant set represents the orig-
inal concept. The sites which preserve complete sets appear to be
distributed over a large part of South India, however excluding the
extreme south (for the latter see section 2.3.2.5). As the dates for 8th
century temples are almost exclusively estimated ones, they have to
be treated with caution. An upper limit for the earlier phase of c. 750
A.D. seems to be most reasonable. In the Deccan it coincides with
the end of the Calukya hegemony, while in North-west and Central
India it marks the time just before the ascent of the Pratihiira
dynasty, which must have flourished in a region with a rich artistic
tradition.
2.3.1.1 The Deccan
The depiction of Indra without any other lnelnbers the group has
already been mentioned for the Ravana Phad~cave temple at Aihoie
(GUPTE1967: pl. 138). During this phase, such an isolated Indra
occurs in the nearby Hucci~nallitemple. Contrary to the former
temple, he is - correctly -placed on the east in the latter one, while
Visnu, ~ i v a - ~ a r v a tand
i , Brahmi are shown on the south, west, and
north, respectively (C.E. RADCLIFFE 1981: ills. 7 16-7 18).
Bcldcl~izi(JambulitigeSvara ten~l~le),
Ndgaral (Nclgarziitlzn teniple),
Mahiikilta (Mallikiirjur~ntenlple)
the ~ f i alld ~ Mahakilp
~ ~ ~ telnples
! with ceiling panels of the
four lol~al,alas, dates around 700 A.D. have been considered in
the ~ ~ z c ~ c l o ~ ~ aofe cIrzdiarz
l i a Tenzl~leArchitect~lre~O~.
The Jamb~l-
liiigeivara temple at Badiimi is similarly dated by an inscription of
A.D. 699. The panels at Nigara! and Mahiktita are very similar to
one another. They are both arranged around a central panel of a seat-
ed Brahmii (figs. 47,48), while the Biidimi ceiling has a central fish-
spoked wheel (fig. 46).
These three lokapiila ceiling panels are the earliest ones with the
four directional guardians in a square arrangement. The re-occur-
rence of the rather uniform motif of only four world gu ard'lans more
than a century after Biidimi 3 is particularly remarkable insofar as
the latter already contained an extended set of more than four
lokapdlas (fig. 10). Later, the group of eight becomes the standard
programme for rectangular ceiling panels.
South
West East
North
Ill. 4: Biidiimi, Jambulinga temple, ceiling of bay in front of
northern sanctum (bold bar - bottom of sub-panel; arrow -
direction of movement of the figure)
BHdiinii, Jambuliiigeivara temple (fig. 46; ill. 4): A solnewhat
flattened wheel with fish-spokes (which originally numbered
24) and a lotus centre is circumscribed by a rectangle, with the
spandrels remaining empty. The raised outer border is decorat-
ed with lotuses of various designs, alternating with small fig-
ures. The four 1okal)Elas Indra (NE), Yama (SE), Kubera*ol
(SW), and Varuna (NW) occupy the four corners in such a man-
ner that Indra and Varuna are seen moving towards one anoth-
er on the northern side, while Ya~naand Kubera are shown rid-
ing towards an encounter on the southern side. Their vehicles
198 BUCIIANAN (1985.444) states 1111sfact In an amaz~nglysimplc way,ju?t reier-
rlng to the namer of the ~ n o d e ~states
n "A set of foul dikj)u'lar seem\ to have bccn stan-
d a d In K ~nataka,
I a rct of elglit 111 And111a P~adc\li"Tlic now ruined Galaganiitha tem-
plc at PattadakalIKaroataka was lnost probdbly an exception, as ~t was obv~ously
des~gnedto contaln a larger set However, what remain\ of the teniple closely resem-
bles the developed Alampur temples, which must have dircctly served as models.
1" "The Svarga Bralima Siva temple [...l was made in lionor of tlie Mahiidevi
(queen) of VinayBditya by Lokddityn Eli-arasa. An inscription on the Jambuliliga tem-
ple at Bidinii, dated three years aner tlicend of Vinayiditya's reign, stales that Vinaya-
vati, who was tlie widow of Vinayiiditya and mother of his successor Vijayiiditya,
installed Bralimt, Visnu, and MaheSvara at that placc [...l it is not certain that Vinaya-
vati was identical with the Maliidcvi ~nentionedin tlie Svarga Brahma temple inscrip-
tion" (BUCIIANAN 1985: 174).
ZrHl K.V. SOUNDARA RAJAN1986: 49 (Maliikfi!a. Mallikirjuna temple). 57
(Nigara!, Njgandtha temple). The temple at Nigara! has been described in detail by J.C.
HARLE(1969).
. - date was later modified to "third or fourth cluarter
. The earlier assipned
of the seventh century" (HARLE1986: 169). 111a more recent article on MahSkULa by
TARTAKOV (1985: 149), a date of c. 715 A.D. for the Mallikirjuna tclnple has been pro-
posed. For further chronological consideriltions sce below.
201 0. D~VAKARAN (1970: 28) was u~iderthe erroneous impression that the ram-
riding god is placed on tlie south-east, and liencc identified it with Agni.
 and attributes are: Indra (without crown, wearing a topknot) - South
elephant, standard; Yama - buffalo, club; Kubera - ram, club;
Varuna - rnnknrn, noose. The figures partially overlap the
empty spandrels of the rectangle.
Yama
Nlgara!, Nlganiitha temple (fig. 47, ill. 5): Unlike the Bldiimi
panel, which is composed of only two slabs, the one from
Nlgara! and all its successors (with directional guardians) are
composed of three slabs, which easily allows for the division
into a nine panel grid. In this case, the panels are occupied by west Varuna Brahmii East
Brahmii (centre), the four 1oknl)Elns in their correct directions
(centre of each side) and four celestial couples (corner panels).
BrahmL is shown three-headed and four-armed, seated on a dou-
ble lotus and encircled by worshippers. Indra is the only one
who proceeds in the clockwise direction, seated on his elephant
along with two servants, accompanied by three sword-and-
shield bearers, and like the other loknpiilns wields a club in his
right. Yama is seated on a galloping buffalo and is accompanied
by four club-bearing warriors. Varuna is seated on his nzaknrn, North
a mythical aquatic creature with a huge branching tail, from
whose mouth a figure emerges. Kubera is shown seated on his 111.5: NBgara!, Nlganiitha temple, central ceiling panel of rnn!l<lnpn
galloping ram, accompanied by five club bearers. The bands
which divide the panels are particularly wide in the east-west
direction. They are decorated with undulating creepers.
Mahiikiifa, Mallikiirjuna temple (fig. 48, ill. 6): The panel is
divided into nine sections with rosette-decorated borders. Orna- South
mental lotuses occupy the corner panels. Brahmii is three-head-
ed and four-armed, with his proper right hand in the gesture of
fearlessness. He is seated on a double lotus, which has broader
petals than the corresponding one at Niigara!, and he is wor- Yama
shipped by two bearded !?is (above) and other figures (below). 4

Indra proceeds anticlockwise like all other loknpiilas, seated 011

a galloping elephant, holding a short club, and accompanied by
five warriors with clubs and swords and another figure. Yama
West Varuna Brahml Indra East
is seated on a mighty buffalo, holding a staff or club (damaged),
escorted by three warriors. Varuna is seated on a mnknrn, hold- v A

ing a large club, accompanied by three sword-and-shield bear-

ing warriors. Kubera, on a mighty ram, has no visible attribute,
and is escorted by three warriors.

On the two latter panels, the heads of the loknpiilns are directed
towards Brahmii, so that they can be properly seen by the temple vis-
itor when proceeding in the respective direction. The ceiling panels North
have been placed in the centre of the pillared hall (nza!l(lnpn) in the
latter two cases, and directly in front of the northern sanctum of a
111.6: Mahiikilta, Mallikiirjuna temple, central ceiling panel of
triple temple at BLdiimi. Additionally, the Mahiikilta Mallikiirjuna
mn!z~npn
shows a procession of the four loknl~cllnsholding clubs with warriors
on the eastern basement of its nza!z&rl,a (BUCHANAN 1985: 232).
After comparing the typological features of the panels, it shall
be proposed that Bldlmi is the earliest, followed by Niigara! and
Mahiikil[a in this order2o2. The chronological order of the temples as
suggested by BUCHANAN (1985: 236), however, is the exact oppo-
site of this. According to her, the temple at Mahlkilfa is the earliest 21'2 The loknl~cila\arc not placed 111 the^^ colrcct pos~t~on\ on the BHdlm~p~nel.
-- the
nf ---- three. as the so-called Hindu Trinity is not -yet represented
- on This placemetlt ~cem5to be the most archa~c01"the ~ I I I C CS ~ ~ C I I I ~di\cu\\ed
C~S here,
its ceiling. she assumes that the Nfigara! temple postdates it, but sincc their arrangement in thc rcctangular space is apparently not yet standardized.
Their heads point to thc north (Yama, Kubera) and south (Indra, Varuna), respcctivcly.
predates the One at TARTAKoV'S (lgs5: 149) dating' how- At NHgaral, the couples in the conlcrs also have thcir heads oriented towards the north
ever, tallies better with the proposed sequence lokn~alapanels. or h e south. Indra does not confonn to thc anticlockwise direction of his fellow jiods,
He places the Mahlknta Mall~klrjunarelatively late, in the middle d some rlgures squeezcd underhls lotus \e,lt Only
and Brahm2 1s \tlangely ~ c n d e ~ ew1t11
of Vijayiiditya's reign (C. 715 A.D.). at Mahikuta 15 thc layout of the pancl wcll-planned and regular, w~th\trip\ o f u n ~ f o r ~ n
border5 d1v1~11ng the Inner rectlons On the Mahiktta c e ~ l ~ nthe g , dangel of thc contl-
Although the exact dates for the respective temples are
nulty of the pattern\ at the beam jolnts bang ~nte~rupted has bccn ~ n g e n ~ o u plevent-
~ly
unknown, a 'lear to resume the loka~'E1a lnotif On the cd Thc stnps runn~ngparallel wtth the \labs are co~nplctclycarved on OIIC slab, ~nthe
temple ceiling can be observed in the latter part or the end of the 7th cascs ofthc crosrlng stl~p\.a S I I O I ~stretch stalts 011~ I I Cddj~llllngbeam
century. The four directional guardians re-occur on the central
nin!l<lrrl,n ceiling in the Malliklrjuna temple at Pattadakal (fig. 326;
c. second decade of 8th century)203 and on a ceiling on one of the
temples of the Jyotirlinga group at Aihole (c. fourth decade of the
8th century)204, as well as on the walls of the PBpanBtha temple at
Pattadakal (ibid.: 361; 397; 278; see below).
The visual pairing of the gods of opposite directions only occurs
at BBdBmi (fig. 46). This unusual arrangement could indicate an
experimental handling of the lokrrp~ilnconcept.
With this temple (BOLON 1988: 309) and the PBpaniitha at
Pattadakal, both datable well after 700 A.D., the asfndikl~iilns
become a standard iconographic programme for temple ceilings in
the Deccan. Their placement in the Huccappayya temple is once
more an unusual one, as they appear on the sides of the massive
cross-beams which support the ceiling slabs, in the centre of the pil-
lared hall of the east-oriented building. Although they must have
originally numbered eight, only six images are still preserved. The
central ceiling panel, which showed BrahmB or Uma-MaheSvara
( ~ i v awith his wife PBrvati), is not in sitcl any more (BUCHANAN
1985: 391). The Huccappayya temple exhibits another special fea-
ture that is rare in the western part of the Calukya kingdom: name-
ly an image of Varutp on the western side of the tower (Sikl~nrrr), can be judged from the less encrusted surface, are all carefi~lly
near its top (ibid.). No counterparts survive on the other sides. On
the Kumiira BrahmB temple at Alampur, about a century earlier,
Yama and Kubera occur on the Sikhnrn (figs. 23,24).
Indra (fig. 49): centre east, seated on a huge elephant together
with his consort, his four-tusked elephant rising from a mass of
clouds; Indra wears a cylindrical crown and has a halo, is two-
armed and holds his thunderbolt (vnjrn) in his right hand raised
to the shouldcr level.
Agni (?) (fig. 50): on the (south-)east, seated on a galloping
horse together with his consort, wearing a conical crown, prob-
ably with a halo, two-armed, holding an indistinct object (prob-
ably a rosary) in his raised right hand; the background is filled
by cloud or rock formations. The vehicle horse would be unique
for Agni.
Yama (fig. 51): centre south, is seated with his consort on a
mighty buffalo. The animal is well integrated into the architec-
tural frame and co~lspicuouslythrows up its head. Yama wears
a conical crown, earrings, a necklace, and a br%hma~!icalcord
(cll7nvlta). He is two-armed, holding a long, thin staff in his right
hand close to his waist.
Viiyu (fig. 52): (north-)west, is seated with his consort on a leap-
ing long-necked antelope, that turns towards the onlooker. VByu
wears a high, somewhat conical crown, has a halo and is two-
armed. He holds a standard in the left and an indistinct object
near the chest in the right hand. A mass of undulating clouds can
be seen below the antelope.
Kubera (fig. 53): centre north, is seated on a galloping ram with
large horns, wearing a conical crown, earrings and cr/>avitn.He
is two-armed, the left one casually resting on the curly mane of
the animal, the right one holding a club of medium proportions.
ISlna (fig. 54): (north-)east, - considerably obliterated -, is
seated with his consort on a mighty bull.
It is significant that only the reliefs of Kubera and Yama, facing
each other and each being isolated on their respective beams, have
an ornate frame of split gnviik? arches, while the others are bor-
North
.
Kul)cr:i
West
Yama
.
111.7: Aihole, Huccappayya temple, central bay, loknpnlns on the
beams
dered by simple rectangular ridges. The loknl>iilnpanels, as far as
designed and se~lsuouslymodelled, with the graceful movement and
the controlled power of the animals being accentuated.
The PBpaniitha temple at Pattadakal (BOLON1988: 305-307), which
reveals different phases of building activity, bears the unique evi-
dence of a set of four 1oknl)Glas on its exterior (western side) and a
set of eight on its ceiling (eastern pillared hall). As the western hall
(riin!l~n~~a)must be the older one205, it can be stated that the subse-
quent stages of four and eight directional guardians have been
recorded on this temple in a singular manner.
21)) Here, the directional guardians have been incorporirted into a NaFeSa panel.
Siva is eight-armed and accompanied by Pirvati and his bull vehicle (P. BANERJI 1985:
ill. 12 [mirror-image]; also RADCLIFFE I98 l: fig. 857; BUCHANAN 1985: fig. 549). There
are two vertical rows of divine attendants: on tlie left (from top): Viy!u, Indra, Yama
(?);and on tlie right: Bralimi, Kubera (?), Varul!a. BUCIIANAN (1985: 352) suggests an
earlier date for tlie Mallikiirjuna te~nplc,wl~icl~ had been assigned to c. 74015 A.D. She
rcgards it as the carliest of tlie tlirce large southern-style temples at thc site and assurnes
that its construction started "in tlic middle of Vijayiiditya's reign", i.e., c. 715 A.D.
204 BUCIIANAN (1985: 397, n. 183) identifies this pancl as "the first indisputable
Cnlukya cciling representation of Siva surrounded by dikl,dlns formally positioncd in
their directional domains." Siva is said to be acco~npaniedby Parvati, GaneSa. and
KBrttikeya. BUCIIANAN suggests that thc composition is "almost certainly inspired by
Huccappayya temple iconography". See also RADCLIFFE (1981: 409, and fig. 798).
TARR(1969: 225) in liis description mentions eight diklx7lns. However, the corner
figurcs are gand/ran~asaccording to the former authors.
205 According to G. MICIIELL (1973: 200), the rrra!~c/a/)osarc coeval, dating from
c. 750 A.D.,and tlie cella shortly before that. BOLDN (1988: 305) and BUCIIANAN (1985:
275-6; 409) suggest different versio~lsof a thrce-pliase cllronology. I11 both, the west-
er11 nro!r<!nl,n would bc carlier than tlie eastern, annexcd hall. BOLON'S proposed tinle
brackets for thc subsequent phases of construction are: I. 720-30, 11. 730-4; 111. 735-
50 A.D.,while BUCIIANAN suggests: I. 690-2; 11.692-700; 111.74015-? A.D.MICHELL lias
observed tliat the eastern wall of tlie western riro!r<lapo was, surprisingly, not designed
as an outer wall. Tlris fact is neglected by BOLON, wliile BUCHANAN incorporates it by
assuming that the building activity had stoppcd after tlie construction of the first
niches cast of the porches, on the north and south respectively. It is difficult to believe
that this state was maintained for about 40 years, and that the design of tlic pediments,
etc., was reproduced on tlie eastern part of tl~ebuilding after such a considerable lapse
of time.
 The ceiling panel (ill. 8; figs. 55-61) is located at the centre of South
the larger, approximately square-shaped iiza!z~apa, which has been
added on the eastern side of the siinrllzEra type temple. The relief has 4 .? 2
the unusual feature of a grid with three elongated panels in the
Nairlta Yan~a Agni
centre, as its central space houses Visnu in his reclining form Anan-
taiiiyi. Although the vehicles have lost some of the liveliness of b
T
4

those of the Huccappayya temple, their rendering is in many details

similar to the latter. A special feature, however, is the realistic por- 5 I
trayal of the riding posture of the deities, which compares well with
Varuna Anantaiiiyi Indra East
the Jambulinga temple at Bndnmi (fig. 46) as well as with the ear-
lier set of four on the same temple, excepting Yama (figs. 62, 63). • Vibnu <

Yama's buffalo with its head thrown back is similar to that of the
Huccappayya temple and was probably inspired by the Mahiikuta 6 7 H
ceiling relief206 (fig. 48).
VSyu Kubcra liiina
Both in the Huccappayya and the PEpaniitha temples, the
lokalxilas are represented with their canonical attributes, in this 4 A 4

deviating from the tradition of the NEgara! and Mahnkiita panels

(figs. 47-48), where they all hold only clubs. North

1 Indra (fig. 55): east, rides his striding elephant (four-tusked).
He wears a tiered crown and holds a goad (arikziia) across the
elephant's head in the left hand and a pointed attribute (?) in
the right hand.
2 Agni (fig. 56): south-east, rides on a ram. His hair rises in a
flame-like manner. He has a moustache and a pointed beard
and holds a bottle in his right hand.
3 Yama (fig. 57): south, riding a buffalo with mighty horns,
with its head thrown up. He wears a tiered crown, spiral-
shaped earrings, and an elaborate necklace and holds an
impressive club in his right hand.
4 Nairrta (fig. 58): south-west, riding on the shoulders of a fly-
ing lzara (anthropoid) with a tight cap or coiffure, who clasps
the rider's legs with his hands. Nairyta wears a conical crown,
earrings and a simple necklace and holds a sword in his right
hand and a lotus in his left hand.
5 Varuna (figs. 58-59): west, riding on his nzakara, which is
reduced to a head and a floral tail only. He wears a tiered
crown and holds an indistinct object (the end of a noose?) in
his right hand.
6 Vfiyu (fig. 59): north-west, riding his leaping antelope (iizrgrgcz)
characterized by massive straight horns. He holds his left
hand on his chest and his right hand raised; both hands prob-
ably hold a flowing scarf, which can, however, only be
assumed here, however.
7 Kubera (fig. 60): north, riding on his mighty striding
ram with a massive neck and curved, but not spiral-shaped
horns, wears a tiered crown and the usual ornaments and
holds a club (gadii, thinner than Yama's) in his left hand.
8 I&na (fig. 61): north-east, rides a bull with a stout triangular
head and wears a tiered crown. He is the only one with four
arms, but of the attributes, only a trident held in the upper
right hand is visible.
The dikpfiln figures Indra, k n a , Varuna, and VByu can be recog-
nized as one approaches from the south, Agni and Nairyta, when
approaching from the north, and the large rectangular panels of
Yams and Kubera can be perceived when coming from the east.
Such an irregular positioning is only paralleled by the similar ceil-
ing panel in the KiiSiviSveSvara temple at the same site (ill. 24).
The 1okal)iihs belonging to the set of four are sculptured in high
relief directly on the exterior walls beside the false porticos on the
south, west and north sides of the east-oriented temple.
111.8: Pattadalcal, Pjpaniitha temple, eastern iizn!z~al~a,central
ceiling
1* Yama: on the south(-east), has as his corresponding figure
a seated BrahmE - south(-west). He is seated in rijalilisaila
(posture of "royal ease") on a throne, with a buffalo crouch-
ing in front of it. The damaged right arm rests on his right
leg.
2 V n d r a : on the (south-)west (BOLON1988: pl. 627), facing
left, corresponding figure: Varfiha - (north-)west. He is
riding on a relatively huge elephant; only the right hand is
raised, probably holding an elephant goad (arikuia). He
wears an elongated crown.
3" Varuna: north(-west) (fig. 62), facing left, riding on a
iizakarn with its body almost merging into the scrollwork;
his right hand raised in the gesture of salutation (vaiirlaiiii-
nzullrii), the left carrying an angular unidentified attribute.
4 q C u b c r a : north(-east) (fig. 63), facing right, riding on a well-
proportioned ram, holds a club in his right hand (broken at
the shoulder) and with the left hand in the gesture of saluta-
tion. A halo is faintly visible behind the head.
3:':Varuna (P) 4" Kubera
2" Indra
Brahms (P) I :l: Yama
Ill. 9: Paftadakal, Papanitha temple, western portion, outer wall
P - fahe port~co(blrcidrugci~~cih,cr)
ZoVThere,however, all an~makhave Llle~rliead\ 1d15ed.On he Hu~c~~ppuyya
allcl
Papaliarha temples, thw I \ a feature of only home of the vcillo~ln\

38
 The directional guardians of this series are - except for Yama -
also shown in riding postures. They are again two-armed. The
gesture of salutation (vnrzdarzanzudrii), not observed on the ceiling
panels, is seen in the case of Varuna and Kubera on the exterior.
This is quite puzzling, as the only difference between the ceiling
set (1-8) and the ceiling panel of the later KiiSiviSveSvara temple
(fig. 129) at the same site lies in the presence of this gesture in the
latter case. Thus, there remains a serious doubt about the posteri-
i ority of the ceiling panel to the wall sculptures207. The gesture of
salutation later became a common feature of the ceiling images of
the dikpiilns.
The peculiar distribution of the directional guardians on the
wall, which is incorrect in more than one case (Indra, Varuna), can
hardly be explained. A possible reason could be a general uncer-
tainty regarding their correct placement in the early years of the
reintroduction of the loknl~iilntheme in the Deccan region, i.e.
around 700 A.D. Regarding the co~nbinatio~~ with the other deities,
the figures of BrahlnB and Variiha that occupy the remailling space,
could also be associated with the directional order of the cosmos.
The boar Variiha would symbolize the nadir and BrahmB the
zenith. Such an expanded set would be quite singular and it defi-
nitely recalls the Vedic notion of the six directions (see section
1.1.1.2).
AIampur on the Tudgabhadrii river in the Mahbubnagar district of
Andhra Pradesh looks back on a glorious history of building activ-
ity. The so-called Navabrahmii group of nine temples bears vivid
testimony to this fact208. Their present names are, proceeding clock-
wise from the north-east: ViSva Brahmii, Vira Brahmii, Arka
Brahmii, Kumiira Brahmii, BBla Brahmii or BrahmeSvara, Garuda
Brahmii, Svarga Brahmii, Padma Brahmii, and T3raka Brahmii209.
Today, Sivali~ignsare enshrined in nearly all of them (that of the
Tiiraka Brahmii has been shifted), with the one in the BBla Brahmii
temple still under worship. The Kumiira Brahmii temple with
~
images of two lok~rl~iilns on its Sikknrn has already been mentioned
as belonging to the previous phase.
l
All the temples have a sdndlldrn plan, with intergrated spacious
l
mn!~da~ms, with the exception of the Driivida-style Tiiraka Brahmii,
l
which is rzirnncllziirn. There is no break in the line of the wall sur-
rounding the ambulatory path ~ r n d n l c ~ i ~ ~ i i ~and
~ a the
t h aadjoining
) tions of tlic ViSva and Vira Bralitiii tcti~pleshave been erroneously intercliangcd. In tlie
~~zn!z<lapa. Therefore, the outer appearance of the building must have
called for an unusual decoration. Windows of various sizes, differ-
ent numbers of niches, floating ~trlgnrllns(pediments), wall sculp-
tures and pilasters were employed here.
alalnpur belongs to the sites with the earliest surviving images
of directional guardians on the temple exterior. Most probably, sets
were fashioned on seven temples (Biila4', Arka, Vira, Svarga*,
ViSva:I:, Garu+% Padrna)210, excepting the KumBra and the Tiiraka
Brahmii temples, which are without any appropriate niches. At
Alampur, the canonical group of eight seems to be supplen~ented
by the Sun and Moon Gods on the east and north, respectively2~~,
judging from the allotment of the deities in the niches on the
Svarga BrahtnB (ill. 11). This is the only temple with all its images
surviving. Images of two forms of Siva are associated with the
Sun and Moon Gods and the eight lokap~ilnsin its outer wall nich-
es. This combination may symbolize, as BUCHANAN (1985: 186)
puts it, "the subordination of the earth and time to ~iva".Altema-
tively, the Sun and Moon Gods could have been portrayed in allu-
sion to the supposedly popular Manu Lokapiila theme (see section
2.4.1.3).
The absolute chronology of some of the temples is quite secure,
as the Arka Brahmii bears an inscription of Vikramiiditya I (A.D.
65415-681) and the Svarga Brahmii preserves a record of the time of
Vinayiiditya (A.D. 681-696; PRASAD 1988: 322; 330). The relative
chronology is not entirely fixed212, but the chronological sequence
of the temples with surviving loknpdla images hardly remains open
to question (in the following chronological sequence): Bala
Brahmh, Svarga Brahmii, ViSva Brahmii, and Garuda Brahmii. Con-
cerning the sizes, the slabs with the guardian images have average
heights of 81 cm on the Biila Brahmii, 95 cm on the Svarga Brahmii,
and 87 cm on the ViSva Brahmii temple. The considerable height of
the Svarga Brahrnii slabs is due to the integrated "floating socles" of
the images, which are a peculiar feature of this temple. The shaping
of the figures varies from temple to temple, sometimes consider-
ably, so that they can be easily distinguished (table XII).
The conceptions of proportion and style must have changed
rather rapidly during this time of an extraordinarily rich art produc-
tion. However, a comparison of the images from the AIampur
temples shows that distinctive types of loknpiilas have been created
here, which, as in the case of Viiyu, solnetimes combine both
North and South Indian features (billowing cloth and standard,
respectively).
207 In this light, MCI~ELL'S opinion (1973: 200) is probably correct. According to
him, the westcrn and eastern 111o1,1<1opos have been constructed at the same time. Mod-
ifying his theory to fit the present findings, I would postulate tliat the carving of the out-
er walls contitiued for sotiie titiie after the completion of the interior.
"'8 0.DIVAKARAN (1971) has devoted an exccllent article to tlie Alatnpur tetnplcs.
It gives a dctailcd atialysis of their significant architectural and iconographical featurcs.
Tlic directional guardians have for the first time been correctly identified by her (ibid.:
61; 66). However tlie less intact buildings, like the Padtna Bmlitnl temple, have been
sotnewhat neglected. For other publications see [lie following foottiole.
2119 Much confusion has been created by inadvertently mixing up tlic names or
positions of the buildings. In the earliest description published in tlie ADNDAR (1926-
27: 7-12), however, the same names, which arc alrcady knowti from a 16th cctitury
source (BUCIIANAN 1985: 116). may Iiavc been assigned to the temples differently.
Therc, the Svarga Braliml temple is referred to as Solka Bralima, the Pad~iiaBralinii
as Taraku Bralitna, the Vira Bralimi as Arka Bralitna, and the ViSva Brahtnl as Vira
Brahma. Some plans arc incorrectly labelled in P.R. RAMACIIANDRA RAO(1977: 25-
28): it must be Vira instead of Arka Bralima, and Padma instead of Garuda Dralitira. M.
RAMARAO (1961: 369) inadverten~lymentions the "Arkabrahmi~" twice, "Svarya
Bralitni" bcing corrcct in the first instance, as forming a group with the Padnia and
Garuda Brahmi tctnples. In tlic rccent publication by MICHCLL (1989: 344) the posi-
Eitc)clol~oedioof 111din11Ter~rl~le A,rhifccfrrre (BOLON1988: 338, fig. 133) the ground
plan of the Garuda Bralitiil temple goes under tlie guise of "Pa?yani. Siva temple".
21Ll An asterisk indicates, where iliiages of /okrr/~irlosstill remain. Even if the nich-
es never containcd the actual imagcs, tlicy were ~iiostprobably designed to Iiold tlietii.
Tlie supposedly earliest tetiiplcs with a nuniber of exterior niches arc tlie Arkn and Vira
BraIimZ ten~ples.Their tell ~iicl~es sccm to IIWC bccti planned for tlie same ten deities
as on tlie Svarga Bralitni (ill. 11).
"1 G. MICIIELL (1973a: 84). who first described a set of the Alampur loko/~cilrrs.
comments on this fact: "That Sftrya and Candra are also included with tlie rlik/)ri/o.s
recalls the pre-Prrrri~jictexts where these deitics were considered themselves as
loknlxil(~srather than as 'planets'. This suggests that tlic iconographic sclietiie at Alam-
pur liiay belong to a transitional stage bcfore the stereotyping of the lokn/~cilrlists in
the Pl~rrirlrr.~had occurred." Tlie author unjustifiably presupposes tliat the sculptures
werc dircctly dependent on the scriptures.
212 Tlic Navabralimi group seems to be a classic exa~iiplefor the transition frotii
the simple to the intricate form. Tlie less evolved tctiiples of those under discussion are
tlie Bila. Arka and Vira Bralimi temples. the most evolved oncs tlie ViSva, Pad~iiaand
Garudn Bralimi temples. Tlie Svarga Bralimi temple liolds an intermediate position.
DIVAKARAN (1971: 73) proposcd tlic following clironological scquence: Kutnira, Bila,
Arka, Viru, Svarga, ViSva, Garuda Bralimi, however otiiitlitig tlie Padlila Bralitnl
tetnplc. Later scholars expressed differcnt views, e.g. placing the Blla aftcr the Vira
Bralitni (PI~ASAD 1988: 326), or placing the Garuda and Padtna aftcr the ViSva Bralimb
(RADCLIFPE 1981: 567). So~netiniesthe decision about tlie actual sequence has been
avoidcd by proposing overlapping or identical bracket dates. which is also acceptable.
 Table XI1 - Scuptural typology of dikpnlns at Alampur

Deitv I Attributes I
A - Arka Brahmri; B - Brila Brahmii; G - Garuda Brahmii;
S - Svarga Brahmii; V - ViSva Brahmii
Vehicle I Characteristics
1 I
Indra (B, S) small vajra (bolt) r.; elephant; high crown; S: large dccorated halo;
B: flower 1.; S: nk;arizdld (rosary) 1. B: frontal S: horizontal third eye
Agni (B, S) aksan~BlB(rosary) r., - flames behind the whole body,
kanza;z~lalu(waterpot) l., crown of flames, long hair,
standard floating behind left arm sash worn like ~rl~avitct
(sacred cord)
Yama B, S: staff r.; buffalo; high crown; B, S: two treestone tree in thc
(B, V, V: staff l.; S, V: with background, no ~lpavita,Citragupta on
S -damaged-) S: ? -damaged- 1. raised head the right; V: Citragupta on extra socle to
the left, holding manuscript
Nairrta (S, club or sword r.; skeleton, S: peculiar hairstyle, faintly rese~nbling
loose in A213, A: sword r. standing ilrcllzvakes'a (hair standing on end);
fig. 83) frontally no ~ ~ l ~ n v A:
i t acrown
;
Varuna pdia (noose) r., - no ul)avita, halo, nioustache'?
(B, S) iariklza (conch) 1. B: jczfdrn~rktr/a(hair crown)
VByu (B, S) B: billowing scarf r., flag 1.; -
moustacl~e;B: curved portion of
S: scarf in both hands, streamer r. scarf visible, halo;
S: curve of scarf not visible
Kubera gadd (mace) - staff with bulging top r.; wearing a draped garment
(B, S, V) S also: neck of of the sack l., below the breast, halo, disc earring r.;
on which he is seated B: moustache, no rrl>avita,long hair?
S: asymmetrical face, petalled halo,
throne behind sack; V: only throne
Soma ak;arrzald r., kanza!zdalu 1.; - S: large halo with inscribed crescent,
(S, V, G?) V: broken r.*" G: additional standard waterpot held from below near
behind right shoulder (like Agni) the shoulder
&na (S) 4-armed: handled object lower r., - ja;iinzultu/a
-damaged- tris'illa (trident) upper r.,
serpent? upper l.**
SClrya holding a bud in each hand rectangular breastplate,
extra large halo, boots

** The other hand rests on the hip or thigh.

The partly surviving Ga!aganiitha temple at Pattadakal con-
forms closely to the Alampur type. It may have been known under
the name Lokaprileivara, as mentioned in an inscription at the site
(BUCHANAN 1985: 344). Unfortunately, no images of IoIcal~das
have survived here at Pattadakal.
Bsla Brahm5:
The Bcila Brahmci temple dates from the middle of the 7th century
or slightly later215. The remaining images show that it must have
been originally provided with a full set of eight directional
guardians plus sun and moon (ill. 10). It is the earliest of the
Alampur temples with such an iconographic programme, which was
repeated on all of the later temples, as far as it can be inferred from
2 ' 3 BUCIIANAN (1985: 272) tentatively assigns this image to thc ViSva Urahmri
telllplc, along with an :lllegcd iniagc ofl~idrai~nagckept in thc samc place. RADCLI~'PE
(1981: 259) regards this and other imagcs (ibid.: ills. 456,457,459-461) as probably
belonging to the Garuga BrahtnB. One of the images in the site museum most probably
represents TSBna (ibid.: ill. 457).
The Si~ryaimage of the Bala BrahniB tc~nplehas bccn regarded as a later
replacement. However, this still confor~nswith tlic original concept. An image i~ssigned
to the ViSva Bral~mite~npleby M.R. SAICMA (1972: 5 I) with its height of 1 10 cm (A.W.
KHAN1973: no. 7) is too large to lit into a niche only 87 cm high.
215 Tllc dates suggested for the BBla BrahmB lcmple ;VC: mitltllc of 7th century
(DIVAKARAN 197 1: 73); 660-690 (RADCLIPI:E 198 1 : 567); 660-670 (BUCIIANAN 1985:
184); 670-681 A.D.(PICASAD 1988: 326). Tlic Arka and Vira BrahmB tc~nplcshavc thc
same number of niches which may have contained the lokc~~~rilts. Only fcw scliolars,
however, regard these teluples as predcccssors of the Bila Brahmi (e.g. PIZASAD 1988:
326).
I the number of niches, now mostly remaining empty. The rlikl1E1n
images which are extant on the Biila Brah~niitemple (Indra, Agni,
Yama, Varutp, Viiyu, Kubera; the images of &na and Siirya are
probably not as 0ld216) exhibit sturdy proportions. Their features,
postures, attributes, and particularly the absence of vehicles in the
majority of cases reveal a remarkable originality. In fact, no direct
predecessors are known for most of them. It appears as if every deity
was treated in a consciously innovative manner:
I Indra (fig. 64), seated astride his crouching elephant, frontal-
ly faces the viewer. The only known parallel, also quite close
in time (7th-8th centuries), is the enigmatic elephant rider in
the centre of the eastern wall of the Arjunaratha at Miimalla-
puram/Tamilnadu217. Although the elephant is given much
more space in the latter example, there may be some formal
relationship between the two images, as the rider's right hand
is rendered in a similar position and the lower part of the
elephant's trunk also curves to its right. This similarity be-
tween the two sculptures does not necessarily imply the same
identification of both images. The Bala Brahrnii Indra wears
a high cylindrical crown, the usual ornaments and the
br3hmanical cord. In his frontally shown hands he holds a
small, horizontal thunderbolt (proper right) and a small fruit
or flower (left).
2 Agni (fig. 65) confor~nsbasically with the much older
KusInaIGupta concept of the deity (fig. 7) in that he is stand-
ing without a vehicle, surrounded by flames, and holding a
rosary and a waterpot. Some 11ovelfeatures, however, are the
flame-crown, long, flowing hair, and the standard (stnoke-
banner?) emerging from behind his left arm. Similar to
Varuna's attire, sashes are visible on either side of his legs.
3 Yama (fig. 66) is shown seated on his buffalo. This type has
occurred earlier in the Deccan (figs. 10, 13, top). Here, two
new features have been added, viz. two palm trees (a symbol
of the south?) on either side in the background, and his scribe
Citragupta on the right side of the panel. Citragupta appears
to be emerging from behind the bucfalo's head. He most
probably holds a stylus in his raised right hand. The I T I L I C ~
smaller figure has been skilfully incorporated into the com-
position. Later depictions of Citragupta at AIampur (figs. 72,
80) and at Kudnveli (fig. 85) show him - rather clu~nsily-
floating in the background or standing on a small socle above/
behind Yama. Tbe buffalo wears a bell on his neck.
4 V a r u ~ p(fig. 67) represented without a vehicle is very rare218.
He is seen standing in the same rigid srm~nl~Si(/nsthdr~nkn pos-
ture like Agni and Viiyu. A uniquc feature is the introduction
of the conch shell as his attribute in the left hand. Another
notable detail is that the inner part of the loop of the noose,
which he holds in his right hand, is not hollow, but remains
solid. This recurs on the Svarga Bmhmii temple and in
Pratihiira art as well.
5 VIyu (fig. 68) is also represented without any vehicle. The
well-known Viiyu iinage from Piihiirpur (fig. 236) offers
an interesting parallel. Here, a unique combination of
"North Indian" (billowing cloth - viiyviil,irritn-vastrn [Vcll~P
111.58.11) and "South Indian" (standard) attributes219 occurs.
The image is Inore consistent than its successor on the
Svarga BrahmP temple (fig. 75) in that each hand holds only
one attribute. The individuality of Viiyu has been further
emphasized by a helmet-like crown and a demonic face,
resembling Varuna and Kubera, with large, bulging eyes and
a big moustache.
6 Kubera (P) [Candra] [ISiina]
5~ a y u [Siirya]
4 Varuna I Indra
(P) 3 Yama 2 Agni
111.10: AIampur, Bala Brahmii temple, outer wall niches
6 Kubera (fig. 69) has broad lips and seems to smile gently,
in spite of his large bulging eyes and moustache like Viiyu.
His prominent belly is further emphasized by a curtain-like
drapery. The latter probably represents the "northern" style of
dress as prescribed in some texts. Kubera's left leg is sup-
ported by a drapery-covered pouf. Such a seat often occurs
in North Indian images of Kubera. The club (gnrlrl) as his
weapon, on the other hand, is very rare during this period in
North India, but is found in South India.
The most significant characteristics of the early lokn/~iila set
of the Biila Brahmii temple are the variation of postures and the
seemingly arbitrary inclusion or absence of vehicles. In some cases,
the particular concept of the deity was determined by earlier images
that must have served as models. In two cases, i.e. Indra seated
frontally on Airiivata and Varuna standing without iizaknrn, no
earlier model could be traced. In the cases where earlier tnodels are
known to us, some modifications have been introduced at AIampur.
21h BUCHANAN 1985: 186-188. The iSana image is described as "somcwhat
s~iiallcrthan tlie othcrs [...l He holds a long-sliaftecl attribute in liis rear right hand, and
pcrhaps his rcar lcft hand holds a rosary. A pot is placcd at liis lower left." For an illus-
Iralion see RADCLI~;E 1981: ill. 327, labcllcd ''Dhalri?'.
2'7 The identificiltion of tlie figure, which is devoid of ally attributes, as lndra lias
bcen questioned long ago. L'IIERNAULT (1978: 141; pl. 124) lias proposed an idcntifi-
cation as Skanda, although slightly hesitant. S.L. HUNTINGTON (1981: 63) argues for an
interpretation oftlie deity as Aiyanar-$3~13.Presently, it is impossible to find an undis-
putable solution. This is probably what HUNTINGTON (ibid.:60) wanted to indicate when
she rcferred to the "frequent use of double-ctitendre [ambuguity] in Pullava art ancl
epigraphy".
218 At Osi3fi/Rajastlian, Varu~ylias becn shown without a vfihar~crin two cases
(figs. 221,222), wliich could probably have becn due to a lackof space. In tlie first casc,
Varupa stands with his legs crossed - tlic left Icg behind tlie right; in the second case,
lie stands witli his fcct placcd on tlie belly and brim portions of a waterpot lying on the
side, which occurs instead of a vehicle here.
2" This division occurs in tlie period under discussion (until c. 900 A.D.),with
only few exceptions. The Orissa region confornis to the Deccnr~arid tlie resl of South
India. In looking for parallels, it liiay be significant to observe that in the Pralili5ra
region in the north, a few i~iiagesoccur in which V3yu's cloth has becn combined will1
a stafflield in tlie other hand, with tlic cloth winding round tlic stafl(fig. 318). Even the
helmet-like crow11 reseiiiblcs tliat of the prcsclit cxample. IIowever, tlie figures do not
appcar to be relatctl in any way.
Svarga BrahmP:
The construction of the Svarga Brahmii must have taken place
sometime in the last two decades of the 7th century220. It is the best
preserved example at Alampur. This temple has two additional wall
niches, which contain images of ArdhanHriSvara (north) and Hari-
hara (south). These syncretistic forms of ~ i v ahave been placed
between Candra and TSana on the north221 and between Agni and
Yama on the south. The order of the set of directional guardians with
the associated Sun and Moon Gods, however, has not been changed.
The placement of some images directly on the wall between the
niches is an innovation. The 1oknyZilns appear stiff and somewhat
archaic in comparison to these larger and more lively conceived
figures. When compared to the earlier BBla Brahmii images, their
bodies are pronouncedly slim; so much so that even Agni's and
Kubera's paunches are hardly visible. However, on the whole, the
Bgla Brahma concepts have been followed. The development seen
in relation to the respective images on the Bala Brahmii temple (if
extant) can be gathered from the following comparative description
of images:
1 The image of Indra (fig. 70) differs significantly from its
counterpart on the Biila Brahma temple. He is not shown rid-
ing frontally as on the Bala Brahma temple, but stands in front
of his vehicle which has been reduced to a moderate size. He
has a large, decorated halo and wears a band across his belly
(udarabnrzrlha) like Tiiina on the same temple. He holds the
thunderbolt in his right hand and a rosary in his raised left
hand.
2 Agni (fig. 71) closely resembles his Bala Brahma counter-
part. His left arm is extended downward, holding a larger
waterpot, but in this case, the standard (smoke banner?)
behind this arm touches the ground.
3 The image of Yama (fig. 72) is also very similar, with only a
few differing details: he appears to be sitting above the buf-
falo instead of properly riding on it; the buffalo's head is
raised; and only one tree is shown in the background.
4 Nairgta (fig. 73) faces frontally and is shown seated on his
anthropomorphic vehicle. The corresponding image is miss-
ing on the Biila Brahmii temple. This figure is reminiscent of
the image in Cave 29 at Ellorii on the Nateia panel, where
Nairrta is also depictied on an anthropomorphic vehicle
(fig. 21, extreme left). The only difference is that Nairrta
wears a flat coiffure instead of a high crown at Ellorii.
Nairyta's traditional weapon, the sword, can be seen at &am-
pur, but not at EllorS.
5 Varuqa (fig. 74) holds the conch (jankha) in a manner dif-
ferent from that of the Bala Brahmii temple. Here, he grasps
its mouth instead of holding it from behind. Besides, his
crown is not made up of braids as on the earlier temple.
6 With only marginal alterations, the image of VByu (fig. 75)
has been modified to portray his element, air, more convinc-
ingly. The deity stands on a narrow socle. The extension of
the loop of the billowing cloth beyond the framework creates
the illusion of lifting the image from its base. Parts of two
attributes, cloth and standard, are quite inconsistently held in
the right hand.
7 In Kubera's (fig. 76) case, the absence of a prominent belly
is the most apparent difference, although a slight paunch is
visible. Apart from a much less opulent appearance of the
deity, other changes are the introduction of a treasure sack
instead of the pouf under his left leg, leaving a part of the low
throne visible, as well as the distorted face with a tilted line
7 Kubera NaLeSa S Candra Ardha- 9 ISfina
(P) 11HriSvara
6 Viiyu
Tri-
vikra~na
(P)
I l
4 N a i r ~ t a - (P) 3 Ynlna Harihara 2 Agni
Ill. 11: AIampur, Svarga Brahma temple, outer wall niches
of the nose. The throne with lathe-turned feet reveals that the
artists were familiar with the Central IndianIOrissan tradition
(cf. figs. 25,2(i, 42). Whether the slight distortion of the face
was actually intended is somewhat doubtful, but the descrip-
tions of Kubera often mention that some of his features are
asymmetrical or even deformed.
8 CandraISoma (fig. 77) exhibits a calm, transcendental
expression. His large halo with a crescent inscribed in the
lower part, is positioned such that the crescent is directly
behind Candra's head. The crown is of medium height. The
double-row pearl necklace and the ~rj~avita studded with gems
seem to hint at the pearl-white lustre of the moon. The vessel
held in his left hand near the shoulder creates a perfect
equilibrium in his stance.
9 The image of z~5na(fig. 78) has been badly damaged. He is
shown without a vehicle, four-armed, and clad in a rllloti with
the typical - for AIampur - sashes hanging down on both
sides.
10 Siirya (fig. 79) is characterized by a large undecorated halo.
His face and body are unusually broad. Even the niche itself
is comparatively wider. In the case of the body this is empha-
sized by a flared lower garment with a sash hanging down on
both sides. Snrya wears a high crown, an elaborate necklace
with a breast ornament reminiscent of a breastplate, a chain
girdle like Candra (8) and boots.
The set of directional guardians with the - structurally, but not nec-
essarily conceptually - associated Sun and Moon Gods on the Svar-
ga Brahmii temple represents an impressively homogeneous series
of individually conceived deities. Their stiff postures form a marked
contrast to the lively shapes of the sc~~lpturesplaced directly on the
wall surface. The miniature flanking figures on the innermost band
(STiklzii) of the door-like niches generally do not have any affinity
with the deity contained within it. Busts of praying serpent deities
are most often depicted in this zone. The fact that the set on the
Svarga Brahma is completely preserved makes it one of the most
outstanding testimonies of early temple decoration.
221' The follow~ngdate\ have been suggc\ted for the Svarga Brdhrna te~iipleaftcl
680 (DIVAKARAN 1971 73), 690-696 (RADCLII FL 1981 567), 682-690 (BUCIIANAN
1985 243), 681-696 A D ( P ~ A S A1988 D pl 658) For ~nacr~ptlonal ev~dencecf
note 199
221 Thc adjacent posltlon\ of ArdhaniriSvara nod iGna ha\ led MICI~I-LL (1973
85) to the hypothcs~sthat the V ~ ~ i ~ r ~ d l t n ~ ~ ~ t o r r cInr wli~ch
r ~ / ~ ~ the
r ~ cfor~iier
i ~ ~ n ,aeellis to
tepldce the latter, was known to the arcli~tects
 ViSva Brahmii:
~ The three extant images of loknj~nlnson the highly embellished and
~~
probably later (late 7th century) ViSva BrahmB222 temple still betray
the ancestry of the BBla Brah~nBiconic types. The modifications, as
on the Svarga BrahmB, mainly concern the stylistic treatment of the
figures (more plasticity and better proportions) and the layout of the
sculptural panels. The type of niche conforms to that of the Svarga
l
Brahma temple.
i In the 22 niches of the ViSva BrahmB temple, 14 images must
I
have been added to the group of agadikpiilas, thus clearly outnum-
bering them. Most probably, the corner positions of IndraIAgni,
Nairyta/Varuna, VByuIKubera (the latter still in that position) and
ISBna/Sfirya were maintained, while Yama and Candra (the latter
still remaining in this position) would have occupied the central
niches of the longer sides. Only Yama (fig. 80) seems to have been
subsequently moved from his place, which is also indicated by the
damaged border of the panel.
I The layout of the Yama panel (fig. 80) is a mirror image of
the Svarga BrahmB panel (fig. 72). Apart from this, some
ininor differences can be noticed in the leaping attitude of the
buffalo, the position of Yama's remaining, damaged arm, the
absence of the palm, and the manuscript held prominently by
Citragupta (damaged on the Svarga BrahmB), whose body is
shown in a slanting position.
2 Kubera (fig. 81) is characterized by a protruding paunch as
on the BBla Brah~nBtemple (fig. 69). The garment draped
over the belly is similar to the two preceding images. How-
ever, here the hand holding the club is raised to the shoulder
in a threatening attitude. The throne with curved feet is a nov-
el introduction here. Kubera's seated posture with a pendent
also differs from those of
right leg (dak;i!ziil~rnlarlzDa~~rTdn)
his predecessors.
3 CandraISoma (fig. 82), whose much smaller halo has again
an inscribed crescent, stands rigidly, with the left hand rest-
ing on his hip. The water vessel appears to be missing here.
The right hand is raised to the shoulder and shows traces of a
rosary.
2 Kubcra - -(P) - 3 Candn - - [iSina]
[vayui
- (RADCLIPPE
Gangs-
dliara (P)
- plcs. It combines both early (lack offirkn~~dscr
[Varuna]
[Nairlta] - - (P) - [Yania] - - l Yam
[&nil
111. 12: AIampur, ViSva BrahmB temple, outer wall niches
Althougli the ViSva BrahmB exhibits an extremely elaborate wall
decoration, there was again little innovation with respect to the
lokapiilns. However, the images of Yama with his leaping buffalo
and Kubera with his raised right arm are more dynamic in expres-
sion than their known predecessors.
Garuda Brahmii:
The Garuda BrahmB temple dates from c. 700 A . D . ~and ~ ~is the
southernmost of the NavabrahmB group. Only one niche image
survives on an inaccessible part of the building. Although the panel
is considerably smaller than the niche, it was probably not added
much later (fig. 84). With the visible attributes being a rosary (r.)
and a waterpot (I.) and located on the north side, one would natu-
rally expect this figure to be a representation of Soma. However,
there is no halo. Moreover, an object emerges behind the figure's
left arm, recalling the manner in which the standard is shown behind
Agni. As the latter is characterized by the same attributes, it could
also be a representation of Agni. Therefore, with neither halo
(Candra) nor flames (Agni) present, no indisputable identification is
possible.
It seems to be a sign of decadence that one of the latest temples
of the Navabrah~nBgroup at AIa~npurhas produced only a lifeless,
somewhat hybrid image of a directional guardian.
Kudnveli SaliganieSvam tenll~le,Recorzstr~icteda t A l a n ~ ~ ) ~ ( r
The SangameSvara temple probably dates from the late 7th cen-
tury224. It was formerly situated at the confluence of the rivers
KrsnB and Turigabhadrii, but has recently been rebuilt at AIarnpur
(re-consecrated in 1990). Its sii~~cllziiralayout (with integrated
ambulatory path) and its Northern type superstructure compare
well with the basic temple type at AIampur. However, there are
also major differences, like the southern (Dravipa) niche type, the
absence of large windows and false porticos, and the absence of
a Suknniisn (antefix) on the tower. The well-preserved enclosure
wall which encircles the temple has two integrated shrines on the
north(-east) and south(-east) and is decorated by a frieze of sculp-
tural panels225.
222 Thc proposed time brackcts are: 681-696 (DIVAKARAN 1971: 73); 690696
1981: 567); 691-700 (BUCIIANAN 1985: 264); 696-734 A.D.(PRAMD1988:
pl. 673).
223 Therc is a clear tendency to date this temple in the early 8th century: after 696
(DIVAKARAN 1971: 73); 681-690 (RADCLIPPE 1981: 567); 700-710 (BUCHANAN 1985:
264); 696-734 A.D.(PRASAD 1988: pl. 669).
224 The date of this templc is even morc disputed than those of the Alanipur teni-
and large windows) and somewliat later
features (e.g. a large number of niches). The proposed dates are: early 7th cent. (1.K.
SARMA 1993: 352, 355); founded somc tinie after 631 (BUCHANAN 1985: 127-128);
~iiiddleof the 7th cent. (DIVAKARAN 1971: 73); 665-681 (PRASAD 1988: PI. 652); 690-
696 A.D.(RADCLIPPE 1981: 318). The latter autlior discusses so~iiesculptural fenturcs
this temple sliares with the Durga temple, Aiho!c, which support her relatively late date,
in a separate article (publislicd under a different namc, BOLON1986). However, the
sculptures appcar to have been inserted in [lie niclics and could be posterior to the Innin
structure itself.
225 BUCIIANAN (1985: 257-258) dates theprcikdrcr wall later (Vinaysditya's reign:
681-696 A.D.)than tlie main building (ibid.: 127-128: founded in tlie 630ies). Slie also
placcs some of the s ~ u l p l ~ r of
e s the lliain Structure, among wliicli are the Ya~iiaand
Kubcra figures discussed below, in the later period. BOLON(1986: 51) does not assume
sucli a large time gap between the temple (c. 700) and its enclosure wall (c. 700-720
 Regarding the depiction of directional guardians, the evidence
is rather meagre. Most of the niches contain standing male figures,
which cannot be identified. They most likely do not represent
lokn/~nlns.In the second niche from the east on the southern side, panels, some individual variations have been introduced. As on the
a small, later inserted image of Yama survives (fig. 85). In height
(90 cm) and also in some other details, the sculpture conforms of Airiivata, but female attendants have been added (BUCHANAN
closely with the Yama images at Alampur. Yama's crown with a
hatched design resembles that of the same god on the BIla Brahmii
temple (fig. 66). The layout of the panel and the separate socle of
the record-keeper Citragupta are strongly reminiscent of the ViSva
Brahmii temple (fig. 801, however. Its sculptural style (stout figure,
soft, sensuous modelling of the features) and the ornaments of the
deity are shared by the other niche figures at Ku?aveliZZ6, thus
suggesting that it was no import from AIampur. The same charac-
teristics are also shared by a fragmented Kubera (fig. 86)227 at the
Alampur Site Museum, which has already been attributed to
Kudaveli by BUCHANAN (1985: 257). The single-row pearl neck-
lace, the spiral-shaped keyurns (upper arm bracelets), the
briihmanical cord, the band across the belly (udarnbandlzn), and the
semicircular line which indicates the paunch are similarly rendered
on the Yama icon. Only the backslab is partly missing in Kubera's
case. Kubera's attributes are the club (r., broken) and a sack
(l., broken).
After having identified Yama and Kubera as matching pieces, a
possible reconstruction of their original positions will be attempted.
BuCHANAN (ibid.) speculates that they may have belonged to an-
other temple in the vicinity. I am inclined to believe that Yama
and Kubera were the only directional guardians present on the
SaligameSvara temple, as on the Kumiira Brahm3 temple, AIampur.
Their place may have been either in the smaller niches flanking the
entrance, where the tzirlltis (dwarfs symbolizing particular treasures)
would have been later inserted, or they may have belonged (0 the
separate shrines on the south and north, attached to the enclosure
wall.
On this wall, the only member of the lokal~iilngroup is a small
Kubera (fig. 87) on the northern face near the north-western corner, Agni (fig. 88) stands in the rigid posture sanznlxi~insthiitlnIca,with
which corresponds to the position of Kubera on the Alampur tern-
pies. Kubera's attributes are a club (r.) and a sack (I.). His coiffure huge flaming bow. Near the right arm, which should have held a
of "Buddha-like" curls also resembles that of the Alampur Museum
piece (fig. 86), where the curls are visible behind the diadem. More
dikl-'61aswere not depicted on the compound wall, which once more
confirms the rudimentary stage of this sculptural theme on the
Kudaveli SangameSvara temple. This feature may well be subsulned
the other "early" features of the temple, like, e.g. the absence lnetrically crosses the torso at Satyavolu. The long lower garment
of a bknniisn (see above).
~ ~ t y a v o lR6nznlirigeSvnrn
~l, te~~zple
satyavolu lies to the south-east of AIampur and Kudaveli in District
prakasam. It preserves the only examples of latitzn tCigara-~ikhnra rosary and a waterpot has been transformed by portraying him with
(Northern type) temples with rziratzcNz6ra plan (without ambulatory
path) in Andhra Pradesh (BOLON1988: 335). RAMARAO(1959-60:
78) dates this temple on the basis of an inscription in 7th century
characters on a nearby smaller shrine. However, the unusual layout
of the temple (ill. 131, with an oblong pillared hall (tna!z~lalm) in
front of the pOrticus(arztnr61n)and the sanctum (garbhagrlzn)' may
suggest a successive construction of the lemple in the (fig. 70) and Agni (lig. 71) on tlic Svalga Bralirna templc wear a slrn~lar~ y p cof
early 8th centuryzz8.
What is left of the six images On the mn!z<krpn shows that a rudi-
mentary set must have been represented, which followed a similar
distribution images On the the Brahmiimn!l<lnlJa.
However, the ambulatory path @rnclnksi!ziipatha) is missing, and
it seems that Nairyta, Varuna, Viiyu and Kubera were not included
in this set. Only Indra, Agni, Yama, and Sarya still survive intact.
Regarding the rendering of the figures and the composition of the
Svarga Brahmii temple at AIampur (fig. 70), Indra stands in front
1985: 261); Agni (fig. 88) is uniquely shown with bow and arrow;
Yama (DIVAKARAN 1971: fig. 43) rides his buffalo to the right as
observed at AIampur, Viiva Brahmii temple (fig. 80), and Kudaveli
(fig. 85), but has one extant female attendant; Sarya is also shown
in the company of other figures, riding his chariot (ibid.). On the
sanctum walls, an intricately conceived Daksinii~nilrti~ i v apanel
(DIVAKARAN 1971: fig. 44) also survives.
- window -
- Surya
A
Daksinii- Indra
miirti
Yama window Agni
111.13: Satyavolu, Riimalingeivara temple, plan (without later
t~~a!~~lnl~a)
a raised right hand (damaged) and an extended left arm holding a
rosary, is a floating arrow, pointing upwards. Agni wears a crown
of flames with large encrustations of gems. The broad necklace, the
band across the belly (uclnrnbnncll~n)and the thick briihmanical
cord (~ipavitn)are similar to the ornaments of the Svarga Brahmii
Indra (fig. 70). A second belt, meant for holding the quiver, sym-
(dlzotf) is held by a different type of girdle, but sashes hanging down
on both sides recall the dress of most of the lolcnl~iilasat Alampur
(e.g. Agni, fig. 71). On Agni's right stands a large spouted vessel,
very similar to the one held by the Svarga Brahmii Agni. It is
remarkable to see how Agni's character of a humble ascetic with a
a flaming bow and arrow. These devices give him a somewhat
22"e~ B. DAGCNS (1984. p115 399-41 I ) T ~ 11i1ck
C make-hke k c y r ~ ~(uppc1
~rs a ~ m
bracelets), colhng around thc arm, arc typical for thc Ku?,~vel~ \culpturc\ Thcy also
occur 0 1 thc imngca of the Durga templc, Alliolc 11 I \ rcmalkable Ihat Indra
jewellery, but W I I a~ th~nnerhcctlon
227 So far, I ~ imagc
C has becn ~dent~ficd a\ ~ ( I ~ ) Raksakablia!n
c I (M.A.W.KIIAN
97'l0 42)3 l"otecrive 'pirtt.
2 2 " u ~ l ~1985: ~ ~261.
~ For
~ the date, the end of Vlnay:id~tya's,A.D.696 ( h i d . .
259; DIVAKAIIAN 1971: 73). or Vijayjdltya's, A.,,. 733134, rcign have bccn I'avouretl
(BOLON1988. p]. 676). TIKground plan IS publ~\hcdin PRASAD 1983. I I ~1 0(W).
warrior-like character. The niche and socle of the panel correspond Position God Attributes Veliicle
to the Alampur idiom, the only difference being the lack of any 1 outer .Colthd. r. vayu standard? two addorsed
decoration at Satyavolu. antelopes
Unfortunately, the images of Indra and Yama could not be 2 outer .CokhZ, r. ISiina staff, top broken bull
studied with all their details. 3 outer h k h 8 , r. Kubera (probably broken) bird chariot
4 innermost .<okhd,r. Y ama ? (diminutive buffalo
size)
2.3.1.2 Central India Indra ? (diminutive elephant
size)
The evidence for complete sets or directional guardians from the G outer SnkhZ, I. Agni flower? sheep
period under discussion is rather meagre. So much so that it is 7 outer inkhd, I. Varu~!n noose jnlebl1n (water
doubtful, whether the scheme of eight deities (asfidikpiilas) had elephant)
already gained popularity during this period in Central India. It has S outer Snkhd, l. Nairrta (?) sword or staff bird (?) chariot
often been hypothesized that temples earlier than or contemporary or throne
with those at AIampur, possible models for them, IIILIS~have exist-
ed22Y A further speculation could be that extant monuments in
Central India merely exhibit the "provincial" development, with The obliteration due to erosion or whitewash, as well as the intri-
the exception of the grand Teli-kii-mandir, which may have con- cate, fine carving have resulted in the fact that Inany details cannot
tained the directional guardia~lsin its wall niches. It is very likely be clearly identified. The headdresses in most cases seem to be
however, that a separate development occurred in the Gurjara- turbans. Kubera (3) with his left arm complete, but his right arm
PratihSra region, starting with the directional guardians over the broken, can be identified by his large, obese figure. It could be con-
door at the very beginning (PalSri), subsequently occupying the sidered significant that he rides an aerial chariot similar to the one
kapili (vestibule) walls (at Amrol), which is still in the entrance of Nairrta (?; S), with whom he shares demonic affiliation. The
zone, and finally leading to the fully developed distribution of addorsed, leaping antelopes of VSyu (l)also belong to an aerial
asfadikpiilas in the corner niches of the ,,Harihara no. 1" at OsiSfi chariot (virillina) or a throne. All animal vehicles are shown in a
(Western India). Several temples with incomplete sets of direc- leaping or "flying" attitude with their legs pulled towards the torso.
tional guardians in the Gurjara-Pratihiira region seem to testify to Yama's (4) buffalo is characterized by mighty ribbed horns.
the relatively late development of ostadilcpiila representation in Yarna (4) and Indra (5) are depicted flanking the doorway on
Central India. their leaping vehicles, below the two groups of three guardians.
In contrast with the groupings encountered so far, neither the They are located directly above the parasols held by the attendants
staff-bearing deity DandapSni, as in the eastern parts, nor both the of the river deities, while the other lokol16las are situated above the
Sun and Moon Gods, as at MundeSvari and AIampur, seem to have parasols of the river goddesses. Since the figures of the two pairs of
been associated with the Central Indian lokapiila sets. The ~ i v tem- a females are exceptionally large, relatively little space has been
ple at Indor, Dt. Guna, preserving the earliest full set of a ~ ~ d i / ~ ~ ) i i lprovided
as for the directional guardians. The other themes included in
i n its exterior niches as well as the Ical>iliniches, has been plausibly the door decoration at PiiIBri are associated with the god Siva:
dated to after 750 A.D. From the architectural remains of the 7th-8th NateSa; the wedding of ~ i v aand Piirvati (Siva VaivShikamCirti);
centuries, it can be inferred that Western India was more progres- AndhakSsuravadha; and Lakuliia surrounded by four disciples,
sive than Central India. which are on the outside flanked by Brahmii (I.) and V i y y (r.).
With regard to the PalSri group, two important factors should
Paliiri, Si(1~llzes'varaTerliple be noted: The complete canonical cls!adikpiila group is represented
here, however, only four of its members (2,4,5,6) ride their tradi-
The site is located on the MahBnadi river in the Raipur district, tional vehicles. The particular deviations in the niode of con-
south-eastern Madhya Pradesh. The west-facing Siddheivara veyance are unique and were not followed in the later Central Indi-
teinple230 has been regarded as one of the later examples (c. A.D. an tradition. The "split alignment" of deities on both sides of n
675-700) of the brick temples of Dak,si~!a Kosala that arc dated central group above the entrance has a parallel in the roughly con-
between c. A.D. 600 and 725. The "stellate" (star-shaped) plans of
some of these nirar~~ll~iira (without integrated ambulatory) temples,
viz, at PalBri, Kharod, and Dhobini, are highly complex and
innovative. It is an important negative evidence that on those '?".g.: "a conspicoous similarity is noticed in the Pratiliira templcs of central
surfaces, which more intricately interact with space than simple India and the group of Navn-Bn~limbte~nplesat Alampur [... The latter] are quite sim-
ilar lo the templcs of the first phase of tlic Pmtiliirn period" (R.D. T~uver~t 1990: 3).
square or rectangular walls, the directional guardians have not been Naturally, one would cxpcct an influence in the architcclurc of Northern type tcmplcs
depicted. to run from tlie north to tlic south. Tlicrcforc, some Nonh Indian or Central Indian
This is also the case at PalSri, where eight dikpiilas simply flank anlcccdenls for the Alanipur teniples in ~ h south
c would be cxpccted.
the stone doorway in two groups of three plus two single ones231 23" D.M. STADTNER 1981; DBVA1988: 243-346. STAI).~NER (I98 I: 49) :lssigns
mther narrow time span for tlic tcmplc, c. A.D. 675-85, wliicli is extcndcd lo 700 by
(fig. 89). They do not appear in their ideal arrangemellt, according
DBVA(1988: 246), while DONAI.DSON (1985: 192) is of thc opinion that "stylislically
to the order of the directions. The visual concept is reminiscent of I...tlle SidtllicSvara and relatcd temples] could date from tlic 8th century".
the way in which directional guardians are employed on figural pan- 2" DEVA(1988:246) is not aware ofthc fact that two more guardians havc bccn
els, particularly at EllorS. Their characteristics and identification are scpalntcly dcpictcd, when Iic writes: "obovc tlie heads of llic river-godtlesscs li~rvesur-
as follows (fro111the proper right to the left side)232: vivcd six ligures of the DikpBlas, three on cucli side." STADTNI?II (l98 1: 52) rccognizcs
only one of the separately depicted loknl)riltrs, i.e. Indra.
232 STADTNEI~ (1981:52) coniniils sevcral errors in the identification of Ilie dircc-
tional guardians, wllicli seen1 to be ~nainlytluc to the facl Illat lic ignorcs tlic atlributes.
Only VByu (I) and Indra (5) havc bcen corrcclly identificd by Iiini.
temporary wooden ~ a k t temple
i at Chatriirhi/Himachal Pradesh,
but it has not yet been conclusively determined if they are actually
loknpnlns233.
Anzrol, RGnze$vnra MahcTrleva and Diinebabn Temples
The historical remains at Amrol, District Gwalior, were first noticed
in 1929130 (ADGSAR 1929-30: 13). The better preserved
RiimeSvara Mahgdeva temple233 (ill. 14; figs. 90, 91) is situated
2 km to the north-west of the present village, while the ruined so-
called Diinebiibii temple235 (ill. 15; figs. 92-97) lies l km to the east
of the latter. The Diinebiibii temple, which has completely lost its
vestibule portion and the major part of its superstructure, obviously
postdates the RiimeSvara Mahiideva temple by a few decades.236
We would prefer to date both temples in the first halr of the 8th
century, the RiimeSvara Mahiideva to the beginning and the
Diinebiibii temple to the end of that period.
The sculptures of the DBnebiibii temple are inferior in quality
and tend to be stiff and lifeless. A development of the RiimeSvara
Mahiideva scheme is seen in the increase in the number of lokapnlns
from three to at least five plus Siirya, who appears to be incorpo-
rated in this group.237 A further development is also the introduction
of karna (corner) niches adjoining the knpilt (vestibule) wall, the
latter now lost is suggested by the shape of the plinth. It would have
been of considerable assistance to know whether the lokap8la
images also occupied the kapili niches. The omission of even one of
them would have rendered the lokapiila group incomplete. How-
ever, such an incomplete set does occur on a similar temple at Diilig
(ill. 26; figs. 145-151), where the knpilt niches do not contain the
directional guardians.
On the basis of the temples at Amrol, a hypothetical develop-
ment of the representation of the a~fadikpiilatheme can be postu-
lated. The structural predecessor of the RiimeSvara Mahiideva tem-
ple is most probably the temple no. 23 or Durgii temple at Naresar,
Dt. MorenalM.P.,238 the only one at that site with karnn and kapili
niches - with dancing gnnn figures (dwarf followers of ~ i v a ) .
Except for the dimensions and the orientation (the RBmeSvara
Mahiideva is larger and faces east instead of west), the plans of both
temples are almost identical. However, in the RiimeSvara Mahiideva
temple, three of the dancing gana figures have been replaced by
directional guardians, while on the west-facing Diinebiibii temple,
all formerly ga!zn-containing niches, as far as they are extant now,
exhibit directional guardians plus Siirya. In all the above cases,
GaneSa (S), Kiirttikeya (E or W), and Piirvati-paiiccTgizitapas (per-
forming penance amidst five fires, one being the sun; N) occupy the
bhadra (central offset of the wall) niches. The probably slightly lat-
er temple at DZiig still has a mdimentary set of lokapiilas, that are,
however, confined to the kar!za niches only. Thus, we can trace an
independent development of a~~adikpiila representation in Central
India that differs from the concept found at Alampur, particularly
the tendency to accommodate the regents only in the karna niches.
Moreover, a development from the seated (RlmeSvara Mahiideva)
to the standing (Indor, Dt. Guna/M.P.: ill. 27; figs. 152-159) type
of lolrnp6ln can be traced, the intermediate stages being manifest in
the Diinebiibii temple, Amrol, the temple at Diing, and the Sfirya
temple no. 1 at Osiiiii.
During the earlier explorations by the Gwalior State Department
of Archaeology under D.R. Patil, most of the loknpdlas were cor-
rectly identified: Agni, Yama, and ISiina for the RiimeSvara
Mahiideva temple, and Viiyu, fSiina, Siirya, Indra, and Agni for the
Diinebiibii temple (ADGSAR 1942-46 [1949]: 20-21). Here and
in subsequent descriptions of the latter ternple,239 the still extant
Kubera image in the (north-)west karnn niche has not been men-
tioned. Starting with the RBmeSvara Mahiideva temple, the charac-
teristics of the directional guardians are as follows:
Gana Piirvati Gana
Gana
Kiirtti-
keya
Gana I Agni
2 Yalna GaveSa Gana
Ill. 14: Amrol, RiimeSvara Mahgdeva temple, plan
1 Agni (fig. 90), southern lcapili niche: seated on a couchant
goat in (~~l~i!ziipralnnzbapddn (right leg pendent), wearing
jariinziiku~a(hair-crown) and a broad decorated sash across
the obese torso, holding a water flask in the proper left arln
resting on the knee; he most probably held a rosary in the
damaged right hand. His body is tilted to the (proper) right,
and the usual flames - here in "bundles" - appear behind his
head and shoulders. Agni, like f~iina,wears a twisted sash
around his knees. This portion is damaged in the case
Yama.
2 Yams (fig. 91), south(-western) k n ~ niche,n squatting On "
couchant buffalo with knees turned outwards, is badly dam-
aged. He holds a staff (&!zda) in the left hand (damaged).
3 i62na (VIENNOT1976: ph. 54; originally four-armed?), north-
ern kapili niche: seated on a couchant bull in v6iizal2ra1anz-
b ~ p c d a(left leg pendent), with the face damaged like the
other lolcapalns, wearing jn;im~lkufn,necklace, earrings, and
a brzhmanical cord; a sash of twisted cloth is visible near
right hip and extends to the left knee. He holds a trident in the
Proper left hand which rests on the left knee. The right band
held near the right shoulder is badly damaged.
DEVA1991: 102; PI. 209, Tlic two groups of deities arc shown seated on both
sides of a sct of eight planetary dcities, i.e. two on the left and three 011the right side'
Only the second figure frorn the right is identifinble as Viyu, since he holds a
ing cloth. Below the figure on thc cxtrclnc sight, Indra is depictcd as the uppermost
figure of a vertical align~ncntof standing deitics, thc I.e~nainderare however noL
tiollal guardians. Apart from V5yu and Indra, 110 other [okopcilas can be identified
probable date of the teinple is 710 A.D.
234 ADGSAR 1929-30: pl, 111; ADGSAR 1942-46: 20-21; R.D. TRlVEDl 1990:
70-75; DEVA1991: 12-15.
235 ADGSAR 1942-46: 21; TRIVEDI 1990: 86-89; DEVA1991: 15.
'" (1991: 15) believes that both temples belong to the first quarter "le
8th century, while TRIVEDI (1990: 75; 89) datcs the Rgtnc$varaMah$.deva telnple "le
middle and the DBneb5b5 to the end of the 8111 century. VlENNOT (1976: B 7 ) 'ls0
assigns the D$.nebXbB temple ("T. ruin6 ou no. 2") to a later phase of her relative
chronology.
237 For reflections on the incorporation of Silrya or Candra into thc group of direc-
tional guardians SW C. WESSELS-MEV~~~~N 1994: 607-609.
'" Tluveor 1990; 59-63; DEVA 1991: 11-12.
~ 1 88) describes (he deity as a "standing Varuca", while DnvA
2 3 " ~ ~ ~ ~(1990:
(1991: 15) fails to mcntion it at all. He starts with Vjyu in the adjoining niche.
011theDiinebiibii temple, the deities are arranged in the following As a general rule, all the seated deities, except for Vayu who raises
manner: both hands, raise the respective hand, which is opposite the vehi-
cle's head. This concept was obviously employed in order to create
a visual balance.
5 Viiyu Piirvati G Soina?

The well-known, earliest extant temple at Gwiilior, now known as

Teli-kii-mandir, overlooks the Gwiilior fort on the Gopagiri Hill.
Unfortunately, it is bereft of most of its sculpture, and nothing has
I Kiirttikeya been preserved elsewhere. Only small relief figures remain on the
building, the most interesting ensemble being a frieze of 1 13 figures
on the basement (ndlzighE~m)portion. Relying on the palaeography
of some inscriptions on the temple, among other indications, the
Teli-kB-mandir has been assigned by DEVA~~O to abo~itthe middle
of the 8th century A.D., and by T R I V E D to
I ~the
~ ~ latter part of that
8 lndra
century.

GaneSa 9 Agni
4 Kubera
Ill. 15: Amrol, DiinebiibB temple, plan 1 5 Kubera
I I r..

