Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Phāṁsanā in Western India

Author(s): Michael W. Meister


Source: Artibus Asiae, Vol. 38, No. 2/3 (1976), pp. 167-188
Published by: Artibus Asiae Publishers
Stable URL: http://www.jstor.org/stable/3250133
Accessed: 18/01/2010 17:21

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=artibus.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Artibus Asiae Publishers is collaborating with JSTOR to digitize, preserve and extend access to Artibus Asiae.

http://www.jstor.org
MICHAEL W. MEISTER

PHAMSANA IN WESTERN INDIA

he temple in north India normallyhas a curvilineartower made up of vertical bands (the


typicalLatinaor mono-spired Two othertypesalsooccur:Valabhi,a
Nagaraikhara).
barrel-vaultedroof perhapsbest knownfromthe Teli-ka-mandir at Gwalior,andPhramsana, a
pent-roofof severalstoreysperhapsbest exemplifiedby its earliestexample,the templeat Gop
on the Saurastracoast(fig.6).I
In studying the ruined structure of the Hariharatemple no. 3 at Osian near Jodhpur in
Rajasthan(figs. 9-Io) I came to the conclusion that it had originallyPhamsanaroofing rather
than Valabhi as had been suggested by several past writers.2To support that conclusion, how-
I choseto expandmy investigationto include
its broadersignificance,
ever,andto demonstrate
the use of the Phamsanaform throughout WesternIndia. The results of that researchI present
here.3
is used in medievaltexts to refer to
The term "phamsana"(meaning"wedge-shaped")
pyramidalstructureshaving several layers. Such structuresin the medieval period are used to
roof mayadapahalls ratherthan the temple's principalsanctum; their multiple layers often take
the form of curved cornices (as example, a late use of Phamsanain WesternIndia, the roofs of
the Madhav Vav, a i3th century step-well at Wadhwan in Gujarat-fig.8). The layers of a
a cornicewith quarter-
roof, however,cantakeone of threeprofiles:straight-edged,
Phamhsana
round profile (kapota), or a reflexedcornice (kapotadll).Of these the straight-edgeseems most
accuratelyto reflectthe originalform.4
Roll corniceswere used as earlyas the Guptaperiodto buildup towersrepresentingthe
Medieval roll-
multiple storeys of a palace (a type I have elsewhere labeled bhbmi-prasada).5
cornice Phamsanasadapt that type's cornice layeringto the lower pyramidaloutline of a pent-
roof. The condensedbhbmi-prasddas of the Gupta period were meantto representmany-storeyed

I Phamsanaand Valabhiare termslargelyresuscitatedby the works of M.A.Dhaky.For chapterson both see his The
PrincipalFormsof IndianTempleSuperstructure, Varanasi (1975 still in press).
2 See references given in note 14.
3 This article was first written in summer of I973 and was included as part of a chapter in my doctoral dissertation Form in
theNorth IndianTemple(unpublished dissertation, Harvard University, Cambridge, 1974). All photographs and plans are
by the author save where other acknowledgments are given.
4 Phamsameans "wedge" in modern Gujarati.For the application of the term to temples in Vastu texts see J. Nanavati and
M. A. Dhaky, TheMaitrakaandtheSaindhavaTemplesof Gujarat,Ascona I969, p. 27 note 70; see fig. 9 for the three principal
types of tiers. Dhaky gives a reference to the late I xth century Pramanamanjar! which calls the sloping "wedge-shaped"
roof of a house phansakara.The term is used in the Samarigganasatradhara, the T.rkdrnavaas well as forming the subject
of an entire chapter in the Aparajitaprccha.
5 "An Essay in Indian Architecture", RoopaLekha, XLJ (1973), pp.3s-47, notes; and the dissertation cited in note 3,
Chapter i.

