Tradition and Transformation: Continuity and Ingenuity in the Temples of Karnataka

Author(s): Adam Hardy

Source: Journal of the Society of Architectural Historians, Vol. 60, No. 2 (Jun., 2001), pp. 180-
199
Published by: University of California Press on behalf of the Society of Architectural
Historians
Stable URL: http://www.jstor.org/stable/991703
Accessed: 02-12-2015 20:28 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

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 [email protected].

Society of Architectural Historians and University of California Press are collaborating with JSTOR to digitize, preserve
and extend access to Journal of the Society of Architectural Historians.

http://www.jstor.org

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
Tradition and Transformation
Continuity and Ingenuity in the Temples of Karnataka

ADAM HARDY
De Montfort University, Leicester

he architecture of Indian temples-the monu- municated,and, second, by its explanatorypower-its abil-

mental masonrystructureserected from around ity, for manyexamples,to describeand explainarchitectural
the sixth century-is based on complex patterns compositionsas a whole.
of form involvingvariousregional traditions.This is espe- The temple forms evolve, metamorphosinggradually.
ciallytrue of the temple shrineitself, the vimana(for south- I shall argue that they evolve in a particularway that is
ern temples) or the mulaprasada(for northern temples), linked closely to the natureof the individualcompositions.
consisting of the sanctum and a towering superstructure. Both the compositionsand the ways in which they develop
Relativelyindependent of "functional"developments,this are closely analogous to recurrentpatternsin Indian phi-
part of the temple was treated as the sculpturaland sym- losophy,religion,andmythology,to the extentthat the tem-
bolic focus, befitting an image of the cosmos and the ples can be saidto embodyor concretizemetaphysicalideas.
embodimentof the divine. Its formalpatternswere not the But let us leave aside, for the moment, the question of
result of fixed types but were created through the combi- meaning. In order to understandthe patternsof develop-
nation of a numberof compositionalelements,in a way that ment through time, visualizationis again a prerequisite.It
allowed numerouspermutations. is necessary to take a broad enough view, a long enough
Traditional treatises (s'stras), besides conveying the time span,to see the overallshapeof a traditionbeyond the
mythologicalcontext of temple buildingand specifyingrit- lulls and vicissitudes.Takingall the availableexamples,the
uals, contain elaborateclassificationsof temples and exten- existence of evolutionarytendencies can be substantiated
sive naming of parts. However (perhapsbecause no early through systematicformalanalysis.2
illustrationssurvive, or because the texts were written by The intention in this articleis to examineone temple-
priestsand not architects),these texts have not, so far,been buildingtraditionin some detailto illustratethe patternsof
veryhelpfulfor modernscholarsseekingto understandhow form and transformation,while focusing on the relation-
temples were conceived or designed.1 This has to be ship between a characteristicmode of evolution and indi-
deducedfrom the monumentsthemselves.The main chal- vidual artisticchoice.
lenge is to see the architecturein a way that is true to its
conception, and this involvesrecognizingthe principalele-
ments and understandingthe relationships among them. "Karnata Dravida" versus "Vesara"
The validity of a way of seeing can be tested, first, by its The tradition to be examined can be called the "Karnata
abilityto be visualizedand for this visualizationto be com- Dravida" tradition, one of the two main branches of