3 VS!

4 ILubera (fig. 92), (north-)western knr!zn niche: standing in 2 Varuna

dblzarign (body slightly flexed), with a damaged face like all
the other lokal~iilnsand a badly eroded body, has an obese,
stout figure. He wears a japinzukuta, earrings and necklace
damaged, and keyurns (upper arm ornaments); with an ele-
gantly swinging money sack held in the proper left hand near
the left thigh, and the right hand resting on the belly proba-
bly holds a small cup, attended by an animal-headed small
figure (probably his vehicle) standing on his proper right
side.
5 VByu (fig. 93), north(-western) Icnr!zn niche: badly eroded, \
seated on a crouching" antelope (?) in va~izn~~rala~~zbal~iida, l Agni
wearing jatiinzukl~tn,and holding a billowing cloth (vdyu-
111.16: Gwiilior, Tell-kii-mandir, plan (after DEVA1991: fig. 7 )
Epilritnvnstm) in both upraised hands.
6 Sonia? (fig. 9 4 , north(-eastern) knrtza niche: seated on a
couchant quadruped in ~~ak~i!~~pralnnzbnl~Edn, wearing
There is no indication whether directional guardians once occupied
jntdr~zulcu/a(damaged), earrings, necklace, and keyuras; the
the corner shrines, which form a special feature of the temple. The
proper right hand is raised, holding a rosary, and the left one
resting in the left knee, holding a waterpot. images of the frieze on the mztara1)nffnmoulding are the only source
of information as to whether directional symbolism was intended at
7 Siirya (fig. 9 9 , (north-)eastern knrtza niche: standing in a
rigid snnznpr?& posture, with the high crown damaged, wear- Gwiilior. Among the 113 figures, which are all in a fairly good state
of preservation, four or five loknpdlns can be identified: Agni ( l ) ,
ing ornaments, the long sash typical for Sfirya, a breastplate
Varuna (2),Viiyu (3),Kubera (4,5; depicted twice); and probably
over the breast, a long lower garment, and boots, holding two
SomdCandra (6). It is a noteworthy fact that the lokq5ln group,
lotus flowers, accompanied by his usual attendants Danda
which also included guardians of the intermediate directions,
(proper lelt) and Piilgala (proper right).
remained incomplete. The guardians' positions are generally cor-
8 Indra (fig. 96), (south-)eastern knrtzn niche: seated stiffly on
rect, but they have been placed in a rather irregular and unsystenl-
his couchant elephant in vii~~la~rnln~lzbal)dda, with his cylin-
atical manner. It is unlikely that Candra was regarded as a genuine
drical crown damaged, adorned by a thin neclclace, earrings
member of the group of directional guardialls at Gwiilior, but the
and keyuras. The raised proper left hand must have held the
deity may be regarded as associated with the group here. The
thunderbolt, raised to the shoulder, and the right hand is rest-
ing on the right knee.
9 Agni (fig. 97), south(-eastern) kartzn niche: seated on a com-
paratively small goat in (laksi!zdl~rnla~~~ba~~iida,with neatly 240 DEVA 1985: 162; 1991: 17.
241 TRIVEDI 1990: 95. As in tlie case of Amrol, TRIVEDI hcre prefers a later date.
arranged flames encircling the head, has a ja/iinzukuta, and
In the opinion of most scholars, the Teli-kj-mandir should be posterior 10 Itle
ornaments like the other guardian deities, a broad brahmagi- RjmeSvara Mahjdeva temple at Amrol, although there is no actual evidence otlier tlian
cal cord, and a yognpngn ("yogic band") around the knees, its no nu mental size and intricacy in stlucture and decoralion to sub~tantiatethis. Keep-
holding a small bottle in the proper left hand resting 011 the ing in mind that the RimcSvara Mahidcva temple was "seemingly built by a guild relat-
left knee and probably a rosary in the damaged right hand ed to tlie one which collstructed the Teli-ki-mandir" (DEVA1985: 162), both temples
could be more or less coeval.
held at shoulder level.
guardians are all shown standing, with the exception of the addi-
tional Kubera (5). The figures show a sensuous modelling of the
body. Even the roll of fat just above the lower garment has been
depicted, similar to that on the RiimeSvara Mahsdeva temple,
Amrol. This feature also occurs on the Diinebiibii temple at Amrol.
The sash worn around the legs by Agni (I)and Kubera (5) is again
paralleled by the RiimeSvara Mahiideva lokapalas Agni and ISsna.
The genuine directional guardians on the Teli-kii-mandir (1-4) all
occupy knrr~apositions. Their faces are damaged.
1 ~ g n(fig.i 98), arztarnpa@ frieze, no. 19, south(-east): stand-
ing in dbharig~(slightly flexed) with his body tilted to the
proper right. He wears ja!dnzsLkll!a (hair-cr0wn), a simple
necklace, earrings, a faintly visible briihmanical cord and a
sash tied around the legs with an ornate kllot. Agni seems to
hold a water bottle (1.) and a rosary (r.; both damaged);
bundles of flames are visible behind the head; the goat
cle sits to his proper right.
2 Varuqa (fig. 99), nntarapntp frieze, no. 69, (north-)west:
badly damaged; Varuna's PostLlre is similar to Agni's, but it
already represents an extreme flexion of the body (atibhfi-
ga). He stands in front of a large goose (Iznnzsn), which he is
feeding with grass. In his raised right hand is a noose that
curves to his proper right. Varuna resembles Agni in his
coiffure and the modest ornaments, among which are keyurns
(upper arm ornaments).
3 VByu (fig. loo), aiztarapnttn frieze, no. 70, (north-)west:
This is the most dynamic representation. VByu stands in an
extremely flexed (atibharign) posture, the flexion being in the
opposite direction to that of Varuna. His lower garment and
his ornaments correspond to those of Varuna, but his coiffure
is the demonic clrdlzvakeia (raised) hairstyle. Both hands are
raised and grasp the cloth floating over Viiyu's head. A
caparisoned horse stands behind the god, with its head turned
towards him.
4 Kubera (fig. 101), arztarqat;a frieze, no. 75, north(-west):
The figure is also badly damaged. He can be identified o~lly
on the basis of the remaining outlines of a long sack which he
holds in his proper
- left hand. The posture resembles that of
Agni and Varuna. The hairstyle may have been ja!~iinuku~a.
The proper
- - right- hand is raised and holds a dish or cup.
5 Kubera:':,seated (fig. 102), mztarapntta frieze, no. 80, north,
facing west: The deity has a paunch. He is seated in
clnk;i!zdl~ralanzbapdda, with his left leg on a cushion. His
hairstyle is the demonic clrcllzvalceSa. The necklace is longer
and more ornate than that of the above figures. An elegant
sash is wound around the right leg. The proper left arm rests
on the lower leg, while the right hand, which is raised, holds
a drinking bowl.
G Candra (fig. 103), aiztnrapn![a frieze, no. 113, east, north of
entrance: stands in a rigid posture, wearing a cap or crown,
large earrings, a double-row necklace, keyuras, and the
briihmanical cord. A rather thin crescent appears behind the
head. His proper left hand may have held a waterpot, his
raised right hand could have held a rosary (both damaged). A
stylized shrub can be seen on his proper right.
We cannot conclusively state whether this incomplete, haphazardly
positioned set of directional guardians on the Teli-kii-mandir still
belonged to the early phase of a gradual introduction of the lokal~nla
set, in view of the evolved and carefully planned structure of the
temple. The general course of development in Central India makes
it very likely, however, that the depiction of the complete group
eight directional guardians had not yet become a common practice
in the region.
2.3.1.3 westernIlldia
someimportantevidence ofthis period colnes from osi2fi. AS in the
other regions discussed so far, some variations in the treatment
the dikl7r7la subject are clearly discernible. Some partic~llarl~ tricky
identifications of LlikIJclnshave not beell solved by other authors so
far (,.g., cf. c . p , A T ~ ~ E R1997:
T ~ Nfig. 5 with ill. 21). Apart from
the culmination in the aesthetically balanced complete set at
osiiifi, H~~~~~~~ temple no. l, near the end of the period (i.e. 750
*,D.), Some valuable and ullique evidence has been preserved in the
two temples at Cittaudgadh, the Kiiliksln&=, and the ~ ~ ~ b h a i y g m a .
The date assigned to tllese telnples has been pushed back
and that of the Kiilikiimiiti in fact to the 7th century.242 ~ittaudgadh
most probably represents a more genuine regional developlnellt
than Osiafi and pipad, wllich appear to have received impulses
from Central India. The three ilnages of directional guRrdia~1sstill
remaining at Bithi1 resemble those at Cittaudgadh. They are all
shown standing, unlike most of the clikl~8lasof Central Indian
inspiration.
Cittfll~~~g~gn~~Iz,
Kiilik?irizBt~iarzd Kunzbkaiyaiiza Tenzples
The Cittaudgadh fort in Rajasthan preserves some early evidence of
temple architecture. The large sdrzcl/zdra (with integrated ambula-
tory) temples, nowadays known as K$lik2miiti and KumbhaSy2Ina9
contain an i~npressive wealth of architectural and sculptural
including representations of the directional guardians. The
size of the temples and some of their features which appear to be
more developed,244 make it difficult to believe that the ~ g l i k g m ~ ~ '
temple belongs to the latter part of the 7th and the K u ~ n b h a S ~ i i ~ ~
temple to the first half ofthe 8th century, as they have been assigned
However, the fragments of the rather securely dated
(636137 A . D . ) temple
~ ~ ~ at Kusumg show that the art formulas
242 The period of 643-742 /\.D.(bascd on the Saitkar gh@ inscription, MEISTER
1975-76: 44; 1991: 273) forms thc tinlc span for the const~-uctionof thc ~ j l i k i m s I" t~
this publication, tlic author narrows this period to laltcr part of the 7th century
which, in his opinion, is corrobornted by mason's marks on the telnple wall (ibid.: 291).
LIN-BODICN (1980: 60) suggests that thc te~nplcwas built "probably in Lhc sccolld
tcr of thceighth ccntury", on thc-rather shaky-basis ofso~lleobscrvatio~ls011 the gen-
cral stylistic dcvelopiiient.
'" M U S T ~1991:
I ~ 285-297. In alalerpublication, the authorcompareS ~ ~ C S ' O U " *
plans of both templcs and discusses some poitlts of particular intercsl(l993: 346-350;
fig. 9).
244 VICNNOT (1976: 248), on the basis or particular architcctu~lfealures, lists
three tclnples at Ciltau<lga?h as belonging to tllc sixth period of her scheme. Most
the other temples assigncd to this period belo~lgto the 9th century. VIBNNOT'Scatego-
rization is rather inexacl, as is revealcd by thc fact tl~att{>ethird te~npleat cittaudgadh.
lhc Ksemaitkari ("le~i~ple oti borrl bnsshz rle~,anrLe KilikZ ~ i f c i "MElSTER
; 1991: 302-
304), which is clearly latcr than lhc Kiliki~nitiiand KLllnbllaSy3~na temples, is reck-
oned under thc same category. HARLE(1986: 145), who refers to the assignment of Ihe
latter tc~nplcsto the second quartcr of tl~c8111 century, sceptically remarks: "If so,
new kind of~iiche[with rolllid pilasters and an awning] appears llerc at least three cluar-
LcrS of a cenlury earlier than at Osiah,"
245 ME~STBR 1991: 291,297; cf. also 273. CC. note 141.
21"lle Collllcction betwce~lthe datcd "KutSiSra~~a inscription", preserved at the
Magdor Museum, and the extant Riimacandraji temple, however has been doubted by
C.G. LIN-BODIEN (1980: 50). She tentatively assigncd thc tcmple, on stylistic groullds,
to the latter par1 of the 7th ccntury (ibid.: 60).
 particularly reflected in the KBlikImiiti,247 both in sculpture and
. architecture, were prevalent in Western India during the early 7th
century. It is a remarkable fact that a number of differences, includ-
ing the rendering of the dikl~dlas,separates the Kalikamiitii from the
KumbI~aS~iima temple, which are nevertheless very closely related
structurally. In this connection, MEISTER(1993: 347) has observed
that:
In plan and architectural detail, to some degree in iconography, and
to a much lesser degree in sculptural typology, these two temples
exemplify a distinction made some years ago by M.A. Dhaky
between styles in parts of western India to which h e gave the geo-
graphic names, 'Mahl-Maru' and 'Mahl-Gurjara'.
The directional guardians on the K151ikiimBtii temple OCCUPY the
1car;za (corner) niches of the garblzagrlza (sanctum). They represent
a complete group of eight, however with SomdCandra replacing
Kubera on the north, which is a rather rare occurrence.
6Vayu Silrya 7 Soma
7
5 Varuna 1 8 iSsna
3Y a m S~irya 2 Agni
111.17: Cittaudgadh, KHlikZmltB temple, gnrblzagrl2a plan
Additionally, some ofthe djkpdlas appear in a symmetrical arrange-
ment above the garblzagrlza door (see below).
The Moon God Soma, according to some traditions is the sole
Guardian of the north (see section 1.3). This fact as well as the fact
that the temple was originally dedicated to Silrya, who is likewise
regarded as a "planet", could explain the preference for Soma
instead of Kubera.
Most of the directional guardians are two-armed, except for
Agni and IiBna, who are four-armed. A characteristic feature of the
Cittaudgadh cliklldlns is that they stand with their feet apart when
are not represented in the rigid san~nl~Zdn posture. Vsyu (fig.
108) is a particularly good example for an extreme form of this pas-
ture. The free leg is stiffly stretched to the side, the foot turned out-
wards. This posture is typical for Western India.249 In Central Illdia,
the legs of lokapdln images tend to be placed close together.
h o v e posture, which could be referred to as tribhalisa (triple-
flexed, if one accepts this expression), has been somewhat modified
in the KumbhaSyZma sculptures. Here it is even more exaggerated,
'esulting in a kind of mannerism. The free leg in that case is
Stretched out at a wider angle and has to be bent in order to reach the
ground, creating a rather inelegant effect.
1 The image of Indra on the (south-)east is unfortullately not
documented.
2 Agni (fig. 104), south(-east), is shown four-arn~ed,wearing
ja?Znzuku@ (hair-crown), a moustache and goatee, and stands
in the rigid s a r l l ~ p ~posture.
& He is adorned with heavy ear-
rings, a necklace, bracelets and a valzalizdlii (garland of leaves
and flowers) reaching down to his knees. Strangely, his
upavita (brahmanical cord) seems to be wound around his
belt buckle. Behind Agni's head are two concentrically ar-
ranged, elongated, flaming haloes. His upper right hand holds
a fan, which was used to kindle the fire. The lower right hand
is damaged; it may have held a fruit. The lower left hand
holds a pot with handle, and the upper left hand holds a long-
handled attribute resting on the ground, possibly a standard.
On Agni's lower right is a ram crouching on the ground. Two
plants symmetrically flank the god.
3 Yama (fig. 105), on the south(-west), stands in the charac-
teristic posture described above. He has a curly hairstyle,
probably representing the demonic Irdlzvakeia (upraised
hair). A perfectly round halo is visible behind his head. His
ornainents are similar to those of Agni, except for the pres-
ence of anklets. I11 addition, he wears a string of pearls around
his neck, with chains hanging down from the girdle over both
legs. He also wears keyuras (upper arm bracelets), and a sash
that falls down across the thighs. The ulxzvita is worn in areg-
ular way. Yama holds a staff topped by a skull in his right
hand, and in his raised left hand a large lotus flower with a
bird perched on it. A massive varzallzdlii extends down to the
knees. The diminutive buffalo is depicted standing behind its
master.
4 Nairrta (fig. 106), (south-)west, is a mirror-image of Yama's
posture. He wears a high crown with an intricate halo behind.
On the whole he resembles Yarna, however, he has a third
necklace with little bells, and the varzmizfild extends even fur-
ther down. The god wields a sword in his right hand, and the
left hand is resting on the sash which runs across his thighs.
A tiny figure of a worshipper on his right probably represents
his anthropomorphic vehicle, while a female attendant is
standing on the other side.
5 Varuna (fig. 107), (north-)west, whose head has been com-
pletely destroyed, resembles the other dikpiila images (here
only one necklace), except for one unusual characteristic: his
upavita extends from the right shoulder down to the left side.
The brihmanical cord is worn in this way, as opposed to the
regular manner, only during rites connected with the dead, a
connection which cannot be established for Vamna. The staff-
like attribute in the right hand is badly damaged, but it seems
to be a specific type of noose. Where the old surface is pre-
served on the "handle", a twisted structure is discernible. The
upraised left hand holds a lotus flower, which is curved
towards the face of the deity. On the bottom, on the proper left,
the gaping mouth of a nzalcnra (mythical aquatic creature) is
clearly discernible. Except for one visible paw, the rest of the
body is made up of large scrolls, seen on Varuna's left.
6 The image of V2yu (fig. log), north(-west) is particularly
impressive for its beautifully flowing shawl attribute. Unfor-
247 MEISTER (1975-76: 44) writcs: "Overriding points of similarity exist betwcen
Kusu111,iand the K,ilikZ-mBti temple at Cllittor [...l, yet in term of idiorn they are far
apart." He does not specify the points of similarity. Collcer~lingthe sculpture, the soft,
flowing contours of the bodies, the squarish faces, and the seemingly boneless, elastic
fingers (ibid.: fig. 30), clearly connect both templcs.
248 Unfortunately, the image of Illdra could not be studied.
249 111the following, it is assulncd that the posturc with the foot of the free leg
sharl~lyturned to the side originated in Wester11 India. So~lleearly examples are
illos(rated in SCIIASTOK1985: PI. XIII, figs. 21, 22 (doorkeepers fro111~,imal,iji,dated
c. 539 A.D.); pl. XXXI, fig. 52 ( h a , next to Brahmi~!i, perhaps from Tinyoi, dated
c. 550 A.D.).
 tunately, his head has also been completely damaged. VByu's
right hand, raised to shoulder level, holds the broad piece of
cloth, which flows behind his crown and is wound around the
left forearm. The ends of the cloth are attached to the pillars
flanking the niche, in the form of volutes. The left hand rests
gracefully on the thigh of the free leg, just above the hori-
zontal sash. On VByu's right is an antelope with straight horns
standing frontally. On the other side, a kneeling worshipper
is visible.
7 Soma (fig. 109) on the north(-east) has a serene expression.
He stands in the rigid samapiida posture, the right hand hold-
ing a rosary. Behind the huge jatiinzukuta coiffure is a large,
undecorated halo. He wears enormously large circular ear-
rings. A peculiar feature is the unusually long upavita, even
covering the sash behind which it disappears. The sash worn
over the legs is tied in a big knot beside his right thigh.
Soma's left hand extends downwards, grasping the neck of a
pot (gesture and pot are similar to that of Agni, fig. 104). Tiny
figures of worshippers kneel on both sides of the Moon God.
Just above the figure on the proper right side is the head of an
animal, probably a dog. The body of the animal is unnatural-
ly elongated, as the hind legs are visible on the left beside the
god's legs. A predecessor of this image of Soma at the relat-
ed site of KusumHzso was originally paired with a SGrya image
(MEISTER1975-76: fig. 32).
8 ISFina (fig. 110), on the (north-)east, is again four-armed. He
wears jnfiimukuta. In posture, ornaments and the position
of his upavita, he resembles most of the other dikpiilas. The
lower right hand elegantly holds a small flower or bud and is
uncharacteristically placed in front of his chest. The upper
hand grasps a trident, with its upper part visible behind the
pilaster capital, echoing its form. The upper left hand holds a
serpent, the lower part of which is wound around the wrist.
The lower left rests on the sash falling over the thighs. Two
dwarf figures flank h a on both sides. The one on the right
seems to carry an offering for the god, while the one on the
left feeds the bull vehicle with balls of sweets. Among the
pilasters flanking the deities described so far, IS3na's are the
most lavishly decorated ones. In contrast to the other ones, the
shafts are completely covered by ornamental leaves, and
below the upper ribbed zone a row of lotus petals embellish-
es the capital.
In the KBlikamBta temple, a few images of directional guardians
appear amongst a large number of figures on the garbhagrha (sanc-
tum) overdoor relief (MEISTER1991: 290; pls. 655, 657). Here,
Yama is found paired with Kubera, who is excluded from the group
of dikpiilas in the niches. On the opposite side, Varuna appears
together with VByu, and Indra seems to appear twice, i.e. once on
both sides. Both pairs correspond to those generally found among
the group of Manu Lokap3las (cf. section 2.4.1.3).
On the KumbhaSyBma temple, two differently composed sets of
directional guardians probably illustrate an incipient stage in the
development of the theme. The interior set in the karna niches of the
sanctum, consisting of nine directional guardians, represents a
unique scheme. Only three dikpiilas show individual characteristics,
thus distinguishing them from the "ISBna type" which is rendered
with or without a bull vehicle. This puzzling set has been com-
mented on by MEISTER(1 991: 295-296) as follows:
Both of the east-facing dikpBlas have no vihanas. All have four
arms and hair tied into a jatimukuta. Only Niryti, Varuna, and Vsyu
have been differentiated by attributes and vBhanas. The remainder
take a form suggestive of ~ i v a - I i i n a[...l The intention of this
imagery would seem to be to suggest that all ala~nkira-dgvati~ .
["ornamental" attendant deities] are manifestations of ~ i v a-
including Dikpiilas and apsarases [heavenly nymphs] - as well as
the "isfadEvatBs" on the bhadras (a fact further emphasized by the
repetition of ISina on the northeast pratiratha, where elsewhere only
apsarases have appeared).
The autho! further suggests that the original concept must have been
that of a Siva temple, which was converted to that of a Visnu tern-
~ 1 after
% a temporary suspension of work some time in the first half
of the 8th century (ibid.: 297). The identification of the guardians in
the western k a r ~ aniches as Nairrta and Varuna is rather doubtful,
as their sets of attributes do not follow those traditionally described
for these deities. All the guardians have been portrayed with relaxed
postures, the free leg almost invariably directed towards the corner.
917 The karga niches of the east wall are hardly accessible. The
directional guardians there belong to the simple variant of the
L'-*-
Isam type, without vehicle", resembling the image of Iiftna
of the K5likBmSta temple (fig. 110), except for the vehicle and
the attendants. The sculptures have been executed in a rough,
simple manner. The (north-)eastern image is the better Pre-
served one. The distribution of attributes and gestures seems
to be the same for both images: lower r. - lotus; upper r. - tri-
dent; upper 1. - serpent; lower 1. - katihasta (resting on the
hip).
I
14 VByu 15 h a 16 ISBna
B type type
9 ISina
type"
111.18: Cittaudgadh, Kumbhaiyfima temple, garblzagyha
(sanctum) plan
(B -central bhadra niche [empty]; * without vehicle)
I O $ I I , On both the south (figs. 111, 112) and the north walls
15~16(MEISTER1991: pl. 676, 1st and 3rd image from the right) of
the mnctum are found two i&na type dikpiilas. The south-
ern ones have an additional attendant with a trident on their
Proper left. Their postures are exactly symmetrical (i.e. rnir-
rored by the counterpart on the opposite side), and the detail
of the k ~ e garment
r being shorter on the side of the free leg
The figure was erroneously identified as Kubera by MEISTBR (1975-76:
33). Both SOrya and Candra are shown in situ on an earlier pholograpll (MelSTcR
74: fig. 12). The Moon God resembles his counterpart on the KBlikjmjtti [emplc wit"
respect to his coiffure, round halo, and the attributes (there must have been a rosary in
his right hand). The vanatiicilci is also present. The dcity is seated without vehicle and
holds the vessel from below. The form of the pilasters flanking the niche has survived
in the architecture of the Kjlikfimlt~and KumbhaSy,ima temples.
 also occurs in both cases. The distribution of attributes in the
upper hands (r. - trident; 1. - serpent) is the same as for the
other "PSBna type" dikpdlas. The head of the bull always
appears on the proper right side. In ornaments and dress the
guardians largely resemble those of the KBlikBmLtB temple.
The south(-eastern) guardian (fig. 111) has his front right
hand raised, and holds a rosary and lotus. The front left hand
is held in katihasta. The south(-western) dikpala (fig. 112)
has his front right in the latter position, his left hand being
raised and gracefully holding a small lotus.
12 The ~ a i v aDikplla (fig. 113) on the (south-)west is four-
armed and wears ja~Znzu1cu;a (hair-crown), like all other
dilcpdlas on the sanctum. The lower right hand is partly bro-
ken. It could have been held in the rather unusual gesture
varacln (bestowing boons). The upper left hand holds an
indistinct object, probably ~lamaru(small drum), the upper
right grasps a snake, reminiscent of the "&B'na type". The
lower left hand holds a trident (with an additional skull like
his attendant on his left?). A lion vehicle in a relaxed,
attitude crouches on his proper right.
13 Varuna? (fig. 114), on the (north-)west, is characterized by
a makara or fish. However, there is no other clear indication
of his identity. The coiffure also seelns to be ja~dinukufa.The
attributes are: lance (lower r.); lotus (upper r.); rosary (upper
l.); kafihasta (lower I.). The absence of a noose makes the
identification as Varuna doubtful. The deity is flanked by
attendants on both sides with one behind the fish vehicle on
the proper right. The relatively small mouth of the animal and
also its strange vertical position suggests its identification as
a fish rather than a nzakara. This has also been mentioned as
Varuna's vehicle in the literature (MtP 286.94.
14 The information available on Viiyu could only be taken from
MEISTER(1991: 296). The deity probably has an antelope
vehicle as in the KBlikImBtI temple. He holds a flag here,
unlike the image in the latter set.
The outer walls of the temple which is now roofed by a much later
superstructure, preserve the origillal set of eight directional guar-
dians. It is unusual in that it incorporates KBrttikeya, probably orig-
inally on the north(-west).251 The figures are (clockwise, starting
from the [south-least):Indra, Agni252, Kirttikeya,Naiwa, V a m ~Yams,
, 21 Nairrta
Kubera, ISBna. Some of the images appear to have been partly recut
in the 15th century (MEISTER1991: 294). The images of this set SO
closely resemble those of the first one, that chronologically, both sets
do not seem to be far apart. However, there is a slight tendency to-
wards some awkward features like squarish faces, narrow shoulders
an unnaturally flexed free leg. The background of the panels is
often blank. The generally simpler conception is also visible in the
fact that all deities except for Ksrttikeya and &na are two-armed.
18 Indra (fig. 115) stands in a posture similar to that on the
K B l i k i i ~ n temple
~ t ~ and wears a large cylindrical crown. He
holds a large v a j r ~(thunderbolt) with the right hand at waist
level and has his left resting in katilzasta (hand on the hip). A
miniature elephant on his left is the only attendant figure.
19 Agni (fig. 116), who is - at least at present - placed ir-
regularly on the south(-west) karrza (corner) niche of the
giiflzama!zdaI~a (pillared hall and ambulatory), s ~ m to s 253 from the images on the exterior of the KumbhaSyiima temple it is clearly vis-
have a recut face. He is bearded and wears ja@nzukuta (hair-
crown). His right hand at waist level holds a staff-like attrib-
ute, which may represent a trirln!zcla (triple staff, attribute of
ascetics). The left hand rests on the hip which is the usual
posture of the guardians of this set (except for Varuna). A
goat's head is peeping out from behind his proper left side,
while a tiny male attendant stands on the other side. The
god's head is flanked by two large leaves. Vertically
arranged flames are visible above these.
20 Skanda (fig. 117) on the south(-west) is here also described,
although this deity rarely appears among the dikpdlas. He is
shown three-headed and four-armed. His flexion at the hip
is particularly strong (atibhariga). His three heads wear
ja;dnzukutas. His front right hand holds a long-shafted spear
and the rear left hand an object with a handle, probably a '
mirror. His rear left hand is held close to the floral capital of
the pilaster and grasps a fruit, probably a mango, which looks
like a part of the capital. His lower left hand rests on the hip.
A female figure, probably his wife Devasenii, stands on his
left side. On the right side beside Kgrttikeya's lotus socle are
the remains of his peacock.
21 Nairrta (fig. 118), located on the (south-)west, probably has
a recut face and crown. He holds a - largely broken - sword
in his right hand. His anthropomorphic vehicle kneeling on
his proper right side looks up at him with its hands folded in
aii,ialirnudrE.
22 Varuna (fig. 119), like Naiqta, remains in his proper posi-
tion on the (north-)west. His face and ja.tLfnzuku/a seem to
have been recut. In his right hand he holds a stiff object,
obviously a stylized noose. It is not complete on the left side,
and it seelns to have been cut to the proper size after the com-
pletion of the pane1.=53His left hand does not show kafihas-
ta, but is gracefully held beside the hip, probably in the
process of offering food to the nzakara with open jaws visi-
ble beneath.
23 Yama Kseman- Ardhaniri CBn1~1nd3 24 Kubera
kari
I I
22 Varuna 25 ISina
A giidlza-
ilin!zdapa
18 Indra
20 Skanda Lakuliia Harihara I9 Agni (Agni)
111.19: Cittaudgadh, KumbhaSyBma temple, simplified plan of
outer wall
251 Both this sculpture which is now on the south(-west) and the image of Yama,
now on the north(-west), violate a rule which appears to have been impleinented on both
temples under study: When the statue is not shown in samapddn (feet together), the free
leg is always directed towards the nearest corner of the structure. Only Kiirltikeya and
Yama of the KumbhaSyima are exceptions to this rule. Therefore it seems likely that
they have been placed on the te~nplein positions other than those originally planned.
252 A "151h century" figureof Agni now occupies the south(-eastern) kapa (MEIS-
TER 1991: 293) niche which must once have belonged to that deity. However, the old
sculpture of Agni is locatcd on the south(-west) k a y a of the gridliania!i~/apa. Probably,
this inconsistency testifies to a change in the concept of the te~ilpleduring its construc-
tion. A number of errors have occurred in MEISTER'S distribution table of images (ibic[.),
ible that the niche figures have been carved together with the right pilaster, the left
pilaster being most probably affixed later, ill situ. In such cases, there are often consid-
erable morphological differences between the left and right pilaster. In one exalnple
from the Kilikiimiit3 temple (fig. 107), the left pilaster is completely missing.
 23 Yama (fig. 120), north(-west), was most probably intended
to occupy the south(-western) niche. He also has the typical
squarish face and a jnpinzukufa. His right hand holds a
klzafvarign (skull-topped staff). A relatively large buffalo
stands with its head almost squeezed between the pilaster and
the vannrndlii (garland of leaves and flowers), to the right of
the deity.
24 Icubera (fig. 121) on the north(-east) generally resembles
the other guardians. He wields a large ribbed gadcl (club)
with a protuberance at the top. Near his left foot is an indis-
tinct object, probably not a representation of his vehicle.
Thus, Kubera is the only dikpdln at Cittaudgadh without any
vehicle. This feature has been continued at OsiHii. A club as
an attribute of Kubera rarely occurs in North India.
25 &na (fig. 122) on the north(-east) closely resembles the
"ISHna type" images inside the temple. His upper attributes
are trident (r.) and serpent (l.), his lower attribute is probably
a lotus (damaged; r.), and the lower left hand rests on the hip.
The bull's head appears to the left of his master. A11 unusual
element is the large bowl with sweets in front of the animal.
It is quite difficult to interpret the intricate, somewhat puzzling evi-
dence from Cittauggadh. A regular dikpdln scheme, like the one on
the sanctum exterior of the earlier temple, the KdikiimHtH, does not
appear in the KumbhaiyHma temple. Moreover, the latter temple
contains two differing sets of dikpdlns. The interior set seems to be
characterized by the multiplication of PSHnat~iva,while the p s i -
tioning of the guardians is also irregular in the case of the exterior
set. The choice of Kubera as guardian of the north in the exterior set
seems like a conscious departure from the KHlikHmHtH scheme,
where Soma has been represented in his stead.
There are some striking facts which support MEISTER'Searly
(latter half of the 7th century) dating of both Cittaudgadh temples:
two points of similarity connect them with the already discussed,
securely dated (c. 681-696 A.D.) Svarga Brahmii temple at AIampur.
Both these features are clearly understandable in their context at
Cittaudgadh, but appear to be less logical in the AIampur context:
- The lance-like object or standard placed on Agni's proper left
side is also found in the KHlikHmStii temple, where it is held
in one of Agni's four hands. At Alampur, the two-armed Agni
has no free hand to hold the standard (fig. 71), and hence, it
remains floating in the background.
- The syncretistic deities Ardhaniiriivara on the north and Har-
ihara on the south are both represented on the Svarga Brahms
(ill. 11) and on the KumbhabyHma (ill. 18) temples. On the
latter, their presence suggests a visual "pun", referring to the
meeting point of two architectural elements. These images do
not have the same function at Alampur~54since there is no
unevenness in the structure of the wall.
The second of the two points was elaborated earlier by MEISTER
(1979: 228):
here seems to be little doubt that the priest or architect at Chit-
taur, searching for a schema within which to place his images,
saw the relation between his conjoint images of S i v a - ~ ~ v a t i the layout of the panels m d their iconogrvphic features, they Can be
[Ardhaniiriivara] and Siva-visrlu [Harihara] and the juncture
between priisdcla and i?za!z~n17nin which he put them.
just likeat Amrol and
at MundeSvari9Cittaudgadh
Preserves two with dikl?Ela (or related) sets. At all these
sites, the later temple significantly deviates from the earlier one in
the composition of the dikpiila group as well as the distribution
the figures on the temple Galls. This phenomenon seems to be
restricted to the experimental phase in the development of the
dilcpaln scheme.
Uniquely during this earlier phase, on the sanctum of the Kurn-
bhaiyHma temple at Cittaudgadh, all the dikpalns are depicted with
four arms, combined wit11 a strong ~ a i v aspect.
a
Bt!hii, Mnhiidevn ter?zple
The Mahiideva temple at Birhfi255 is particularly interesting for its
unusual ground plan. The site is located in the Pali district*
Rajasthan. With an estimated date of c, second quarter of the
century (MEISTER1991: 170), the telnple lnight postdate both tern-
ples at Cittaudgadh. The dikpala images at Bithfi rese~nblethose of
Cittaudgadh, however, sufficient d i f f e ~ n c e sexist to suggest that
they werenot execu~edby the same scllool of craftsmen. The frames
of the niches are differellt here, alld tl1e only feature which is remi-
niscent of the Cittaudgadll telnples are the lotus brackets. Unfortu-
nately, only three directional guardians survive in the directional
scheme at Bi!hfi.
I Varuija (MEISTER1982: fig. 8) occupies the (north-)weSt
lcarrzn niche of the outer face of the wall enclosing the
prarlnkc~i~iipntlza (ambulatory). The deity stands in a natur-
ally flexed iibhnrign posture with the legs held together and
not apart as at Cittaudgadh. It is unclear as to whether he
wears a crown or jnfa,xultutn. Some ornaments are dis-
cernible, but he seems to lack the up,avitn (brzhmanical cord).
The squarish face is reminiscent of the ~ittaudgadhscu'p-
tures. In his right hand Varuna holds a staff-like, twisted
as at Cittaudgadh. The left hand rests on the hip. An attendant
is seen on Varuna's right side. On his left side, the outlines
a rlznknrn can be faintly discerned.
2 VIyu (ibid.: fig. 9) north(-west) exhibits the s a n e
relaxed Ebkaiiga posture as Varuca. A piece of V ~ Y U ' S
lowing cloth is visible where it is wound around his rightfore-
arm. It is probably wound around the left forearm as well, but
the shapes are indistinct. It is a rare feature that both his hands
are held in a low position (1.: kafi/znstn). As in Varuna's
a round halo is still faintly visible. AI, elegant, long-flecked
antelope can be seen behind Vfiyu's legs. He is flanked by
two attendants, who appear to be miniature replicas of him-
self.
3 Soma (ibicl.: fig. 10) on the north(-east) also occupies a p ~ ~ i -
tion corresponding to the KBlik;TmHtB set. He stands in
sar~tnl~iidn.
The ends of a crescellt moon project from ~ 0 ~ ~
shoulders. His right arm seems to be raised. He is flanked by
two, probably female, attendants. The rest of the image lS
very badly damaged and does not reveal any details.
The sculptural style of the Bithfi (/ikp&s is less "m0rILImenta1"
than at Cittaudgadh and is characterized by softly flowing contours2
representing a rather independent idiom. H ~with regard
~ to ~ ~
254 BUCJIANAN (1985: 49) regal& a combination of Har~haraand ~ r d h a n i r ~ ~ ~ ~ ~
as "popular with Calukya patron,",
255 Milsrra 1991: I Q - ~ ~ Q M
; , . IV82
~ ~ ~mdyzed~ tell,P~eand i]Iustratt
ed the remaining d k,,E,u V,
clearly grouped with their counterparts at Cittaudgadh. The place-
ment of Soma on the north is particularly reminiscent of the
Kiilikiimiita temple.
Like Bithii, Pipiid and the following site of Osinii are located in the
Jodhpur region, Rajasthan. The Pipliid~niitiitemple has a rzira~zdhdra
(without ambulatory) plan, with projecting knpili walls (enclosure
of the vestibule). The original superstructure is lost. The karrza (cor-
ner) niches contain an incomplete, partially substituted set of direc-
tional guardians. Unfortunately, the images are in a bad state of
preservation. The temple has been assigned to the second quarter of
the 8th century by MEISTER(1991: 139).
Several factors indicate a Central Indian influence on the
Pipl5dmiits temple, such as the broad Icnpili walls and the motif of
paired geese appearing in a cloud formation (ibid.). The rendering
of some of the dilq~iilas,particularly that of the north(-eastem)
image (fig. 128), is strongly reminiscent of a Central Indian model,
i.e. 011 the Teli-kg- lnandir image at Gwiilior (fig. 100).
6 V3yu Gaja- 7 T&a?
laksmi
5V a r u ~ a ? I Indra
Kertti-
keya
4 tlikpdn? 2 Siirya
3 Kubera Varjha GayeSa
111.20: PipId, Pipliidmiitii temple
1 Indra (MEISTER1991: pl. 326), on the (north-)east, occupies
a different position than at Cittaudgadh (KIlikiimItii garb-
habhifli; Kuinbhaiyiima outer wall; ills. 17, 19), which can
also be assumed for Bithii. On these temples, the sequence
starts with Indra 011 the (south-)east. Here, at Pipnd, the god
is portrayed as seated on his elephant, with his c011~0rtgrace-
fully seated behind him, partly turned to the rear. It is not clear
whether Indra has an attribute in the raised right hand. The
left hand is resting on the thigh.
2 Siirya's (fig. 123) position on the (south-)east is noteworthy.
The only parallels for a coinbination of Indra and Siirya 011
the east wall are found at AIampur (ill. 11) and Amrol (ill-
15). As in the case of Indra, some space has been left elnPtY.
The Sun God stands in samcry~idnand wears a prominent sash
at the height of his elbows, which extends dowll to his kllees.
All other features have been destroyed beyond recognition.
Only the remains of a large round halo can be seen. Two
attendants, apparently Danda and ~iilgala,stand in relaxed
postures at some distance on either side of Silrya. The ilnage
seems to strongly resemble the one on the Diinebiibs ternplc
at Amrol (fig. 95).
3 Kubera (fig. 124) on the south(-west) has an unusual P s i -
tion. The figure is co~nparatively]age. The god is seated in
a relaxed posture, wears a large crown and holds a bag in his
left hand resting on the knee. He resembles the Kubera
images at OsiLii (cf. 2.3.1.3).
4 Dikpila? The figure on the (south-)west (fig. 125) probably
does not belong to the dilcpdla scheme. It is shown in a
crouching posture and seems to be animal-headed. The left
hand is resting on the left thigh and the other touches the right
knee.
5 The deity on the (north-)west (fig. 126) is most probably
Varuya, although the gracef~~l posture with the right arm
held parallel to the body creates the impression that it is a
female. The posture mirrors that of the Teli-kii-mandir image
of Varuna (fig. 99) however. The deity stands in iiblzariga,
with the pivot leg on the left. Its left hand is raised and most
likely holds the loop of a noose. Perhaps the right hand orig-
inally held some grass as fodder for his vehicle as at Gwiilior.
This would make sense only when the vehicle was the goose
(Izal?zsa).However, nothing remains of a vehicle on the prop-
er right side of the god. A11 indistinct fragment in the left cor-
ner could be either the remains of a vehicle or a floral orna-
ment.
G VGyu (fig. 127) on the north(-west) can be easily identified.
Although the upper portion of the image is completely dam-
aged, there is no doubt about its identification, as the upper
contours are still visible and the vehicle is preserved. The
Vziyu image on the Teli-kii-mandir (fig. 100) could have
served as a model for it. The figures of Viiyu and the rela-
tively large horse behind him are rendered in the same way,
except for a few details, like the raised right front leg of the
animal and the raised, now missing head of the horse. At
Gwiilior, the horse is tilted to the right, while at Pipad, the god
himself is quite unnaturally tilted to the same side. He must
have held a billowing cloth in his hands.
7 The severely damaged, seated figure on the north(-east) (fig.
128) can be identified as k i n a or it could possibly be a mis-
placed Yarna. It is not clear whether it was originally four-
armed. A certain similarity with the k n a image at OsiBii,
Harihara temple I (fig. 195) is, however, quite obvious. The
latter's rare feature of a serpent held on the right side may be
paralleled at PipId, where the figure holds a long, slightly
curved object in the (front?) right hand. However, since the
serpent is usually held in one of the rear hands, this identifi-
cation may be doubtful. The left portion is not very well pre-
served.
The PjplBdm2tii temple once more provides evidence for a fonna-
tive stage in the development of the agadikpiila theme, which is
traceable in Western India. The clear affinity with some Central
Illdial1temples - the Diineb5bii temple, Amrol, regarding the distri-
bution of dikpdlas; and the Teli-kg-mandir, Gwiilior, regarding the
rendering of some of them (Varuna, VIyu) -is particularly signifi-
cant. On the DInebiibI temple, both seated and standing dilcpalas
occur. This combination is an indication of the for~nativeperiod
which can be witnessed at Amrol, Pipiid, and also at AIampur dur-
ing the latter half of the 7th and the first half of the 8th century. It is
s~lrprisinghowever, that a complete set of a~fadikpdlasis found in
the KAlikIiniitI temple, presumably before 700 A.D. A plausible
explanation could be that the large, royal temples at Cittaudgadll had
sources of inspiration other than the Central Indian influence found
at Pipiid. A clear difference also exists between the Izorror vncui
treatment of space at Cittaudgadh and Bithii and the images which
"float" in the einpty space as at PipId and in Central India.
Osidfi, "Siirya" tenzple no. l and "Harihnra" temple no. 2
At the important temple site of Osiiifi, Jodhpur districtIRajasthan,
temple construction started some time around 700 A.D. As in Cen-
tral India, several stages of the formation of the dikpdla theme can
be observed here. Sbrya temple no. 1, presumably the earliest of
this series, does not yet have a full set. A complete canonical set
occurs for the first time in the Harihara temple no. 1, where the
a~[adilpdlasappear in seated (Viiyu: riding) postures in the eight
karna (corner) niches. Until this stage, a strong Central Indian in-
fluence is traceable, particularly in the goose (hamsa) vehicle of
Varuna. While on the Harihara temple 1, some of the viihnnas are
clearly related to Central India (quadruped for N a i ~ t a hamsa
, for
Varuna, horse-like antelope for Vgyu), the postures of some of the
dikpdlns (Viiyu, Varuna) on SBrya temple 1 with nearly parallel but
tilted legs are strongly reminiscent of Central India and clearly
differ from those of the other figures shown in similarly relaxed
stances.256
Like most of the other temples of the 8th-9th centuries at this
site, Sbrya temple no. I2s7 has no ambulatory. It has its own plinth
and an open nznndapn (hall) in front. It faces west like most of the
other temples at Osiiifi. This orientation was not altogether unknown
in Central India during that period, as the Diinebiibiitemple at Amrol
shows. While there is no definite clue regarding the absolute date of
the temple, the recently proposed date accommodates it shortly
before or after 700 A . D . ~Concerning
S~ the directional guardians, it
is extremely unusual to find them rather regularly positioned on the
western and southern walls, but practically absent from the eastern
and northern walls. The seated Kubera ( l ) in the (south-)eastern
k a r y niche most probably does not belong to the dikpdla concept
on SBrya temple no. 1. However, it has a similarly placed counter-
part at Pipiid, where the astadikpiila concept is likewise irregular. It
is a significant fact that all the "true" dikpiilas on the temple are
shown standing. The most likely explanation for the absence of
dikpiilas on the east and the north would be that the design concept
of the temple had been changed during its construction. In any case,
these images have apparently been sculptured separately and insert-
ed into the niches.
Ardha-
niiriivara - Nateia
6 Viiyu
Piiriva-
niitha?
5 Varuna
A SBrya
- IS
1
4 Candra I / 1 Kuben
3 Yama Ganeia 2 Agni
111.21: Osiiifi, Sarya temple no. l , plan
I The image of Kubera (fig. 182) on the (south-)east has some
rare features. He is shown squatting on a cushion or sack. His
headdress consists of thick rows of lirdhvakeia (upright
standing hair) curls, which are held together by a diadem or
ribbon. His face is damaged. His ornaments are earrings,
keyurns (upper arm ornaments) and a necklace hanging well
below the neck. The briihmanical cord is depicted as a broaP
band (vastropnvita). Kubera's belly is pronouncedl~big. HIS
Proper right hand holds an unusual object resembling a short
staff. The left hand holds an elongated purse. Flanking hiln in
the lower corners is a pot turned upside down with something
issuing from it. Above on both sides, branches ofmango trees
with bunches of mangoes are visible.
2 Agni (fig.183), positioned on the south(-east) in a rigid POS-
ture, doubtlessly acts as the guardian of the south-east. He has
an unnaturally short torso. Small, neatly carved flames
emerge from behind the head and shoulders. The headdress
jatiiilzukutn (crown of hair), and hc wears simple ornaments.
The broken right hand raised to shoulder level must have held
a rosary. The broken object just below the left hand inusthave
been a water vessel. ~t seems to rest on a small plant below.
On Agni's right is a small, damaged image of a goat moving
towards the front.
3 Yams (fig.184) also occupies his canonical place, the south
(-west). He is seated on a striding buffalo. The tangled
JrdhvakeSa curls are held together by a ribbon. A s s h Is
wound around Yama's staff and arlns and flows behind
shoulders. The left hand rests on the left leg, and the right
hand resting on the right knee holds a damaged staff. The face
is damaged.
4 The figure ill the narrow (south-)west k a r p niche (fig.18')
has been recently identified as by lne.25"t exhibits
no specific feature of Candra, but no other identification of
this deity seems to be possible. Candra resembles A&ni i.n
many ways. The posture and coiffure are the same. The POs1-
tion of the arms are likewise identical with the latter, and the
attributes (most probably a rosary - r., water-bottle - 1.)
seem to be the similar. A bri]~manicalcord has been clearly
indicated. A tiny vehicle stands behind the god. It appears
be a rather fragile antelope. In view of his unusual position it
is unclear whether Candra functions as a dikpiila here.
5 The image in the northern kapili niche (fig. 186) has
also recently been identified as Varuy.26fl he occurrence
''' The difference becomes particularly collspicuous when comparing twOjuxta-
Posed illustrations in MEISSER1991. on the llorlh(-west) (ibid.:PI. 309)
stretches the frce leg stimy to the side, with the foot strictly parallel to the socle The
same posture, which has also becn chosen for the doorkeepers (ibid.: 3141317'
318), is already familiar to us from the dikpilns or the KJlikiIrn513 temple at cittaud-
(e.g. fig. 108). Varuna (fig. 186) stands in a slightly flexed postusc (~~hfliifi')llke
ArdhanariBvara, but the legs and feet are kept in an almost parallel position Although
wit11 the extrelnely short torso and the broad pace it resembles the latter image, the
and the thick lower end of the short garment are much more naturally modclled VQy.u
(fig.187) shows an extreliie version of this stallce with parallel legs and feet.
comlnon in Central India (cf. fig. 99).
257 MEISTER1991: 132-137; BHANDARKAR 1912: 110; HANDA
Strangcl~,the temple has not been included in VIENNOT'Sstudy (1976).
55-58'
MUSTERin the lext of the Encyclopaedia (1991: 137) assigns it to the
quarter of the 8th century, but in a plate caption (p]. 303) the date is given as "C. l"'
seventh-early eighth century A.D.".
259 WESSELS-MEVISSBN 1994: 599.
'" MEISTER(1991: 310) identified the deity as KBrLlikeya. In that Case, the bird
would have to be a peacock. However, the noose attribute is unknown with Karctikeya'
Identifying the god as Varu~~a (Wcss~~s-M~v~sse~ 1994: 600) on the other hand leaves
the peacock-like bird unexplained. With its crest it does not represent the typical goose
which is found with Varuna in Central India. Perhaps, a misconceplion about the bird
had occurred, which would not be the only instancc ill the case of unusual vehicles'
On the slightly later Harihara tc~nple1, Varu~a'sgoose has been rendered lifelike
(fig. 192).
 a dikp6ln in a knpili niche is elsewhere only found at Amrol
(ill. 14) and Indor (ill. 27). V a r u p has a stout figure, a broad
face, and mighty hands and feet. His iirdlzvakeia curls are
held together by a ribbon. Small earrings, a simple necklace
and bracelets are his only ornaments. A decorative sash is
wound around both his arms. His right hand is raised and
holds a large noose with a solid centre (cf. fig. 99). His left
hand rests on the hip. The lower garment ends above the
knees. On Varuna's left side, a large bird is shown moving
away from the deity in a graceful posture which probably
represents courtship dance. On the proper left side, the leaves
of a shrub balance the carefully conceived scene.
6 Vlyu (fig. 187), the Wind God, appears to be uncomfortably
squeezed into the narrow (north-)western karpa niche. His
coiffure is very simple. His posture is an unnaturally flexed
dbhaliga. A sash is tied around his thighs, just above his
knees. The Agni image at Gwalior also wears a sash in a sim-
ilar position (fig. 98). In his two raised hands, VByu holds a
billowing cloth. A small jumping antelope can be seen to his
proper right.
Harihara temple no. 1261 (ill. 22) represents a culmination in the
early temple art at Osiafi. The temple has been constructed as a
pflficdyntnrzncomplex, consisting of one central and four laterally
arranged shrines. With the Harihara temple 1, lcnpili walls cease to
be constructed at this site. Pratiratlzn niches have now been intro-
duced between the [carpa (corner) and central blzaclrn niches 011 the
exterior wall. The distribution of deities in the niches has been made
in a more careful, conscious manner, with the complete a$adikpdla
set starting on the (south-)east with Indra. It is a peculiarity of the
Harihara temples nos. 1 and 2 at Osiiifi, that the Sun and Moon God,
SLirya and Candra, who are found associated with the dikpdla sets
at some of the early sites,262 have been shifted to the pratirnthn
niches and thus structurally separated from the group of directional
guardians. It seems that after this development, the inclusion of
Candra andtor SBrya in the group of dilcpdlns ceased in North India
and the Deccan.
I Indra (fig. 188) on the (south-)east resembles his counterpart
on the Danebaba temple, Amrol (fig. 96), in that he is seated
on a reclining elephant with its head in profile. Both images
of Indra wear a simple necklace and exhibit a roll of fat above
the lower garment. Indra wears his typical cylindrical crown.
His right hand resting on the knee must have originally held
a thunderbolt, his left hand rests on the left knee. Some inter-
esting details are a tight sash that holds Indra's right leg in a
slightly upright position, and a saddlecloth and a headband
for the elephant.
2 Agni (fig. 189) on the south(-east) has many features in com-
mon with his counterpart at Ainrol (fig. 97). The positions of
the hands and legs are largely identical. The sash used as a
kind of yognpnttn (band to keep the legs in place during med-
itation) occurs in both cases. Agni has a ja~di~zulcutn (hair-
crown). His attributes are a rosary (?) in his raised right hand
and a decorated water-bottle in his left hand held near the
lower part of his leg. A departure from the Amrol example are
the long flames that completely fill the background, and the
figure of the goat covering the breadth of the panel. The low-
er garment hardly reaches the knees. The brahmanical cord is
absent.
3 Yama (fig. 190) 011 the south(-west) squats on a couchant
buffalo. He wears a ribbon around the u'rdhvakein curls. TWO
rows of beads serve as a necklace and as keycirns (upper arm
ornaments). Yama wears asymmetrical ear ornaments. He
holds a staff with a bulging top (skull?) in his left hand. His
right hand is damaged. It must have rested on the right knee.
An important detail is the small bird (cock?)263 shown be-
tween the feet of the reclining buffalo.
6 Vtiyu Brahmj. Narasi~pha Maliisa- 7 Kubera
rnardini
5 Varuna 8 !SBII~
Sirya
Hari-
hara
Visvu
4 Nairrta l Indra
3 Yarna GaneSa Trivikrama Candra 2 Agni
Ill. 22: OsiBii, Harihara temple no. 1, central shrine, simplified plan
4 The figure of Nairrta (fig. 191) appears to be squeezed
into a very narrow niche on the (south-)west. He wears
jafiimukufa and a necklace with two rows of beads like Yama.
His right hand holds a long sword, and his left hand rests on
the respective knee. In the lower section of the niche, parts of
the body and the head of the quadruped vehicle are visible,
which seems to be a dog or jackal.
5 Varuga (fig. 192) on the (north-)west has been allotted a nar-
row niche like Naiqta. He wears ja@nzukuta, a double-row
necklace, and a short lower garment. He squats on a goose
(hal?ua). His right hand is raised and holds a noose made
from a rope. The left hand holds an indistinct object, which
may be a fruit.
6 Vlyu (fig. 193) on the north(-west) appears riding on his
antelope. He wears his hair in upright or u'rdlivakeia curls.
His earrings seem to be asymmetrical, and he wears a simple
necklace. Vayu's symmetrically raised hands hold an inflat-
26' MElSTER 1991: 158-167; BIIANDARKAR 1912: 101-103; KALIA1982: 3 4 ;
HANDA1984: 29-3 1.
262 Both Siirya and Sonia have been structurally treated as dikpcilns on tlie Svar-
ga Brahmii (ill. 11) and other temples at Alampur. Only Skya appears among the
dikpBlas (i.e. in a corresponding kar~japosition) on the Ddnehiibii temple at A~nrol(ill.
15) and at Pipid (ill. 20). SomdCandra is probably included on Siirya temple 1 at OsidB
(ill. 21). On the Kiilikimiti temple at Cittaudgadb (ill. 17) and at Bithii (see above)
however, he has a regular function as a guardian of the north. Both Surya and Candra
are essential members of the group of "Manu Lokapdlas" (see section 2.4.1.3), where
they replace Nairrta and iidna.
263 This detail has also been noticed by HANDA (1983: 55). He identifies the bird
as a cock, which has been mentioned as Yama's attribute in some texts. Parallels from
Celitral India also show a similar position of the bird in front of the buffalo (figs. 147,
154). It should also be noted that a closely similar bird forms part of the Ya~nailnage
of the Kilikiimiti temple at Cittaudgadh (fig. 105), where it is placed on the lotus held
by the god.
 ed cloth which curves around his head. The god sits astride
a stiffly conceived, fat antelope. It resembles a mule rather
than a swift animal. This is an interesting feature, since the
contemporary Central Indian models depict VSyu on a horse.
7 Kubera's image (fig. 194) is placed on the north(-east). He
wears his hair in nicely dressed curls, but has no crown or
diadem. Kubera's ornaments, particularly his necklaces, are
unusually lavish and intricate. His face and that of his anthro-
pomorphic vehicle squatting below him are largely damaged.
His right hand is broken, and his left hand grasps an elongat-
ed purse.
8 The image of (fig. 195) on the (north-)east is the only
four-armed figure. This panel is only slightly damaged and
exhibits some interesting features. His headdress consists
of an elaborate jatiimukuta, and he wears symmetrical
earrings. Apart from a double-row necklace, the torso is
embellished by a vastropnvitn (broad briihmanical cord).
While the keyuras are rather simple in design, a long wzdld
(garland) made out of skulls and bones encircles the entire
body. The god sits in dak~ip~pralambnpiicla(right leg
pendent), with the left leg resting on the hump of the bull
vehicle. The front right hand raised to the shoulder level holds
a rosary and probably some other object (or the tail of the ser-
pent?). The rear right hand holds a serpent, and the rear left
hand grasps a trident, the shaft of which consists of a spinal
column. The front left hand rests on a part of the garland.
No other temple shows such a close relationship between the devel-
opment of dikpiila representation in Central and Western India as
Harihara no. 1. However, in the following phase, the supposedly
indigenous Western Indian tradition, as it is iconographically
reflected in the dikpdlas at Cittaudgadh and Bithii, became domi-
nant. This will be demonstrated by the evidence of the temples sub-
sequently constructed at OsiLB.
Lnmba, Old Tenzple
Only four of the originally six knrna-niches of the temple at
LLmb1,264which is located south-east of Jodhpur and north of BithCi,
survive. Of these, only one contains a dikpdla image: namely V5yt1
on the north(-east). An image of Siirya on the (north-)east co~11d
have belonged to an early configuration of dikpdlas. However, the
present sculpture has probably been inserted later (MEISTER1991:
173). The temple has been regarded as "contemporary with the old-
est temples at Osia" by DHAKY(1966: 145) and tentatively dated to
the "second or third quarter of the eight century A.D." by MEISTER
(1991: 170).
VIyu (ibid.: pl. 387) sits on, or rather, above, a large, realisti-
cally rendered antelope with its head raised. His left knee is visible
above the snout of the animal. The wind god is shown frontally,
holding the usual billowing cloth. The arrangement consciously
makes use of the rather elongated niche format. Inside the temple,
there is another depiction of Viyu which is included in a frieze of
Manu Lokapilas (cf. section 2.4.1.3) shown in procession (ibill.: p].
393c; ill. 31). Here, the deity is less stiffly rendered, the billowing
cloth swaying gracefully sideways behind his crown. He is seated
On the back of the which leaps forward in an ele-
gant manner. The stylistic differences suggest either a chronologi-
cal gap between the two images of Viiyu, or they may have been the
work of two different schools of artists. It is also important to note
that is the with regard to the depiction
of the the Manu Lokapiilas.
2.3.1.4 Eastern India and Bangladesh
The only site with spatially arranged directional guardians in the
eastern part of the subcontinent in this period is Dev ~araniirak
(Shahabad district) in north-western Bihar. Rather unusually, a
group of only four members is represented here. What may be
regarded as the legacy of the nearby site of MundeSvari with its
cornerstone set of Manu Lokapiilas continued only in a number of
friezes of this particular group of deities, some of which Inay belong
to the period under discussion (see section 2.4.1.3).
Dev Bnra!zdrak (Deo Barzt@irk)
Unfortunately, this unique arrangement of directional guardians is
not well-published, and therefore, a detailed and analytical descrip-
tion cannot be given here. The lolcn~~iilas
are carved in high relief on
an isolated pillar carrying
... two sets of sarvatBbhadrik5 images [facing the four lllaill
directions, respectively] -at the base representing Siva, Pirvati,
Sfirya (?), and Ga~j?.Sa,at the top, four LGkapilas. The pillar
shows square, octagonal, and circular sections and small stand- 1
ing figures ofas[agrahas [planetary deities] in low relief. (DEVA
1988: 122)
An old Archaeological Survey photograph, re-published by ASHER
(1980: p]. 125), shows that they appear seated above their vehicles.
It is clearly visible that Varuna has har?zsn as his vehicle, which is
his typical characteristic not only in Central India but also in the
eastern part of the country, until C. the 9th century (see section
2.5.5). A short description of the images has been recently given by
B.K. JAMUAR (1985: 60-61):
(i) Indra seated on an elephant and attended upon by a figure
carrying [an] umbrella (east),
(ii) Kubera seated with his right leg raised up and holding a bag
of molley in his left hand (north),
(iii) Varuna seated on a bird (harhsa?) and holding a flower in
each hand (west), and
(iv) Yama seated on a buffalo and holding a flower in each hand
(south).
2.3.2 Later Phase (c. 751-1000 A.D.)
Although a consolidation of the clikpdla theme265 was achieved in
most of the regions during this period, variations in the placement
and arr~wzementof the group did occur on and off. These, howev-
er, may have been nothing more than variations on a the~newhich
was already more or less widely known at that time. Gradually, the
directional guardians had to give up their rather dominant positions
on the temple walls. Likewise gradually, the representations became
rather stereotyped and manneristic. Both these features, of course,
were results of the general developments in the mediaeval Indian
7.64
Ig(j(j: I 4 N 4 5 ; Mrlrrrn 1991: 170-174, ~ i ~ ~ tllcr~
~ i aref IIOi ~ ~ ~ ~
knrt~nniches on the front (west) side, which seelns to bb an early feature.
'G S. MARKEL (1995) in his book on ano~Iierimportant group of ~ninorgods,
lanet tar^ deities, subsumed their develop~nentunder three main headings: "Tile
forlnative period", "Innovations in imagery", and "Variations on a tl~e~ne" (ibid.: 1.11).
Acco~lingto aim, tllese p l l a ~correspond
~~ to the 6th. ilh, alld 8th cenudcs, respec-
tivcly.
 Art. The dikpiiln images of this phase can be traced to some earlier
models (Pattadakal, Piipaniitha temple; Cittaudgadh, Kiilikiimrit3
temple).
It would be important to ascertain the date of the acceptallce of
the canonical agadikpEln concept in the different parts of the coun-
try. However, the extremely sparse availability of absolute dates
makes it very difficult to establish all these datum lines. The degree
of regional variation, which can be more easily traced, is also of con-
siderable interest.
One of the newer findings is that the wider acceptance of the fu1-
ly developed n.~tadikpCj/nconcept in the North-west started with the
Harihara temple no. 1 at Osinii. This complete set, the earliest in the
region, seems to have been introduced as the result of a close inter-
action between Westem and Central Indian traditions. This contact
Inarks the beginning of a regular dikpdla representation in the
region. It is, however, noteworthy that the prototypes of deities
influenced by the Central Indian modes of representation, were soon
discarded and replaced by local versions.
Due to the great number of temples with installed asfaclik~8lns
in this period, they can be treated only briefly and may even remain
incolnplete in some cases. Apart from the general lines of develop-
ment, some significant prospects for the subsequent period will be
included.
2.3.2.1 The Deccan
A unifying trait of most parts of the Deccan (including the related
sites of Dharmapuri and Armamalai in northern Tamilnadu) is the
'epresentation of as;ndilcl~alnson ceiling panels. This was the con-
tinuation of an earlier practice, already foulld in the 6th century
Cave 3 at Biidiimi, and continued from the end of the 7th cellturY
Onwards. Certain characteristics of the relief panels exhibit purely
regional concepts that continued at least until the end of the 10th
Some of the ceiling panels are exquisite works of art.
The concept of depicting dik~~iilas on the temple exterior did
gain wider acceptance in the Deccan. 0111~a few examples
Occurred, which were obviously based on the model of the
Papanatha temple at Patfadakal.
A Partly related concept was the p~acenlentof the guardians on
the gol>urn(part of the western screen wall) of the Kailba temple at
Ellor%. Although unique during the period under discussion, this
forln of represelltation continued in the later C6la art.266Further, at
~ k ~ ~ dstill
l a sfind their place among the attendant figures
On the upper portion of lnonumental relief panels.
Another rare position ofdikpdlas was near the top of the vifl7iirzn
tower above the sanctum). It is found only in one example
In the region under discussion, viz. at PratakotdA.P. The installnent
Of clikc~alas at such an extreme height must have been inspired by
the Kumiira Brahms temple at Alalnpur or by other, now extinct
Shrines,reflecting the salne tradition. Another rare representation
the directional guardians is seen 011 the door lintel frieze at the
No!alnba capital, HemBvati/A.P. (DoddeBvara temple).
The depiction of directional guardians at ~ikkavolll/A.P.
(GOlinge~~ara temple) represents a different concept, showing
of them on the exterior, and not placed in their canonica1
positions. Agni, and VByu appear in a seqllence, framed
Other deities, on the southern wall of the vinziirza. At Kelbr near
AihO!e, an isolated V a r ~ l ~occurs
a in the wester11 niche of one the
temples.267
At Ihkkaniir (Navalifiga group, older shrine), the uniforrn
group of deities in the exterior niches could not possibly represent
dikpnlas, as has been stated by K.V. SOUNDARA RAJAN(1986: 144),
since dilc1kilns are always shown with individual features, which is
not the case here.
Starting with the prominent tradition of agadikpiiln ceilings,
several facts have as yet not been noted:
- A number of different modes of layout occur, in which
the riding deities (generally in profile) have been arranged
(ill. 23).
- There tend to be regional preferences for particular orders as
well as for other characteristics of the deities, their arrange-
ment, and the design of the panel. Thus, four-armed
dikpZlas are restricted to the eastern part (Alampur region),
while the presence of female consorts is common in the
western part of the Deccan.268
- Generally, the lower portion of each sub-panel forins the
border of the whole panel, but there are a few exceptions
with the lower sections directed towards the centre (ill. 23
b, i). The lower part of the central panel is in most cases
placed in the direction opposite the sanctum, but there are
exceptions to this rule as well (e.g. ill. 23 b, h, i).
- The central sub-panel is generally occupied by the god ~ i v a
(in his dancing form Na!eSa, mostly four- or eight-armed;
Kalyiinasundara; Umiisahitamnrti; or Somiiskandamurti).
The only exceptions are found in the Jaina temples (ill. 23
e, f), and once at Aihoje (fig. 130), where Brahmg is repre-
sented in the centre. The distribution of panels with
Umiisahitamurti is significantly restricted to the eastern part
of the Deccan (ill. 23 c, i).269
The n~~nclikl>,iilnceilings of the Deccan form good material for an
in-depth study. The results could be correlated with other features
of temple architecture and may even assist in establishing the
chronology of the respective monuments. The example of the often
published relief in the KalleSvara temple, Ara!aguppe/Karnataka
(ill. 23 b; fig. 136), which forms the most ornate part of the temple,
shows that the clikpiila ceiling can be assumed to convey a signifi-
cant theological meaning. The fact that Nateia in the majority of
cases appears in the centre also supports this assumption. His eight
arms obviously allude to the heavenly directions and emphasize
Siva's role as the dancing Lord of the Region~.27oTypoIogical]~, the
developinent of the NateSa-cum-nstadikpCjld(z theme evolved froln
elements, which were already known in the early to mid-8th centu-
ry shrines at Patladaka1 (Mallikiirjilna and Pripaniitha) and Aihole
(Jyotirlinga group; see section 2.3.1.1). In the Mallikrirjuna temple
itself (fig. 326), the main constituents can be found: namely, an
eight-armed NateSa, accompanied by four dikpiilas (plus V i g and
Brahmri, which is somewhat renliniscellt of the early ceiling relief
266 J.C. HARLE 1963: 103-138 (Cidambaram) and F. L'HERNAULT 1987: 35,118,
pl. ico. IV (DBrBsuram).
267 Kelilr, Rjrnaliligeivara telnple group, temple no. 2 (RADCLIITE 1981: 474, fig.
881). Varu~nis described as four-armed, llolding a noose in the lower (right?) llatld alld
carrying two fly-whisks in the upper hands. He stands on a mnkam, which is uliique,
The southern and liorther~iniche of the temllle both contain an i~ilagcof goddess,
respectively.
268 An exception being Indra accompanied by his wife at Bhavanssi Safigaln
(Alampur region) and at Dhannapuri1T.N.
269 Further spccirnens with U1n3sallitalnorti at the centre cxist at Siddllepalli
S ph. 853) in the Alarnpur region, at Nandi (A.L.COIIEN
( D A ~ E N1984: 1989: 180) and
Avani (ibid: 205,209) in the Kolar region, mid at Dharmapuri (il)icl.: 233).
270 Tiru~nillar's Tirririicrririrurii,Stanza 2728 (2774). The text is assigned to the
7t11-8t11 centuries by K.V. ZVELEBIL(1985: 47; cited: 49).
 South South
South

.
Nairrta yma Agni
.
Nairrta Yama Agni
.
Nairrta Yama Agni

West Varuaa Natela Indra Um3-

Vamaa sahita- Indra East
(8 arms) A
marti I

Vayu
v
Kubera
C .
ISaa V2yu
C
Kubera
C A Vayu
I
Kubera k3na
I

a) Alur, ArkeSvara temple (G); after b) Aralaguppe, KdleSvara temple (N); after c) Bhavanai Sangam, Rapala SangameSvara
~ pl. 70.
1.K S A R 1992: 1.K SARMA1992: pi. 45. temple (R); after B. DAGENS1984: ph. 805.

.
Nairrta Yama
4
*Pi T'a
.
Yams ~gni
4
Nairrta Yarna
4
A V

West

L
.
Vamna Jina Indra
Vamna
Dha-
ranendra
Yaksa
Indra
A
East

Viiyu
. Kubera
C
YSha VZp
C
Kubera
C
. Iirina
V3yu Kubera
C
IS3na
C

A -
d) Dharmapuri, Mallikajuna temple (N); after e) Karnbadahdli, S2ntinltha bati (G); after f) Karnbadahalli, Adinatha basti (G);
after
K KRISHNAMURTHY1987: pl. XLI. SARm 1992: pis. 112, 129. I.K. SARMA1992: pls. 121.

.
Nair~ta yam,
v
Agni
. Nairrta
4
Yama
4
~gni
4
Nairrta Yama
4
Agni

West Varupa NateSB Indra Uma-

Varuna NafeSa Indra East
@arms) A
Varuna sahita- Indra
(4 arms) miirti
-
A
A A

Vayu Kubera ISana

C C Vap
C
Kubera
C
fSaa
C
V a ~. .
Kubera fSya

North
~ .-.
--

North North
g) NarasmangduR
temp'e (G); h) PapanaSanam, group 9 (R); after B. D A G ~ S i) Pratakota, NQeSvara temple (R?); after
after LK SARMA1992: pl. 56.
1984: ph. 450. B. DACENS1984: ph. 746.

111.23: Ceiling panels of the Deccan region, c. 9th-10th centuries (sketches)