I67
structures.Theydifferfrom medievalPhamsanaformulain thatthey are crownedby dmalaka
(ribbed stone) and not by ghatad(bell), and have tall rather than shallow silhouettes (a later
examplefromAbanerimimickingthis pre-NagaraformI illustratein fig.5).
Medieval texts specify that "Phamsana' be built with "no space" between layers,6which
seems to distinguishPhamsanafrom bhbmi-prdsddatemples, which often have miniaturestoreys
betweentheircornicelayers.We can,in anycase,be surethata laminated-cornice structureis
meantto be Phamsanaonlyas it is madeto conformto the lowerprofileof the pent-roofandis
crowned by gha.ntdratherthan amalaka.The medieval Phamsana,though it adopts the cornice
layersof theprasdda wasintendedto mimica pent-roofandnot the storeysof a palace.7
structure,
The earliestpent-roof Phamsanapreservedis that of the temple at Gop, its double-layered
roof faced by bold gavaksa(or candrasala)dormers, its two gabled levels crowned by a large
ghan.tdbell (fig.6).8A numberof temples in the 7th centuryuse a pent-roof structureas decora-
tive pediment over wall-nichesbut none shows a crowningghatad.9Shrine models over door-
ways at Aurangabadand Ellora, however, show structurescrowned by ghatan;and the side
faSadeto Cave 9 at Ellora (fig. 7) must certainlyrepresenta pent-roofed meeting hall.
Gop representsin stonewhatStellaKramrisch hascalleda "pent-roofof laminatedboards"
andwhichshefelt "didnot lenditself [in stoneandbrick]to greatdevelopment... due to [its]
meagrenessas a plastic form".Io Several writers have tried to connect Gop to a "Kashmiri"
type,but the only connectionbetweenGop andIashmir templeswith gablesuperstructures
is
their common attempt to make a translationinto stone of a wooden original. They are in diffe-
rent styles,had differentmodels,achievedifferentresults,and are of differentdates,so past
attemptsto link themcan be discarded.,I
It is possible still to see in house architectureof parts of Northern India some parallelsfor
the Phamrsanaform:a villagehouse,however,showsonly a low pent-roofwith an upperlevel
allowinga kindof clerestoryfor ventilation(ratherlike the mavdapa
hall attachedto the Para-
suramesvaratemple). Near Aharaurain Uttar Pradesh, however, I have seen a large wooden
structurebuilt over a saint'stomb whichshowedthreetiersof pent-roofsupportedon wood
6 Dhaky, PrincipalForms...,
op.cit., quoting Aparajitaprcchb.(I thank Mr.Dhaky for allowing me to read his manuscript
while still in the press.) Nanavati and Dhaky, op.cit., pp. 27-29, in discussing their "Pharhsakara"class, do not distinguish
between shrines of low or high profile or crowned by gha.ntaor amalaka.Dhaky, in the more recent PrincipalForms...,
points to the prescription in Aparajitaprccha requiring that Phamhsana be crowned by ghanta.In discussing the three basic
awning types used to build up Phamhsana roofs Dhaky comments that "though used with the same intention and for the
same functional purpose they are morphologically quite distinct from each other and each has its independent origin".
7 Nanavati and
Dhaky's Phiamsakaraclass, op.cit., pp.43-5I, pls. 7-24, would seem to contain true Phamhsana structures,
with low profile and crowning ghanta,some roll-cornice structureswith low profile but amalaka,and temples with a tall
profile and crowning amalakawhich are "pre-Nagara"in form, or even, if one wishes, "pseudo-Nagara" in that they
exist contemporaneously with developed Nagara structures (as did the curvilinear bhnmi-pr&asda temples at Jagesvara).
8 c. 600 A. D.: Nanavati and Dhaky, cit.,
op. PP.33-40, 77-78, pls. 7-8; James Burgess, Reporton theAntiquitiesof Kathiawad
andKachh(Archaeological Surveyof WlesternIndia,II), London I 876, p. 187, pls. LI-LIII; H. Cousens, Somanathaand Other
MedievalTemplesin Kathiawad,ASI New Imperial Series XLV, London I93I, p. 37.
9 Mundesvari, Kusumi, Alampur, Aihole (Gaudargudi, Durga
temple), Pattadkal (Galagnath), Bhubane{vara(Parasura-
mesvara, Bharategvara).Some 6th century fragments from Sondani and Nagarl suggest also the existence of pent-roof
forms.
10 The Hindu
Temple, Calcutta I946, p. 220.
1 Burgess, Report..., loc.cit., made no mention of Kashmir. In TheAncient
Monuments,TemplesandSculpturesof India, Lon-
don I897-911, pt. II, p. 35, he does draw the Kashmiri parallel. This is stated more strongly by Cousens, Somnatha...,
op.cit., p. 6, and in TheAntiquitiesof WesternIndia, London I 926, p. 13; H. D. Sankalia,TheArchaeology
of Gujarat,Bombay
1941, P. 57; and by Percy Brown, IndianArchitecture:Buddhist,Hindu,Jain, Bombay 1959, pp. I60-I6I.
hallsoutsidetemplesin Balialso give some reflectionof the wooden
posts. The dance-drama
modelson whichsuchtempleswereprobablybased.Templesin Keralawith woodenroofsstill
providea pent-roofmodel,havingalsogavdksadormerson eachlayer.12
In Rajasthana small number of pent-roof Phamsanastructuresfrom the 8th centuryare still
preserved. Some of these are miniature shrines, either independent or used as corner kutas
is used in referenceto
(aediculae)on larger structures.In texts on architecture <'akara-kuta"
set on the cornersof a largerstructure(figs.13-I4,19).Thesecanbe either
smallshrine-models
Nagaraor Phamhsanain form."Ku.ta",however,mayonce havereferredonly to a smallshrine
having pent-roof structureand a squareplan; its later, more general applicationmakes it diffi-
cult, however, now to utilize the term in so limited a sense.,3 Beyond these small kfta-structures
more complicatedpent-roofs did exist, one of which was the originalstimulusfor this discussion.
Thoughmoreoften preservedin stoneas a roof for hallsthanfor the sanctumof temples,
the Pham'sanaform was used to roof the sanctumof some temples, including that of Harihara
templeno. 3 at Osian(figs.9-IO). Manyscholarshave beenled to conclude,by its rectangular
plan (fig. I), that this temple must once have had a Valabhiroof.14The preservedfragmentsof its
form.
however,leavelittle doubtof its actualPhamhsana
superstructure,
Figure9 shows the Hariharatemplefrom its long (west) side. Vedibandha mouldingsof
kugmbha, half-lotuses,andwith
kalasa,andkapotali(thelatterheavilydecoratedwith candrasalas,
kalikd-budssuspendedbeneath)supporta plainwall with ornamental appliedniches.The wall
shows a band of half-lotusand pearlchainsabove, just beneaththe vara;.dikd-cornicewhich
carriessculptedscenesof Krsna-Lila.Above the bhadraandpratiratha(centerand flanking)
offsetsappearthree niches undera ribbedawning (khuracchadya)-the centralone showing
Yoganarayana flankedby Cakrapurusa andSafikhapurusa.On thecornersappearsmallminiature
phatsana-kf.taswhich are connectedto the centralniches by a vedikd-railing with capping
corniceswhich simulatea surroundinghdra(figs.io, I3). Above this appearthe lower half-
of a giant simhakaryawhich would have faced the entire superstructure.A second
candrasalads
smallersithakarpawould have been set above the awning which shadesthe niches.Is
The small phimsana-kutason the corners (fig. I3) show two pent-roof layers crowned by
andkalasa.The pent-roofsarestraight-edged
dmalasarikg,
gha.n.ta, and ribbed.Udgama patterns
face eachlayer.Above and to the right of this karna-k.ta(figs.9-Io) a cornerof the pent-roof
12 K.R. Srinivasan, Templesof SouthIndia, New Delhi 197I, pl.25 (the Vadakkunnathartemple at Trichur).
13
Suggested in conversation with M.A.Dhaky. Coomaraswamy, "Early Indian Architecture: III...," Eastern Art, III
(I93I), p. I9I, writes: "The term kttdgdra,especially when combined with Jala,may sometimes mean an entire building
with a peaked roof... but I am not convinced of this; and usually the ki.tdgdrais a self-contained and separatelyroofed
pavilion on any storey of a pdsdda,either a gabled pent-house on the roof... or more often a gabled chamber on other
does not exclude the idea of a domed or barrel-
storeys..." On p. 193, however, he notes that "it is clear that kuttdgdra
vaulted roof".
I4 S. K. Sarasvati, in M:ajumdar,ed., TheStruggle for Empire (Historyand Cultureof the IndianPeople,vol. 5), Bombay I95 7,
PP. 555, 578; Krishna Deva, Templesof North India,New Delhi I 969, p. 3I. Nanavati and Dhaky, op.cit., p. 42. doubt that
the Kadvar temnplehad a Pharhsanasuperstructurebecause of its rectangularplan, stating that "the superstructurethat
would have been fitting is the valabhi,the wagon-vault type which occurs in a number of slightly later monuments in
other parts of the country, and was indeed habitualfor oblong structures".On the other hand, in "The Temples of Osia",
Archaeological Surveyof India, Annual Report, 908-09, p. o4, D.R. Bhandarkarsuggests a parallel between the Harihara
no. 3 superstructureand that of the small Phamrsanishrine two miles south of Osiani(my figs. I6-I7), but only on the
inconclusive ground that both show a heavy khuracchddya-awning. Hariharano. 3 in fact shows this awning only over the
bhadra-nichesof the superstructure,not as a transition from shrine to superstructureas on the Devi shrine.
15 Diagonal lines can be seen on the slabs against which these simbakarpas were once set in fig. 9.