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
Dravida, or south Indian, temple architecture. Indian medievalDeccani temple type is what is meant in contem-
(Hindu, and alsoJain)temple architecturefallsbroadlyinto porary architecturaltreatises, at least those of northern
two categories, the Nagara and the Dravida, associated India,by the term "Vesara." A numberof texts give a three-
respectively,but not exclusively,with northern and south- fold classificationof temple architecture:Nagara, Dravida,
ern India. andVesara.Since, etymologically,"Vesara" impliesa "mule"
Drawing from a common artistic heritage, including or hybrid, it has been tempting to identify the term with
the monumental architecture of the Buddhists, the two this kind of temple, situatedgeographicallybetween north
strands were becoming distinct by the seventh century. andsouth and formallysharingsomethingof Nagaraas well
TamilNadu, in the farsouth, is often thought of as the true as Dravida.
home of Dravida, or southern, temple architecture-and The Deccan, Karnatakain particular,is the part of
indeed is where it has lasted longest. Nevertheless, it is in India where both Nagara and Dravidatemples were built,
Karnataka,in the southwest Deccan around Badami,that sometimes side by side. Moreover, at certain times inge-
the formativestagesof this architecturallanguagecan most nious mixtures of the two languages were created.5This
clearly be traced, among the seventh- and eighth-century kind of hybridityis differentfrom the normal late Karnata
monumentserected under the EarlyCalukyadynasty. Dravida or "Vesara,"with its various Nagara-seeming
While the scale and extent of temple building varied aspects. Nagara temples and experimentsin hybridization
according to political circumstances, a gradual but dra- were common in the Early Calukya period. Under the
matic transformationof architecturalform continued for Rastrakitas only the Dravida is found, but from the
six centuries.3 The architectural language was fully eleventh century on the latest Nagara forms are known
mature in the magnificent Early Calukyatemples at Pat- again. Full-sized Nagara temples were built, and many
tadakal.Activity here was interrupted by the invasion, in more Nagara compositionswere tried out in the miniature
757, of the Rastrakutas, whose architectural legacy is "shrine models" that formed niches in Karnata Dravida
chiefly at Ellora (Maharashtra), where India's largest temple walls. In these miniaturetemples, as well as in the
monolithic temple, the Kailasa(late eighth century), can full-scaleworks,are found not only the simple Latinaform
in broad terms be considered as part of the Karnata of Nagara,with its curvedsikhara(spire),but also the newly
Dravida tradition. But the tradition is carried forwardin developed Sekhariand Bhumijamodes, the formerwith its
far more modest shrines, built under Rastrakita hege- interpenetratingclustersof ikhara forms,the latterwith its
mony in and around the former Calukya heartland. A verticalchains of kuta-stambhas (sikharason pillars).At this
burst of activity gathered momentum after the Later time, games of hybridityare played once again,with great
Calukyas of Kalyani came to power in 973. Though the sophistication.
temples were much smaller than the last Early Calukya The issue of whether "Vesara"is an appropriateterm
monuments (see, for example, Figures 10, 18), the tem- for the typical eleventh-centurytemple form of Karnataka
ple-building activity of the Later Calukyas and their is inseparablefrom the question of how this form relatesto
feudatories was prolific and spread over a much wider the earlierDravidatemples. In a recent study,AjayJ. Sinha
area. The final creations of the tradition, a few as late as has argued that a "conceptualshift" occurred during the
the fourteenth century,were under the Hoysalas in south- eleventh century,indicating"a deliberatesearch for a new
ern Karnatakaand in Telangana (northwestern Andhra organizingprinciple,giving temples a consciouslymodern
Pradesh) under the Kakatiyas. form."6Sinhastressesthat "Vesara"temples arenot merely
By the earlyeleventhcentury,the architectureof a typ- derivativehybridsof Nagara and Dravida,and that, while
ical temple of the region had become transformedto the based on the Dravida, they "cannot be fully predicted by
extent that, without being able to trace the intervening previousmonuments."7We shall returnto these questions:
steps, it is not immediately recognizable as Dravida. It is whether there was a "conceptualshift," and whether the
quite distinctfromthe Dravidaof the EarlyCalukyaperiod, later temples can in any sense be "predicted"by the earlier
and from the more conservativeTamil Dravida, in which ones.
architecturaldetail, though not temple scale and planning, For a proper understanding of the late Karnita
would remain relatively unchanged for centuries.4While Dravida,or "Vesara,"forms, it will be necessaryto explain
every aspect of the "new"Karnataform evolved from the briefly the developments of the Early Calukya and
earlier Dravidaof the region, it has certain characteristics Rastrakita periods, before analyzing a number of Later
reminiscentof Nagaratemples. For this reason,it has been Calukyatemples. But firstit is necessaryto summarizecer-
widely held among scholarsin the last sixty years that this tain principlesthat can be deducedfrom the monuments.

TEMPLES OF KARNATAKA 181

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
Composition, Evolution, and Meaning
Figure 1 Generalized
The understandingof Indian temple architectureoutlined eaji patternof evolutioninthe
here holds true, I believe, for both Nagara and Dravida KarnataDravida tradition
' .
temple architecture.8Four related principles are summa-
rized,which may be acceptedor rejectedsingly or together:
(1) aedicularcomposition, (2) centrifugalgrowth, (3) cycli-
cal evolution, and (4) cosmic manifestation.

1. AEDICULAR COMPOSITION. The whole temple

form is made up of many small templeforms of various kinds,
conceivedthree-dimensionallyand as if embeddedor interpene-
trating. When a temple is seen in this way, its whole composi-
tional structure can be understoodat once.