g? . .'::-*w%.-"G+" . . . .- - .- . . . - ... - .-.-. .....,-..-.
. .- ,.. . . .
. ",..."_*
W
'-
-.
at BadImi, Cave 3); and a grid of nine compartments, separated by
high ridges, however filled with minor deities.271 The association of
females as observed in the western part of the Deccan seems to go
back to the Huccappayya temple at Aihole (figs. 49-54).
The earliest panel of the period, dating to the third quarter of
the 8th century, is located in the R,?strakfifa KaSiviSvancitha (or
K ~ ~ i v i ~ v e i v atemple272
ra) at Pattadakal (ill. 24; fig. 129). It is has
been clearly modelled after the AnantaSIyi ceiling in the eastern hall
the Papanatha temple at the same site (for the asfarlikp31a frame
Seefigs. 55-61). Although chronologically, the temples are not very
far apart, some significant modificatiolls have bee11 made on the
Kaiiviivaniitha ceiling: Indraysand Vamqa's compartments are of
the same size as those of Yams and Kubera (the latter two being of
much larger size on the Pftpallatha ceiling); the two-armed dikp3las,
with their hands raised in valz&nanzudra (gesture of salutation),
Seemto be devoid of their attributes; most of the guardians and their
viilzflllas are rendered in three-quarter ~rofile.However, most of
these innovations, as well as the fact that all the dikpala sub-panels
of unequal size and bordered by plain ridges, are not found %ail1
in the later temples. Somjskandalnfirti at the centre is likewise
unique. Umbal~itamfirti,however, might be regarded as all icono-
graphically related successor p i the theme.
All other panels under discussion date from the 9th-lOth ten-
turies.273The one at Aihoje/Karnataka and those in present-day
Andl.lra Pradesh appear to belong to the RIstrakfita dynasty, and
those in Karnataka and in norther11 Tamil Nadu belong stylistically
the art of the Galiga and Nolamba dynasties.274The temples of
this period are generally quite small, 1zirnrzdhZra (without ambula-
tory), and provided with a square-shaped hall made up of ninebays*
with the asfadikpnla ceiling over the central bay.
The dikpclas are always found in their canonical directions.275
are shown in profile, and moving in a particular direction.
Apart from some rare cases, in which the peripheral gods - Or
a least some of them (ill. 23 c) - are represented facing the centre
241, the directional guardians move in a more or less
circle aroulld the central deity. In the Alampur region,
there prevails an obvious preference for the arrangement two
groups of three deities, which respectively move in the same direc-
tion (mostly on the north-south axis; ill. 23 h).
South
[Nairyta] *"' Agni
4
T
Som5-
West
Varuna skanda- Indra
rnfirti 4 ly the saliic order and orientation of deitics (H. COUSENS
I L
VIyu Kubera ISIna
4
C A
A structed as conforming exactly to the otber mentioned ones.
North
'l1. 24: Pattadakal, Ksiiviivan~tha
temple, ceiling panel
. . ... . . .
- ...
. .
.. ., , - - .- ... .-. .. . , ... .- . . --.. .'.-. . - . ..
-.-- , .- .-
-. .
-.._..
..>:
-L_-; - .-
.,
.- W
At AlurIKari~ataka,such a grouping occurs on the eastern and west-
ern side of the panel (ill. 23 a). At GokIWKarnataka, the same lay-
out appears, but the feet of the central Nap& are turned to the east
(DHAKY1996: pl. 650). In the NEgeSvara temple at Pratakota1A.P.
there is a peculiar variant of this type (ill. 23 i), while at Aihoie,
this concept with two 'parallel' sides may have been elnployed
(figs. 130-135), but the western portion of the panel is much too
damaged to be sure about this.
The second mail1 type is the "svastika" order, with two deities
aligned on each of the four sides of the central panel. At Ara!aguppe
(ill. 23 b), where the lower sections of the sub-panels are directed
towards the centre, there occurs an order that seems to be unique for
this period.27Wuch more frequent is the variant with the same
directions of movement (corresponding to the clockwise circu~n-
ambulation), but with the lower portions placed on the periphery
(ill. 23 e, g)277. There exists only one example in which the deities
of the main directions are aligned with those of the following inter-
mediate direction, thus creating the illusion that they are at the poi~lt
of leaving the panel (ill. 23 f). Finally, at the southernmost site
Dharmapuri/Tainil Nadu (ill. 23 d; same order of dikpiilas in the
neighbouring Devi temple), an interesting mixed type is found,
showing an alignment of three directional guardians on the easter11
side.
The types of borders dividing the panels also reflect regional
preferences. The eastern part, the Alampur region and Dharmapuri,
shows a predilection for facetted ridges with a central square por-
tion, which corresponds to the shape of the ends of each segment. In
the same region, however, dividing borders are often absent.
Although the iconography of the images varies in detail, only
a few examples will be described. The north-eastern temple of the
so-called Kunti group at Aih0je2~~ bears the latest specimen of a
dikpdla ceiling in the "ancient triangle" BIdami-Aihole-Pattadakal
271 See RADCLIWE 1981: fig. 856; P. BANER11 1985: ill. 12 (mirror image). The
centre shows a rnirhrirln (in erotic attitude) couple according to BUCHANAN (1985: fig,
531), but Umisahitamiirti according to RADCLIPFE (1981: 464) and TARR(1969: 169).
The outer figures are heavenly beings. The ceiling relief is positioned in the antecllam-
ber of the gnrbhflgrllfl.
272 DHAKY 1988a: 346-349. The ngaclikpdln ceiling's central themc has been
errolleously identified as "Gauri-Salikara" (ibid.: 349).
273 111 lliost cases, the probable dates and the relative chronology are still opell to
discussion (e.g. DAGENS 1984: 24-28; SOUNDARA RAJAN1986: 185-214).
274 ~ 0 t kingdoms
h werc closely related both politically and artistically. In sonle
cases, there is disagreement about the classification as Ganga or No!a~i~ba. In this con-
nection, olily the KalleSvara at Aralaguppe needs to be mentioncd. COHEN (1989: 153-
160) and SOUNDARA RAJAN(1986: 213) regard it as No!aniba, while SARMA (1992: 88-
91) classifies it as Ga~iga.
275 In one instancc, a wrong ordcr has been creatcd due to an unqualified recon-
str.uctjon. The relief is lioused at the Alampur Museu~n(A.M. GASTON1982: pl. 31).
The two outer slabs would have 10 be exchanged in order to place the deities in their
correct sequence. GASTON classifies thc piece in her Period 1 (c. 550-800 A.D.)because
of the preselice of a halo. However, this featurc continued to be depicted in the region
ulldcr discussion (see below). As a probable date for the Alampur picce I would sug-
gest the 9th century.
East 276 A later ceiling, of the Hoysa!a period, at Ba!ag,imvelKarnataka, shows exact-
1926: pl. CXIV).
277 Exactly the same order as at Kambadalialli, Slntinltha basti (ill. 23 e), is round
in the Mailareivara temple at Pathasivara~iianear HemlvatilKarnataka (fig. 137). It is
peculiar to find the east-west axis being maintained here by the central Na!eSa, in spite
of the temple's orientation towards the south. At Hemiivati itself, three loose slabs
forlliing an os!adikl?dln ceiling have been found. One is kept in the Government Muse-
um Madras (C. SIVARAMAMURTI 1983: fig. 7-58; detail on fig. 140), and the two others
are still at the site (COllEN 1989: PIS. 23, 24). The order of sub-panels can be recon-
278 Tlie north-eastern temple of the Kunti group at Ailio!e has been referred to by
differetit numbers. SOUNDARA RAJAN(1986) calls it "no. 3" on pp. 134-135, but ill
fig. 54 on the following pagc of the same publication, it is labelled as "no. 1".
(figs. 130-135). Only five dikpalns are still discernible, seated on
their vehicles which are shown in profile (except for the frontally
rendered head of IiBna's bull) and in different ways of movement.
The two-armed dikpiilns hold their left hands in vanrlnrzanz~~drE and
wield an attribute in the right. k n a is four-armed, which is his char-
acteristic form both in North and South India. His upper left arm is
hidden by incrustations. The attributes of Indra and Kubera are
obscured, while Agni carries a club (?), Yama has another type of
staff or club, and in case, only a triiGln in the upper right
hand is recognizable. The background of the figures is plain, with
the exception a dwarfish attendant figure behind Indra's elephant.
The Nojamba Kalleivara temple at Aralaguppe (Tumkur Dt.1
Karnataka) undoubtedly contains one of the masterpieces of ceiling
sculpture in the Deccan (fig. 136). As is common for the temples in
southern Karnataka, the dikpzlns are accompanied by their consorts.
A few peculiarities or inconsistencies of the KalleSvara panel may
point to the fact that it is a relatively early specimen in the Ganga-
Nojamba series:
- The central NateSa (SOUNDARA RAJAN 1986: pl. 633) is only
four-armed.
- His feet are directed towards the sanctum, which is unusual
in southern Karnataka.
- Naiqta and his iakti ride a horse (perhaps meant to represent
a donkey - klzarn, cf. section 2.5.4)279.
- Only the panels of Indra and Varuna have a background of
stylized clouds, and the attributes held by these gods (leaf
baskets?) differ from those of the others (buds).
- Only and his consort are accompanied by attendant
figures, i.e. celestial garland-bearers, musicians and GaneSa. ridges wit11 a central square portion) and are also related in some
The unique presence of GaneSa (son of h a ) is indicative of
Iic?na9sidentity with ~ i v a .
- Indra's elephant is shown with four tusks, as it is also depict-
ed on the 8th century n~~adilcpdln ceiling in the PBpanatha
temple, Pairadaka1 (fig. 55).
So far, however, the KalleSvara ceiling relief and sculptures have
been regarded as particularly developed versions and therefore be-
longing to the later Nolamba temples. COHEN(1989: 160) suggests
the middle or end of the 10th century as probable dates.
Another nsfndikl~iilnpanel rendered in a particularly charming
manner is carved on the ceiling of the Ri~pBlaSangameSvara temple
at Bhavanasi Sangam/A.P., dated C. second quarter of the 9th ten-
tury (SOUND*R* 134; 'g' 138; 23 C)' all
its deities can be viewed correctly from the western side (i.e. the side
where the sanctum is located). There are no dividing borders
between the figures, as is common in the AIalnpur region. As ill the
former the dikl'Liias One hand the 'ear 0")
in the gesture of salutation (valzclnllanzlldrz)and an attribute in the
other. Prominent attributes appear only in the cases of Yama (staff
or club) and VByu (standard). All other gods, including the four-
armed iizna, hold only small objects, which are difficult to identi-
f ~In ' some their hands seem be empty. characteris-
tic details of the RDpBlH Sangameivara panel are the flame-like
coiffure of Agni (cf. PBpanItha, Paitadakal, fig. 56) and the fact that
only Indra is shown in the company of his consort, which recurs at
Dharmapuri (fig. 267).
An important feature at Bhavanssi Sangam is the presence of
dikpiilas sculpted on the exterior walls of both temples (Rfipsla
Sangameivara and BhujangeSvara) in slightly recessed knr!?n (tor-
ner) niches' without any parallel
for early 'li'c1'51fl representations on the exterior wall is at Aihoie,
temple no. 52 (fig. 139), dated to c. second quarter of the 10th ten-
tury.281 There, the gods are again positioned in the karcza sections,
but they appear at the very bottom of the wall, where they are placed
without any recess or frame.
The typical four-armed dikplilas of the .&lalnpur region are
known in a number of stylistic variations (figs. 141-144).282 They
usually hold the upper left hand in vandnnnmudrd and have their
torsos turned towards the viewer. A halo (round or oval) is present
in most cases. Only solne of the vehicles show a sign of movement.
It is quite likely that this schematized and obviously very colnnlon
theme for ceilings persisted for some time. So far, it is impossible
to determine the relative chronology of this group of images.
A specific, quite different group of reliefs is formed by those
dilcpiila ceilings which contain a larger number of figures. This
group occurs to the south-west of the regioll discussed so far (e.a
fig. 140). The dikp& images at Hemavati, Betfir283, and ~athasi-
varama (fig. 137) are accompanied by their consorts and surround-
ed by a large array of figures (particularly at Betfir), which results
in a Izorror v n c ~ ~effect.
i At Hemiivati, stylized clouds form the
background of some of the panels. In these cases, the relatively large
size of the vehicles is noteworthy.
The southernmost site, Dharmapuri (Dist. ~harrnapi1rilT.N.)
has two ceiling panels in which the deities are arranged in an iden-
tical order. The execution of tile reliefs was obviously carried out
after the same blueprint. Except for the superior artistic quality in
the MallikBrjuna temple (figs. 266-274; ill, 23 the central .
image (UmLsahitamiIrti in the K$mBksyamma) is the only point
difference between the two ceilings. Although the reliefs have
some "easternv featllres (no consorts except for Indra; separating
ways to certain examples in southern Karnataka, it is qoite remark-
able that important characteristics such as tlje proportions and the
empty background are more relniniscent of tile Aihole piece
27"0r a considerable period aftcr Bidi~ni,Cave 3, the horse re~naineda "floaL-
ins'' vehicle in thc western Deccan. ~t Aiho!e, H u ~ ~ a ~ ~ ~ a ~it ~was a g probably
udi,
"trihted toAgni (fig. 501, here at Aralaguppe to Nair!.ta, and (p~~obably later) at Kam-
badahalli (SOUNDARA RAIAN1986: pl, 619) and other sites (C.& REA1896: CV*l ) it
becalne Kubera,s vehicle,
'" ~ O U N D A R ARAIAN1986: pis. 395,404,405; DACENS1984: ph. 799. As Ear as
can be judgcd Crorn the pliotographs, ,hc sequential order oftIle deities starts with Indra
0" the (north-)east. The animal vciharlns appcar to bc s~nailerand stiffer as colllpared
10 the ceiling panels.
"l SOUNDARA RAIAN1986: 147, p]. 447 (~outllwall). On the soulh(-cast) is,a
figurc of Agni; itrnlgcly, tile south(-wcst) has a figure riding m rnlhmpr~n~rphlc
vciharzo. This could bc cithcr Yalna on an exccptionnlly an~~ropol~orphiscd
rnrlhi,?a Or
- InOrC likely - Naivta. It is likcly that Yams has been omitted, as ii seems that there
enough spacc for dikpdlas on thc wcstcrn face of the tcrnplc. The order starts
wit11 iS5na on the (north-)east (fig. 139).
2. ~ h c Inay
y local idion. o~pa,.t~cu,rrw o ~ s h o p sbut
, chmnologi('l
differences. B O ~ Iaspects
~ also have caused tl,c varialions occurring at the srte
Pratako!a/A.P., a sitc, which has yieldcd five sets oldikIjci,ns (three scts on ceilings, lor
tIleolhers see below; DAGCNS 1984: phs. 735,736,746,752-755.758.759)
rcveai particular differences regarding the proportions (vehiclc included), attire,
ber of arms, presence of a halo, etc. Thc already mentioned complete set in the
pur Museum (GASTON 1982: pl. 31) belongs to the salno group ofreliefs.
283 KalleSvara temple, see SOUNDARA RAJAN1986: 214. The telnple is classified
unc"~' "N~!"llb= of I-lEmivati", but it is not rcgardcd as a Nojamba telnplc by CO11w
1989.F~rtlledik~"~"~eL.~~. SIIUKLA 1972: pls. XLIII (yams), XLIV (Nair!.ia), xLV
("Vs~u". but it sllould be Kubera, who rides a horse in this region). The celltre of ''le
ceiling has an unusual representation of a sixteen-am1cd Siva-~aj~ntalta (SouNDARA
RAJAN1986 637).
284 TIie second ceiling is located in the KBm,iksyammntetnp1e, which is adjacent
10 the MdlikSrjuna tcmple. I do not agrce with Colrrx~ (1989: 258) that
KSlnlksyalnma tclnple is earlier than thc Malliklrjuna. I would suggest that if any
significa,~chmnologicrl gap exists all, llle ~~~i temple slloLlldbe posterior, as the
ceiling re\icfrelns e be a copy of the sllperb Mallikjrjunil
 (figs. 130-135). The attributes held in the respective right hands are:
Indra - small thunderbolt; Agni - spear or torch; Yama - staff or
club; Nairrta - sword; V a r u ~ p- noose; Vayu - flag; Kubera - club;
IBBna - trident and an indistinct object in the lower left. The latter is
four-armed as usual.
Armamalai285 in the North Arcot district of Tamilnadu is situat-
ed close to the border of Andhra Pradesh, in the Palar river valley.
A unique testimo~lyis found in a Jaina temple built in a natural cave,
viz. a ceiling painting depicting dikyllas in an oblong grid. Unfor-
tunately, very little can been seen on this part of the western ceiling.
Only parts of Agni and Ya~naremain, who, along with their con-
sorts, ride on their vehicles amongst the clouds286 (SARMA1992:
PI. 162). Agni and Yama are shown heading in different directions,
which is not a rare feature in comparable ceiling reliefs (cf. ill. 23 a,
d, f, i). Judging from the published photographs, it looks as if the
two deities were separated by narrow ridges. According to
BA~KARAN (1975: 29), however, they are separated only by a thick
black line. The paintings have beell assigned to the 9tI1110th cen-
turies (ibirl.: 30).
Summarizing the facts about dik1~8laceiling reliefs, the large
variety in some details of the layouts of the panels is the most
remarkable fact. The group " ~ i v a - c u 1 n - a ~ f a d i k ~ l Imust
8 ! ~ "have
formed an important religious concept and was thus given particu-
lar attention. The depiction of Jaina subjects in the central panel in
the Jaina temples does not contradict this fact.
At least one more concept of the placement of dikp8las in the
Deccan region is found in the NBgeSvara temple at Pratakota,
Nandikotkur Taluk, Dist. Kurnool/A.P., the ceiling relief of which
has already been mentioned. Tile temple belongs to C , the 9th-10th
centuries (DAGENS1984: 28, 508). On the temple tower, a few
of mouldings below the iiklzara, images of four seated
dilv8las have been found (ibid.: phs. 743,752-755). On the (north-
)east, Indra is depicted, with a narrow portioll of a frontal elephant
head sculpted as a seat for him (ibid.: ph. 752). On [he soutll, Agni,
Yams (ibirl.: phs. 753,754, respectively) and Nairrta (fig. 275) have
been placed in their respective positions, a vehicle bei~lgdiscernible
Only in Nair~ta's case (the anthr~~olnorphic nara). The ilnages
resemble the deities on the exterior wall of the temple phs.
750, 751). They wear high crowns, have a slnall halo bellilld the
head and are of good workmanship, wit11 well-balanced proportiolls
""d serene expression. Their attributes cannot be identified at all.
Finally, in the NfigeSvara at Pratakota, two more separate
ilnages are extant, sllowing Varupa and VQLI (figs- 276, 27%
which must have ollce belonged to a third set in the temple. They do
conform to the ilnages on the vir718rn, as they are four-arlned and
have larger vehicles. In this they more closely resemble the ceiling
Sculptures in this temple (ibid.: ph. 746). The slabs are, however,
slnaller than would be expected for ceiling panels. The sizes of the
64 and 66 cm, respectively, would be more appropriate for
niche images.
A unique occurrence of a clikpcla frieze 011 a lilltel is
from the Do?deSvara telnple at Hemrivati, C , second quarter of
the 10th century (SOUNDARA RAJAN1986: 210-213, PI. 631). The
lintel represents the main decoration of the llla!z~fll~adoorway
the west-facing temple. Bordered by the scrolls of rflakarfl tails
both sides (the left one being that of VarLll!a's nzal(ara), the
centre of the relief is formed by a group of musicians and other
flanking an eight-armed Nate.$a.287 The dikl?Glas on their
leaping vehicles are grouped according to the orientation the
Indra, Agni, Yams, and Nairrta (from the left) placed On
the southern side, and Varuna, Vgyu, Kubera, and k n a appearing
On the norther11 side of the lintel. All are shown moving towards
the centre. The guardians salute the dancing god with their rear
hands and do not carry any prominent attributes. They are not
shown in the company of their consorts, which is exceptional in
the region.
Another frieze has been preserved at Alur (SARMA1992:
pls. 75-77), one of the sites with ceiling reliefs (ill. 23 a). It is a con-
tinuous frieze on three walls of the Nandi nza!zdapa, with all the
deities moving to the proper right. On the eastern face, the central
figure is the god KBrttikeya, the only deity without his female coun-
terpart, and the central god on the south is - appropriately - Yama.
On the north, Indra seems to be in the centre, but, strangely, an
elephant rider appears on each of the three sides. The animals are
all shown in a trot. Those on the east and on the south step on a row
of clouds.
The art of the Rristrakota period at Ellorii has produced a few
noteworthy representations of directional guardians. A monunlental
variety without ally early parallels is the frieze of saluting dikcl~iilas
on the outer (western) face of the gopiira or the KailBsa temple,288
dating to approximately the third quarter of the 8th century (fig.
278; ill. 25).289 They are depicted in high relief and are separated by
neatly carved pilasters. The guardians are depicted unusually large,
but their forms reveal a lack of aesthetic feeling and a rather poor
rendering of proportions.29o111the wall segments beneath the riders,
other figures - most probably celestial couples - had been sculpted,
but little survives of them. The southern side of the western facade
is badly eroded. Only parts of the images of Indra (no. 16) and Yama
(no. 17) are discernible there. Interestingly, Indra seems to be
accompanied by his consort291Space for two more dik]?Glas remains
on the south (nos. 18, 19),292but the carvings are not recognizable
any more (HESTON1982: figs. 1, 2). While Varuna (no. 1l), Vgyu
a ~ ~9;~fig. 278) are depicted on the 11orth
(no. 10), and K ~ b e r (no.
together with KBrttikeya (no. 81, three gods, Agni, Nairrta and Piiina
are missing. It seems therefore that the set was left incomplete. Most
probably it did not include Tiiina. It appears that Kiirttikeya replaces
Piiina, as he occupies the place in the sequence "Varuna, V Q ~ ,
28s S.T. BASKARAN1975; SARMA 1992: 201. Further refcrc~lcesare given by the
authors.
286 I do not agree with BASKARAN (1975: 29) that Yama's consort D l i i ~ is~ n ~ ~ ~ ~
seated in front on the buffalo. It is quite apparent that she sits behind as is commoll.
287 COIIEN (1989: 90), who dates the temple slightly carlier, c. 920-930 A.".,
describes the scene (ibid.: 74): "to Na!iSa's right, Gal@ da~lccsand Pjrvati wa(cIlesn.
The presence of GaneSa could not be ascertained.
2% Thc individual panels of the screen-wall with the integrated gol)ri,n have becn
numbered by BURGESS (1 883: 35-36). These numbers will be referred to here. Regarrl-
ing the identification. thc autllor is obviously wrong in labelling tlle deer-rider as
Varu~y(no. 10) and tlic 11lcrkaro-rideras Niqti (no. 1I).
2x9 SOUNDARA RAJAN1986: 111-124.
290 This is pxticularly reflected by the proportions. I11 the case of Vriyu, the god's
head appears disproportionally lmgc (llluc11 larger than Kubera's and K S r t t i k ~ ~ ~ ' ~ ) .
Kiirttikeya rides his peacock ill a very clulllsy manner. Morcover, sincc the bird appears
to be standing on solid ground it is odd to see the legs ofthc rider reaching beyond those
of his vehiclc.
291 Although I was not able to detcct a female figure, all authors describing this
pancl have accepted this identification so far (e.g. BURGESS 1883: 36).
292 In SOUNDARA RAJAN'S table (1986: 1231, only tllrce dikpcila niches have been
crro~lcouslyad~nittedon the southenl side. The last niche is nowadays hidden behind
~lioder~i wall segnient that was probably erected to preservc the corncr portioll fro],,
being destroyed (HESTON 1982: fig. 6).
293 Since the days of BURGESS (1883: 351, this god has been identified as ~ g ~ l i .
However, it is quite apparent that only deities representing the "northcm spectrummof
the directions have been chosen. The sanlc attitudc js reflccled on tlle S i v a - ~ o g ~ i ~ ~ ~
panel inside the temple (fig.280)-which also fi~ceswest. Thcrc, 011the right side (soutfl),
Indra and Yania are shown juxtaposed in the same mainer, Agni being supcrilnposed to
Nairrta on the exlrerne left. On the left side (north), the dcities appear to bc the salne as
011 the exterior (of course, with Karttikeya being replaccd by igli~la),but this sequence is
exactly opposite that on thc gol)liI'cl (from 1. tor.): Varulp, Viyu, Kubera, iSina.
Kubera" appropriate for Tiiina. The frieze has been regarded as
coeval with the other parts of the Kailiisa temple and dated to c, third
quarter of the 8th century.
no. 17 no. 18 no. 19
Klrttikeya Kubera Vlyu
I
no. 11 no. 16
Varuna Indra
Ill. 25: EllorB 16 (KailBsa temple), gopurn, exterior, sketch of
dilcprlla niches (numbers after BURGESS1883)
A stylistic comparison with the contemporaneous ceiling panel of
the KiiSiviivanfitha, Paftadakal (fig. 129) proves the rendering of
the deities and their vehicles to be much inferior at Ellori. It is
indeed likely that the dikprllns as guardians of the gopura have been
carved earlier than their counterparts on the eastern (inner) side of
the same architectural unit (fig. 279). The sculptural treatment of the
figures is much better in a third group of dikpiilas, situated even
deeper within the rocky mass, on the main temple, opposite the Nan-
di rnan&pa (figs. 280-282). The group on the gopura exterior is
clearly the most stiffly executed and least artistically convincing
one. The deities are almost completely frontally represented with
their legs held in a manner that would require the vehicles to be
shown at least in a three-quarter profile. However, except for
KBrttikeya's peacock, which is shown frontally, they are rendered
in full profile. The almost frontal head of VByu's mrga is only a
slnall attempt to overcome this dichotomy. By portraying the deities
frontally, the artists avoided the depiction of a torsion in the lower
body of the riders, which had been already achieved -more or less
successfully - on the NafeSa panel, EllorB 29294 (fig. 21). A certain
variation is seen in the different attitudes of greeting. Kfirttikeya is
exempt from this attitude, since he belongs to the higher gods. He
holds a short lance (r.) and a cock (I.). Among those of the northern
dilcpnln group, only Varu~!a holds an attribute, a flower bud.
It is difficult to decide whether the absence of iSBna occurred
because of his identification with Siva (as is obvious in Ellor%29,
see above), the temple's main deity. In any case, KBrttikeya, his son,
has been portrayed as the general of the army of directional
guardians. They are prominently shown as they come to worship the
main god, h a , and, at the same time, protect the entrance by their
monumental presence.
In three other instances at Ellorii 16, the dikprllas appear again
as witnesses of important mythological events: Durgg MahisBsura-
mardini's victory over the buffalo demon (fig. 279); Siva's medita-
tion (figs. 280-282); and RBvana's shaking of Mount KailBsa
(SIVARAMAMURT~ 1976: fig. 65). The two former cases are the only
panels at the site, where all eight guardians are present.
On the MahisBsuramardinT relief (gopura, northern side of the
entrance, facing north), the sequence of the deities clearly refers
to the directions. Five flying couples in the lower row colnplete
the assembly in the upper sphere. Regarding the dikpalns, there are
(fig. 279)295:
upper row (from l.): Indra, Agni, Yama, Kubera, Nair!+ta,
Varuna
lower row (from 1.): fiiina (? two-armed; extrerne left), Vsyu
(extreme right).
Thus, the guardians of the east and the west are placed in the correct
corners (left and right, respectively). The guardians of the south and
the north are, interestingly, portrayed as movillg towards each 0th-
er near the central axis of the panel. Only Yama is armed with aclub.
The small objects held by the others in their lowered hands are not
discernible, but noteworthy is the fact that none of them raises his
hand in the gesture of salutation. As on the exterior of the goPLlrn,
the dik~~iilas wear high crowns, a belly band (mdarabarxdl~a),the
brahmanical cord (~il~avitn) and the usual ornaments necklace,
bracelets and keyuras.
In this relief, the sculptors have collvincingly portrayed some
the guardians moving to the proper right (Kubera, N a i ~ t a in) three-
quarter profile, along with their vehicles. Of the others, either the
vehicle, or the rider, or both are shown in prorile.
The Siva-~ogiSvara296relief probably represents the apex of the
successflll inclusion of the directional guardians in a monume1lta'
relief panel (figs. 280-282). A dramatic contrast exists between the
central, static, eight-armed god and the dynamic element, the
dik~nlas,who are the only attendant figures. In order to emphasize
this contrast, the dynamism of the group had to be stressed, in which
aspect the sculptors were particularly successful. BONER (1962:
201-202) gives a somewhat poetic description of the directiona1
guardians' function:
The DikpBlas sunounding Siva and the Tree are arrayed in "
semi-circle like the dome of Heaven. As elnbodimellts of
eight directions of space they represent the ubiquity of Heaven+
and by analogy, the all-pervasiveness of the Spirit. Their move-
ments are aerial, quick and lively, unhampered by weight Or
other material contingencies.
The panel faces west and it reflects - with a few lnodifications -the
same sequence of directional guardians as on the gopurn, which lies
on the same axis. The dikprllas are, from left to right (north to south):
Varuna, VByu, Kubera, Iiiina, (central figure), Indra, Yarna9
Agni, Nairyta.
As on the Nafeia panel in Cave 29 (fig. 21), the dikprllfls do
share the upper sphere with any other godlings. All deities along
with their vehicles have been portrayed from different angles, most
of them shown frontally or in a dramatic three-quarter profile (VByU9
K~lbera,i58na, Indra, Yama), The full profile had been chosen
only for some of the peripheral figures. Likewise, the gestures are
greatly varied, including the rare folded hand gesture (nfijfllinz~ld~~:
V a r u ~ aVByu,
, Kubera, k n a , Indra, Nairyta). fiBna is four-armed as
usual. No attributes have been depicted, except for Yama's staff, It
is important to note that ISiina and Indra are shown with their
consorts, which inay already reflect the convention coinlnon in
9th-lOth centuries in the western Deccan.
'" sigllificantly, on the Kalyinasundara panel in the same cave VARMA
l988: phs. 146-148). the legs of so~neoftlie dikpcilns are rendered in the sallie way "
on the Kailasa gollum. Thcy also have been co~nbinedwith vcihnt~crssliow~llargely
profile (Yama, Indra, Varuna). But there, the skilful artists did not allow such a
dichoto~n~.
295 BLJ~~cBss (1883: 29) has identified so~ncof the dikl,iilns incorrectly. proceed-
ing in a rather unsyste~naticway. O~iceagain, he has not rccogllized Kubera's presence'
Furtllcr, he has identified Nair~taas Vis~u,and Vayu :IS iSana, and hc was also liot
aware of the fact that a s~llallridcr appears in the lower row on the extreme Ief.
'" "me authors, as A. BONER(1962: 195-208), refer to il as JRBna- or Yoga-
Daksin~m~lrti. This lnay be allowed only when using the [ c r ~ lin a broader sense
(cf. B * ~ E ~ l 470-472),
n~: as the image is not facing the south.
 The great, extremely deep-cut Riival?Bnugrahamnrtirelief on the
southern side of the temple (reference above) contains two groups
of figures in the upper sphere, with the one on the right consisting
of dikpdlas. They are placed among dramatic cloud formations.
Viewing the group from left to right, an eroded portion creates the
impression that the first guardian has been destroyed. This is not the
case, however. Viiyu (? vehicle damaged) follows, and behind him
Indra. On Indra's right, who is still largely shown in profile, the
shapes are extremely difficult to identify. It is impossible to detect
any dikpalas, which may have been carved in that part of the relief.
In any case, the group of eight must have remained incomplete. On
the left side of the central divine couple, a group of four seated fig-
ures is also depicted as salutillg Lord Siva. Shown bereft of vehi-
cles, it is unlikely that they were conceived as a part of the dikpdla
group.
At Ellor%,two more occurrences, both fragmentary, point to the
Popularity of the directional guardians in the region. In the so-called
Dagzvatiira cave (no. 15) at Ellora, some of the rlikl~dlasframe the
Kalyanasundara panel on the northern wall (C. BERKSON 1992: 297-
299). The group seems to be nearly complete, but the individual
Sods are difficult to identify. In accordallce with their assigned
directions, Indra appears on the right, perhaps followed by Agni and
Yams, while Varuna seems to be on the left, together with VRyu and
K~lbera.It is doubtful whether Nairgta and h a have been includ-
ed. The excavation of the temple is dated around 7 5 0 A . ~ . ' ~ ~
The southern wall of the GoliAgeSvara temple at Bil<kavolu/
A.P., belonging to about the lniddle of the 9th century, has an inter-
esting but rather unique, incomplete group of dikpdas as its central
theme. Indra occupies the central position of the jariglzz (fig. 283).
His body exhibits a very slight flexion (iibhaliga), while Agni on the
left and Viiyu on the right side are seated above their vehicles
(SOUNDARA RAJAN 1986: 168, 527). The fact that KBrttikeya on
the west is flanked by a serpent couple, which belongs to the aquat-
sphere, probably reveals an attempt to include directional sym-
bolism, but so far this cannot be confirmed for the mentioned grollP.
While a pot-bellied Agni squats above a goat or sheep, holding the
usual rosary (r.) and waterpot (1.),298 v~gyuis seated above his ante-
lope in a graceful but unrealistic manner (fig. 284). He holds a bil-
lowing cloth, which is quite unexpected so far in the south (cf. sec-
tion 2.5.6). The images are not surrounded by any frame. They are
carved directly on the wall surface and have an integrated, thin socle
Portion.
The image of Indra is found within an ornate niche, the most
prolninent parts of which are a kirttinz~tlclza("face of glory") on
and the tails of two i~lward-facingn l a k a m ~with riders. The stand-
illg deity with a rather small, unspectacular crown has broad shoul-
ders and holds a thunderbolt (vajm) in his right hand, in front of
the centre of his body. The left hand rests in katihastfl. Indra is
Separated from a small elephant below by a narrow socle.
The Bikkavolu images were either inspired by an llnknowll
Source, Or they represent an independent concept. If 'lad been
Conceived as a separate group, Indra and V2yu were perhaps
chosen to represent deities of the Vedic period.
2.3.2.2 Central and Northern India
Judging from what has come dowll to us, it seems that the complete
astfldikpfila scheme was introduced slightly later in Centra1 India
(Indor, District Guns) than in Western India (Cittaudgadh,
Ki?likiimHtHtemple; Osi2fi,H H - ~ )The
. complete destruction of the
important site ~ ~however,
~ ~Ilave effaced
~ formerly
~ j exist- ,
ing, metropolitan beginnings of the theme. Evidence is particularly
scarce in the northern part of the region.
There are two highlights of dikpdla representations in Central
India: the first one is the uniquely conceived, innovative Siva tem-
ple at Indor (District Guna1M.P.) with the earliest extant colnplete
set; the second one is the Caturbhuja temple at Gwalior, which is
important because it is securely dated to A.D. 875. The artistic qual-
ity is much higher in the former case. Thus it seems that, parallel to
the Western Indian development, the prototype at Indor exhibits a
subtle inspiration, which was lost during the subsequent depictions
of the theme. This decline can be clearly observed in the Caturbhu-
ja temple.
Harihara Piirvati 6 Varuna
5 Vayu Goddess
Kiirtti-
keya
A
Gana
4 Nairgta
3 Yama GaneSa 2 Agni
Ill. 26: DBlig, Siva temple, plan
The development of the later period should start with the small Siva
temple at Dang (District Bhind/M.P.), not only because of certain
archaic structural features and its sculptural decoration, but also,
because the dikpiila set is still in an incomplete stage. The temple
has been assigned to the late 8th century by T R I V E D I ?however
~~ it
should be dated somewhat earlier, perhaps just after the middle of
the 8th century.300
The temple shares its incomplete dikpdla represelltation with
the two temples at Alnrol (particularly, thc DBnebriba, ill. 15) and
the PiplftdmiitVg temple at Pipiid (ill. 20). Like these temples, it also
includes Sarya in a karrza position on the eastern side. The compo-
sition of deities in the karrza niches differs in all three cases. ~t
DBng, apart from Siirya ( I , fig. 1451, Harihara and an unidelltifiable
goddess on the (north-)east301 (fig. 151) seem to be - uniquely -
combined with the dikpalas. The five directional guardians are Agni
(2, fig. 146), Yama (3, fig. 1471,N a i ~ t (4,
a fig. 148), Varuna (6,fig.
297 Thcre is an i~iscriptionof tlie Ri:~!~akO!aDantidurga in the lllalJ~/fl,,ain front
of the cave (SOUNDARA RAJAN1986: 11l), referring to a donation lnade to [hc temple,
~1t110ugIi the posqibility re~nainsthat it predates the niiddle ofthe century,tile shrine is
treated under the heading "Later Karrga style, c. A.D. 750-880". This ciassification is
also regarded as appropriate here.
2% See M. RAMARAO 1964: 8, figure no. 7. The author does not recogllize the
deity as Agni. Its counterpart V5yu (no. 9) has been identified will1 Agni instead,
299 TRIVEDI 1990: 79. The author is tlie only one who desclibed the tclnple $0 fLlr
(j/~irl.:75-89). VIENNOT1976 and EITA 1991 do not include it.
300 The proposed date is only based on the relative chronology of the telnples at
Naresar ''ld Anlrol'
301 The upper part of the image is damaged, and its lower part is concenlcd by a
cloth. The figure has a pronounced halo, a large coiffure and appears to hold a lnilror
its face on its left side. Although there is no co~nparablecvidellcc known to us. it
may be hypothesized that the deity is associated with lunar ~ymboliy~n. ~t be a
~ e r s o n i f i c a r i o l l o ~ m ~(cn11drir71ci).
~~~~gll~
150), and VByu (5, fig. 149). Their most common headdress is the
demonic Crcllzvnkes'n, the only exceptions being a "hair-crown"
(jn;dnzuku!n) for Agni and a topknot for VByu, who stands in the
rigid sainnpiicln posture. The artistic quality of the images is rather
poor, with the legs more or less cylindrical and stiff, and the vehi-
cles rather crudely rendered, except for Varuna's goose (fig. 150)
which has a very lifelike sliape.302The bird (cock?) perched in front
of Yama needs particular mention. Similar birds characterize the
Yama images on the Harihara temple no. 1 at 0s'iiifi(fig. 190) and
on the Siva temple at Indor (fig. 154). Both temples probably date
from the middle or the third quarter of the 8th century, and this is
also the likely date for the DBng temple.
As typical Central Indian vehicles, the quadruped for Nairrta
and the goose for Var~lnaare clearly recognizable. Whether V ~ y u
was accompa~iiedby a horse is not certain, because of the poor exe-
cution and the severe damage to the image. However, this was prob-
ably the case, since the god's scarf leaves no space for the horns of
an antelope (fig. 149).
111.27: Indor, Siva Temple, plan (after MEISTER1982: fig. A).
After Dlrig, which appears to be a relatively late example of a
hybrid, formative clikpdln set, the ~ i v atemple at Indor (District
Gu11alM.P.)presents an impressive picture of the progress made in
~likpiilnrepresentation in the latter part of the 8th century.303 The
sanctum rests on two circular plinths and rises in an intricate geo-
metrical form with nine blzndrn (projecting wall) portions alternat-
ing with ten right-angled piers (ill. 27). Knpili walls enclose a
llzukhanza!z~apnwhich leads to the square satxctum. Six of the nine
blznrlrn niches contain dkpdlns. The circular shape would have
allowed for a more satisfactory spatial order the directional
guardians. However, this was not carried out, and the cardinal direc-
tions are, as usual, occupied by the members of the main god's
family (parivdrn). Indra and fiiina are installed in the wider kapili
niches. In this aspect the temple follows the earlier model of the
RlmeSvara Mahldeva temple (ill. 14), where two dikpdlns have
been placed in the knpili sections. The standing dikpiiln images at
Indor (figs. 152-159) are characterized by softly modelled bodies,
squarish faces, and a generally relaxed, almost graceful, individual-
ly varying posture. It is quite remarkable that their build seems to
have been consciously varied, probably in accordance with the char-
acter of each deity. Thus, Vgyu (fig. 157) is shown particularly slen-
der, which conforms well with his fast-moving nature. At Indor, all
the three typical Central Indian vehicles can be found: a four-legged
creature for Nair!ta3O4 (fig. 155; corresponding to Yams's, both
more closely resembling a bear than a buffalo), a goose for Varuna
(fig. 156) and a horse for Vlyu (fig. 157). All the other deities have
vehicles which were also colnrnon at Osiafi during the same phase.
Regarding the attributes, the noose in Nair~ta'sleft hand (quite
tinct from Varuna's noose), the round fruit in Varuna's right band
and a similar fruit (?) in Kubera's right hand in front of his chest are
noteworthy. These objects have been added to the ones already
known from OsiBii. kina's attributes differ froin the OsiLli scheme:
they are, from the lower right, the boo~i-bestowinggesture (varnlJ*),
rosary (?),trident305 and waterpot. The fine set of individually ren-
dered deities is stylistically closely related to the clikpiila ilnages
the RBmeSvara MahHdeva temple at Alnrol (see below).
Two loose images from Kola in the neighbouring ~hivpllri
trict, one of Yams (fig. 160)306and one probably of PSlna,307are
close parallels. The rendering of these latter figures is even more
subtle and refined, and it could be surlllised that they are roughly
conte~np~rane~ils with the Indor set.
A loose relief of Varuna squatting on a lotus above a large
goose (fig. 316)308 is closely related to the Indor i r n a g e ~ . ~ ~ ~ ~ ~ ~
differences between this and the Indor dikpiilns lie in the boldly
cuted features, the even broader face and the seated posture of Ihe
former. A 6th century date as given by some authors is ]nuch
early. It may at best be slightly posterior to Indor, which could be
justified, among other arguments, by the rule "first seated, later
standingW.3l0
Illdor represents the apex of early clikpiila depiction in central
India, not only concerning the artistic cluality, but also regarding the
occurrence of the three indigenous vehicles: quadruped for Nairgta'
goose for Varuna and horse for Vlyu. A few decades later, solne of
these vehicles were replaced by others, which were also in vogue In
Western and Southern India. Thus, the goose was replaced by the
mfllcnr~,and the horse by the common anteIope.31i ~airyta'sfour-
3" Another, s~nallcriniage of Varu~!a with a bird is found in a secondary P ~ " ~ ~
above the door (TRIVEDI 1990: pl. 204, centre). 1" this case, the bird is show11 with a
thin neck and a long tail, resembling a peacock.
"' MEJSTER 1982a; TI~IVEDJ 1990: 95-100; DEVA1991: 17-19. TluvoDr
a date of about A.D. 800, while DNA favours the third quarter of the 8th century. ,
"O The lack of co~nparableexamples led K. BRUIIN (196O: 190-191) to identlfY
the Nair!.ta i~iiageas Yama.
We hnvC obscrved that in Western Illdia, the videnl is generally carried On
igsna's right, whilc in Central India, it is placed on his left side. At Osisfi, HH-',
exccptioll with the tridcnt on the left side (fig. 1')5), thus f~lrthercorroborating lhe
olllcr Central Indian features of this di/p<ln set.
"' A Peculiar feature here is the object held in the deity's right hand. A
(wheel) is the most probnblc identification, as it is pcrfeclly round. The ~ t h e r ~ ~ ~ ~ i ~ ~
it^, a noose, was not a typical characteristic for in thc pcriod under discussLon
A cokrfl, however, would bc absolutely singular. The staff held on the left side is not
an unusual feature.
"' The Sculpturewas published as "Siva" ill the Annual Rept. of ~ r c 1 l a e o ~ ~ g ~ ~ ~ ~
Dept. of the Gwalior State, 1923-24, pl. 111 b. 11 is presently housed in the
Arc1laeological Museum, Gwalior, acc. no. 62. The image has an cxtrfle1y sl!nder
body - similar to the Indor Viyu (fig. 157) -and also closely resembles the 1ndorlSana
(fig- 15%. Solne of the dirfering details arc [he absence or a watcrpol (here kfl!illnsla)
alld the front right hand fccding thc bull. A wcll-knowll figurc OSa dancing lndrB1!i
(s~~nemuscam, no. 65; BIIAWACIIARYYA 1991:pl. 55) belongs to t11c same site. Unfor-
tunately, it is not known to which te~nplcthe images of i&na (?), Yanla, and lndrani
once belonged.
"' The figllre Was first published and identified as V a r u p by VAN l'OOIJ
1985: fig. 10.
'" Its c0iffitl.ecan be best cornpared with thc Indor Kubera (fig. 1561% Ihe
attributes and the manner in which they are lield is exactly like the Indor Varuna (fig'
158). The costume and ornaments appear co be somewhat poorcr, as the kc)'u''cls and
the sash are absent.
'l0 This seelns to bc true for Ccntral India. The dikjx7lns on the ~ l l
MahPdeva temple at Arnrol are seated, and their successors on the DXncb.3ba telnple at
[he %me site are shown partly scated and partly standing,
3 1 1 A rclniniscencc of the horse remains in the form o l a donkey as V5yu's vlih'"n
on the north-western subsidiary shrine of (hc Laksmn!la temple at ~hajuraho(TR1pATHr
1965-66: 120).
 legged vehicle, however, continued until after 900 A . D .At ~ ~the~ his vehicle; and fSiina carries a serpent as an additional attribute in
same time, the association of Kubera with his anthropomorphic his hands). The figures are well-modelled with clear, partly exag-
vehicle gradually disappeared, so that none of the guardians had gerated features.
Possessed a human-shaped vehicle for quite some time. During the Another one of the Central Indian Sfirya temples of the 9th cen-
9th century, Kubera was often shown with a pot below instead of a tury is found at Umri. This structure has been assigned to c. secoild
vehicle, or even without any associated object at all, except for his quarter of the 9th century by DEVA.The two-armed dikpilns appear
attribute. in their usual positions and are - with the exception of TSana who is
The temple site Batesarii in District Morena1M.P. is an impor- shown standing - seated with the right leg pendent (figs. 162-169).
tant landmark in the development of the dikpdln theme. Unfortu- The sculptures share their bold, sensuous shapes with the Keldhar
nately, the images of the nearly 40 structures could not be fully eval- images. At Umri, however, the figures are much more uniforlnly
uated for the present study.313 The main temple, variously called rendered. The only pec~liarities3~~ are the absence of Agni's vehi-
BhfiteSvara Mahiideva or BafeSvara Mahiideva and assigned 10 the cle (fig. 163; the lower part seems to be filled by "smoke"), the re-
late 8th century A . D . , ~shows
I~ the rare feature of four-armed introduction of the allthropornorphic vehicle for Kubera (fig. 168)
clik~dlas.Like the earlier four- armed directional guardians at Cit- and the fact that ISiina is only two-armed (fig. 169). The two-armed
tauclgadh, KulnbhaSyiima temple, it is apparent that they are not form of PSiina is likewise known in Western India in the 9th centu-
direct predecessors of the four-armed dikpiiln representations that ry. As at Indol; the headdresses of the guardians vary to some
started in the 9th-10th centuries. Thus, Agni (DEVA1991: pl. 39) degree. Indra wears a cylindrical crown, which resembles Varuna's
does not as yet carry a sacrificial ladle, which becomes a common at Indor (fig. 156). Yama, Varuna and Kubera wear a crown of
feature from the 9th-10th centuries onwards.315 Likewise, Kubera's braided hair (jn[din~ikcr~n). Nairyta and Viiyu have their straight hair
Pot, held in his rear left hand at Bafesarii (TRIVEDI 1990: 381, does tied into a silnple topknot, while Agni's and ISiina's coiffures are no
not belong to the later set of attributes as ]mown from K h a j u r a h ~ . ~ ' ~longer recognizable. When compared with the earlier images, there
His vehicle still has a human form (~znrn).Viiyu's vehicle, however, is a greater tendency towards representing the dikl>dlas with
has changed from a horse to an antelope (ibid.: 43). briihmanical cords, which is also seen on other temples. Only
One of the neighbouring shrines of the BateSvara Mahiideva Kubera and ISiina can be seen without the sacred thread here.
temple, which has been referred to as no. 4,317 has an unusual con- The Catunnukha Mahiideva temple at Nacnii "combines sculp-
figllration of deities. In the north(-western) knrnn niche, Varuna tures and architectural members from three different periods and
stands along with his vehicle, while Kubera on the north(-east) is enshrines a divinity older than the present structure" (DEVA1991:
characterized by his new emblem in this region, the pot (fig. 161). 69). Its construction is assigned to the middle of the 9th century or
In the (north-)eastern knrpz niche, however, an Ekapiida Siva slightly lz~ter.3~~ The extant dikl?dkns on this temple, Agni and
appears (DEVA1991: pl. 46) instead of Iiiina. Another nearby shrine, Kubera (TRIVEDI 1990: pis. 128,129), reflect an advanced type since
referred to as no. 3 (ibid.: pl. 44), shows Nairyta on the (south-)west they are four-armed. Agni's attributes are, starting from the lower
right: rosary, flower, sacrificial ladle and waterpot. In Kubera's
VByu on the (north-)west, thereby following the placement on
the Siva temple at Diiiig (ill. 26). It is also importallt to note that case, the objects are difficult to make out. His lower left hand holds
on temple no. 7 at Bafesarii, the knrrzn niches frarne scenes from a large rinlczilnkn (purse), and he probably holds a club in his upper
Krsna's life (ibid.: pl. 48). Therefore, even without studying the right hand. An image of Varuna housed in the Bhiirat Kalii Bhavan,
Batesarii evidence Inore closely, it can be said that, as far as the Varanasi (fig. 170), ]nay have once formed part of the set. It has
similar flying garland-bearers at the top and attendant figures below.
dik~filnimages are concerned, an experimental attitude seems to
The deity is four-armed and exhibits a serene expression, with the
have prevailed at this site. Since the lower parts of the telnples are
generally sparsely decorated (wit11 the exception of the almost inac-
cessible BateBvara Mahgdeva), it is very difficult to fit them into a
312 TRIPATHI (1965-66: 115) lists two types of quadrupeds, a bull-like animal a~ld
relative chronology. Their inception might have been even earlier, a dog, as Nairyta's vehicles at Khajuraho. The third variant is a reclining human figure,
around A.D. 775, as suggested by DEVA.This would make them 313 One reason for the lack of illclusio~iis the fact that it is very difficult to take

with thc much more ornate Indor temple.318Althougll they
lnaY actually be more or less coeval, it is difficult to measure the
"Ore "provincial" or simpler conceived group of temples at Nare-
Sar, Amrol, DHlig and Batesarii on the same scale as the "imperial", '.
intricately conceived Teli-kg-mandir at Gw6lior and the Slva
telnple at Indor.
The well-preserved Silrya temple at Umri, District Tikamgadh* known
the Siva temple at Keldhar, District Shivpuri, are good exam- ilnagc is probably later (see below). Admittedly, at Khajuraho, the association of a srlrk
ples of the stage of development reached at the beginnillg of thegth
century in Madhya Pradesh. Tile west-oriented ~ i v telnple
a at Keld-
which has been dated to the early 9th century, reveals a few
exceptional and illformative features of the directional guardians.
The dikpnlns are shown two-amed and standing, with the exception
Indra, who rides his elephant (TRIVEDI 1990: pl. 101), Kubera,
W110is depicted seated without a vehicle (ibid.:pl. loo), and ISBna?
who still appears in his four-armed form. Some details can be COm-
Pared with the Indor dikljalas (the bird at Yarna's feet; the sword and
in the hands of Nairyta), but others signify a departure from
photographs of the i~llagesof the main temple, because of the narrow space betweell
the outcr wall and the wall of the ambulatory. Another unfortunate fact that should
be mentioned here is the liberal destruction of Lhc heads of the images, which lllust
have occurred in relatively recent years (cf. VIENNOT 1976: ph. 85 with T~lveor1990:
pls. 38 and 39).
314 TICIVEDI 1990: 79-82; DEVA1991: 19-22. The latter prefers a date of c. 775 A.D.
315 The Caturniukha Mahddeva at NacnB (District Panna/M.P.) llas the earliest
example (DEVA1991: pl. 39) in which a large ladle is one of Agni's four attrib-
utes. The construction of the temple has been tentatively dated to c. 850 A.D.,but the
is not without exceptions (TRIPATIII 1965-66: 109). An early occurrence of the sacrifi-
cial ladle carried by a two-armed Agni is altested to at Keldllar (sce below).
316 TRIPATHI 1965-66: 122. Olleexception is llis no. 13, but here, thevesscl is llcld
in the front left hand.
317 "NO. 4'' is the designation used in the plate section (Dlivn 1991: pls. 45, 46),
but it has bee11omitted in the text (ibid.: 23).
318 VIENNOT (1976: 247) in her typological shdy actually assigned the Ba!eSvara
Maliddeva tenlple ("no. I") at Ba!esar?i ("BateSvar") to thc sanle period as the Siva
temple at Indor.
319 Tltlvanl 1990: 112-1 15. The telnple is llot included in V~ENNOT 1976 and DBVA
1991.
320 Unfortunately, no information concerning the vehicles of Varuna alld VSyu at

Indor model is seen riding; Kubera is seated without a Keldhar could be obtained. Thus, they cannot be compared to those at Umri.
vehicle; Agni holds a ladle in his right hand; Nairrta has a as 321 Dev! 1991: 151; Tluvenl 1990: 129.
eyebrows slightly raised, however. Varuna wears a ja!Sirnukufn preserved in the Jhansi M~~~eum.323 The published images conform
(hair-crown), a necklace, bracelets, anklets, a sacred thread, a leaf- to the features described above. Solne other fine specimens loose
garland (vanamdld) and elaborate ornaments adorn his lower gar- dikpdla sculptures are the fragment of an Indra image of fine work- 1

ment. A miniature inakara head and a kneeling attendant flank the manship from :Badoh, District Vidisha,324 and an image of Varuna
deity on the lower right and left, respectively. Behind them, two from Kjrital. (DEVA 1988: pl. 322), District Jabalpur (both in
attendants stand on each side. The lower right hand is held in the M.P.).
boon-bestowing gesture (varadamudr~),and the upper right holds a Two loose and completely dispersed, slightly oblong
large noose which is rather clumsily represented in two parts. The reliefs of Varuna and iijna (figs. 177, 178) must have once
upper left holds a large lotus bud and the lower right a waterpot. The belonged to a diki,filn set in Mathur,?/U.P. or its vicinity.
rather superficial treatment of the ornaments along with the slight- deities are seated in the same posture (dnksinapiprololdap~da), how-
ly aogulw, schematic shaping of the extremities suggest a date most ever, with little space below the right leg. Their ornaments are mod-
likely near the end of the 9th or the beginning of the 10th century. est, except for the large round emings which they share (&ina: only
These dikpdla images stand on a IOW socle, and those still found on right ear). Large ]eaves emerge sideways from behind their heads
the temple walls appear to have been loosely inserted into the niches. (Iijna: only on the right side). Their hands show graceful gestures
1t could be that the dik~filaspostdate the temple by a few decades, and their facial expression is benign Varuna holds a noose in
~n important, dated landmark in the latter half of the 9th centu- right hand and sits on a goose (fig. 177). has a trident in h s
ry is the rock-cut Caturbhuja temple at Gwiilior (ill. 28)322, situatedleft hand and holds a rosary in his lower A rather small bull
half-way up to the rock fort. The date of the foundation inscription crouches below. The set sllould be dated to the late 8th or early gth
is 875 A.D. 1t has already been briefly mentioned that no space was century. An G'early''bature seems to be the presence of the goosy
available to accommodate Indra and on the east (entrance) but the two-armed form of Tisna is a "later9. feawrc. Varuna S
side. An certain crudeness characterizes this temple. demonic coiffure (cirdlzvakeia) also points to the 9th century The
This structure is very crude in its sculptural and architectural deities are both bordered by a simple frame which is slightly dent-
embellishment, perhaps due to the poor quality of the local stone ed above ISiina's head.
(DEVA1991: 55). Froin the impressive rock-cut telnple complex at ~ a s r ~ r 9 ~ ~ ~
District KangraIHimachal Pradesh, a few loose sculptures remain
Thus, the details are mostly blurred (figs. 171-176). There are some from what must have been a complete dikpdla set. One fragment
conspicuous peculiarities of the regularly positioned deities: t\gni3s a slightly obese figure of Agni, wearing a diadem, with
vehicle seems to be either anthropomorphic or missing (fig. 171), flmes behind his head.326 While the standing image of ~ g nisi two-
Nairyta's four-legged vehicle resembles a lion (fig. ITS), Varuna's armed, Varuna seated on nzakam was probably f o ~ r - a r m e d . ~ ~ ~
head is encircled by a cloth resembling Vjyu's typical attribute (fig. sitting posture and the position of the n~aknrabears comparison with
174), and Kubera is shown without any vehicle or pot in substitu- the image of Varuna at Umri (fig. 166). It is likely that both images
tion (fig. 176). Of the specifically Central Indian vehicles, only more or less coeval, although the date "c. A.D. 800" has been
Nairyta's quadruped is still present, though not as the "boar ver- assigned to Masrur by DEVA(1991: 118). The ~ot-belliedfigure7
sion". Again, the guardians have been portrayed with a sacred probably of Kubera, is depicted seated on a lotus-throne
thread, with relatively sparse ornaments and with individually var- ddOrsed lions underneath (V.C. OHRI1991: pl. 3.18). The arrange-
ied coiffures and postures. The dikpdlas of the Caturbhuja temple ment with the lions is quite unique, but a parallel can be seen in the
conform to the characteristics of the directional guardians subsumed Kubera image at Umri (fig. 168), where this pot-bellied deity sits in
by TRIVEDI (1990: 43) for his third phase of Central Indian Pratihara manner on a high lotus socle.
temples (c. 850-950 A.D.): A rare instance of a late but somewhat or rather
'reduced' group of diki,Silas occurs at Gyaraspur, District Vidisha9
... in the telnples of the third phase they are invariably shown M.P. The Jaina Mjljdevi temple of 850-900 A . D features in its
two-armed standing with or without their vehicles.
~rotrlldingka~lzaniches on the i?za!t&pa's north. side Indra, seated
(l.), and Kubera, standi~lg(r.; K. BRUHN2000: 312; figs. 9, 10).
5~ i i y u Trivikrama 6 Kubera
BRUHNobserves that Indra's right, pendent foot is resting on a pot
which may form a reference to Kubera's pots - two small jars are
4 Varuna
depicted with the latter.

322
1990: 135-143; DEVA1991: 54-56. The in the latter publication
cn$. 27) obviously simplified, while TRlve~l'splan (fig. 28) reveals that the
Is
is ~rregularlys1]apcd.
323 S.D. T ~ l v t i 1983:
~ l figs. 52-55, 61, 65-66,
3 Nairyta 324 The aPParcnlly two-anned deity has a fine oval halo with a border Of lotus-
petals Indra wears an intricately shaped crown and exhibits sharply cut features ''
Is housed in the Central Archaeological Museum, Gwalior, and can be dated to "le
2 Yama GaneSa l Agni 9th-10th centuries.
325 DnvA 1991: 116-118,
Ill. 28: Gwalior, Caturbhuja temple, plan sketch of sanctum 326 In the publicatiol~by D.K. CHAKRADARTIIS.~. HASAN(1984: p]. 12) the sculp-
turehas been tentatively identified as aynk+, Tile bulging eyes lnay point to this
Ilfication.However, we do not know if the grotesque features were part Of the
of Agni in that particular region.
Apart from the evidence, which has been preserved in situ, there are
327 D n v 1991:
~ PI. 269; CRAKRADARTI/HASAN 1984: p]. 11. The dig~lificdportrait
a few sites which have yielded loose sculpt~~res. A particularly of Varu?a shows a diadem, heavy earrings, two necklaces, and anklets. He has a
prolific site is Siron Khurd, District Lalitpur/U.P., from which a behind the head. The god is seated on a boldly carved, voluminous maknra Unfor-
collection of dikpdlas of different phases (9th to l lth centuries) is tunately, the Panel has bcen severely damaged. A d i c f 0fIndi-a seated on his elepllant'
probably four-armed, has also been found,
 A very fine example of dikpiila representation, the Barwiisiigar
lintel frieze, marks the latest phase of the period under survey
(fig. 179). The rather well-preserved Jariii-kg-math temple of about
A.D. 900 or a slightly later date328 is located in the Jhansi district of
Uttar Pradesh. There are two dikpdla sets on the temple: one is the
overdoor frieze, where the deities are arranged in their correct
sequence from left to right. Regarding the number of hands, their
attributes and vehicles, the dikpdlas of the frieze largely conform to
their counterparts in the niches.329 All are shown two-armed. On the
frieze, the guardians are seated in rlnk,ri!ziil2ralanzbapiida. Their
vehicles are (from left to right): elephant, goat or sheep, buffalo,
quadruped (jackal?), mnkara, antelope, ram, bull. Their attributes
are also those commonly shown in the latter part of the period, with
the exception of an additiollal cup (? or bird?) in Yama's right hand.
The free hands are held in cinmudrii (thumb and index finger touch-
ing) ill front of the breast. Two tiny attendant figures are discernible
in the background of the panel. On the (proper) left behind Varuna
is a personage squatting on a fish, and on the (proper) left behind
Kubera, a wine-waiter stands on a lotus socle. The niche figures are
all shown in a standing posture.
The Barwiisiigar frieze can be compared with two other friezes
of standing dilcpalas which are no longer iiz situ. The first one from
Badoh belongs to the Gwalior Museum collection (fig. 180). Since
the deities here are four-armed, it is likely that it postdates the for-
mer frieze. The Badoh frieze is severely darnaged. It is bordered on
both sides by a female figure turned inwards. The clikl~iilasare shown
standing in identical flexed postures, with their vehicle or emblem
on the proper right. However, all the figures have haloes behind their
heads, unlike those at Barwassgar. The attributes have been com-
pletely destroyed, with the exceptioll of a severed head (lower left)
and a shield (rear left) in Nairyta's hands, and the jars in the lower
left hands of Agni, Kubera and I i ~ n a The. vehicles largely Cone-
spend to those at Barwjsiigar, except in the case of Kubera. He has
a single pot by his side with a tiny figure peeping inside or clasping
it. A third, fragmented panel should be mentioned here. It is a well-
'
known piece from the Heeramaneck ~ollection.330Thispartially pre-
served frieze represents the figural sequence only from Nairrta to
Kubera and was obviously an ~verdoordecoration. The large seated
figure was probably intended to represent PiZna, who would be
'?issing otherwise. The guardialls are all two-armed, except for the
Siva-f~Znafigure, which is four-armed. The occurring combination
Wester11 Indian and Central Indian traits confirms a probable
origin of this panel in the Kota region, Rajasthan, as suggested by
MASON(1993: 236-237). The vehicles correspolld to those of the
Bad011 panel, including the quadruped for Nairrta and the pot for
Kubera. A significant feature is the fact that VByu holds a standard
instead of the cloth, as would have normally been expected in cen-
tral India. The standing postures with the feet placed wide apart is
also "western" as is the fact that the trident is held in the rear right
hand. The panel with the two-armed dikl~zlashas
been assigned to the second half of the 9th century by the authors.
In Central India, the general introduction of four-armed
dikpiilas does not seem to have taken place before c. A.D. 950. An
development in the 10th century is the elimination of
niche frames for the sculptures (e.g. at Nohta, DEVA 1995: PI. 200).
A remarkable case or a hybrid dikpala image is fmcountered at
Sesai, District Shivpuri/M.P., on the Stirya temple of about the sec-
ond quarter of the 10th century. A deity occupying the usual posi-
tion of Viiyu, the north(-west), exhibits traits of both Varuna and
Vgyu (fig. 181).331 A rather representation of the billowing
cloth encircles the head of the deity. In this it differs from the Western
Indian tradition, where the scarf generally does not rise that high
(e.g. fig. 210). The vehicle of this "Viiyu" figure, however, is a
~nakara.The development towards a cloth-bearing Varuna was a]-
ready indicated at Ulnri (fig. 166) and at Gwiilior, Caturbhuja temple
(fig. 174). Thus, this hybrid form seems to have gradually developed.
Along with the absolutely stereoptyped postures at Sesai, this case
of a hybrid deity may be regarded as an indication of an increasing
lifelessness332in the old system of two-armed dikpdlas, which final-
ly led to their substitution by newly conceived four-armed direc-
tional guardians from the latter part of the 10th century onwards.
An interesting counterpart of the Sesai Viiyu is a loose image
characterised by the noose as Varuna, but combined with a gallop-
ing antelope, from Uttar Pradesh.333 The presence of this second
example of hybrid representation leads one to surmise that during a
particular phase in North-Central India, the vehicles of Varuna and
Vayu may have been interchanged.
In Central India, a possible teriizilzus a d quellz for the introduc-
tion of four-armed dikpdlns is provided by the Laksmana temple
at Khajuraho, which combines two-armed334 and four-armed
guardians. The date of this temple is c. 950 A.D.
2.3.2.3 Western India
Shortly after 750 AD., the dikpdla theme seems to have reached a
state of consolidation in the region. Only one temple of this period
at Osiiifi (Stirya temple no. 2) has an incomplete group of dikpdlas.
All other later temples with elaborate sculptural decoration contain
the directional guardians. There was also very little variation in the
position of the diklxilas. At Osiiifi, the most prolific site, two tem-
ples (Pipaliidevi; Visnu temple no. 2) show an unusual placement
of these deities, but on the whole, their order remained stable in the
following period.
The available evidence at Osiiifi335 is indeed invaluable for
dikpdla studies. Here, a local gradual development can be traced
with all its interdependences, innovations and probable relapses.
The only regrettable fact is the complete lack of an absolute date for
any of the temples.
328 TRIVEDI 1990: 168-175; DEVA1991: 56-60. The former author dates the
temple to the middle of the 10th century, at the very end of the Pmtiliira period in
Central India.
329 The niche figures are llol in a good State of preservation. A significantly dif-
ferent representation occurs in Vlyu's case. The deity holds an inflated cloth ( " ~ J ~ u -
~pfiri/a-vnstrcr)with both hands in the frieze, but in tlie niche has a standard in his left
hand. Thc transition from the clot11 to the standad also takes place in Western India,
in the 9th-10th centuries. For the niche figure of Kubera see TRIVED~ 1990: 111. 191.
330 It is presently kept in the Los Angeles County Museum, Museum Associates
Purchase M.74.5.6. The measurements are 73x99 cm. In the top register, the nine
planetary deitics are depicted.
331 Tliis has been overlooked by TRIVEDI (1990: 163, pl. 181).
332 In fact, tlie represe~itatiollof dikpdlas at Sesai is conlpletely stereotyped alld
uninspired.
333 K. DEVA/S.D.TRIVEDI1996: Cat. 100, fig. 84 (10th century; Allahabad
Museum, acc. no. 1385).
334 The siridlidr'a type Laksmana temple has sets on its sanctum exterior and on
its four subsidiary temples, but not on its outer walls. L.K. TRIPATHI (1965-66: 104,
11.2) observes: "Images of Indra and TSjna, both on the SE and NW shrines, alld those
of Yama, Nirgti and Agni on the NE shrine are two-anned."
335 The colnmon designations for the temples will be abbreviated as follows:
Harihara temple=HH; Mah3vira temple=MT; Pipaljdevi temple=PD; Surya tern-
plc=ST; Surya-Vis?u telnple=SV; Siva telnple=$T; Visnu temple=VT. Recellt pub-
licatio~is011the temples are HANDA1984; MEISTER 1991: 140-146, 174-209,263-265;
TRIPATHI 1971-84; and WESSELS-MEVISSEN 1994 (flikl~dlns).BIIANDARKAR 1912 gave
the first detailed account of these temples; for a concordance of these and later labels
see WESSELS-MEVISSEN 1994: table 1. A few telnples at OsiBA are without clikl>dlas:the
Satya-NirByana temple, the now "missing" temple and ST-2.
Table XI11 - Scuptural typology of dikpiilns at OsiZiB

HH-2,3 - Harihara nos. 2 , 3 ;

OSIAR M - MahBvira; PD - Pipalidevi;
(C. 75 1-900 A.D.) ST-2 - SOrya no. 2; ST-l - ~ i v no.
a 2;
SV - Silrya-Vis~u;VT-I, 2 - Visnu nos. 1 , 2
DikpBla Attributes Vehicle l'articular Characteristics
Indra vczjrn (bolt) r.""; elephant cylindrical crown;
PD: vajrn 1."" M: diadem
Agni akqamdlZ (rosary) I., goat ,jnffirrzuku?n(hair crown),
lcania!~~alc~
(watcrpot) or flask 1. flames behind head and shoulders,
vastropavitn (broad sacred cord);
M: diadem;
M, ST-l: no ctpavita;
VT-1: simple ul~avita
Yama (staff)r.**; buffalo crown; VT- I, PD (?):
PD: darzdc~l."" Erdhvakes'a (hair standing on end)
Nairrta klza(1gn (sword)r.**; flower? 1. anthropomorph ilr~llivakein;ST-2: crown
Varu~a pZSn (noose) r.** vznlcnm (aquatic animal); crown; PD: ja!dri~ukufa;
HH-2, SV: ST-2: awkward position of
no vehicle; VT-1: the left arm, which is
pot instead of a vehicle wrapped in a sash;
VT-2: platfor~nabove vehicle -
Viiyu billowing cloth in both hands; deer crown
VT-2: additional two banners in front hands
Kubera cup r.", purse 1.; anthropomorph; curly hairstyle, often with diadem;
PD: lotus 1. ST-2, ST- 1, VT- I : HH-2, ST-2: only diadem;
no vehicle; VT- 1: ilr~lhvalceSn;
VT-2: m e ~ n(ram) HH-2, HH-3, PD, SV:
platform above vehicle
ISZna (clockwise from below r.): bull; jn{Ei~?~~kug~
padmn (lotus)", triiilln (trident), M: anthropo~norphic
snrpn (serpent)"" ; vehicle
PD: only two arms (triSE1a r.*:k)

V o s t l indistinct.
~
The other hand rests on the hip or thigh, generally holding a loop of the sash. Nairrta in two cases shows a particular gesture (kntnkdir~lldrZ?)in front
of the chest (HH-2, VT-2).

a the Mahivira temple could not be included. On VT-2, Yama, Nairrta, Varuna, V q u , Kubera, alld igalla are four-anned. All the dikl?filfis
vnyu and ~ u b e r of
of the telnple are in a bad state of preservation, so that their attributes and specific features co~lldnot be evaluated except in a few cases. On PD, the head-
dresses of the dikpdlns are difficult to classify.

It is also extremely difficult to establish a relative chronology,
since most of the temples seem to be chronologically very close to
each other.336 Thus, the temples HH-2, ST-2, MT337, VT- 1, and ST-
1 have been assigned to approximately 750-775A.D., the two latter
ones with a tendency towards the end of this time period (ST- I prob-
ably even later). The last quarter of the 8th century is the period sug-
gested for the HH-3 and the SV. The Pipakdevi temple has been
assumed to date from the first quarter of the 9th century, and the VT-
2 in the third quarter of the 9th century. This chronology postulated
by MEISTER(1991) is not be contradicted by the results of an analy-
sis of dikpdla images at the site. Among the dikpdlns, clusters of
close similarity indicate that the same concept for a particular deity
has been applied to a number of - most probably coeval - temples.
One of the earlier temples, ST-2, still lacks a complete group of
directional guardians. It shows Nairrta (fig. 196) and Varuna (fig,
197) in relatively wide niches in their usual positions (south-)west
and (north-)west, respectively, on either side of the entrance. A
seated Kubera on the south(-west) (fig. 198) cannot be regarded
a di$iiln. The isolated depiction of Nairrta and Varuna Inay hme
been due to a change in the concept of the temple which was prob-
ably originally planned without dilpdlas. This temple also seems
have been the last one without prntirntlza niches.
33" ~ ~ e l a tchronology
i~e in the form of a proposed secltlelice of telnples has been
postulated by TRIPATRI
(1971-84: 299), out of which only temples with dikl~dasha",e
been chosen here: HH-I, HH-2, HH-3, SV, VT-1, ST-I, ST.~,MT, VT-2, PD. ST-l Is
not included in thc list. TRIPATIII'Ssequcnce is not colivil~cingill some respects,
ularly in the chronological placement O ~ S T -andI ST-2 after HH-1-3.
"' A cllronological hint is given in an inscriptioll of 956157 *.D.on the Mahivira
temple (MEISTER 1991: 184).In the text, the ruler VatsarAja (mentioned in the contem-
porary rccords of A.O.778 and A.O.783) is rcferred to. However, it is not specinca1lY
expressed that the MT was foundcd by Vatsarija. The dates known to us have, there-
fore, generally bcen rcgarded as a (enuirl~lsarlte que,n. The MT is an exception,
only here, the dikpcilas are situated on thc outer wall of thc sci~fdlriira
complex. Asthetr
placelnent does not include the mn~;<ln,~n, VByu and Kubcra arc, strangely, accojllmO-
datcd inside the gudl~an~a!~~lal~~.
 The dikl~dlasNairyta and Varuna clearly postdate those of the transition from HH-l to HH-2, the affinity of the dikl>iilaimages
HH-1, as the following characteristics indicate: on the different temples was in some cases very close. Some of the
guardians must have been regarded as models, but even in such
- They are shown standing and not seated. This alone would, cases, the whole set was not copied. The positioning of the deities
of course, not suffice, as Candra (fig. 185) and VHyu (fig. in the majority of the sets and the composition of all the sets were
187) on the relatively early ST-1 are also standing in corre- very similar, however with some minor variations in the rendering
sponding niches on the facade of the temple. of the deities.
- In their proportions (as well as the relationship in size Some of the images with a particularly close affinity are:
between the guardians and their vehicles), their dress and
their crowns, they resemble the rlikl~dlnsof HH-2 (e.g. fig. - Indra: MT and VT-1; HH-3 and SV;
207), which is a complete set. - Agni: HH-2 (fig. 206) and SV;
- The most striking argument, however, is that the unusual, dis- - Yama: HH-2 (fig. 207), HH-3 (BHA'ITACHARYYA 1991:
tinct "Central Indian" vehicles, a jackal or other quadruped pl. 49) and SV;343
for Nairyta, and a goose for Varuna, which occur on ST- l alld - Nairyta: MT (MEISTER1991: pl. 416) and ST-l (fig. 216);
HH-1, have alr~adybeen replaced by the more common - Varuna: HH-2 (fig. 209) and SV (fig. 221).
vehicles, a human being and makara, respectively, as known These exceptionally close affinities in the representation of certain
from the I<,?lik,?m,?tatemple at C i t t a ~ d g a d h . ~ ~ ~ deities suggest that the respective temples are largely coeval.
The replacement of lllost of the "Central Indian" vehicles after the An analysis of the particular standing postures reveals that a
formative phase at Osigfi is an indubitable fact. Only the anthropo- lllixture of "Central Indian" and "Western Indian" postures seems
lnorphic vehicle of Kubera, which also appears to have Come froln to have occurred soon after the supposed relapse of HH-2 to the
India (fig. 92), continued for some time. A special forlnula KHlik%m,?tHmodel. The dilcpdla set of HH-2 (figs. 205-212) - wit11
Seems to have been invented to distinguish it from Nair~ta'svehicle, the exceptions of Nairyta (fig. 208), V a r u ~ p(fig. 209)344, and
i.e. a platforln was inserted betweell the god and the carrier under- Kubera (fig. 211) -reproduces the scheme with the free leg point-
neath (figs. 200,211).339 For some reason, this formula fell into dis- ing to the nearest corner. In contrast to the K%lik%n~$t$ model, at
and Kubera remained without a vehicle on a few teln~lesat OsiHfi, variations in the leg positions were relatively frequent. It is,
Osigfi (figs. 201, 219). Later, 011 VT-2 (fig. 202) and other temples however, quite obvious that the temples with a more or less consis-
Outside Osisii, the anthropomorphic lzara was replaced by the rain tent depiction of the free leg pointing to the nearest corner are con-
('lzesfl). It is noteworthy that the gnr.blzab1zitti set of the K,?lik~ln~tfi centrated in Western India (at OsiHii: HH-2, HH-3, SV, PD). A typ-
which could be regarded as model for the "old order"340in ical "Central Indian" atiblznrign posture, a strong flexion of the body
India, does not include Kubera but Soma instead. Thus, an with parallel legs and feet, can be found, e.g. in the figures of Agni
Original model of a djklJcla Kubera may not have existed in this (fig. 206) and Nairyta (fig. 208) on ST- 1.
region. It seems likely, however, that Kubera had received special
in the preceding period. A cult image of Kubera is known
Osi,?fi (fig. 199), which was probably once worshipped alollg 338 See fig. 107 for Varuna. The co~npositionof the panel is strikingly similar.
Howcver, a few differences also occur, the major one bcing the position of V a r ~ n a ' ~
with two other deities, Gal@a and Laksmi. The Cup attribute in propcr left hand and the absence of an attribute in this hand at OsiiA. Nairrta at Cit-
Kubera's right hand is also a regular feature of the dilc~dlaKubera taudgadli (fig. 118)also has Inally similar characteristics.
o ~ i 5 f(see
i table XIII). In Celltral India, this attribute occurs less 339 This device is rcstrictecl to the temples (HH-2, HH-3, SV), where a pedestal
frequently. has been employed to rcduce the height of the niches. Thus, the size of the niche could
Apart from Kubera's vellicle, the images of the directional be adjustcd to the proportions of the image. Significantly, on PD. both the clikl,n/ns ill
pratirothcz niches are ~liountedon pedestals (figs. 203,204). This creates a visual hier-
guardians do llot vary to a great extent, but differ mainly in minor archy among them and relates them to the vertical axis, which is already e~liphasized
details. AS far as the spatial arrangement is concerned, two exceP- by the pilaster-like form of the prarirnthas. Actually, their irrcgulnr scquencc (from
tiOns are noteworthy: the Pipal,?devi temple, where four d i l c ~ ~ ~to~north: ~ Nairrta, Varuna, Kubera, VHyu) may have been determined by the fact
south to north: Nail~ta,V a r u ~ aKubera,
, VHyu; figs 2037 204) [hat for Varuna and Kubcm, a sock with integrated vehicle is niost appropriate. This
have been depicted on the western side; and the Vispu temple 2, "tier system" was used in all allnost extreme form on the devnkrrlikCs of HH-1 and HH-
2 (MEETER1991: pls. 368,398).
where the order starts with Indra on the (norlh-)east (see cover) 340 Sllould it turn out that the dating of the Kalikanlatl and KumbhaSyi~natcln-
instead of the (south-)east. While the first case is unique and ~ l e at
s Cittaudgadh is not as early as what has rccently been assigned (see above), tIley
Only be explained by the lleed to accom~nodatethose additional could very well belong to the first phase of architectural activity as OsiM. The direc-
guardians, for whom no place was available on the front (northern) tional guardians of the Cittaudgadh tcniplcs would then have to be regarded as expres-
sions of an experi~nentalstage in the develop~nentof a local idiom of dikl,Cln repre-
Part of the temple, the second case has a few parallels outside Osirlii sentation. The close rese~nbla~ice betwecn the image of Vat.una of the KjlikPmBtj.
(e.g. Khidarpurg, M E ~ ~ 1991: T ~ Rpis. 548, 549). Here, a different telnple (fig. 107) and that of the Hariham temple no. 3 at OsiBA (fig. 223) could indi-
arangelnent of di/cl>Elnsstarting on the east wall with the northern cate a si~nilardate of the two.
341 Tllis fact is also reflected by MEISTER'S (1991: 182) statement: "Hariliara
niche, might have been regarded as conceptually more consistent. no. 2 and the Mahivira temple, with which it is closely related, represent new guild
With regard to the uniformity in the rendering conventions at Osiiii, yet considerable respect has been shown for previous production
'stfldikp~lns at Osigfi, it can be that some clusters of at this site."
Felationship exist. The first conspicllous fact is, of course, the affin- 342 cf.also HANDA1984: pl. 106 (Agni) and MEETER1991: pl. 588, 589 (soutll

ltY between HH-1 and HH-2, where the iconographic ~ ' ~ 6 ~ " " " ~ and wall, respectively). The hierarchy in size between the small kor!in andpmti-
and the much larger bliadrn image is remarkable.
had been identically reproduced, jar. copied on the latter (ill. 22). On ,u,hn343i~nages The raised head of the buffalo is the particular lertiirriz corii,,nrnriorIes. It is
HH-2, however, the rendering of the figures was quite diffemt.341 illtcresting to note that the buffalo of HH-2 is the 011ly one rendered realistically. ~1~~
The change of vehicles in solne cases has already been mentioned' heads of both the other specimens are stylized, and the position of their tails is also iden-
Another important difference is the transition from the seated Or tical.
344 Nairrta and Varu~pare located in very narrow niches on the side of the
' ~ position to the standing pOStUre9 which in the
riding (in V , ? Y ~case)
entrance (west), thus leaving very little space on the sides.
phase was only abandoned on VT-2 (fig. 202).342After
 It is rather peculiar that both Kubera (fig. 201) and VByu (fig. images of Vjyu and Agni (fig. 225) closely resemble each other'
218) are sometimes shown in the same stiff, clumsy posture, inspite Since the latter is still in situ, there is no doubt that both guardians
of their varied characters.345 The samapdda (feet together) or a relat- originally belonged to the garblzablzitti.
ed, clumsy posture for VByu is also found at Abgneri (B. SAHAI: Judging from the few extant sculptures, it appears that both sets
fig. 8), DBng (fig. 1491, and PBhBrpur (fig. 236). Kubera generally of dik1,dlas are characterized by rather stiff postures, n feature that
retained his obese features and stocky proportions. This is empha- does not recur in any other set in this region or in Central India. The
sized by a posture in which the legs are held slightly apart (e.g. immovable, monumental appearance of the deities, in fact,
fig. 233). the images at AIampur. At the same time, it is clear that the deep1Y
Another feature shows that at OsiBfi, individual solutions were undercut figures at Abgneri with their excluisite proportion in^ "ld
employed for particular conditions. This is the unusual and varied modelling are much 1noi-e evolved than their counterparts at
rendering of dikpdlas in extremely narrow niches on the facade, pur. The presence of attendant figures is sirnilar to the ~ f d i ~ ~
which occur on a number of temples (HH-2, SV, VT-l, VT-2). In temple at C i t t a ~ d g ~ dbut h , the lavish decoratioll ("lz~rror
some of these cases, particularly Nairrta and Varuna in HH-2 characterizing this temple is not reflected at Absneri. An interesting
(figs. 208, 2091, the dikealas in the narrow niches resemble detail can be observed 011 the image of Agni (fig. 2251, namely, Ihe
dviirapdlas, i.e. door since their bodies are turned separate socle for tile goat vehicle behind the god.
towards the door, the free leg pointing in the same direction. Here, Besides the K u i n b l i a ~ ~ ~atmCittaudgadh
a and the ~ a r s ~ ~
on HH-2, Varuna is shown without any vehicle and N a i ~ t a ' ssword at Abgneri, a third, well-preserved temple with two sets of clirec-
is - uniquely -resting on the ground. Another interesting example tional guardians can be found at Khed in district Barmere The
is provided by the SV, where both Naiqta and Varuna (fig. 221) s d n d l ~ a r(with ~ ambulatory) concept of the ~ a ~ c h o dtemple' ji
stand in identical postures, with the left leg bent and crossed behind which has beell assigned to the middle of the 9th century, fofollows
the right one. Once again, Varuna does not have a vehicle. On the tradition of a few other temples in Western India.351At Khed*it
VT-1, Varuna (fig. 222) is - uniquely - shown standing On a spilling can be observed that the rendering of the dilyalns is quite silnilar l"
pot instead of a vehicle. Thus, it appears that in the cases, where the different sets. Obviously there was no intelltion of creating
dikplilas had to be accommodated in narrow niches, the sculptors particular distinction betweell the two sets. The deities of
created new, imaginative concepts in the representation of the garbhqrlIn (sanctum) set (figs. 226-2331 stand in lifelike, sliglltly
directional g~lardians. stiff postures in relatively narrow niches. Their bodies are well-pro-
In summing up the evidence from OsiBfi, it can be stated that the portioned, with broad faces, and they wear rather heavy ornaments'
group of temples probably dating to a phase soon after 750 A.D. The vehicles are colnparatively small, The figures of Agni and
(namely HH-2, HH-3, SV), contain very similar, standing rlikpalas. Kubera are conspicuously obese. hi^ is interesting with regard
The deities are generally shown standing in clear-cut postures (par- the symmetrical placement of both deities on the south(-east) and