169
whichmadeup the centralsuperstructure canstillbe seen.It alsois straight-edgedandribbed,
paralleling the miniature kuta below. A second pent-rooflayer would have appearedabove,
crownedby a largeghatda.(The largeghatadappearingin fig.9 is on the maydapa behind,not
on the mulaprasdda.)
To arguethata rectangular planprecludesa Phamsanaroof is not appropriate for Western
India,at least in Maru-desa.The strangetempleat Teori (fig.12), the Kamesvaratempleat
Auwa (figs.2, 27-28), and the side sub-shrinesat Dhamnar(fig.22) all arerectangularshrines
with Phamsanaroofs.16 It would seem,in fact, that this was the only acceptablesolutionin
Maru-desa, fromwhichno Valabhishrinesareknown.17
An interestingarchitectural drawingof a Phamsanastructureis scratchedon the backface
of the kaksdsana of the Hariharatempleno.2 at Osian(fig.I I). It showsno facingsimhakarpa,
the
leaving pent-rooflayerscompletelyvisible.Its angleof batterduplicatesthatof the Maha-
vira templehall (fig.13), but in placementof decorativecandraiaads it perhapsmore closely
the
parallels peculiar structureat Teorithirteenmilessouth of Osian (fig. 2).
The Teoritempleprobablydatesfromearlyin the gthcentury.It showsno basemouldings,
andno khuracchddya
aboveits wall,onlya heavykapotali
decorated
withcandrasiads
andhalf
lotuses. This corniceis separatedfrom the first storeyof the superstructure by dentils;this
fromthe secondby a square-and-diamond pattern.A furthercornicedecoratedwith candrasaads
acts as skandha supporting an dmalaka ratherthanghantad.The whole is so rude as to almost
warrantthe conditionin whichit is found.18
Let me comparebrieflya seriesof smallpent-roofku.tas.The kar.a-k.tason the Harihara
templeno. 3 andthe Mahaviratempleat Osian(figs.13-14)both representopen pillaredstruc-
tures, the first showing "piled-pot"pillarswith dmalakacapitals,the latterwith a reduced
ghata-pallavatype of pillar.The firsthaskapotadl-cornice abovethesepillarsanda roof of ribbed
phamsanad-layers. The latterhas a ribbed andthephawsanad-layers
khuracchaddya-awning, areplain.
The kutasof the Mahaviratemple(fig.14)have additionalhalf-udgamas set to eitherside of the
centraludgama (reflectingthe halfsimhakaryas of the mainroof-fig. I9). Neitherof theseku.tas
show basemouldings.The Mahavira templeaediculaehousedancersandmusicians.Theyare,in
a suggestiveway, the early equivalentof the open corner-pavilions of much later Rajput
architecture.
The two tiny structuresacrossthe roadfromthe Harihara templesat Osiani(fig.I5) arefull
miniaturetemples,with base mouldings,bhadra niches, and in one casea decorateddoorway
(withNavagrahaover the door).The wallsaretoppedby a heavykapotdli-cornice as at Teori,
but alsoby a khuracchadya awningroughlyscoredratherthanwith the carefulribbingshownby
the awningof the Mahaviratemplepavilion.Thephthsotnan-layers show singleandhalfcandra-
sJldsrather than full udgama as decoration.