2. CENTRIFUGAL GROWTH. The interrelationships

of the small templeforms expressa processof centrifugal emer- C3
gence, expansion,and proliferation,and simultaneouslya process AL
of dissolutionand reabsorption.This process appears to orig-
inate at the finial, or at an infinitesimalpoint above its tip,
continuing downwardand outwardfrom the vertical axis
of the shrine, radiating all around, but especially in the
four cardinal directions. As the forms evolve downward,
the summit recedes and the whole monument grows.9
Movement is explicitly and illusionistically portrayed
through a number of mutually reinforcing architectural
means: projection(pushing forward of embedded forms),
(nZad3
staggering(multiple projection through offsets), splitting,
bursting of boundaries, progressive multiplication (similar
forms increasing in number), expandingrepetition(similar
forms increasingin size), andpictorialrepresentation
(sculp-
tural depiction of movement).
iir=
,^^CT^^H^ 8
3. CYCLICALEVOLUTION. This pattern of growth,
expressedin a single temple, isfound in the evolutionofarchitec-
turalforms through the courseof a tradition. Here the pattern
alsogrowsfrom unity to multiplicity,simultaneouslytending back
to unity througha processof dissolutionandfusion (Figure 1). In
this way it is cyclical. Cycles can occur at different times
and at differentratesin differentpartsof temples, as well as
in differentregions and traditions.The overallprocess can
be understood in terms of a number of evolutionaryten- 4. COSMIC MANIFESTATION. This dynamicpattern
dencies, which can be observed systematicallyand rigor- makesconcrete the ideaof a worldof multiplicityevermanifest-
ously if enough examplesare studied over a wide enough ingfrom unityandeverdissolvingbackintounity.Through its
time span.Briefly,these tendenciesare:aedicularity (increas- dynamic structure, a vimana or a mulaprasada becomes a
ing), aediculardensity(shrineimages getting closer together symbol of manifestation:10 of the enshrinedgod and its var-
with increasing interpenetration), proliferationandfragmen- ious aspects (inherent in the aedicularimagery,'1given a
tation,centralemphasis(along cardinalaxes, in parts as well temporal dimension by the sense of movement) and, in
as whole), movement(increasing sense of), staggering(in anothersense, of the universe.Manifestationis to be under-
parts and whole), continuity/alignment
(horizontally,verti- stood not as a single event but a continuous process. Such
cally, radially), abstraction,assimilation. clear analogiescan be drawnbetween the patternsof form
182 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
and metaphysicalideas about manifestation, recurrent in Nadu.'3 But harasintended to be read separatelyfrom the
Indianthought, that these ideas can be saidto be embodied walls below were superseded by a more fruitful concept,
in the forms. implicit already among the earliest Dravida shrines,
whereby the pavilions of the parapet are placed above
pilaster-boundedprojectionsin the wall to form a two-story,
Formation of the Dravid a "Language," and the
three-dimensionallyconceivedshrineimage or aedicule,of
Early Calukya Period which the kuta,sala, or other pavilion form is the super-
The imageryof both Nagaraand Dravidatemples-on one structure.
level representingmultistorypalacesof the gods-is of tim- A "kzta-aedicule"and a "sil/-aedicule"are illustrated
ber-and-thatchconstruction.Even the earliesttemples are in Figures 2 and 3. They show the full sequence of super-
not copies of wooden buildings,the imageryof timbercon- imposed moldings that has become established(for parapet
structionhavingbeen translatedinto a symboliclanguageof and plinth) in the later temples of the Early Calukyas(see
masonry(stone or brick).12 Buddhistarchitecturepredating Figure 2). The parapetmoldings representa thatchedroof
any surviving Hindu temples depicts, in variousways, the (sikhara-a dome for a kuta, a barrel roof for a said), the
kind of storied palatialarchitecturethat the temples evoke. "neck"(griv,), or inhabitablespace below the roof, a sur-
Relief carvingssuch as those on the gatewaysof Bharhut rounding railing (vedi),and the edge of a floor, with joist
and Sanchi stupas (second century B.C. to first century A.D.) ends (vyalamatld,where decoratedwith mythicalcreatures).
show tall mansions and city gateways with barrel roofs, This last molding sits on the roll cornice (kapota)that
horseshoe gables, half-round canopies of thatch, and bal- depicts thatchedeaves.
conieswith railingsandbracket-bearingposts.These ingre- The success of the aedicular concept was certainly
dients are still present in the palace-like facades of the helped by the fact that the images depicted by the original
fifth-centuryrock-cutviharas(monasteries)and caityahalls
(prayerhalls)at Ajanta,becoming rhythmicand formalized
to framethe now numerousBuddhaimages.Variousproto-
Dravida elements are found here: the basic form of the
Dravidapilaster-a relief representationof a wooden post
(thus differentin origin from the traditionof rock-cut pil-
lars),and the kapotamolding-a half-roundcornice repre-
senting thatchedeaves,with horseshoedormers.Two types
of stylized pavilion essential to Dravida imagery appear
here: the silat,with a horseshoe-gabledbarrelroof, and the
kuta,with a square domical roof. The facades of Caves 1
and 19 are crowned by kapdtascarrying haras(chains) of
salas,in the formercase terminatedat one end by a form of
kuta.
A typicalDravidavimana,the shrine properof a south
Indian temple, and the partfocused on in this article,con-
sists of garbha-grha,or "womb chamber"(a dark cuboid
sanctum, housing the main image), surrounded by thick
wallsthat, togetherwith theirmolded plinth and their para-
pet, form the first tier (tala)of a stepped pyramid.One or
more furthertiers (the interiornormallyinaccessible,often
solid) rise to a solid "dome"and a pot finial.The parapetof
each tier comprisesa haraof pavilionimages, squarekutas
at the corners,and, normally,silas in between.
The art historians' way of seeing and explaining
Dravida temples has been to look at the horizontal divi-
sions, describingthe plinth, the pilasteredwall zones, and
the stringsof parapetpavilions.This interpretationreflects Figure 2 Kuta-aedicule(K),showing moldings, Sahgamesvara temple,
a concept that certainlypersistedas one alternativein Tamil Pattadakal,early eighth century