ticularl~on HH-2, figs- 205-212) with the free leg turned towards north(-east). Other notewortlly characteristics the short flames
the nearest corner. A second group of temples comprises MT, rendered in high relief behind ~~~i (fig. 2271, a pointed attribute
ST-I, and VT-I. On these temples (e.g. ST-I, figs. 213-220), the (flower?) in Nwrlta's left hand (fig. 229). the flag attribute of 'liYu
poses of some of the figures have become rather exaggerated (fig. (fig. 231; note also the flower in thenright hand), which seems
220) and ungainly (figs. 201,218). A tendency towards an ill-con- have become a comlnon feature in the 9th century, tile long, thin
trived and awkward twisting of the body can be noticed on PD ( B p . purse of Kubera, held in his right hand (fib 232; a in the left),
203,204, cf. VIENNOT 1976: ph. 236). A later Concept appears on and the newly introduced pot in the lower left hand of &*la (fig
VT-2, where the dik~dlasare shown seated on relatively large 233). ~t is a peculiar fact that the vehicles of the dikl7Clas at Khe.d
vehicles (MEISTER1991: PIS.588,589). Here, the niches, which had belong to the "mainstream" tradition at osijfi, even including the
become narrower in the course of this period, are again wider. anthropomorphic nara for Kubera, while the broad faces and the
Apart from Osiiiii, a few exquisite examples of dikpala sculp-
tures are known from the HarsatmBtB temple at Ab~neri,viz. Agni
(fig. 224), Indra (fig. 225), and V ~ Y U ~The ~ ' vast
. ruins of a once 345 In thc case of Kubera, this feature is understandable, sincc he is characLeristi-

imposing temple, probably of V a i ~ n a v or a ~~~i ~i kaffiliation,

t~ con- tally obese and hence clumsy. I" the case of the wind god vayu, the only explm"3ion
is the fact that the snmn,,iidrr posture matches the symmetrically draped cloth, for wJ1ich
tain only a few surviving specimens of dik~dlas,most probably earlier models might have been knowll ~ ~guardiinls t h exhibit the same ~ l u l POs- ~ ~ ~ y
belonging to two different sets of directional guardians. The first ture on VT-1 (MEISTER1991: p,. 342).
quarter of the 9th century has been proposed as an estimated dale349 34"~1 WESSELS-MEVISSEN (1994: 603-604), a ~ C W]nore cases of d i k with ~ a~ ~ ~ ~ ~
for the s511clhEra temple (with ambulatory), which was probably Second" f u n c t h ~as dviirn1~irlns(door guardians) have bcen described
347 For VLYUSec B. SAMAI 1975: fig. 8. The prcseat whcrcabouts of image
once a paAcdyatana complex (consisting of five shrines; F. REITZ
could not be ascertained.
1994). Its sculptures are not stylistically related to those at Osinfi nor 348 C.P. ATHERTON 1995: 207 Sqq, A doubtfulpoilll in her arg~lmentis the libera1
to any of the temple sites discussed above, but represent a distinct idendfication of a colnlnon ~ i s (VBsudeva) ~ u sculplure as ~ ~ i r u d d h(ibid.:
a p'. 25),
idiom. without any real iconographical justification (ibid.: 209: "His iconography is quiteclose
It is very likely that two diki~dlasets embellished the garbha- to Viisudeva's").
bhitti (sanctum exterior) and the outer wall of the prada1c;i~dpatha
., MEISER 1991: 237,
a" This has been expressed by x ~ (1993: ~ T 208-209):
~ "Jedoch existier2
(ambulatory), respectively, here, as is the case at Partadaka1 Bildnische mit Indra [...l,der man die gleiclle Funktion an der LJrngal%smauer
(Piipanatha temple) and Cittaudgadh (KumbhaSygma temple). ~h~ besagten Tempeln in Chittor [Cittaudga$li] und Osian zuweisen kanll".
type of the niche, which is preserved in all three cases, provides an (1995: 203-204) also assumes such an original position of directional ~ u but ~ ~ ~
fails to note the different niche frame for Indra: "With othcr mbulalory structures' ['.'l
distinguishing feature. R E 1(1993)
~ has proposed a plw-
these corllcr piers are custolnari~y provided niches dirrciionaI gu"dFn
sible classification and functional analysis of the niche types. The (dikpiilri) figures 011their central orhetS. UnfortungeIy, no guardian images surv've
sculpture of Indra (fig. 224), which is now in a secondary place 'hat would serve as candidates for these corncr positions". M ~ l s T e n(1991: 236) 'pe"'
(REITZ 1993: Abb. 2), most probably belonged to the gcdha- the "image of Indra, that now replaces the pilaster south of the antarala"* but 'On-
ma?&P tinues to assign it tentatively to another, no longcr extant shrine.
The niches with the two other surviving
,,, Manin 1P91:25e258.
 large bodies in relaxed, gracefully curved, awe-inspiring stances is the increase in the number of guardians to ten. This inclusion of
more vividly recall the Cittaudgadh images.3s2 BrahmI and Ananta, guardians of the zenith and the nadir, respec-
These features appear to be less marked among the images of tively, however, seems to have been restricted to Western India. Due
the outer set on the iniilnprdsdrln (MEISTER1991: pls. 572, 573). to a lack of documentation, only two examples can be mentioned
They more closely resemble the fine dikpdla set of the KImeSvara here: GhInerIv359 and KBleS~ari-ni-nLl~~~, both belonging stylisti-
temple at AuwI in the Pali district of Rajasthan (ibid.: PIS. 477,478). cally to the 10th century.
Here, the bodies are also slightly less imposing, and the gracefully The stagnation in the cEikpLZla representation in the late 9th and
flowing contours have their own charm. The deities, as at AbIneri, early 10th centuries seems to have been followed by an innovative
are Surrounded by worshippers and attending figures. They are all development of the late 10th century. However, antecedents for the
two-armed like those of the PipalIdevi temple at OsiHii. Some new concepts may be traced to the earlier period. For the present it
other features, such as the slightly limp posture (ibid.: 21 l), are should suffice to recall that four-armed dikpdlns - other than fSIna
reminiscent of this presumably contemporaneous t€mple (c. A.D. - had been introduced at Cittaudgadh (KIlikBmItI: Agni; Kumb-
800-825). halyiima: garbhagrlza set), Osisii (Visnu temple no. 2: all except
The development of the dikpdln representation seems to have Indra and Agni), and at BatesarI in Central India.
reached a point of stagnation during the 9th century in Western
India. The dikpdla sets on most of the temples353 of this phase in 2.3.2.4 Eastern India
Western India are generally two-armed (in some cases including
Iezna as weI1354) and standing in relaxed postures, often with the free Evidence of the canonical dikpala images appears relatively late in
leg turned towards the nearest corner. The end of the 9th century Eastern India (Bihar, Bengal, Orissa and the adjacent areas of north-
witnessed the creation of some better shaped images of perfect Pro- ern Andhra Pradesh). Some early traces, MundeSvari, NIlandI and
portions, with delicate surface treatment and standing in a complete Dev BarunLrak in Bihar and Bhubaneswar (ParaSurBmeSvara tem-
eClui~oi~e.355 Here, both tendencies, the bold, statuesque appearance ple) in Orissa, have already been discussed. None of these early
and the slightly limp, flowing form have been perfectly blended. depictions seem to have had any direct descendants.361
The result is a very subtle rendering of a divine personage. In a vari-
a t form of the depiction, the deities appear seated On relatively 352 These features, however, are not rare among the dikpriln images of the 9th

large vehicles as at B h u ~ d I n S(ibid: pls. 557, 558) and at Osisii, century in Western India (cf. the set at Goth-Miiliglod, c. 825-850 AD., MEISTER1991:
Vi!nu temple no. 2 (fig. 202). M O S ~of the members of the latter set pls. 564,565).353 Exa~nplesof Western Indian 9th century temples exist at Kliidarpurj (MEIS-
four-armed. TBR 1991: PIS. 548, 549). Bhavjnipur (ibid.: PI. 554). Bhundina (ibid.: pls. 557, 558).
On Some temples, irregularities in the placemellt of the direc- Goth-Midglod (ibid.: pls. 564, 565), Cittaudgadh (sati ground, no. 1: ibid.: pl. 684;
tional guardians can be observed. In some cases, however, it cannot Ksemalikari: pl. 686), Mlndalgadli (ibid.: pl. 691). (triple shrine: ibid.: pls. 695-697;
wcst group, no. 1: ibid.: pl. 730), and Candrabhjgj (ibid.: pl. 727).
be ru!ed out that these are not caused by subsequent disturbance. 354 These cases are the Pipalldevi temple at Osilfi, the K31neSvaratemple at Auwi
The Siva temple at Khidarpurj follows the same pattern as the Vis?u (see above) and the Nakti~niitiitemple at Bliavinipur (MEISTER1991: 243).
t e l n ~ l eno. 2 at OsiBfi with the set starting from the (north-)east. In 355 E.g.aloose sculpture of Viyu, now in theBrooklyn Museum, acc. no. 86.183.3,

these cases, the sequence of deities has not been changed - only measurements: 111.7x43.5 cm, published in D. MASON1993: cat. no. 8.
356 Interestingly, this is not a unique case. It occurs in Central India at Djlig and
their placement varies. In other cases, however, the canonical
Batesari as well (already citcd in WESSELS-MBVISSEN 1994: 604). Nairga and Vsyu are
Sequence of dikpdlns has been disturbed. Thus, Varuna 11% taken both assembled on the western wall at Baresari (ibid.), Mii~dalgadh(sce below) and
V ~ ~ U place ' S and vice versa356 on the NaktirnBtI temple at also on the Pipalidevi temple at OsiBfi, where Varuna and Kubera appear i n p r ~ t i r n t h ~
Bhavjnipur (ibid.: 242). On the same temple, Indra and fSjna are positio~is(figs. 203,204). A hybrid form of V a r u ~ and ~ a VPyu is attested at Sesai, where
found on the eastern side of the edifice, in niches carved 011 pilasters the deity with a floati~iggarment on the north(-west) has a rltnknm vehicle (fig. 181).
357 On a rock-cut te~iiplewith a similar ground plan, the Caturbhuja at Gwllior,
SuPPorlillgthe roof of the 1fl~lklzai7znn&7pn (porticus). This appears no provision was made for the directional guardians on the east side. Although they
be a logical arrangement, sillce the pilasters have left no space for could havc been accomlnodated on the nzukl~o~i~arj~npa pillars, both ISiilia and Indra
on the ~fzag&vnra(exterior wall) itself.357 were omitted (see below).
Two disorderly sequences are found on the telnple no. 1 the 358 An important tertliinwspost qrreln for the general introduction of four-armed
c/ikpHlas in Western and North-Central India is providcd by the Ambikimlti temple at
ground at Cittaudgadh and on the JaleSvara temple at Mj?dal- Jagat, District Udaipur, dated A.D. 961. Two of its three sets consist of two-armed
gadha On the former, the deities are j&ina, Yarna, Nairrta, Varuna? guardians (prcisridn, prn!lri/a~llfl!l(~flpa), but one set already shows four- armcd deities
V j ~ uand , Agni, enumerated clockwise from the north(-west) the ~)raveJnnta!ida1>a).The most important study of four-armed dikpdlas has been lllade
South(-wesl), the western wall remainillg devoid of directional by L.K. TRIPATHI (1965-66).
359 The Mahjvira temple at Gliii~~eriiv in District Pali, Rajasthan, dated to about
Guardians. In order to explaill this aberration, MelsTER (19g1: 301)
!lie lniddle of tlic 10111century, is a north-facing sdndlidm complex. DHAKY (1968: 328-
'las Proposed the theory that the temple was originally planned with 330) reports t11al the additional dikprilns Brahml and Ananta are represented on the
""ntrallce 011 the east instead of the west. That would mean that nortlier~iside, "on wall pilasters in the Mukhamandapa that stand in the immediate
lhese Parts of the building to be completed before assemblillg vicinity of the karyas of the Gcdhamandapa".
360 H.R. GAUDANI/M.A. DI-IAKY 1969. The site is located in Sabarkantha district/
theln iiz situ. However, the positiolls of f S ~ and a Agni cannot be
Gujarat. Some of the images of the original temple have been amxed to the present day
in this way. structure. An image which can be plausibly identified as Ananta is illustrated 011fig. 5.
At Mii?dalgadh, it is quite probable that the temple had The deity has been provided with a cobra-hood and an u~iidentifiablcvehicle. An illlage
rebuilt around the year 1560 A.D. (ibid.: 304-305) The sequence, of Braliml has not been preservcd.
361 ~ l thel sites mentioned liave very distinct modes of representation: at
Starting frolll the (south-)east, is: Kubera, dik1,Gla with ram'? (ibid:
306, it is stated that Aglli has been shifted to the exterior the Mu?deSvari, the Manu Lokapala group has been depicted on cornerstones, and a niche
set including Dandaplni occurs also here; at Dev Baru~>irak,a free-standing pillar
porticus, kapili), Yams, Nairyta, Vgyu, a dikpdln without a ve11icle9 sllows the four loknpcilns on its four sides; and on the ParaSurilneSvara temple, a wall
Indra, and fi&. frieze exhibits a - probably incomplete - dikpola set plus DandapBni. The baselnellt
After the period under survey, two important developrnents frieze of the Stone temple no. 2 at Nllandi, however, Inay be compared in sollie respects
Occurred. The fint one is tile gelleral introduction of four arms for will1 the evidcnce from Pahirpur (to be discussed), since both liave an incomplete set.
AII the deities at the earlier sites were depicted seated, with or without a vehicle,
"le dil@~n images in the 10th century.35"he second development
 While all eight dikpdlas were depicted together on a frieze in the
9th century at Gaur (?; fig. 240), the group of directional guardians
was incompletely represented on separate panels on the late 8th
century Buddhist structure at Pnhirpur, Bangladesh (figs. 234-237).
In both cases, the guardians are shown standing. In Orissa and the
adjoining areas of Andhra Pradesh, an increasing number of - gen-
erally incomplete - dikpnla representations are known from about
750 A.D. onwards. It is difficult to precisely ascertain when the corn-
plete sets emerged on the walls of the temples, but most probably
not before the 10th century. There are some signs of a Central Indian
influence in Orissa, but this was not very dominant. Apart from the
standing form of dikplilas, a riding variant appeared already at the
beginning of the 9th century, at Biirambs and Mukhalingam, which
may represent a South Indian (Deccani) influence.
One of the important characteristics of dikpalas in Eastern India
is the fact that they were generally two-armed (except for TSiina).362
Some additional attributes, i.e., the noose for Yama and the elephant
goad for V g y ~remain, ~ ~ restricted
~ to the region of Orissa. Gener-
ally, however, all these results of the survey should be treated with
caution, since there are only relatively few records distributed over
a vast part of the country. This adherence to the more traditional
two-armed form, especially in Orissa, is another indication that the
eastern part did not follow the developmel~taltrends in Western and
Central India.
The so-called Somapura Vih2ra at Piihnrpur (District Rajshahi,
~angladesh;ill. 29) has provided an extremely important evidence
regarding the representation of directional guardians in North-east-
Candrn 60
l-
29 Indra
111.29: Piihjrpur, Somapura VihLa, plan sketch (panel numbers
after DIKSHIT)
ern India in the late 8th century.364In fact, some of their features
point to an earlier source.365The images of Indra (no. 29; fig. 234),
, 39; fig. 236) belong to a series
Agni (no. 34; fig. 235), and V l y ~(no.
of 63 surviving stone panels on the basement of the cruciform ten-
fral Structure at the h d d h i s t site. Images of iiiina? (no. 40) and
SomdCandra (no. 60; fig. 237) ]nay have been associated with this
group, but that cannot be conclusively proven. The rlik1)iila fullction
of these deities is rather doubtf1ll, since the relief panels are not
placed in the "canonical" directions. Moreover, the set seems to
have remained incomplete. The image of Candra on the (north-)west
slightly differs from the others, since it is of d i f r e ~ nmateri-
t
al, black basalt.
There is a doubtful image of &na (DIKSMIT 1938: PI. X X X I ~ ) ~ ~ ~
next to the Vjyu image. He stal~dswitl~outhis vehicle. Among the
securely identifiable dikpGlns, only Indra is accompanied by his
vehicle (fig. 234). He has an ornalnented halo and holds a fruit in
each hand. His smiling face with the horizontal eye on the forehead,
his pearl necklace, cr&mbancl/2a ("belly ban$) and chain girdle
have all been delicately modelled. The large elephant behind him
leaves no space for any attendants. Agni (fig. 235) has boldly
incised flames behind his body, starting at hip level. He is pot-bel-
lied, wears a broad necklace and a vastropnvita (broad br3hma?ica1
cord) like Candra, and holds rosary (r.) and waterpot (I.). It is a gen-
eral feature of Eastern Indiall Agni ilnages that the left hand does
not extend below the elbow. Aglli is shown wit,lout a vehicle. He
was probably originally flanked by two attendants.
The image of Vsyu (fig. 236), which had been identified
as Yama,367 has clear-cut symmetrical features alld wears individ-
ually conceived ornaments. Like Agni and Soma, he stallds in
sanzapiidn (feet togetller). The flowing vastra is beaLltif~llydraped
over his head. Its lower ends almost touch the two attendants, male
and female, who likewise hold pieces ofCloth.368HiS hair is done in
a neat ilrdlzvabia (demonic hairstyle) coiffure. The "clot11 type"
Vjyu must have been popular in this region, since two fine terra-
cotta p~aquesof seated Wind Gods with a flowing scarf above the
head have been found. On one of the pieces, also from Psh5rpur9 a
tiny antelope is visible beside the god (fig. 239). The other exa1nple
from Mahrlsthiin, Bogra districtlBanglades11(T.N. R A M A C H A N D R ~ ~
1936-37: p]. XVIc), corresponds to the Piihsrpur plaque ill lnal1Y
details. It has a similar, square shape, a ~ l dthe arms of VByu are
not positioned in a symmetrical manner, the left hand beillg held
m c h lower than the right one. Only the antelope is missing. Keep-
ing the three described VByu irnages in mind, it is significant to note
that these are the only known representations of the Wind God hold-
ing a billowing cloth in Easter11India.
0111~a few exceptions have come up, such as the image of I~idraat Madall
Kjmdev/Assarn, assigned to the I lth cent~lry(M. JJUTTA 1990: p]. 9G), and the image
Agni at K.jmikhy3lAssam (BHATTACHARJEE 1978: 120) of about the salme dale.
Both deities are four-armed.
3" This has not been recognized by T.E. DONALDSON (1987: 1146) in his general
of Orissan dikpiila images, perhaps due to ,he often damaged state or the
sculptures.
3" DIKSIII'I. 1938 (011the history sce p. 5); DWA 1991: 403-404.
3f
l5 The fact that Indra holds only fruits but no thunderbolt is rctniniscent lhe
ilnpressive Indra image from Gay2 of c. 7111 century (SAIIA~ 1975: fig. 2; PAUL 1985:
133, n. 75) who holds a fruit in exactly the same manner in his extcnded right halld. 'l1
the latter case, the god is seatcd on an elephant tlironc in ~ / ~ k , ~ i ! ~ r i / ~ r n / c ~ 1(rig'1t
~1/>~1/7~(~~
leg pendent).
'" with the trident held on the proper left side and tile serpelit earring on the right3
ilnage resembles the iSana i~nagcsat Caurjsi and Gal)eSwarpor/Orissa ( D O N A L ~ ~ O ~
1985: rigs. 708; 726). Dilferences are the absence of the vehicle at PSliB~'pur,alld lbe
differing - positions
. of the right hand (gcsture of bcstownl of boo~isat P3h~rpUr3 rosary
held at medium lieiglit in Orissa).
367 G, BIIA~.ACIIARYA (1987: 64; also 1997: 780) on this andrefel.s
to M.T. DE MALLMANN (1963: 133, n. 5) as the first scholar to arrive at the
identification. Nowadays, with more studies on iconography, in this case Wticularly
TANARE (1090), such a mistakes arc preventcd.
Although thc upper portion rese~nblesa noose, the two lower ends show 'liat
they are different ,,errions of an
 In the case of Pahiirpur, the group of dikpdlas seems to have
remained incomplete, and some of the images are irregularly placed.
The positioni~lgof Indra and Agni was correct (ill. 29), however.
Moreover, on the south-east, two terracotta images of a seated Agni
can be seen above the stone sculptures.369 Among the dilc1>dlas,Agni
has been most frequently portrayed in the later art of Bihar and Ben-
gal (e.g. B.CK CHHABRA 1935: pl. VII, 5-7).
A complete stone panel, probably from GaurIBihar, showing a
dikpdla frieze (fig. 240) is a rare piece of evidence from the 9th cell-
t u r ~(G. BHATTACHARYA 1987: 3). They are all shown standing in a
slightly flexed posture, accompanied by Ganeia on their right, two-
and mostly holding only one attribute. Their tiny vehicles
"Pear in the register below them, which is quite ullusual (from 1.:
elephant, goat, buffalo, reclining man, ~izakam,gazelle, ram, -dam-
"ed-). The attributes are as follows: 1ndra- damagedlindistinct @.);
Agni - rosary (r.), sacrificial spoon (l.); Yama - skull-topped staff
(klzfltvdrign,r,); Nairyta - sword (r.); Varuna - noose (r.); Vayu -
standard (l.); Kubera - probably club (r.) and pursc (~~alc~ilalca, 1.);
ISBna- rosary (r.), trident (I.).
The Bihar panel cannot be Inore precisely dated. Some aspects,
as the individually rendered crowns and body proportiolls,
recall those of the dikpcila set at Khed, Rajasthan (figs. 226-2331>
of about the middle of the 9th century. In this set as well, Vgyu
holds the standard, which had not yet become a common feature in
Central India. Further, the elongated money bag, draped around
Kubera's shoulders like a scarf at Khed (fig. 232), could be related
to the elongated object in Kubera's right hand on the Bihar panel,
which is too small to be convillcingly identified as a club. The tri-
dent held in Ti~na'sproper left hand a l s points ~ to a Central Indian
influence.
Furtller east, the evidence becomes even more scarce. A 10th
example of V a r u ~ ais a stone sculpture at HajotAssam
( D 1990: ~ pl. ~46). The~ deity~ is seated in padnzasarza (lotus
posture) on a frontally facing makara, holding a serpent noose
(~~dgfll2dia) in the left hand and a damaged object (collch?) in the
right hand.
BenisBgar or BenusSgar, in Si~lghbhi~m districtmihar, and situ-
close to the border to Orissa, has yielded a good number of
One such is all interestitlg image of the staff-bearing
Da?dapBni wjth an obliquely held stick, of about the 7th
century A.D. ( A s ~ l n1980: ~ PI. 103). Three loose clikpda ilnages of
wet belongillg to the 9th or lot11 century have been described by
D o ~ ~ (1985: ~ ~ 246; s fig.
o ~598 [Agni]; fig. 599 [Kuberal). Their
appear extrelnely stout and flabby. In fact, Agni's vehicle on
his left also reflects this obesity. Agni has elongated earlobes and a
beard9 and holds a rosary (r.) and a waterpot (I.) at elbow
Both Agni and Kubera wear an ~llmvita.Kubera is uniquely repre-
sented with four inverted pots surrounding him, in the corners.
Streams of precious objects, most probably coins, discharge fro111
the two lower pots as well as froln the purse held in his left hand.
The presence of several vessels containing treasures (nidl1i17fitra)
"" the olnission of a vehicle already correspo~ldto the OrisSan tYPe
Of Kubera. Vgy~l]~o]ds his standard and is accompanied by an ante-
lope at Be11is~gar.371Whether the "standard-bearing type" of VSyu
had always been common ill Orissa and its northern borderlands -
" in the Deccan - cannot be incollclusively ascertained, due to the
of extant images. It is, however, a fact that no image of the
'clot11 type" has come to light in this region at all.
111Orissa and at the related sites of Mukhaliligam, N5raya1!apu-
ram and G a l I ~ v ~ linl i llortll-eastenl Andhra Pradesh, the instalment
Of guardians in the corner niches of the telnples can be
traced back to the 10th century, lnost probably startillg in the early
decades of that ~entury.~72 A few earlier depictions, which are not
yet "canonical", are listed below:
- Bhubaneswar, Markandeyeivara temple, c. 750 ~ . ~ . : 3 Agni 73
and Varuna flank the door of the sanctum; Agni (proper rigllt
jamb; DONALDSON 1985: fig. 125) holds a rosary (r.) and a
waterpot (? l.), while Varul>a(proper leftjamb; ibid.: fig. 126)
holds a noose (r.). Both deities wear the sacred thread
(upavita) and are provided with a halo. The upper part of
Agni's body is - as usual - encircled by flames. The gods,
who are both seated without a vehicle, probably did not serve
as directional guardians here.374
- Mukl~alirigam (District Srikakulam, A.P.), Madhukeivara
temple, c. first quarter of the 9th century:375 On the northern
wall of the jngarizolza~zais a richly adorned figure of Indra
(DoNALDSON 1985: fig. 333), standing in front of his ele-
phant. He wears a broad crown and holds the thu~lderbolt
(vajra) in his right hand, the left resting on the large knot of
his sash. A noteworthy feature is the olpavita made up 01three
rows of pearls. Indra obviously does not serve as a dikpnln in
this context as well.
- BIramba (District Cuttack), Simhanatha temple, probably C.
850-7h.o. :376 Agni can be found in the south(-eastern) cor-
ner niche of the jagnnzolza~la(fig. 241), and Yama is depict-
ed in an oma~nentalframe (gavdk;a, in Orissa: vcdra-11zasta-
ka) also 011 the south, but on the temple tower (ga!idi; fig.
242). Both deities are realistically portrayed as riding their
vehicles moving to the proper left. Agni holds a rosary (r.)
and a waterpot (l.), supported in his hand. Y a m holds a staff
(S.)and a noose (1.).377 The same attributes are carried by two
attendants on both sides of Yama. The vehicles are shown in
motion.
- Mukhalinga~n:loose, fragmented frieze of dikl~cilas,probably
roughly coeval with the riding clikpdlas at BBrambB, viz. 9th
century (GRAVELY/SIVARAMAMURTI 1939: pl. XVI, top).378
Four gods of an incompletely preserved set are extant, riding
to the proper right: Yama, Nairyta, Varu~!a, Viiyu. Their
attributes are respectively: lloose (r.), skull-topped staff
(kl7atvdrign, 1.); ? (r.), sword (l.); iloose (r.), fruit? (l.); stan-
dard (r.), elephant goad? (l.). The most significant detail is a
donkey as Nairrta's vehicle. A model for such a four-legged
369 Gerd Mevisse~i1997, personal communication.
370 This appears to be a typical Eastern Indian characteristic, where the vase is
supported by the hand held at elbow level. This distinctive feature of supporting [hc
water vessel can alrcady be observed in the early 7th century, viz. at MundeSvari/Bihar
(figs. 36, 38), and on the ParaSurZmeSvara temple, Bhubaneswar (e.g. DONALDSON
1985: fig. 79).
371 The image lias been published by K.C. PANIGRAHI 1956: fig. 111. A stout ante-
lope stands on the proper left of the god, who holds the broken portion of his attribute
on the samc side.
372 Tlie authors P. ACllARYA (1953) and H.C. DAS(1974), who have specifically
dealt with dikprilns in Orissa, did not yet have sufficie~itdata to be able to trace the ori-
gins. DONALDSON (1987: 1141-1 143) was the first one to describe tlie outlines of the
develop~ne~it of clikpri/ns in this region. The partly unsettled chronology of Oris-
San telnp]cs is clcarly reflectcd in [lie various works by the authors DONALDSON (1985-
87). MITRA(1991) and W.S M I T l l (1994). DONALDSON (1985: 274, 282) and SMITH
(1994: 135) differ in their dates for thc Viiriihi telnple at CaurZsi and tlie Paficapiil!dil-
va temple at GaneSwarpur that contain the earliest extant oriented dikpdlns, by 1110~~
than 50 years. As the latter author's view is based on an in-deplli study of a particular
telnple (MuktcSvara temple, Bliubaneswar) related to some of the temples under dis-
cussio~~, his assignment lias been accepted here (c. 970 A.D. for Ga~~eSwarpur, c. 985 for
Caurjsi). DONALDSON (iOi(f.) places CaurBsi in the first quarter of the lot11 century,
Ga?cSwarpur in the scc011d.HARLB(1986: 164) still adhered to tlie earlier opinions,
assigning thc Cauriisi temple to the 9th century.
 vehicle could only have come from Central India. The other
vehicles are the traditional ones.
- Mukhalingam, SomeSvara temple, c. 900 an isolated
image of Agni as laliitabinzba (crest figure) above an Ardha-
nlriivara panel, jarigkii (wall-frieze), (south-)east (MITRA
1991: pl. 945). Agni is bearded, stands in the rigid sanzapdda
posture, wears a simple necklace and a long sacred thread,
and holds arosary (r.) and a waterpot (1.) slightly above elbow
level. He is flanked by attendants but has no vehicle. The cor-
responding parts of the temple (laliifabirnbas of the niches)
do not contain any dikpiilas.
- Charda (District Bolangir), KapileSvara temple, c. early
10th century:380 a loose relief fragment of a four-armed Agni
image (DONALDSON 1985: fig. 496). This is a rare example of
a four-armed dikpala - other than fSiina - in Orissa. Another
unique feature is the extended right arm holding the water-
pot.381 This feature points to an influence from another
region, most likely Central India, where four-armed dikl3iila
sets were produced from the latter part of the 9th century
onwards (e.g. at Nacnii, see above). The garland-like loops
below the girdle also point to such an affiliation. A significant
aspect about this "import" is that it was not copied or repro-
duced in Orissa afterwards. The corner niches of the temple
are now empty, but it is very likely that the Agni image was
once placed in such a niche, probably forming part of a
dikpiila set.
- Niiriiyanapuram (District Vizianagaram382, A.P.), Nilak-
antheivara temple, around 900 A . D . :frieze
~ ~ ~ of dilc~iilason
the doorjambs of the sanctum door (MITRA1987: figs. 19-
21), starting from below, (proper) right. As was common with
the other groups of deities, the directional guardians also
occupy oblong panels on the outer doorframe, bordered by
decorative strips. As the relief panels, which are in a poor
state of preservation, show the earliest extant complete
dikpiila group, MITRA'S(1987: 202-203) description shall be
cited here at full length:
Starting from the bottom niche of the southern jamb are:
(i) two-armed Indra with a thunderbolt in his left hand, a
high mukuta, a halo behind the head and the dangling
ends of the scarf, seated in paryanlciisana with his
viilzann elephant below the seat;
(ii) pot-bellied two-armed Agni with flames around his
head and shoulders and a water-pot (kamnndalu) in his
left hand [, seated] inpavynrikiisana with his ram below;
(iii) pot-bellied fierce-looking384 two-armed Yama holding
a da!zda with his right hand, seated in nznhiiriijalild,
with a humped bull below;
(iv) fierce-looking Nirgi (lower portion broken, the animal
viiharza partly preserved) with a sword [r.];
(v) Varuna (partly preserved, on the top of the northern
jamb) with apdia;
(vi) two-armed Viiyu bearing the staff of a banner in his left
hand, [seated] in rnahiiriijalild, with his mount deer
below;
(vii) pot-bellied two-armed Kubera with a halo behind the
head [, seated] in malz8riijalilii (viilzarza damaged) and
(viii) two-armed fSiina holding a trident in his left hand [, seat-
ed] in mahdriijalilii, with his bull below.
The noteworthy details of this early assembly of dikpiilas are the
bull vehicle for Yama and an animal vehicle for Nairfta. The latter
could be cornpared with the donkey of the Mukhalingam frieze
mentioned above. The author failed to mention the attribute in
Kubera's right hand, which seems to be a club, and the usual rosary
in Agni's right hand. There were probably other additional attrib-
utes, which have since been obliterated. Another important feature
is the two-armed form of TSnna, which was co~nmonin the 9th
century (and until c. 950 A.D.)in Central India, of which a few exam-
ples exist in Orissan art.
In the vicinity of Mukhaliligam and Nhriiyanapuram, another
dik@la-bearing temple has only recently been uncovered: the
KiimeSvara at Galliivalli (District Vizianagaram), which has been
variously assigned to the 9th or the 1lth century.385 After evaluat-
ing all the available evidence, I have come to the conclusiorl that this
temple should be assigned to c, first half or middle of the 10th
century, taking into account the close resemblance of its sculptural
"'MITRA1991: caption of plate 894.
374 They may have been c~nployedto convey a fire-water (on a more subtle level:
sun-and-moon) symbolism, or probably exemplify the roles of the respective planetary
deities above them: SOrya in the case of Agni, and R3hu in the case of Varuna. The
former two are both connected with the positive, non-destructive aspect of fire,
latter two Inay be regarded as potentially dangerous.
37s MITRA1991: 430. Dates assigned by otlier authors vary between the second
half of the 8th century and the middle of the 9th century,
376 The dates for this telnple also differ considerably. We refer to the recellt date
O~.SMITH (1994: 135). The figure of Yama on the lower cc~itralpart o i tlie tower can be
seen on pl. 903 in MITRA1991.
377 These attributes are typical for Yama in Orissa, but do not occur anywhere
else. D~NALDSON (1987: 1145) is unsure about the noose and describes it as a rosary in
the god's left hand. However, what he regards as a rosary must be the end of the noose.
'78 The authors propose a tentative 10th century date. However, a few details dlf-
fer from the Birarnb3 representations. Since Ya~nais depicted at both sites, a C O W a r -
ison reveals that the buffalo's head and horns are much bigger and diflerenll~shaped
at Mukhaliligam. The god here also seems to have an ugra (demonic) face and wears
cr~lltvakeia(hair standing on end) unlike his counterpart at BBramb3. His staff is a
khn!valiga (skull-topped staff), and his legs are held in a less angular positioll. 111
cases. the noose is -unusually - held in thc right hand of the deities, although inOrissa,
the mJose is generally shown in the proper left hand. Nairyta's sword is also shown On
the left side, which does not occur anywhere else. Thus, the published pliotogra~hy
the frieze may be a mirror image ol. the original.
379
date has been assigned by SMITII(1994: 135). whoin I prefer to follow in
the cases of the templcs discussed by him.
"O SMITII1994: 135; cf. also to his discussion of the temple pp. 111-112.
AS has been stntcd earlier, thc feature of Agni's left arm not extending below
the level of the elbow is a gcneral sign of Eastern Indian images. It already occllrs In
the earlier art of Bhubaneswar, e.g. on a narrative frieze of the SatrughneSvara temple
(DONALDSON 1987: fig. 3637).
LINDA(1990: 232), in her account of the temples at this site, Iocatcs it in Ille
Srikakulam district. This may represent an older district division.
383 Both MITRA (1987) and LINDA(1990) independently date the group of telnples
to the end of the 9th century or early 10th century. The latter author's main argunlcnt
is the similarity with the SomeSvara te~nplcat Mukhaliligam, which is also assiglled
SMITII (1994: 135) to c. 900.
3" This Cxprcssion of MITRA apparently includes the ~ r ~ l / ~ v a kcoiffure.
eia
In the first publication by KRISIINA KUMARI (1985-86: 58; the site is referred
to as "Galiavalli") the temple was placcd in the l lth celltury on the basis of several
common features with the SomneSvara at Mukhaliligam. Later, iconometric studieswere
undertaken, which, although based 011 the same icono~netricalchart (JAYA SREE19g8-
89: 57; 1. V~JAYA LAKSIIMI/M. KRISIINA KUMAR~ 1991: 117), led the authors to differ-
entconclusions and chronological assignments. JAYASRCE (1988-89: 58) believed that
iconolnetric system of the GallBvdli itnages more closely rese~nblesthat of the
MadhukeSvara at Muklialingam and therefore proposcd the first hall. of tlie 9th century
as the probable date. In the monograph on the temple, VIJAYA LAKSIIMI and KalsklNA
KUMAllr (1991: 116) saw thc closcst affinity between the BhimeSvara at Mukhalihgam
and the temple under discussion. Tlicy proposed the first half of the 1lth century as its
date. They have made no reference at all to JAYASREE'Sarticle. Both the latter publi-
cations should not have relied exclusively on iconometry for assigning a date to the
KimeSvara temple.
 iwkF!~ with the NilakantheSvara at NBrHyanapuram and the
SomeSvara at Mukhalingam.386
However, it should be noted that the placement of directional
guardians is quite irregular, and that both Indra and Yama are absent
here (ill. 30). A general characteristic of the Gall&vallifigures is a
fresh, innovative attitude towards the concept of dikpdlas. Every
single panel has been caref~lllyconceived, and except for a certain
flabbiness in their bodies, these figures and the other details could
surpass their counterparts on the well-known RBjarBni tern-
PI^ at Bhubaneswar.387
The temple consists only of a tower (deul) with rather simple
base mouldings (VIJAYALAKSHMIIKRISHNA KUMARI1991: P]. 3 4
" O V ~ a plinth. It faces east. The dikpdlas are found on the southern,
and northern walls, the corresponding niches on the
east enclose the river goddesses Gang2 on the (south-)east and
Yamun&011 the (north-)east. The river deities also flank the sanctum
doors on the related temples ~ilakantheSvaraat NBrByanaPuram
ad some~varaat Mukhalihgam. It is therefore apparent that the
of the large-size river goddesses, presumably existing on
earlier temples, was maintained, while room Was made for the
dik~dlafigures only in the six remaining niches.
5 Kubera Mahisiisuramardini 6 ZSiina
7 5 Kubera (fig. 247) is depicted on the north(-west).388 He
3 Varuna
I I
2 Naiqta GaneSa I Agni
'l1. 30: Gall$valli, KSmeSvara temple, plan sketch
The members of this nearly callonical djkl~rilaset require a detailed
description:
l the deity.
(fig. 243) on the south(-east) stands in a hieratic Pose,
with his legs apart. His hair is tied in jn@lrzuku@(hair-crown).
He is bearded and without earrings. His arms are bent at elbow
level and carry a large, round rosary (S.) and a pot (1.). A g n i ' ~ exactly the same: cf. VIJAYALAKSIIMI/KRISHNAKUMARI1991: pl. 8 with LINDA1990:
lower garment is particularly long. Like the other dikpdlas, he
a necklace, a long chain upavita (sacred thread), key1lras
arm bracelets) and a girdle. Flames encircle his head
"d ~houlders.On the proper right side, they extend further L~~SRMI/KRISHNA
dowll and seem to be issuing from the goat vehicle standi?g 2601). ~t is quite obvious that the same workshop which constructed the SolneSvara
below. As an unusual feature, two fire altars flank the god
An attendant on the proper left side holds a lance (later temple at sarapaIIi, DONALDSON
" "tribute of Agni in Orissa) in his right hand.
dagger or short sword pointing downwards in front of his
body. His hair is standing on end (IrdlzvakeSa) and he wears
the usual ornaments, including bangles. The deity stands on a
prostrate man-vehicle, with the unique attributes of a sword
and a shield. Unlike most other representations of this anthro-
pomorphic vehicle, he does not support his head with his hand.
3 Varuna (fig. 245) on the (south-)west also has a few unusual
features. Like all the guardians except for Agni, he sta~ldsin a
relaxed stance. His right hand holds a ribbed vessel at elbow
level, the left hand rests on the hip and holds a noose, the loop
of which appears besides Varuna's head. A female attendant
holding something indistinct (also a noose?) stands on the
proper right, beside of the open mouth of a nzakara. The god's
element, water, is represented by wavy incised lines on the
plinth below. The vessel attribute may also point to the same,
aquatic symbolism.
4 Viiyu (fig. 246), on the (north-)west, holds two attributes,
which typically occur in Orissa. His left hand holds a mighty
standard, while his right wields an elephant goad. A faint allu-
sion to his sphere, wind, is made by the fluttering end of his
sash. On his proper right stands an animal-headed attendant,
leaning on a mace and probably holding a fly-whisk in his left
hand. The figure could represent the personified vehicle of the
god, but the horns are rather short for an antelope.
has a particularly obese body, which is further exaggerated by
his stance. Like Varuna, he holds a ribbed vessel in his right
hand. He wears the customary ornaments, but an uncomlnoll
feature is the mighty club in his left hand. Another unusual
feature is a small tree on the proper right side, which extends
below Kubera's arm. It seems to have a carved face, which
would be unusual, and obviously represents kalpavrk;a, the
wish-fulfilling tree. The ka1pavrk;a motive also appears on
the RgjarLni temple (DONALDSON 1985: 315), but the mode of
depiction is completely unique here. A pot-bellied dwarf,
holding a cup like Kubera, squats below the tree.
6 ISgna (fig. 248) on the north(-east) is shown two-armed as the
others. His right hand is extended in cimizudrd (thumb and
index finger touching) in the direction of his bull vehicle. The
trident in his left hand has a particularly long handle and is
supported by a dwarfish figure. TWOrosettes flank the head of
386 The basic concepts of the images of KBrttikeya and Mahi;jsuramardini are
25 and DONALDSON 1987: pl. 3033 for KBrltikeya; cf. VIJAYALAKSHMI/KRISHNA
KUMAR[1991: pl. 9 with LINDA1990: @l. 28 and DONALDSON 1987: p]. 3196 for
Mahisisuramardi?i. Moreover, a Set of plalletary deities (rtavngmhas) with an unusual
figure of Ketu with folded hands (aijali) is sharcd by the KBnieSvara (VIJAYA
KUMARI1991: p]. 12) 2nd the So~neSvara(DONALDSON 1986: fig.
temple at Mukhalidgam was also responsible for the erection of several other shrines
ill a region extending at least as far south as the viciliity of Vizianagaram (DibbeSvara
1985: figs. 619-6271,
387 This impressive dikpcila Set of about 1030 A.D. has been illustrated in

Nairtta (fig. 244), on the south(-west), wielding a sword in DONALDSON1985: figs. 821-828. There are quite a few colinnon traits between the
his right and carrying a severed head in his left hand9 repre- ~ ~ l l j ~and~ Rjjari~?i l l i clikp6las, but it seems inipossible to judge from their features
only, which was the earlier set. The Galllvalli set is certainly not a poor copy of the
sents a type which we have not yet come across here. Like the ~ j j set,~I am~ inclined ~ ~to believe
i that in the latter the concept of the former one was
members of the set, his lower garment is shon9 "ld ilnproved upon, by o~iiittingsome of its too Inally details. Obvious differences are: the
he wears a crown (probably the c'basket'' type kara!zdonzu- four-armedform of ighna and the much shorter rf/)avitas(sacred threads) of the dikl)i/os
ku!a). Fangs can be seen protrudillg from his mouth. on the R'qjarhl!i temple. Later images of the directio~ialguardians in Orissa are always
nanking the god, is a pair of rosettes, which also sometimes seated. 1x8 In the mentioned publications, Kubera has been wrongly identified as i g j n a
Occurs On the panels at Ngrjyanapunm (LIND*1990: P'S' 26' and vice versa (e.g. VlJAYA LAKSHMI/KRISHNA KUMARI 1991: 65).
a
27. 29). On Naiqb's proper right side, a stout warrior
It is difficult to say why the placement of dikpdlas is so unusual at Agni
GallZvalli. It could be an irregularity which often occurs in early,
experimental sets. The distribution of dikpdlas at Galliivalli, how-
ever, recalls their positioning on the Svarga Brahma temple at
j\lampur (ill. 11). The directional guardians occupy the correspon-
ding karna (corner) niches there, but some of those present at &am-
pur are missing at Galliivalli: Indra, Yama, Soma, and Siirya.389 The
basis for this resemblance, however, could be the supposed starting
point of the set: at both the early dikl~rllasites of Cauriisi and
Ganeiwarpur (to be discussed), there are sets starting with Indra on
the (north-)east. All guardians except for Agni would be placed in Yams
their appropriate positions, provided, this order had been employed
at Gallavalli. With the east wall not included in the dikpdla concept,
the set should have started with Yama on the south(-east). Such a
starting point might have been avoided, because of Yama's inaus-
picious nature.
There are two more temples with early standing dikpdla images
in Orissa: the Viiriihi temple at Cauriisi (District Puri), and the
Vaisnava Paiicapandava complex at Ganeiwarpur (District Cut-
tack), which have been regarded by DONALDSON (1987: 1142) as the
earliest evidence for the placement of the dikprllas in their canoni-
cally assigned directions. In the latter case, the dikpdlas are pre-
served only in the karrza niches of two subsidiary shrines (south-
eastern and south-western) of a patictiyatarza complex, both of
which have their doors on the northern side. The only other surviv-
ing subsidiary temple (north-west) shows scenes from the life of
Krsna in the respective niches. At Cauriisi, the directional guardians
are installed on the jaganzohana. The recently assigned dates are c.
970 A.D. for GaneSwarpur and c. 985 A.D. for Caurbi.390 It is quite
difficult to place the extant dikpdlas in a relative chronology, as the
partly corroded figures in the rather small niches (heights of about
26 cm) can hardly be compared to the larger, more elaborate panels
at Galliivalli.
The female counterparts of the dikpdlas, installed in the neigh-
bouring niches, respectively (DONALDSON 1985: figs. 704-70g),
have an early occurrence at C a ~ r B s i . 3A~ ~rare feature at this site
which DONALDSON observed (ibid.: 1145) is that Nairrta is not yet
accompanied by the prostrate human vehicle, which becomes com-
mon later in Orissa, but by a dancing figure. A corresponding danc-
ing figure also appears on the Riijariini temple at Bhubaneswar
(ibid.: fig. 824).
The order in which the clikpdhs were shown is of considerable
significance. At Caurisi and on the south-eastern shrine at
Ganeiwarpur, the sequence starts with Indra on the (n0rth-)~a~t.392
However, DONALDSON (ibid.: 1142) failed to notice the fact that the
south-western temple bears a set which is arranged in the usual way,
with Indra situated on the (south-)east. Further, there are a few sig-
nificant variations in the rendering of the dikpdlas on the south-east-
ern and south-western shrines at Ganeiwarpur. The images will
therefore be discussed - deity-wise - in detail. Their typical stand-
ing posture is with a slight flexion (dbhariga):
Indra The figures of Indra (figs. 249,257) resemble each oth-
er largely. In both cases, there is no trace of a thunder-
bolt in the damaged right hand. Perhaps some other
object was held by the god. His left hand rests on the hip
(kafihasta). The rendering of the elephant vehicle is sig-
nificantly different. On the south-eastern temple, the ani-
mal seems to be seated on Indra's right side (perhaps
shown partly anthropomorphic), while on the south-
western shrine, only the forepart of the striding animal
has been portrayed, held by a rope on the god's left side.
Both versions of the fire-god (figs. 250,258) must have
shown him carrying the typical attributes rosary and
waterpot (not preserved on the south-western shrine) A
striking difference is the hieratic posture on the south-
western shrine, which is reminiscent of Galliivalli
243). In this case, and probably also in the other version,
fires seem to have been depicted floating above as at
Galliivalli. The vehicle and an attendant must have
accompanied Agni on the south-western shrine. Both are
apparently absent on the south-eastern temple.
In the depiction of Yama, there are also obvious differ-
ences (figs. 251,259,260). The god occurs once on the
south-eastern temple, with a stout, obese figure, and
twice on the south-western shrine, where be is shown
with an ideal body: on a recessed panel in the bhadra
position (projecting wall portion) above the cornice
(fig. 2% and in the regular position on the south(-west)
(fig. 260). His attributes are generally a staff (r.) and a
lloose (l.), but in the latter image, the right arm is extend-
ed in the direction of the vehicle and does not hold an
attribute. The vehicle is absent on the south-easter11tern-
~ l ande in the upper version on the s ~ ~ t h - w e s t eshrine. rn
The latter image is directly flanked by two attendants and
also by two large attendant figures in separate panels.
Nairrta His images are very similar on both subsidiary temples
with a sword (r.) and a severed head (damaged, 1.) qS
attributes (figs. 252,261). The cir&valceSa coiffure (har
standing on end) is nicely preserved on the south-west-
ern temple. The most important features, however, occur
on Nairyta's right side. On the south-eastern temple, a
human vehicle appears to be standing beside the god's
socle, which is unique (fig. 252). On the south-western
shrine, it appears as if an animal vehicle once stood in the
same place, but sharing the socle with the god. The shape
of the socle is indistinct, possibly as the result of erosion.
Varuna The images of Varuqa (figs. 253, 262) also closely
resemble each other. The right arm is extended towards
a small nzalcarn in the (proper) right corner, and the left
hand, resting on the hip, holds the rope of the circular
noose. The position and attribute of the left hand corre-
spond with the counterpart at Ga1l;lvalli (fig- z4')'
Varuna wears a conical crown and large round earrings
like most of the other clikpdlns.
V i i ~ u The images of Viiyu (figs. 254,263) also exhibit very
similar features, except for the position of the left army
which is held lower on the south-eastern shrine.
'"
and SiIrya, who are associated with the di$tiln group at Alalnpur'
appear flanking the door, which is elnbellished by an n.F!aciikI~dlnset, at ~ ~ ~ ~ y ~ n ~
ram, NilakaniheSvara temple, in the vicinity of Gallfivalli (see above). Thus, solne k'nd
association of these deities also occurs there,
SM~TII 1994: 135; DONALDSON'S date is first quarter of the lOlh century
Caurssi (1985: 274) and sccond quartcr of the 10th century for Ga~!eSwarP~~ (ibi(l':
282), thus differing not only in the dates but also in the chronological sequence ''le
temples.
'" The felnale f0rms generally carry the same attributes. As an interesting deF1'
the trident is held on different sides in the case and his consort. He holds It In
his left, while she has it on her right. Tllere may be some gcnder-specific ~ y " b ~ ~ ~
behind this. A colnparable sculpture, probably Nairyli, the fc~nalecounlerpart of
Nairlta, is known from Khiching (near Bcnisjgnr, Bihar; D~NALIJSON 1985: fig. 587)'
The context and date o l this figure are uncertain, Iiowever.
In this connection, it should be noted that the Indra image of tlie visnupSda
temple at GayYBiliar (cf. note 483), is presently wors\lippcd in a shrine in the
eastern corner of thc compound.
 attributes elephant goad (r.) and standard (1.) correspond
to those at Galliivalli (fig. 246). The figures in the lower
portion is quite indistinct. On the south-eastern shrine at
least, there seems to be an attendant and a vehicle flank-
ing the god.
Kubera The stance and the positions of the arms are the same in
both cases (figs. 255,264). His right hand holds a club,
the left arm is broken. The shapes below are again indis-
tinct. It is probable that an attendant stood on the (prop-
er) right side and that the forepart of a leaping ram was
depicted on the god's left.
kna The two-armed &na images (figs. 256,265) also do llot
exhibit any significant differences. His right hand is
raised to the elbow level and probably held a rosary. His
left hand holds a long-handled trident (top damaged), the
lower end of which is supported by a dwarfish attendant.
This detail is also present at Galliivalli (fig. 248). The
bull vehicle appears to the right of the god. A character-
-istic feature at GaneSwarpur is the serpent earring in
ISZna's right ear.
The variation in the images depicting the same deities at temples for "related" deities).395 Out of the four deities Agni396,
GaneSwarpur as well as a varying placement of the gods Inay be
considered as indicative of an early stage of dikl)iila represelltation.
The theme was handled in an individual manner and not according
to a Stereotyped concept. Regarding the later Orissan ~ 0 n ~ e l l t i oofn cation as Cat~dra.~OO
showing Nairyta's vehicle as a reclining figure (e.g. DONALDSON meant to function as a dikpiiln, and he is found on the northern side
1985: fig. 824), the evidence at GaneSwarpur sl~owsthat this had not
Yet been introduced here. GallZvalli has a utlique version related to
the later concept, showing Nairfta with a human figure under his
feet (fig. 244), who is characterized as a warrior like the figure
Standing besides him.
Some features indicating a significant correspondence betweell
Gall%valli,Caurasi and GaneSwarpur (i.e. the two-armed form of
ISZna, always with an attendant dwarf on the left; the extended left
hand of Varut!a bearing a noose; Vsyu carrying two attributes) are
not continued in the 1lth century Riijariini temple. Here, in the case
Varuna, the hand remains higher, and ISiina is four-armed and
without a dwarf attendant. Although " the evidence is somewhat
meagre, it is quite likely that the three sites with early clikl~iilasets
On the jarighii (wall frieze) portioll share a traditioll of rlilyiila
representation, whicll was inspired by the Central Indian art of the
late 9th/early loth centuries. As the lively imagery at Gall%valli
~llows,some other, probably indigenous, elements were employed
to render the theme in an allnost epic manner, elements that are
missing in the somewhat lifeless dikIxila sculptures of that period in
Central India. Botl1 traits, 'borrowed' and indigenous, of the OrisSan
region were ingeniously perfected on the l lth century Riijariini
at Bhubaneswar.
2.3.2.5 Southern India
There are only a few pieces of evidence in the modern state of
Tamilnadu from the period under discussion, which witllesses the
earliest occurrence of directional guardians in the extreme south of
the subcontinent. Although the Kailiisaniitha temple at Kaficipuram
Was constructed prior to 750 A.D., it will be considered as a supposed
direct predecessor of telnples with dikpiila representations in the
South in this sectioll. The absence of any securely identifiable com-
plete set of four or eight directional guardians before 1000 A.D.
makes it doubtful that the group as a whole was known at this stage
at all. The Dhannapuri and Armamalai ceiling panels, which have
been discussed above, are situated quite far to the south (northern
Tamilnadu), but they form a stray evidence, following the Deccani
tradition. However, the as;adikpalas were a standard programme of
most of the southern temples at least from the 1 I th century onwards.
The most impressive example for this full-fledged stage of dikpdla
representation, the great RiijariijeSvara temple at TaiijSvGr (c. 995-
1010 ~ . ~ . ) 3 9was
3 , begun at the very end of the period under discus-
sion. It will be briefly included here, as well as its immediate suc-
cessor, the Riijendrac6liSvara at GailgaiI<6ndac6lapuram.
On the other hand, the few early dikpiila images from the south
may have been just a faint echo of the tradition of their representa-
tion in the Deccan, and there are good reasons to assume that some-
what different concepts about heavenly guardians might have pre-
vailed in the region.
The Pallava Kailiisaniitha or Riijasiq~heSvara temple, a well
known architectural gem of about the first quarter of the 8th cen-
tury,394 according to R. NAGASWAMY (1969: 13) has representations
of the 32 padadevatiis, including dikpiilas, of a viistupurusn-
nza!zdala (foundation diagram) portrayed on reliefs in the outer
priikiira, which contains a sequence of 58 devakulikiis (miniature
Yaina397, Soina and h a , which he identifies, the probable Soma398
(fig. 285) and ISiina399(fig. 286) images are illustrated here. The
presence of a halo in the former case points to a probable identifi-
However, the Moon God was not necessarily
of other early southern temples also.401The personage on the north-
east (fig. 286), seated and leaning on a mighty club placed on the
ground upside-down, exceptionally without a female partner, could
be TSiina, but there is no parallel evidence for such a form of the god.
In any case, the hypothetic installation of a colnplete group of direc-
tional guardians in thepriikiira of the KailQaniitha temple remains
doubtful.
A conscious differentiation of the directions in the same temple
becomes apparent in the choice of the images adorning the lower
394 For a descriplio~isee K.R. SRINIVASAN 1983: 5 9 4 4 . An early account of the
te~nplehas been made by A. REA1909: 18-42.
395 REA(1909: 29-35) counted only 55 cells. Hc did not count three of them,
because hc regarded them as structurally incomplete.
396 Between cell 3 and the south-cast corner recess, according to RDA(1909: 30).
It is very likely that the figure's hair appcars in thc for111of fla~nes(as detcctcd by
Dr. R. Nagaswarny, personal comm.). Adjoining this relief on llle right, very close to
the south-eastern corner, there is an image of Gal@ (not numbered by REA).Thus,
Agni does not exactly occllpy the corner position.
397 NAGASWAMY 1969: 22, bctween cells 12 and 13. No such deity or divine cou-
ple, which should be located between Trivikrauia (REA'Scell I l ) and the OIJJ!.IOI,~~I~-
rha~lascene (between cells 11 and 12) is described by REA(1909: 31). REAsuggests
that Yania is represe~itedamong other figures on a panel in cell 10. Tlie presence of
Ya~naon the outer wall (south) of the slightly later MukteSvara temple at tlie sa111esite
has beenpostulated by L'HERNAULT (1978: plan V). It is likewise doubtful, and no otlier
probable dikl~cilncan be iden~ifiedthere. In this connection, it may be significant to note
that Yama appears in the list of parii~~m~levntcis (sorrounding "related" deities) at
Tirupparaitturai of c. 885 A.D. (TRIAGARAJAN 1989: 240). which does not include any
other ~neniberof the dik~~n'lfl group.
398 NAGASWAMY 1969: 23. REA(1909: 34) describes the lialoed image as located
betwccn cells 43 and 44.
3g9 NAGASWAMY: ibid. According to REA'Slist, tlie i~iiageis located bctween cell
52 and the north-east corner.
400 D, TIIIAGARAIAN (1989: 172; cf. pl. 28) colnmcnts: "Althougli he Inay be
Candra [,l attributes associate him equally with Silrya in the Pallava period II".
401 E.g. on the third tnln of thc DIiannarSjaratha at M3mallapura1i1,c, mid-7th cen-
tury (SRINIVASAN 1983: 40); likewise on the PBndya Ve!!uvalik6vil at Ka!ugumalai,
sccond taln, grl~ol~i!~<liof c. 800 A.D. (TIIIAGARAJAN 1989: 222-223).
parts of the shafts of the pillars flanking the oblong shrines in the
south and north walls and of the gopura tower on the west (now
closed) of the priikiira. Thus, five-hooded nLEgas (serpent kings)
with hands in aiijalimuclrci are restricted to the western gopura tow-
er (HARLE1963: pls. 11, 12). Their presence alludes to the aquatic
character of the western region, which is perceived in Buddhism and
Hinduism alike.
A third instance of a possible reference to the dikpiila complex
is the depiction of heavenly riders ascending above a VinBdhara
Siva with attendants (REA'Scell 46, north; fig. 287). Both riders are
accompanied by their consorts, seated behind them as usual. The
couples sit close to each other and are shown in profile. The ani-
mals are - doubtless - bulls, so that both riders can be subsumed
under the "kina type". One of the possible meanings could be the
concept that the heavenly guardians are all identical with Siva, a
concept, which is also expressed in the garbhngrha set of the
KumbhaSyiima temple at Cittaudgadh. A frieze on the upper por-
tion of the rear wall in the shrine of the AirBvateSvara temple at
Kiiiicipuram (REA 1909: pl. CXVI) is reminiscent of the ascending
riders of the KailBsaniitha temple. Here also, riding couples appear
on bulls. There are altogether six of them, three proceeding to the
left and three to the right.
Another interesting instance of heavenly riders, which, like the
latter example, exhibits a remarkable degree of symmetry, occurs on
a Jaina panel at Kaiugumalai (DE LIPPE1978: p1.194). The well-
known rock-cut temple VettuvankBvil is the earliest major stone
monument in the extreme south (formerly Tirunelveli dist.).402 It has
been assigned to c. 800 A.D. (SOUNDARARAJAN1983: 83). Above an
unconspicuous Jina figure seated on a lion throne, the usual three-
tiered parasol and two rows of creepers are depicted. The upper row
consists of dancers, musicians and mythical sages. It is flanked on
both sides by a flying male figure which seem to represent Sfirya
and Candra. Above this section, the upper space is exclusively filled
by riders, which may be understood as heavenly guardians, plus two
standing attendants. The central figure is shown frontally, seated on
a huge elephant and attended by two standing males. On both sides
of this central group, four riders, arranged in two tiers, proceed
towards the centre. Their vehicles are: lion (upper row, inller fig-
ures), serpent (1.) and bird (r.; upper row, outer figures), yiiii (myth-
ical creature; lower row, inner figures), horse (lower row, outer fig-
ures). All the riders have one arm raised. It looks as if some of them
are holding banners. The central elephant rider could be Indra. How-
ever, in South India, the god KLttikeya has been represented seat-
ed on an elephant, generally on the rear side of a temple, from about
the middle of the 7th century onwards (L'HERNAULT 1978: 141-
142).403 Although Indra was traditionally worshipped in the region,
early depictions of this deity are, strangely, wanting303. Thus, the
Jaina relief at Kajugumalai appears to depict a group of four or
eight divine warriors or guardians led by Kiirttikeya as a general of
armies.405The vehicles of the riders, however, have nothing in com-
mon with the usual dikcpSila vehicles. They might have been chosen
arbitrarily, but they might also represent a set of directional animals
which was common in the region.
There remains a faint possibility that among the numerous
sculptures on the superstructure of the Vettuvank6vil (e.g.
SOUNDARA RAJAN 1983: pis. 51-52), some might have functioned
as directional guardians. One of the sculptures has been identified
as a probable image of Indra by P.Z. P A ~ A B I R A (1975: M I N pl.
CLXVI, fig. 1). It is located on the first tala, south(-east), beside a
female figure on the corner. However, there is no particular reason
to suggest that the figure should be identified as Indra. On the first
taln adjacent to the corner figures, there are altogether six un-
specified male images, two each on the south, west, and north. The
eastern side is devoid of any sculptures.
On the NBgeSvarasvBmi temple at Kumbak6nam (Tanjavur
dist.), assigned to about 886 A.D. (DHAKY 1983: 161), a group
of eight dikpiilns on the second tala of the superstructure has
been noticed by T.V. MAHALINGAM (1937: 33-34), but it is quite
apparent that the rather large figures consist of stucco and are not
original. For the sake of completeness, the author's description is
quoted here:
In between the corner figures carrying the cornice and the pilasters IS
a four-armed image [on each of the four sides]. The images which are
now painted are of stone [?] and wear necklaces, urlnrrrbanrllzn, va!a~clst
kafikn!?~is,1cn;isCtrr: etc. While their upper of arms hold icon'-
graphic cognizances, the lower pair are in the nbhnya and k ( l ! ~ n v ~ ~ ~ ~ -
bits, the only exception being the first image on (he southern side whose
lower left arm holds a vnjra. Coming in a clocl<wise direction the dtrlb-
utes see11in the upper arms of each of the images are as follows: a bud
and an indistinct object [Yama?], sword and shield [Nairyta?], ilkti and
noose (?) [Varu?a?], an indistinct object and noose [ V ~ Y U ?sword
], and
shield [Kubera?], parni~iand n z p [TSiina?]. [...l The projected pXt
[~~~l(a~lfis(i]
of tile vinzcina is carried by an image at the southern and
northern sides, each with four hands, the upper right of the one i n
South holding a vnjrn and, that of the North holding a pZh. The lrnage
on the South has flames around his head. On either side of the proJec-
tion in the East is a carving [?l, standing wit11 four hands and Orna-
merited with necklaces, yajilol,nvrrn etc.; the carving in the southern side
has two heads [406] with fla~nesover them and hence nlay be Agni
while the details of the corresponding northern image could not be stud-
ied. These two images [...] make the set of the dik,,cilns complete.
ThejnfigIzSiimages of the same vin?n/za,where an isolated image of
Agni appears on the north(-west), have been interpreted in
context (D.T. SANFORD 1994: 50; fig. 12). The impressive standing
figure was first identified by S.K. GO~INDASWAMI (1935). The
~haracteristicfeature that indicates his identity as the Fire God is the
erect flame-like coiffure.407 In this connection, the early occurrence
of this motif in the PBpanjtha temple, Pattadakal (fig. 56), is called
to mind. The Ceylonese text M a i i j ~ l s ' r i b I ~ a ~ i t n - v z s t ~ ~ v i d ~ ~ ~
265a also describes this form of Agni ("$iklziijvSila~z").
4"2 Accordillg to the ncwly introduced district divisions, ~ajugulnnlaiis "OW
situated in the Chidambaranar district (Map of Ta~nilnadu,Governmentof India,
right 1994).
4" The well-known elephant rider on the eastern (rear) [ace of the Arjulla
at Mimallapuram (Chengalpnttu dist.) represents most probably KZrttikeya (L'HER-
NAULT 1978: 141).For an illustration sec I<.R. SRINIVASAN 1983: pl. 4, c. mid-7tb
tury. The strongest argument for such an identification lics in thc fact that the
rider wears 3. conical crown with a broad base which is typicalfor K.ir~tikcp
HUNT~NGTON'S (1981: 63) identification of the sallle figure as ~ i ~ a f i B r - g iiss tquite
~
ingenious, it must be kept in ~nindthat it is not supported by any other early icon'-
grapllic evidence. Morcovcr, H U N ~ N G TWas ON not aware of L'HERNAULT~S study *''
other depictions of frontally shown elepliant-ridcrsappear to bc modelled the
image on thc Arjuna rcrrha. The absence of ally attribute is similarly only reported*'Om
the exalnplc at Koc!umbi!fir, Mhvarkcvil, dated c. 880 A.D. (L'HEIzNAULT 1g7': ph'
127). Later images of elephant-riders are equipped with KZrttikeya's attributes.
404 HUNTINGTON (1981: 61) mentions the $ilq~~,nc[ik~rar~~, traditionally a"ig"d
10 the 2nd-3rd centuries AB.,as referring to shrines and a festival of Indra. For a prob-
able South Indian image of Indra, standing willlout vehicle, of the 9th centoryl See
below (TIIIAGARAJAN 1989: pl. 105g).
405 An association of Kiruikeya with the dikl76lns has been rcported from A!kor
VWCambodia, 12th century (K. B r l ~ l - r ~ c r ~ a1961: n y ~ 139-141); cf. also Ellora
(Kailisa; fig. 278).
40C
The two-headed form of Agni is certainly later. The Agni images
Cidambaram &!O/JltmS (late 12th century and later) are probably the earliest ones
tYPe (C.&!. HARLE1963: 108-109; pl. 143).
407 It is likely that the rather straight slrallds of flame-like hair indicate Agni"
~ e a c ~ f nature.
ul A lnandorla of expansive, flickering flalnes is in Soot11 India'' "t
mstrictcd to ferocious dcitics (e.g. B. LEGRAND 1987: ,,h. 81, Bhnirava at Ki!aiyfir, l"'
9th ccntury).
 Before discussing a group of loose, seated images from differ- the short upavita, which only reaches as far down as the ~ r h r a -
ent sites, a rather enigmatic group of nine standing images from bandlza ("belly band"). The deity sits in viinzapralanlbnpaclnb (left
Tirukkadaiynr (Dist. Tanjavur), Amytagha~eSvaratemple, will be leg pendent) and keeps the right hand in a peculiar variant of
introduced (THIAGARAJAN 1989: pls. 105a-h; 281-283). They abhayanzudrii (bestowal of protection) with bent fingers. The left
doubtlessly form a group and have been assigned to the "late mid- hand rests on the left thigh. The bench-like throne is very simple
die of the 9th century". The images all have a backplate, which is in and does not have a finished surface. The side-view shows that the
most cases rounded at the top. The figures stand in a hieratic pos- back of the sculpture remains completely unfinished. A parallel,
ture, generally with the left hand in katilzasta. They are clearly dif- which seems to be chronologically close (first half of the 10th ten-
ferentiated by their crowns or coiffures and their attributes. An orig- t ~ r y ) ~ l lexists , on the Sundareivara temple at MElappaluvnr
inal directional arrangement may be as the figures are (Tiruchchirappalli dist.). The deity is again Agni (LEGRAND 1987:
llowadays distributed in the eastern, southern, and northernprfikiira ph. 99), who sits on a similar throne with the right leg pendent. This
of the temple. For the two Aivins llowever, a directional guardian- image shows a number of rare features: The backslab has an orna-
ship is very unlikely. Without trying to force the figures into a pre- mentation of sinall fires on its border; the deity is bearded; he car-
conceived scheme, it is suggested here that there may have been a ries the same attributes as are known from sites further north, rosary
directional arrangement with ,$firya (ibill.: p]. 105a) probably on the (r.) and waterpot (1.). The broad flames of the coiffure have beell
east and Soma (?408, ibid.: P], 105h) on the north as at Alampur, symlnetrically arranged. AS another conspicuous feature, Agni
Svarga Brahma temple (ill. 11). Indra409 seems to be a member of wears two chains of rudrfilc~aseeds cross-wise around his upper
the group, wearing a crown and holding a small thun- arms. The loose ilnage is located on the south-east side of the tan-
derbolt (ibid.: p]. 105d), filrlher Yams with his staff (ibid.: PI. 105g), ple, facing west. While this could be an important informatio~l,it
""drobably Nairrta wielding a sword (ibicE.: pl. 10.5~).Two other is, unfortunately, impossible to decide whether it was originally
figures are, according to THIAGARAJAN, both provided with a lotus located there, as BALASUBRAHMANYAM (1963: 27) states: "The tern-
bud (1.) and a trident (r.). However, in case (ibid.: pls. 10.50, the ple has undergone repairs more than once, and its ancient architec-
trident is particularly large and lollg-handled alld the alleged lotus tural features cannot be inferred froln the present state of this
bud, which is very large and ribbed, certainly represents something structure".
else. It could be VL?yu7sbanner, because a fluttering element appears Finally, a high-quality set of Agni and Viiyu froln Tirunelveli
in the background. The second image (ibid.: pis. 105e) has a differ- (Tirunelveli dist.) shall be discussed (figs. 291-293). The images
ent type of trident or lance with short side-prongs and holds a nar- should be assigned to the 9th-10th centuries. Significantly, apart
TOW bud in his left hand. with this of light and the relative- from the posture and size, every effort has been made to individual-
ly long vertical strands of hair, the deity Inay be Agni. THIAGARAJAN ize the inlages, which lnay reflect the intention to express the dif-
(ibid.: 282-283) idelltifies the set as the sun god Sfirya with his ferent natures of the Fire and Wind Gods. Their physiognonlies are
entourage on the basis of textual evidence, which is not wholly con- clearly distinct. As with the National Museum Agni, parts of the
vincing. It is, however, very difficult to come to a final assessment, original stone block remain, but only in patches, around the lower
The hieratic standing postllre and the colnbination of lcatikasta portions of the arms. It seems that there was some fear of instability,
(hand resting on the hip) and one attribute is silni]arly found on a if the ilnage were coln~letelyreleased from the stone. The faces are
ntlmber of loose images, which are not very attractive, show little oval and the countenance serene. Agni has slightly roulldish eyes
v"iation, and are difficult to date (fig. 288). These figures with their (fig. 291). He wears the usual flames on his head, which are accu-
immovable appear to be guardians, The set considered rately shown terminating at his round halo. He wears all the orna-
"eve could also be assumed to have functioned as guardians.41° ments, including anklets. He is somewhat more obese than vsyu.
Moreover, some melnbers of the Tirukkadaiyfir set are bearded and His right hand holds a rosary with bent fingers. A peculiar profiled
probably h, fallgs, ~ lit is possible
~ ~ that~ it expresses
, an other- peg connecting ]land and breast seelns to be designed specifically to
which now rob- stabilize the hand. The left hand is closed and rests on the thigh. The
wise ullrecorded tradition of directional
ably remains incomplete. ~h~ AS^^^^ could llave served as atten- throne consists of a double lotus. Like Viyu, Agni is seated in
~lakSi!z~~~ra1anzbapdcla (right leg pendent).
of Sfirya and may have beell placed at his sides.
for the develop- The image of V ~ Y L (fig.
I 293) had not yet been identified tor-
Another important category beanng
rectly. The lollg, slightly sloping object in his right hand may at first
ment before the end of the loth celltLlryis a group of loose images
from different sites, which are at least close predecessors of those glance resemble a sword or axe,4I2but the undulating streamer issu-
dikp~[as, ing from it shows that it is a flag. Vsyu wears an elaborate, high
are known from the slig~ltlylater set in the Tafijivfir
RZjaqegvara temple (figs. 294, 295). ~l~~deity is in this category crown. Some other pieces of 0rnalnents as in Agni's case have been
shown seated on a lotus or bencll-like throne, gellerally a devoted attention. Thus, the heavy necklace, the u~>avita(sacred
Vehicle (figs. 289-293). are sometilnes added, but may thread) and the belt-buckle reselnbling a flower are pafiicular]y
elaborated. His throne is bench-like. A graceful lion figure reclines
"SO remain absent. Tile IleigIlt generally varies between 70 and
beside Viiyu's foot. It is difficult to decide whether the lion was con-
cm. As most of these images depict Agni, it remains doubtfu1
they were originally conceived as part of a colnplete group
Of directional guardians. 4118 TI-IIAGARAJAN is not sure about her identificatio~iof this image as Candra, as
The National Museum in New Delhi houses an ilnpressive the deity does not liave a halo like tlie supposed Stirya image. Thc bud attribute is quite
example of this group (figs. 289,290). Much like the image on the similar in both cases.
Jfl~ig/~fi (wall frieze) of N ~ temple (D.T. SANFORD
~ ~ 409 TI~IAGARAJAN s(ibill.: 282-283) identifies
~ the figure with cln!l<ln (staff) as Indra,
~ ~ ~
since she regards the staff as one of the attributes of Indra. This is, however, not true,
fig. 12), flames rise straight above his head to form his cOif- 410 A function as vcistt~l~acla-del~oIBs, as assunled by NACASWAMY for some of the
fure. The National Museum piece, however, shows a ribbon and a gods in the KeilBsanBtha temple at KBiicipuram (see above), is also probable here.
SmallpOrtioll of hair (?) inserted below this zone. In both cases*the 411 B. LEGRAND 1987: 21.
412 R. RUSSEK(1986: Abb. 120) erroneously identifies the dcity as ParaSur~~~la.
face is youthful, beardless and has a serene expression. The loose not correct.
Tile indicatedheigllt 156 cm is
Sculpture wears finely executed ornanlents, A significant feature is
ceived as Viiyu's vehicle. The fact that his counterpart Agni has no
vehicle indicates that this had not been a regular part of the concept.
Therefore, it could represent an allusion to a lion throne.
In the absence of any other dikprila sculptures, which might
have formed part of a larger set, the Tirunelveli pair remains an iso-
lated evidence of partly unknowll significance.
The great royal Riijariijeivara temple at Taiijjavnr, built shortly
after 1000 A.D., demonstrates what these icons may have been meant
for, if they had been conceived as directional guardians. The temple
preserves a unique arrangement of three-storied subsidiary dikpdln
temples with octagonal Sikharas (crowning members), which are
integrated into the prdkiira wall of the temple complex.413 The
shrines of the guardians of the four main directions are located with-
in the cloisters on the respective sides (for Indra see below), and
those of the guardians of the intermediate directions are situated
directly in the corners. The devakulikiis (miniature temples for
"related" deities) all survive with their Sikharns, mostly showing an
image of the deity414 and figures of vehicles at grlvd ("neck") level
(figs. 297-300).415 Out of the dikpdla icons which were installed in
them at ground level, only those of Agni (fig. 294), Varuna (fig.
295) and iSiina (fig. 296) are extant iiz situ. An exception has been
made in the case of Indra. His shrine is built into the northern side
of the inner east gopura tower, i.e. on its northern side. The soutll-
ern side has a corresponding structure, in which an image of a
Niigarjaja (serpent king) has been enshrined in the cella at ground
level. Images of both deities416 appear on what corresponds to the
third storey of the devakulikds (PICHARD 1994: ph. 352)?17 The
southern shrine has been, most probably erroneously, generally
assigned to SCirya, as an inscription records the installation of a
Sarya image in the eastern gopurn t0wer.~l8In the structurally relat-
ed, later temple at Gangaikdndacdlapuram, a similar shrine seems
to have existed only in the southern portion of the ruined gopura
tower (PICHARD et al. 1994: ~ h339). .
he small grivd reliefs ficing the four cardinal directions (figs.
297-300) show the dilcpcllns either in a seated or standing posture
and generally carrying one attribute. In the case of Agni, it can be
seen that the god is depicted on the Sikhnra as standing, while he is
portrayed as seated inside (fig. 294). The characteristics of the
dikpclln images at TafijjavCir are largely the same as those of the loose
images discussed before: the large sculptures are seated with sparse
attributes, a throne, no vehicle, and an individual treatment of
details.
It is worthwhile comparing the image of Agni (fig. 294) to those
of the same deity described earlier. Points of similarity are largely
restricted to the MBlappaluviir Agni (LEGRAND1987: ph. 99), which
also belongs to the Cd!a dynasty. The TaiijSvcr Agni is clearly rec-
ognizable as having developed from the earlier image. He is par-
trayed bearded, with flame-hair, thick lips and obese, holding a
water vessel in the right hand from below (in a manner typically
occurring in Eastern India). The left hand is damaged. Agni squats
with both legs close to the trunk, perhaps held in their position by a
yognpa;ga, on a high lotus throne. The most important detail, which
particularly betrays his dependence on his counterpart at MElap-
paiuvfir, is the pointed arch (top portion lost), which has a very sim-
ilar border of small fires. At M6lappaluvi1r,the pointed backplate is
still solid.
Probably, the two other images of Agni (figs. 289-292), which
exhibit his youthful, beardless form, may belong to a different local
idiom or dynastic affiliation (perhaps PSn$ya)419. Unfortunately,
they cannot be used for establishing a relative chronology.
At Tafijiivcr, iiiina (fig. 296) and Varuna (fig. 295) both are
seated in rlak~i1,zcl~~rahnzbapdrla (right leg pendent). Some of the
major characteristics are the very high jntiiilzukt~ra(crown of braid-
ed hair) for k n a and the simple, low coiffure for Varuna. Both
deities have long hair falling on their shoulders. iiiina's four arms
are all broken. His trunk is slightly bent to the leCt, which may point
to the fact that he originally carried a bulky object in his front arms
(vine?). He exhibits a beautiful broad smile.
To enumerate in short the characteristics of all the dikl)dlns at
Tafijgvcr: Indra has a hatchet (;m&) or thunderbolt, the vehicle is
not shown; Agni holds a waterpot on the right and probably origf-
nally a rosary, his vehicle being a goat; Yama carries a staff, veh'-
cle: buffalo; Nairrta is not depicted, antl~ropomorphic vehicle
(fig. 297); Varuqa holds a noose on the right, vehicle: mflknra
(fig- 298); Vjayu has a flag in his right hand, vehicle: antelope (fig.
299); the haloed Soma seems to hold two lotus buds in his h!ndsv
vehicle: lion (fig. 300);420 the attributes of the four-armed ISfina
image cannot be detected (arms broken), vehicle: most probably
bull.
The largely lost cloister of the Rfij&ndracb!iSvara temple at Gai'-
gaikdndacdlapuram (Tiruchcl~ira~palli dist.), which followed lhe
RsjarSjeivara temple after a period o[ about 20 years in the 1lth ten-
tury, is assumed to have contained dikpiila shrines after the ~ a f i j ~ " ~ "
model (PICHARD et al. 1994: fig. 40). The extant reinaills provide a
good basis for such a hypothesis (ibid.: phs. 363-365). A new and
most probably unique feature of the temple is, however, the Pres-
ence of dikpriln figures on the first upper storey of the vinziilzn (mala
temple). The images in most cases are not easy to recognize, as the
contours are very soft (ibid., starting from Agni and proceeding
clockwise: phs. 113, 117, 122421,129422, 131423, 135, 138)*The fig-
ures are executed in a rough, unrefined manner. The dikl~iilasare
two-arlned, except for ii8na, who is four-armed. Indra is missing.
The deities are generally shown in the correct direction, stxting
4'3 For a recently drawn, exact plan see PICIIARD et 01. 1994: pl. 4. Regardingthe
dik/~alatemples, "Therc is also a reference to these eight shrincs in an inscriptio"
dated in the third year of Riijcndra I. Fro111this illscription we get to know that for
of the eight Dikpiila shrines, kalaSas (finials) were gifted, and from the decipherable
~orliollsof the inscription, we are ablc to glean the other nanics, such as Nairutli,
and Varuna" (VENKATARAMAN 1985: I I 1-1 12).
4'4 111the case of Nairrta, this is missing (fig. 297).
4'5 For the s'ikhcrrus pertaining to Agni, Indra and iiiina see PICHARD f a ' 1994:
1lhs. 359-362 (in thc ordcr: south - Yalwa; south-east - Agni; north-east: I&ina) O' r
Indra sec below.
41h The image of Indra is recognizable by its attribute hatchet ((nrika;
1963:
23) or thunderbolt (vnjra). As on the jirivds of t]lc (/~vnklrlik~?s, both gods are without
any vehicle.
4" Ending in kir</l~ arches at the lowest ham level, integrated shrines are much
higher than their counterparts in tIieprikam wall. The third storey is rclativcl~ a:d
has been used to acco~n~nodate irnagcs of the enshrined dcilics shown in bold re"e f
(PIC~IARD ef 01. 1994: ph. 3.52). Thc portion corresponds to the grivii lcvcl of Ihe
frce-stantlillg Sikhnr~sonce more shows a figure (probably the same) in relief. In the
lower CenhC of the kli<lliabove, a third image appears 011 each side (later stucco).
4'R VBNKATARAMAN 1985: 112; HARI.~ 1963: 23. The lattcr author has
lloticed the discrepancy bctwcen the expected Shrya image and the surviving N'?rii'a
image: "A contemporary inscriplion in the gopum rccords the installation of an 'm'ge
or S C ~ Ybut ~ ,no aureole is shown here and Surya is usually represented with I~~"'''~'
not a rlilo//~ala,in each hand".
419 his assumption is based on S A N F ~ R D ' ~(1994: 58) ascriptioll of the
NfigeSvarasvii~lli temple at Kutnbakallam to the Pfqdya king Srimjra $rivallabha,
420 K.R. SRINIVASAN 1983: 240 erroneously mention? a mongoose as Soma's
vehicle.
421 Strangely, no object is to be seen in Nairrta's hands.
422 It is much morel~robablethat this figure represents Varya, as it is
and not four-an~icdlike t h e f i g ~ ~ r e ~ r o ~ino sthcpublication
cd (ibid.: ph. 1 2 . 5 ) . ~ h ~ s l f ~ ~ ~ l
bulbous object in its right hand resembles thc noose of thc large image at T ~ ~ J
(fig. 295).
423 This figure, which scems to carry a flag, is a bcttcr for VsYu than "le
image proposcd in the publicatio~i(ibid.: ph. 129). 1" any case, the identificatioll ""
Virabhadra is erroneous, as the attribute is apparently not a sword.
 with Agni on the (south-)eastern corner. Agni is the only seated one. ill. 21). A singular case with additional deities occurs on several
The image clearly betrays the Taiij3vGr model. The deities, vari- temples at Alainpur, where both SLirya and Candra have been
ously Set in Icar!za, salildntara or pratibkadra positions, alternate integrated into the already complete set of eight dikpiilas (e.g. ill.
with somewhat similarly portrayed but four-armed r ~ ~ d figures ra 11). Another unique case of substitution is found at Cittaudgadh
and other deities. (Kt~mbhaiyiimatemple, ill. 18), where five members of the dikpdla
Subsequently, standing or seated images of dikpdlas appear 011 group have been obviously replaced by an "kina type" image. The
Sop~lras,"~~ but they are also found on various levels of the vinziirzas. - generally earlier - group of four directional guardians appears in
On the latter, they are sometimes portrayed as riding their vehicles. most cases as a complete set, but even here, in a few cases, a sub-
Thus, at Garigaik6pdac6iapuram, the vinziilza contains riding stitution of some of its members can be observed (Bgdiini 3,
dikpdlas at the corners in the uppermost storey which is crowded mandapa, figs. 15-18; Alampur, Kumgra Brahmii, iiklzara, see
with sculptures. They have been assigned to the 17th Century section 2.2.1).
(PICHARD et al. 1994: ph. 202 [4th fake-storey]). The variable appearance of the dikpiila group is difficult to
A few parallels from Southeast Asia425 bear testimony to the explain. Its main reasons must lie in the somewhat intricate devel-
Popularity that the early seated images once enjoyed. In the case of opment that led to the representation of the complete group, but
a probable Nairrta from Central Java of the 9th-10th centuries probably also in the often variable theoretical concepts of direc-
(FONTEIN 1990: cat. S), parallels are quite striking. The god wears tional guardians or lolca~~dlas (see section 1). There may have been
similar ornaments and kam!zr/am~~kuta(basket-like crown) and sits also aesthetic reasons, arising from particular concepts and princi-