16 Teori: Archaeological Survey of India, Western Circle, Progress Report, Igo906-07, pp. 35-36. Auwa: ASIWC, PR, I908-og09,
pp. 49-50; I 914- 15, pp. 80-8 I; M.A. Dhaky, "The Old Temple at Lamba and Kamesvara Temple at Auwa", Journal of the
Asiatic Society, Calcutta, VIII (I966), pp. 145- 148 and figs. 5-9. Dhamnar: ASIWC, PR, I 9I I-I 2, p. I 7; I92-I3, p I.I
The Auwa temple measuresc. 4.8 by 5.7 meters (corner to corner); c. 2.5 by 3.3 meters in the sanctum.
17 The only Valabhl shrine I know from Rajasthainis from Satwas (east of Kaman near Mathura)which is more related to
the style of Madhyadegathan to that of Western India (ASI, WC neg. no. 5382)?
18 My fig. I2 is Bhandarkar'sphotograph of Teori taken in I906. Villagers report the temple still standing in spite of its
seemingly precarious condition.

I70
JAGATI

GARBHA GRHA
*

RANGA MANDAPAo e

BELOW ABOVE one meter


a sanapattaka ksanapattaka

Fig. i Osiani,Dist. Jodhpur, Rajasthan.Hariharatemple no. 3 (ca. late 8th century), groundplan.
SUPERSTRUCTURE

SANCTUM

I I
I I
i _ _J

asanapattaka asanapattaka
Fig. 2 Auwa, Dist. Pali, Rajasthan.Kamesvaratemple (ca. mid gth century), groundplan. one meter

Fig. 3 Osian, Sri Satyanarayanatemple (ca. early 8th century):


A) plan of superstructure; B) groundplan of sanctum.
\ ii
;I 1
I mo

Fig. 4 Osifan,Sri Satyanarayanatemple, Fig. 5 Abaneri, Dist. Jaipur, RajasthIn.


plan showing ancient parts of present temple. Shrine model (bhzimi-prasgdatype) (ca. 800 A.D.).

Fig. 6 Gop, Dist. Jamnagar, Gujarat. Old temple (ca. 60o A.D.),
superstructure.Photo courtesy M.A. Dhaky.
Phriamsana
Fig. 7 Ellora, Dist. Aurungabad, Mahiarstra.Cave 9 fagade (ca. early 7th century).
Photo courtesy American Institute of Indian Studies, Varanasi.

s ~~4A, I_
5.'I
S^'0 '-^
Fig. 8 Wadhwan, Dist. Surendranagar,Gujarat. Madhav Vav Fig. 9 Osian, Hariharatemple no. 3, view from west.
(I 3th century A.D.), Phiamsanaroofs over step well.
Fig. Io Osiani,Hariharatemple no. 3, northwest corner from west. Fig. 12 Teori, (Tivri), Dist. Jodhpur, Rajasthan.Phamsanashrine
(ca. early gth century), called Khokri-Mata-ka-mandir.
Photo courtesy Archaeological Survey of India,
Western Circle negative number 2806.

Fig. I I Osiani,Hariharatemple no. 2 (ca. late 8th century), inner face of south seat-back,sketch of Phiramhsana
roof.
Fig. I3 Osiai, Hariharatemple no. 3, kuta on northwest corner of shrine. Fig. 14 Osianfi,Mahaviratemple (ca. late 8th century), phadMsana7
kzta on northeast corner of mandaparoof.

Fig. 15 Osian, Pharhsanashrines west of Hariharagroup.


Fig. I6 Osian, Devi shrine two miles southwest of Osiani
(ca. mid gth century), view from east.

Fig. I8 Arna, Dist. Jodhpur, Rajasthan.Phamrhsana


shrine
(ca.late Ioth century), west view.

Fig. 17 Osiani,Devi shrine from west.


Fig. roof a temple, g a-maapa, Ph san east.
from

Fig. 19 Osian, Mahavifa temple, gudha-mandapa,Phamsana roof from east.

-
i,.wt-

.
t . add --N

Fig. 20 Osiani,Mahaviratemple, view from northeast (tower later restoration).


Fig. 21 Dhamnar, Dist. Mandasor, Madhya Pradesh. Rock-cut Dhamnesvara temple (ca. mid gth century),
view from northwest.

Fig. 22 Dhamnar, oblong shrine north Fig. 2 3 Kotai, Dist. Bhuj (Kutch), Gujarat. Siva temple (ca. early loth century),
of main temple, view from southeast. view from south. Photo courtesy ASI, Baroda Circle.
Copy photo of ASI Western Circle Neg.
No. 3736 courtesy India Office Library.
Fig. 24 Osian, Sri Satyanarayanatemple,
view from south (walls above platform modern).

Fig. 25 Osiian,Sri Satyanarayanatemple, superstructurefrom southeast.

Fig. 29 Chatsu, Dist. Jaipur, Rajasthan.


Miniature Phamhsanashrine (ca. late gth century).
Photo courtesy ASI, Western Circle neg. no. 3496.

Fig. 26 Osian, Sri Satyanarayanatemple,


superstructure,front (west) view.
Fig. 27 Auwa, Dist. Pali, Rajasthan.Kamesvaratemple (ca. mid gth century), superstructure,front (west) view.

Fig. 28 Auwa, KImesvara temple, view from southeast.


Fig. 30 Osian, Saciyamatatemple,
southwest sub-shrine
(ca. late Ioth century),
superstructurefrom northwest.

Fig. 3I Osiani, Saciyamatatemple,


northeast sub-shrine
(ca. early IIth century),
Sarhvaraniroof of mandapa,
view from south.