TEMPLES OF KARNATAKA 183

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 Figure 3 Sa/5-aedicule(S),
Sahgamesvara temple, Pattadakal,
early eighth century
Figure 4 Buddhashrine from
Gandhara(BritishMuseum), an early
form of alpa vimanaor kuta-aedicule

aedicule types were ancient, recognizable types of shrine. or without attachedhall (mandapa)or porch. The molding
The kzta-aediculegoes backat leastas faras the secondcen- sequencein the superstructureis that of the fully developed
turyA.D., to BuddhistGandhara,hundredsof miles north of kuta-aedicule.Because of its broaderproportions, an alpa
the Dravidaregions. Gandharanrelief carvingsshow kuta- vimanagenerallyhas a pairof intermediatepilasterson each
aedicules sheltering the Buddha, with Corinthianesque side (except for the entrance elevation),in addition to the
posts and leaf-thatchedkapotaand circularkuta(Figure 4). corner pilastersof the more slender kuta-aedicule(Figure
Beforethe earliestsurvivingstructuraltemples,these forms 5). Alpa vimanascan also be rectangular,with a il/a super-
are found in Early Calukyacave temples of the late sixth structure,or rectangularwith an apsidalrear,entered lon-
century:kzta-aediculessheltering fat dwarfseither side of gitudinallyand crownedby an apsidalsala/.'5
the Ravalaphadicave,Aihole, and a chain of kuta-and sald- The seventh-century Mahakftesvara, Mahakuta,
aediculesover the sanctumdoorwayto Cave 3, Badami. belongsto a stagebeforethe aedicularprinciplehad become
Furthermore, the small wooden shrines depicted in all-pervasive.16Thus, its first tala (surroundingthe circum-
reliefs have full-scale masonry equivalents. This, too, is ambulatory path around the sanctum) has a projecting
alreadythe case in ancient Gandhara:a squareshrine,with aedicule only at the center, containing the sculpted deity.
a kapotacornice supporting a circular kuta, is one of two This projectionforms a s/da-aedicule,a very early example
shrine types linked to form a cloister aroundthe "courtof of the staggeredtype, with four pilasters,that alreadylooks
the stupa"at Takht-i-Bahimonastery.14 This, in essence, is forwardto the double-staggeredkind characteristicof the
the form that becomes the alpavimana(minorvimana),the later KarnataDravida(Figures 6, 7). The remainingpara-
simplestvimanaform in Dravidaarchitecture.Just in front pet pavilions, including the corner kutas,have no corre-
of the Ravalaphadiat Aihole is the earliestalpa vimanain sponding projectionsor paired pilastersin the wall below,
Karnataka,probablyof the earlyseventhcentury.The con- and thus can be read only as part of the horizontalhara.
cept of an alpavimanais of a large kutasitting on a kapata The comprehensive use of aediculararticulationhas
over the squaresanctuary,the latter open on one side, with been achieved in the design of the Bhutanatha,Badami

184 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 Figure 5 Vimanaof the
Bhutanathatemple, Badami,
showing composition based on
the idea of embedded aedicules:
(a) alpa vimana,(b) kuta-aedicule,
(c) sa/a-aedicule

Figure 6 Staggered sa/a-aedicule(S1), Navalihgatemple, Kukkanur, Figure 7 Double-staggered sa/a-aedicule(S2), Yellamagudi,Badami,

tenth century early twelfth century

TEMPLES OF KARNATAKA 185

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
Figure 8 Bhutanathatemple, Badami,c. 730 temple, Pattadakal,c. 742
Figure 9 Mallikarjuna

(Figure 8; see Figure 5). The vimanahas three talas, the found at the center of kutasandsalasand in the intervening
lower two with kuta-aediculesat the corners,sia/-aedicules stripsof "cloister"in a hara.
at the center: K-S-K. The top tala, with its big crowning Over the sanctuaryrises a three-tiersuperstructure.In
kuta,is in the form of an alpavimana.This concept, typical the Virupaksathe composition is essentially that of the
among earlierKarnataDravidatemplesandneverabsentin whole vimanaof the Bhutanatha(see Figures 8 and 10). In
TamilDravidatraditions,createsan "uppertemple,"a tem- the Mallikarjunathe alpa vimana forming the top tala is
ple form at the summitof the vimana,justas the alpavimana replacedby wallswith three projectionssupportinga circu-
form itself has a kutahut as its superstucture. lardome. The arrangementof the moldingsover the kap6ta
Early Calukya developments culminate in the great indicatesthat these were intendedto be the lowermoldings
Mallikarjunaand Virupaksatemples at Pattadakal,each of kutasandsilas,carryingthe K-S-K aedicularschemeright
builtby a queenofVikramadityaII aroundA.D.742 (Figures up to the dome. The multiaediculartop tala, as opposedto
9-11). Though the lesser twin, it is the Mallikarjunathat the unitaryalpa vimana,is one extension of the aedicular
takesevolutionarylogic a step farther.In the first talawalls scheme thatwould graduallytakehold, becominguniversal
of both temples, which surroundan ambulatorypassage,a by the end of the eleventh century.
five-projectionpatternis introduced.Between the central Comparedwith earliertemples, here we have a much
sal-aedicule andcornerkita-aediculeareintermediatepro- denserconcentrationof aedicules,not only becausethe pro-
jectionsin the formofpanjara-aedicules, the thirdbasictype jections are much more closely spaced,but also due to an
of Dravidaaedicule(Figure 12), here used for the firsttime extensionof the aedicularlanguagethroughlower levels of
as a primary component: K-P-S-P-K. The panjara, the compositionalorder.A greatincreasein iconic sculpturein
horseshoe-arch-toppedcrowningpavilionof this aedicule, the walls,where almosteveryprojectionandrecesscontains
is a frontalview of the gableend of a sila. Minorversionsare an image,is accompaniedby a proliferationof architectural