cross-legged on a low throne. Two ribbons, which are fastened to ples in art. Thus, particular conceptual types in art by their very
the girdle, fall down over the edge of the throne and embellish the character have the tendency to either promote the completeness of
plain seat. The face is serene and slightly smiling. He holds a curved the group (horizontal panels; friezes), while others (vertical pallels)
sword in the right and a horizontally placed shield in the left hand. for a long time preserve the situation of a partially represented, at
The backplate is pointed like that of the MelappaluvLir Agni, but its best filled-up group of directional guardians. However, there are
shape resembles that of a stylized lotus petal (i.e. squarish). The exceptions also: early ceiling panels inay have escaped the struc-
latter sculpture also has two sashes falling over the throne, however turally immanent demand of completeness of the group by substitu-
at the sides. In conclusion, the Nairrta image from Java adds to the tion of one of its members (figs. 15-18, Ksrttikeya replaces Yama),
sparse evidence, which may point to the fact that complete sets while on a probably more narrative panel a nearly conlplete group
a!tadikl)dlas existed in South India in the 9th-10th centuries. The appeared surprisingly early (fig. 21). The general rule expressed
question remains, whether separate vehicles were provided for them above, however, still remains valid.
as at T a i i j ~ vand
~ r probably once at Galigaik6pdac6!apuram. 111this The mail1 difference has been recognized between group and
colltext, it is significallt to note that the Ceylonese text MVS, which single representation. The latter term refers to the individually con-
describes the flame-haired, one-lleaded form of Agni (see above), ceived dikydla image and not to an isolated sculpture which does
not refer to the dikpiilas' vehicles. not pertain to a dikpiila scheme. In a few cases, however, this can-
In summing up the evidence, it should be mentioned that there llot be clarified (e.g. Bikkavolu, see section 2.3.2.1). In group rep-
is a remarkable absence of riding dilyiilas, which were so popular resentations, the degree of uniformity is understandibly higher. In
in the Deccan. The South Indian dikl,dla images are generally single representations, the images are much more apt to f~~nctioll as
shown as seated or standing and ollly much later appear riding their guardian of a particular quarter, however within certain structural
vehicles.426 It could be regarded as another evidence for limits. In the case of friezes, a directional orientation was only
"Separate developlnent in the depiction of directional guardians that achieved for friezes which are laterally arranged on particular archi-
On the introductioll of the theme and subsequently, Solna occurred tectural elements.
instead of Kubera427 as g~lardianof the north. This was, however, A chronologically early incomplete set should be regarded as
not the case on the Deccani ceiling panels of the 8th through 10th prin~ariIy incomplete or "incipient", while an incomplete set occur-
centuries.
It should be noted further, that in the RsjariijeBvara temple at
Tafij~~Lir, the layout of dikphla subshrines represellts an ideal 4" A plan of the distributio~lof deities 011 the four Cidambaratn goptrrvs has beell
prel~aredby L'HERNAULT 1987: pl. ic0. V. For the D3rPsuram gol7um see ib;c/.: pl, ice,
wit11 the guardians positioned in their respective directiolls in
IV, and for tlie Timvanaikka g o p ~ r asee ibid.: fig. 5. On each of the six gopums, the
"Curate manner. ~t seems as if this order was modified only reluc- complete group of dikl)cilas had bccn originally installed, o~lly at Tiruvanaikka
tantly later on. is reportcd to substitute Agni. It is significant that the correct sequence
clYajfin~ntirti"
of c/ikpd/as and their general orientation has always been observed, but no fixed rule
appears to have determined their placcmcnt on the structure.
425 From the 13th century, there is evidence from Candi Singhasari, East Java
2.4 Concepts of dikpiila Representation (VAN LOIIUIZEN-DE Leeuw 1955: pls. 1-6). Here, it is an important feature that figures
of veliicles have been integrated in the throne or socle. The most interesting sculpture
In Order to describe and the various modes of represell- is [hat of a probable guardian of the nadir, Ananta, on a large tortoise vehicle.
426 E.6 G.J.R. MEVISSEN 1994: pl. 4 (DBrBsuram, 12th century, set with Sollla as
tation of the directional guardians, the major ~011Ceptsof spatia1 guardian of the north).
"rangemm will be discussed and analyzed. Directional guardians 427 In the MVS, which has already been referrcd to, an image of Sorun is described
have been represented either as a complete group (e.g. Osisfi, Hari- 011 the north instead of Kubera in 2.78 (however, Kubera is lnentio~iedin 2.100 a),
hara temple 1, ill. 22) or as an incolnplete group, the latter cases Kubera replaced Sonia probably from the second half of the 12th century onwards.
being often difficult 10 explain. In a few cases of incomplete groups, M~vlsseN(1994: 419, n. 13) has shown that at Dirasuram, both sets with Solna alld
Kubera ss guardian of the north occur.
One may speak of a partly substituted or an arbitrarily "filled-u~" 428 In no case it can be dccided whether an intentional substitution of a particular
gr0U~,428 c ~ ~ n p r i sInembers
i~g other than the canonical set of c/ilik/,i/a took placc or whether tlie group had been arbitrarily fillcd up by another deity,
tional guardians. As deity, most often S f i r ~ ahas as the rest is not complete or has not been colllpletely preserved. Apart fro~nthe sub-
been employed (Satyavolu, ill. 13; ~ m r o lDiinebZbB
, temple, ill. 15; stituting deitics Sarya and So~ndCandrain the Manu Lokapila group (to be discussed),
no case of regular substitution occurs.
@rig, ill. 26), but rarely also Soma/Candra (OsiZii, SBrya temple ' 9
ring posterior to the introduction of the theme in the respective
region should be regarded as a secondary incomplete, reduced
group. The latter sometimes serve as a wisual reference to the
theme.429 In every case, however, it should be considered whether
the deities of an incomplete set were intentionally chosen and what
may have been the criteria.
An interesting case, which almost forms a category of its own,
is the group of "Manu Lokapiilas" (see section 2.4.1.3) which
consists of Indra, Agni, Yama, Varuna, VByu, Kubera, Sbrya and
SomdCandra, with the order differing. It represents a constant, but
incomplete (or partly substituted) set of dikpcilas. Its popularity
seems to cease with the introduction of the canonical dikpiila group
in almost all the known contexts in the 10th century.
2.4.1 Group Representation
In group representation, a great variety of concepts can be observed.
The largest category are the vertical reliefs, which- to some degree
- all have a narrative aspect. Horizontally placed reliefs generally
lack a narrative aspect, as they were not primarily conceived for
contemplation. The horizontal reliefs have a no re diagrammatic,
mandala-like character, and consequently, the dikpdla set is gener-
ally shown complete and arranged in a somewhat uniform way. The
little known foundation tablets also belong to the category of hori-
zontal reliefs. They represent the ritual, symbolic aspect in its purest
form, as they are are hidden below the temple structure.
The category of friezes consists of horizontal alignments of
directional guardians, which are not directly placed in any wider
context, narrative or other. They are either arranged on one plane or
form the decoration of an architectural ekment (e.g. pillar or pitha
- ~latformfor offerings). Friezes are designed to portray the totali-
ty of a particular group. Unfortunately, only few dik~Glafriezes pre-
date the 9thcenturY9 when the group had been firmly established in
art. However, alignments of f~urdire~ti~nalguardians seemto have
existed already in the Kur2na period (see section 2.1.4). The order
of deities generally follows the directional sequence, starting with
Indra for the eastern direction.
Friezes which are positioned above or around a doorway will
be treated separately.
2.4.1 .l Vertical Panels
When attempting to portray a spatial arrangement of deities on a flat
surface, it is clear that only a certain section of the directional order
will be visible. At the same time, the surface itself will be "orient-
ed" in the true sense of the word. This is what takes place on the ear-
ly monumental panels at Elephanta, where Indra and Varuna flank
the central deity in accordance with their respective regions (see
section 2.2.1). Panels containing other dikpiilas in the same cave do
not completely follow these rules, but they reveal a fact, which is a
very important one: Indra and Varuna at Elephanta are not meant to
represent the totality of the directional guardians, but only a partic-
ular, relevant section of the group.
The same intention of an absolute placement of dikpclas is
found on very few representations outside Elephanta (e.g. EllorH 16,
figs. 280-282). It is somewhat surprising that this collceptual type
did not become more popular.
There are three other forms of incorporation of the directional
guardians on - mostly narrative - relief panels: as a bunch or troop,
more or less aligned, and as a proper frieze. The central themes
of the reliefs are generally NateSa, Siva ~ a l ~ L ~ a s u n d a r a / ~ a i v g ~ ~ ~ ~ -
mnrti, AnantaSayana Visnu or ViSvarbpa Visnu. However, there
seems to be hardly any context where members of the d i k ~ d groupa
are absolutely essential, except perhaps for the ViSvampa theme A
theme, which has been apt for dilcp,dla only during
a very early phase, is the Churning of the ocean (Pawiyii torana'
fig. 9; BLdSmi 3430). 1, this context, the group consists of four
regents at the most. From a rather late period (10th-l lth ce11turies)?
a Mah2vira panel from Palma/Bengal uniquely includes the
a.yJadikpiilnslined up vertically on both sides of the central figure9
riding their vehicles (G.J.R. MEVISSEN 1998).
Some other themes, Mahisjsuramardini (fig. 279) and
YogiSvara (figs. 280-282), incorporated llikpdlns in the 8th
ry, but no visual tradition seems to have existed before or after this'
A similar case is that of a fine relief of a reclining mother breast-
feeding her child ("kcna-janma") including lively depicted riding
Guardians in the upper register in two groups, showing
aligned from left to right, colnes from Hilig15jgadh, C. lothe1 Ith
centuries431. In some cases, the dik@lns appear amoiig other celes-
tial~,often some of the main gods, but there are also a number of
exceptions.
Characteristics of the inclusion in the four most common themes
will be briefly mentioned before discussing then, With
NafeSa, dilqp81as are often combined, but not in a compulsorY
The group is already nearly complete on early NafeSa pane1 at
Ellor3 29, c. late 6th century (fig. 21). The of NateSa
l
with the directional guardians was shifted to the ceilings in the Dec- l
l
can from the 8th century onwards (see below), but vertical panels O f I
this theme were again carved in ~~~t~~~India in the 10th century
and later. Regarding the theme of the wedding of $iva and pgrvatl
(Siva Vaiv$hikamfirti), it is remarkable find that for a long time,
Nairrta and ISina remained excluded. In the 8th century, SfirYa and
SomdCvndra completed the group to form the Manu ~ o k a p l a ~ .
while the piece from ~ t ~ 10th h , century, incllldes the ar;adikpalas
and thus demonstrates that the depiction of the Manu ~ o k a p i i l a ~ i
probably formed only an intermediate stage in the development. The I
guardians appear on icons with AnantaSayana Visnu in a few
instances in the period under discussion, mostly showing them seat-
ed in a line, while one example includes a peculiar battle scene.
Finally, there is the incorporation of directional guardians int0 rep-
resentations of ViSvari~pa.Here, the dikplas are often difficult
trace and to identify. In early examples, the group was probably
never shown completely.
- A complete set of as[adikpiilas in this
?
text developed in the 8th-9th centuries.
This brief outlook reveals the remarkable fact that the
tionship" of the rlikpiila group wit11 a particular theme has been
variously conceived by the artists.
429 such references are found on panels representing holy places. One example
a so-called~ra~~gnpa!!a (N.P. Josrrr 1991: plate), where in the upper right corner,
is depicted seated, and just below, V q u with his inflated an
is seated next
unidentifiable deity. A few other clikpirlas might be recognized in the unspecified 'l&-
ures in the upper right corner, but in any case, the group must have remained incorn-
~lete.The piece is assigned to the 9th-10th centuries. o n a stele from Tewar ne"r
Tri~uri/M.P.,C. I lth-12th centuries, only the dikpdlns lndra and Agni are shown 'ld-
ing (heir vehicles on the left side of the second lowest tier (v.S. P~THAK/S.K. ~u~~~~~~
1991: pl. 17).
43U BANERII 1928: ~ 1XXIII
. c: the relief shows the battle between rlevas and
as after the process of churning.
431 GUITA1979: pl. 2211; Central Museum Indore, acc. no. 157. The ~ u ~ r ~ y ~
are almost thrce-dimensionally conceived alld shown in profile. The d i k , ~ ~ ~S'ta sIn
riding position, with their legs touching the ground, as their vehicles are only squ"'ing'
 From an early period onwards, the depiction of the dance of Siva The complete group of directional guardians was somewhat reluc-
included a representation of directional guardians, though the as- tantly included in this theme. The starting point is about the same
Sociation was not an inseparable one. On the earliest panel with in time and space, but already markedly different in quality and
dilcpdns, located in the western annex of Elephants 1 (fig. 20), only content. The important early panel at Ellorii 29435 includes Yama,
two of them are shown, directly flanking the dancer. On the proper VByu, Agni and Kubera on its proper right, while Indra and Varuna
left is Indra, and on the proper right there is a goat or ram rider, appear juxtaposed on the Proper left. Visnu, several celestial cou-
either Kubera or Agni. The placement should be viewed in the ples and one unidentifiable ascetic in the proper left corner corn-
context of two other panels in the same cave (ArdhaniiriSvara and plete the animated scenery in the upper sphere. As on the NateSa
Gangiidharamiirti, see section 2.2.1), where Indra and Vamna act as panel, the placement of the regents does not follow their spatial
"~osmiccoordinates". In this case, however, the choice of these arrangement in any way. The most conspicuous fact, which is the
Particular dikpiilns is difficult to understand. One may speculate that absence of both Nairrta and f i ~ n ahas , been pointed out above.
the mythical place of the enactment of the dance is located in the the light of the later development, this is not accidental. Although
north-eastern direction, which and Kubera represent, rather the absence of both deities can be explained in some way or the
than on the ~ o u t h - ~where , panel has been carved in the cave other (negative impact in the case of Naiqta, identity with siva in
~ ~ tthis
temple. Interestingly, the NafeSa relief at Ellor%,Cave 21 (fig. 22), the case of fiBna), the later concept including the Manu Lokapalas
likewise includes Indra and - most probably - Kubera as the only is clearly foreshadowed.
dikpdas. The third one of the early NafeSas is extant at Ellor% The rich production of the 8th to 10th centuries in North India,
Cave 14.432Typologically, it ranges after the other two, as the group whe11 featuring more than just one or two "referential" dikpdlns (see
of four dikp8lns on the proper right of seems to be complete. below), seems to be unanimous in including the group here referred
Another NateSa panel probably dates before 600 A.D. is to as Manu Lokap2las (Indra, Agni, Yama, Varuna, Viiyu, Kubera,
located in Cave 29 at Ellor2 (fig. 21). Here, the unique situation SLirya and Soma). Four complex panels from the Mathurii region
arises that eight deities (seven of theln riding a vehicle) appear on the and from Kanauj belong to this type43G(figs. 301-305). The popu-
upper level, along with an additional figure holding a staff, probably larity of the Manu Lokapaas appears to be restricted to the central
Dandapiil!i. Piana is the only member of the aStadikpalas which is and eastern parts of Northern India. However, Nairrta and f&na are
missing. It is a noteworthy fact that the dikpalas are not arranged in also not included in the lokapaln group on the Kal~iinasundar~ pan-
a sequence relating to their spatial order. At E]lorj 29, the intention el in Cave 15 at E l l ~ r i iAS . ~O ~n~the earlier panel at Ellor2 29, the
to show the totality of cosmic guardians is clearly expressed. regents are divided in two groups (one on each side), but here, they
There is no extant sculpture of the Dancing God and the direc- alone dominate the celestial region.
tional guardians from the 7th century, In the 8th century, such This domination can also be observed on four loose panels of
images appeared on temple ceilings in the Deccan. This is an impor- about the 8th through 10th centuries from North-Central India. All
tant step, demonstrating that a conscious effort was made to find an panels have the comlnon feature that the sphere of clouds has been
"equate aesthetic expression and arrangement for a cosmic theme. indicated, on which or above which the guardians appear. On what
In the 10th centliry, the tlleme was taken up on an even grander can be regarded as one of the most impressive Vaiv2hikamfirti
S,cale in Bengal. ~h~ first one of these relief panels comes from reliefs, the Kanauj panel (fig. Sol), the narrow, undulating band of
Sankarbandha/Bangladesh and is housed at the Dacca M~seum.~33 clouds of the deep-cut relief leaves two solnewhat cave-like hollow
The a;[adikpalns are placed singly above the aureole of the ten- spaces in which Siva and piirvati stand. There is much space above,
""ed Siva in the shape of an arch, riding their vehicles. The vehi- which the lively conceived loka~)dlasdominate, along with Cavesa
cles are the usual ones, but Kubera has an anthropolnorphi~vehicle, and Kirttike~a. On typological grounds, this may be identified as
and Varuna is - remarlcably - portrayed on a goose The second the earliest of the four reliefs from Kanauj. Sarya and Soma/Calldra
piece is kept in the Museurn.434 NateSa is twelve-armed in are portrayed standing at the sides, just below the level of stylized
this case and again dances on his bull. The shapes are Somewhat clouds.
clumsy in this case, limbs are much more angular and appear The reliefs from Klmiin (figs. 302, 303), Miinasigalig$ (fig.
less elegant. The nstadjkPalas flank the god in two vertical rows and 3 0 4 , and Mathua (fig. 305) all Portray the divine couple in front of
are again provided'with separate On the proper right side, clouds, which consists of four superi~nposedlayers in both the latter
starting from the top, there are rndra, Nairrta, Varuna and Kubera, cases. As the lokapala group has been restricted to the sectioll above
while on the proper left yams stands at the top (even higher than the clouds, they appear somewhat crowded in the allotted space.
Indra), followed by ~ ~Teana~and Vnyu.
i , On the first panel, the Regarding the emphasis on the central wedding couple, however,
regents are portrayed in motion, while on the second, they are stat- this spatial division is more satisfying than that of the Kanauj ver-
ically conceived. ~h~ only difference in tile vehicles lies in Varuna's sion. A hypothetical typological-~hr~n~l~gical sequence is pro-
fOur-legged mnkara, in both cases holds a billowing scarf. The posed here, in which Kamall comes after Kanauj, as the lokapfilns
Second panel belongs to C. 10th-l lth centuries.
Evaluating all available facts, it becomes a p p ~ e n thatt there 432 SIVARAMAMURTI 1974: 173, fig. 10.
was a tendency in the Deccan and lateralso in Eastern India to make 473 SIVARAMAMURTI 1974: fig. 171. Na!eSa wields a sword in Ilis rear riglit hand,
the cosmic dimension of Nateia's dance visible. It still has be which is a very rare feature.
individually clarified whether there was a more narrative or a more 434 GASTON 1982: p]. 38.
it was intended to 435 VARMA 1988: phs. 144-148.
symbolic aspect behind this concept, i.e. 4% In an important article, M.M. MUKMOPADRYAY 1967-68 has discussed four of
show that the dikpalas were present to witness his dance, as a cos- the reliefs treated below: Kanauj (p]. I, fig. 1); Mathurii (pl. 11, fig. 3); KBlllin
11,
fralne denoting the space of the whole universe, Or fig. 9, not mentioned in the text); Etah (p]. 111, fig. 5). Regarding the identification of
the fact that His arms are the directions and that Siva embodies the dikpiilns, we disagrce with the author in a few cases.
437 BERKSON 1992: 298-99.
a!!adikpdlns in himself.

The theme of Visnu reclining in his yogic sleep has much less fre-
quently been combined with directional guardians. The few extant
cases are nevertheless quite instructive regarding the treatment of
the rlikpdln group in different contexts.
It is significant to observe that this theme incorporated some of
the dikpiilas or proto-dikp8lns at a very early stage: at Deogadh,
Indra, Varuga and Viiyu appear above Visnu, along with Brahmii
and ~iv,?ll&jna?445
Varuna and Viiyu are found among the members of an incom-
plete group on the 8th-9th century-panel from Baran.44"t is one of
the rare cases in which the clilcpdas, here only five, are shown
ellgaged in a fight, which is depicted in the upper tier of the panel.
The dilc~dlascol~leriding from the left, while the foot soldiers occu-
PY the right half of the panel. The former are (from the left): ? (dam-
aged), Viiyu on a deer, Indra without a vehiclc (lower portion hid-
den behind Visl!uYs cakm), Varuna on lnnknrn and Yama 011 the
buffalo. The latter three wield their attributes in the right hand, while
Vgyu holds only his inflated cloth, which can hardly serve as a
weapon. It relnaills unknown, which criteria led to the choice of
these particular dik13Elns.447 The dynamic movement of the group
creates the impression that they come rushing to defend the sleep-
ing Visnu.
The next panel from Baroli has been assigned to the second
quarter of the 10th century.438 Like the former piece, it belo~lgsto
eastern Rajasthan, and it shows a remarkable continuity in its layout.
In this case, the upper tier is separated from the lower part by a plain
ridge. It is filled by a dn&vatdra set on the left side, which occupies
more than half of the length, and a row of seven seated dikl2Elas on
the right side. A very rare case of lnisplan~lingmay have led to the
fact that there was no space left on the right to accommodate I65na.
Most likely, a slnall figure just below the upper tier, squeezed in
between the leaf capital on the right and an attendant figure, repre-
sents this "missingn guardian. He holds a trident in his left hand.449
The seven deities above are clearly characterized as (from the left)
Indra, Agni, Yama, Nairyta, Varu~a,Viiyu, and Kubera. They hold
their common, mostly promi~lentlyrendered attributes in the right
hand and in the left they all appear to carry a waterpot.
A panel of Visqll AnantaSayana now kept in the Museum fiir
Indische Kunst Berlin (fig. 327), which can be, on the basis of its
~haracteristicsshared with with the Baran and Baroli panels, respec-
tively, typologically placed between both these, likewise includes
an aas~arlikpiilngroup. A closely related piece at Dhamniir, Dist.
Mandasor, U.P. (MEISTBR1991: pl. 724), makes it probable that the
Berlill piece originates from western Madhya Pradesh or southeast-
ern Rajasthan (I<ota region). As for the Dharnniir AnantaSayana, a
date of c. mid-9th century is suggested. The Berlin piece shows a
seated dikpdln group aliglled in its upper register, on the left side,
W i n with a solnewhat displaced ISiina, who - in all probability -
had been the first in the row (now colnpletely destroyed), followed
Indra, Agni, Yams, Nairrta, Varuna, Viiyu, and Kubera.
The above evidellce shows that the a~fndikpdlaset makes its
aPPearance in the context of the Reclining Visnu during c. 9th cm-
tllr~,450 although visual references to the group ~ ~ ~ u rearlier
r e d in
this context. It may be noted that h a in is two cases irregularly
placed.
The connection of the dikpElns with Visnu in his multiple, cosmic
is irregular and difficult to understand. In the evolved
Viivariipa icon, it seems that one or more members of the group of
directional guardians (but probably not always in this capacity) have
always been incorporated in the divine assembly. The works of
T.S. MAXWELL (particularly his 1988 monograph) provide clues to
the development of the theme which have been helpful in the pres-
ent context.
From the so-called Classical Phase of Viivariipa representation
(ibid.: 113-143), the Gupta Period, there is a fragmented Viivaropa
image from Bhaiikari, Dist. Aligadh/U.P., which includes a partly
visible figure of Agni in its lower right portion (ibid.: pls. 49, 52,
54). The identification of a figure, resembling the Buddha, as Agni
is indicated by the flame nimbus and, particularly, by the similarity
with a late Kusiina Agni image from the Mathura district (fig. 7).
Above Agni, there is a personage which holds a rod with a bulbous
top in the right hand. It could be Yama (ibid.: pl. 52) who later
becomes one of the most frequently included members of the
dikpdln group in this context.
Another early ViBvarUpa image from Viiriinasi,451 dated to the
4th century, preserves a fragmented figure of Viiyu. Only the head,
encircled by a scarf, is still extant below. Viiyu belongs to the second
vertical row on the left side of the central portion, which consists of
superimposed forms of Visl~u.Significantly, the Wind God appears
in generally similar positions on at least two later pieces.
One of these later reliefs is an elaborate ViSvariipa icon, locat-
ed irz situ at ViSriimaghiit, ~iimaliiji.According to SCHASTOK (1985:
p]. XX, fig. 34), it should be dated to c. 535-40 A.D. MAXWELL
(1988: 148) has attempted to identify all the figures contained in its
large nimbus. He recognizes two members of the dikpdln group:
Varuna (P, second figure from the upper central one on the left) and
Agni (Q, symmetrically placed on the right side). The identificatioll
of the latter figure as Agni remains doubtfi~l,as there are hardly any
flames visible. In the case of the supposed Varuna, who carries a
short noose or loop in his right hand, doubts also arise, as he is
provided with a serpent hood. The figure below the probable
Varuna is perhaps Yarna or Nairrta (L), and beneath this is certain-
ly Vayu (H) with his inflated cloth. The probable Yaina or Nairrta
holds a long object in his right hand,452 which can be interpreted
as a mace, sword or a skull-topped staff (klzntvdriga). He seems
to wear his hair in irdzvnkein fashion. There is some probability
that two more lokn~~dlns, Indra (M) and Kubera (K), are represent-
ed inside the populated nimbus and together with Siirya and
CandrdSoma, who flank the uppermost figure, form the Manu
445 Cf. section 2.1.3. It cannot be coinpletely clarified in which capacity the gods
Indra, Varuna, h a , and Viyu were portrayed. VANKoou (1985: 698) argues that [hcy
represent different classes of deities.
446 R. PARIMOO 1983: fig. 29. It is presently housed at tlie Archaeological Musculn
Kota.
447 Neither the usual sequence of dikpdlfls call be recognized, nor is there any
for~nationof pairs as is gelierally the case with Manu Lokapilas (see ill. 31).Tlicclioice
illay have been made on tlie basis of aesthetic criteria.
44% MASON1993: 266-67, no. 72. Another illustration is published in PARIMOO
1983: fig. 27.
449 If the identification of the two-armed figure as is correct, another possi-
ble reason for its shifting could be considered. There may have been a conflict between
the Central Indian custom ofportrayi~igibfina's trident on tlie proper left side of the god
and thc required uniformity of the upper group ofdikpdlns, which all hold their pron~i-
nent attributes in their right hands.
450 Later sculptures, when i~icludingthe directional guardians, have an o!rfldjk-
p810 set, e.g. PAIUMOO 1983: fig. 30, above, right side.
451 T.K. BISWASIB. JHA 1985: 41; pl. X, 16.
452 In his drawing, MAXWELL (1988: 148) has misinterpre[ed the sword as the
figure's raised arm, eildiiig in a fist. This mistake beconies obvious when coinparing
different photographs of the sculpture.
L o l t a p l l a ~ . ~However,
~3 this is impossible to prove. The selection
of directional guardians for the Viivarnpa images often differs, and
there is no parallel for the depiction of Manu Lokapiilas in this con-
text. In this case, after discarding all the doubtful identifications,
there remains only VByu, who is securely identifiable on the basis
of his distinctive attribute.
A piece of the 7th-8th centuries from Deogadh1U.P (MAXWELL
1988: pl. 63) shows a significant step towards the inclusion of the
vehicles. Interestingly, this sculpture abounds with the depiction of
animals, particularly cows. MAXWELL(ibid.: 234) has identified
three dikpdlas in the upper part of the panel (D, E, I;?, seated astride
their vehicles. Indra, Agni and Yama can be recognized. Agni, on
the right side, is somewhat doubtful, as the god usually does not
carry a rod-shaped attribute. Below the probable Agni, a human
head appears above what looks like the head of a horned animal.
This figure, which is not referred to by MAXWELL, may represent
another dikpdla.
Among the figures from the 8th-9th centuries in India, an aver-
age number of about five dikpdlas is included. They are often
depicted on both sides of the central V i W . In some cases, one side
is preferred. ~ l t h o u g hat about the same time, the earliest extant
ViSvarilpa image with a full set of directional guardians occurs
(figs. 309,310), the selection of dikpdas still had not become stan-
dardized.
A fragment of c. 8th century (MAXWELL 1988: pl. 65) shows the
group of eleven rudras, above which (from the left) ~ a m and a 1ndra
ride their vehicles. These gods hold their usual attributes in the right
hand. They are sensuously portrayed. AS an informative detail,
Yama's reins are shown leading through the nostrils of the buffalo.
Much space has been provided for the dik~dlafigures. One can only
speculate whether Varuna and Kubera had completed the group.
Both Kubera and Varuna are very rare members of the Viivarfipa
assemblage. However, in Nepal there is a well-preserved instance
of ai coturlokapdla group (four members) on the ViSvar~pastele
from Changu NSrgyana, c. 8th-gth centuries (DEVA1984: pls. 27-
29). o n the left side, beside Visnu's sword, four gods are arranged
in two tiers, who are ncognizable by their distinctive attributes.
Their coiffures are also partly varying. Above left is Yama with his
rod, to the right Indra with his thunderbolt, below left is Varuca with
a stem-handled noose, and below right Kubera is characterized by a
bag. The Changu Njrjyana Vigvarfipa impressively demonstrates
that a reference to the Various components of the cosmos was
intended. This representation of the cosmic elements is further com-
pleted by the depiction of the four diggajas (directional elephants)
in the lower part of the panel.
One of the most important pieces of the later phase is a well-pre-
served ~ i i v a r f i p afrom Kanauj (MAXWELL 1988: pl. 69)454. There
are two riders to be identified as dikpdlas, in this case placed in the
lower part of the populated nimbus. VEyu has taken his "tradition-
al" place on the lower left side, with Yams or Naima below him
(their vehicles resemble each other in this period).
The next 8th century panel, the Stuttgart piece, also originates
India (fib 307-308). Here. the upper portion (left side)
depicts the family of Siva, while the right side showed probably five
of the dikpdlas. The latter all seem to be riding their vehicles, ~t the
very top, MAXWELL (1994: 103) identifies Indra on his elephant fol-
lowed by Agni on his goat, the latter being almost completely dam-
aged. Indra is Yalna on his buffalo, wielding the
usual klzatvdriga staff in his left hand. MAXWELL believes that this
deity is Virfipiik!a on a camel. A stylized, somewhat caricature-like
rendering of the buffalo was7however, common in Rajasthan at this
time. There is a good example at Osijfi, Harihara temple no. 3.455
To the right of Yama is a figure riding a smaller animal, who holds
a damaged attribute, most probably a sword. It must be Nairyta*who'
in the Central Indian fashion, rides a quadruped. Below, between
Yama and N a i ~ t a ,a squatting personage is holding "
flattish cup at shoulder level. This is undoubtedly Kubera,
probably also carries a money bag and rides a vehicle, ram Or 'zora
(damaged).
Typologically, the Stuttgart Viivarfipa is likened to the Tumaln
piece (figs. 309,310), which it most probably chronologicall~pre-
cedes. The stele from is significant for its incl~lsion a
group of qgdik/>dlas, depicted with only one small i 1 ~ ~ g ~ ~ ~ ~ !
!&a is replaced by or shown in the form of Siva with his famlly'
Siva, Pgrvati and GaneSa are seated on the top left, while Kgrttikeya
is depicted slightly below. They are provided with a solid so:1e'
which probably symbolizes the KailHsa mountain. On the right 'lde
of the central Brahmg, a procession of dikl~alasis shown
towards the centre. They are (from top to bottom): Indra on the ele-
phant AirBvata, Agni on a goat, Yalna on a buffalo, Nairrta On a
quadruped, Varuna on a goose, Vgyu on horse?, and Kubera On an
anthropomorphic figure. The arrangement is characterized by "
peculiar asymmetry, as no counterpart for the dik@a group appears
on the other side. Perhaps it was this asymlnetry which prevented
this visual concept from becoming a standard type.
There are three more Viivarfipa images which aPPeN be
later thall this one. Two images from Kanauj date c. 9th
(MAXWELL 1988: pis. 66,681. They resemble each other closely7but
even here, regarding the dikl,aks, certain differences are apparent'
The regents' position in the upper ponion is, however, identical'
Both reliefs belong to the type of ViSvarlpa icons with I d r a and
Yama on the right side, like the two pieces just described. Two of
the other pieces discussed above show lndra and Yama on the left
side (MAXWELL1988: pis. 63, 65). On the two later ViSvaraPa
reliefs from Kanauj, not all of the di,cpcins are provided with v~''"
cles, and tile representation is not clear. Kanauj no. 1
or Vjyu on a horse,
pi. 66)45~includes, from left to right:
Sorna? with waterpot? (1.1 and msary (r,), - central figures Indra -9
with the thunderbolt, seated above an elepllant, Agni with pot (I.)
and rosary (r.) and characterized by a beard, Yams or Nairrta stand-
ing with staffoi sword. Knnauj no. 2 has, in the same order: Kubera
on an ~ n ~ ~ r o p o m ofigure,
r p ~ ~~~~i
c on a ram, again characterized
by a beard, - central figures -, lndn squatting on an elepllant, yams
(k/la!v&iga), and Nairyja
on a buffalo, wielding a skull-topped
with shield (1.) and sword ~h~ latter is the only one who Is
depicted frontally.
Thus, on both pieces, Indra and - most probably - Nai'Tt"
appear on the right side, while the progralnme of the left side vanes'
The inclusion of Soma (and Surya) in one case is unique
and may have been inspired by the Siva vaivjhikamfifii
The last piece of interest is the c. 10th century ViSvarfiPa
SuhSnis, kept at the Archaeological Museum, Gwalior ( M A X ~ ~ ~ ~
1990: fig. 44). Here, the hosts of gods have been neatly arranged in
453 rhe probable lndn ( M ) wCarS a hig,, Crown l i b surja, md he may hold a thull-
derbolt (rather indistinct) in his right hand. H. is pot-bellied. The pmbable K1lbfn (*
is also obese. He is shown with loose hair crowned by a topknot. He holds afrult In "ls
right hand and a bag or bottle in his left hand.
451 A good illustration of this piece has been published by A. G o s W A M 1 1956:
16,
45s B H A ~ ~ A C I I A R Y1991:
Y A pl.~g.~hebufi~oishiglllystylizcd,wi~h~n~-s'laped
tail. Its long, slim neck is raised. The horns arc very short and curved. Yet, when
p"i% the animal with its counterparts at Osiiii, there rclmains no doubt that a buf'a'o
was intendcd to be shown here.
.,, ,, I TS, Muwcll for kindly providing me with rolo~lr
photogmplls of ,llc reliefs Kanallj I md 2.
 three tiers. A unique order has been chosen for the nstadikpiilns,
who occupy the middle tier. Two seated figures, which belong to the
group of twelve Gdityas above, frame the group on the left and on
the right side. The dikpalas are (from left to right): Varuna on a
goose, Viiyu on an antelope?, Kubera on an ailthropomorphic fig-
ure, k n a on a bull, - central figure -, Indra on the elephant, Agni
on a goat, Yama on a buffalo and Nairrta on the usual unidentifiable
cluadruped.457 All the dik~~nlas are shown moving towards the cen-
tre. The group starts with Indra as the first one on the right side and
ends with f i ~ n awho
, is the first one near the centre on the left side.
The guardians are apparently shown approaching the centre from
the stations of an imaginary circun~ambulatorycircle brndak-
sinGpntha).458 In this way, ,'$iv&igna7splace close to the centre is
also maintained. The Suhjni j ViSvarfipa is the only one of all the
Panels discussed so far, which presents a consistent arrangement of
nsfadilcpiil~s.
Summing up the evidence, it can be stated that until the loth
century, the liberty of the artist with regard to the dikpcila represen-
tation must have been The impression arises that a
picture puzzle was created, without, however, leaving the pat11 of
tradition established by earlier sculptures. The period of new,
experimental creations continued until the 10th century (SuhiiniB
Viivarfipa), when an aesthetic concept was developed which con-
vincingly accommodated the complete group. The actual introduc-
tion of the complete nspdikl~dlaset in the Viivarfipa theme, how-
ever, is marked by the Tumain stele (c. 8th-9th centuries).
2.4.1.2 Horizontal Panels
The main characteristic of the horizontal panels was, of Course, the
fact that the directional guardians could be ~ortrayedadequately,
i.e. positioned in their respective region. At the same time, there
Was less possibility for variation, and the application of different
aesthetic concepts remained significantly limited.
The category of horizontal panels has two sub-groups, the first
0% however, being almost completely unknown. These are the
designs applied on foundation bricks or slabs, which represent a
cosmic diagram fig. 1, see section 2.1.2). AS such diagrams
were hidden below the structure, the design did not need to be dab-
Orate. However, more elaborate designs inay also have been
employed. Theoretically, a mirror-image of a ceiling panel could
served as a foundation stone.
The ceiling panels depicting dikpdlns are more or less confined
to the Deccan region. As they are mostly preserved and perceived
they have been discussed at length in the sections dealing
with this region (2.2.1, 2.3.1.1, 2.3.2.1). The general line of devel-
Opment, which can be inferred froin the extant remains, will be
sketched below.
Ceiling Pnizels
The know11 starting point of this tradition are the ceilings of Cave
temple 3 at Biidjmi, dated *.D. 578. As has been described earlier
(figs. 13, 15-18) are found
(secti011 2.2.1), sets of four loka~~iilns
here side by side with an early set of eight directional guardians
(fig. 10).
It may be hypothesized that new phases in the development of
the astadikpiiln thelne on ceiling panels were marked by panels,
which to a certain extent incorporated a narrative or dramatic aspect.
The intention to incorporate such an aspect could have prolnoted the
exPerilnental search for new aesthetic concepts.
No predecessors are known for the first stage in the develop-
ment of a~tadikpdlnceilings, the caturlokapaln panels at Biidiimi
(ill. 1, figs. 13-18). It is a remarkable fact that, although an incom-
plete set of a~;adi/cpdlas occurred in the same cave temple (fig. lo),
caturlokapliln groups still decorated the ceiling panels at BBdBmi,
Niigara!, and Mahiikiita (ills. 6-8; figs. 46-48) more than a century
later. It could be hypothesized that the concept of the set of eight
was not developed enough to be employed or developed further by
the later sculptors of BiidLmi and that both this early set and the later
caturloknpdla set at Bjdjini, Jambuli~igeSvaratemple, contain a
dramatic element, which indicates a fresh impetus. Regarding the
proto-n;fadikp6la set (fig. 101, there are a few striking facts which
suggest an underlying narrative or dramatic aspect: when compared
with the similarly constructed Brahmii medallion (fig. 13), it be-
comes obvious that the central Visnu is exceptionally accompanied
by an attendant couple and is allotted more space than the central
Brahm2 in his panel; all surrounding figures are arranged in such a
way that they can be perceived without turning the head - with the
exception of Yama, who is turned at a right angle -, and those,
which are shown in motion, are heading towards the centre. This dif-
fers considerably from the Brahmii panel, where the bottom portions
of all peripheral figures point towards the outer frame and the move-
ment is basically circular. With its centred, dramatic impact and the
individually conceived layout of the relief which extols Visnu, it did
not serve as a model for the time being. It seems that in the 8th cen-
tury, one of the earliest a~fadikpdlareliefs with an angular layout
takes up this concept - perhaps consciously - again (ill. 8), signifi-
cantly portraying Anantaiayana Visnu in its centre.
The Biidiimi Jarnbulingeivara panel (fig. 46) contains a unique
dramatic element. Again, there is an aspect of centripetal move-
ment, but in this case there are two dramatic scenes, which cannot
be perceived froin the same position. In the corners of the frame of
a fish-spoked wheel, dikpiilas are depicted riding towards their
counterpart, the guardian of the opposite direction. It cannot be
decided whether there is a symbolic meaning other than the usual
parade of dikpdlns, who rush towards the centre (the enshrined
deity) to pay their reverence. One would not expect the dikpdlas to
fight among themselves,459 but the scenic arrangement seems to
imply this. The ceiling panel of the Jambuli~igeivaratemple proba-
bly recedes or at least marks the occurrence of two other catur-
loka1~alnpanels, at NBgara! (fig. 47) and at MahIkfita (fig. 48), all
of C. late 7th century, where a rectangular grid of nine coinpartme~lts
has been employed for the first time. The ancestry of the two latter
reliefs, however, lies in the two Brahmii panels at BBdiimi 3 (figs.
13-18). While in all of these types of grids it would have been pas-
sible to depict eight dikl~hlas,this was not done. The basically cir-
cular movement of the regents was also retained.
The first half of the 8th century was obviously an experimental
phase, which finally led to the rectangular asfadikpiiln grids of the
457 MAXWELL (1990: 136) did not recognize the deities as a~!adikl~dlas,but s u e
gested to identify them as as[anlntrkas (eight mother goddesses).
458 Tlie arrangement is identical with the order of agadrkpdlas on the Hen13vati
lintel frieze (SOUNDARA RAJAN1986: pl. 631), and it is almost identical, but for the
inversion of Agni and Yama, with their arrangement on the Siva YogiSvara panel at
Ellor3 I6 (figs. 280-2821. In all lhese cases, the quite successful attempt has been
to project a thrce-diniensional configuration into the two-dimensional picture. A simi-
lar, but much less satisfactory attempt is recognizable on the Siva Vaiv3liikam~rtirelief
from Etah (fig. 306). In this case, the group on the right side is not shown heading
towards the centre like the group on tl~eleft side. The sequence of the fonner group also
does not reflect the intention to porlray a basically circular arrangement of dikpd/as.
459 IF S L I C an
~ i~~ipression
is intended, there could be an allusion to powerful killgs
fighting for suprelnacy in thc world.
later phase (c. 751 A.D.-1000 A.D.).Three different starting-points conceived relief at Bhavaniisi Sangam is the latest of the particular-
can be discerned. The first panel, which definitely has some narra- ly inventive ceiling panels of the Deccan. It can be regarded as the
tive, dramatic elements, is the eight-armed NateSa Siva accompa- true perfection of a centripetal order as it was basically already
nied by Piirvati, his bull, Brahmii, Visnu and the four loknpiilas460 intended on the Visnu panel at Biidiimi, Cave 3 (ill. 10). Here, how-
in the Mallikiirjuna temple at Paffadakal (fig. 326).46' The appear- ever, the extolled deity was Siva, who in the meantime had become
ance of the directional guardians is very modest in this case. Be- the deity most often associated with the astadikpiila group.
cause of their small size, they are perhaps hardly perceivable from An interesting later example (Bhubaneswar, P~~aniiSini templet
below. Yet, this relief represents the most important predecessor of c. 1 lth century)"3 shows two of the dikpiilas, Agni and join-
the very popular NateSa-cum-a;tadikpdla ceilings in the Deccan ing Siva in his dance. They appear at his sides, imitating his dance
which were widely distributed in the 9th-10th centuries. Its datemay posture. This partial representation of the group can be regarded
be as early as c. 710 A.D (see section 2.3.1.1). a reference to the complete set, or it probably portrays A ~ n and i
The second innovative representation of the first half of the 8th V&y~l in their capacity as rulers over particular elements, viz,
century is carved on the vertical portions of beams supporting the and air.
ceiling in the Huccappayya temple at Aihole (ill. 7; figs. 49-54).
This unique arrangement does not belong to the category of hori-
zontal panels, but it is regarded as a three-dimensional frieze. Unfor- 2.4.1.3 Friezes
tunately, it is not known which deity was depicted on the central
ceiling. BUCAANAN (1985: 391) regards Brahmii as the most proba- The friezes generally represent the dikpiilns in a formal alignment.
ble candidate for this position, which would have meant a "conser- Although a processiol~or a linear battle forlnation could be a
vative" combination of the directional guardians with BrahmZ. rewarding motif, a narrative context can be assumed in very few
The third example of the experimental stage is perhaps the ear- cases. Nevertheless, the occurring variety is quite surprising, indi-
liest ceiling panel depicting the common astadikpiila group. It is cating that fancy and invention was employed also in this catego-
located in the eastern mandapn of the PEpiinZtha temple, Pafladakal ry. Most of the pieces to be discussed are loose relief panels.
(ill. 8; figs. 55-61). The uneven size of the outer sub-panels may Among the cases in which the original position is lcnown, a plxe-
express the intention to emphasize the guardians of the four cardi- ment near the door occurs particularly often. Only in two cases, a
nal directions. The combination of Anantaiayana with the dikpiilas position on the wall other than the doorframe is reported. Friezes
has a ~arallelin the above-mentioned 8th century panel from Baran, in the door area are treated once Inore as are also group
Rajasthan.462 Its origin lies in the ancient concept of Visnu abiding representations in adjacent niches and three-dimen~i~nally
at the centre of the cosmos. What is highly noteworthy, however, is arranged friezes.
the fact that in the Piipaniitha temple, with the exception of Viiyu, Unlike the categories discussed so far, that of the friezes is 4~~~~
all of the dilcpiilns are shown moving towards the centre or the diverse, which is not surprising when considering the large chrono-
central axis (ill. 10). This feature is surely reminiscent of the Visnu logical span, almost one millenium, and the vast space of their dis-
ceiling of BBd&mi 3. The Piipaniitha panel is closely followed in tribution, the modern Indian state. The number of roughly 20 extant
time by the a$dikpdla ceiling in the KiiSiviSvaniitha temple (ill. 24; friezes is rather small llowever, so that a further subdivision would
fig. 129), both ~ a n e l dating
s c, shortly before and after the middle hardly yield Inore satisfactory results, Regarding introduction of
of the 8th century, respectively. Here, the centre is formed by a the ngaclikpiiln theme, the friezes solnetimes provide importa11t
Somaskanda group, which is a rare motif in this context. information, particularly, when they come from regions where other
1n the following ~ e r i o d the
, evenly spaced grid of nine com- conceptual types of dilc,iilas are rare Kosaltl; Bihar).
partments was firmly established. Although some variation con- The basic facts about the friezes are already revealing. The hey-
cerning the central motif occurred, the most common theme was the day of the n;tacli/cpcln friezes occurred ill tile 8th-gth centuries.
eight-armed NaieSa (see ill. 23). Even in the latter case, consider- Only few examples survive from an earlier date, ~~~i~ differences
able variation was expressed in a number of details of the individ- exist regarding the sequence, the relldering of the figures (standing,
ual rendering, and it be of a completely seated, with vehiclelsymbol), and the fact whether there is a S p a -
stereotyped representation of the a;!adik~Gla theme. A rare exam- rating device betweell the lnembers of the group. he friezes can be
ple of the incorporation of a dramatic element in the later period divided in two groups: firstly, showing a sequence of deities refer-
'is found at ~ h a v a n l s iSafigam, Rapall SafigameSvara temple, c. ring to the directions they preside over, and those with a different,
second quarter of the 9th century (fig. 138). It could well stand at to
"illogical" sequence, This criteriuln is of no value with
the beginning of the specific tradition of the Alampur region, which
the Manu Lokvpjlrs (see below; ill. 31), as the directional assign-
is characterized the absence ridges (e.g. figs. merit is llnknown in this case. The rules determining their seq~lence
141-144)+ However, in the absence any reliable dates for the
are only partly understandable. ~~~~t from the frequelltly occ~~rring
north-eastern group of ceiling panels, it is impossible to establish a Manu Lokap:lia set, a few
groups with substituted Inembers
chronological sequence. The Bhavaniisi Safigam panel is unique for
are known, e.g. at BZdjlni, nmnclapa of cave 3 (ill. 1, figs. 15-18)
the fact that all figures can be viewed from one side. There are no and at Ellorj 16 (ill. 25, fig. 278), in both cases, =grttikeya
separating ridges, but the figures are nevertlleless confined to a par- has replaced one member of the respective group. A characteriza-
ticular, limited space. Indra and Varuna are both rendered bigger
than the other dikpiilns. The scene impressively shows how the
directional guardians rush towards the divine couple, Siva and
4" The atlending divine personages number eight, which corresponds to the
piirvati? seated on a lotus throne in the centre, to P ~ their
Y reverence. number or i i r a ' s arms, i a a r , le pildlclism generally exists betwren
The panel has been designed in such a way that all of them, except Na!eSa's arms and the cr~!arlikpdl~~s.
for Indra and Varuna, are heading lowards the centre. Indra and 46' See also BANERJI 1985: ill. 12 ( 1 n i r r 0 r - i ~ ~ ~ ~ ) .
Varuna move in accordance with the circ~mambulator~ 4" PARIMOO
@m&/<- in a battle, 1983: fig. 29. Therc, only a selection of dikpcilns is shown, tn&"gcd
?i?u) order, which is generally foll~wedon ceiling panels. The well-
463 ,, ,,,,, ima: ,67, ll,osi probably ;l ceiling
 FRIEZES OF MANU LOKAPALAS ( sequential, aligned)
Chechar, frieze (DILIPKUMAR (I( y) (A 1) (v Vs) (So? Sii?)
1986: PI. VIII, 3) ,

Visnuplda temple, Gayl, frieze, (A 1) Y V K - - -

H: 23 c n ~W:
, 51 cm (SAHAI1970: plale)
Old temple, LSinbYRajasthan, G I (V5 v) (y K) - -
frieze (MEISTER1991: pl. 393c)
Mathurl, Govt. Museum Sii? (A 1) v5 - - - -
Mathura, no. 3 1.2 110, frieze
H:11.5 cm (S.D. TRIVEDI 1987: pl. 1)
Southern Bihar, Museum fiir (Sii so) (v5 v) ('4 1) (K Y)
Indische Kunst Berlin, frieze (fig. 311)
Bihar, State Archaeological (sii so) (vii v) (A 1) (y K)
Museum Patna, frieze (fig. 312)
So) (v5 V) A - - -
Southern Bihar, State Archaeo- (sii
logical Museum Patna, no. Arch.
_ 11265, frieze (BI+A~ACIIARYA I 993: fig. 3)
Guneri, B o d h g a ~ aSite Museum (fig. 313) - - (V5 v) (A 1) (K y)

-A ~ A N u L o ~ ( A ~ON
I AV
~~A
~ I V ~ l (sequential,
~ RELIEFS ~ ~ ~not strictly
~ ~ aligned)
~ l
S k y a temple, Bargaon, (SO So) (V V5) (A 1) (Y K)
- - - - P

p-
Sii (V va) - (Y
-K)
-
(1 A) -
So

Sii (1 A) (v5 V) (K Y) So

A Agni K Kubera V Varuca ( pairs of deities

G GaneSa So Soina vfi VSyu - not extant
I Indra Sii Sfirya Y Yama

111.31: Representations of Manu Lokapiilas (8th through loth centuries)

one could assume that the remarkable number of panels of this phase
lion Of the friezes in their chronological order will now touch all would be the consequence of the introductio~lof a;fadilvihs in a
Important aspects of this conceptual type.
wider area, this cannot be the only reason. About half of the num-
Even when the group of Ku$"a panels showing four stereotype
ber of panels depict the already mentioned, partly substituted group
figures could be finally proven to represent the ~ ~ t u r l ~ k ~(see ilas
of Manu Lokapiilas (Si~rya,Candra, Indra, Agni, Yama, Varuna,
Section 2.1.4), there seems to be no direct link with the later frjezes
" "l1. On later depictions, the directional guardians have always
been differentiated.
Vsyu, Kubera). Thus, it seems that both versions were vying wit11
each other for importance. It must be recalled in this place that the
Manu Lokapalas occurred only once in a directional arrangement,
the 7th century, when the earliest a;tadikl)Gla sets were
at MundeSvari Hill (ill. 2, figs. 36-41), which is also the earliest
Conceived, two friezes survive, At Bhubaneswar, ~ a r a 6 ~ r i i m e i ~ ~ ~ ~
known exa~nplefor the depiction of this group, c. early 7th century.
(fig. 42-44), belonging to the earlier part of the century, the
On the level of directional symbolism, they were not competing for
Original llL1mber of guardialls seems to have been eight. It inclllded
donlinance with the usual dikpdlns, except probably forMundc6vari
Danda~ii$,who is a guardian of the site. TWO of the deities
itself (see section 2.2.3). The Manu Lokapiilas were quite dominant
"'Provided with vehicles, while in the case of ~agdapgni,
in the category of friezes, at least in the region of their occurrence,
vehicle has ever been shown. The figures are placed in individual
i.e. North India. Unfortunately, only one of the Manu Lokapiila
which are separated by pilasters and crowned by udPi7las
In reliefs is preserved iiz situ (LiimbdRajasthan)464, and there is 110
relief. They are depicted seated in different postures.
of the particular indication whether they had ever occupied a place above the door.
'Ithough there is no sign for a The most striking feature about the Manu Lokapiilas is the fact
cept Of the ~likpdlason the ParaSuriimeS'vara temple, the door deco- that they are arranged in a binary manner, i.e. particular pairs of
at Nbiiyanapuram, early 10th century, is in some aspects deities recur. However, the sequence of pairs as well as the order
relniniscent of it. Tile second frieze, of the latter part of the 7th a particular pair may change. This binary system might go
Or Somewhat later, is the sole evidence for clik1)Gla represents-
back to an earlier directional arrangement, but there is only one
'lon in its region, viz. D a k s i ~ aKosala (M.P.). The door decoration
indicatioll for it, the - now loose and incomplete - MundeSvari Hill
at Palari (fig. 89) accommodates a colnplete set of a!?anikpi1as9
Ihough not all of them are aligned in a row. The guardians are seat-
ed On. vehicles, thrones, or a combination of both. The distinctly
l n d l v l d ~ 1rendefing
~~
464 M&lsTeK 1991: pl. 393~.The panel belongs to a series of reliefs carved on

of the gods, besides their irreglllar sequence, lintelsin the rarigamarjda~~a (ibid.:pl. 393~-0.The lhe~nesare not linked in ally way,
a'e Interestingly,Agni is shown separately, on a Liligodbhavamtirti panel, standing in
most significant aspects of this set. NOW,the gth and dbhnriga posture in a ~nandorlaof flames(ibid.: pl. 393%third figure from the right),
CenturiesWitness a large p&~lction of dilc~dlafriezes. AlthoL1gh
set (ill. 2, figs. 36-41). The pair SQrya and Candra is in some cases
arranged symmetrically on the extreme left and right sides of the
rest of the group. This is so far only known from a few ~ i v a
Vaiviihikamiirti and Visnu Anantaiayana panels (ill. 31).
Among the Manu Lokapiila reliefs, further variations occur: in
some cases they are depicted without vehicles (e.g. figs. 311,312),
and in other cases they are portrayed as seated on their vehicles
(Chechar, Gayii; cf. ill. 31). The variants at LiimbB and Mathurii
seem to be comparatively rare. The former piece also shows the
Lokapiilas seated on their vehicles (except for Kubera, who is shown
without), but there is some dynamism in the group, which creates
the impression of vigorous movement. On the latter piece, only the
first figure on the left is seated on horseback, while the others are
standing without vehicles, which is quite rare. The frieze is appar-
ently incomplete. It may have belonged to a narrative panel. The
Manu Lokapiila group was perhaps still depicted in later times,
which is indicated by its presence on a visnupatta, dating to c. l lth
century .465
Apart from the representations of Manu Lokap2las and the
dikpdla reliefs in the vicinity of doorways (see below), there are
only two more cases of about the 9th century A.D. Both have been
mentioned in the systematic part of this study. Besides Bikkavolu
(see below), there is another case in Andhra Pradesh, viz. Mukhal-
ingam. It is a loose, fragmented frieze466showing Yama, Naiqta,
Varuna and Viiyu riding in procession on their prominent vehicles.
All of the remaining reliefs of the 8th to 10th centuries are part
of or once were part of doorways, or are friezes on other architec-
tural members. In summing up the evidence gathered SO far, the
most striking point is the absence of any uniformity in the mode of
representation of dikpalas on friezes, except for some of the Manu
Loka~iilareliefs (figs. 311, 312, and "Southern Bihar", Patna
~ u s e u mcf., ill. 31), which are geographically and chronologically
closely related.
Group Representation in Adjacent janghii Niches
he case of the ~araSuriimeSvaratemple at Bhubaneswar (ill. 3;
fig. 42-44) has been discussed at length, representing an important
early occurrence of dikpdlas. The directional guardians nowadays
appear in adjoining jarighii (frieze of images on the wall) niches in
two groups of three which are separated by a window. Originally,
before the insertion of the window, which has violated the sequence,
there may have been eight members aligned in a row.
In the case of ~ i k k a v o l u l ~ n d hPradesh,
ra the depiction of Agni,
lndra (fig. 283) and Vgyu (fig. 284) could be viewed as a partial rep-
resentation of the dikpdla group. The images are found juxtaposed
on the southern wall of the GoliligeSvara temple of c. 9th century,
with the absence of other dikpdla representations in this region, the
concept could be interpreted as a faint reflection of earlier dikpdla
sets on walls as occurring at Alampur. It is remarkable that here, a
is
member and in the central position.
portraying Indm as the largest
Doors
The is the also otherwise remarkable
gathering of eight dikpiilas at the sides and above the door of the
Siddheivara temple, Pal2rilM.P. (fig. 89). As has been described
above (see section 2.3.1.2), the group appears divided in two batch-
es of three on both sides of the lintel, with the remaining two
ing the date, there are two suggestions, "c. 675-85 A,D." and "8th
century". A completely unusual concept prevails in the represents-
tion of the dikpdlas. Strangely, the group did not occur afterwards
in the region of Daksina Kosala. Most of the deities are turned
towards the door opening. It can be noted that the closer they are
placed near the door, the stronger is their turn towards the door. In
this consistent concept, the two outer guardians are rendered fronta1-
ly. This fact can be interpreted in the way that the asfadikpfilaswere
employed as guardians of the door, towards which they rushed frorn
different directions, though their arrangement is not systematica1
according to the usual sequential rencering of the directions.
Before discussing the North Indian examples, the only two
South Indian occurrences shall be mentioned. They differ COm-
pletely. The 8th century gopura-like entrance structure of the
Kailiisa temple (Ellor5 16, ill. 25, fig. 278) has a large frieze of
dikpdlas, most probably not the canonical group, which can be
regarded as door guardians, quite in line with the Paliiri group.
Those which are extant, are all turned towards the door opening, rid-
ing their vehicles. The dikpdlas are probably led by the god of war
KBrttikeya as their general, who alone is depicted frontally. Their
order is not systematic, as far as it can be perceived now.
At Hemjvati, Doddeivara temple, an astndikcl~iila-~um-NateS"
frieze embellishes the door lintel. Here also, the directional
guardians ride their prominent vehicles. The scenery seems to be
rather joyful than martial. S~VARAMAMURTI (1974: f i g 67) in his
description expresses the idea that the dik1281ns are depicted clanc-
ing like the central NateSa. As has been briefly mentioned above, it
is a remarkable achievement that a particular order has been estab-
lished here, rendering the &kpfilas as coming from both sides, from
the stations of an imaginary circumambulatory path (pradak-
siniipatha). This order can be regarded as an important development
in the period under discussion. The Doddeivara temple app'0xi-
mately dates to the second quarter of the 10th century.
Belonging to the Orissan style but located in northern Andbra
Pradesh, Niiriiyanapuram has a unique example for the extensive
d~corationof doorjambs with dikpiilas.467 In continuation of the Yth
century dikpdlas with vehicles at Bhubaneswar, paraSuriimeSvara
temple (fig. 42), particularly the representation of Varuna, the gods
are rendered separately in squarish compartments, with small velli-
cles underneath what seems to be the throne. One of the apparent
modifications of the early 10th century temple at Nfiriiynpapuram
are the haloesbehind the heads. The sequence of dik17Zlas starts with
Indra below on the left jamb (third $&ha) and on the right
jamb with Varuna, the alignment proceeding from top to bottoln
here. This is a rare example for a vertically arranged dikpiila frieze
in the period under discussion.
The remaining friezes from Nortll India ulliformly render the
astadikl2dlas in their "proper" order from left to right, starting wit11
Indra as the guardian of the east. However, the individual features of
the panels are still remarkable. The NO impressive friezes of stand-
ing deities from Gaur? (fig. 240) and Badoh (fig.180), which both
measure more than two metres in length, are
assumed to have served
as overdoor decorations of 9th or loth century temples. The forlner
panel starts with GaneSa on the left side and has pilasters as SePa-
rating elements between the deities. This visual concept creates a
certain uniformity in the set-up, which is somewhat balanced by
rendering the dikpdins in a distinctly individualized manner. T b s ,
4" C. BAUTZE-PICRON 1994: fig. 11. ~h~ eig~t~okapalss sppear in tbelowesltier.
*hey are all squatting in different postures, willlout vehicles. 111 a few cascs, their iden-
tification is difficult. It may be a contaminaled version,
4" MKRA1987: figs. 19-21, eastern door of the m~lkAnMld(vestibule).
 the attributes are held either in the left or in the right hand or in
both, and the physical proportions also vary. The vehicles are only
in miniature size and placed below, in a recess of the
moulding.
The ns[adikpdla frieze from Badoh (fig. 180) is perhaps the
latest one in this category, dating to the late 9th or early 10th centu-
ry- The deities are all four-armed, but survive only in a severely
damaged state. They are standing in an extremely flexed posture
(atibhn@z)and are all provided with a large halo. Of the four hands,
the 10wer left always holds a kanza!z&lu, and the lower right is held
In var~danzudra.On both sides of the frieze, a female figure is
who is also provided with a halo and turned towards the
group. A thin band of floral motives in relief below the ridge forms
the floor level for the dikljdlns.
In the two cases to be finally discussed, the associatioll with the
is clear. The earlier relief of the two is a fragmented Part of
what must have been the upper end of the right doorjamb with
the adjacent portion of the overdoor to the left.4b8 Below a row
Seated planetary deities (Iznvagmlzns), there is a tier with standing
di@iilfls,out of which (from the left) only Nairyta, Varuqa and VByu
Preserved, further a larger seated figure, and, ollce more in the
smaller size, a standing, damaged figure of Kubera. The two-
:rmed directional guardians hold their basically rod-like attributes
ln the right hand, while the left rests on the hip. They stand in
flexed posture ('tiblz&gn) like their counterparts, but the
feet are kept apart, which is typical for the Western region
Ra~asthan).It is assumed that the large four-armed, seated figure
'lva (upper arms damaged) represents i ~ l n aSuch . an inversion of
Kubera and kBna becomes likely when taking into account the fact
that space remained at the right end of the panel, where iS'na
have been depicted. With only three deities missing on the left
'lde (Indm. Agni, and Yama), it is quite obvious that the alignment
Of dilcpalns could not have extended completely Over the door Open-
'gg. Probably, another group of minor deities, a dd5vntGrQ group
Or the like, had ollce completed this tier of the door decoration On
the left side. Perhaps a larger deity in the centre separated the
groups.
The last relief, which is one of the latest (C. 900 A.D.1 friezes
under
is extant in sjtu at ~arwbigar/M.P.,on a temple
J a r ~ i - k i i - (fig. l ~ The figures are deeply cut and
~ ~ ~179).
"
' ln follow the usual sequence is
with Indra on the left. It
nent
remarkable that this order has been maintained, although the prom'-
of the two gmups of four dikpiiias both face the centre
which is formed by two jackal-faced goddesses. The sitting posture
dOf the two-armed dikp51as is dakjiniil,ra~aab~p5d (right 1% pen-
ent)' A remrltable degree of individual, non-uniform treatment,
gives the group a lively Attention has
b paid to details, e.g. two tiny f i g p ~ in
s shallow ihe
relief On
"kground in an almost narrative manner accentuate the funct!on
:particular gods, viz, Varuna and Kabera (see section 2.3.2*2).The
arw5sBgarfrieze is locate; on the tier of the overdoor dec-
. Of the sanctum door. ~~~tabove the frieze, the beams for the
Oration
celllng of a now lost porticw were attached (cf. TRIvED1
'990: fig. 37).
. In summing up the evidence on door decoration, the great
?lon Occurring in this category has to be noted. In a few cases. Ihe
Impression arises that links for the of a pxticular
concept
B are now missing (Bhbaneswar - Narjyaqapllram; ~algri-
arwBf'gar), but often, the treatment of the dikpiila theme seems
be,lmbibed by innovative fresllness. only in a few cases*.the
nadrkl'a~fls might have had a secondari functioll as door guardians
Ellor5 16). Significantly, these cares are the earliest Ones'
Tlzree-di~~zerzsio~za~
Friezes
This conceptual category did not become popular in the period
under discussion. The diverse members of this formal group once
more tell of a conceptual innovation, which was not necessarily
followed by consolidation. For an ideal positioning of dikpdlas in
their respective domains, the depiction on multi-faceted or round
architectural members would be ideal. Consequently, such octa-
gonal or circular arrangements of as~adik~~dlns occur in ]ater
periods.469
The first case is unique for its arrangement of dikpalns on the
beams of a ceiling bay (figs. 49-54). This peculiar concept in the
8th century Huccappayya temple at Aihole at the same time seems
to stand at the very beginning of nsfadikpcila representations on
ceilings. During the subsequent developnlent of ceiling panels, the
&kl.'dlns were included in these, their panels being of the same
forinat as the central panel (e.g. fig. 137).
The second case is somewhat unclear, as the particular deities in
the procession can hardly be identified. The frieze covers the beams
of the Nandimaqdapa on the south, east and north sides470 at Alur/
Karnataka. In date, it could be assigned to the 10th century. A con-
tinuous line of gods (mostly accompanied by their consorts) ride
their prominent vehicles over cloud forlnations. All are shown head-
ing towards the left. It is not clear whether in some cases, the same
djkpdlas are once more depicted in the frieze. Thus, there are three
elephant riders. However, other gods completed the alignment.
One of the outsiders is KBrttikeya, who rides his peacock near the
centre of the eastern side. An obvious sign for the intended direc-
tional symbolism is Yama's mighty buffalo, dominating the centre
of the southern side.
The octagonal portion of a C. 10th century offering platforln
(bnlipQlza) is decorated by affadikpBlns at Kambadahalli/Karna-
taka.471 The bnlipi~lzahas been regarded as coeval with a group of
Jaina shrines and is structurally quite elaborate. From bottoln to
top, it consists of an updrzn and square, octagonal, circular, and
lotus- segments. The octagonal section is decorated by shal-
low reliefs of rlikl~dlns,riding their large-sized vehicles along with
their consorts to the left. The movement thus conforlns to the
13mciflksi!zii concept. The figures lift their hands in the gesture of
salutation. They appear to be bereft of individual attributes. The
squaish relief panels are bordered 011 both sides by vertical deco-
rated bands, the design of which always seem to match with the
adjacent band.
The last member of this category, a free-standing pillar at Dev
~ ~ ~ ~ j ~ ~ k /c.B8th ihar, referred to as dlzvnjnstambhn
(flagpole) by the author, also has both square and octagonal sec-
tiolls. The octagonal portion starts at medium height. It is decorated
468 MASON1993: cat. 110. 58, Heeranianeck Collection, hight: 73 cm, breadtIl:
99 cm, Kota region1Rajasthan.
469 Three bglipi!hn-like S O C ~ are
~ S pleserved from c. 12th century, two of them in
the state Museum Hyderabad and one in the Prince of Wales Museum Bombay, no. s.4
( ~ ~ ~ 1988: ~ l no.
~ 120).
, , ~The~top ofthe cylinder is decorated by a lotus and r.c$isigns
in a circular arrangemeat The cylindrical part (octagonal at the bottom and round at the
top) is carved with n.~!adikl~rilnsriding to the left along with their consorts. Below
varuna, the ,-Ilarioteer A I U I and
? ~ the seven horses of S k y a are carved in shallow relief,
on level of the square base-slab. Thus, the lnost prominent part of its content is solar
Allother example, the circular anangemellt of seated lnetal images of the
d;k,,6las above tile base of a dl~vnjastatilbhaat Achchankovil, has been illustrated by
R,P. GOSWAMI (1979: p!. 7).
470 T K. SARMA 1992: pls. 75-77.
471 Ibid.: 155; p]. 118.
472 A s l l ~1980:
~ pl. 125. Unfortunately, thc rear side of the pillar has, to our
knowledge, not been published.
- most probably - with agagrahns (eight planetary deities). The
upper part of the pillar shaft is again square. The relief panels on the
four sides are bordered by ridges. As far as can be judged from the
photograph, they contain representations of lokapnlas seated on
rather large vehicles or thrones. The pillar design may bear a cos-
mic symbolism, in which the postitions of the guardians correspond
to heaven or to the peaks of Mount Meru.
2.4.2 Single Representation
The conceptual type of single representation is a category which
contains a high degree of internal variation. The greatest diversity
concerns the choice and placement of the represented deities. Other
variations occur with respect to the frames of the panels, which are
often sunken into niches, and, of course, the concept and layout of
the panels.
Naturally, one would expect the dikpdlas to be positioned in
accordance with their respective regions. Yet, this was, strictly
speaking, never the case with the images on temple walls.473 Al-
though the ideal positioning Was more or less achieved, compro-
mises on the basis of certain architectural and theological rules were
always made. The hierarchical order of deities must have been the
reason for the fact that dilcpdlas were generally not permitted in the
blzadra (central offset) position on the wall.
In this category, three sub-types will be formed: jarigkii (wall
frieze) images, Siklzara (temple tower) images, and sculptures in-
the-round. There are also representations on other architectural ele-
ments, the exact positioning of which is not clear in some cases (e.g.
MundeSvari, see section 2.2.3).
The jaiiglzfi images are doubtlessly the kIrgest and most varied
category. In the chronological part of this study, these representa-
tions have been discussed at length, So that a general summary of
the results will suffice here. The second category, Sikhara images,
are a very interesting group, starting quite early, but only few
images of this type belong to the period under discussion. The
category of separate, individual (loose) icons, which were kept in
the temple's prr7k8ra or in particular shrines, are restricted to the
extreme ~0~1th. The relatively few remains of this group have been
treated in section 2.3.2.5 on South India.
2.4.2.1 Jaliglzii (Wall Frieze) Images
It is a11 established fact that, unless the assigned dates of some of the
early temples in Madhya Pradesh and Rajasthan would have to be
modified, the KBlikiimiitii temple at Cittaudgadh and some of the
temples at Alampur (Biila Brahmii, Svarga Brahmii) must be regard-
ed as the earliest extant shrines with dikpiilas. They all belong to the
latter part of the 7th century. As has been noted above (see section
2.3.1.3), there may have been some kind of interaction between the
two sites, and a notable common feature is the positioning of Soma
on the north. However, there are basic differences in the represen-
tation of the deities. While at Cittaudgadh, the basic concept and
layout of the individual panels (standing deity with small vehicle at
its side) remained the same for centuries to come, the concept of the
dikpiilns at A l a m p ~ r as
" ~well
~ as their positioning in wall niches did
not occur later in the Deccan.
The temples of Cittaudgadh and Alampur may have had solne
predecessors, which are now lost, but still, the evidence preserved
at both sites is uniquely important. The most remarkable aspect of
both these early sites with regard to the dikpiilas is the fact that the
sets play an important aesthetic role on the temple exterior. Inter-
estingly, there were other lines of development without such avisu-
al domination, particularly that of Osiiifi.475 Even the early occur-
rence of incomplete sets at Pipiid, ?gi~g,Amrol (~iimeivaratemple)
and Osiiifi (Sfirya temples 1 and 2) lnay be interpreted as a reluc-
tance to let the dikpdlas dominate the telnple walls.
During the early development of the a?tndikpiila the*ne On
walls, a large variety of individual concepts was created
special cases may be mentioned at first:
- The dikpiila sets on polygonal temples (Mundeivari; Indor),
and
- those dikpcln ilnageswhich are integrated in colltinuous wall
friezes (NiilandL, Temple no. 2; Gwiilior, ~ e l i - k g - m a * ~ ~ ~ ;
PHhBrpur).
Both categories provide valuable information on the position of
the clikp81as among the hierarchy of gods as well as on the rllles of
their directional positioning. The earliest example, the octagonal
MundeBvari temple (ill. 2; cf. fig. 351, unfortullately does not sL1r-
vive intact. In can be only hypothesized that the dikyiilas Once occu-
pied the lateral niches on the four oblique sides of the temple. That
implies that a certain compromise with regard to the ideal directions
had been lnade here as usual, although the temple would have pro-
vided the possibility of a correct placement.
The Siva temple at Indor, c. latter part of the 8th century, is
eleven-sided, two kq,ili (walls of the vestibule) niches adding to
nine portions, which have been arranged in the round.
Here also, it can be demonstrated that the Ucorrectw of
dik~alashas not been given priority, as the wall portions in the
cardinal directions have been allotted to the usual pariviira
(family members of the main god) GaneSa, Skanda, andPpvati. The
directional guardians have been distributed in the two Itapili niches
and six of the nine blzadrn T,,~~,concerning their individ-
ual orientation, the same compromises have been made as on rec-
tangular strllctures.
Like the polygonal temples, the colltinuous wall friezes, incluf-
ing figures other than di$a[ns, belong to a relatively early stage tn
the depiction of directional guardians. In all three cases, ~ 5 l a n ~ ~ ~
Stone temple no. 2 (see section 2.3.3), Gwslior, ~ e l i - k i i - m a n d ~ ~
161, and Plhiirpur, Solnapura Vihara (ill. 29), the set of astadikl,nrlas
is not complete. Chronologically, NBlandrI belongs to the 7th centu-
rY, ( h I l i 0 r to the 8th centurv. and Pgharpur can be assigned the
late 8th century. In the first two cases, the wall friezes are
pletely preserved. Thus, it is clear that some of the dikpalns h"e
been actually missing. At Piih.hkpur,the northern portion of the wa11
frieze has been completely destroyed. Judging from the irregular
distribution of the remaining guardians, it seelns quite likely that the
group was originally incomplete. In a final assessment, it must be
stated that although the conditions would have allowed a cofleCt
directional arrangement of dilcpiilns in continuous wall friezes, the'r
placement was obviously haphazard.
473 The ideal order would be: Indra stationed in [lie east, Agni stationed il: the
south-east, and so on. This was ,he armngcment on ceiIillg panels, on foondation bricks'
and on thm-ditnensionally arranged friezes,
474 Viewing the Alalnpur images in a wider geograptiicnl context, however, "ley
show a tendency which seems to bc typical for tllc southern part of India, viz. par-
tray di/c/)cilaswithout any vehicle whet1 they were not conceived as riders,
475 On the Hariliara tc~nplcsnos. 1 and 2 (ill. 22), the introduction of f>r[ffirafha
(intermediate) niches has led to the loss of dominance of the cjikpalns in llleir kor!'a
niches.
 The most frequent depiction of dikpiilns on temples is the
Positioning of the guardians in k n r y niches, i.e. next to a corner.
On east-oriented temples, the sequence generally starts with Indra
on the southern (proper right) side of the entrance. This was already
the case on the earliest, just mentioned temples at Cittaudgadh and
Ahmpur.476 Regarding the kartza position, it Inay be assumed that
the dilcpiilas were employed to protect the corners, which are
traditionally regarded as vulnerable It is also probable that
they were intended to protect the walls from the sides, which
seems likely in view of the "doorkeeper (dvnrrryiiln) pose"477 of
the dikpiilns on a large number of temples in Western India
(Rajasthan).
The first exception to be mentioned is the inclusion of the
kapili (walls of the vestibule) niches ill the placement of dikpdlns.
Such Cases occur in the 8th century, both in Madhya Pradesh and
in Rajasthan. At Amro1,478 Rslnegvara temple (ill, 14), and OsiLii,
Sbrya temple no, 1 (ill. 211, some of the members of early, incom-
~ l e t esets have been placed on the kalJili wall in a non-systemati~
manner, while at Indor, the krrpjli niches seem to have been con-
sciously chosen for the most prolninent dilcPalns of a complete
Set, i.e. TSsna2nd (ill. 27). ~t is an fact that the
dikpiiln representation in knpili niches was only an intermediate
Stage in the development alld was never taken up again after the
8th century.
A completely unique case is the early 9th celltury PipalBdevi
temple at OsiBii. ~t is the only case of &kl,~las placed in prntiratlzn
(intermediate) positions. l-his placement occurs only on the west-
ern wall. There, the two dikl)61as which should have been accom-
modated on the north, have been depicted in the prntiratlzn
niches, adding to the usual dikpglas of the wester11 side. Moreover,
inversion of and ~~b~~~ has taken place (see section
2.3.2.3).
The second substantial exception is a in the positio~lsof
the directional guardians. such a set, starting with Indra on the
occLlrsat pipiid (ill. 20), at Osififi, Visnu temple no. 2
(MCISTE~ 1991: 265), at caLIrgsi,and at GaneSwarpur, south-eastern
subsidiary shrine ofthe paficapfindava temple (for both OrisSan sites
see section 2.3.2.4). ~ h i d ~(ibid.: ~ ~241),
~ ~such f an
i order is also
implied, willl !isna surviving on the nolth(-east) and Yams on the
s O ~ ~ t h ( - This
~ ~ ~positionillg
t), may have been the result of particular
C O n ~ i d ~ ~ ~ resewing
t i ~ ~ ~ the
, wall for the first two mem-
hers of the set, but this concept did not become popular.
In solne cases ofincomplete sets, it is impossible to decide about
R$meSvara Mahfideva,
the originally intended order
ill. 14; Pipfid, ill. 20; ~ ~ l l ~ ill. l l i , in some other cases
~ ~so),
doubt about the ullderlying
of incomplete sets, there remains
(e.g. Gwiilior, Caturbhuja temple, ill. 28).
where single representations
Regarding the parts of the
one or more sets of dikl)alas occur, a number of different concepts
C""e distinguished:
1 Only one set, on the gnrbl7agrlza (central shrine) exterior of
a ~zirnrzdziirn(without ambulatory) temple;
2 one set, on the gnrblwrizn exterior of a sz,ld/lzro
ambulatory) temple;
3 only one set, on the l12n!z$a1~n (pillared hall) exterior of a
s&~rNziirntemple;
4 two sets, one respectively on the gnrblzagrlla and nzn!ldq~n
exteriors of a siiizdziirn temple;
S only one set, on the ~~zn!zdnpnexterior of a ~zirmzdlziiratemple;
G Only one set, on the exterior of the central ~trUCtLlreof a
13~~ciiyntnrzn (consisting of five sanctums) complex;
7 several sets (theoretically up to four) on the exteriors of the
subsidiary temples of a paficdyatana complex, adding to at
least one set on the exterior of the central structure of the
complex.
Concept no. I is the most frequent one, as it refers to the prevailing
temple type in the northern part of India during the period under dis-
cussion. The only example for Concept no. 2 is the late 7th century
KBlikfimBtB temple at Cittaudgadh (ill. 17). The slightly later tern-
ple at the site, the KumbhaSyiima (ills. 18,19), significantly belongs
to Concept 110. 4. The third concept is best illustrated at Alampur,
where all teinples except for the KurnLra Brahmii and the Taraka
Brahm5 seem to have been provided with dilcpdln sets on the
nin!zdnpn (hall), but not on the gnrbhngrlzn (sanctum). A very spe-
cial variant of Concept 110. 3 Occurs On thc PfipanHtha temple at
Pattadakal (ill. 9). Here, the four directional guardians, along with
Varfiha and BrahmB, "guard" the bhndriivnloknrzns (false porticoes)
on both sides. Their placement is partly irregular, with Indra p s i -
tioned on the west.
The occurrence of two Sets on different architectural elements
of a s6rzdzdm temple (4)is reported only from Rajasthan: Cittaud-
gadh, KumbhaSyBma temple (ills. 18, 19); AbBneri, and Khed (see
section 2.3.2.3). On the earliest of the three temples, the Kum-
bhaSyL~na(c. early 8th century), the two sets differ substantially.
The set on the outer wall seems to be quite regular, whereas the
gnrbhagrlzn set ~lniquelyincludes a number of nearly identical
deities ("'TSfina type"). At Khed, the images of the two sets are very
much alike 226-233).
The rare Concept no. 5 is well illustrated at Satyavolu, Rfilnal-
irigeivara temple (ill. 13). The dikl,dln set m~tsthave remained
incomplete, as the (north-)eastern one of the available eight niches
is occupied by a Sbrya image. At Caurfisi/Orissa, the same mode of
representation occurs for a complete nstadikl~slnset, which starts
with Indra on the (north-)east (see section 2.3.2.4). Concept no. 6
appears on the relatively early l~aficriyotnrzntemples at Osijfi,
Harihara nos. 1 and 2, C. middle of the 8th century.
Concept no. 7 was probably employed at G a n e S ~ a r ~ u r l ~ r i s s a
(see section 2.3.2.4). The south-eastern and south-western sub-
sidiary temples each have a complete set of nmdilcyiilns (figs.
249-2653. However, there is a difference in the distribution of
Ikpdlos, as the set of the south-eastern temple shows the excep-
tional order starting with Indra on the (north-)east. The north-west-
ern subsidiary shrine exhibits Krs~aliliiscenes instead of the dikl,sln
theme. The central shrine of thepnficiiyntnrzncomplex survives ollly