Fig. 32 Osiani,Saciyamatatemple,
northwest sub-shrine,
Sarvarana roof from southeast.
The small Devi shrine two miles south of Osian, which Bhandarkar thought the model for
the Harihara
no. 3 superstructure
(figs.I6-I7),19revealsits relativelatenessin severalways.The
doorway seemsadLvanced in style; the niche is enlarged,its pedimentcrossingthe uppermoulding;
the vedibandha mouldings have lost some of the squarenessseen in earlier shrines; and half-
lotuses with sharp triangularformat decorate the kapotadli-mouldings. The sunken pattern be-
tween the two tiers does not appear on earliershrines at Osian. Thegha?ta is ill-formed; the
khuracchadya relatively larger and with greater spread than that on earlier shrines. Still, for a
temple hardly of human height, there is a dignity, especially when viewed from the front
(fig. I7).
Though the upper tier of this shrine is straight-edged,reflectinga pent-roofed model, the
lower tier has taken on a curvedprofile,showing a degradationof the originalstructural
reference,and a returnto the overridingpatternof roll-cornicelayers.
A small Ioth century shrine at Arna, some ten miles from Jodhpur (fig. 18), completes my
surveyof miniatureshrines.20
This shows the preservationof the kuitatype, but is built up
totally of kapotadi'
layers. The plain khuracchadyahere is particularlyelegant, giving a reflex
curveto the overallprofileof the temple.
Though wooden pent-roofs may once have been used in ancientIndia to roof large assembly
halls (as the fasade to cave 9 at Ellora would seem to suggest-fig.7), in later periods the
Phamsanaroof was used in stone primarilyto roof mandapas which stood as entry-hallsbefore
the main shrine.2:These mediaeval Phdamsana structuresoften were built up of roll-cornices,
losing the boldness of the slope-roof. Two earlier halls in Western India, however, preserve
properpent-roofphamsanas.
The Mahavira.temple at Osian provides perhaps the boldest manydapa roof in all of India
(figs. I9-20). Its hLeavytri-bhumaPhamsana,like the sails of a schooner,carriesthe templefor-
ward. On each corner of the first bhimi are set s.rngs, on the second tier phadsana-kutas. An
excellentghantad crowns the third tier, with admalaka
and kalasaabove that. Niches with a ribbed
chadyabase the superstructure.
Checkerboard
grilleseparatesthe first bhbmifrom the chadya
below and from the second bhbumi
above. The third bhbmirises above a band showing a row
of pillarets(vedikd)with diamondsbetween.
The projecting bhadrabalcony below supports two srngasand a series of niches which
themselvessupportascendingsimhbakarna
pediments.The two upperlevels of the pent-roof
also carry central simhakarnadecoration. The first and second levels in addition carry half-
simhakarnapatterns to either side spread and arrangedin such a way as to suggest a single
giantsimhakarvabut without obscuringthe structureof the basicpent-roof.Such clarityof
form and balancebetweenelementsmakethe structureaestheticallystrong-perhapsthe most
perfectexampleof this class.
The mukha-mayndapa
projectingin front also is roofed by a (two-tiered)Phamsanaroof as is
the mukha-catuski
in frontof that.The mukha-catuski
bearsa crowningghanta,dmalasdrikd
and
kalasawhich, witl the ghatadof the main Phamsanaand the assumedadmalaka
over the original

19 ASIAR
I908-09, p., I04.
20 ASIWrC
906-07, pp. 33-34.
21
By the I Ith century even this use has disappearedin Western India, though Phramsana
roofing occasionally is encountered
later, as in the series of pyramidalroofs used to cover the Madhav Vav at Wadhwan, ASIWC I 898-99, p. 5 (fig. 8).

I83
sanctum-tower, form an ascending chain (fig. 20): that progression of structures which so im-
pressesat Khajurahoand for which there is so little evidence before the Ioth century.
Another large pent-roof Phamsana covers the g.dhamaydapaof the rock-cut temple at
levels are
Dhamnar(c. mid gth century)betweenMandasorand Kota (fig.2i). Thephamsana
somewhat compressed.The s.rngas on the firstlevel are large, standing out from the body of the
Phamsana;the pent-roof levels are withdrawnand rathershallow. An additionals.rngahas been
added in the center of each face, behind the simhakarya-pediment over the bhadrabalcony; this
seemsan awkwardaddition,destroyingthe balanceseenat Osian,andthesymmetrical
advance-
set to eitherside.
mentof the halfsimhakaryas
The phbdsanad-ktas of the second bhbmihave been reducedvirtuallyto the tilakasor kutakas
of later medieval structures.Thegha.tadshows little nobility, and the heavy ribs of stone extend-
ing bothfromtheskandha andfromthe secondpent-roofas supportfor thesimhbakar.nasgive the
effectof greatplanksset on hobbyhorsesto supporteachtier. This effectof tiers restingon
boards is even more strikingin the rectangularside-shrines(figs. 21-22) where the bhbmishave
a curvedprofile.
The totalcomplexat Dhamnar,with fourNagarasub-shrinesset at the cornersof the exca-
of the prasdda,is impressive;but the
vation, three rectangularshrinesopposite the bhadras
Phtamsana
srngasof the ma;ydapa's roof so echo both the Nagarasub-shrinesand the central
Latina tower, amplifying the pre-eminence of Nagara over Phamsana, that the may.dapa's
shallowPhamhsana roof is reducedtruly to a neutralcushionbefore the masculinityof the
Nagaraspire.22
Followingthese few shrines,the true pent-roofedPhriamsana disappearedin Maru-desa,
replacedby a Phamsanamadeup of roll cornicesor even ribbedchadya-awnings. In Gurjara-
desa,however(thepresentGujaratandsouthernRajasthan), a modifiedformof straight-edged
phadtsan-layering continueduntil the Ioth century.In the 8th centuryat Roda23Phramsana
roofs had been used as superstructure for antarala-porticoes, actingas ukanadsa for the shrine.
This bold use of pent-roofphdasanan graduallyis replacedby a morelaminatedform,but one
which preservesa straight-edged profilefor its elements.Thisformis used in Gujaratboth as
sukandsa for the templeandas roof overtheg.dhaman.dapa (thetwo meldedtogetheras nowhere
elsein India).This phaseI illustrateby the Sivatempleat Kotai,Kutch,ca. earlyioth century
(fig.23). (Phanhsana pedimentshereareusedalsooverthe bhadra projectionsof the mainshrine.)
TheKamesvara templeat Auwa(Palidistrict,Rajasthan), built on a rectangularplan,usesa
navdnydaka formfor its sikhara,but with apham'sana centralplatformratherthana Nagaraspire
(figs.27-28). The high roll-phadsandabove the secondlevel of srngas(fig.27) and the single
skandha-layer abovethatareoriginal.The threeupperlevelsandcrowninggha.ntd seenin figure
28 are recent additions replacing an ancient gha.n.ta2The single large S'urpa-shaped
phdasana
has a squareprojectinglip instead,whichis
showsreflexcurvature;the shallowupperskandha
the standardkapotaliformulathroughout the 8th and much of the gth century.On this skandha-
as "neuter", Latina as "masculine", and Valabhi as feminine; Nanavati and Dhaky,
22 Medieval texts refer to Phamrhsana
op.cit., p. 27, note 70.
23 See U. P. Shah, "Sculpturesfrom Samalajiand Roda", special issue of the Bulletinof theMuseumandPictureGallery,Baroda,
XII, (1960).
24 Added by the village Panchayatin c. x969-70. The original gha.nt which was still partially in place at the time of both
Bhandarkar'sand Dhaky's visits (see note 6) now lies in pieces at the base of the temple.