186 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 K-S-K
K-S-K L I I I I I
0 5A
K-S-K
K-P-S-P-K

Figure 10 Virupaksatemple, Pattadakal Figure 11 Mallikarjunatemple, Pattadakal,showing probableoriginal

intentionfor top tala

niche surrounds,mainly monster-flankedarchwaymotifs of wall push forwardin stages, at the same time tending to
(makara-t6ranas) and secondarypanjara-aedicules.Exam- dissolve the individual elements by breaking down clear
ples of the latter,housing deities, are containedwithin the boundariesand creatinghorizontalrhythmsand ripples.
primarypanjara-aedicules(see Figure 12). At a yet smaller One "prophetic"occurrencecan be mentionedhere as
scale areminiatureku.ta-aedicules (i.e., alpavimanas)within a further illustration of how the "Vesara"represents the
the horseshoe arches of both the primaryand secondary development of long-held aspirations and possibilities
panjaras. inherent in the architecturallanguage.17The sila, forming
The tendency toward increasing central emphasis is an "overdoor"inside the early-eighth-centuryMallikarjuna,
apparentin the main differencebetween the vimanasof the Aihole (a predominantlyNagara temple), has a second pair
Virupaksaandthe Mallikarjuna.In the latterthe centralele- of gables emerging laterallybut without the majorpahjara
ments, especially in the first story, leap dramaticallyfor- eventuallyfound at the center(Figure13). Beforethe devel-
ward,stronglyreinforcingthe sense of centrifugalswell. In oped type is arrived at, this form appears again in the
this outwardsurge, partslose their individuality,the whole Rastrakutaperiod, crowning a full, primaryaediculeat the
impressionbeing of the kind of dynamicfusion character- Siddhesvaratemple, Mudhol.
istic of later monuments in the region. This quality is
enhanced by the "staggering" of corner and central
aedicules,where offsets are introducedup to and including The Rastrakuta Period
the roll cornice (kapota),with correspondingpilasters.The The evolution that culminated,under the Early Calukyas,
same treatment is used in the Virupaksa,but more insis- in the Mallikarjuna,Pattadakal,continued in the Karnata
tently in the Mallikarjuna.The effect is to make portions heartland during the two centuries of Rastrakita rule. A

TEMPLES OF KARNATAKA 187

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 staggeredform of sala-aediculebecame widespreadat this
stage (see Figure 6). Already foreseen, for example, at
Mahakutesvaraand in the third tala of Mallikarjuna,this
type projectsfrom the wall in two stages,the offsets(defined
in the wall section by pilasters)runningup into the crown-
ing s'la. With the panjaraat the center of a isad made into a
full-scale, primaryparapet-pavilion,a full panjara-aedicule
now sits at the center of the sia/-aedicule.Here we havethe
germ of an idea-of primaryaediculesnot merely embed-
ded and emerging from the wall, but interpenetratingwith
and unfolding from one another-that reached its height
of expressionin the eleventh century.
A comparabledevelopment takes place in the plan of
the vimanaas a whole. Generallyit is still usual for all the
components of the wall to project equally,but the central
element begins to project farther than the others, and in
five-projection schemes a progressive stepping forward
from corner to center eventuallyis established.This plan
had been achieved in the eighth-century Kasivisvesvara,
Pattadakal,the culminationof Early Calukyadevelopment
of the Nagara, and at the Kailasa,Ellora.The first Dravida
vimanain Karnatakawith fully staggeredwallsis that of the
Jain temple at Pattadakal,built perhapsat the turn of the
tenth century(Figures 14,15).
The PattadakalJain temple will serve to illustratevari-
ous developmentsat this stage. It is the last temple in the
region to have an ambulatorypassagearoundthe sanctum.
The outerwallsare lost, but the plinthremains,fromwhich
the five-projectionvimanawall scheme can be deducedby
comparison with the aedicular components of the hall
(mandapa). Recallingthe neighboringeighth-centurytemples,
thiswouldhavehadstaggeredsa/a-aedicules at the centerand
panjara-aedicules betweentheseandthe cornerkuta-aedicules
(K-P-S1-P-K).The projectionsswell forwardfrom the cor-
ners to the centeralong eachcardinalaxis.Overthe sanctum
risesa two-storysuperstructure containingan uppersanctum,
somethingnot unusualinJain temples.The superstructure is
an advancedversionof a two-talavimana compositioncom-
mon at thistime,found,for example,in most of the nine small
vimanasof the Navalingatemple,Kukkanur(Figure16).This
is K-S1-Kin the firsttala,surmountedby analpavimanaform.
Generally,at this stage, the first step of the staggerediald-
aediculeadvancesno fartherforwardthan the corner kzuta-
aedicules,but in the Pattadakal superstructureit comesfarther
out, staggeringthe three-projectionwall as a whole. In alpa
Figure 12 Panjara-aedicule(P), containinga lesser parjara-aediculeas vimanas,which originallyhad flat walls, the centralbay has
a wall shrine, Mallikarjuna
temple, Pattadakal,c. 740 increasinglybeen projected,at first simply,and here with a
stagger,so that the wall as a whole becomes "double-stag-
Figure 13 Staggered s/la from overdoor at Mallikarjuna
temple, gered";moreover,these offsets are now carriedup into the
Aihole dome.
188 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 l I I II
0 5M