with empty niches and it is known whether they once contained
dikpdlns as well. The GaneSwarpur exarnple of the late loth centu-
ry once more demonstrates that the multiple occurrence of basical-
ly similar dik~dlnsets ill a temple in the course of time seems to
476 how eve^, Ala~npuralso provides the only known example for a posltlonlllg of
drkl~dlnsIn the ce~ltleof a wall (Yama on the soulli, and SorndCandm on the llorlh),
thereby changtng tlie older of those in the collier pos11101isTak~ngthe Svarga Brahrnj
as an exa~nple(ill. 11). it becomes obv~ousthat out of the riglit usual d~kpilos,flve of
them (Nal~yta,Varuna, VByu, Kubcra, i h a ) occupy posit~onsthey would have had
when the sequence had stalted with Indra 011 the (nolth-)east. Howevcr, I n d l ~1s placed
0, ,I,, (s0utl~)eastand Stirya, as an associated de~ty,holds the (no~th-)eastemposltlon
477 cf. [lie p o ~ t l l r e o f h ~ ~ ~ aat~Ol S~ I/ ~a BsS,n ~ y templeno.
a 1 ( M e s ~ t n1991, PIS
312,314). wlilcll IS typical Tile pivot 1% is posltloned next to the door wli~chlie p ~ o -
tects, and the free leg is 011 the other slde. The same posture 1s generally shown by
d,kprilcl~In Rajasthan, with the free leg polntlng towatds the nearcsl colliel
478 ~ l l seco~ld
e 8th century temple at Arn~ol,tlte rurned DSncbBbri, prob,lbly had
also knl~rliwalls, but ~t cannot be declded any lnole whether any d~kl~ciln i~ilagewa$
pos~t~oned there (ill. 15).
have increased. On the Khajuraho temples, not discussed here, Again about hundred years later, c. 730 A.D.,the H u c c a ~ ~ ~ y Y "
concept no. 6 became the usual temple decor. When multiple sets temple at Aihole was designed with an incompletely preserved
occur on earlier temples, they generally belong to different modes a$adikl,filn set in its interior (figs. 49-54) and at least one the
of representation.479 four lokn~~iilas placed in an arched frame at the upper end of the
Finally, another case will be mentioned which cannot be classi- r?zadI~yalatfi(central vertical band) of the s'ilckara. unfortunately9
fied in one of the above sub-categories. At BBran~biVOrissa,only the image of V a r u ~ aobserved by BOLON(1988: 309) on the west-
Agni and Yama have been included in the lavish temple decor (see ern side was never documented. No trace remains of any probable
section 2.3.2.4). They are not depicted on the same portion of the counterparts in the other directions.
temple. Agni is situated on the hall (jngarnohana) and Yama on the The 9th century example at BZrambdOrissa has been already
tower (ga!zdi) of this east- facing nirarzdhtirn temple. As bothdeities mentioned above, as its counterpart is positioned in a jnhshd (W"'
appear in the c~rrectpositions, one cannot deny their dikpiila role. frieze) niche. Here the image of Yamariding his buffalo andaccorn-
This is a very peculiar case of an intentional selection of two panied by attendants holding his attributes (fig. 242) appears On the
clikpGlas from a phase (c. 9th century) in which complete sets had lower part of the s'iklzam, framed by a gnvdlc~aarch. This arch
not yet been introduced on Orissan temples. the topmost part of a pediment of the central image of the sollthern
side of the garblzngrlzn (sanctum) wall, which has been extended
into the second storey (blzcitizi) of the s'iltlznrn.
2.4.2.2 Siklzarn (Temple Tower) Images The last example of this category is formed by a set of images
in-the-round, which have been placed at grivfi ("neck") level On a
Dikpiiln images on iikharns are a rare, little noticed type. It is quite Southern style temple tower (vimiinn). In this case, the 1 l 1 o r ~ ~ ~ ~ l ~ g -
probable that some still extant cases have not yet been uncovered. ical link with the aforementioned depictions is less than wit11 the
The rareness of this type raises doubts whether there was any lowing category (see below). The evidence appears to be absolute-
continuity at all or whether some of its representations had been ly unique. On the vir~zdrznof the NageSvara temple at ~ratakota/~.*.?
independently conceived. It is quite likely that a certain continuity which belongs to c. 9th-loth centuries (see section 2.3.2.1), a few
existed, considering among other reasons that the placement of courses of mouldings below the iiklznm, individual images in-the-
dilcpdlas "above" was an apt representation for this group of heav- round of ~ O Lseated I ~ dikpfilas are extant. Indra occurs on the (north-
enly guardians. It should also be noted that out of five known )east, with a narrow portion of a frontal elephant head sculpted s' '
occurrences, four probably represented the cnturlokapfiln group seat. On the south, Agni, Yams and Nairrta are placed in the correct
(four guardians of the cardinal directions). This may be regarded as positions, an - anthropomorphic - vehicle being discerllible only in
an early trait. The choice of the group of four regents may also Nairyta's case (fig. 275). ~ 0 t the h sparse indication of vehicles and
have been determined by the structure of the tower itself, which the plasticity of the images may point to south Indian relations
generally provides only space in its central portions, orientated (cf. the following section).
towards the cardinal directions. A development, which occurred
after the period under survey is the positioning of astadikpiilas near
the corners in the upper storeys of South Indian viriziirzns (e.g. at 2.4.2.3 Individual Icons
GangaikBndacB!apuram, later addition, PICHARD et al. 1994: 89;
ph. 202), where they are distributed in the same way as on wall In section 2.3.2.5 on South Indian dilq,Glns, this type has been dis-
representations. cussed at length. It never, at least as far as the evidence goes%
The five examples belong to five successive centuries, starting i~lcludesa vehicle. In the case of seated images, no complete set IS
with BhGmarii in the 6th century. Bhbmar5IM.P. is indeed a key site. known. In tile case of the standing images, however, a group of nine
The Siva temple has yielded the earliest evidence for the single rep- sepante images from the 9th century ~ ~ ~ t ~ ~ h ~temple { ~ Satv a r a
resentation of directional guardians. With only the images of the Tirukk@aiybflamilll;ldL1could be regarded as a dikpfila set with a
rlikpdlns Yams and Kubera surviving (figs. 25,261, the concept was few substituted members. In this case, a directiollal arrangement in
probably restricted to the four guardians of the cardinal directions. the pfikarcl seems to be partly preserved.
However, the portrayal of the group may have been incomnplete. It may be hypothesized that early, incomplete sets of aS!ndlk-
~ssociatedwith this group was a guardian figure, which has been ~ ~ d l were n s individually ensllrined, as attested by the later evidence
here referred to as DandapZni (see section 2.5.10; fig. 28). The of the RZjariijeSvara temple at Tasjgvfir of the early l lth century.
placement of the fine panels framed by elaborate gnviik~ns("cow's where chapels for the di/cl,filas were provided in the proper direc-
eye" arches) on the superstructure can be safely assumed, but it can- tions. The absence of vehicles has been observed ill south
not be decided from the extant remains, where exactly the gnvfik~as India, could have been colnpensated by the presellce of the respec-
had once been fixed. Apart from the snrbhogrizn (sanctum) roof, the tive vehicles at grip5 level as it occurs at Tafijsvilr. However, there
mal,dapn (pillared hall) soof may also be considered as a probable is no evidence whatsoever of sepamte &k13iiln sh,.ines before the
location (DEVA1988: 41), albeit a less likely one.
time of the RZjariijeSvara temple, which already postdates the period
When dating the Bhbmarii temple to c. 520-530 A.D., the under discussion.
Kumiira Brahmii temple at Alampur postdates it by about hundred
Years. As on the former temple, the dikpiila set on its $ik]lnra only
consists of Yams and Kubera (figs. 23,24), but here, the evidence
is completely preserved, with NateSa and probably Pirvati oc-
cupying the corresponding niches on the east and on the west, 47' An add~tlonalgioup icprcPelitatloll occuls at Ba1wjs3gar (see scctlon 2 3 2)'
( B u C 1985: ~ ~122).~ Yams
~ ~ and Kubera have been a" at LLjlnb3 (see qectlon 2 3 1 3). whele friezes add to Ihe dcp~ctlonof dfk171ilnsin
installed in the Correct directions. Yama appears seated on his n c l l e ~At Alhole, Huccappayya temple, a fr~ezeon beams in the lntellol Is
buffalo, while Kubera is seated without any vehicle or symbol with at lcast one drkpcila image on tlie towel (i~khnm).Ano~hergood example IS
the p1pcln3tha temple, Paktadakal, w h c ~ esets of tour and eight drkl~ciln\are found
At 'lampur, the relief panels are framed by trefoil arches. ,.espcc,l~e,y
LIIC cxtcrlor alld lnterlor ~cell,ng),
 2.5 Iconography of the di1rpIlas
It has become apparent in the course of the foregoing chapters
that the iconographic deterlnining factors, of which attributes and
vehicles are the most ilnportant ones, were subjected to a certain
variation. While the variations sometimes appear to be individually
conceived, general lines of development as well as substantial
'egiollal variations can be deduced from the available data.
The analysed images generally belong to the mode of single
(see section 2.4.2), as in the case of group represen-
tation the individual rendering of the deities is often less pro-
?ounced. An unusual lack of individual characteristics occurs on the
Siva panel at Ellod 16 (figs. 280-282), where the dilc~~dlns do not
c"rY any attributes. This is also the case on ceiling panels. The
crowded scenes and the small size of the deities often make an evd-
u"i~n very difficult in the case of group representation. The reason
behind the less individualized treatment may lie in the absolute
Subordination of the gods to the centrally depicted main deity,
lesser importance.
In the morphological analysis of the dilcpiilns, non-serial con-
texts should be excluded. Generally, however, it becomes apparent
[hat the visual concept of a particular deity does not change wit11 the
Particular context. A good example is the god Kubera, who must
have been popular outside of dikpdln contexts. The formula "seated
vehicle", which is oftell applied in these cases, has some-
times entered dilcpdln co11texts.480
For the tables, a selection of &Icpdla images has been made. The
most significant, "key" samples have been chosen, particularly in-
cluding early evidence. They are generally taken from the same sets!
if Possible, so that the iconograpllic traits of the complete sets can
be easily viewed together and with those of other sites.
The last column "Companyv has been omitted, for Want of space,
in the case of Agni. This aspect has been only considered in the
discussion.
2.5.1 Indra (table XIV, see p. 96)
Indian image from Tirukkadaiytir (D. TRIAGARAJAN 1989: pl. 105d).
The omission is not surprising in the latter case, as the vehicles of
the dikpdlas were not yet introduced in the extreme south in the peri-
od under discussion.
Among the various postures assumed by Indra, the riding pos-
ture has been chosen significantly often, but not in the majority of
cases. At Pipad (M.W. MEISTER1991: pl. 326), this concept is par-
ticularly remarkable, as no other deity rides its vehicle there (figs.
124-128). The motivation to show Indra on the back of Airavata
may have been to present an impressive picture of the mighty king
of gods, who is so closely associated with his vehicle rese~nblinga
rain cloud. In the frontal rendering, the visual unity with his vehicle
is particularly appealing (fig. 64). It is difficult to decide whether
the similar concept on the much later Cidambaram gopuras (S.R.
BALASUBRAHMANYAM 1979: pl. 11) has developed from the earlier
images.
Indra's headdress, the broad cylindrical crown (kirita nzukzqn),
occurs significantly often. It obviously represents a continuation of
an early - KusHna - tradition (P. PAL1979: figs. 1, 10, 14), to which
the oblique third eye on the forehead of the god (ibid.; cf. fig. 70)
also belongs. However, both characteristics were not always shown
together in one image.
As with the other directional guardians, no regular entourage
other than the vehicle occurs in the case of Indra. The attendants
found at Indor (fig. 152) must have been included because of the
large size of the lcnpili (exterior wall of the vestibule) niche. They
are, in this case, male and female, while two male attendants are
depicted as PSgna's entourage on the same temple and placed in the
corresponding, spacious niche (fig. 159).
2.5.2 Agni (table XV, see p. 97)
In the case of the Fire God Agni, the survival of a visual concept dat-
ing from the Kusiina period is particularly obvious (cf. section
2.1.5).484 It includes the attributes rosary and waterpot as well as
the stiff, hieratic posture of the deity who is encircled by flames485

As wit11 lnally of the dikpclas, the King of the gods, possesses One
480 G, BHATTACHARYA (1996) discusses Kubera in his recurring associatio~iwith
particular, largely indispensable attribute, his thunderbolt ( v ~ j m ) . ~ ~Laksmi
' and Ga~!eSa. A silnilar Squatting Posture, which enipliasizcs tlie big belly, and
The thunderbolt appeared already ill pre-dik~"ln image: of [llc absence of vehicle (cf. fig. 199) occur in carly sculptums, in wllicll Kubcra does
Indra (see section 2.1.5). Apparently, an Eastern Indian tradltlon not yet play the role of tlie Guardian of the north (figs. 124,182,198), but solnetilnes
omitted the v,jm (cg.fig. 234). However, ilnages with vfljrn like- ,]SO in latcr images, where tlic other tnelnbers of tlie dikl~cilagroup are provided with
wise occur in this region at about the same time (fig. 311, 3rd figure vehicles (fig. 306, top row left image; TICIVEDI 1990: p]. 100, Kuben at Keldharf

the right, vojrn broken off). AS with the main attributes 0 f l l 1 ~ ~ ~ ~ ~ . ~ ~ : ! ~ s i B n i oofnVnj,.a shapes SCC AA (1981: figs, 1993:
other dikpfilas,482 the vnjrn is generally carried in the right X"-XXI). Howevcr, the study does not reveal any particular lines of typologicnldevel-
hand, but Sometilnes held in the left hand of the god. ol'ment. Regarding the 11ajr.asin images of the dikpcila Indm, these are often damaged,
AInong the attributes foLlnd single representations of and Uie remaining ones show considerable, almost individual differences.
482 Amo~igthe cases in which each dikpzla clearly has one main attribute, Nairl.ta
Indra, roundish fruits occur at Pnhgrpur (fig. 234) and here
stands out, for he always carries his attribule, Ihe sword, in tlie right hand. Herc, it Illay
ln
of the vnjm, and probably at Alampur ('g' 64)' be understood that the sword was a colnlnon weapon, which had to be wicldcd ill the
The fruit as an attribute has not been mentioned ill the early texts right hand, h~terprelingthis evidence, all olher weapons (e.g. noose, club) 11lust]inve
(tables IX-XI), another variant, exceptionally occurs ill been regarded as purely sylnbolic devices, so that tlie ~ieccssityof showing them in the
earlier period, lndra carries a large elephant goad in his left right 483 hand was
This C. 7th ce~itury-reliefconies fro111 an unknown context (B. SAIIAI 1975:
hand on the ceiling relief at Pattadakal, pspaniitha temple (fig. fig. 2). It depicts I~idraseated in ~ ~ n k ~ i ~ ~ c i l , ~ a l a n ~(with
b a ~ ~the
c i (right
I n leg pendent)
It Is not clear, whether he also wields his vfljrfl here. goad a throne, which is supported by two elephants, a ~ carrying ~ d a fruit in his rigllt hand,
(alik~i~fl) has been mentioned alollg with the vnjrn and the lotus in is held in the wish-bestowing gesture.
4x4 The concept of the image most probably can be traced back to ''the early tra-
the VdlzP (3.50.4).
wllich dition of two-armed Yaksha statues" (R.C. AG~~AWALA 1965: 151). which is supported
N~ alternative exists for Indra,s vehicle, the by the recent discovery of a huge ")ak;a" Agni from Bliava Kalan (SRINIVASAN 1990:
bears the individual name Airnvata and was oftell illcluded in the figs, 3A and B).
S c u l ~ t ~ l and
r a l other representations. Among the examples selected 48s Flatiies emerging from the body are not restricted to tlic god Agni, but occa-

for table XIV, it is Inissing in the Mundeivari illlage (fig. 39), which sionally apl>car wit11 other deities as well. A rare image of Kziltikeya wit11 flanles

,
belongs m MallLlLokap lowe ever, md ill L e South
bcliind his head has bwn published by R.c. AGRAWALA 1968: fig.1.
Table XIV -Iconographic Chart of Indra

Site Attributes Vehicle Posture Headdress Carupany -

Alampur, Bala vajrn (bolt) r. elephant riding cylindrical -
Brahmi (fig. 64) flower or fruit 1. (shown frontally) crown
AIampur, Svarga vnjrn r. elephant salnnpddn conical crown -
Brahmi (fig. 70) rosary? 1. (feet together)
Amrol, vajra 1. elephant seated in vdnzn- cylindrical -
Dinebib5 (fig. 96) prnlailtbn11dcln crown
(left leg pendent)
on the vehicle
BBdimi, Cave 3, vajrn r. elephant riding conical crown two male
nzanac~pn(fig. 15) (shown in profile) attendants
Cittaudgadh, vnjru I. elephant flexed posture, cylindrical W

Kumbhabyima, left hand on the crown

exterior (fig. 115) left thigh
Indor (fig. 152) vnjrn r. elephant flexed posture, left hand ? flanked by a male realld
P______
on the lert thigh
MundeSvari, vnjra 1. - squatting cylindrical -
cornerstone (fig. 37) crown -
Osilfi, HH-1 (fig. 188) vajrn r. elephant seated on the vehicle, cylindrical W

right leg held by yognpa!?~, crown

left hand on left knee
Osilii, HH-2 (fig. 205) vcira r. elephant flexed poslure, left hand -damaged- -
on the left thigh
Pnhirpur (fig. 234) a fruit in elephant sanznl~drln cylindrical W

each hand? (feet together) crown

Pip%< vnjm r. elephant riding - damaged - female consort
(MEISTER
1991: pl. 326) (? indistinct) (shown in profile) sits behind the god
Tirukkadaiylir vajrn r. - sn17zupiida slightly -
(THIAGARAJAN 1989: pl. 10%)
conical crown
Umri (fig. 162) vnjrn r. elephant squatting on the vehicle, -
cylindrical
left hand on left knee crown

(figs. 7,8). A popular Vedic concept, however, according to which

Agni has three powers, three bodies and three tongues (e.g. RV
1II.20.2), may have been reflected in a unique piece of about the
8th century, in which Agni act~~ally appears three-headed (K. D ~ v d
S.D. TRIVEDI 1996: fig. 73). It belongs to a non-serial context. The
shape of the image with a tenon at the underside and the
missing vehicle make it improbable that it was part of a dikcpclln set.
A two-headed form of Agni (e.g. HARLE1963: pl. 143) became pop-
ular in South India after the 10th century and has therefore not been
included here.
Agni's right hand holding the rosary is generally raised, while
his left hand carrying the waterpot is held in a lower, extended
position. It has already been noted that the exceptional feature of
holding the left hand supporting the pot at shoulder level is clear-
ly concentrated in Eastern India.4g"nterestingly, among the con-
sulted texts (tables X-XI), this concept of Agni appears to be only
secondary to the concept of the deity carrying a spear (Snkti) or, in
one case, a trident (tris'ula). As far as we know, the trident does not
appear in art. The spear occurs in later (10th-11th centuries)
images in Orissa,4g7and it also represents Agni as a pictogram on
foundation stones (e.g. fig. l). A standard occurs floating behind
the god at Alampur (figs. 65,71,84), and a similar one is held by
the deity at Cittaudgadh, KBlikamBtB (fig. 104), which might be
regarded as symbolizing a smoke standard (dh6mncilzna). This
peculiar characteristic has been mentioned in connection with
Agni's chariot in the VdlzP (3.56.2). A somewhat different, un-
hasysma temple at Cittaudgadh (fig. 116), and a floatillg arrow
characterizes Agni at Satyavolu (fig. 88). For the unique occur-
rence of bow and arrow, one literary prescription of these attrib-
utes is known from a Jaina context.488 Except for the images from
the extreme south (figs. 289,291), the waterpot is never missing9
while the rosary has been at least once substituted by andher
attribute which looks like a fruit (fig. 104). It Inay be regarded as
a sign of impoverishment in the representation, w h e i ~Agni appears
with only one attribute, a lotus (?), in a c. 10th century image fro"
an unknown context (fig. 315).
Among the examples chosen for tablc XV, there is 110 alterna-
tive for the goat vehicle, which is rarely. absent. The unique
at AihOle
bination with a horse occurs in an early
(fig- 50). This vehicle ( h q a - horse) has been actually rnentiolled
4R"he sites are: MundeSvari (fig. 38; Asllelt ,980: pl. 53), P3hXrpLIr (fig. 235)'
BellisiEar (DONALDSON 1985: fig. 598), Gall3valli (fig. 243), Muklialinganl
SON 1985: fig. 61 1).
487
an image from Bhubaneswar of [he l lth century, which sllows a
armed Agni sealed on his vehicle, holding a rosary in the right and a spear with '
lnassive handle in his left (Cntc~logue~o,?dort1982: cat. 110.94).
4XX
GOSWAMI (1979: 288) in this connection rcras to [lie 15th cenlLIry J"na
text A c ~ r ~ ~ l i l l f l kbya m
Vardhamlnasnri, alLh0ug1~hc is llot awarc of the exislcnce
image 01"Agni with the attributes bow and arrow. AS (hc ~lmalidge$val.atemp? at
Sat~avoluhas no Jaina affiliation and its probable date is lllucl~earlier, this
reference cannot be linked to that particular image. Perhaps, both pieces of C V ~ ~ ~ ~ ' ~

identifiable elongated object is carried by the god in the Kumb- point 10 a common pool of ideas which were rarcly and u~~systcnialically expressed in
the art media.
 Table XV - Iconographic Chart of Agni

Site Vehicle Posture Headdress Peculiarites

Attributes
Alampur, rosary r. - saliznl>iida flaming hair halo of flames behind
Bala Brahmi (feet together) with ribbon the whole body, including
waterpot I.
(fig. 65) the head; moustache
floating standard 1.
AIampur, - snritapZ(la flaming hair halo of flames behind
rosary r.
Svarga Brahmi with diadem the whole body, including
waterpot 1.
(fig. 71) the head; ~noustache
floating standard 1.
goat seated on the vehicle in ja?Z/iiuku!a nimbus of flanles arising
dnksi!~al~ralnnfba]7~~la (hair crown) from the shoulders; goatee
(light leg pendel10, - head damaged -
with yognlla?!fl P-

vara MahZdeva
7 seated on the vehicle in
dakSi!i5prctlamb(~~~~d~7
ja!iiiizukw;c~ separate flames arising from
head and shoulders; goatee
walerpot 1.
(fig. 90) with yogapfll!a -head damaged -
~ariiapddfl ja~5aiuku!a double-layered nimbus of
Cittaudgadh, fruit? lower r. goat
Kiilik~m%t~ flames arising from the shoulders;
fan upper r.
(fig. 104) goatee, moustache
standard upper 1.
waterpot lower 1.
flexed posture, ja;alii~rk~r;a small flames?
Cittaudgadh, oblong object 1. goat
Ku~nbha~~ima, left hand on the left thigh goatee, moustache
exterior (fig. 116)
samap5da ja!iiinuku!a separate flames arising
Pang (fig. 146) rosary? r. goat
from head and shoulders;
waterpot - head damaged -
- (? damaged) l.
salllfl]~l,d~ jnfiiiizuku~a flames arising fro111the arms,
Gall%valli(fig. 243) rosary r. goat
shoulders, and from the goat;
waterpot 1. two separate fires above; beard
sanznlldda ja!iinz~rksr!a separate flames arising
Indor (fig. 153) rosary r. goat
from head and shoulders; goatee
- waterpot 1.
simple
Mul!de~vari, rosary S.
- squatting flames arising from
hairstyle head and shoulders '

siln~le halo of flames behind

(left leg pendent) hairstyle the whole body
niche figure (fig. 34) waterpot l.
on a low throne
seated on the vehicle in ja;iirilukuffl halo of flames behind
Osiiiii, HH-1 rosary r. goat
claksin~l>ral~n~bfl~7iidfl, the whole body; goatee
(fig. 189) waterpot 1.
with yogapa!!a - head damaged -
flexed posture j ~ flames
k arising from
Osiili, HH-2 rosary r. goat
head and shoulders
(fig. 206) waterpot 1.
ja!iinzuku!n flames behind the whole
pi'lall~~r(fig. 235) rosary r. goat snmnpdda
body, starting above the hip
- waterpot 1.
flaming hair separate flames arising
Sat~avolu(fig. 88) rosary (? damaged) r. - sanzapridfl,
with diadem with diadem from the whole body and
floating arrow r. fro111 the bow;
pitcher standing moustache?
at the bottom r.
- flaming bow 1.
- seated on a lotus in -damaged - faintly incised flames arising from
Ulllri (fig. 163) rosary r. head and shoulders -head damaged -
- waterpot 1.
dAkSi!~zpml(rnzbn~Zda

a 4n9 The line drawing of the image published by R.C.AGRAWALA (1965: fig. 2)
in 'he South Indian text KlLilidlzi/cim (15.16), which may point must have been prepared on the basis of the photograph in the first publicatio~l(S.C.
'lttk known, distinct tradition. KALA1954: pl. VI). However, something seenis to be concealed by the shadow in the
A vehicle is missing at AIalnpur and Satyavol~lin the soutll, at background, probably the goat vehicle. The image has been assigned a date around
Munde~vari,and, surprisingly, in a later image at Umri, where Agni 700 A.D.If nly identification of the attendant figurc in the LachchBgir image as an
ant~lropo~norpllic vehicle is corrcct, it resenibles the well-known "Geddes pane]" in the
thrones on a lotus (fig. 163). In two cases, an anthropomorphic van- cleveland Museum, assigned to c. 9th-10Lh centuries, O n which a theriomorphic goat
ant of the vehicle appears: at Lachch3gifl89 and at ~ i i r a d a k ~ n d a ~ and
~ ' two goat-headed humans occur side by side (J.Ph. VOGEL1933: pl. I).
in U.P.). Both are loose images and probably do not represent 490 VOGEL 1933: pl. 2a. The goat-headed (damaged),pot-bellied attendant, ,>rob-
*gni in his capacity as a dikpiilo. he much mutilated image of Ilolding the same attributes as the god, stands on the proper right of Agni. As on
the LachchBgir image, a human attendant is depicted on the opposite side. Both cases
mOnulnentalsize from SBrnBth/U.P. (fig. 314) could bear a similar ofan~]lropomorphic goat veliicles have been noted by P.K. AGRAWALA (1989-91: 7-8).
date as the one from Lachchagir, c, late 7th century. It forms allother
rare example for the absence of Agni's vehicle. Because of its large
size, which is not suitable for a niche image, it can also hardly be
regarded as a directional guardian. The two depicted birds, peacock
(r.) and cock (l.), are positioned behind the two personages flanking
Agni (probably representing his two s0ns)~91and are thus related to
them. The right arm of the god, which is only partly preserved,
hangs down, which is a rare feature.
The usual attributes of Agni, rosary and waterpot, give the deity
the apppearance of a Brahmin priest. Some of the other common
features, which stress this aspect, are the ja?6nzuku~a(crown of
braided hair), the beard and the stiff, hieratic posture, which evokes
an awe-inspiring presence of the god. Among the standing images,
the hieratic salnapdrla posture clearly prevails. The headdress
ja@nzukufa typically occurs in the north, while in South India, the
hair is always shown in the form of flames (figs. 56,65,71,88,268,
289,291,294). The earliest mention of "shining hair" (Soci;keSa-)
is already in RV 1.45.6. A later, Ceylonese text MafijuSrtb1zd;ita-
VastmvidyaSEstra 2.65a also describes this form of Agni (s'ikh6j-
vdlarn).
As most of Agni's features were determined to a considerable
degree by tradition, the artistic freedom was restricted to a few
details of the representation. The flames behind the deity which are
rarely missing, provided a good opportunity for individual varia-
tion. This explains the surprising range of non-schematic versions
for the flames, which sometimes assume the form of a halo, but
often vary in shape and size. The flames have a visual counterpart
in the frequently occurring goatee, an effect, which is obviously
intended. On the other hand, however, Agni's beard, which may
also be substituted by a moustache, is a sign of his priesthood and
his role as a mediator between the humans and the gods. This aspect
dates back to the Vedic times.
A feature which points to the ascetic aspect of Agni is the yoga-
paffa,a ribbon used for keeping the legs in a fixed position during
meditation. It is depicted in the images at Amrol, Diinebiibii temple
(fig. 97), and Osigii, HH-1 (fig. 189), which can both be dated
c . middle of the 8th century.
2.5.3 Yama (table XVI, see p. 99)
The main determinants, attribute and vehicle, of Yama's images
show extremely little variation. In most cases, a few individual
details in the rendering betray the wish to create a powerful, indi-
vidually conceived image of the mighty God of Death. One of
the earliest images, dating not much later than the middle of the
7th century, has an unusually intricate layout. In this image from the
Bjla Brahmii ten~pleat Alampur (fig. 66), Yama's record-keeper -
Citragupta as well as two palms are depicted in the background,
which lend the panel a partly narrative aspect. The depiction of
Citragupta is locally restricted to the neighbouring sites AIampur
and Kudaveli492 (figs. 66,72,80,85). These images are chronolog-
ically not far apart.
No alternative occurs to Yama's staff or sceptre, which is only
rarely supplemented by other attributes. The staff is mostly carried
in his right hand, while the mode of holding it in the left is largely
restricted to the early phase in Central India (figs. 9, lower row; 41),
with stray cases at Osiiifi, HH-1 (fig. 190), and Bhubaneswar
(fig. 42). The staff is typically, but not always, topped by a human
head, which was, after about 750 A.D., substituted by a skull. Unfor-
tunately, the respective portion rarely survives.
An additional attribute, the noose, is mentioned in some of the
early texts, viz. MtP and HayPa (see tables X-XI). It may have been
understood as a device to seize the souls of the departing persons.
Interestingly, the noose occurs only in Orissa, mostly as an
tional attribute (figs. 242,251), but in a singular case also as the only
attribute (fig. 260).493 Sword and shield, mentioned together in the
VdltP (ibid.), have not been portrayed in art.
Yama's awe-inspiring vehicle is the water buffalo, which al-
ways accompanies him in a;fadikpCla representations. The mighty
dark-coloured buffalo represents Yams's qualities to such a
degree4g4that it can be regarded as an indispensable part of his
hnage. Only in one case, at Indor (fig. 154), the identification of the
associated animal as a buffalo is doubtful, but the shape of the vehi-
cle is only roughly executed. It has a bear-like appearallce and
resembles N a i ~ t a ' svehicle on the same temple (fig. 155).
Yams's posture varies. He is found seated in various positions
and also riding his vehicle. However, no image is known to us ln
which he stands in the hieratic posture sanzrpEda, a fact, which is
quite remarkable.
A considerable variation occurs in the headdress of the god.
The cylindrical crown, which appears on relatively early images
(figs. 25,66,72), causes a certain resemblance with Indra. In the
first example, Yama is portrayed as throning in the "European fash-
ion" with legs kept apart and pendent, which illustrates his royal
character. A large nimbus has been added. In North and central
India, the demonic hairstyle irrdlzvakeia (hair standing on end) has
been mostly chosen for Yama, in order to emphasize his fierce
nature.
As a last point to be discussed, there appears another locally and
temporally restricted element in the image: a little bird perched
at the feet of Yama, which occurs at two Central Indian sites, D3lig
and Indor, and one Western Indian site, Osiiifi (figs. 147,1549 190)'
All these temples date to about 750 A.D. A similar bird is found
perching on Yama's attribute, the lotus, on an image at ~ittaudgadh
(fig. 105)$95 and as the only attribute of the god at ~ h a v i i n i p ~ ~
(C.P. ATIIERTON 1997: pl. 111). At least in the former two Cases, the
bird-motif is clearly connected wit11 an episode mentioned in MtP
11.11-17, in which SGrya gives his son Yams a cock ( k & ~ v ~ I ~ ~ )
removing the pus from his foot, which was injured as a result "
punishment by his mother.
2.5.4 N a i r ~ t a(table XVII, see p. 100)
The name of the fierce guardian of the south-west refers to a fede
deity mentioned in early texts, i.e. Nirrti, the ancient goddess
destruction and decay. However, the dikpzla called variouslY NirFtl
or Nairrta always appears as a male in art as well as in the COrre-
491 P.K. AGRAWALA 1994: ill. 83.
492 At Kudaveli (fig. SS), the small male figure floating in the background O"
socle SWangely holds a flywhisk in his right hand. The objcct in his left 'land
cannot be identified. At Alampur, Citragupta a pen (figs.66,72) or a manuscript
(fig. 80).
4" The reason for the substilution of the staff by the nooFe may probably lie in a
of the concepls of Yama and Varuna at the same temple (fig. 262). 'Ow*
ever, nearby r~presentationof Yatila only holds the staff (fig. 259).
494 Cf. L.VAN DEN BOSCII 1982: 45-48. The author refers to the dangerous aspect
buffaloes, which still adheres to d ~ m e s t i ~ n animals,
(~d and explains it as
"lame buffafaloesoften have wild blood, because wild bi~llsfi.equcntly mate with tame
COWS."
4'5 Similally, a skull Cup surmounted by a bird is held in Yama's left halld in a
loose image of about the 9111 ccntury, also from Rajas(llnn (Catalogue ,!.orld~lz1982:
cat. no. 93).
 Table XVI - Iconographic Chart of Yama

Site Attributes Vehicle Posture Headdress Cornpally

AIampur, staff r. buffalo riding cylindrical crown male attendant (Citragupta)
B& Brahmii (fig. 66) in the background 1.
Alampur, Svarga staff r. buffalo squatting on the vehicle cylindrical crown male attendant (Citragupta)
Brahmii (fig. 72) in the background I.
Alampur, ~ i ~ v a staff 1. buffalo squatting on the vehicle slightly conical crown male attendant (Citragupta)
Brahms (fig. 80) in the background I.
Amrol, RPmeSvara staff 1. buffalo squatting on the vehicle -damaged - -
Mahgdeva (fig. 91)
Bhcmarii (fig. 25) fruit? r. seated in the "European cylindrical crown two female attendants
staff 1. fashion" on a tilted throne (flywhisk-bearers)
Cittaudgadh, staff r. buffalo flexed posture lirdlzvakeia -
. Kalikslniit~
(fig. 105) lotus with bird 1. (hair standing on end)

Cittau?gadh, KumbhaB- staff r. buffalo flexed posture, left hand conical crown -
Ygma, exterior (fig. 120) on the left thigh
P i n g (fig. 147) staff r. buffalo flexed posture, left hand lirdllvakeia perching bird r.
on the left thigh
Indor (fig. 154) staff r. buffalo (? unusual slightly flexed posture, simple hairstyle with pecking bird r.
appearance) left hand on the left thigh skull diadem
MundeSvari, - squatting irdlwakeia -
fruit? r.
cornerstone (fig. 41) staff 1.
. Osilfi, HH-l (fig. 190) buffalo squatting on the vehicle irdllvakeia perching bird, centre
staff I.
OsiSfi, HH-2 (fig. 207) buffalo flexed posture, left hand diadem -
staff r.
on the left thigh
Satyavolu buffalo riding slightly conical crown female figure in the back-
staff r.
( D 1 1971: 43)
~ fig. ~ ~ ~ ~ ~ ~
ground 1, right side damaged
Umri (fig. 164) buffalo seated on the vehicle in iirdlzvakeia -
staff r.
daksi!ziipralanzbapa
1 I I l (right leg pendent) I I I

'p0nding texts.496 In the male version, he is known as the king of breast. However, the visual collcept of the dikpdlns in the narrow
r'k!asas, a particular class of demons. Rarely, the guardiall of the niches of the west-oriented temples at Osiiii must have been regard-
ed as particularly apt for variation.s00
South-west has been referred to as the god VinipRksa, with Nirrti as
consort ( ~ d h 3.57.1-3).
p This demollstrates a lingering ambigu- Nairyta's vehicles mostly appear in human shapes, but animal
vehicles likewise occur. Strangely, in the latter case, it is sometimes
ltY in N a i ~ t a ' sidentity with regard to his relationship with the god-
not clear if the snrize animal was intended to be portrayed, or
dess N i ~ t i .Among the images of Nairyta, the greatest degree of
whether a certain degree of variation occurred among the the&
variation Concerns the vehicle. This association wit11 differellt
vehicles is also expressed in the texts (tables X-XI). omorphic category. The anthropomorphic vehicle is generally de-
The indispensable weapon of Nairrta is the sword, which is picted as a simple male figure in various attitudes, adoration (fig.
and which is most frequently497mentiol~edin the texts. 106, 118), reclining (fig. 1961, or appearing in a curious "four-
The sword is always held in the right hand, which tells of a "rea-
legged" posture (figs. 203, 216). It is uniquely represented as an
armed warrior at Gallivalli (fig. 244); however, this variant form of
'lstic" collception of a common weapon. The defensive shield is,
however, generally absent. Only two images of those chosen for
XVII depict the deity with more than one attribute. Those out-
Stmding images belong to Indor (fig. 155) and Gallivalli (fig. 244).
In the first case, a noose498 appears in the left hand, while in the
case, a severed human head is carried by the god, as at
..
Gane~warpur in the same region (figs. 252,261). The noose is ran-
496 Even in the context of directional guardianship, the feminine gender is i ~ n p l i ~ d
in a few texts (table VI: MSS, last colun~n- intennediate directions; table XI:
'"Iscent of the iron bonds of the goddess Nirrti ($B VIL2.1.10)1 who Yogayalra).
"SO holds a noose according to V&P 3.57.3. It never gained any 497 When the guardian of the south-west is called Virfiplksa, his concept seems to

Popularity as Nairyta's attribute. The severed head, however, which
testifies to the fierce nature of the deity, Was introduced in
Orissa as well as in Northern and Central India during the following
period.499
The image at Osiiii, Harihara temple 2, is unusual illsofar as
be solllcwhat different, as he is once described carlying a staff (cln!tda;VdlP 3.57.1-7,).
495 It should be noted that in this case, the noose only consists of a loop alld has no
knot as in the case of the associated, neighbouring Varuly (fig. 156).
499 E.g. an image from Sulilni?dM.P., c. I lth century, published in
19-37-38: pl. Xb. The deity carries a shield in his upper left hand.
500 Another case of a unique concept of a dikpcila in such a position is the image

Nairrta leans on the sword which rests on the ground. The left ]land of Varuna on the Visnu temple 1 atosijfi (fig. 222).Here, the god is portrayed as stand-
ing on a pitcher.
lS also held in an exceptional position and gesture in front of the
Table XVII - Iconographic Chart of Nairrta

Site Attributes Vehicle Posture EIeaddress cornpally

Alampur, Svarga sword r. human skeleton riding open curls of -
BrahmL (fig. 73) (shown frontally) hair or cap
Cittaudgadh, sword r. anthropomorph flexed posture, left hand large, helmet- female
KLlikLmiitL (fig. 106) on the left thigh like crown attendant
Cittaudgadh, sword r. anthropomorph flexed posture, crown
KumbhaSyiima, left hand on the (later added)
exterior (fig. 118) left thigh
Dgng (fig. 148) sword r. unspecified four- flexed posture, left hand ir(l/zvc~lcei(~ -
legged animal on the left thigh (hair standing on end)
GallLvalli sword r. warrior samapdda conical crown standing
(fig. 244) severed head 1. (corpse) (lee1together) warrior r.
Indor (fig. 155) sword r. unspecified four- flexed posture LirdlzvalceSa -
noose 1. legged animal
OsiBfi, HH-l sword r. unspecified four- seated on the vehicle in ilrdhvalteSn
(fig. 191) legged animal rla1c~i~iil,ralan1I~a~~ri~ia
(right leg pendcnl) -
OsiLii, HH-2 sword r. man standing, right leg flexed, irdhvnkeia -
(fig. 208) (corpse?) left hand in front of breast
Umri (fig. 165) sword r. boar seated on the vehicle simple hairstyle -
in ~la/qi1~Epralanzba11Eda (damaged) -
an anthropomorphic vehicle is not unusual in O r i s ~ a The . ~ ~singu-
~ donkey (khara) is mentioned as Nairyta's vehicle. This variant of
lar version of an emaciated man or skeleton is found at the rather the theriomorphic vehicle seems to have an isolated occurrence at
early -late 7th century - site F\lampur (figs. 73,83). Aralaguppe in South India (fig. 136).
The male figure should be regarded as the anthropomorphic The stances of Nairrta vary in such a way that no particular
nara,so2 a creature, which has been mentioned in a number of texts seems to have been followed. Regarding the headdress, a predilec-
(table X). It clearly has a supernatural, demonic aspect, which is tion for open, upraised curls (crd/zvakeSn)can be observed in
probably alluded to in art by the Way the male figure is sometimes India. This coiffure characterizes the deity as demonic and pote1?-
peeping towards the onloolcer from behind. However, it can be tially dangerous. The facial features normally do not follow IhlS
regarded as a corpse ( b ~ a in) those ~ ~ ~cases where it is shown characterization, with one exception at Keldhar in central India,
reclining as at GallZvalli (fig. 244). NaiGa's vehicle is explicitly where Nairrta has a grotesque face with bulging eyes and a beard
described as a corpse in the South Indian text Uttarakiii?zikiiga~m (R.D. TRIVE~I 1990: pl. 103). Massive helmet-like crowns are some-
(cited after GOSWAMI 1979: 129), which does not belong to the cat- times worn by Nain:ta and the doubtful ilnage at ~ittaudgadh(figs'
egory of early texts considered here. A reclining corpse would actu- 106,113). In Soutll India, the deity has been portrayed with the
ally not be suitable as an active carrier of Nairrta. This is probably a1 collical crown (e.g. s ~ u RAIAN N 1986:
~ ~ 447, ~ left
~ side).
the reason why it was rarely depicted. However, it is possible that No particularly notewollhy cornpanion can be reported f
the anthropomorphic vehicle has to be regarded as a corpse (Sava) Naima, except for the standing at ~ ~ l l s ~ which
~ l l i almost
,
or ghost @ r e t ) "in action" in some Cases, and these are a befitting duplicates the vehicle of the deity (fig. 244). Like the latter. he wears
companion for the frightening god of destruction and decay. At the fird/zvake$ncoiffLlre.
Klampur (see above, figs. 73,83), the emaciated body of the stand-
ing figure carrying Nairrta may point in this direction.
The theriomorphic category of vehicles is largely restricted to
Central India m d early Osijfi (fig. 148,155,191). Here, in the 8th Regarding early images of the fierce goddess C%n~onc,iB, the pcculinr
t,izatio,l vehicle has been devribed by DoNAlosON(19PI: 123): dagger is
century sculptllres, it is generally difficult to identify the animal, oficn WOSI~at tl~eIlip, while the coiffure is frequentlydishevelled to suggest
''le

while in the 9th century, Nairyta's companion is clearly a boar (fig. corpse is a warrior OS LISLI~LL(demon)."
165).504 The specimens classified in table XVII as "unspecified 502 This translation ol the term, which also denotes simply "man", is COntaincd l"
standard dictionaries. E. RAVEN (1988) has discussed in detail the role of the as
four-legged animal" resemble a dog, ajackal, a bear or even a lioo.
ofthe di k,,zla Kuben below),
For the bear there is literary evidence in the HayaSir;apaii- 503 0,) [he jL,m.ua,lann as ..comr-vehicld. of the goddess, see T.E.
carEtra (table XI). The lion appears with a seated image from DONALDSON(~~~I).
Haryana,s05 the doubtful image of Nairrta at Cittaudgadh, Kumb- 504 0t11cr cxa~nplesoccur at Keldhar (TRIVEDI 1990: pl. 103) and Marlthero
haSy5ma temple (fig. 113) and in a much later, south ~ ~instance
d i (co'nmunicatcd
~ ~ by Prof. Dr. A.J. Gail), both assigned to the early and to the lntc 9th
at Tiru~afi1A.P.(K.V. RAMAN1985: 234, first of the group on century, respcctively.
mi c (1996: [is 4 ) mis image,seated on a lows a rnlining lion,
the right side). The somewhat related animal tiger is referred to only holds S W W ~(I.) and noose (I.), with its lloose as as solne other aspects bearing
in South Indian In the Indian mainstream tradition, the some similarity with the standing imngc at III~OIC (fig. 155). the vehicle of which'lay
dog is an emblem which belongs to the sphere of fierce deities. be a lion.
However' the association of a (log Or jackal with Nairrta Occurs 50"The expression iacirdliln occurring in thc Kl?ilirdhiltBra (15.24)
at OsiBfi9 Harihara K+aL[j7i&jrn (5.146). however,has a mulling includingtiger, pmthn, leap
1 (fig. 1 9 1 ) In AgP 96.28, the nrd, a,ld (lle ~ a b u ~ ocreature
us $arnblla,
 Table XVIII - Iconographic Chart of Varuna

Site Attributes Vehicle Posture Headdress Company

Alarnpur, noose r. - snrilapdda ja?cTntuku?a
- &]a B r a h ~ n (fig.
i 67) (feet togelher) (hair crown) -
co~lch1.
Alampur, Svarga - sarita~~drln conical crown -
noose r.
B r a h m (fig. 73) conch? 1.
Bsdirni, Cave 3, noose r. I lilaknm (~ilythical I seated on the vehicle conical crown -
I ll~fl!?&q~n
(fig.17) 1 floating trident r. I aquatic anilnal) I I l l
- bowl with gems? 1.
-
Cittauggadh, ii~nknm flexed posture -damaged -
noose? r.
Kglikimiti (fig. 107) lotus I.
Cittaudgaph, noose? r. ninkrrrn flexed posture, conical crown
Kumbhaiyirna. left hand extended
exterior (fig. 119) towards the vehicle
flexed posture, left hand lirdlzvakeia -
Pifls(fig.150) noose r. goose
-. the left thieh
nn (hair sta~idilipon end)
I I
l
GallPvalli (fig. 245) 1 wateroot? r. ( rliaknrn snn~upddn conical crown female figure in the
background r.
noose 1.
flexed posture cylindrical crown -
Indor (fig. 156) fruit r. goose
noose 1.
seated on the vehicle in ja?iirnuku[n -
Osizfi, HH-l (fig. 192) noose r. goose
fruit? 1. dnkSi!zriprnln~iiba/?Zd~
(light leg pendent)
slightly flexed posture, left cylindrical crown -
OsiBfi, HH-2 (fig. 209) noose r. mnknrcl
hand resting on tlie hip
seated above the vehicle lirdlzvnkein -
Umri (fig.166) noose (? damaged) r. riinknrn
ill (lak~i!~dpralnrilbn~~dda
+

2.5.5 V a r u ~ a(table XVIII)
Varuna, the God of the Waters, like Indra carries his ancient Vedic
attribute,in this case the noose (p&). With a rare, early exception,
which supplies him with a lot~ls(fig. 1 3 , this is his unmistakable
sign.
The noose was the god's illstrument to bind sinners. In AV
1V.16.~,Varuna's nooses are described as an intricate systenl of
'lings: "seven by seven, [they] stand triply relaxed, shining." In later
texts, the noose has oftell been characterized as consistillg of ser-
pents (ll~gfll.'d~a), but in art, this feature only rarely Solne
Peculiarities in the representation of the noose should be 11lelltioned
here. An early ceiling panel from the region of present-day Kar-
(M.BUSSAGLI~C. SIVARAMAMURTI 11.d.: fig. 21 6 ) interesting-
ly sllows the I I O O S ~floating in the background behind Var~lna'sright
arm. In the right hand, he carries a bunch of jewels. That is proba-
bly the ocean has been regarded as a source of jewels. The
god's left hand is empty. In another instance in the same region,
!here seems to be a tridellt floating behind Varul?a9sright arln, but
lt is rather faintly incised (fig. 17). The trident wo~lldbe ullique as
" recalling the anciellt Greek and Roman deity (Poseidon'
presiding over the ocean.
A general distinction can be made regarding the shape of the
Either, the noose has been conceived as a stiff, staff-like
Object with a handle or as a rather realistically conceived lass0,
however, wit], its upward extensio~~ counteracts gravity.
One
R-. -. completely realistic portrayal of the noose comes from the
aJara~eSvaratemple, Tafijsvfir (fig. 295). Here, it is not puffed up
" usual. In the 8th celltury in Central India, the mode of represen-
tation was also realistic, in that the loop portion of the lloose started
just above the hand of the god (e.g. figs. 156,316). The stiff version
("staff type") typically occurs in Rajasthan. Early examples exist
from the late 7thlearly 8th centuries at Cittaudgadh (figs. 107,119)
and Birha (MEISTER1982: fig. 8). Further, it occurs at Osisii
(figs. 197,203,2221, Khed (fig. 2301, Jhalawar Museum (fig. 317)
and at Markhe~a/M.P.~O~A singular, very early example is depicted
on the lintel from Pawiiyii ill Central India, c. 400 A.D.(fig. 9). The
"staff type" is not a I~omogeneousone. There are still - almost
individual - differences concerning the "handle" and the upper
part of the attribute, which sometimes takes the shape of a carpet-
beater.
Apart from the noose, there are a few objects which clearly
underlay regional predilections. At Biidiiini (fig. 17) and on the
loose panel from Karnataka, probably jewels are being carried by
Var~ina,while he holds a collch in the left hand on the ,&lampur
panels (figs. 67,74). 111Central India (figs. 156,316) and at Osigfi
(fig, l%), a globular fruit is the colnplernentary item, at least in the
8th century images. This fruit obviously attracts the attention of the
goose vehicle.
The waterpot as an attribute is not regionally restricted, occur-
ring at Gall,?valli (fig. 245) and in the image at the Jhalawar Muse-
um (fig. 317). A vessel filled with jewels (ratizap8tra) is actually
so7 On two Eastern Indian reliefs, Varulp holds a serpent (fig. 240; BIIAT-
T A C ~ ~ A R Y1993:
A fig. 3, fourth from the left). An iniage in tlie Orissa State Museum frolll
~iranllacandrapur-6B~ana. C. 14-15th centuries, shows llle noose represented by a
wound like a rope (DONALDSON 1987: fig. 3042). The image is re~narkablefor
the rare goose vehicle and has so far been wrongly identified as K5rttikcya.
'~l,ggnl~g.f~"
and related terms see tlie linguislic study by M.B. EMENEAU
so8 Colnlnu~licated by Prof. Dr. A.J. Gail; tlie date is c. 9th ccntury.
(1961)).
described among the four attributes for Varuna mentioned in the
VdhP (111.52.5; table X). The club (gadrl) however, which twice
occurs in the texts, and which was used as a weapon in combination
with the noose, is not found in art.
Only recently, it has been recognized that Varuna's vehicle in
art is not always the mythical aquatic creature makara, but that the
deity has also been portrayed with a goose, hamsa (VANKOOIJ
1985). The goose has been mentioned as the vehicle in the VdhP
(III.52.1), the BrS (58.57), and in the Nat (111.64). It is a remarkable
fact that this probably earliest injunction was applied in art in
Central India (figs. 99, 150, 156, 192, 316) and in the Mathurii
region (fig. 177), at the related Western Indian site OsiBii (fig. 192),
at Dev BarunBrak in Eastern India (ASHER1980: pl. 125) and at
Bhubaneswar in Orissa (fig. 42). This convention, which is already
traceable in a pre-dikpGla context at Deogadh in the 6th century
(fig. 3),509 in some of the mentioned regions seems to have ceased
in the 9th century. A much later (c. 14-15th centuries) Orissan image
raises the question about an actual survival of the early conven-
tion.510 All other images generally have the makarn as a vehicle.
A unique support for Varuna exists at OsiBii, V i ~ n utemple l (fig.
222). There, the god stands with his feet on a spilling jar, which
could be regarded as a symbol of the ocean.51' Exceptionally, at
Osiiiii (figs. 209,221), Varuna appears without any vehicle. As in
the just mentioned instance of the jar, the niche is particularly nar-
row. At Alampur, the god likewise stands without a vehicle (figs.
67,74), but in this case, some other deities also lack their carriers.
In an early (late-7th century) frieze at Paliiri (fig. 89), Varuna rides
an aquatic creature Galeblza - "water-elephant") consisting of the
forepart of an elephant and the hindquarters of a n~akara.Another
early, 6th-century piece from the ruined Jethiini temple at Tiilii in
the same region, Daksina Kosala, shows a unique version of Varuna
with two makaras in front of the god's breast, with their jaws
opcn.512
The postures of Varuna vary. No particular tendency can be
recognized.
There is also no general predilection for a special headdress of
Varuna. Such different characteristics as the demonic hairstyle
irdkvakeia and the royal cylindrical crown have even been chosen
on temples, which are not far apart geographically as well as chron-
ologically (Pang and Indor, figs. 150-156).
2.5.6 VByu (table XIX, see p. 103)
In the case of the Wind God VByu, some interesting observations on
different iconographic conventions can be made. Most remarkable
is the fact that an early convention in the portrayal of Viiyu
(fig. 318), with the billowing cloth or garment in his hands, has
survived side by side with another concept, at first clearly South
Indian, which assigns the flag or standard to this deity. In an often
cited description of V2yu contained in the VdhP (3.58.3), the god
is described with a gaping mouth and dishevelled hair. The inflated
cloth or garment, which he holds with his two hands, is here referred
to as vrlyu-rlpiritavastra (3.58.1-2). The Wind God Oado, portrayed
as running with a billowing cloth floating behind and with open
mouth and hair, frequently occurs on KusBna coins (e.g. TANABE
1990: figs. 4, 5, 17). As demonstrated by TANABE, this image has
been originally created on the basis of mainly Hellenistic models.
He has also shown that a simplified representation of the cloth
forming an arch above the deity must be a later Kusiinadevelopment
(ibid.: 62; fig. 19). In the above-mentioned c. 2nd-3rd century-
Gandhara relief, in which the Wind God has a somewhat dynamic,
but not running posture (fig. 318), the realistically conceived
lowing cloth is still reminiscent of the Greek prototype.
When it comes to the representation of the dikpiila VsYL1, the
arch-like, not the three-dimensional, version of the cloth has gener-
ally been chosen. With rare exceptions;l3 it is so much stylized that
wind as a natural phenomenon remains an abstract idea. The run-
ning movement and the dishevelled hair, which once supported the
impression of wind, has been abandoned in later art. Even in the
conception of VByu riding his vehicle, where movement could have
been indicated (figs. 193,320), the representation was strictly sty1-
ized and therefore rather stiff. A rather realistic treatment of the
billowing cloth can be found in the early 6th century image at
Deoga?h (fig. 3, extreme right), which is a pre-dikpdla context (see
section 2.1.3). Although Viiyu is shown in a flying attitude here,
both the cloth and the neatly arranged open hair remain colnpletely
unaffected by the breeze.
As a general rule, it can be stated that the "cloth type" of
is North Indian and the "flag type" South ]11dian,51"~owever includ-
ing Orissa. The second type was later introduced in North India,
where it seems to have gradually superseded the forlner.
Process probably started in the late 9th century. The "cloth type",
however, survived. One of its latest exalnples has been scl~lptedOn
the legs of the Yajiia VarBha at Khajuraho, c, middle of the lot''
(D. DESAI1996: 69). On the other hand, two Cases of OC-
currence of the "cloth type" can be observed in Southern India The
earlier one at Alampur is particularly important, as it represents "
rare combination of the cloth and the flag (figs. 68, 75).5'5 The
second occurrence of the "cloth type" in the south is found at
Bikkavolll in north-eastern Andhra Pradesh (fig. 284).
A loose image of a seated Valul~afrom Madhya Pradesh (fig. 316) has bzen
assignedto the 6th century by VAN KOOIJ(1985: fig. 10). The representation SO
IYrelated to the one at Indor (fig. 156) that it canllot bc regarded as chronologically
apart from the latter. In the loose imagc, the deity holds a similar noose in the left ha"d,
and the riglit also has a fruit, which attracts the attention of the goose. Both ilnagcs are
carved in relatively low relief, The face is palticuIarly flattish in the loose image, bUL'
with the simple, somewhat clunisy bhapes, this fcature seems to belong to the
rustic", provincial character of the image, which could belong to tlie 8th century as
510 The image of Varuna on harnsu from Birarmaca,drapur-6i5ana is kept in the
Orissa State Museum, c. 14-15th ceniuries (DONALDSON 19871fig. 3042).
5 1 1 The horizontally placed jar could refer to the celestial ocean with
Varuna is accociated in RV5.85.3 (cf. G.V. VAJRACJIARYA 1999: 55-56).
'l2 L.S. NIGAM 2000: pl. 21; see p. 52. Varunawears akind of turban andabroad,
ornate necklace. In tlie largely damaged panel hc is accompanied by a fclnale figure1l''
a male figure in arijcdi (worshipping gesture). =.S. RAO(2000) in the Sane volL'll~c
regards a sillgular large-sized standing multi-faced figure as a form of Varu?n In IllS
capacity as a lord o l magic (L.S. NIGAM2000: PIS. 33-38), whicl~is, for want any
parallel evidence, not very likely.
513 A SOlnewhal more realistic rendering of the inflatecl cloth - in a l ~ w c r ~ ~ ~ ~ ~ - ~
- has ~rcvailedin present-day Rajasthan: e.g. at Cittaudgadh (fig. 106), at
Harihara temples I and 2 (figs. 193,210) and at ,&beneri (B.S~rrAl1975: f i g 8). In
another rare exception, Vgyu is shown seated on a leaping antelope, the cloth
ing llaturally in the wind. This lively ilnage belongs to the Manu Lokapsla frieze at
LsmbgRajastllan, of about the middle of the 8th century (DIIAKY1966: fig. 4).
5'4 Here, PalSri in southern Madhya Pradesh (Dak$ina Kosala), forlns an
tioll. The late 7th century-frieze of r[i$ci[as includes the flag type of Viyu (fig. 89.
extreme left).
5'5 It SeelnS as if the artist has consciously blended the two traditions Aefihetl-
tally, the combination of both attributes in two hands looks awkward and is hardly sat-
isfactory. After Alampur, which belongs to the late 7th century, a similar phcnonlenon
of about the middle of the 9th century [Osigii, Visyl tel~lple2) most probably marked
the beginning of the introduction of the flag in North India. An interesting colltempo-
raneous case of intermingling both attributes can be seen in the same rcgion, e . 6 in Ihe
Jhalawar Muscum imagc [fig. 319). Here, the cloth, floating behind the hcad as is
in the region, is wound round a staff held in the god's left hand, as if alluding
flag.
 Table XIX - Iconographic Chart of VByu
Site Attributes Vehicle Posture Headdress Company
Alampur, inflated clot11 held in - satizapiicln slightly conical crown -
Brahml (fig. 68) both hands plus flag 1. (feet together)
Alampur, Svarga inflated cloth held in - sainapiida slightly conical crown
Brahmii (fig. 75) both hands plus flag r.
Amrol, inflated cloth held horse? seated in viitiia- -damaged - -
Dgnebibn (fig. 93) in both hands praln~n6apiida
(left leg pendent)
Cjttatudgagh, inflated cloth held antelope flexed posture, left - damaged - small worshipper r.
K l l i k j n l l t ~(fig. 108) in both hands hand on the left thigh
P26g (fig. 149) horse snnzapii~la central hairknot -
inflated cloth held
in both hands (damaged)
Gallfivalli (fig. 246) anthropomorphic slightly flexed conical crown -
goad r.
flag 1. vehicle? (head of a goat?)
Indor (fig. 157) inflated cloth held horse flexed posture firdlzvakeia
(hair stand~ngon end)
in both hands
Osiafi, HH-l (fig. 193) inflated cloth held antelope, which riding Crdl~vakeb -
in both hands resembles a horse
flexed posture conical crown -
OsiSfi,HH-2 (fig. 210) inflated cloth held antelope
in both hands
P2h5r~ur(fig. 236) - satnapiida firdlzvakeia male 1. and female
inflated cloth held
r. attendants
in both hands
holding scarves
pp
(fig. 293)
flag r. lion seated on a throne in
&kSi!ziipralai?ibapdrla
conical crown -

(r~ghtleg pendent)
-z

Umri (fig. 167) antelope

seated on the vehicle in -damaged - -
inflated cloth held
daksi!~@~ralaiiibal~dlla
in both hands

(fig. 181). Another singular version is known from Tirunelveli in the

The stylized billowing cloth in the course of time seems to have
been misinterpreted by t]lose who created the images. In one case of extreme south, in which case VByu has a reclining lion (fig. 293).
9th v ~ y u ' sposture and headdress have not been subjected to a par-
the arched object above VByu looks completely
ticular, wide-spread conception. There is also no regular companion
id and obviously represents a serpent (in the process of sloughing?;
T R 1 v e1983:
~ ~ fig. 62). to be observed. Only in one case, at PBhBrpur (fig. 236), a pair of
attendants is present, who similarly hold pieces of cloth. Their
the billowing clot11 u v ~ y u - ~ I ~ ~ i r i t a v aiss tdescribed
r~"
attributes resembling nooses must have contributed to the earlier
Only in the Vdl~t'(see above), d/2vaja(standard) or ~ a t 6 k a(flag) is
the attribute for Vgyll in the surveyed texts. In One
of the central deity as Yama.517
'nst"~ee, in the MtP, an elephant goad is also mentioned (table X).
In the Period under discussion, it was only employed in Orissan
Yt7 viz,at Gall,?valli and GaneSwarpur (figs. 246,254, 263)7 2.5.7 Kubera (table XX, see p. 104)
blned with the flag. A variant, a staff-and-~lotl~ combination, which
Kubera, the Lord of Riches, has sometimes been replaced by Soma
flag, can be observed 011 the Jhalawar Museum image
(Fig. 319). as the guardian of the north (see below). The great number of images
of this apparently very popular deily, which not necessarily always
Regarding V,?yufs vehicle, a horse was common in Central
portray him in his capacity as a dikpda, provides a good basis for a
India during the 8th century, This has not been noted before? al- study of the aesthetic concept of Kubera.
though K B H A T T A C H ~ ~ :~14'2)
(1961 ~ has made some Observatiolls The iconic appearance of this deity with its numerous variations
On this asPect.516 All other regions associated the Wind God with
is rather difficult to comprehend. His different vehicles and supports
antelope, Which is also mentioned in the texts. However*
In or, the absence of these devices, make Kubera the
cases in Rajasthan, where V ~ y uis depicted riding his ante- diklj61a with the greatest variatioll. C ~ l l ~ e r n i nthe g attributes, there
lope (figs. 193,320), the shape of the antelope is reminiscent of a is one main type prevalent in North India. Besides, a few other,
horses AS the first example belongs to the Harihara temple l at
Osip> which was obviously largely inspired by Central Indian
it seems plausible that the visual model of a horse had be 516 He observes that on the A~lkorVBt tempIelCambodia (first half of the 12th
adapted to the local tradition, which prescribed an antelope vehicle. Vjyu appears with a horse, but he is unable to cite any parallel case froln
"teelope appean earlier in Westem India, &g. at ~itta~ldgadh India.
(lig' log). The ollly surprising deviatioll in the vehicle occurs at 517 ~ 1 ~ ~ p j h j r p u r V B y ubcen corre~tlyidentified by DEMALLMANN
has (1963: 133,
n, 5) and BIIATTACHARYA(1987: 64, and m. 9, IO), who both refer to the earlier
Sesai in the early 10th c e n t q , where the associated rliakara points
interpretation as Yalna.
to the peculiar fact that must have been confused with Vam?a
Table XX -Iconographic Chart of Kubera

Site Attributes Vehicle or Emblem Posture Headdress Company

Alampur, club r. - seated on a throne (?) in large curly -
Bala Brahms (fig. 69) sack l. dnk;inZl~ralnnzbapZda coiffure
(light leg pcndent)
Alampur, Svarga club r. - seated on a throne in slightly conical -
Brahm5 (fig. 76) sack I. c l n k ~ i n ~ ~ ~ r a l ~ ~ n z O o p crown
~dn -
Alampur, ViSva club (? damaged) r. - seated on a throne in slightly conical -
Brahma (fig. 81) - left side damaged - (1ak;i~Z~~rnlarnbaI~cT~/acrown
Amrol, purse 1. anthropomorphic slightly flexed, right hand curly hairstyle -
DBnebibs (fig. 92) vehicle? (head of a goat?) resting on belly (damaged) (damaged) -
Biidimi, Cave 3, cup 1. goat or ram riding curly hairstyle female r. and male
nzal$<lnpa(fig. 18) with diadem flying in erotic attitudes-
BhEmari lotus bud r. three h r g e vessels besides seated in the "European curly hairstyle -
(figs. 26,27) and beneath the god fashion" on a tilted throne with diadem
Cittaudgadh, club r. ? slightly flexed, left conical
KumbhaSyBma, hand- on the lcft hi eh crown
exterior (fig. 121)
Gallivalli (fig. 247) small vessel r. tree with face? slightly flexed conical cup-bearing dwarf
club 1. (upside down) crown
Indor (fig. 158) fruit or cup? r. purse l. anthropomorph flexed posture simple hairstyle -
MundeSvari (fig. 40) small vessel r. - squatting simple -
purse 1. hairstyle
Osiiiii, HH-1 cup (? damaged) r. anthropomorph seated on the vehicle in curly hairstyle -
(fig. 194) purse 1. (/ak;inq~ralnnzDa[~dd(z
Osiifi, HH-2 (fig. 211) cup r. long purse 1. anthropomorph slightly flexed large diadem -
Umri (fig. 168) fruit or cup? r. anthropomorph seated on the vehicle in Lirdlzvakes'a (hair -
purse 1. llak;i~Zpralanzb~il~Z(/n standingon end)

completely different concepts occur in the north. The situation in
South India is less complex than in North India, with one type
clearly dominating there.
In the "Northern" type, Kubera is conceived as holding a cup in
his right hand and a purse in his left (e.g. figs. 40, 199, 321). The
shapes of the cup and purse vary, presumably in accordance with
local predilections, and the CUP is sometimes held on the left. This
'Urinking type" of Kuben is always depicted seated. In a closely
related type, the cup has been replaced by a fruit (F.M. ASHER1980:
pl. 187).It is difficult to decide whether this is a parallel or aderived
type, which might have developed due to a misinterpretation of the
- often very small - cup. In the "Southern" type, Kubera is simply
~rovidedwith a club (e.g. fig. 63). However, this seemingly clear
situation is somewhat obscured by the fact that, in a few cases, the
club is held together with the Purse in the nonh (e.g. figs 2, 182,
2 4 0 ) ~as~well as in the south (figs. 69,761. In this connection it is
important to note that the studied texts unanimously mention the
club as Kubem's attribute (tables x-xnI). Further, there are a few
images of Kubera which show a rare attribute or a rare combination
of attributes, e.g. only one lotus (fig. 26), or the cup and the club (fig.
247). In a singular image at Naresar in Central India, Kubera is par-
trayed holding a lotus bud in his right and a club (damaged) in his
left (fig. 322).519 Uniquely, a warrior with a similar club represents
the of god. He is flanked by
carrying the two
main treasures Kubera, the Conch (jahkhn, r.) and the lotus
( p d m n , l.).
In the "Northern" type, one particular feature illdicates that an
cup can be traced at least from the well-l<nown 2nd century P"i
Kherabowl support(HUNTI~~TON 1985: ill. 8.39). ~tcontinuedunt1'
the Kubera image in the dikpnln frieze at Barwiisiignr, c. goO A'D'
(fig. 179). The felnale figure clearly fits into the concept according
to which Kubera represents the enjoyment of rather worldly pleas-
ures.
The more freqllently sollthern 66clubtype9' also occurs at Cit-
taudgadh and AUW j (c.P. ATHE~TON 1997: pl. 56). On the exterior
of the 1<umbhviylma temple, the god cnnies a lnighty club
right hand as his only attribute (fig. 121). With a few otl~erfeatures
suggesting southern connections at Cittaudg;ldh (see section
2.3.1.3), some kind of influence or inspiration from the south
be assulned here. Finally, there are afew osibutes which appear
may"'
the texts but play no role in a n (tables X-XI), the spear and
conch, in addition to the respectively.
Regarding the vehicle, there are again basically nOrthern yd
southern tendencies, In the north,. the snthropomorph'c
being G6sora'¶ has usually been chosen, while in the south, tile
518 Sce R.N. MISRA 1981: figs. 8 (first figure from the right), 67. The
are not distributed ill a unifor~nway in two cases (fig. 2 ancl thc f o r ~ n e r a ~ ~ ~ ~ ) ~
club is held in the left arm with a purse in that band. In the second case (the latterabove)'
a smdler club is held in tllc right and ,hc pL,rsein tbleft hmd, ~ i l ~ f i images,
pa
of course, pda. the fii,,ilxiiu tradition.
51y The image is contained in the western bhadrn niche of temple no. l7 (TR'VE-
D1 1990: 51-53). The head portion has been badly damaged recently, as can be inferred

old concept was consciously retained for several centtlries. ~h~ from the illustration of the still co~npletcsculpture in TRIVEDI'S book (ibirl.:PI. 7)'
52" The female figure has been tentatively identified as the wine-goddess
figure of a wine-waitressSz0which appears by the side of the by C,B , ~ ~ ~(1996: , , fig,
~ . ~ ~ ~ ~
 (llze?a) generally appears as the carrier of the god. The ram was
later, c. during the second half of the 9th century, introduced in
the north as well, e.g. in the BarwhBgar frieze of C. 900 A.D.
(fig- 179), and at Osiiifi (fig. 202). Actually, the vehicle rzara is
by Nairrta and Kubera (cf. tables X-XI), which are both
characterized by a chthonic and an - at least potentially - fierce
" W t . It is therefore significant to observe that the representation
Nairrta's and Kubera's vehicles converges in some CaseS. Thus,
at Khed, Kubera's man-vehicle (fig. 232) is shown crouching and
peeping from behind much in the same manner as Nairyta's anthro-
Pomorphic vehicle appears at Osiaii (e.g. fig. 216). At Umri (fig.
168), it is shown reclining, which is usual in Orissan art521 and there
it is also found accompanyillg Nairrta (see above). In these Cases
convergence, Naiqta's vehicle is clearly differentiated from
Kubera's.
There is also a number of images, in which, quite peculiarly, a
has been omitted. Apart from AIampur (figs. 69, 76, 8 1 ) ~
where vehicles are missing for a number of other dik@las, too,
this is particularly striking in a few cases at OsiBA (figs. 201,219),
where vehicles are otherwise present with hardly any exception.
The platform separating the human vehicle from the god above at
the same site is likewise noteworthy (e.g. fig. 211). Another variant
shows Kubera instead of a vehicle associated with one or more
pots symbolizing his treasures (e.g. fig. 321). The 6th century
Image at Bhiimara (fig. 26) is the earliest of this kind from a &$ala-
Later examples also occur in ~ o r t h -Central India, even
ln cases where the dikpZ,8las are aliglled in a frieze and all the
Others are accompanied by a vehicle (fig. 180, 3rd figure from the
right).%
Another rather rare feature, the elephant as a carrier of Kubera7
has a north-western distribution,523 Its date of introduction is c-
half of the loth century. It does not appear on the temples
under survey.
The seated postilre can be regarded as particularly characteris-
tic Kubera. Thus, Kubera is so1netimes shown seated, eve11 when
the other clik]>iilas appear standing, as at iilampur (figs. 69,769 81).
Where this case occurs at Osisfi (figs. 182,198) and Pip5d (fig. 12419
however, the Kubera ilnages are probably not portrayed as direc-
tlO1lalguardians, as they are not positioned on the north. This main-
ly to the early phase ofdi/qjZla representation. Later, Kubera
is also depicted in similar stances as the other dilc1~iilas,but gener-
"IY not in the rigid snnznpaclu posture.524
Kubera's headdress is noteworthy, since the curly hairstyle
dominates (e.g. fig. 194). Even in the south, where general-
ly the rather undifferentiated high crown had been in Use, ilnages
Kubera with an elegant, curly coiffure are known (&.S. fig. 18,
69).52s These are apparelltly restricted to the early phase. Other
of headdress, including the demonic hairstyle, also occur in
the north.
Regarding Kubera's company, particular mention must be made
Of a f l ~ i n gcouple with erotic gestures, which flanks the god on both
'Ides in the nza!zc/al,a at BBdBmi, Cave 3 (fig. 18). Except for the
hove-mentioned figure of a wine-waitre~~, which generally does
Occur on panels of the dikpiila Kllbera,526 no regular entourage
"Pears.
2.5.8 Soma (table XXI, see p. 106)
Tflere are only three securely idelltifiable cases in the period under
dls~Ll~sion, in which the ~~d Solna (in other contexts genes-
" 1 ~called Candra) is represented as the sole regent of the north, viz.
at BithB (MEISTER1982: fig. 10), Cittaudgadh, KBlikBmBtB temple
(fig. 109) and Tafijjaviir, RBjarBjeSvara temple (fig. 300), probably
also at BBdBmi, Cave 3 (figs. 10,13), and KBficipuram,KailBsanBtha
temple (fig. 285). The loose terracotta image, inscriptionally
referred to as "Soma" (fig. 451, may either depict him as dikpdla or
Soma as a member of the related group of Manu Lokapalas, which
he belongs to as a regular member (see section 2.4.1.3). At Alam-
pur, Soma appears twice as a kind of additional guardian of the
North (figs. 77,82). At OsiBfi, the Moon God occupies a position on
the proper left side of the main deity and is therefore generally locat-
ed in the south (WESSELS-MEVISSEN 1994: 608). Here and at a few
other sites, representations of the Moon God also appear in associ-
ation with the dikpiilas (figs. 94 -doubtful -, 185), but the relation-
ship is sometimes not very clear (figs. 103,237).
The iconic appearancc of the deity does not seem to be affected
by its varying function. The god is generally characterized by a
crescent behind or above his head in North India, while in South
India, he has a round halo like the Sun God Stirya. At Alampur, both
characteristics are combined by integrating the crescent into the
halo (figs. 77,82). In the image of the Siirya temple 1 at Osiah (fig.
185) and in the loose terracotta image (fig. 4 9 , the crescent is
absent, while at Cittaudgadh, the god has been simply distinguished
from the other deities by a large halo as it is cornmon in the south
(fig. 109).
Regarding the attributes, a significant similarity between Agni
and Soma exists. Both deities generally carry the waterpot
(/camn!z&zlu) in the right hand and the rosary (aksar~ziild)in the left.
When this configuration is combined with the crown of braided hair
(jatiinzuIcuta), it clearly describes the deity as an ascetic or priest.
~ 0 t with
h Soma and Agni, the right hand is sometimes shown sup-
porting the waterpot from below (figs. 36,77). The vessel seems to
be absent only in the case of the ViSva BrahmB temple at Odfifi,
where the right hand of the damaged sculpture is resting on the hip
in kntilzasta (fig. 82).
Soina is often portrayed without any symbolic animal, but in a
few cases, different vehicles appear. Chronologically, these start
with Cave 3 at BadBmi (578 A.D.; figs. 10,13), where the guardian
of the north is associated with the lion. However, it cannot be
proven in this case, whether the deity is actually to be
called Soma. At Cittaudgadh, KBlikHmBtii temple, the sole occur-
rencc of a small animal resembling a dog or fox can be observed
(fig. log), while on the Harihara temple 1 at Osiafi a unique ar-
rangement of two geese carries the seat of the god (D. HANDA1984:
PI. 100). A single goose, which occurs on a loose sculpture from
Pengore in the same is probably also depicted on
521 For the representation of the iava-vfihaita (corpse-vehicle), cf. T.E. DONALD-
SON 1991, specifically fig. 15.
522 E.g. at Benisagar (T.E. DONALDSON 1985: fig. 599). Here, the standing deity
is flanked above and below by four inverted pots, with different substances issuillg from
I J I ~lower ones. Interestingly, the god holds a floral attribute like his counterpart at
Bhtimar; (fig. 26).
523 A. KUTAR(1983: 367-370) has collected sonie pieces of evidence froln
10tll/l Ith century-temple sites in Rajasthan, where the elephant is Kubera's exclusive
vehicle: Ahar, Mirl B8i temple; Iswal, Vi?fllt temple; Tusa, Stirya temple; Jagat,
Ambiki temple; Kiradu.
524 A singular exaniple for the sariiopiida posture exists at Gwslior, CaturbhLlja
lemple (fig. 176).
525 The curls are ralher small. SonletiIneS, tlle ilnpressioll of a wig or even a helmet
arises.
526 A exception occ~irsin tlie Ba~wiisigarfrieze of c 900 A.D. (fig. 179).
527 See R.C. ACRAWALA 1977: fig. 6; unfolt~tnalely,it is not known, which are tile
associated On the temp1e.
Table XXI - Iconographic Chart of SomaICandra

Site of SomaICandra Attributes Vehicle Posture Headdress Con~pany

(as Dikpiila, Associated
deity or Manu Lokapila)
Alampur, Svarga rosary (? damaged) r. - samapiida conical crown -
Brahmi (fig. 77) A (feet together)
Alampur, V i ~ v a rosary r. - santapiida conical crown -
BrahrnL (fig. 82) A waterpot 1.
Bitha (MEISTER - damagedlindistinct - - darnagedlindistinct - sama,fjda - damagedindistinct - two attendants
1982: fig. 10) D (indistinct)
CittauGgadh, rosary r. dog or fox? samapiida ja?cimukufn two kneeling
KLliklmZtti (fig. 109) D waterpot 1. (hair crown) worshippers
Gwilior, Teli- rosary? r. waterpot? 1. - samaniida conical crown?
kl-rnandir (fig. 103) A? (both hands damaged)
MundeSvari, rosary r. - squatting cap-like -
cornerstone (fig. 36) M
Osiiii, HH-1
waterpot 1.
rosary? r. waterpot? 1.
coiffure
-
-
two addorsed geese squatting ja~iinzuku[a
(HANDA1984: pl. 100) A (both hands damaged) below a moon seat
OsiiR, HH-2 rosary r. lion or goose? flexed posture ja?iimuku[a -
(HANDA1984: pl. 101) A waterpot 1. +

Osigii, HH-3 rosary r. - flexed posture jn[iinzuku~n -

(HANDA 1984: pl. 102) A waterpot 1.
Osiifi, ST-1 (fig. 185) A
--
rosary r. waterpot 1.deer or gazelle snmal~drln ja@nzukufa -
OsiLR, SV rosary r. -
-damaged - -
(unpublished) A waterpot 1. crossed legs
PBhLrpur (fig. 237) A? rosary r. waterpot 1. - samapiida ja$irnuku!u -
"Uttar Pradesh" rosary r. - seated with hairknot and -
(fig. 45) M? D? waterpot 1. crossed feet open curls

Harihara temple 2 at Osiafi (ibid.: pl. lol), but this is doubtfu1.528

other dikpdlas on the sanctum of the Kumbha6ybma temple at -
Uniquely, a rather tiny antelope or gazelle is depicted behind Soma
Cittaudgadh, Rajasthan (ill. 18, figs. 111,112). An explanation for
on Siirya temple l at Osiiii (fig. 185). In the case of the north(-east- this exceptional representation could lie in a existing
ern) image of the DiinebBbii temple at Amrol, which may represent
theological concept, according to which all ihe directional
Soma as an additional guardian of the north (fig. 94), the squatting
guardians are basically identical with the high god ~ i v a In .
quadruped below the seated deity is impossible to define. The
Ynnection, it is informative that the theo]ogical predecessor
attributes of the god, who is characterized by a yogapatta (yogic
Siva, Rudra, is known as diSd~?lpati (e.g. TS IV.5.2 a), Lord of the
band), are both damaged, but they could well have been waterpot Directions.
and rosary.
k n a is mostly four-armed and thus recognizable as a form of
Soma's posture is mostly the hieratic samapdda, which lends
h a . This, however, has not been the case in Orissa. he
the deity the aspects of dignity and stability. In two cases, at BirhG
form of k n a is referred to in some of the early texts (tables X,
and at Cittaudgadh (fig. 109), a symmetrical configuration has been
and lGna as the only four-armed god among the regents has been
created by the lateral placement of two attendant figures. The
described in the HnyaPa (28.12-13).
headdress often consists of a hair crown, but the conical ornament-
Unfortunately, at the early (late-7th century) site of Alampur?
ed crown also occurs in a few cases. Exceptionally, an individual
complete image of &ina survives. It is only clear that the Svarga
coiffure has been chosen for Soma (fig. 36).
BrahmB image is four-armed (fig. 78) unlike the rest of the group'
On the ceiling of the Piipangtl~atemple at pattadakal, f 6 m has been
likewise distiguished from the other dikpdldi by his four arms (fig'
2.5.9 & ~ n (table
a XXII, see p. 107)
61). Some other relatively early images, at Aihoje (fig. 54) and
Relatively few complete images exist of iSBna, who has been
represented as largely identical with the god Siva.529 Due to this
visual identity, loose images cannot always be safely identified as
a dikpZla. i l n a ' s main attribute is his trident (triifila). Interesting- 528 What KALIA (1982: 129, fn. 62) takes for a goose on the proper left side of
ly, an east-west distinction appears in Northern India, with the god may be actually a shrub. On the proper right side, there is an object which.hasbeen
interpreted as a small lion by HANOA (1984: 140). Because of the rough, lndisLinCt
trident being carried on the (proper) right side of the god in the
it is ilnpossible to come to a final decision in this case.
west and on the left side in the east. This distinction also applies to H. MEINHARD (1928: 26) has workcd on I&na and his rclationshi~will1
other images of Siva. 1 is further noteworthy that in an absolutely He states that i ~ ~ nasa a, directional guardian, bears close resembla~~~e with the latter'
unique instance, images of the k n a type replace some of the In contexts, iGna would be rather verblaJt, i.e. withoutany parlicularchracter's-
tics.
 Table XXII - Iconographic Chart of R3na

Site Attributes Vehicle Posture Headdress Company

Alampur, Svarga - snnznpdrla conical crown -
handled object lower r.
Brahmn (fig. 78) triiLiln (trident) upper r. (feet together), lower left hand
serpent (? damaged) upper I. on the left thigh
Amrol, Ram&- bull seated on the vehicle in jn~dntultu!a -
tris'ila 1.
vara Mahsdeva (damaged) r. viimnpralanzba~~dda (hair crown)
(TRIVEOI1990: pl. 26) (left leg pendent)
Cittaudgadh, lotus bud lower c. bull sljghtly flexed, lower left jafcinzukup worshipper,
K5liktim~t;i hand on the left thigh person feeding
.. r.
triiLila uuoer
(fig. 110) serpent upper 1. the bull
Cittaudgadh, bull slightly flexed, lower left ja!iirnukuin -
lotus bud lower r.
KumbhaSyiima, triiila upper c hand on the left thigh
interior (MEISTER
1991: pl. 670)
. .. l.
servent uooer

Gallfivalli (fig. 248) triiiln l. hull very slightly flexed, right hand ja!dntukn!n dwarf suppor-
extended towards bull (cbtiiiudrd) l
ting- the trident
Indor (fig. 159) bull slightly flexed, lower right jaidnz~~ku!a two male
rosary upper r.
halid extended downwards in attendants
triililn upper 1.
waterpot lower 1. the wishfulfillilig gesture
Osiiii, HH-I bull seated on the vehicle in ja!dniukufa -
rosary lower r.
(fig. 195) serpent upper daksit~H)~rnlani bal~ddn
- - r.
l I triiiln upper l. 1 I (right leg pendent) I I I
- fruit lower l.
Osijii, HH-2 flexed posture, lower left hand ja@~~tukuia -
lotus bud lower c. bull
(fig. 212) triiLiln upper r. on the left thigh
serpent upper 1.
Umri (fig. 169) bull slightly flexed, left hand on - damaged - -
triililn (? damaged) r.
the left hip

Amrol, RiimeSvara Mahiideva temple (TRIVEDI1990: pl. 261, how-

ever, show the deity as Because of the absence of the
trident, the north(-eastern) image of the DBnebiibii temple most
Probably does not depict &ina but Solna (fig. 94). Another Central
Indian two-armed example of Iiina is found at Umri (fig. 1691,
C.
quarter of the 9th century.
In the four-armed type, both the trident and the serpent are
generally held in (he upper hands of Ii311a. The trident appears on
-Proper - right in the west and in the south, while in Central and
+tern India it is canied on the left side. As a local variant, \~lhich
IS restricted to the region of modern Rajasthan, Iigna
a lotus flower in front of or beside his chest (figs. 110, 111,
122,212,220 (?), 233; D. MASON1993: cat. 58). This is mostly
'leld in the right hand, while in one image of the type" at -
Cittauclgadh it is carried on the left side (fig. 112). An image of
the same series additionally holds a rosary in the Same hand
(fig. 111).
A Central Indian ilnage from Indor roba ably depicts iigna with
"w"rp0t in his lower left hand (fig. 159). However, the scarcity
of this guardian of the north-east makes it impossi-
ble to tell whether this was the usual custom in the region.
Regarding f i ~ nposture,
~ ' ~ no particular position seems to have
been preferred. The rigid snmnIl~& as it occurs at Alampur (fig. 78)
Seems to have been rather rare. ~ 0 t the h vehicle and the headdress
Of k n a generally correspond to the usual signs of the god
Viz. bull and jatiinzukutn, respectively.
No recurring companion can be observed.
A inale deity holding a staff horizontally in both hands over his
knees occurs in association with the early dilcpiilns at BhfimarB
(fig. 28), MundeSvarI (fig. 3.9, and Bhubaneswar, ParaSuriimei~ar~
temple (fig. 42, right). It has been tentatively identified as
"Dandaplni".531 An association between Dandapiini and the
directional guardians existed in the 6th-7th centuries,532 but was
lost later on. It is not clear whether Dandapiini or Dandai~Byak~,
as he may have been alternatively called, was allotted any par-
ticular direction, although in the Viimnnal~urii!zn (32.22-23),
530 Later two-anned images come from Mathurl (fig. 178), Auw5 (M.W. MEIS-
TER 1991: pl. 478), the Pipaliidevi temple at Osiiili (A. KALIA
1982: 140) and Uniri (fig.
169).Exatnples for the two-armed Orissan k n a i~uagesare at Galllvalli (fig. 248) alld
GaneSwacpur (figs 256,265).
531 After I had considered such an identification sonic years ago, this name was
also proposed in a Museum catalogue (K. DEVA/S.D. TRIVEDI 1996: 50-51). Another
designation, or rather classification, suggested recently is ''si~ltl/~opu,u.p"(VS.
PATI]AK/S.K. SULLEREY1991: 5 1).
532 Tile associaliori appears to be llie closest at Bliubaneswar, where the god sits
directly besides Yama, i.e. on his proper left (fig. 42). At BhCtmarB, the image (fig. 28)
is larger than those of the extant other clikl~iilasYania and Kubera (figs. 25-27). At
Mut!?eSvari, the image of Dandap5ni (fig. 35) tliost probably belongs to the series of
which only a seatcd Agni relnains (ASIIER 1980:131. 53).
"Dandaniiyaka" is mentioned for the south533 in a directional 16 KBlaiijara (fort)/U.P., relief slab ("~aikunfhapatfa") show-
scheme. The remaining - all incomplete - dikpdla sets which ing deities including Dandapani, in sitii, no. KF 2389
include DandapBni, unfortunately do not contain any clue to a par- 11th-12th centuries, N.P. JOSHI 1992: pl. 49; S.K
ticular directional positioning of this deity. SULLEREY 1993: pl. XIII.
The identification of the so far unnamed deity as Dandapiini or 1 7 Tewar near Tripuri/M.P., relief slab showing deities
Danknayaka could be made on the basis of the mythology con- ~ i probably Indra and Agni, sit& c. 131h
ing D a ~ d a p i i and
nected with the ancient city of Vkznasi. A strong tradition links century, V.S. PATHAK/S.K.SULLEREY 1991: pl. 17.