I84
platformoriginallyrested a bold gha.td, the pieces of which still lie near the base of the temple.
This would have completedthe steep,straight-edged profileof the originalspire,the Nagara
and
sub-spires recedingsimhbakar.as buildingup to theghav.td
crown.
As a balanceduse of the navdaJdaka
formula, combined with a singlecurvedphasana-level,
to roof a rectangularshrine,the Auwa templeis unique.As a varietyof Phiamfsana aloneit
warrantsonly slight mention,so dominatedis it by the panatisof the narvaedaka
form, by the
srngasand receding sim'hakarnas.
Still, it follows the Osian pattern of choosing Phamsanarather
than Valabhifor roofing a rectangularplan; and it more successfullybalancesthe elementsof its
spirethan any otherin the gth century.It has a cohesionand unity otherwisenot
navdandaka
found in anekdadaka
spires (at least until the Visvanathatemple at Khajurho, ca. oo002A.D.,
forcesthe Naigara
centralspireto coherewith the surrounding
srngas
by compressingthepankti-
levelsandthe centraltowerinto a singleform).
A much earlier small temple at Osianii(figs. 3-4, 24-26), the Satya Narayanashrine on the
Saciyamatahill (dating earlyin the 8th century), provides also a rectangularplan crowned by a
peculiarsuperstructure which, while properlyneitherPhamsananor Valabhi,is, in a way,
ancestorto the Auwaroof. A bold sitmakara(framingan image of
Gan esa)frontsthe sukanasa
on the temple'swest side(fig.26). Behindthat standsa secondsimbakaryawhichfrontsa pent-
roof dormerprojectingfromthe broadfaceof the rectangular A seconddormer
superstructure.
appearson the oppositeface,forminga cross-plan(figs.3, 25). The simhakarna-patternswhich
were to face south, east, and north have only the lower level carved; the upper part, partly
restored,is only blockedout (fig.25).Zs At the crossingof the pent-roofa simpleskandha-plat-
formrests,supportingan archaicg ta anda kalataandbyiapuraka finial.
Both the longer rectangle and the crossing dormers of the superstructureare divided into
two levels. Separatedfrom the upper cornice of the wall26by a narrow,undecoratedrecess and
a band of chain-loops and half-lotusesis a curved Sjrpa-shapedlevel which correspondsto the
lower half-candrasadlds
of the sibhakarya antefixes. The edge of this s'rpa is decorated with
florets. A narrowneck and band of sawtooth decorationthen separatesthis curvedlevel (which,
on a very ancientmodel, correspondsto the curvedroofing of the side aislesof a caityastructure),
from an upper hip-roof.
Were this structuresimply a rectanglewith a sukandsaprojecting in front, it might pass for
Valabhi, though it does not have a keel-roof. (There is some relationshipbetween this and the
Valabhi entry-structurefronting the small stupa-shrineat temple site no. 3 at Nalanda,27as to
the sukandsaprojectionsof a number of temples.) The Osiarnstructure,however, shows a cross
plan. Simhakarnasface pentroof projections on all four sides, and a skandha-platformwith
ghata is supportedabove. Becauseof these crossed pent-roofs and the crowningghatadI should
prefer to consider this temple relatedto the Phamsanaform. Though close to Valabhi, simha-