K-S1-K
K-P-S1-P-K

Figure 14 Jain temple, Pattadakal,c. late ninthto earlytenth century

The wall zone of the crowning alpa vimana can be

describedas "compressed,"but the pilastermoldingsare of
the normalproportions,and it is more appropriateto see it
as embeddedin the talabelow,buriedup to its chest. Thus,
IL2:l
LrL
in the pattern of downward and outward expansion, a
sequentialdetelescopingof the talasaccompaniesthe pro-
gressivesurgingout of theirwalls,each stage more revealed
and bulging out fartherthan the one behind.
A strikingaspect of temples in northernKarnatakain K-S1-K
this period,when comparedwith earliermonumentsor with
contemporaryworksof otherregions,is the generalabsence
of sculpturalimagesin the walls.If this was due initiallyto a
drainingof resourcesfrom the region, it correspondedto a
new preoccupationwith niche surrounds,or "wallshrines,"
as architecturalelements,symbolizingdivinepresencewith-
out necessarilyenshriningimages of gods. In addition to
archwaymotifs (makara-toranas),wall shrines mirror,at a
smallerscale,the kuta-,sial/-,andpahjara-aedicules thatform
the primarycomponents of the temple exterior.This has Figure 15 Jain temple, Pattadakal
alreadybeen seen in EarlyCalukyatemples,butnow the wall
shrines are more consistently assimilatedto the primary Figure 16 Navalihgatemple, Kukkanur
aedicules,in detailandplacement,so as to form a secondary
register in the composition. It can only be the difference curves,and the imageryof timber and thatch begins to be
betweenthe proportionsof the secondaryaediculesandthose lost in rhythmiclayersof light andshade.The shapesof the
of the principalcomponentsthathaspreventedscholarsfrom blocksfrom which forms are carvedare more conspicuous,
noticingthat they arethe sameformsand fromhere extrap- often left in their unelaboratedstate.The growingabstrac-
olatingthe whole aedicularconcept. tion is enhancedby an atrophyof the veranda-likerecesses
During the Rastrakiuta period, the Dravidamoldings of parapetand molded base (grivaandgala),which eventu-
graduallyevolve toward the shapes characteristicof the ally are treated not as separate courses but as mere slots
"Vesara"stage. The process is one of abstraction,accom- incorporatedinto the molding above. The resultinghori-
panied by enhancedcontinuity,horizontaland vertical,all zontality(seen also in the "compression"of wall zones and
contributingto the increasingsense of fusion.The plastic- in the leveling of parapetpavilionroofs) is complemented
ity andlushnessof the earliertemples graduallygive way to by a new vertical continuity created by aligning certain
precisionand sinuousness.Roundedshapesbecome double detailsone above the other.