this god to the well-known pilgrimage ~ e n t r e . ~On
3 ~a number of 18 P%tan/Nepal,ViSveSvara temple, niche image, south(-west)'
reliefs with schematic representations of this site (see below, nos. A.D. 1627, A. GAIL1984: pl. XXXIV, 4.
12, 13), Dandapsni is included, often in a position close to the cen- 19 P%tan/Nepal,Archaeological Garden, Lalitpur, loose image'
tre. This remained, however, unnoticed by the authors. Nowadays, c. 17th century, Catalogue Essen 1959: no. 32.
at VLBnasi, the deity is worshipped merely in the shape of a staff 20 Deo PBtan/Nepal, ilnage on a temple tower (s'ikhara), Fob-
(M.A. SHERRING 1868: pp. 62-63). The small temple housing it was ably 17th century, A.W. MACDONALD/A.V.
once located probably west or south (D.L. ECK1982: 2; 121) of the ill. 45.
original ViBvanatha temple which forms the spiritual and geograph- 21 Vjra~asi,Ke&areSvara temple, late (16th century or later?)'
ical centre of Viiriinasi. figure in niche inside the temple, squatting (com1nunicated
In order to strengthen the connection between the staff-bearing by Dr. Falk Reitz).
deity and the name Dandapiini, all known examples for its depiction
will be enumerated here, including later images as well. The list of A number of miniature shrines in the Allahabad Museum, dated
images will precede the iconographic description of Dandapsni: c. 10th century, exhibit an ilnage of the typical squatting ~ a n d ~ p ~ n l
above the door.535 It has been hypothesized that these miniature
1 ~ h k m a rloose~ , image with gavdksa frame, c. 520-530 A.D. temples "generally enshrining $ivalihga, [...l seem to be memorial
(fig. 28). monuments of funerary character" (K. DEVA/S.D. TRIVED1
2 MundeSvari, loose image, c. early 7th century (fig. 35). 1996: 51).
3 Bhubaneswar, ParaBur2meSvara temple, southern w d l of Da@ap?ini, as far as we know, does not possess any vehicle.536
wza!zGapa, small niche image, c. early 7th century (fig. 42, He is always shown seated. ~ 1 early 1 images, which are linked
right). with rlikl~nlasets (figs. 28,35,42), as well as the Benisjgar
4 Unknown findspot (North India), Bhiirat Kala Bhavan, (no. 6) sit on a low throne with lathe-turned feet. In the latter case'
Varanasi, loose image, c. 6th century (fig. 323). the staff is held in an oblique manner. As all other details, including
5 North ~ndia,National Museum, New Delhi, acc. no. 67.106, the position of the hands on staff, to the general ch"-
loose image, C. 6th-7th centuries, R.C. AGRAWALA 1970: acteristics of Dandapsvi, the image has been rnelltioned here'
fig. 1.
Perhaps a seated figure, which is conceptually associated with the
6 Benisagar (Benusiigar)lBihar, Patna Museum, acc, no. 61, directional guardians in the upper portion of the Naie6a pane1 at
loose image, c. 7th century, ASHER1980: pl. 103. Ellor2 29 (fig. 21, extreme right), also depicts ~ a n d a p a n i .
7 Vfiianasi, Bhirat Kala Bhavan, Varanasi, act. no. 1S, loose Apan from (he - generally - h o r i z o n ~ a ~placed
~ y staff, a regular
image with gavdkra frame, c. 6th-7th centuries, T.K. B-. feature of most of the Danaapjoi images is a hairdo.
WASIB.JHA lg85: P I XXXVIII, fig. 92; cf. K. DEVA1991: However, this has not been always portrayed in a uniform manner'
PI. 182 (here dated to the 8th century, which may be too Either the hair hangs down in tile form of short, big curls or strands
late).
(nos. 2,3,4,5,7,21), or it is given the shape of a curly wig (resern-
near RobertsganjlMirza~urdt. (U.P.), Allahabad bling that of a judge, nos. 1, 6). sometirnes the hair forms a small
Muaum, act. 9057
niche image on architectural knot on top, which also occurs in a larger version ( n o 4,20). R ~ Y *
fragment, C. 6th-7th centuries, K. Deva/S.D. TRIVEDI 1996: the deity wean a crown, i.e, a cylindrical crown at ~ e l l i ~ ~ g ~
fig. l56 (here dated 5th which may to be too early). (no. 6) and Baijnith (no, 12, f i 32fj), ~ the latter of which represents
Bhubaneswar, MukteSvara south wall, figure in a syncretistic icon. A crown of braided hair, jo?rinruku!o, occurs in
front of lattice window, c. 970 A.D., W. SMITH1994. one of the Nepal images (no. 19) and at Hiligl5jgadh (no. 11, fi6.
1 0 Siron Khurd/U.P., loose headless image inscribed on the
324). An explanation for the conspicuoLls treatment of the hair
pedestal, c. 10th century, S.D. TRIVEDI 1983: fig. 76.
11 Hiligl%jgadh/M.P.,framed image on architectural fragment,
c. 10th century (fig. 324).
12 ~ a i j n l t hKangralHimachal
, Pradesh, ~ i v temple,
a compos- 533 E.W. HOI'KI~S (1915: 112) to epic passages in which "Dnndapzni"
ite deity consisting Of h a , Visnu, and Dsu;ldapani, c. 13th Selves as a name of Yalna and Antaka. Theoretically, the variant form Da?danayaka
century (?l, V.S. OHRI1991: pl. 3.41 (fig. 325). could also refer to Yarna, particularly, as the dcity is positioned in the south here. 'OW-
13 Bad0h-Pa!hiiri/M,p., Archaeological M ~ Gwalior,
~ ever.
~ ~seelns 10~be no examl~le
~ for ,ssucli a referenc. in the dictionary (MoNra-
WILLIAMS). On the other hand, the extant dikpcilo sets at BhLimarB (figs- ZSdZ8)
285y "v'riinasi-iild~a?!~l" including DandapZni, c. Bhubmesuar (fig. 13 ~ ~ l oL~,at v yams ~~~d~~~~~a,,pear side by side Da?daPal?l
1lth century, N.P. 3 0 ~ 1991a:~ 1 p ~ I. and 1992: PI. 46. as a guardian of VBriipsi (sec below) clearly has a ficrce aspect, as hc is closely
14 Unknown findspot, Rani Mahal collection, Jhansi, act. ciated with Kila Bhairava (D.L. ECK1982: 198-201). This aspect befits the souLherp'
no. 13997, "V6rSinasi-s'ild1~attn"including Dandapsni, c. south-western direction.
".L. ECK1982: 198-201. According to KEJi K/zfl!f<ln32.162 of tile skr''7da-
l l th-12th centuries' JosH1 1991a: P" and 1992: ,XW~!;Y. the guardian deiLyD ~ ~ is&said, to~be stniol,ed~ ~ ~to llle sou[ll of viiveSvar"
pl. 47. the form of Siva who is thc main god of the city.
l5 Unknown findspot, State Museum, Lucknow, act, no. H, K. DEVI\IS.D.TILIVEDI 1996: figs. 157,323.
l989 relief slab showing deities including Dagdapcni, c. 53"t the feet of image no. 7 so~nclhingis seen provuding, which faintly
1lth-12th centuries, N.P. JOSHI1992: PI. 48. bles an animal head. However, even by a perso~lalexarninatio~~ of the piece, this could
not be verified.
may lie in the concept of the god, which has been regarded as
with HarikeSa Yaksa, whose name points to a particular
("yellowish") type of hair (ECK1982: 200).
Each one of the following features is shared by several images:
- pot-belly: nos. 4,5, 7 , 9 ;
- YogapntF: nos. 10,18,19,20;
- halo: nos. 4, 6; in some cases not clearly discernible;
- cross-belt over the breast: nos. 1,2, 7?;
- crossed legs: nos. 13,18,19,20.
Da?dap~l!i has often been portrayed with bulging eyes, and some-
tilnes also as bearded. Thus, some of the recurring features point
l0 his grotesque nature. His ilnage is not likened to that of the
cOmln~~l ynk,m type, which is depicted standing, but it lnaY be
regarded as related to the ilnage of Kubera, the ynksn king, who
is similarly shown squatting and pot-bellied (cf. R.N. M ~ ~ R A
1981:
ill. 7).
Apart from the known [unction of DandapBni as a guardian of
the site at VBrBpasi, its ilnage contains some elements of static
St"ilit~, which have been enhanced during the later iconographic
development by the introduction of the - horizontal - yogaPflfta
"d the cross-legged sitting posture.
His grip on the staff has been portrayed in different manners.
Mostly, both the hands hold it either from below or from above, but
there are exceptions. This shows that attention has been given to
the way in which the staff was held. Besides, there is at least one
instance of an obliquely held staff (no. 6).The horizontal or nearly
horizontal position of the staff may have signalled that the entry of
negative forces to that particular site was prohibited.
Conceptually, DandapBni, as a guardian of a particular direction
or of the site itself, seems to have lost its associatio~lwith the
clikpclns shortly after the introduction of vehicles for the latter. At
Bhubaneswar (fig. 42), different modes of support occur, including
Varuna as seated on a throne with his vehicle underneath, Yalna
(centre) as seated on his buffalo and DanGapBni as seated on a
throne.
Although the n~ythological and some of the iconographic
aspects (absence of crown, grotesque features) of Dandapiini make
him a minor deity, his fusion with Siva and Visnu at BaijnBth (no.
12, fig. 325) probably testifies to his close relationship with the
higher gods. Regarding his original function, it can be hypothesized
that DandapB~!i or DandanByaka was once a guardian of the south
or the south-west in a collfiguration of minor deities encirclillg
ViSvanBtha, i.e. ~ i v as
a Lord of the Universe, ultimately forming a
part of this universal deity.
 . . .. .. -
-W

-. --p-

" ' - .,.. --.L...I:"::--

" -. . - - . . .... ... --
. - -.- -. .... . . .. -
. ...-. ..- - F.>.i-3g:-
. . +S
.'-.
'-I.--.

.-.--__._ d.
 CO~C~US~O~
Both literary evidence and the visual representations of the dik-
/lakapfilas reveal certain degrees of variation. In the early Vedic
scriptures, the four main directions, generally including the zenith
the fifth coordinate, were clearly differentiated and assigned to
the guardianship of particular deities. Altl~oughthe positions of
these deities or overlords varied, Yams as the southern and Varuna
as the western guardian appear significantly often, especially in the
At~~arvavetln (table I). Most of the main gods of the Vedic pantheon
were classified in the directional scheme, which was later enlarged
"d filled up with other deities. There is no available evidence for
the existence of a parallel tradition of minor or folk deities during
this early period. During the time of the later Vedic texts, however,
the evidence from the two great epics can be regarded as a kind of
Parallel tradition, with tile four loknl,clas - the assigned directiolls
V"~ing - clearly dominating. The traditioll of four guardian yn1c;as
(minor tutelary deities), for which there is some evidence, may have
been another parallel tradition in this period.
In the PurBnic texts of c. first millenniuln *.D., the degree of
variation is definitely lesser than in the Vedic texts. Obviously the
c " ~ n i ~group had evolved during a process of consolidation, and
the individual characteristics of the rli/cl,filas then remained rather
constant. However, the V i s ! z u c ~ l z a n l z o t ~ n Y aa lvery
~ ~ rinforma-
~~~~,
and probably relatively "early" text, illcludes a number of rare
details in the description of the directional deities, some of which
have never been mentioned in any other independent text nor repre-
Sented in iu.t (e.g. a chariot drawn by parrots for Agni; table X). The
M ~ ~ t ~ ~ ais~particularly
, ~ ~ ~ f informative
i ! ~ ~ regarding the variety of Chatriirhi in Central India. From this inception onwards, the theme
contexts that the dilCpalas appeared in. They are llot only mentioned
inhabiting their particular cities in [lle mythical landscape which
Served as a cosmic scheme, but they also took an active part in
fighting evil forces, were worshipped during elaborate rituals, and
there existed several versions of descriptions of their characteristic
features.
Regarding the absolute c~lrono~ogy of textual references, there reliefs and on the temple exterior was common, the latter placement
is little evidence: A Mfl~al,lr,.&n passage describing the group of
four lokq~las (MP 124.19-24jprobably dates to the 5th century,
while explicit lnention of the canonical astndikpajas is made by the
6th CentLlry-authorvar%hamihirain his yogayfitr; (ch. 6, table XI).
It is, however, very likely that the assigllmellt of deities to the eight
started earlier or at least had earlier roots.
The deveIoplnellt of dikl,& representation in the period under
gradually underwellt different stages, referred to as phases.
The knowledge &out tile process of development depends on the
Wailable evidence, is particularly scarce in the period of
"Early Development", c. 400-650 A.D, ~~~i~~the sLlbseq~lent
centuries
" the evidence in some regiolls is not continuous, e.g. in Orissa,
where, after the first occul-rence in the early 7th century, a second
introduction of the theme seems to have taken place in the late 9th
century.
has been represented on
A group of four directional
Separate panels in the ~ ~ of which
~period, ? several
g examples
~ rule must have determined the placen~entof the Moon God on the
~ are
known (section 2.1.4). is a peculiar fact that the deities in this case
have identical features. A ~ such a ~concept ~is also expressed
~ ~~
in a few textLld passages, no actual iconographic prescription is
known for this case. rt is possible that tllese panels reflect the above-
mentioned folk tradition of four guardian yaksas, who were not
with individual traits since this was not prescribed foryflksas.
The depiction of individualized dikl,filns started with the lintel
relief from paw%ygol c. 400410 A,D., on wllicll the guardians
appear lined up beside a mytllological scene. Apart froln this very
early lintel, other depictions from the 5th-6th centuries occurred at
Bhirmarii, Elephanta, Ellorii and Bidiimi, showing the dikpnlas in
various modes of representation and visual concepts. While some of
these heterogenous beginnings had no successors, Elephanta, E]lors
and BBdimi became the models for future development, with cer-
tain modifications. A1 Elephanta and Ellorii, a limited (not corn-
plete) number of guardians is shown arllong the entourage of the god
~ i v on
a monulnelltal relief panels, which continued to be a mode of
depiction for the following centuries. At Biidiimi, the inception of
a subsequently very popular mode of dikpala representation, i.e. on
carefully laid-out ceiling panels, can be observed. Both sites, Ele-
phanta and BBdiimi, present some variations within certain limits,
which call be understood as an indication of the creative effort of an
incipient Stage in the development.
The evidence from the early 7th century shows that the direc-
tional guardians were mostly depicted on the telnple wall (wall
frieze -jalighz), which continued to be the common mode of rep-
resentation. This importallt developmellt occurred first, almost
simultaneously in the first decades of the 7th century, at MundeSvari
Hill, Bhubaneswar (ParaSurBmeSvara temple) and at Nsland:,
although ollly on the plinth of the temple at the latter site. There is
one Occurrence of about the same tilne of (only two) dikpzlas placed
higll on a temple tower (iikhnm) at Alampur, which had no lasting
influence in art, although other depictions on temple towers did
occur at a later date.
According to the available evidence, complete sets of the canon-
ical astadikl~cilasappear only after c. middle of the 7th century, first
at Cittaudgadh in the north-west, at Alampur in the Deccan and at
was introduced successively in other regions of India, though not
necessarily through a direct transfer of concepts. The number of
representations also substalltially increased. Only in the extreme
south, where matters remained unclear for quite some time, a full set
of dikpdlas cannot be conclusively ascertained before 1000 A.D.
While in the Deccan, the placement of dik1,dlas both on ceiling
Was, almost exclusively, practised at the temple sites in Central and
Western India. In Central India, there lnay have been older repre-
sentations, now lost, as successors of the early depictions at Pawsy%
and Bh6mari, but a separate development starting with the instal-
ment of only three d i l c ~ a l on
~ s the temple exterior at A~nrol,can be
clearly discerned in Central and Western India (PratihBra kingdom)
in the century.
At Alampur, the site in the Deccan with a number of complete
dilcpfila sets on temple exteriors dating as early as the latter half of
the 7th Century, both the Sun and Moon Gods are associated with
the dikj,iilas, as they have been placed in the corresponding niches.
Strangely enough, further to the north and slightly later, images of
either the Sun or the Moon God (the latter not in the position of a
guardian of the nortll) Were often associated with the rlikpdlas. This
occurred in the 8th century in Central and Western India, inostly OS,
temples with inconlplete, formative sets. In these cases, a distinctive
left side of the idol in the garbhngrlza, while the Sun God, when
present, was positioned on its right. Thus, a left- and right-hand
~ ~ I ~
symbolism was involved in the placement of the Sun and Moon,
which actually never influenced the positioning of the dikl>&rs
proper.
The indispensable inclusion of both Sun and Moon, however, is
found in the case of the related group of Manu LokapBlas (ills. 2,
31), which was depicted from the 7th through 9th centuries, but
generally in the form of a frieze and not in a specific directional nature of the dikpHlas was such that the group did not have to be
scheme. The Manu Lokapalas have been classified as a related necessarily complete, since its single lnembers already conveyed
group with partly substituted members here. Other groups of the intended symbolism. Moreover, the group of directional
dikpnlas with substituted members also occur (e.g. figs. 15-18, guardians was probably so popular that everyone must have been
where Kkttikeya replaces Yama), but these did not become as able to immediately recogllize them on a diagrammatic or narrative
popular as the Manu Lokapalas. panel.
Regarding the fully developed set of a.gadikpiilas, there were The absence of the necessity in depicting the complete SrouP of
still a few variations in their placement and their individual charac- directional guardians must have been a lingering phenome110n7
teristics. A number of examples are known for the distribution of the which for a relatively long period allowed the incomplete, substi-
group which started with Indra on the (north-)east, others, where tuled or otherwise varied representation of rlilpiilas, which
some members have not been depicted (ills. 28,30); further, spa- gradually ceased because of the structural ~~ecessity to accornmo-
tially restricted representations of dikpiilas (figs. 203, 204) and date eight related deities at the corners of the te~nplewalls or
seemingly degenerated, hybrid images (e.g. fig. 181) have also been eight outer panels of a decorated ceiling bay, This lack
noticed. of numerical determination of the group c o ~ ~ actually
ld have pro-
There was obviously a gradual increase in the number of dikpiila moted its representation, as ample scope was left for variation.
sets on a temple or temple complex, which, in the case of a which eventually concerned not only the nulnber or deities but 'ls0
paiciiyatana set-up, could have accoinmodated up to five sets on its other aspects of the depiction (also ellhanced by varied literary
walls. ~ o w e v e rfor
, this theoretically possible number, no example descriptions). ~ h uthe ~ dikpclas
, for quite some time held the posi-
can be cited. The choice of several diklxila sets must have been tion of "variablesv in art.
made in order to protect all parls of the temple by the guardians. When a set has been described here as "formative", this is on'y
ino or variations in their iconography were often made (e.g. at from the chronological point of view: in all the enumented
Ganeiwarpur, figs. 249-265).
of representation of the directional guardians, the tendency towsds
~ikpHlaswere also included as attendant deities on panels the depiction of a complete is clearly visible - it was only
depicting one of the main deities of Hinduism. In this mode of achieved at different times, and even given up in cenain cases
representation, it is a significant fact that the complete number of later on.
guardians was rarely shown. This fact could be interpreted in the
Regarding the relnaining two deities, ~ ~ for the~zenithhand ~ s
same way as the fact that from the Very beginnings of the depiction V i g y for the nadir, their inclusion has only been reported fr0In
of the theme, a restricted number of members, sometimes only two,
GhlinerSvIRajasthan. It is noteworthy, however, that in the
quite often 0c~urred.In complete contrast, (?.g, to the depiction of
diagrammaic ceiling panels, ~ ~ clearly
h dominated
~ the
a central
the group of planetary deities (~zavasrahas),which were always had been
in a set*this
position (fig. 13, 14, 47, 48), which he most
was probably sufficient allotted becaLlse of his guardianship of the zenith. In the examp?
for a reference to the group of directional guardians. A concept
from the Pth-IOlh centudes, the central deity is generally god SIVa
that has been suggested by the 6th century text YogayHtr~,where it
or, in the case of laina temples, a lina or Yaksa (ill. 23 a, f). Flow-
is described that a military expedition had to carry an image of
ever, one case of a central Brahma is known from the later
the guardian of the direction in which it was proceeding: the very
period (fig. 130).
 Merkmalen, darunter die verschiedenen Welten-Spllaren, den Rich-
tullgen zugeordnet, was offensichtlich die Vorstellung von den be-
treffenden Gottheiten nicht nur als Hiiter der jeweiligen Himmels-
Die illdischen Richtungsgottheitell werden im Sanskrit als dkpfilas richtung sondern auch als Hiiter eines bestimmten Welten-Sphare
(Richtungshiiter) oder als loltnpdlas (Hiiter der Welt oder einer be- begriindet hat.
~timlntenWelten-Sph$re) bezeichnet oder mit noch anderen, ~neist Die hier als ,,Manu LokapBlas" bezeichnete Gruppe, die in der
verwandten, Namen belegt, Der Begriff asgdikpdlS!z bezeichnet bildlichen Darstellung eine gewisse Bedeutung erlangt, und die sich
die am hiiufigsten auftretende Gruppe der acht Richtungshuter gott- von den oben genannten Gottheiten dadurch unterscheidet, dafl in
licher Natur: ihr anstatt von Nairrta und ISana Siirya und Soma enthalten sind,
sleht besonders deutlich in der Tradition der Spharenzuordnung.
Indra - Osten In der literarischen Uberlieferung des ersten Jahrtausends nach
Agni - Siidosten der Zeitenwende sind das MatsyapurC!za und das VisnucElznr~notta-
Yama - Siiden rap~irfi!zabesonders erhellend fiir die Tradition der as[adikpfilas. Im
Nairyta, auch Nirrti genannt - Siidwesten ersteren werden die Richtungsgottheiten in unterschiedlichen Kon-
Varuna - Westen texten geschildert, die ihre vielfaltigen Rolle im mythologischen
vsyu - Nordwesten Bereich und ihre grolje Bedeutung in bestimmten vorgeschriebenen
Kubera oder Soma - Norden Ritualen der Zeit darlegen. Im letzteren Werk, das in der Fach-
ISBna - Nordosten Iiteratur bereits vie1 Beachtung gefunden hat, sind besonders aus-
fuhrliche Beschreibungen der Richtungsgottheiten enthalten, die
jedoch ungewohnliche Ziige offenbaren und in vielen Fallen ohne
Zuwei~enwerden die acht horizontalell ~ichtungskoordillatendurch
parallele sind. Im Vi~!zu(~Izar~izottarap~ird~za sind auffallend viele
die Koordinaten auf der vertikalen Achse erganzt, denen ebenfalls
der dikpdlas (Indra, Agni, Yama, Varuna, ISana) als vierarmig be-
Gottheiten zugeordnet sind.
schriebe~l(sonst genere11 im ersten Jt.n.Chr. zweiarmig). Sie treten
lneist Init einer Partnerin, z.T. auch noch Init anderen Begleitfiguren
Brahma - Zenit
Ananta - Nadir auf.
Als Beispiel fiir eine knappe, aber haufiger anzutreffende Kon-
ent- Zeption der astadikpiilas sol1 eine Stelle aus dem Matsyapurfi!ln
Aus der friihvedischen Uber]ieferung 13Rt sich bereits
( ~ t 260.65-261.23,
p S. table X, p. 15) angefuhrt wer den:
wickekes BewuBtsein fiber die Himmelsrichtungen, erganzt durch
Zentrum, Zenit und Nadir, sowie iiber bestimmte, den Richtungen Gottheit Richtung Gefiihrt (vtihana) Attribute
zugeordnete Oberherrn oder Schiitzer (ci&il)atis), ableiten. Indra Osten briinstiger Elefant Donnerkeil, blauer Lotus*
Schutzbediirfnis kann sowohl auf die magisch-religiose als such Agni Sudosten Ziege Speer
letztlich auf die weltlich-milit$rische Ebene bezogen werden. Yama Siiden Wasserbiiffel Stock, Schlinge*
In der vorliegenden Arbeit, deren Schwerpunkt auf der bild- Nairrta Siidwesten anthropomorphes Schwert, Hammer*
lichen Darstel~ungder Ric]ltungsgottheiten im ersten ~ahrtausend Wesen, bzw. von
".Chr. liegt, werden sowohl schriftliche als auch bildliche Zeug- diesem gezogene
herangezogen, erstere jedoch haupts8chlich zur Erhellung des Kutsche*
Varuna Westell Fisch* Schlinge
SPezifischen kulturelIen Hintergrundes, aus dem die Bildzeugnisse
VByu Nordwesten GazelIe Flagge, Standarte
hervorgingen. Die Behandlung der s~hriftlichenQuellen ist unl- Norden v011 Keule
Kubera
fangreicher und starker analytisch ausgerichtet als vergleichbare anthropomorphen
VOrlaufersludien (besonders ware zu nennen die Dissertation zLlm Wcsen gezogene
gleichen Thema von R.P. GOSWAMI 1979537). Kutsche
Die Gottheitell, die als Hiiter der Richtungen fungieren, zeigen ISBna Nordosten Stier Dreizack
die gleichen Charakteristika wie alle indischen Gottheiten des brah-
manischen Pantheons: Sic besitzell vorziiglicl~e~orpermerklnale~ In diesern Beispiel sind alle Informationen iiber die c/iklxjlas ent-
halten in ihren Handen spezifische, ihrem Charakter ulld her llalten, wobei im Falle des Attributs haufig sogar zwei angegeben
Fullktion angemessene Attribute und werden rnit eilleln Reittier, sind, entspechend beiden Hiinden der zweiarmigen Fornl. Ein
das zuweilen durch ein unbelebt.es Objekt ersetzt werden kann, Asterisk (*) kennzeichnet hier die in anderen Beschreibungen sel-
dargestellt. ten vorkornmenden Klassifizierungsobjekte.
Wahrend der frfih- und der sp$tvedischen Periode (ca. 150°/ Die sich in den Puranas weitgehend stabilisierende Konzeption
1200 v.Chr. bis zur Zeitenwende) lassen sic11 verschiedene, neben- der astadikj~dla~ findet, mit einer gewissen Variationsbreite, ihren
einander existierende Versionen des Systems der ~immelsrichtun- Niederschlag in der Kunst. Die Wurzeln der dikl78ln-Darstellung in
gel1 und der ihnen zugeordneten Gottheiten erkennen (tables I-V1!. del Kunst der brahmanischen Religion liege11erstens in der friihen
Die Anzahl der Gottheiten betrggt zun$chst nur vier oder fiinf, mlt Konzeption von vier yaksas (iibermenschliche, halb-damonische
dem dazugerechneten Hiiter des Zenit, aber es wurden such sechs Wesen) bzw. ya1c;a-Konigen, die Schutz gegen Gefahren aus den
Oder siebell Richtungslliiter beschrieben. Dabei scheint die Uberlie- vier Himmelsrichtungen bieten; die friihen Darstellungen der
ferung des Atlmrvnveda (table I) relativ eng mit der spater kanoni- yaksas Kubera und VirCidhaka am buddhistischen stiipn van
Schen Version (S. eingallgs) "erwand[ zu sein. In der spatvedischen ~ h ~ r h ausu t dem 2. Jh.v.Chr. sind die wichtigsten Zeugnisse hier-
Literat~lrfinden sic11 das illteressante Phiinomen der doppelten Zu-
Ordnung von Goltheiten zu einer Himmeldchtung, sowie bereits
537 Der Autor lrug ein u~nfangreichesMaterial sowohl textlicher als auch bild-
weit entwickelte Ansatze zuIn Achtersystem der Richtullgen und
der Zugewiesenell Einzelgottheiten. In dem sic11 zunehmend licher Art zusammen, bearbeitete es aber nur in deskriptiver, nicht in vergleichender
ulld analytischer Weisc auf.
verkom~li~ierellden Richtullgssystem wird eine Vielzahl
 Gruppe, aber auch andere, wohl ka~imassoziierte Gottheiten treten
in korre~~ondierenden Nischen auf. Dieses Phanoinen gehort in die
Entwicklungsphase der rudimentken dikpBla-Gruppierungen, die
nur in Zentral- und Westindien belegt ist.
Ein Fall von ,,stabiler Substitution" tritt bei der den dikpalas
verwandten Manu Lokapslas (hier SO benannt aufgrund ihrer einzi-
gen textlichen Erwahnung) auf. Hierbei fehlen Nairrta und IiBna,
Und SCirya und Soma, Sonnen- und Mondgott, komlnen hinzu. Die
G m ~ ist ~ eein relativ hhfiges Thema von Bildzeugnissen und
erscheint zuweilell mit dikpclas an eineln Tempel. Soweit bekannt,
~ i n ddie Manu Lokapglas keinen bestimmten Richtungen zugeord-
net - nur in Munde$vari Hill hat ursprunglich eine rhmliche Anord-
bestanden, die nicht erhalten ist. Ein eigentiimliches Merkmal
der Manu Lokapila-Gruppe ist die Bildung von stabilen Zweier-
paaren (SGrya und Soma, Indra und Agni, Yama ulld Kubera,
Varuna und VByu), die such bei der hbfigsten Abbildungsform, der
einfachen Aufreihung, erhalten bleibt (ill. 31). ~nscheinendverlor
die Manu Lokapda-Gruppe im 10. Jh. an Bedeutung.
Gegen Ende der bearbeiteten Periode, im 9.110. Jh. in Zentra-
lindien, treten einige Bildzeugnisse auf, die man als hybrid bezeich-
nen kann: und zwar wurden Merklnale der Gotter Varuna und VSyu
miteinander kombiniert (fig. 181). Dies betrachteten wir als Merk-
mal einer gewissen Degeneration der althergebrachten, iiberwie-
gend zweiarmigen dikp&s, die gegen Ende des 10. Jhs. in Zentral-
UndWestindien von einer neuen Generation von vierarmigen Rich-
tungsgottheiten abgelost wurden. In Orissa bestand auch weiterhin
die zweiarmige Form, wahrend sich im extremen Siiden nun erst die
friihesten vollstandigen Gruppiemngen von dikpalas zeigten. Im
Dekkhan blieb die Darstellungsweise, die schon in BBdBmi 3 be-
griindet worden war, weitgehend stabil. Die Richtungshiiter in ihren
quadratischen Decken-Kompartimenten wurden im 9.110. Jh. schon
haufig vierarmig gezeigt.
Wahrend die Abbildung am TempeliiuBeren der tatsachlichen
Richtungszuordnung der Gottheiten nie exakt Rechnung tragen
konnte, war die Abbildung an der Tempeldecke eine Darstellungs-
weise, die einer korrekten Ausrichtung in idealer Weise entgegen-
kam. Das Zentrum dieser Deckenanordnung nahtnen die Getter
BrahmB, visnu oder ~ i v ein.
a Bedeutungsvoll ist dabei, daB Brahmg
und V i p , die schon in BBdBmi auftreten, weiterhin mit einer spe-
zifischen Anzahl kombiniert werden: BrahmB ist zunachst wieder
von vier lokaphlas (ills. 5, 6), Visnu aber, wie schon in BBdBmi,
dann auch in Paftadakal, von acht Gottheiten umgeben (ill. 8). ~ i v a
als Mittelpunkt des dikphla-Deckenpaneels wird von der Mitte des
8. Jhs. an popular. In dieser Anordnung erscheint er oft als NafeSa,
der aufgrund seiner Vielarmigkeit symbolisch mit der Herrschaft
uber die acht Himmelsrichtungen in Verbindung gebracht wird.
Die extrem seltene Abbildung aller zehn Richtungsgottheiten ist
nur am Jaina-Tempel in GhBnerBvIRajasthan belegt (bisher nicht
dokumentiert).
Die Anzahl der abgebildeten dikpiila-Gruppen pro Tempel-
komplex steigt in einigen Gebieten im Verlauf der beschriebenen
Entwicklung allmahlich an, so daB in GaneSwarpur am Ende des
10. Jhs. wohl vier von sechs an den AuBenwanden moglichen Grup-
pen reprasentiert waren. In dieser Weise wurde der symbolisch-
dekorative Schutz der Tempelwand durch die Richtungshuter auf
alle Elemente eines Tempelkomplexes ausgedehnt.
 Bibliography Hayas'ir~apaficardtra[HnyPa]
HayaSir;apaAcardtra ( A d i - ~ i i ~ d aed. ) , Kali Kulnar Dutta Sastri,
with iconographical notes by Kalyan Kumar Das Gupta (Bib-
PRIMARY SOURCES: liotheca Indica, 306), Calcutta 1976.
Jainziniyn-Brdlz11zn!za [JB]
AgllipurG!za [AgP] $ri~lnddvail~Gya~za~lzu~zipra!zita~n aglzi~urci!zam, Jairnirziya Brdlzrna!za of tlze Sanzaveda, ed. Lokesh Chandra/
~ 1 1 a n d 8 i r a m a s a m s ~ t a S r a ~ ~ t41,
h ~ Poona
l i h 1900. Raghu Vira, second revised edition, Delhi 1986.
Aitareya-~r~lznza!za [AitB] Jainziniya-Upalzisad-Brdlznza!za[JUB]
Das Aitareyn-Bralzmn!za. Mit Allsziigelz sus de17z Conzazerztare Jnin~ifl?yar;eya-,Jailliilziyopar2i;ad-BrdIz111a!zas,critic. ed. Be]-
von jdya!ziiciirya u~zdarzderelz Beilagen, ed. Theodor Aufrecht, likoth Ramachandra Sharina (Kendriya Sanshit Vidyapeetha
Bonn 1879. Series, Nos. 5-6), Tirupati 1967.
Aitareya-Uparzi~ad[AitU] Kqi.~;lzala-Ka~ha-Sni~zI~itG [KapS]
Aitareyn Ul~nlzi;n~, ed, and transl. Lilian Silburn (Les Upan- Kapisthaln-Katlza-S~1nlzit6, ed. Raghu ~ i r a~, a h o r e1932.
ishad, Texte et traduction ~011sla direction de Louis Renou, X), Kds'iKlza!zda
Paris 1950. KGii Klza!zda. Gurumandala GrantharniiIByI, No. XX, Vol. Iv,
A17ast~zba-~lzar11zasitra [&DlzS] Calcutta 1961.
Al>astanzba Dlzarnlasatra. Wit12 the Corl~nzerztaryUjwala by jrl Kiiflzaka-G!"lzyflsitra [Kii117GSl
Hnra~lattaMisra (Kashi Sanscrit Series, 93), Benares 1932. Tlze K ~ l z a l a g ~ h y a s ~ iwith
t r a extractsfronz tlzree conzrizentaries,
Al)flsta~zba-~rautasfitra [&is] an appendix and indexes, ed. Willem Caland (Daygnanda
Apasta~izbairautnsfitra pvitlz tlze ~17irtasv6nzibhdsya,ed. A. MahiividyBlaya Sams&tagranthamalii, 9), Lahore 1925.
Chinnaswami Sastri, Vol. 1-11(Gaekwad's Oriental series, 121, Kii!ha1ca-Sar?21zitG lKS1
(Yojurvediya) Kii~liakasarizhitd,ed. Sripad Damodar Satvalekar,
142), Baroda 1955, 1963.
Parni 1983.
Art11aJ'L7straof Kautilya [Artla]
Ka~ltiliyan~ Artliaidstranz, ed. R. Shalna Sastry, Mysore 1924. KauSika-Sitra
TIze Kau~ika-SCtraof the Atlzarva-Verln, pvitlz Extracts frollz tlze
A s ' v a l ~ ~ a ~ ~ a - ~ r ~ z [As'vGS]
Yasitra
C ~ l l z l l ~ ~ l ~oft Dzrila
~ r i e ~and Kecava, ed. Maurice Bloomfield.
AS^^^^^^^^^^^^^^, the commentary of Haradat-
tacharya, ed. T. ~ ~sastri (Sri ~ Dass
~ Orielltal
~ Series, ~ ~In: JCUrna1t thei American Oriental XIV, 1890.
Kausitaka-Grlzyas2tra [Kaz~;Gs]
291, Delhi 1985.
Atlzarvavedn-~a~?zhitii [AV] Kau~italcaGrIzya Sitras, ed. T.R. Chintamani, New Delhi 1982.
Kausitaki-Briiknzn!za [Kau;B]
Atlzarvavedasm?zlzit~,ed. Damodar Satvalekar, Pardi n.d.
Knusitaki-Briihnza!za, ed. Sreekrishna Sarma (Verzeichnis der
Bh~gfl~atapur~~ia
$rinzad Blziigavatallz of @!za-Dvnil~clynrza Vydsa, ed. A-C. Orientalischen Handschriften in Deutschland, Suppl. 9,1),
Wiesbaden 1968.
Bhaktivedanta Swami Prabhupada, New York, etc. 1972.
Klziidira-Grlzyasitra [KIzGS]
B1zdradvdja-~rlzyasitra[BIzGS]
B l l ~ r C I d V ~ j ~ g r / 2 y a S-~ gze
t r a 1Doalestic
1z Ritrrnl Accorrhzg to The Grilzya-SClras. Rules of doiltcstic cere~nolzies, pt. 1:
the Sclzool of BhZradvdja, ed. Henriette J.W. Salomolls, New ~Bi~khii~ana-, Aivabyana-, Pargskara-, KhBdira-Grhya-Satras,
transl. Hermann Oldenberg; (Sacred Books of the East 29),
Delhi 1981.
Oxford 1886.
Brlzad~ra!zynka-~~)alzi;ad [BA U ]
B ~ l ~ ~ ~ l ~ r ~SUrejvara1s ~ ~ ~ ~Viirtika ~ ~ O012 I MI d~z ~I l ~ i ~ ~ ~ d
BrdIzina!za, ed. and transl. K.P. JogIShoun Hino, Delhi 1988. ~rivaiklziirznsn-bhngavacchastre Bhrgul~rokta!zKlzilfidlziMralz,
ed. Parthasiirathi Bhatfa, Tirupati 1961,
Brl~at~ar?zhit6 [BrS]
V a r d I ~ a ~ ~ z i h i ~ ~Sorjl/zitG,ed. and transl. M. Ramakrishna Kri~adlzikdra
Bhat, 2 parts, Delhi 1981, 1982. Vaikharzasiya Kriyadlzikara (Bhrgu Sa~!zhitii),ed. The Vaikha-
nasa Agama Pandit (Sri Venkatesvara Oriental Series, 40),
C1z~~zdogya- ~ ~ 3 a r z i ~[Clz
a dU ]
Tirupati 1953'
Tlze Cl~aadogyo~ p a l z i s / 2repr. a ~ of the edition Calcutta 1849-
Kumiimsanzbhava
1850 (Bibliotheca Indica, 3), Osnabriick 1980.
Dlz~ri?iasar?zgralzn [K2lidZsa:] Tlze KurnGrasa~zbhavntvitlz t11e Co~lzriterzta~i~~ of
Tlze Dhnn,2rrsal~zgraha. ancielzt Collecliolz of B~rddlzist
teciz- Mallillatlz (Sarga 1-8) and of Sitnriim (Sarga 8-17), ed, Kasi-
nath Palldurang Parab, 3rd edition, Bombay 1893.
lzical Ternzs, ed. Kenjiu Kasawara/F. Max ~ i i l l e r l H Wenzel .
(Anecdota Oxoniensia, Aryan Series, Vol. 1, Pt. 5), Oxford Liliga!'urdna[LiP1
1885. Liligapurii!zmlz ~ r i r i ~ a ~ z ~ ~ z a l z i i ~ ~ i l c ~ ~ ~ a d v a i ~ ~ d ~ a ~ z a v
Gobhiln-~rlzyasitra[GGS] taut (G~ri1lnandalagsa11tha-mB1L, 60), Calcutta 1960.
G ~ b h i l a g y l ? y ~Wit/% ~ ~ BIza@ndrGyalza
t~~~. e d MaAabhdrata
'S Conzrlze~~t~rY,
Sriniarz Mahiibhiiratanz. Accordiizg to Soutlzerrz Recerzsion Based
Chintamani Bhattacharya, New Delhi 1982.
on tlze South Indiarz Texts. Witlz Footnotes and Readings, 8 vols.,
Hir~!z~akeii-~rlzyns.tm [HGS]
Tile ~ ~ i i z y n - s f ~~l~~
i t ~ ~@hnlestic
~, cerenzolzies, Pt. 2: Gobhi- e d T.R. Qishnacharya and T.R. Vyasacharya, Delhi 199 1.
]a, HiraIxyakeiin, Apastamba, transl. Hermann Oldenberg; Maitr@a!2i-Snr?2hita[MSI
Moitriiya!zism?zlzitG,ed. Damodar Satvalekar, Pardi n.d.
(Sacred Books of the East, 30), Oxford 1892.
Harivamia [Hv] MaitrGya!zi-Uporzi~ad[MaitrU]
Tlze Harivarizia, Beilzg tlze Kllila or Sul)17le"zfl" to the J.A.B. VAN BUITENEN, TIzf Maitriiya!ziya U1)anisad. A Critical
Essay, with Text, Translatiorz and Corn~~ze~ztary (Disputationes
Mo/z~b/2drata,ed. parashllram Lakshman Vaidya, Vol. I: Criti-
Rheno-Trajectinae, G), s1Gravenl1age 1962.
cal Text, Poona 1969.
 AGRAWALA, R.C. 1993 Images, Attributes &Motifs. Studies in Early Indian Art
1960 A Newly Discovered Relief from Rajasthan, Jourrzal of and Nunzimzatics, 2 vols., Delhi.
Indian History, 38 (I-111), 191-192, Trivandrum. BANERJI, Projesh
1962 Interesting KusIna Terracottas and Sculptures from 1985 Nataraja - The Darzcirzg God, New Delhi.
Rajasthan, IrzrEinn Historical Quarterly XXXVIII, 4, BANERJI,
260-266, Calcutta. 1924 The Tenzple of jiva at Blzurnara, Memoirs of the
1965 Agni in Early Indian Art, Journal of I~zdialzHistory Archaeological Survey of India No. 16, Calcutta.
XLIII, Pt. I, 151-157, Trivandrum. 1928 Basreliefs of Badanzi, Memoirs of the Archaeological
1968 GandhIra Skanda with Flames, East & West, N.S. 18, Survey of India No. 25, Calcutta.
1-2,163-165, Rome. 1933 Eastern Indiarz School of Mediaeval Sculpture, ASINIS
1970 More Sculptures from the National Museum, New Delhi, 47, Delhi.
East & West, N.S. 20, 3, 35 1-355, Rome. BARUA,Benimadhab
AGRAWALA, V.S. 1937 Barlzut, Vol. III, Calcutta.
1963 I~zdiaas ~llowlzto Pa!zini [A Study of the Cultural Mate- BASKARAN, S.T.
rinl in tlze ~ ~ l ~ ~ ~ 2nd & ~revised
a ~ iand] , enlarged 1975 Paintings and other Remains of the Armamalai Caverll,
edition, Varanasi. Art arzd Arclzaeology Papers, December 1975, 24-33.
1964 Wlleel Flag qf India Clzakra-Dlzvaja, Varanasi. BAUTZE-PICRON, Claudine
1984 VnrnllasiSeals andSealirzgs,Indian Civilization Series, 1994 The "Visnu-LokeSvara" Images from Bengal, An East-
No. XXVIII. Varanasi. ern Version of ViSvarCpa According to the Piiiicarstra?
A G ~V.S./MOTICHANDRA
~ ~ ~ ~ ~ , Festschrift Klaus Brulz~zzur Vollerzdungdes 65. Lebws-
1960 Yaksha Worship in Varanasi, MatsyapurIna (Ch. 180), jalzres dargebraclzt von Sclziilern, Freulzdeiz ulzd Kolle-
PurGna 1, no. 2, 198-201. gen, ed. Nalini BalbirIJoachim K. Bautze, Reinbek,
ALTEKAR, A.S. 129-170.
1949 State and Governnzelzt in Ancient India, Banaras. BECK, Brenda E.F.
A ~Vinayak ~ ~Mahadev 7 1976 The symbolic merger of body, space and cosmos in
1954 Social and Religious Life irz the Grilzya Siitms, 2nd Hindu Tamil Nadu, Contributions to Irzclialz Sociology,
edition, Bombay. N.S. 10, 2,213-243, New Delhi, etc.
ARBMAN, Ernst BELL,H.C.P.
1922 Rudra. , T J ~ zunz~ nltindischerz
~ ~ ~ ~ U I ~I ~ ~ 19 14
Glauberz / ~
Arclzaeological
~ ~ ~
Survey
~
of~ Ceylorz,
~
Arzlzual
~
Report 1909,
Kultw, Inauguraldissertation, Uppsala. Colombo.
ASHER,Frederick M. BERKSON, Carmel
l980 Tlze Art of Eastern India, 300-800, ~inneapolis. 1992 Ellora. Concept and Style, New Delhi.
A ~Cynthia Packert
~ ~ ~ ~ ~ BHANDARKAR,
~ , D.R.
1995 The HarFat-Mjte Temple at ,&banefi, Levels of Mean- 1912 The Temples of Osii. Anrzual Report of the Archneo-
ing, Artibus Asiae LV, 314,201-236. logical Survey of India 1908-09, 100-1 15.
1997 Sculpture of Early Medieval Rajastlzarz, Studies in BHAnACHARJEE,Arun
Asian Art, Vol. XXI, Leiden. 1978 Icons and Sc~~lptures of Medieval Assanz, Delhi.
AWASTHI, A.B.L. BHAVACHARYA, Gouriswar
1976 studies ilzskanrh ~ ~PnrtIK ~ ~ ~ ~~ l ~ a~ Art !a n i ~c a1987
~ ~and l ~ A Solitary Illustration of Eight Lokapslas from South
Bihar, Soutlz Asian Studies 3, 63-70 [= Essays 011
Icorzogrylzy, Lucknow.
BAJPAI,K.D. Buddhist Hindu Jairz Icoizograplzy & Epigrcrylzy,ed.
1947 A New Inscribed Image of a Yaksa, bzdia AlztiPa. A Enamul Haque, Dhaka 2000,205-214,564-5651,
v~~~~~~~ oriental studies preseltted by lzis Friends 1993 Sonla - the Hilldu Moon God, Orientations 24, 10,
and Pupils to Jean Plzilippe Vogel, Kern Institute, 84-87, Hong Kong [= Essays on Buddlzist Hirzdu Jairz
Leyden. Icorzograplzy & Epigraphy, ed. Enamul Haque, Dhaka
B * ~ ~Krishna
~ l , Dutta 2000: 303-307,591-5931.
1997 Depiction of lndra in ~~~l~ Indian Art, Dinze~zsiolzsof 1996 Trio of Prosperity, A Gupta Terracotta Plaque from
Ozrliarz History arzd Culture. Dr. Subirilal Chalzdra Bangladesh, Soutlz Asian Studies 12, 39-47.
~~~l~~~~ i centenary~ ~ VolLlnze,
l ~ ed. B. Prasad/B.P. 1997 A Unique Bronze Image from Bangladesh, South Asia~z
RoyIJ. Mishra, New Delhi, 290-296. Arclzaeology I995, ed. R. Allchin, B. Allchin, Vol. 2,
B ~ ~S.R. ~ s ~ ~ ~ U.S.A./New~ DelhiICalcutta,
~ ~779-788. ~ ~ ~ ~
BAATTACHARYA, Kamaleswar
l963 Four Clzola Temples, Bombay.
1979 ,rater C I ~ ~~ E ~ ~~~~~~~t~~~~~~ ~ I to ~Rfljelzdra~ III ~1961 l Les religiorzs ~ bralzr~zarziques
~ , darzs l'arzcien Canzbodge,
(A.D. 1070-1280), Madras. d'apr2s l'kpigraplzie et l'iconograplzie, Publications de
B * ~ ~ Jitendra
~ ~ ENath ~ , 1 ' ~ c o l e
franqaise dYExtr&me-Orient, vol. XLIX, Paris.
1956 The Developlnenf of Hbl& Iconograp/zy. second BHATTACHARYYA~ D'C'
Revised and Enlarged Edition, Calcutta. 199 1 Pratinz6laksa!za of the Vigudlzar~izottara.New Delhi.
B * ~ ~Ar~~ndhati
~ ~ l , BHATTASALI, N.K.
198 1 Origin, ~ ~ and ~ ~~ ~ ~ lEvolution~~ ~ i of c ah l 1929 ~ I c o l l o ~ rl a p of l ~ ~Buddhist
~ and Bralzmarzical
~ Sculptures
~
vajra in India11 Art, Roopa-Lekha LII, nos. 1&2, 15-279
iiz the Dacca Museunz, Dacca.
New Delhi. BISWAS,S.S.
1981 Terracotta Art of Bengal. Delhi.
BISWAS,T.K./JHA,B. CHHABRA, B.Ch.
1985 Gupta Sculptures, BIzarat Kala Bhavan. New Delhi. 1935 Agni, Joi~r~zal of tlze h i l a n Society of oriental Art 'I1'
BLOCH,T. 45-47.
1898 Buddha worshipped by Indra, a favorite subject of COHEN,Andrew L.
Ancient Indian Art, Proceeding 4 the Asiatic Society 1989 Ten21,1e Arcl~it~cture and Scubture in ~ o ! ~ ~
of Bengal VII, 186-1 89, Calcutta. PhD. dissertation, Chicago, Illi~lois,2 vols.
BODEWITZ, Henk W. 1995 Unpublished Nolamba Ceiling Panels from He'navat1'
1973
1982
Jain~iniyaBrGlznzn!za [I,]-651. Leiden. Sri NiigcTbhir~mzrlanm(Dr. M S . Nagaraja *"
The waters in Vedic cosmic classifications, Indologica sclzr@), Essays on Art, C ~ l l t ~ ~Hi~torY~
re, ArclzfleoloSY'
Taurirzensia X, 45-54, Torino. Epigraphy and Conservatio~iof C~blt~irfl( Prol'"ty
1984 The Fourth Priest (the Brahmin) in Vedic Ritual, Select- hidin andNeigbbouring Countries, ed. L.K. srinivasad
ed Studies on Ritual ilz tlze Indian Religions, ed. R. S. Nagaraju, Bangalore, 321-324.
Kloppenborg, Leiden, 33-68. 1998 Tenzple Arclzitecture and Sculpture of tlze
BOLON,Carol Radcliffe (see also RADCLIFFE) (Nintlz-Terztlz Centuries), New Delhi.
1986 The Durga Temple, Aihole, and the SangameSvara COLLINS, Charles Dillard
Ternple K~davelli,A Sc~ilpturalReview, Ars Orientalis 1988 Tlze Icollogmq,hy (znd Ritllal of Siva at E l e New ~ ~ ~ ~
XV, 47-64. York.
1988 In: EITA Vol. 11, Pt. 1, ATortlz Illdia, Foundatiorzs of COOMARASWAM~, Analldn Kentish
North Indian Styb c. 250 B. C . 4 . D . 1100, ed. M.W. 1928 Yaksns, Part 1, Washington.
MeisterMA. DhakytK. Deva, New Delhi. 1928a Early Indian Iconography, I. lndm, with qecial Refe-
BONER,Alice
rence to "Indra's Visit", Eastern Art 1, no. 32-41?
1962 Prirzciples of Coinposition in Hindu S c ~ ~ l pCave t~~~, Philadelphia.
Tenzple Period, Leiden. COUSENS, Henry
VAN DEN BOSCH,Lourens P.
1926 The ClzGlukynn Architecture of the Kannrese Districts'
1982 Yama, the god on the black buffalo, Visible Religion ASINIS Vol. 42, Calcutta.
(Ann~l*lforReligiollsIcofzogrPlzy) I, 21-64, Leiden. CZUMA, StanislaWJ./MORRIS,~ ~ k h ~
BRERETON, Joel Peter 1985 K ~ s k a nScul~tz~re, 11nagesfroin ~ a r l India.
y cleveland'
1981 The Agvedic Adityas, American Oriental Series, vol. DAGEN~,
63, New Haven. 1984 Entre Alanzpirr et Sris'ailam. ~eclzerchesarclzgoloSc-
BRUHN, Klaus
qiles ell Andlira Pradesl~.Publications de l'lnst'tut
1960 Distinction in Indian Iconography, BDCRI 20 (,yllslzil
Frangais dlIndologie no. 67, 1 et 2, ~ o n d i c h i r ~ .
Kurnar De Felicitation Volunze), 164-248. DAMAIS, Louis-Charles
2000 The Gmn~marof Jina Iconography 11, Berliner Illdo-
logische Studien 13114,273-337.
1969 A propor du couleurs symbo]iques des points
dinaox, Bulletiri de ~'~colefrarangaise
car-
d '~x~r~ns-orlelit
BUCHANAN, Susan Locher
5675-1 18, Paris.
1985 Caluk~aT e ~ l e s ,History and Iconogrqlzy, p h . ~ . R.N.
dissertation. The Ohio State University. Ann Arbor, 1979 Select ~ ~I vec1ici~ ~ ~ ~ h~ ir n c~t s~. l~e l h ~i . ~~ i ~~
Michigan (University Microfilms International). DAS,H.C.
BURGESS, James
1974 A Set of Astadikpala Images in the Orissa State
1877 Rock Cut Temples at Badami in the bldiarz
urn, Orissa Historical Researclz Jourrzal XX1'I> 1-4,
Antiquary 6,354-366, Bombay.
67-7 8.
1883 Report on Ehrfl Cave Temples and the BralIninni- D E ~D~~~~~~~~ ~ ~ ,
and Caves in Western Inclin, completing the 1996 ~h~ ~ ~imagery of~lzajuraho,
~ i Project
~ for Indiani ~
Results of tlze Fifth, Sixtlz, and Seventh Seasons1 olJer-
Cultural Studies, Publication IV, Mumbai.
ations of the Archaeolosical Survey 1877-78.1878-79, DESA~, ~~l~~~~ S.
1879-80 S1lpplementaV to L'olunze "The Cave 1973 I ~ o f ~ s n LN~~
l ~, ~ ~ l h i~ . ~ ~
Te1721'1es ' f India': ASINIS Vol. V (= Archaeological
DEYA,Kri~hndAGR*WALA, V.S,
Survey of Western India, Vol. V). London.
BUSSAGLI, M./SIVARAMAMURTI, C. 1950 The Stone Temple at N l a n d l , journal of the His-
n.d. torical Society XXIII, 1-2, 198-212, Lucknow.
5000 Years of the Art of lrzdia. New York.
Catalogues DEVA,Krishna
1980 Stone Temple (No. 2) at Nglandl, ~ocrrnaloftlzeI"d1nn
1959 5000 j ~ h r e Kunst aus inrlierz, Essen, villa ~ u ~ ~ l
(14.5.-30.9.19.59). Society of Orierztnl Art N.S. XI, 80-84, Calcutta.
1962 1984 h n q e s of Nepal, Archaeological Survey of India9New
Art of India Stone Sculpture, Asia Society I ~ ~ . , Delhi.
New York. Introduction by Aschwin Lippe.
1982 l985 Mu~deivari Temple, Ramgarh; ~ e l i - k f i - f l ~ " ~ ~ ~ '
Ill the Inzage of Mart, London.
1988 Aninzals in Indian Art, Tokio. Gwalior, Indian Epigmplzy. Its ~ e a r i n gon the
CHAKRABARTI, Dilip K./HAsAN,S.J. of Art, ed. F.M. Asher1G.S. Gai, 125-127; 161-163'
1984 The Antiquities of Kangra. New Delhi. New Delhi, Bombay, Calcutta.
CHANDRA, Pramod 1988 In: EITA Vol. I[, Pt. I, Nortlz Inclin, ~ o u J z ofd ~ ~ ~
1970 Nonh hzdian Style c. 250 B.C.-A.D. 1100, ed. M'w'
S t ~ n eScull7t~lrein the Allalzabad MlueunI,poona.
Meister1M.A. DhakyIK. Deva, New Delhi.
 1991 In: EITA Vol. 11, Pt. 2, Nortlz India, Period of Early FISHER,Robert E.
Maturity, c. A.D. 700-900, ed. Michael W. Meisterl 1995 Noble Guardians. The Emergence of Lokapalas in Bud-
M.A. Dhaky, New Delhi. dhist Art, Oriental Art, N.S. XLI, 2, 17-24.
l995 Temples of India. 2 Vols. New Delhi. FONTEIN, Jan (with essays by R. Soekmono, Edi Sedyawati)
DevA, I<rishna/~nIv~D~, S.D. 1990 The Sculpture of bzdonesia. National Gallery of Art,
lgg6 Stone Sccllptclre irz tlze Allalzabad Museum, Vol. 11, New Washington.
Delhi. FOUCHER, Alfred
DHAKY,M.A. 1905 L 'art grico-bouddlzique du GaizrJlzara, Vol. I. Paris.
GABORIEAU, M.
The Old Temple at Lamba and KiimeSvara Telnple at
Auwa, Jourrzal of the Asiatic Society, Calcutta VIII, 3, 1992 Des dieux dans toutes les directions. Conception indi-
141-148. enne de l'espace et classification des dieux, Pllrusdrtlzn
15,2342.
Some Early Jaina Temples in Western India, Shri
Mahavira Jaina Viclyalaya Golderz Jubilee Volume, GAI,G.S.
290-347, Bombay. 1985 The MundeSvari Temple Inscription of the Year 30 and
lg88 The Dravidian Sculptures in Pre-Imperial RBGrakiifa Its Bearing on the Temple, Indian Epigraplzy. Its Bear-
ing orz the History ofArt, ed. F.M. As11erlG.S. Gai, 129-
Cave-Temples in Ellora, Ellora Caves, Scull~tureand
131, New Delhi, Bombay, Calcutta.
Arclzitect~ire,ed. R. Parimoo/D. KannalIS. Panikkar,
GAIL,Adalbert J.
438-445, New Delhi.
lga8a 111: EITA Vol. 11, Pt. 1, Nortlz Iizclia, Folirzdatiorzs of 1980 Cosmical Symbolism in the Spire of the CeyIon
Dagoba, The Stlipa, Its Religious, Historical and Archi-
Nortlz Iizdiaiz Style c. 250 B.C.-A.D. 1100, ed. M.W.
tectural Sisizificarzce, ed. A.L. Dallapiccola, Beitrage
Meister1M.A. DhakyIK. Deva, New Delhi.
lgg6 In: EITA Vol. I, Pt. 3, South India, ~p12erDrdvidadcia) zur Siidasienforschung, Siidasien-Institut Universitst
Heidelberg, Bd. 55, Wiesbaden, 260-266.
Later Phase, A.D. 973-1326,2 Vols., New Delhi.
DIKSHIT, K.N. 1984 Ternpel in Nepal (Band I), Ilcolzographie Izi~zclclistisclzer
1938 E ~ ~ v a t i o l zats Pdkdrpur, Bengal. Memoirs of the Pagode~ziiz Pnfa~zKatlzii~aizdutal,Graz.
Archaeological Survey of India 55, Delhi. GARDE,M.B.
DILIPKUMAR 1924-25 Excavations at Pawaya, Annual Report of tlze Arclzaeo-
1986 logical Survey of India, 165-166, Calcutta 1927.
Archaeology ?f Vaislzali,New Delhi.
D 1 ~ AOdile ~ ~ ~ ~ ~ , GASTON, Anne-Marie
lg70 1982 $iva in Dance, Myth and Ico~zogra~~lzy, Delhi.
Le temple de Jambuliflga (datC de 699 ap. J.-C.) g
GAUDANI, H.R./DHAKY,M.A.
BSdBmi, Arts Asiatiqcles 21, 15-39, Paris.
1971 Les temples d781ampuret de ses environs au telnps de 1969 Sculptures from Kaleshwari-ni-nal, Jo~tr~zal of the Ori-
ental Institute Baroda XVIII, 4, 360-364.
Ciilukya de BBdjmi, Arts Asiatiques XXIV, 51-lo19
GOICHALE, Shobhana
Paris.
1976 Elephanta Hoard of Copper Coins of KysnarSja, T l ~ e
D O N A L ~Thomas ~ ~ ~ ,E.
l976 Jourizal of tlze Nt~~izisrnaticSociety of IrzrEin XXXVIII,
Doorframes on the Earliest Orissan Temples, Artibus
pt. 11, 89-91, Varanasi.
Asiae XXVIII, '213, 189-21 8, Ascona.
1985-87 Hindu Teniple Art of Orissa, Vol. I (19851, Vol, I1 GONDA.Jan
1966 Lolca. World and Heaven in the Veda. VNAWL, N.R.,
(1986), Vol. 111 (1987), Leiden. LXXIII, 1.
1991 The ~ a v a - ~ ~ h aasn a Puru,sa in OrisSan Images,
1967 The Mearzirzg of tlze Sanskrit Ternz DhEmaiz. VNAWL,
CBmuqds to Kgli/Tiirii, Artibcls Asiae LI, 112, 107-141.
N.R., LXXIII, 2. Amsterdam.
D o w s o ~John
,
1974 Tlze Dctal Deities in the Religion of the Veda. Amster-
1879 Classical Dictiolzary of Hind~tMytlzology London.
dam.
DunA, Monoranjan Triads irz the Verln. Amsterdam.
1976
1990 Sculpture of Assanz, Delhi.
1977 Tlze Ritual Slitras. A History of Indian Literature, Vol. I,
ECK,Diana L.
1982 Fasc. 2. Wiesbaden.
City of Light, Princeton University Press.
GORAKSHKAR, Sadashiv V.
EMENEAU, M.B.
1960 1981 A Harihara Image recovered from Jogeswari and the
NBgapiiSa, Nzgabandha, ~arpabandha, and Related
problem of Dating Gharapuri (Elephanta), MADNU.
Words, BDCRI 20, Suslzil Kunzar De Felicitatioiz Voz-
Recent Researclzes iiz Iizdiarz Arclzaeology arid Art
unze, ed. N.G. Kalelkar, Poona, 291-300. History, Slzri M.N. Deslzl~arzdeFestschrift, ed. M.S.
FA^^, Harry Nagaraja Rao, Delhi, 247-252.
1987 V i s y im Veda, Hilzduisnzus urzd Buddhisnzus.
GOSWAMI, A. (ed.)
schrift fur Ulrich Schrzeider, ed. H. Falk, Freiburg, 12-
1956 Indian Teniple Sculpture, Calcutta.
133.
l997 GOSWAMI, Rajeshwar Pandurang
The Purpose of Rgvedic Ritual, Inside the Texts,
1979 Astadikpdlas iiz Literature and Art, unpublished P11.D.
Beyond tlze Texts. New Approaclzes to tlze S t ~ l dof~ the
dissertation, Deccan College, Pune.
V e h s . Proceediizgs of tlze ~nterllational Vedic
GOVINDASWAMI, S.K.
shop, Harvard University, Jurze 1989, e d Michael
193.5 Note on a Stone Image of Agni, Jo~irrzalof tlze Iizdian
Witzel, Cambridge, 69-88.
Society of Oriental Art III,48-50.
GRAVELY, F.H./SIVARAMAMURTI, C. HUNTINGTON, Susan L.
1939 Illustrations of Indian Sculpture mostly Southern. 1981 Iconographic Reflections on the Arjuna
Madras. Kalcidariaiza. A~nerica~z Studies in the Art of India, ed'
GRONBOLD, Giinter Joanna G. Williams. New Delhi, Bombay,
1984 Die Mythologie des indischen Buddhismus (Worterbuch 57-67.
der Mythologie, Bd.V, ed. H.W. Haussig), Stuttgart. 1985 The Art of Aizcieizt India, New ~ o r k / T o k ~ o .
GUPTA,Nila JAMUAR, B.K.
1979 ParamLrakalin kalZbtiyom mem I(r~nakath2 kB 1985 The Ancielzt Temnples of Bihar, New Delhi.
SilpBmkan, Art of the Pararniiras of Miilwii, Proceed- JAYASREE,K.
ings of the UGC Sponsored All-India Seminar held at 1988-89 Chronology of the Temple at ~ a l l ~ v a l lAn i , 1cOnomet-
Prachya Niketan, Centre of Advanced Studies in Indol- ric Study, Jourtzal qf Ancient Indian History XV1ll'
ogy & Musicology, Bhopal, January 21-23, 1978, ed. 1-2,49-61.
R.K. Sharma, Delhi, 116-117. JOHNSEN, Gosta
GUPTA,S.K. 196566 Varupa and Dhytarastra, Indo-Iranian Journal IX?245-
1983 Elephant in Indian Art and Mytlzology, New Delhi. 265, The Hague.
GUPTE,R.S. JOSHI,J.R.
1967 The Art and Architecture of Aihole, Bombay. 1973 V&-Vgyu, BDCRl XXXIII, 14,227-248,
GYANI,R.G. JOSHI,Nilakantha Purushottama
1950-51 Identification of the So-called Silrya and Indra Figures 1972 ~ ~of the ~ i ~ ~~Sculpt~lres
h i in ~ the ~State
~ ~ ~ ~
in Cave No. 20 of the Bhajs Group, Bulletin of tlze Museunz, Luclcnow (Part I). Lucknow.
Prince of Wales M U S ~ U of ~Western
I India 1, 15-21, 1991 Pray$ga-pa!!a, A Slab in the British Museum depictFP
Bombay.
Prayaga and the Fish Incarnation of Vi?I;lu, A ~ n y a l z l v l '
HANDA,Devendra Essays presented to Dr. Debala Mitm in adlniration
1983 Two Interesting Yama Images from Osian, East & West her scholarly contributions, ed. G. ~ h a t t a c h a r ~D~'~'''
aj
N.S. 33,53-56. 203-2 11.
1984 Osiak Histoy, Archaeology, Art Architecture. Delhi. 1 9 9 1 ~ Varanasi ,$ilipa!!as and Similar slabs in ~ e d i a e v ~ '
HARLE,James C.
Sculptures of Central India, A Preliminary
1963 Temple Gateways in South India. The Architecture and Puriitan 8, 17-23.
Iconongraphy of the Cidambaram Gopuras, Oxford. 1992 Deva-Pattas, A Less-Known Chapter of Medieval
1969 Le temple de Ngganritha h Nagriil, Arts Asiatiques XIX, Hindu Iconography, Easter11Approaches, ed. TsS.
53-83.
well, BombayICalcuttdMadras, 133-140.
1986 The Art and Architecture of the Indian Subcontinent, ~~h~
Harmondsworth. 1982 Art of Osian Tenzples. Socio-economic and
HARTEL,Herbert
Life in India, 8th-12th Centuries A. D. New Delh'.
1971 Die Kunst Indiens. Indien und Siidostosien, Propyljen
KHAN, d. Abdul Waheed
Kunstgeschichte Band 21, ed. H HlrtellJ. Auboyer, 1973 Stone sculptures in t/le iilaInpirMuseu>,l,The Goverl'-
Berlin, 11-96. 399
ment of Andhra Pradesh ~rchaeologicalseries,
HAUSWEDELL, Ernst
Hyderabad.
1961 Allkfion 110, 27.11.61 (Samml~ngK. Stavenhagen), K H A ~Karl ~ ~ ~ ~ ~ ~ ~ ~ ,
Hamburg. 1990
HAZRA,R.C. The Island Shrine Elephalzta. Its Date and DynaSt'C
Afiliatioiz, Government of Andhra ~ r a d e s hArchaeO-
1940 Studies in the Puriinic Records on Hindu Rites and cus-
toms, Dacca. logical Series, No. 65, Hyderabad.
KIRFEL,Willibald
1981 The Historical Backgrounds of the Maruts' (or Rudras7) 1920 Die Koslnogral,his der lnder 12achdel., ~ ~ ~ l l&ge- etz
Association with Indra and Rudra, and of the purgvic
stellt. Bonn, Leipzig (Repr. Hildesheim, Zurich, New
Story of their Origin, Puriina XXIII, 2, 101-148,
Varanasi. York 1990).
1927 Das Puriiga PaficakaIqana. versttclz eirzer Textgem
HEESTERMAN, J.C.
sclzichte, Leiden.
1957 The Ancient Indian Consecration. The R a s i i y o 1954 Purcina vom weltgebirude (Bhnvanavilzy~sa). Die
described according to the Yajus Texts and Annoted.
Disputationes Rheno-Traiectinae 11. 's-Gravenhage. ko~mographischen Tralctate der Purci!za's. versucia
HENSGEN, Hans einer Textgeschiclzte, Bonn.
KLAUS,Konrad
Bemerkungen kosmologischen Abschnitt des 1986 Die a~tilzdische ~ o s l l ~ o ~ oNncil
g i e , den ~ ~ ~ b n a !dar- zas
Brahm Wda-, Vgyu-, und Matsya-Purgpa, Zeitschrift der
DeutschenMorgenlandiscbelz Gesellschafr 108, Heft 1, gestellt. Indica et Tibetica 9, Bonn.
Neue Folge, Bd. 33,161-163.
K ~ ~ ~ ~
HESTON,Mary Beth 1981 Vaisravana in North-West India, MADHU. R"e1zt
Iconographic Themes of the Gopura of the Kailasan~tha Researches in Indian Archaeology w d A r t H i s f o v ~''lri
Temple at Ellor5 Artibus Asiae XLIII, 219-235. M.N. Deslzpande Festschrift, ed. M S . Nagaraja RaO'
HOPKINS, E. Washburn Delhi, 253-262.
K o ~ o wSten
,
1915 Epic Grundrifi der Indo-Arischen Philolo- 1930 ~~t~ on ~ ~ jAcia Orieniaiia
~ ~ 1 1 1p, a ~s IVf ~
gie und Altertumskunde, Band 111, Heft 1 B, Strassburg,
31 1-317, Leiden.
 1989 Silpa versus ~Zstra,Shastric Traditions in Indian Arts,
ed. A.L. Dallapiccola, 2 Vols., Stuttgart, 5-15.
1990 The ViBvaMpa Sculpture from Suhania at Gwalior,
Makararzda. Essays in honour of Dr. James C. Harle,
Sri Garib Dass Oriental Series, No. 105, ed. C. Bautze-
. Picron, Delhi, 131-144.
1994 A Gurjara- PratihHra Image of Visnu ViSvarOpa, Art -
The Integral Vision. A Volume of Essay in Felicitation
of Kapila Vatsyayan, ed. B.N. Saraswati, S.C. Malik,
Madhu Khanna, New Delhi, 97-1 10.
MEINHARD, Heinrich
1928 Beitrage zur Kenntnis des jivaismus nach den alten
PurGnas, Baessler-Archiv, Bd. XII, Berlin.
MEISTER,Michael W.
1973-74 A Preliminary Report on the Siva Temple at Kusumz,
Archives of Asian Art 27,76-91.
1975-76 A Field Report on Temples at KusumZ, Archives of
Asian Art 2 9 , 2 3 4 6 , New York.
1979 Juncture and Conjunction, Punning and Temple Archi-
. tecture, Artibus Asiae XLI, 213,226-234.
1982 BithO, Individuality and Idiom, Ars Orientalis 13, 169-
4 or
1982a Analysis of Temple Plans, Indor, Artibus Asiae XLIII,
VON M ~ ~ E R W A L L N
302-320.
1986 Regional Variations in MZtrkH Conventions, Artibus
Asiae XLVII, 314,233-262.
1988 In: EITA, Vol. 11, Pt. I, North India, Foundations of
MONIER-WILLIAMS,
North hdian Style, c. 250 B.C.-A.D. 1100, ed. M.W.
Meister1M.A. DhakylK. Deva, Delhi.
MOODY,Timothy Floyd
1991 In: EITA, Vol. 11, Pt. 2, North India, Period of Early
Maturity, c. A.D. 700-900, ed. Michael W . Meisterl
M.A. Dhaky, New Delhi.
1993 Style and Idiom in the Art of Uparamiila, Miiqarrzas 10
MUKHOPADHYAY,
(Essays in Honour of Oleg Grabar), 344-354.
MERH,Kusum P.
1996
Yama-The Glorious Lord of the Other World (Recon-
NAGAR,M.M.
structing History & Culture, no. 12), New Delhi.
MEVISSEN,
Gerd J.R.
1994
An Interesting Frieze at Diirssuram, Festsclzrift Klaus
Bruhn, zur Vollendung des 65. Lebensjalzres darge-
bracht v012Schulern, Freunderz und Kollegen, ed. Nalini
BalbirIJoachim K. Bautze. Reinbek, 413-436.
1998
Eine jinistische Skulptur mit DikpZlas, Indo-Asiatische
Zeitschrijt 2,20-24, Berlin.
MICHELL,George
1973
Dating an Important Early CZ!ukyan Monument, The
PiipanZtha Temple, Patadkal, Oriental Art N.S. X I X , 2,
192-20 1.
1973a The Regents of the Directions of Space, A Set of Sculp-
tural Panels from Alampur, aarp (Art arad Archaeology
Research Papers) 4,80-86.
1989
The Penguin Guide to the Monunlents of India, New
York.'
MISRA,Ram Nath
1981 Yaksha Cult and Iconography. New Delhi.
MITRA,Debala
1960 An Early Representation of Varuna at Bhubaneswar,
The Orissa Historical Research Jollrnal IX, 1&2, 1-7,
Bhubaneswar.
1963 Varsha-Cave of Udaygiri - An Iconognphic Study,
Journal of the Asiatic Society V, 3&4,99-103, Calcutta.
1972 Excavatjolzs at Tilaura-kot alzd Kodarz arzd Explor-
ations in the Nepalese Tarsi, unpublished manuscr1ptf
Calcutta.
1973-74 The Relics of Sagrahawa - A Study of the ~ o u n d a t i ~ ~ -
deposits and the Ispk~-ny,asa,Joiimal of the Aldhra
Pradesh Historical Research Society XXXIII, 2-49
1987 Nilaka~fieivara Temple at Narayanapuram (District
Vizianagaram, Andhra Pradesh), I~zvestigatirzgIndian
Art. Proceedings of a Symposium on the ~ e v e ~ o ~ l ~ ~ " ~
of Early Buddhist and Hindu ~conographyheid at
Museunz of Indian Art Berlin in May 1986, e d "
YaldizIW. Lobo, Berlin, 191-21 1.
1988 In: EITA Vol. I[, Pt. I, North bldia, F o u ~ i d a t i oof~ ~ ~
North hdian Style c. 250 B.C.-A.D. 1100, ed. M.w'
Meister1M.A. DhakylK. D e w , New Delhi, 251-275.
l991 1:. EITA Vol. 11, pt. 2, North India, period of
Maturity, c. A.D. 700-900, ed. Michael W M~~~~~~~
M.A. Dhaky, New Delhi.
n.d. Observatiolls on Some Carved Slabs and ~ e ~ o s i t - b o ~ ~ ~
in Museum Pusat, Jakarta, Indian Studies EssayS i'a
Hor~our of Prof. Nilwrralzjan Ray, ed. A. R~)""
Sanyal1S.C. Ray, Delhi, 43-51.
Gritli
ER,
l989 Yaksas of Ancient Mathurg, Mathurfi. The c~~~~~~~~
Heritage, ed. Doris Meth Srinivasan, New Delhi, 368-
382.
Monier
1899 A Sanskrit-English Dictioizary, New Edition.
1980
The Agnyiidheya, Establishnzent of the sacred
Ph.D. dissertation, McMaster University,
Ontario.
Mihir Mohan
1967-68 KalyFi~asundaraTheme in North Indian Art, J~~~~~~
the Indian Society of lrzdiaiz Art, N.S. 2,4-15.
1942 Mathura Museum Notes, Journal of the ( i P ~i~~~~~~~~
Research Society XV, l , 115-121, Lucknow.
NAGASWAMY, R.
1969 / ~ ~ ( A Guide). ~ a r n i l n a dS'"~e
The K a i l a s a ~ a a ~Temp/p
Department of Archaeology, Publication No. Madras'
47
NIGAM,L.S. (ed.)
2000
Riddle of Indian Iconography (Zetetic 01%Rare Icon
frorn TfilB),Delhi.
CONNO NOR, Stanley J.
1966
Ritual Deposit Boxes in Southeast Asian ~ a n c ~ ~ " ~ ~ '
Artibus Asiae XXVIII, 1, 53-60, Ascona.
car~llel
O ' F ~ Wendy ~ ~D o~n i g~ e r~l M ~~ c ~~, e~o~r g~e, / B ~ R ~ s O N f
1983 Elephants, The Cave of Shiva, Princeton, New Jersey'
0 ~ ~Vishwa
1 , Chander
l991 Sculpture of the Westerit Himalayas (History and
Stylistic Developmerzt), Delhi.
PAL,Pratapaditya
1978 The Ideal Image. Gupta Sculptural Tradition ard Its
Infl~ierzce,New York.
1979 A Kushiin lndra and Some Related Sculptures o~~~~~~~~
Art, N.S. XXV, 2,212-226, London.
1988 A Catalogue of the Los Angeles County Museullz of Art
Collection, Vol. 11. Berkeley, Los Angeles, London'
19g7 A Collecting Odyssey. Indiarz, Hir~zalaya~Z, find
east Asian Art from the Janaes olzd Marilynlz A~~~~~
Collectio~z,Chicago.
 PANDE, Alakh Niranjan RAO,T.A. Gopinatha
1958-59 Role of the Vedic Gods in the Gyhya-Sotras, Jour~zalof 1916 Elenzerzts ofHindu Ico~zogral~lzy, Vol. II, Pt. II, Madras.
the Ganganath Jha Research Institute XVI, 1-2, 91- RAVEN,Ellen M.
133, Allahabad. 1988 The Secret Servants of Kubera, The Naras or Guhyakas,
PANIGRAHI, K.C. Studies irz Soutlz and Soutlzeasf Asian Archaeology.
1956 The Archaeological Remains at Benissgar in the Essays offered to Dr. R. Soek~noizo,ed. H.I.R. Hinzler,
Singhbhum District of Bil~ar,Jourrzal of tlze Bihar Re- Leiden, 105-145.
searclz Society 42, 1, 1-1 1. RAY,Amita
1961 Arclzaeological Renzairzs at Bhubarzeswar, Bombay etc. 1983 Life and Art of Early A~zdlzradesa,Delhi.
PARIMOO, Ratan REA, Alexander
1983 Sculptures of 3esaGyi. Vis!zu, Baroda. 1896 Clziilukyarz Architecture; Includi~zgExanzples frorn tlze
P A ~ uSasanka
~, Sekhar BallEri District, Madras Presidency, ASINIS XXI,
1976 Kuruksetra in the Vcinzarza Puriina. Calcutta. Madras.
V.S./SULLEREY, S.K. 1909 Pallava Architecture. ASINIS XXXIV, Madras.
l991 Tripuri-Tirthapa!~a, a Sculptural Mythic Represelltation REITZ, Falk
of Tripuri, Art of the Kalaclzuris, ed. Kalyan Kumar 1993 Die Bildnischen am Harsatmiitii-Tempel in Abaneri -
ChakravartyJRaj Kumar Sharma, Bhopal, 49-54. eine Typologie, Berliner I~zdologisclzeStudie~z7 , 207-
PATTABIRAMIN, P.Z. 225.
1975 Sa~zctuairesRupestres. II, Tanzilrzd& et Kerala, Publi- 1994 The Harsatmiitii temple at Abaneri, originally a Devi-
cations de 1'Institut Fran~aisdYIndologieno. 42, 2, paiiciiyatarza complex, Soutl~Asian Arclzaeology 1993,
PondichCry. ed. A. Parpola, Petteri Koskikallio. Vol. 11, Helsinki,
PAUL,Debjani 645-656.
l985 Antiquity of the Visnupiida at Gaya. Tradition and ROCHER, Ludo
Archaeology, East & West 35, 1-3, 103-141. 1986 Tlze PurGgas. A History of Indian Literature II,3. Wies-
P 1 ~Pierre et al.
~ ~ ~ ~ , (Franpise L'Hernault, Fran~oise Boudignon, baden.
L. Thyagarajan) ROSENFIELD, J.M.
1994 yngt ans apr& Talzjavur, Gn~zgaiko~zda~hola~uranz, 1967 The Dynastic Art of the Kusharzs. Berkeley, Los Angeles.
Ecole fran~aised'ExtrEme-Orient, Memoires archCol0- RUPING,Klaus
giques 20,2 Vols., PondichCry. 1970 Anz~ta~izantlzarza u~zdKiir~iza-AvatGra.Eirz Beitrag zur
P ~B. Rajendra
~ ~ ~ ~ , purarzisclze~z Mytlzen- und Religio~zsgeschiclzte.Schrif-
l983 Clzalulya~ztenzples of andhradesa, New Delhi. tenreihe des Siidasien-Instituts der UniversitHt Heidel-
1988 In: EITA, Vol. 11, Pt. l , Nortlz India, Fouizdations of berg, Wiesbaden.
Nortlz Indian Style c. 250 B.C.-A.D. 1100, ed. M.W. RUSSEK, Rend
Meister1M.A. Dhakyl K. Deva, New Delhi. 1986 Hirzduis~nus.Bilderlcanorz und Deutu~zg,Miinchen.
R A D ~ ~(see ~ lalso
i ~Radcliffe BOLON),Carol E.
SAHAI, Bhagwant
1981 Early CIzalukya Sculpture, 2 Vols., Ph.D. dissertation, 1967 Agni in North Indian Art, Jour~zalof tlze Bilzar Researclz
Society LIII, I-IV, 143-154, Patna.
New York University.
RAMACHANDRAN, T.N. 1970 An Interesting Stone Panel from the Visnupada Temple,
Dr. Unzeslz Mislzra Conznzenzoratio~zVolunze, ed. B.R.
lg36-37 Excavations at Mahasthan, An~zual Report of the
Saksena, Ganganath Jha Research Institute, Allahabad,
Arclzaeological Survey of l~zdin,5 1-54, Calcutta 1940.
RAMACHANDRA RAO,P.R. 709-715.
l977 Alanzpur. A Stlldy in Early Clzalukyarz Art, Hyderabad. 1975 Icorzograplzy of Minor Hindu and Buddhist Deities,
RAMAN, K.V. New Delhi.
1985 Some Aspects of Gopura, Tirumalai and Tiru~ati 1994 Dikpiilas in the Art of Khajuriiho, KlzajurEho irz Per-
Temples in the Light of Epigraphical Data, Indialz spective, ed. Kalyan Kumar Chakravarty1M.N. Tiwarit
Karnal Giri, Bhopal, 137-142.
Epigrplzy. Its Bearing on the History of Art, ed. F.M.
Asher1G.S. Gai, 225-230, New Delhi, Bombay, Cal- SANFORD, David T.
1994 Ramayana Portraits. The Nageshvara Temple at Kum-
cutta.
RAMARAO,M. bakonam, MGrg XLV, no. 3,43-60.
lg60 Cii!ukyan Telnples of Satyavolu, Trarzsactiolzs of the SARMA, I.K.
1992 Tenzples of tlze Gaiigas of Kar@ilaka. Architectural Sur-
Arclzaeological Society of So~lth 1959-609
vey of Temples, Number 6, Archaeological Survey of
Madras, 72-78.
1961 The Temples of Alumpur, Jour~zalof I~zdialzhis to^ India, New Delhi.
1993 Alampur Temples, Rare Evidences on Constructional
XXXIX, Pt. 111,369-391.
Modes and Consecrational Rites, Sri Riirizaclzandrikd
1964 Eastern CElukyan Tenzldes of A~zrNzra ,Andhra
(Professor Oruga~zti Rdrnaclzarzdraiya Festscl~rif).
Pradesh Archaeological Series, 19, Hyderabad.
RAO,L.S. Essays 012 I~zdia~zArchaeology, History, E12igraplzyy,
2000 Numis~izatics, Art and Religion, ed. A.V.N. Murthy1I.K.
A Mysterious Sculpture from TB18 - A Rare Form
Sarma, Vol. 11, Delhi, 351-368.
Varuna, in: L.S. NIGAM2000, 121-126.
SARMA, M. Radhakrishna
RAOv Manjushree
l994 1972 Tenzples of TelirigEtza. TIze Arclzitecture, Iconograplzy
Sarzclzi Sculptures (An Aestlzetic and Cllltural Study).
and Sculpture of tlze Ciilukya and KEkattya Tenzl~les,
New Delhi.
Hyderabad.
SCHASTOK, Sara L. Felicitation Volunze, ed. Anand Krishna, varanasi798-
1985 The himaliji Sculptures and sixth century Art in West- 104.
era India, Leiden. 1983 Vedic ~ u d r a - ~ i v Joarrzal
a, of the American
SCHMIDT, Hanns-Peter Society 103, 543-556, New Haven.
1968 Brhaspati und Indra. Untersi~chungenzur vedischen 1990 Newly Discovered Inscribed Mathud Sculptures.
Mythologie urzd K~~lturgesclzichte, Wiesbaden. Probable Doorkeepers, Dating to the K?atrapa
VON SCHROEDER, Ulrich, with the collaboration of the Archaeologi- (Epigraphic Analysis by Lore Sander), Archives of
cal Department Asian Art XLIII, 63-69, New York.
1990 Buddhist Sculptures of Sri Lanka. Hong Kong. SRINIVASAN,
K.R.
SEN,Chitrabhanu 1983 In: EITA, Vol. I, Pt. I , So~~tiz
India, ~ o w e ~r r z v i ~ l ~ ~ " ~ '
1978 A Dictionary of the Vedc Rituals. Delhi. 200 B.C.-A.D. 1324, ed. M.W. ~ e i s t e r 1 M . ADhaky'
SHARMA, R.C. New Delhi.
1976 Mathura Museurn and Art. 2nd revised and enlarged SRINIVASA~,
p.~.
edition, Mathura. 1991 Mundesvari Temple Inscription, bzdian ~ r c l z a ~ ~ ' ~ g ~
SHASTRI, Ajay Mitra Heritage. Shri K. V. Sozrndnra Rajnn festsclzr@9 ed'
1964 Heracles in front of the Infantry of Porus, Jocrrnal of Margabandhu et al., Delhi, 313-31 6.
Indian History 42, Pts. 1-3, 119-126, Trivandrum. SRIVASTAVA,
K.M.
SHENDE, N.J. 1986 Discovery of Kapilavastu. New Delhi.
1948 The Foundations of the Atharvanic Religion, BDCRI SRNASTAV~, slIashi ~~l~
IX, 1-2,197-414, Poona. 1993-95 Dikpala Images in North lndian ~ e m p l e s .our^^^ of lhe
SHERRING, M.A. Indian Society of Oriental Art, N e s sXXIl &i XX1ll'
1868 Berzares, The Sacred City of the Hindus, art accourzt of 95-106.
Benares in ancient and modern times, London (Reprint STAAL,Frits, in collaboration with C.V. somayajipad and M' Itti
Delhi 1975). Ravi Nambudiri
SEIUKLA, L.K. 1983 Apzi The Vedic Ritual of the Fire Altar*
1972 A St~ldyof Hindu Art and Architecture with especial Berkeley.
Reference to Terminology, Chowkhamba Sanskrit STADTNER, Donald M.
Series, Vol. 82, Varanasi. 1981 The Siddhesvara Temple at PalPri and the of Kosala
SINGH,Chanderpal
during the Seventh and Eighth centuries$
1996 The Images of A@idikp5las from Haryana, Culture Orientalis 12,49-56.
tlzrouglz the Ages (Prof.B.N. Puri Felicitatiorz Volume), S U L L E S.K.
~~~,
ed. Sarva Daman Singh, Delhi, 399-405. 1993 The K5lafijan Vaiho!hapa!ia and its historical signif-
SIRCAR, Dinesh Chandra
icance, Journal of the Asiatic society ~ o m b a yN.S'
, 64-
197 1 Studies in the Religious Life of Ancielzt and Medieval 66,1989-1991,238-241.
India. Delhi, Varanasi, Patna. SUTHERLAND, Gail Hinich
1981 Two Early Inscriptions. 1. An Early Inscribed Image of 1992 yakFn ir2Hinduis171andBu&/2isnz, Tlze ~ @ u i ofs the ~ ~
Indra, Inzdian Museum Billletin XVI, 7-9, Calcutta.
Denzorz, New Delhi.
SIVARAMAMURTI, C.
TADDEI,Maurizio
1974 NafarEja in Art, Tlzought and Literature, New Delhi, 1962 An Ekamukhali~igafrom the N. W. F. p. and
1976 ~ a t a r ~ ~ d rVibhllti
i ~ a , of$iva's Iconography, New Delhi. Connected Problems. A Study in lcOnographY and
1983 Palzoranza of Jain Art, New Delhi.
Style, East & West, N.S. 13,4,288-310.
SMITH,Walter
1970 Indierz (Archaeologia Mundi, 4), Genf.
1994 Tlze Mukteivnra Temple in Blzubaneswar, Delhi. TANABE, Katsumi
SOMPURA, P.O. (ed.) 1990 The Kushao Representation of A N E M O S I O ~ ~and O
1960 D&Zr!zava. Palitana.
its Relevance to the Central Asiaii and Far Eastern
SOUNDARA RAJAN,K.V.
1981 Wind Gods, Silk Road Art and Arcltaeolog~ 1-80'
Cave Temples of the Deccarz, Architectural Survey of Kainakura.
Temp1es N o 37 Archaeological S~lrveyof India, New TARR,Gary (see also G,M. T ~ ~ ~ ~
Delhi.
1969 The Architecture of the Early west err^ ~ l m l u k ~ ~ ~ f
1986 In: EITA, Vol. I, Pt. 2, Soutlz India, U12perDrzvj&dzia, dissertation, University of California, Los Angeles.
550-1075, ed. M.W. Meister1M.A. 1970 Chronology and Development of the chblukya
Dhaky, New Delhi.
Temples, Ars Orierztalis 8, 155-1 84, Ann Arbor.
SPARREBOOM, M./HEESTERMAN, J.C.
TARTAKOV, Gary Michael
1989 of setting the sacr$icinlfires accordiizg 1985 Interpreting the inscriptions of Mah&fi!a Art is^^^^-
to the VCdktlln school. Wien.
SPINK,Walter M. cillly, Indian Epigraplzy. Its Bearing on the History of
l983 A n , ed. F.M. Asher1G.S. Gai, New ~ e l h i l ~ o m ~ ~ y ' ~ ~ ' -
The Great Cave at Elephants, A Study of Sources, cutta, 139-152.
0'2 G~LptaCulture, ed. Bardwell L. Smith. 235-
THIAGARA~AN,
~ ~ b ~ ~ ~ h
282, DelhiIVaranasil~atna. 1989
SRINIVASAN, Doris M. Tile Developnzent of Scrya and ~ a v a g r a h aworship
Tarnil Nadu, Ph.D. thesis, University of Madras.
19" Iconological Studies of Recent Mathura Acquisitions in T I ~paul
~ ~ ~ ~ ,
the Bharat KalaBhavan, Chlzavi- 2. Rai Krishsadasa
1951-52 Der Weg durch den ~ i nach ~ ~ ~ l
General Index Baroli 85
Barwiisiigar 3,67,9 1,94 (12. 479), 104f., fig. 179
Batesarii 3, 65, 7 1, 84 fig. 161
Abiineri 3,70f., 93, 102 (n. 513), figs. 224,225 battle scene 15,25, 82
Achchankovil 91 (n. 469) Benisiigar (Benusggar) $73,715 (11. 391), 96 (11. lo5
Qclhipati,4f., 12, 113 (n. 522), 108f.
Aditi 4-6, 8-10, 12 Berlin 85,89, figs. 311,315,318,327
Aditya 5 (n. 20), Gf., 9f., 11 (12. 62), 12 Besnagar 30 (12. 177)
Afghanistan 19 BetDr 60
Ahar 105 (n. 523) Bhiijii 22
Ahicchatrii 22f. Bhalikari 22, 85
Aiho!e 3,25,35,37,43 (n. 224),44 (n. 226), 57,59f., 88,91, Bharatpur 21
94,96, 106, figs. 49-54,130-135,139 Bharhut 1, 18,22, 113
AirRana/Airavata 17,25,27,29,41,44, 86,95 Bharna Kalan 22f., 95 (12. 484)
Aiyank-Siistii 41 (n. 217),78 (n. 403) Bhavaniisi Sangam 3,57 (n. 268), 58,60,88, fig. 138
Aja Ekapiid 10 Bhaviini 7
Alampur 3,23,28,34f., 37,39-46,52f., 55 (11. 262), 57,59f., Bhaviinipur 7 1,98
70,76,79,8 1,88,90,92-107, 111, 114, figs. 23,24,64-87, Bhind 63
141-144 Bhubaneswar 3,28,30f., 33,71,73,74 (11. 381), 75-77, 88-919
Aligadh 85 96 (n. 487),98, 102, 107-109, 11l , figs. 42-44
Allahabad 67 (12.333),108, fig. 25 Bhi~marii 3,23,28f., 30-32,33 (rt. 191), 34 (11. 1971, g49
Alur 58f., 61,91 104f., 107-109,111, 114, figs. 25-28
Amariivati 21 (n. 137) Bhundiinii 71
Amrol 3,45-48,52-55,63-65,92f.,96-99, 103f., 106f., 111, Bikkavolu 3,57,63,81,90, 102, figs. 283,284
114, figs. 90-97 Birarmacandrapur-SBsana 101 (n. 507)
atnrta 29 BithD 3,48, 52f., 55 (11. 262), 56, 101, 105f.
amrtarnalztlzana (see Churning of the Ocean) Bodhgaya 89, fig. 313
Ananta 1,7, 10, 14,20,29 (n. I74), 71,81 (n. 425). 113 Bodhisattva 21
AnantaSayanaIAnantaSgyi 21,38,59,82,85,87f., 90, Bogra 72
fig. 3,327 Bolangir 74
Andhakiisuravadhamfirti 26 (n. 160), 45
Bombay (Mumbai) 91 (n. 469)
Aniruddha 70 (11. 348) Brahmii 1 , 4 (n. E), 7, 13f., 16, 20f., 23-27,30, 35-399 45955'
Alikor ViitlCambodia 19,78 (11. 405), 103 (n. 516) 57,63,71,84f., 87f., 93, 112-1 15, figs. 47,487 13O
antarik;a/anthrik~a 5f., 7 , s (n. 43), 9f., 10, l l (n. 57, n. 62), brakmalz 7-12
12 (n. 72) Brahmiini 49 (n. 249)
Anuriidhapura 18,20 (n. 123) Brahmin 98
Arajaguppe 3,57-60, fig. 136
Brhaspati 4 (12. 8, n. 9), 5-7, 9f., 12, 16 (n. 92)
ArdhanariSvara 26,42,51f., 54 (12. 256),74, 83, fig. 19 Brooklyn 71 (n. 355)
Armamalai 57,61,77
Buddha 22 (n. 144), 44,85
Assam 72 (n. 362), 73
Buddhist artBuddhism If., 13 (n. 77), 18f., 21f., 313 723
6Stipipdliyanz 4
78,113
as.tagralza 56,92
ngalntitrkii 30 (n. 177), 87 (11. 457) Cakrapumsa 29 (n. 173)
asura 25 (12. 152), 29,82, (n. 430), 100 (n. 501) Calcutta (Kolkata) 21f., figs. 28,305
ASvins 79
Calukyas 23,25 (12. 151), 28,35,37,52 (n. 254)
Auwii 71,104,107 (n. 530)
Cambodia 19,78 (1%. 405), 103 (n. 516)
Avani 57 (11. 269)
Ciimunda 51, 100 (12.501)
Candi NgrimbiIIndonesia 19f.
Badiimi 3, 14,23-B,%-38,57,59,60 (12. 279), 8 If., 87f., 96, Candi Singhasiiri 81 (12. 425)
101, 104f., 111, 114f., figs. 10-18,46
Candrabhbgii 71 (1%. 353)
Badoh 3,29 (n. 173), 66f., 90f., 108, fig. 180
Candraketugadh 22f.
Baijniith 108f., fig. 325
caturrnakiirtijikas l8,21
Bajagiimve 59 (n. 276)
Cauriisi 3,72 (n. 366), 73 (n. 372), 76f., 93
Balariima 31
Changu NSriiyana/Nepal 86
Bali (demon king) 24
Charda 74
bali-d61za 11
CharsadddPakistan 22
bali-pi!ha (see also pitha) 91
Chatriirhi 3,46, 111
Bangladesh 56,72,83
Chaulukya 19
Biirambii 3,72f., 74 (n. 378), 94, figs. 241,242 Chechar 89
Baran 85,88
Chidambaranar 78 (n. 402)
Bargaon 84,89
Churning of the Ocean (amrtanzantharza) 25 (ft. 152), 28f''
Barmer 70
77 (n. 3973, 82, fig. 9
Cidambaram 57 (11. 266), 78 (12. 406), 81 (n. 424), 95 Gaurisvara 15f.
Citragupta 40f., 43f., 98 Gay% 72 (11. 365), 76 (K 392), 89f., 95
Cittaudgadh 3, 34,48-53,54 (n. 2-56), 55 (n. 262,n. 263), GhiinerBv 71, 112, 115
56f., 63,65,69-71,78,81,92f., 96-107, 111, 114, Goddess, jackal-faced 91; lunar symbolism? 63 (12. 301),
figs. 104-122 fig. 151
Cleveland 97 (n. 489) Gokiik 59
CBla 57,80 gopura 57, Glf., 78,80f., 90,95, figs. 278,279
Cuttack 73,76 Gosna Khera fig. 2
Goth-Miinglod 71 (n. 352, 12. 353)
Dacca 83 Greek 101f.
Daksina Kosala 45, 88f., 90, 102 grivh 80, 94
Daksin~mfirti 44, 62 (n. 296) Guna 45f., 63t
DandaIDandin 30,47,53 Guneri 89, fig. 313
Plng 3,46,63-65,70,71 (K 356), 81,92,97-103, 149 Gupta period 22,28, 29,30 (n 1 7 3 , 32,41, 85
figs. 145-151 Gurjara-Pratihiira (see Pratithiiras)
Dantidurga 63 (12. 297) Gwiilior 3,29,4648,53,55,63, 64 (12. 307),65-67,
Dbiisuram 57 (12. 266), 81 (12. 424, 12. 426, 12. 427) 71 (12. 3 5 3 , 84,86,92f., 105 (n. 524), 106, 108, figs. 9,
Deity, male, with club 77, fig. 286 98-103,160,171-176,180
Deogadh 3, 18,20f., 85f., 102, fig. 3 Gyaraspur 66
Deo PHtanlNepal 108
devakzilik?i 77, 80, figs. 294-300 Hajo 73
Devasenl 51 Harihara 42,51f.,63
Dev Barunarak (Deo Barugiirk) 3,56,71, 919 lo2 HarikeSa Yaksa 109
Dhamnlr 85 Harsa era 32
Dharanendra Yaksa 58 Hemiivati 57,59 (11. 273, GOf., 87 (n. 458), 90, fig. 140
Dharlnapuri 3,57-60,77, figs. 266-274 Hingliijgadh 82, 108, fig. 324
Dhobini 45 Hiranyakasipu 13
DhfimornB 61 (n. 286) Hiranyaroma Parjanya 14
dhvajastanzblza 9 1 horror vncui ("fear of empty space") 53,60,70
diggajas 86 Hoysaia 59 (12. 276)
double axe l 9 Hyderabad 91 (11. 469)
DurgB (see Mahislsuramardini)
dvdrrrynla/d~n~ii~Z 4 (11. 3), 18 (n. 10% 32 (1%.1 8 3 , 93 709
Indor 3,45f., 55,63-65,92f., 95-104, 107, figs. 152-159
Indore 82 (12. 431), fig. 324
EkapBda Siva 65 Indriini 7,64 (n. 307)
Elephants 3, 23,26f., 82f., 111, 114, figs. 19,20 fivara 16 (12. 97)
Ellor3 3,23,26-28,42,45,57,61-63,78 (12.405), 82f.9 Iswal 105 (12. 523)
87 (n. 458), 88,90f., 95, 108, 111, figs. 21,22, 278-Zg2
Etah 83 (n. 436), 84, 87 (n. 458), fig. 306 Jabalpur 66
Jagat 71 (12. 358), 105 (n. 523)
fire 8, 12 (n. 70), 19, 46, 49, 74 (12. 374), 75f., 79f., 849 Jaina 2,26,57,61,66,78,91,96, 112, 115
88,97 Jalatunda/Indonesia 20 (12. 125)
JamBlpur/MathurB 22
Java 1, 19,20 ( K 130), 81
Gaja-Laksmi 23,31,53
Jhalawar 101-103, figs. 317,319
Gajsntaka 60 (n. 283)
Gall%valli 3,73-77,93, 96 (12. 486), 97, 99-101, 103f.3
Jhansi 66f., 108
figs. 243-248
Jina 58,78, 112
Jodhpur 53f., 56
gaga 30,46,60
GandhBral~akistan 19, 102, fig. 318
Kailiisa (mountain) 13,26,62, 86
gandharva 9, 13,27,31,37 (n. 204)
KBla Bhairava 108 (n533)
GaneSa 30,37 (n. 204), 46,53-56,60,61 (n. 2839 639
Kalacuri 26 (12. 158)
69,73,75,77 (n. 3 9 0 , 83f., 86,89f., 92995 (12. 480)*
fig. 240
KBlakfita 29
Ganebwarpur 3,72 (n. 366),73 (12. 373),76f.9 933 99, KBlaiijara 108
KBleSvari-ni-nB1 71
107 (n. 530), 112,115, figs. 249-265
KBlidLsa 15,21 (12. 138)
Galkg 32 (n. 187),75
kalpav&a 75
Gaf%idharamfirti 26,43, 83
Ka!ugumalai 77 ( K 401), 78
Gangaik~ndacd!apuram 3,77,80f., 94
Gangas 58-60 ~al~B~asundaramfirti/Siva Vai~iihikamtirtilPBrvati-~ari~~~~
(wedding of Siva and Pkvati) 26 (12. 160), 45,57,62
Garuda 26f., 29
(n294), 63,82-84,86,87 (a458), 90, fig. 301-306
Gaur 72f., 90, fig. 240
KBmBkhyI 72 (n. 362)
Gauri-Sankara 59 ( K 272)
Kiimiin 83f., figs. 302,303 Madan Kiimdev 72 (12. 362)
Kambadahalli 58,59 (n. 277), 60 (n. 279), 9 1 Madras (Chennai) 59 (n. 277), figs. 140,288,291
Kanauj 3,63,83f., 86,89, fig. 301 mahiidiina 14
Kiiiicipuram 3,77f., 79 (n. 409), 105, 114, figs. 285-287 MahBkiita 3,35f., 38,87, fig. 48
Kangra 66, 108 Mahiinadi 45
Kankiili p1FdMathurii 22,30 (n. 177) Mahiisthiin 72
Kantaka CetiydSri Lanka 19 (12. 121) MahBvira 82
kapili 45f., 53,55,64.71,92f., 95 MaheSvara 35 (n. 199)
Karachi 83 MahisamardiniIMahisiisuramardini 30, 55, 62,75, 829
Kardama PrajBpati 14 fig. 279
Kiiritalai 66 Maholi 30 ( K 177)
KiirttikeyalSkandaIKumarii 16,21 (n. 133), 24f., 26 (n. 159), Miimallapuram (Mahiibalipuram) 110,211 (n3001,
30f., 37 (11. 204), 41 (n. 217),46,51,53,54 (n. 260), 61-63, 212 (n. 302)
75,78, 81,83f., 86, 88,90-92,95 (n. 485), 101 (n. 507), 112, Msnasigangii 83f., 89, fig. 304
114, figs. 8,16,117 MBnasottara 15
Katrii (Kedaradeva)/Mathurii 22, fig. 8 nzandala 20
KedahlMalaysia 19 Miindalgadh 71
Keldhar 65,95 (n. 480), 100 Mandasor 85
Kelk 57 Mandor 48 (11. 246)
Ketu 9, 16 (n. 92), 75 (12. 386) Manu Lokapiilas 2, 14, 16,23,33f., 39,50,55 (1%.262),
KetumBna Riijasa 14 56,71 (n. 361), 81 (12. 428), 82-86,88-90,95,
Khajuraho 64 (n. 311), 65 (rz.312), 67, 102 102 (n. 513), 105f., Illf., 113,115, figs. 301,3049
Kharod 45 311-313
Khed 3,70,73,93,101,105, figs. 226-233 Marhiii 30
Khiching 76 (n. 391) Markhera 100 (n. 504), 101
Khidarpurii 69,7 1,93 Masrur 3,66
KilaiyGr 78 (n. 407) Mathurii 3, 18 (n. log), 21f., 25, 34,66, 83f., 89f., 102,
Kiradu 105 (n. 523) 107 (11. 530), figs. 2,s-8,177,178,304,305
kirttimukha 63 Mau 108
Kodumbiijtir 78 (n. 403) MElappaluvGr 79-8 1
Kolar 57 (n. 269) Meru 14, 92
Konkan Mauryas 26 (a158) military expedition 16, 112
Kota (Rajasthan) 67,85,91 (n. 468) Mirzapur 108
Kota (M.P.) 64, fig. 160 Mukhaliligam 72-75,90,96 (n. 486)
Kysna 65,76 MundeSvari (Hill) 3,28,30-34,45,52, 56,7 1,73 (!l. 37O)7
Kysnii (river) 43 84%89f., 92,95-97,99, 104, 106-108, 111, 114f.v
Krsnalilii 93 fig. 34-41
K~emankari 48 (12. 244), 51
K~etrapiila 11,20 Nacnii 65,74, fig. 170
Kudaveli 3,41,43f., 98, figs,85-87 nHga 78
Kukkanar 57 Niigariija 18 (12. 1 0 9 , 80
Kumbak6nam 78,80 ( E . 419) Niigara! (Niigrii!) 3,35f., 38, 87, fig. 47
karma-s'ilii 19 Niigiirjunakonda 22
Kurnool 61 Niilandii 3 , 3 1,34,71,84,92, l l l , figs. 29-33
Kuruksetra 18 (n. 104) NiinLghiit 16,21
Kusiinaperiod 16, 18,21-23,41,82,85, 89,95, 102, NLadakunda 97
104 (n. 518), 111 Narasamangalam 58
KusBna-Gupta transition 30 (n. 177) Narasimha 55
Kusumii 48,49 (n. 247),50
Nariiyanapuram 3,73-75,89,91
Naresar 4 6 6 3 (n. 300), 65, 104, fig. 322
Lachchiigir 97
NateSa 26-28,37 (n. 203), 42,454 54,57-62, 82f., 87f.9
~ a k ~ m i l ~16r i(n. 9I), 29,69,95 (n. 480)
94, 108, 115, figs. 20-22,136,137,266,326
LakuliSa 45,s 1
navagraha 16,75 (n. 386), 91, 112
Lalitpur 66
navasanga 20 (n. 130)
LiimbB 3,56,89f., 94 (n. 479), 102 (n. 513)
Nepal 86,108
Leiden 21, fig. 5
New Delhi 79, 108, figs. 289,290,302,303,309, 316
Lirigodbhava 89 (n, 464)
New York 22,34, fig. 177
Lokiiloka 14
Noh 21
London fig. 240
Nohta 67
Los Angeles 67 (n. 330), 84 (n. 443)
Nojamba 57-60
Lucknow 22,108
Oado 102 Sabarkantha 71 (12.360)
Osiiiii 3,35,41 (n. 218), 45f., 48,52-57,63f., 67-71,81,86, Saci 13
92f., 96-107, 113f., figs. 182-223 SagrahawaINepal 19, fig. l
~iimaliiji 49 (12. 249), 85
Piihiirpur 3,41,70,72, 92,9597, 103, 106, figs. Wt-239 Siifici 18,22
paintings 61 Sadkarbandha 83, 84 (12. 441)
Palar 61 Sadkar ghlt 48 (n. 242)
Paliiri 3,45f., 89-91, 102, fig. 89 ~arabha 100 (11. 506)
Pali 52,71 Sarapalli 75 (n. 386)
Pali Khera 104 Siirniith 19 (it. 121), 23,97, fig. 314
Pallava 41 (12. 217),77 ~arviini 7
Palm (tree) 41,43, 98 Satyavolu 3,44f., 81,93,96f., 99, fig. 88
Palma 82 Savitr 6, 9
Pan"c6yatana 93, 112 Sesa 29
Pandya 77 (n. 401), 80 Sesai 67,71 (rz. 356), 103, fig. 181
Panna 65 (n. 315) Shahabad 31,56
Shivpuri 64f., 67
Panyam 39 (n. 209)
Piipanbanam 58 Siddhepalli 57 (it. 269)
ParaSuriima 79 (n. 412) Sikhara 23,28, 32, 37,39,44, 61,80f., 92, 94, 108, 111, 114
PgrSvaniitha 54 $ilaI~I~ndikhra~n 78 (iz. 404)
Piirvati 23, 26 (n. 160), 27, 35, 37, 45-47, 52, 56, G1 (lz.287)7 Singhbhfim 73
Siron Khurd 66, 108
6 8 4 , 86, 88, 92, 94
PBtan/Nepal 108 Sitaliighiiti 22, fig. 6
Siva 1, 13, 14, 16 (rz. 98), 17,20 (rz. 130), 21,23-28,30f.,
Pathasivarama 59 (n. 277), 60, fig. 137
34 (11. 195), 35, 37, 39,42,44-46,49 (rz. 249), 50, 52, 56f.,
PatnL 31f, 89f., 108, figs. 34-41,312
Pattadakal 3, 35 (n. 198), 37-40,57, 59f., 62,70178, 88, g33 60-63,64 ( L 307), 65,67,78,82-88,90f., 95, 106, 108f.,
111,112,115
94 (n. 479), 95, 106, 115, figs. 55-63,129,326
Pawiiyii (Padrniivati) 3,23,28f., 82, 101, 111, 1149 fig. iivaliriga 39, 108
Skanda (see Kiirttikeya)
Pengore 105
Solniiskandamarti 57,59, fig. 129
Pingala 30,47,53
Srikakulain 73
Pip3d 3,48,53f., 55 (n. 262), 63,92f., 95f., 105, 114,figs. 123-128
piflza 82 Sri Lanka 18 (rz. 102), 19
Srimiira rivall lab ha 80 (11. 419)
Polonnaruva 20
Stuttgart 21, 86, figs. 4,307
l-'radak~i!zii/pradaksindPatlzal l (iz. 6.9, 19 (11. 115),
443 527
Sudhiirnii Vairiija 14
70, 87f., 90f.
Suhiiniii 86, 87, 99 (rz. 499)
Prajgpati 6, 7 (12. 36), 9f., 12
Sarya 5(12.12),6,7(rz.31,n.34),9f.,Il(iz.57),12,14,16,
PrCkCra 20,43 (n. 225), 77-80,92,94
19, 21 (rz. 133), 22 (11. 141), 23-25, 27, 30, 32f., 39 (12. 2II),
Prakasam 44
40-44,46f., 49f., 53-56,63,74 (U.374), 76,77 (12. 400),
Pratakota 3,57-59, 60 (12. 282), 61,94, figs. 275-277
78-86,89f., 91 (11. 469), 93,98, 105, 111, 113-1 15, figs. 37,
Pratihiiras 35,41,45, 66, 67 (rz. 3281, 111
79,95,123,145
PryCgapatfi 82 (11.429)
svastiltn 59
Prthivi G, 1 0 , i I (11. 62), 14,16
P~thu 14 (n. 83)
Puri 76 Tafijiivur 3,77-8 l , % , 101, 105, figs. 294-300
Tiirsi 22
Purusasnkta 4
Tewar 82 (rz. 429), 108
9, 16,74 (n. 374) Tikarngadh 65
Raipur 45 Tintoi 49 (n. 249)
Tiruchchirappalli 79f.
Riijendra I. 80 (11. 413)
Rgjghiitl~aranasi 22 Tirukkadaiyilr 3,79,94,96
Rajshahi 72 Tir~tnzaittirar?~57 (it. 270)
Tirunelveli 3,78-80, 103, figs. 291-293
rZksasalraksa 4 (n. 4), 9, 13f., 99
Tirupati 100
Riistrakiita 58f., 61, 63 (n. 297)
Tirupparaitturai 77 (12. 397)
Riivana 62
R~~aniinu~rahainfirti 26,63 Tiruvanaikka 8 1 (12. 424)
Richmond 22 Tripuri 82 (12.429), 108
Trivikrama 29,55, 66,77 ( K 397)
Rohoi 30 (12. 177)
Rohtas 31 Tuinain 86f., figs. 309,310
Roman 101 Tumkur 60
Tudgabhadrii 43
Y S ~ 6, 10, 36, fig. 1 3
Tusa 105 (12.523)
Rudra 5,8, 10-12,20, 106
R~driii;i 7
Udaipur 7 1 (n. 358) Virabhadra 80 (n. 423)
Udayagiri1M.P. 18,20-22 Viraka 13
UmbMahebvara/Um5sahitamCrti 37,57,59f., fig. 138 Virllpiiksa 14f., 16, 18,20 (1%. 128), 86,99
Umri 3,65-67,96f., 99-101, 103-105, 107, figs. 162-169 Visnu 1 , s (n. 12), 6f., 13, 16 (n. 91), 19-21,24-27,
29,30 (12. 178), 35, 37 (1%. 203), 45, 50, 52f., 55,57,
Vaivahikamllrti (see KalySnasundaramlirti) 62 (n. 295), 66, 70 (n. 348), 83, 85, 87f., 108f., 112,
Variiha 20,38,53,93 114f., fig. 10
Varahamihira 15, 111 vi;nupa[fn 90
Variinasi 3,22,65,85, 108f., figs.27,170,306,321,323 ViSvarkpa 22,82,85-87, figs. 307-310
Vardhamiinasllri 96 (n. 488) ViSveSvaralViSvanLtha 108 (12. 534), 109
Viiruni 13,29, 104 (a. 520) Vizianagaram 74
viistu-devn-gnnn or vastu-pada-devata 20,77,79 (n. 410)
v ~ s t u - ( p n r u ~ a ) - ~ n n ~11
z ~(12.
n ~57,
n 1%. 65), 12,20,25 (n. 155), waters 1 , s (12. 3 9 , 13 (n.78), 14 (n. 83), 26,29, 101
77, 113, 114
VBsudeva 70 (12. 348) YajfiamCrti 81 (11. 424)
VBsuki lOf., 29 Yajfia Varaha 102
Vatsaraja 68 (n. 337) YflIcsa (see also Dharanendra, Harikesa) 9, 13, 18,21-239
Vidisha 66 30 (12. 177), 31, 44 (n. 227), 66 (12. 326), 95 (12. 48% 1099
viclyCidlznt-n 3 1 lllf., 113
Vijayaditya 35 (n. 199), 36, 37 (12. 203), 44 (12. 228) yiili 78
vinzdnn 57, 61, 8 1 Yamunii 75
Viniidhara 78 YogW?fa 34 (11. 197),47,55,80,98, 106, 109
Vinaylditya 35,39,43 (n. 225), 44 (12. 228) YogiSvara 61 (12. 293), 62,82,87 (l?. 4581, figs. 280-282
Figures
("For photo credit see List of Figures)
Fig. 2 Kubera, Gosna Khera, Govt. MuseuIn,
Mathura, c. 1st century A.D."