Bhandarkar's photograph (WC neg. no. 2852) shows the superstructure


25 partly collapsed but with enough in place to
confirm the present restoration. The carved lower levels show images of Ardhanarisvara
(S), Brahma (E), and Gaja-
Laksmi (N).
26 This heavy varaVdika(complex cornice) no longer is visible because of the roofed ambulatory now added to the
temple,
built over the original openjagaf-platform since Bhandarkar's visit. The varandika visible in Bhandarkar's
photograph
consisted of two heavy kapotali-cornices separated by a broad
antarapatra-recess decorated with foliate pattern. Beneath
the upper kapotali a single offset is decorated with scalloped pattern as on the sun
temple on the same hill.
27 Prudence Myer, "Stupas and Stupa Shrines", Artibus Asiae, XXIV
(I96I), fig. I.
it seems to show a strong predilectionon the part of
karyareplacingthe sun-window candrasdla,
its builderstowardPhramsana
form.
In a sense, this temple offersa rationalefor the cross of simhakarya-faced projectionswhich,
filled in with srngas,supports the singleph dsana-layer at Auwa (fig. 27). This is not to say that
this peculiarlittle shrine produces or is the root of later forms; only that it shows certainpre-
occupations in its cross-form and rectangularplan, pent-roof and crowningghaftadwhich also
are reflectedin other temples I have been discussing. We cannot fully trace the origins of the
tradition,at leastso faras the com-
form,but one partseemsto lie in the Pharhsana
anekd.ndaka
dormersand the cornerkutas)
plexitiesfound on the Osian Mahaviratemple hall (the simhakarpa
canbe takenas a reflectionof earlierdevelopmentswithinthe Phramhsana
tradition.
From Chatsu,south of Jaipur, the center of a Guhila dynasty,feudatoriesof the Pratiharas
in the gthcentury,comesa smallshrinewhichillustratesa furthervarietyof Phamrsana
(fig.29).28
on the sides appearsithba-
Over the short kapil-projection in front and the bhadra-projections
karas. Between these on the cornersappearkhandascrowned by karndyakas(bhfmni-amalakas,
as in the veyukosaof a Nagara tower) which support a second large simhakaryaon each face.
in the form of ribbed,
appeartwo layersof phdmsand-roofing
Behindtheseuppersithbakaryas
straight-edgedawning. Here the pent-roof has almost disappeared,replacedby storeys not so
anda muchbrokengha.ntda-bell.
muchof roofsas of heavyeaves.Aboveis a skandha-platform
Unusual in this shrineis the use of khandasto supportphansanad-layers,a unique adaptation
of the Nagaraformulato a Phramsana roof.Also of considerable
significanceis the use of ribbed
awningsratherthanthe straight-edged pent-roofsor the curvedkapotalimouldingsof other
Phamsan structures.This substitutionis importantfor the latertransformation of Phamrhsan
into Samhvarana form.
In this respectmentionshouldbe madeof the PhamsanIroof overthegMdha-ma.dapa of the
the Maladetempleat Gyaraspur in CentralIndia(c. 875A.D.)29Thishighhallfrontsa navdandaka
spire.It uses a complexof manyunits to build up height, giving againa sense of multiple
awningsratherthan of the greatpent-roofsfound on the Osianii Mahaviratemple.Pent-roof
unitsstillareindicated,however.Closeexamination showsthelayeringof thisPhamrsana roofto
consistof straight-edged, ribbed,pent-roofswith candrasdli decorationalternatingwith ribbed
khuracchddya awnings.The sequenceis: straight-edged, ribbedphadtsana, a narrowneck deco-
ratedwith saw-toothdecoration,khuracchddya-awning, narrowneckwith square-and-diamond
(possiblyvedika)pattern,then a secondstraight-edged pthasanad.
The boxy Phamhsani roof over the Laksmana templeat Khajuraho(c. 954 A.D.) returns
almostentirelyto the bhRmi pradsdaformula,built up of "storeys"of checker-grille separated
fromthe next level by shallowkhuracchddya awnings.
The southwestdevakulikd to the Saciyamatatempleat Osiani(c. late Ioth century)also uses
phamsand-layers to roof its entry-space.Set betweencrosseddormersfaced by simbhakaryas
(fig.31),thesephamsana layersareribbedawningsratherthanpent-roofunits.Abovetwo layers
28 ASIWC 1909-IO, pp.49-50. WC neg. no. 3496. - Addenda:The small shrine in Fig. 29 from Chatsf is now kept in the
State Museum at Amber where it has the number Ab - (Ab erroneously indicating Abaneri from which many of the
sculptures in the Amber Museum have come).
29 ASIIC 1913-I4, pp.62-63. Krishna Deva, "Mala Devi Temple at Gyaraspur", Shri MahavirJaina Vidyalaya
Golden
The most accessible and also the most satisfactory published photograph of
JubileeVolume,Bombay I968, pp. 260-269.
this man.dapa IndischerKunst,Koln I 959, fig. 256.
roof is in Klaus Fischer, Schopfungen

I86
of awnings, and beneath a kapotadli-layer
faced with candralaids
which acts as skandha,a single
layer of un-ribbed khuracchadya
(or kapotadi)appears. On the skandba-platform rest a shallow,
dmalasdrikd,
widelyflaringgha.n.td, thenkalasawithcrowningcitron.Thecrosseddormers
candrikd,
of this superstructure(fig. 27) can support my point concerningthe small Satyanarayana temple
(fig.25) also on the Sacikadevi hill. These same crossed dormers act as frame for the later
Samvarana
form.
Late in the Ioth or earlyin the I Ith centurya new, complex form of roofing for the manrdapa
was developed. Known as Samvarana,it allowed an enlargedhall while maintaininga low pro-
file for the superstructure(fig. 31). It does so by using offset awnings, one layer much indrawn
from theothe r, the vacantspacefilledby minorghantds.Becauseof the greaterdistancethat
each roof-layeris drawn in, allowing for the ghantdsalong the edges of the lower layer, fewer
levels are requiredto form a peak. This allowed a largerhall while retaininga shallow elevation
for the superstructure, preservingthe integrityand dominanceof the Nagaratower behind.
pre,
Simhakarvas appearover the bhadras,
andlow miniatureSamvarana roofs act as aediculae
on the
corners.This is an extremelyelegantadvance.Built on the decliningbody of the pent-roof
Phamsanait borrowsthe ribbedawningandghasdtagiving new life to an ancientconceptbut
destroyingits progenitor.The Sekharisikharagrows out of the Latinaand anekdndaka forms
of earliercenturies;so also the Samvarana formgrows out of the Phamsana,fullyreplacingit.
I illustrateit herefromthenorthwestandnortheastdevakulikds to the Saciyamata temple(figs.3I
and32) whereit standsoppositeits earlierancestor,spreadingits newlyopenedwingsin the sun.
The resultsof my surveyof Phamsanatemplesin WesternIndia,beyondestablishingthe
natureof the Hariharatemplesuperstructure,
canbe summarized
as follows:
I) In WesternIndiaPhramsana
is usedin placeof the seeminglymoreappropriate
Valabhibarrel-
vault to roof rectangularshrines.
2) The original Phamsana formula seems that of a straight-edged pent roof. Roll-cornice
werelateradaptedfromthe cornicelayeringof otherand earlierstructuresto
phadmsand-layers
the low profileof the pent roof.
3) Roll-cornice phbdsand-layers become confused with, and at times replaced by, ribbed awning
members (chdyas), a substitution signifcant for the later development of the
complex Sam-
varanaroof.
4) The pent-roof Phamsanahall seems by an earlydate to have developed a complex form, with
sinhakarvadormers over porch-projectionsand corner aediculae (khtas). Such complexity lies
at least partiallyat the root of the anekandaka Nagaraform.
5) Three elementsof the Phamsanatradition-this crossed-siznhakarya plan,the originalcrowning
ghba;nt,and ribbed awnings-are carriedover into the formulationof the Sahvaranasuperstructure
which, with its even lower profile,replacesPhamsana above maydapa-halls earlyin the I Ith century.