TEMPLES OF KARNATAKA 189

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
The Later CfilukyaPeriod
The temples discussedhere belong to what may be called
the Later Calukya"mainstream,"both because of stylistic
continuities (in terms of peculiaritiesof detail) with work-
shops active in the Rastrakutaperiod, and because of the
sustaineddevelopmentthey show.Afteran earlyphase,last-
ing between 973 and about 1000, two mainstreamschools
can be identified: the Sudi school and the more prolific
Lakkundischool, both active in the region to the south of
the Early Calukyaheartland.Early in the twelfth century,
the mainstreambegan to be absorbedbeyond recognition
into the heterogeneouscurrentsthatwere springingup over
a much wider region.
Toward the end of the tenth century, corresponding
more or less to the beginning of Later Calukyarule, the
staggeredsd/a-aediculethathad developedin the Rastrakuta
period was further transformedinto a double-staggered
type (Figure 17), the first and most consequentialof many
ingeniouscomposite,interpenetratingaediculeforms.Con-
ceptually,it is a configurationof five aedicules-a panjara-
aedicule(thatis, an end-on sal-aedicule) at the center,with
two sala-aediculesemerging one from the other on either
side. Always positioned centrally,this element is essential
to the dense and dynamic complexity of late Karnata Figure 17 Dynamics of a double-staggered sa/a-aedicule
Dravida vimanas. The forward and sideways motion is
emphasized by the pictorially represented outpouring of
each gable-end nasi (horseshoe arch) from the mouth of a as the top tala is the simplest form of Dravidavimana,the
monster-face finial, whose snail-like buttress body rides top two together a more developed form, and all three
along on the sala roof ridge. If the double-staggeredsald- togethera yet more developedform,so the centralelements
aediculeis repeatedthrough a number of stories, the extra evolve downward from a simpler to a more complex
vertical lines of additional pilasters and gable sides (now aedicule. It is in such ways that the extraordinaryparallel
invariablyaligned)strengthenthe centralspine, which has comes aboutbetweenthe unfoldingof the traditionand the
been interpreted as a Nagara-like characteristic of the dynamicgrowth patternexpressedin a single vimdna.18
"Vesara"temple. The principle of the "unfurling"central projection,
The Naganatha (or Adakesvara)temple at Mahakuta "evolving"in its downwardexpansion,is given its ultimate
follows a compositiontypicalof the earlyphaseof the Later expressionin the Mallikarjunatemple, Sudi (Figures 18b,
Calukyaperiod (Figure 18a;see Figure 17). Though not a 19). With four talas,this is perhapsthe first vimanain the
spectaculartemple, its design demonstratesa subtle way of traditionwith five projectionsin the superstructureas well
thinking about the downwardunfolding of forms. There as the first story. The top tala is an alpa vimana, triple-
are three talas, the upper two essentially the same as the staggered. In the lower three talas the usual corner kuta-
superstructureof theJain temple,Pattadakal:staggeredalpa aedicules flank (in the now time-honored fashion)
vimanaon top, including the dome (one offset plus nasi), intermediatepanjara-aedicules,while the central element
then a staggeredsal-aedicule between kuta-aedicules(K-
(readingdownward)developsfrompanjara-aedicule to stag-
S1-K). In the first tala this expandsout to K-S2-K, with a gered is/ia-aedicule to double-staggered sa/la-aedicule.
double-staggeredsia/-aediculein the center.The nsi in the The west vimana of the nearby Joda Kalasa temple
dome, echoed andprogressivelybiggerin the two salaforms demonstrateswith clarity another permutationwithin the
below, can be seen as the origin of an unfurling growth established rules (Figures 18c, 20). Again there are four
sequence in the centralprojection,through which the sin- talas,now each with five projections, as this vimanapicks
gle-staggeredelement developsdownwardinto the double- up the growing,and eventuallyoverwhelming,trendtoward
staggered, emitting a furtherlateral pair of sadaends. Just abandonmentof the unitary,alpavimdnatop tala in favor
190 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
I i I Figure 18 a) Naganathatemple,
s
Mahakuta;(b) Mallikarjuna
temple, Sudi; (c)Joda Kalasa
temple, Sudi, west vimana;(d)
Kallesvaratemple, Kukkanur;(e)
Mahadevatemple, Ittagi;(f)
Mallikarjunatemple, Kuruvatti

18a
K-S1-K
K-S2-K
18b
K-P-P-P-K
K-P-S1-P-K
K-P-S2-P-K
18a 18b 18c 18c
K-Ps-S2-Ps-K
K-Ps-S2-Ps-K
K-Ps-S2-Ps-K
tA K - P s - S 2 - P s - K

18d
K-S2-K
K-S2-K
K-S2-K

18e
K-Ks-(S2)-Ks-K
K-Ks-(S2)-Ks-K
K-Ks-(S2)-Ks-K
K-Ks-(S2)-Ks-K
18f
K-Ks-(S3)-Ks-K
K-Ks-(S3)-Ks-K
K-Ks-(S3)-Ks-K
18d 18e 18f
K-Ks-(S3)-Ks-K