Fig. 3 Vsy~lon AnantaSaya~~aV i y y panel,

DeogaCh Gllpta Temple, c. 500-550 A.D."
Fig. 4 Loknpiila panel (?), "North India", Stuttgart, Fig. 5 LokapBla panel (?),inscribed, Matl~uriiregion, Leiden, c. lst-2nd
C. 2nd century*

Fig. G Indra,
SitaliighLti,Mathura,
c. 4th centi~ry'~
 7 Agni, Mathurii Dist., Mathura,
c. 3rd century*
 Fig. 11 Indra medallion, Biidtimi, Cave 3, vestibule, ceiling,
Fig. 10 V i s ~ umedallion, Badami, Cave 3, vestibule, ceiling,
A.D. 578* A.D.578*

Fig. 13 Brahmj. medallion, Bndami, Cave 3, vestibule, ceiling,

Fig. 12 V a r u ~ amedallion, BadSmi, Cave 3, vestibule, ceiling,
A.D.57g4:
Fig. 16 Skanda, Bjdjmi, Cave 3, main hall, ceiling, A.D. 578*

Fip. 17 Varuna, Badjmi, Cave 3, main hall, ceiling, A.D. 578'"

Fig. 19 Ardhanlri Svara Siva,
Cave 1, c, mid-6th cent1~ r y *
Fig. 20 NaieSa, Elephants, Cave l , c, mid-6th centuryY'

Fig. 21 Natesa, EllorB, Cave 29, c. 600-650"

Fig. 23 Yams Alampur, Kumlra Brahml Temple, Sikhara, south,
c. 630 A.D. Fig. 24 Kubera (?), Alampur, Kumlra Brahml
Temple, Sikhara, north, c. 630 A.D.
Fig. 25 Yarna, Bhfimara. $iva Temple, Allahabad Museum, Fig. 26 Kubera, BhfimarB, ~ i v Temple,
a c. 520-530 *.D.*

Fig. 27 Kubera (extant fragment), Bhfimarii, ~ i v Temple,

a
BKB, Varanasi

Fig. 28 Dandapgni, BhtimarB, ~ i v Temple,

a Indian Museurn,
Calcutta, c. 520-530 A.D.*
Fig. 32 Indra and his elephant?, Nlandii, Stone Temple 2, panels
22-23,7th century
Fig. 34 Agni, MundeSvari Hill, ruined temple, Patna Museum, Fig. 35 Dandapiini, MundeSvari Hill, ruined temple, Patna
6th-7th centuries* Museum, 6th-7th centuries*

Fig. 37 Sfirya, MundeSvari Hill, ruined Temple, Patna Museum,

6th-7th centuries*

"g. 36 Soma, MugdeSvari Hill, ruined Temple, Patlla Museum,

'th-7th centuries*
 Fig. 38 Agni, MundeSvari Hill, ruined
Temple, Patna Museum, Gth-7th centuries*

Fig. 39 Indra, Mundeivari Hill, ruined

Temple, Patna Museum, 6th-7th centuries'"

Fig. 40 Kubera, MundeSvari Hill, ruined Temple, Patna M ~ ~ ~ ~ ~ ,

6th-7th centuries* Fig. 41 Yam& MundeSvari Hill, ruined Temple, Patna Museum,
6th-7th centuries*
Fig. 43 Agni or Soma? (left), Viiyu, Bhubaneswar,
ParaiuriimeSvara, c. early 7th century

Fig. 44 Doorkeeper (left), Kubera?, Bhubaneswar,

ParaSuriimeSvara, c. early 7th century
Fig. 45 Soma, "Uttar pradesh"9
terracotta. C. 6th-7th centuries'"

Fig. 46 Four loknl,iilas, ~ i i d ~ m i y

Jambuliiige~varaTemple, ceiling,
699 A . D . : ~
Fie. 48 Four lokoiJilas, Mafil<B!a, Mallikaduna Temple, ceiling, ]ale 7th centory*
 Fig. 52 Viiyu, Aihoje, Huccappayya Temple, ceiling beam, c. 722-730

Fig. Kubera, Aihole, Huccappayya Temple, ceiling beam, c. 722-730 A.D."

Fig, 54 i i ~ n aAihoje,
, Huccappayya Temple, ceiling beam, c. 722-730 AD.*
 Fig. 56 Agni, Pa!fadakal, Piipaniitha Temple, ceiling,
C. 735-750 A.D.*

Fig. 57 Yama, Paitadakal, Pspaniitha Temple, ceiling,

c. 735-750 A.D.:~:
 Fig. 59 Viiyu, Pattadakal, PBpalliitha Temple, ceiling,
Fig. 58 Nairyta, Paftadakal, Piipaniitha Temple, ceiling,
C. 735-750 A.D."
c. 735-750 AD.*

Fig. 60 Kubera, Pattadakal, Piipaniitha Temple, ceiling,

C. 735-750 A.D.*
Fig. 61 IStina, Pai!adakal, Piipaniitha Temple, ceiling,
c. 735-750 A . D . : ~

Fig. 62 Varuna, Paiiadakal, Ptipaniitha Temple, exterior, F i g 63 Kubera, Pa!!adPal, Ptipaniitha Temple, exterior,
north, late 7th or early 8th century*
north, late 7th or early 8th centurv*
 Fig. 65 Agni, Alampur, BBla BrahmB Temple, south wall,
Fig. 64 Indra, A]ampur, BBla BrahmH Temple, east wall,
C. 650-670 A.D.
C. 650-670 A.D.

Fig. 66 Yama, Alampur, BBla BrahlnB Temple, south wall,

c. 650-670 A.D.
 68 V ~ Y UAlampur,
, Biila Brahmj Temple, west wall,
c. 650-670 A.D.

Fig. G9 Kuben, Alampur, Biila Brahmii Temple, north wall,

C . 650-670 A.D.
 ----
Fig. 71 Agni, Alampur, Svarga Brah~nriTemple, south wall,
Fig. 70 Indra, Alalnpur, Svarga Brahma Temple, east wall,
C. 680-700 A.D.
c. 680-700 A.D.

Fig. 72 Yama, AIampur, Svarga BrahmC Temple, south wall,

C. 680-700 A.D.
 -.-
Fig. 73 Nairlta, Alampur, Svarga Brahmii Temple, south wall,
c. 680-700 A.D.

Fig. 74 Varuna, Alampur, Svarga BrahmZ Temple, west wall,

C. 680-700 A.D.
 Fig. 77 CandraISoma, Alampur, Svarga Brahinii Temple,
north wall, c. 680-700 A.D.

Fig. 78 &na, Alampur, Svarga Brahmii Temple, north wall,

C.680-700 A.D.

Fig. 76 Kubera, Alalnpur, Svarga Brahmii Temple, north wall,

c. 680-700 A.D.
rlg. Ir burya, Alampur, Svarga Brahmii Temple, east wall, 80 Yams, Alampur, ViSva Brahmii Temple, south wall,
C. 680-700 A.D.
late 7th century

Fig. 81 Kubera, Alampur, Viiva Brahml Temple, north

wall, late 7th century Fig. 82 CandralSoma, AIampur, ViSva Brahmii Temple,
north wall, late 7th century
Fig. 83 Nairrta, Alampur, loose image, probably fro111 the Fig. 84 CandralSoma?, Alampur, Garuda Brahmii Temple,
Garuda Brahmii Temple, c. 700 A.D. north wall, c. 700 A.D.*

Fig. 85 Yams, formerly Kudaveli, Saiigameivara Temple, Fig. 86 Kubera, formerly Kudaveli (?), Alampur Museum,
late 7th century4'
so~lthwall, late 7th century
Fig. 89 Overdoor of sanctum, PalSri, SiddheSvara Temple, r late 7th or early 8th centllrv:k
Fig. Agni, Amrol, RgmeSvara Mahadeva Temple, soutll Ical~ili, Fig. 9 1 Yama, Amrol, RBmeSvara Mahgdeva Temple, south(-
C. early 8th centuryY: west), c. early 8th century*

Fig. 94 &na or Soma (?), Amrol,

Fig. 92 Kubera, Amrol, Fig. 93 Viiyu, Amrol,
Djllebiibii Temple, (north-)west, Diinebabii Temple, north(-east),
Danebiibii Temple, north(-west),
C. mid-8th century*
c, mid-8th century"'
c, mid-8th century*
 96 Indra, Amrol, Dinebfibii Temple, (south-)east,
C. mid-8th century*

Fig. 97 Agni, Amrol, Dinebjbfi Temple, south(-east),

c, mid-8th century'"
 _ ...I .-
p-
--
.---
--
L.. - _- ..- . .
..
.. ..
. _ __ _ _ __
_. .

Fig. 99 Varu~p,GwIlior, Teli-k5-mandir, Frieze Panel 69,

Fig. 98 Agni, Gwjlior, Teli-ks-mandir, Frieze Panel 19,
C, mid-8th century
c. mid-8th century

Fig. 100 VIyu, GwBlior, Teli-k5-mandir, Frieze Panel 70,

C . mid-8th century
 i u l Kubera (standing), Gwalior, Teli-ka-mandir, Frieze
Panel 75, c. mid-8th century

Fig. lo2 Kubera (seated), GwPior, Teli-kI-mandir, Frieze Panel

80, c. mid-8th century Fig. 103 CandrdSoma?, ~ ~ aTeli-k$-mandir.
i ~ ~ Weze , panel
113, c. mid-8th century:"
 Fig. 105 Yama, Cittaudgadh, KBlikL?mL?tL?,
sanctum, south(-west),
Fig. l04 Agni, Cittaudgadh, KBlikzmatii, sanctum, south(-east),
C. late 7th century"
C. late 7th century:':

Fig. 106 Naiqta, Cittaudgadh, KiililtImBtS, sanctum, Fig, 107 Varuna, Cittaudgadh, K%likiimiitL?,sanctum,
(north-)west, c. late 7th century'"
c. late 7th century*
(~~llth-)west,
 rig. 109 Soma, Cittaudgadh, K5lil<r?m%tj,
sanctum, north(-east),
C. late 7th century*

Fig. 1 l0 i g j n , Cittaudgadh, K5likdmBj, SanctUm, (north-)east,

c. late 7th century* Fig. 1 1 1 k i n a type, Ciltaudgadh, KumbhaSy-ama, sanctum,
south(-east), c. early 8th century:"
 Fig. 113 Nairyta?, Cittaudgadh, Kumbhaiyama, sanctum,
Fig. 112 k n a type, Cittaudgadh, Kumbhaiyiima, sanctum!
(south-)west, c. early 8th century4'
South(-west),c. early 8th century*

Fig. 115 Indra, Cittaudgadh, KumbIiaSyfima, (south-)east, c, early

Fig. 114 Varuva?, C i t t ~ d g a d hKumbhaByZma,
, sanctllm,
8th century"
(nOrth-)west, c. earlv 8th centurv*
Fig. 116 Agni, Cittaudgadh, KumbhaS- Fig. 1 17 Skanda, Cittaudgadh, Kumbhai-
yiima, south(-west), c, early 8th century* Fig. 118 Nairyta, Cittaudgadh, ~ u m b h a g -
yima, south(-west), c. early 8th century* yiima, (south-)west, c. early 8th century*

Fig. 119 Varuna, Cittaudgadh, KumbhaS- Fig. 120 Yama, Cittaudgadh, KumbhaS-
yiima, (north-)west, c. early 8th century* Fig. 121 Kubera, Cittaudgadh, ~ u r n b h a s -
yiima, north(-west) c. early 8th century*
" - ". yima, north(-east), c. early 8th century*
Fig. 122 k i n a , Cittaudgadh, Kumbhai-
YBm" (north-)east, c. early 8th century*

Fig. 123 Stirya, PipBd, PiplBdl Temple,

(south-)east, c. 8th century, quarter"

Fig. Kubera, PipBd, PiplBdmBtZ

Temple, south(-west),
2nd quarter of 8th century'"
 W . 126 Varu~a?,Pipiid, Pipliidmiitii Temple, (north-)west,
c. 8th century, 2nd quarter*

Fig. 127 Vjyu, Pipiid, Pipliidmiitii Temple, north(-west),

c. 8th century, 2nd quarter:"
 Fig. 128 TSgna?, Pipiid, PiplSdm2tii Temple,
north(-east), c. 8th century, 2nd q~~arterq:

Fig. 129 Astadik~xilaceiling, Pattadakal,

KiiSiviSvangtha Temple, c, mid-8th century"
Fig. 137 A.~ladik~>fil~
ceiling, Pathasivarama, MailareSvara Fig. 138 A,$adik1>filn ceiling, BhavanBsi Saligam, R f i p ~Saliga-
l~
C. mid-l 0th century*
mehara, c. 9th century, 2nd quarter4:
Fig. 142 Agni, ceiling, Alampur Museum, c. 9th century

Fig. 144 k n a , ceiling, Alampur Museum, c. 9th century

Fig. 145 Sfirya, DBng, Siva Temple, (south-)east, Fig. 146 Agni, Dang, Siva Temple, south(-east),
c. 8th century, 3rd quarter c. 8th century, 3rd quarter

Fig. 148 Nairyta, DBng, Siva Temple, (south-)west,

c. 8th century, 3rd quarter

Fig. 147 Yama, DBQ, Siva Temple, south(-west),

c. 8th century, 3rd quarter
Fig. 149 Viiyu, pglig, Siva Temple, (north-)west), Fig. 150 Varu~a,Dslig, ~ i v Temple,
a north(-east),
C. 8th century, 3rd quarter
c. 8th century, 3rd quarter

Fig. 151 Goddess, Diidg, Siva Temple*

(north-)east, c. 8th century, 3rd quarter
 Fig. 153 Agni, Indor, Gargaja Mahjdeva Temple, south-south-
east, C. 8th century, 3rd qarter:"

Fig. 154 Yama, Indor, Gargaja MahZdeva Temple, ~ ~ ~ t h - ~ ~ ~. t h -

west, c. 8th century, 3rd quarter*
Fig. 155 Nairyta, Indor, Gargaia Mahjdeva Temple, west-south-
west, c. 8th century, 3rd quaite"r'"
 Fig. 157 Vayu, Indor, Gargaja MahLdeva Temple, north-north-
'g. 156 Varu~!~,Indor, Gargaja Mah2deva Temple, west-north-
west, c. 8th century, 3rd quarter*
west* c. 8th century, 3rd quarter*

Fig. 158 Kubera, Indor, Gargaja Mahsdeva Temple, north-north- Fig. 159 ISana, Indor, Gargaja MahLdeva Temple, north kapili,
C. 8th century, 3rd quarter*
east, C. 8th century, 3rd quarter"'
Fig. 160 Yama, Kota (M.P.),
Central Archaeological Museum,
Gwalior, c. mid-8th century*

Fig. 161 Kubera (left

karga), Varu17a (right),
Baresarii, Temple 4, north,
c. late 8th century"
 Fig. 163 Agni, Umri, Sarya Temple, south(-east),
Fig. 162 Indra, Umri, Siirya Temple, (south-)east$ c. 9th century, 2nd quarter4'
C. 9th century, 2nd quarter*

Fig. 165 Nairyta, Umri, SGrya Temple,

 Fig. 167 VIyu, Umri, Surya Temple,
north(-west), c. 9th century, 2nd quarter* 1
Fig. 168 Kubera, Urnri, Siirya Temple, Fig. 170 Varuna, NacnI, ~aturmukha
Fig. 169 k n a , Umri, Stirya Temple,
north(-east), c. 9th century, 2nd quarter* MahIdeva Temple?, BKB, ~ a r a n a s i ,
(north-)east, c. 9th century, 2nd quarter*
9th-10th centuries*
 Fig. 172 Yama, GwHlior, Caturbhuja Temple, south(-west),
Fig. 171 Agni, Gwslior, Caturbhuja Temple, south(-east)l
C. 875 A.D. c. 875 A.D.

Fig. 173 Naiqta, GwHlior, Caturbhuja Temple, (south-)west,

c. 875 A.D.
 Fig. 175 VByu, Gwiilior, Caturbhuja Temple, north(-west),
C . 0 / J A.V.
c. 875 A.D.

Fig. 176 Kubera, Gwiilior, Caturbhuja Temple north(-east),

c. 875 A.D.
ig. aruqi1, Mathuri region?,
'01 seum of Art, New York,
early 9th century*

Fig. PSiina, MathurB, from Yamuna,

Got useum, Mathura, c. early 9th centurj
 Fig. 182 Kubera, OsiBii, SCirya Temple 1, (south-)east, c. early
Fig.. -181
. VByu, Sesai, Sfirya Temple, north(-west),
C . lOth century, 2nd quarter4'
8th century4'

Fig. 183 Agni, Osilli, Sarya Temple 1,

soutll(-east), c. early 8th century4'
Fig. 184 Yama, OsiBfi, Sarya Temple 1, south(-west),
Fig. 185 CandrdSoma?, Osiiiii, Sarya Temple 1,
c. early 8th century*
(south-)west, c. early 8th century*

Fig. 186 Varuna, Osiiifi, Siirya Temple 1,

Fig. 187 Viiyu, Osiiifi, Siirya Temple 1,
c, early 8th century*
(north-)west, c. early 8th century*
 Fig. 189 Agni, OsiBfi, Harihara Temple 1, south(-east),
Fig. 188 Indra, Osigfi, Harihara Teinple 1, (south-)east,
C, mid-8th century*
C + mid-8th century*

Fig. 191 Nairyta, OsiBfi, Harihara Temple 1,

Fig. l90 Yama, Osigfi, Harihara Temple 1, south(-west), (south-)west, c. mid-8th century:':
C. mid-8th century*
Fig. 192 Varuna, Osisii, Harihara Temple l, north(-west),
c. mid-8th century* Fig. lg3 Vayu, Osiaii, Harihara Temple 1, (north-)west, c. mid-8th
century*

Fig. 194 Kubera, OsiBfi, Harihara Temple 1, north(-east), c, mid-8th cent~~rv* Fig. 195 fiiina, Osiaii, Harihara Temple 1,
(north-)east, c, mid-8th century*
Fig. 196 Nairyta, Osisii, Sfirya Temple 2, (south-)west, Fig. 197 Varuna, Osisii, Sfirya Temple 2, (north-)west,
C. 8th century, 3rd quarter* c. 8th century, 3rd quarter"'

Fig. 198 Kubera, Osiiiii,

Sfirya Temple 2, south(-west),
c, 8th century, 3rd quarter*

Fig. 199 Kubera, Osiiii,

Pipalidevi Temple, sanctum,
C. 9th century, 1st quxter*
Fig. 201 Kubera, Osisfi,
Visnu Temple 1, north(-east),
c. 8th century, 3rd quarter*

Fig. 202 Kubera (right),

Osisfi, Visnu Temple 2,
north(-east), c. 9th
century, 3rd quarter*
Fig. 203 V a r u p (left), Nairyta, Osial?, Pipaliidevi Temple, west, Fig. 204 V5yu (left), Kubera, Osisl?, Pipalsdevi Temple, west,
C. 9th century, l st quarter*
c. 9th century, 1st quarter"

~ i206~ Agni,
. Osisfi, Harihara Fig. 207 Yaina, Osiafi, Harihara
Fig. 205 Indra, Osisii, Harihara Temple 2, south(-west), c. 8th century,
2, (soutll-)east. Temple 2, south(-east), c. 8th century,
, . c. 8th century, 3rd quarter*
3rd auarrel* 3rd quarter"
h g . 208 Nairyta, Osiiii, Harihara Temple 2, (south-)west, Fig. 209 Varuna, OsiHfi, Harihara Temple 2, (north-)west,
c. 8th century, 3rd quarter* c. 8th century, 3rd quarter*

Fig. 210 Viyu, Osiiii,

Harihara Temple 2, north(-west),
c. 8th century, 3rd quarter*

Fig. 21 1 Kubera, Osijii,

Harihara Temple 2, north(-east),
C. 8th century, 3rd quarter*
Fig. 212 k n a , Osiiifi, ~ a r i h a r Temple
a 2,
(north-)east, c. 8th century, 3rd quarter*

Fig. 213 Indra, Osiiifi, Siva Temple 1, (south-)east,

c. 8th century, 3rd quarter4:

Fig. 214 Agni, Osiiifi, Siva Temple 1, south(-east),

c. 8th century, 3rd quarter*
 Fig. 215 Yama, OsiBii, Siva Temple 1,
south(-west), c. 8th century, 3rd quarter4:

Fig. 216 Nairyta, OsiBfi,

Siva Temple l , (south-)west,
c. 8th century, 3rd quarter*

Fig. 217 Varuna, Osizii,

Siva Temple l , (north-)west,
c. 8th century, 3rd quarter:':
 .--- &-.". -,,-,&%.rr-t.-.r'daa~w~m"*iil*.rbi2-
Fig. 219 Kubera, OsiiA, ~ i v Temple
a 1, north(-east),
C. 8th century, 3rd quarter*
C. 8th century, 3rd quarter*

Fig. 220 I~iina,Osiiii, ~ i v Temple

a 1, (north-)east,
c. 8th century, 3rd quarter"
 Fig. 224 Indra, Abineri, Harsatmiti Temple, sanctum,
Fig. 223 Varup, Osijii, Harihara Temple 3, (llorth-)west, c. 9th century, 1st quarter"'
c. 8th century, last quarter*
 Fig. 227 Agni, Khed, Ra~chodjiTemple, sanctum, south(-east),
c. mid-9th century*

Fig. 228 Yama, Khed,

Ranchodji Temple,
sanctum, south(-west),
c. mid-9th century*

Fig. 229 Nairyta, Khed,

Ranchodji Temple,
Smctum, (south-)west,
C. mid-9th century*
 Fig. 230 Varuna,
Khed, Ranchodji
Temple, sanctum,
(north-)west,
C , mid-9th century*

Fig. 231 V ~ Y L I ,
Khed, Ranchodji
Temple, sanctum,
north(-west),
c. mid-9th century*

Fig. 232 Kubera, Khed,

Ranchodji Temple,
sanctum, north(-east),
c. mid-9th century*

Fig. 233 PSlna, Khed,

Ranchodji Temple,
sanctum, (north-)east,
c, mid-9th century*
Fig. 234 Indra, PHhIrpur (Bangladesh), Somapura
VihPa, Panel 29, Site Museum, c. 8th century*

Fig. 236 VIyu, PHhHrpur (Bangladesh), Somapura

VihHra, Panel 39, Site Museum, c. 8th century*
 Fig. 237 Candra, Piihiirpur
(Ban~gladesh),Somapura Vihiira,
Panel 60, Site Museum,
c. 8th century"

Fig. 239 Vayu, Piihiirpur (Bangladesh),

Somapura Vihiira, Site Museum, c. 8th
century*
/I
;!:
II~
I ,
!it
11:
II
I!!

Fig. 240 DikpEla frieze,

Gaur (West ~engal)?,
British Museum, London,
9th century"

Fig.. 241 Agni, BBramba,

SimhanHtha Temple,
south(-east), c. 9th
century, 3rd quarter*

Fig. 242 Yama, Biirainbi,

Simhanatha Temple, south,
c. 9th century, 3rd quarter2'
Fig. 243
Agni, Galliivalli,
Ksmeivara Temple,
south(-east), C. 10th
century, 1st half*

Fig. 244
Nairyta, Galliivalli,
KBmeSvara Temple,
south(-west), C. 10th
century, l st half*
Fig. 246 Viiyu, Galliivalli, Kiimeivara Temple, ( n ~ r t h - ) ~ ~ ~ t ,
c. 10th century, 1st half * Fig. 247 Kubera, Galliivalli, Kameivara Temple, north(-west),
c, 10th century, 1st half*

Fig. 248 h a , Galliivalli, Kiimeivara Temple, north(-east),

c. 10th century, 1st half*
Fig. 250 Agni, GaneSwarpur, PaiicapBndava Temple, subshrine,
(south-)east, c. 970 A.D."

Fig. 252 Nairyta, Ganeiwarpur, Paiicaplndava Temple, subshrine,

south(-west), c. 970 A.D."
Fig. 253 Varuna, Ganeiwarpur, PaficapBndava Temple,
Fig. 254 VByu, Ganeiwarpur, PaficapBndava Temple, subshrine,
subshrine, (south-)west, c. 970 A.D."
(north-)west, c. 970 A.D.*

Fig. 255 Kubera, Ganeiwarpur, PaficapBndava Temple,

Fig. 256 IiBna, Ganeiwar~ur,PaficapBndava Temple, subshrine,
subshrine, north(-west), c. 970 A.D."
north(-east), c. 970 A.D.*
Fig. 257 Indra, GaneSwarpur, Paficapundava Temple, subshrine, Fig. 258 Agni, Ganeiwarpur, Paficapindava Temple, subshrine,
(south-)east, c. 970 A.D.* south(-east), c. 970 A.D.*

Fig. 260 Yama, GaneSwarpur, PaficapLndava Temple, subshrine,

south(-west), c. 970 A.D."

d.' --
Fig. 259 Yams, GaneSwarpur, ~aiicapindavaTemple, subshrine.
south blzndra, c. 970 A . D . ~ :
 Fig. 262 Varuna, GaneSwarpur, Paiicapfipdava Temple, subshri-
ne, (north-)west, c. 970 A.D."

Fig. 263 Vlyu, GaneSwarpur, Paiicapandava Temple, subshrine,

Fig. 264 Kubera, GaneSwarpur, Paficapiiv?ava Temple, subshri-
north(-west), c. 970 A.D.*
ne, north(-east), c. 970 A.D."
Fig. 265 i ~ s n aGapeSwarpur,
, palicapandava Temple,
subshrine, (north-)east, c. 970 A.D."

~i~ 268 Agni, Dhannapuri, Mallikiirjuna Temple, ceiling, c. 9th century

pijig. 270 Nairrta, Dharmapuri, Mallikarjuna Temple, ceiling, c. 9th century
v. - O-IA ifsnn nharmapuri, Mallikiirjuna Temple, ceiling, c. 9th century
Fig. 275 Nairyta, Pratakota, NiigeSvara Temple, vinzaiza, Fig. 276 Varuna, Pratakota, Niigeivara Temple, loose image,
c. 9th-10th centuries* C. 9th-10th centuries*

Fig. 277 Viiyu, Pratakota, NiigeSvara Temple, loose image,

c. 9th-10th centuries*
 Fig. 279 ~ahi~iisuramardini,
Ellori, KailSsa Temple, SopUra,
C. 8th century, 3rd quarter"
Fig. 280 Siva Yogiivara, Ellorl, Kailisa Temple, main temple, c. 8th century, 3rd quarter*

Fig. 281 Siva Yogiivara (left side), Ellorri, Kailnsa Fig. 282 Siva Yogiivara (right side), Ellorn, Kailnsa
Temple, main temple, c. 8th century, 3rd quarter* Temple, main temple, c. 8th century, 3rd quarter*
 Fig. 283 Indra, Bikkavolu, GoliligeSvara Fig. 284 VHyu, Bikkavolu, Fig. 286 Club-bearing deity, Ksiicipuram,
Temple, south, c. 9th century'" Golii~geivaraTemple, south, KailHsanHtha Temple, priikdrn, c. 8th century,
I c. 9th century* l st quarter

Fig. 285 Soma?, KHficipuram, KailHsanstha Temple,

]>riiiciira,c. 8th century, 1st quarter
Fig. 288 Two directional guardians?, South India, Govt. Museum, Madras, c. 9th century

Fig. 289 Agni, "South India", National Museum, New Delhi, Fig. 290 Agni (profile), "South India", National
c. 9th century Museum, New Delhi, c. 9th century
 Fig. 292 Agni (profile), Tirunelveli, Govt. Museum,
Madras, 9th-10th centuries

Fig. 293 VByu, Tirunelveli, Govt. Museum, Madras,

9th- 10th centuries

Fig. 291 Agni, Tirunelveli, Govt. Museum, Madras, 9th-10th centuries

 Fig. 296 ISina, Taiijivir, Riijariijeivara, north-east, clevnk~~lilc~i,
c. 1000 A.D."

Fig. 294 Agni, Taiijiivir, RiijariijeSvara, south-east,

clevnkulikti, c. 1000 A.D."

Fig. 295 Varuna, Taiijiivfir, RiijarijeSvara, west,

clevnkulikti, c. 1000 A.D.*
Fig. 297 Devakulikii of Naiqta, Taiijiivtir, RiijariijeSvara, c. 1000 A.D.~: Fig. 298 Devakulikii of Varuna, Taiijgvtir,
Riijariijeivara, c. 1000 A.D."

Fig. 299 Devnk~llikiiof Viiyu, Taiijiivtir, Riijariijeivara, c. 1000 A.D.~:

Fig. 300 Devakulikii of Soma, Taiijiivtir,

RiijariijeSvara, c. 1000 A.D."
Fig. 301 Siva Vaiviihikamfirti, Kanauj, c. 8th century4: Fig. 302 Siva Vaiviihikamfirti, Kamiin, National Museum, New
Delhi, c. 8th-9th centuries

Fig. 303 Siva

Vaiviihikamiirti (detail),
Kamiin, New Delhi,
c. 8th-9th centuries
 Fig. 304 Siva Vaiviihikamiirti, Mlinasigaligii, Govt. Museum, Mathura, 9th-10th centuries*

I
Fig. 305 Siva Vaivshikamiirti, Mathurii, Indian Museum Calcutta,
9th-10th centuries*

Fig. 306 Siva Vaiviihikarniirti, Etah J.P.), BKB, Varanasi,

c. 10th century
Fig. 307 ViSvarCipa, south-eastern Rajasthan, Linden-Museum, Fig. 308 ViSvarCipa (detail), south-eastern Rajasthan,
Stuttgart, c. 8th century* c. 8th century*

- 3 10 ViSvarCipa,
Fig.
Tulnain (detail),
c. 8th-9th centuries

Fig. 309 ViSvarCipa, Tumain, National Museum, New Delhi, c. 8th-9th centuries
Fig. 31 1 Manu Lokaprila panel, Southern Bihar, MIK, Berlin, c. 8th century"

Fig. 312 Manu Lokapiila panel, Bihar, State Archaeological Museum, Patna, c. 8th century*

Fig. 313 Manu Lokap2la panel (incomplete?), Guneri, Bodhgaya Site Museum, c. 9th century*
 Fig. 3115
Agni, Rajasthan,
MIK, Berlin,
-. - ! - -. --- -- c. 10th century*
Fig. 314 Agni, SBrnLth, Archaeological Museum, Sarnath,
c. 6th-7th centuries

Fig. 3 16 Varuna, Madhya Pradesh, National Museum, Fig. 317 Varuna, Rajasthan, Jhalawar Museum, no. 64,
New Delhi, c. mid-8th century c. 9th century*
Fig. 3 18 VIyu, GandhIra region, MIK, Berlin, c. 2nd-3rd centuries4:

Fig. 320 VByu, Rajasthan (?),

c. 8th century4:
Fig. 321 Kubera, unknown findspot, BKB, Varanasi, Fig. 322 Kubera (?), Naresar, Temple 17, west,
c. 6th-7th centuries c. early 8th century:"

Fig. 323 Dandapiini, unknown findspot, BKB, Fig. 324 Dandapiini, Hingliijgadh (M.P.), Central Museum, Indore,
Varanasi, c. 5th-6th centuries c. 10th century*
Fig. 325 Harihara-Dandapiini, Baijnlth (HP.), Fig. 326 NateSa with diklx?lns, Pattadakal, Malliklrjuna Temple, ceiling,
,$iva Temple, c. 13th century* c. 8th century, 2nd decade*

Fig. 327 Visnu AnantaSayana, eastern Rajasthan (?), MIK, Berlin, c. 9th century*