GLOSSARY OF ARCHITECTURAL TERMS


amalaka: 'cogged-wheel' stone crowning the north anekdndaka:multi-spired (with more than one anda
Indian temple spire. or "'egg").
jmalasirika: secondaryamalakaamong the crowning antarala:vestibule.
membersof the spire, bhadra:central offset on the temple wall.

I87
bhfimi:level or storey. man.dapa: hall, often a pillaredhall.
bhntmi-amalaka: ribbed stone used to mark levels in mukha-catuski: front entry bounded by four pillars.
the outer band (ve.nukosa) of the temple tower. mukha-ma.n.dapa: fronting portico or entry-porch.
bhumikha.nda:section of the outer band of the temple the
mula-prasdda: main body of the temple which
tower marked into a 'storey' by the insertion of a houses the sanctum.
ribbed stone marker. Nagara: the north Indian temple-type (defined by
bhtmiprdsdda:temple-typehaving only cornice layers morphologicaldetail, not tower-type, though prin-
or miniaturepillared storeys (cornice layers with a cipally Latina).
shallow necking showing tiny pillars)used to make navn.ndaka: having nine spires (a centralspire and two
a tower. of
layers four sub-spires).
byiapftraka:citron finial. pankti: platform or level which supports the corner
caitya:stupa; the stupa-hallor assemblyhall of ancient sub-spires (srngas)of an anekdn.daka tower.
Indian architecture;a decorative motif resembling Phiamsana:low, pyramidaltower-type.
the sun-window of such a hall: gavaksa,ku.du,or phathsand: 'wedge'-shaped roof-unit of pyramidal
more properlycandrasald. tower.
candrikd:cap-likememberabove the upper dmalakaof phdamsand-kfta: miniature-shrineroof-unit having low
the tower. pyramidal roof.
chadya:awning. prsadda: literally "palace"; used in north India to
devakulikd:smaller sub-shrine. refer to a temple structure.
gavadksa:curved window-form: candraJiid. pratiratha: the wall-offset between the karna and
gha.n.d:"bell'-a fluted crowning member (used on bhadra.
Phiramsanastructures in place of a crowning Samivarana:complex offset pyramidal roofing for a
admalaka). front hall, which uses chddyaandghanadelements.
ghafa-pallava:'vase-and-foliage'pillar-type. Sekhari: clustered (multi-spired)Nagara tower-type.
gtdha-mandapa: closed hall. Jikhara:temple tower.
hara: enclosure, balustradeor cloister. simhakarna:complex caitya(candrasala)-dormer pedi-
kaksdsana:slanting seat-back. ment.
kalaSa:"pitcher"--acrowning memberof the Jikhara; skandha:shoulder course-the uppermost flat stone
also a half-roundmoulding. level on which the crowning membersrest.
kalika: bud decoration. srnga:miniature Latina spire used as one element of
kapili: buffer wall. a more complicatedtower.
kapota: quarter-roundmoulding. Jukanasa:the 'nose' projectionfrom the likharawhich
kapotali:inverted cymarectamoulding (varient: kapo- roofs the entry vestibule.
tapalf). suirpa:'winnowing-basket'shaped element.
kar,na:corner face of the temple. tilaka: squareminiatureshrine-elementin a clustered
karnan.daka: corner dmalaka. sikhara.
karna-kuta:square miniaturecorner shrine. udgama:trifoil niche-pediment.
khanda:segment of the venukosa. Valabhi: barrel-vault or keel-roofed superstructure-
khuracchadya: curved hood or awning; awning-like type.
moulding. vara.n.dik:complex cornice which acts as transition
kumbha:"pot"-a base moulding with perpendicular between wall and tower.
face and quarterround shoulder. vedibandha: base mouldings of the temple-wall.
ktfa: a miniatureshrine, squarein plan. vedikd:a railing or balustrade; a decorative pattern
kU.taka:sub-miniaturekgta elementused in a complex showing a series of miniaturepillars.
Sekharitower. venukosa:the outer, corner, or binding vertical bands
Latina: single-spired tower-type made up of vertical of the Latina tower.
bands.

Matching terms in texts to actual elements of existing temples in recent years has largely been the
work of M. A. Dhaky, on whose work I am here pleased frequentlyto depend.

I88

You might also like