of the multiaediculartop tala,with the use of identical ele- motifs) in the wall's recesses, their smaller scale making
ments in each tier.19This is all part of the drift toward them a tertiaryregister, in counterpoint to the (generally
fusion through proliferation,aediculardensity (enhanced Dravida)wall shrinesin the projections.
by very "compressed"talas),assimilation,and radialconti- A fertile arrayof new primaryaedicularformsare tried
nuity. Each tala has corner kuta-aedicules and double- out in the Sudi temples, particularlyin the mndazpa(hall)
staggeredsala-aediculesat the center. In between is a new walls of the Mallikarjuna,with its towerlesslateralshrines,
form, the panjara-stambha, a panjara over an embedded and the nearbytank,which has walls treatedlike an inside-
stambha,or pillar,insteadof paired,post-depictingpilasters out temple. Kuta-stambhas, adld-stambhas,
and panjara-stam-
(Figure 21). This is not strictly an aedicule, a shelter, but bhas appear, and, more interestingly, the new palette of
remainsan "aedicularcomponent"in the design. aediculetypesbeginsto be combinedimaginativelyinto var-
The pavilion-on-stambhaimage-the Dravida kuta- ious interpenetratingconfigurations.A remarkableaedicu-
stambhawas soon to become the most common-was prob- lar constellation of this kind is the centerpiece of the east
ably inspired by the Nagara kuzta-stambhas
(Latina sikharas, vimana of the Joda Kalasa temple, which faces the west
or curvednorthernspires, on pillarforms) that had by this vimanaacrossan open, pillaredhall. While the west vimana
time become essential aedicularcomponents of the devel- makes a classic statementof currentpossibilities,the east-
oped, composite modes of Nagara, the Bhumija and the ern one is a feat of virtuosity,applaudedin a contemporary
Sekhari.The temples of Sudi and Kukkanurare among the inscription.20On plan it is the same size as the western
first in Karnatakato use Nagara ku.ta-stambhas as wall shrine, again with five projections and identical talas,but
shrines. These are generallysheltered by toranas(archway threenot four.Figure21 showsthe intermediateandcentral

TEMPLES OF KARNATAKA 191

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
Figure 19 Mallikarjuna
temple, Sudi, first half of eleventh century Figure 20 Joda Kalasatemple, Sudi, west vimana,before 1060

elements in the first tala.The formerarepaijara-stambhas, for the more modest three projectionsand three talas,each
puttingforth a form of Dravidakuta-stambha, which in turn K-S2-K. (In fact,this does not actuallybecome a normuntil
emits a smaller kuta-stambha,the whole thus forming an the following century,for the less ambitioustemplesof the
emanatorysequencethatperenniallycropsup in Indiantem- "nonmainstream.") One specialfeature,probablyused here
ple architectureandmythology.In the adjacentrecessis one for the first time, is significant.21It can be called a "double
of the "exotic"(andas yet not thoroughlyunderstood)con- nasz,"in which the centralnasiof a sla/ gives forth a second
temporaryNagara forms found in this vimana:a Bhimija- one that pushes through the parapet moldings and the
esque wall shrine containing a Dravidakuta-stambha. The kapota(see Figure 22). This is a powerful innovation, but
highly inventive central element is a staggereds'il-aedicule, again not so much a "conceptualshift"as another expres-
but the emergentform (with enlargedkapota)in the middle, sion of the perennial sense of manifestation.Down each
instead of a panjara-aedicule,is a special stellate kuta- face of the vimana,as if originatingin the nasiof the main
aedicule-an eight-pointed, rotated-squarestar, of which dome, issues a stepped cascade of nasis, the downward-
one point projectsdiagonallyat the centerof the whole com- pouring made explicit through "pictorialrepresentation,"
position. At the secondarylevel, a pair of kuta-stambhas are with each arch form spewed from the jaws of its monster-
visible in the wall zone of the stellatekuft-aedicule,and the face finial.The spine thus formedis anotherof the Nagara-
latter bursts through a secondarys'aia-aedicule.At a still seeming traits of "Vesara"temples that has fascinated
lower level of order, in the parapet of the stellate kuta- scholars.
aedicule, miniature kuta-aediculesreplace the vyalamalai The Lakkundi school of the Later Cilukya "main-
molding, one directlyabove eachwall pilaster. stream"uses chloritic schist ratherthan the sandstone of
The Kallesvaratemple, Kukkanur,maywell be a prod- Sudi,Kukkanur,and earliertemples.Whereasthe Sudiand
uct of the sameworkshopas the Sudi temples (Figures18d, Kukkanurmonumentsconcentrateon compositionalinven-
22). Like the Joda Kalasawest vimana,it states a virtually tion, with moldings employedfor their rhythmicmodula-
inevitablenorm for the currentphaseof the tradition,here tion, the Lakkundi works enjoy sumptuous surfaces.
192 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 Figure 22 Double-staggered s/la with double-nas?, KalleSvara
temple, Kukkanur

At this time, as aedicules proliferate,wall shrines are

beginning to become multiaedicular,with their superstruc-
tures treatedas miniaturevimanas.Many of the most inter-
esting experiments in vimana design take place at this
miniaturescale, and it is here that Nagara shrine forms are
depicted in certain temples. Both the Bhumija and the
Sekhari modes of Nagara are displayed at the
Kasivisvesvara-the Sekhariprominently,over the (origi-
nally image-housing)centralniches of the west vimana.
These niches are probably the first of a series of
extralargewall shrines, and the latest step in a process as
these elements become more pronouncedover time. Even
Figure 21 Joda Kalasatemple, Sudi, west vimana:panjara-stambha
(Ps) (left) and stellate kuta-aediculeemerging from .sa/-aedicule
before becoming multiaedicular,wall shrines have pushed
forward in the plinth, gaining their own projection (not
present, for example,in the Naganatha,Mahakuta).Then,
Sometimes, as at the Kasivisvesvara,self-conscious inven- through "burstingof boundaries,"their eaves have strayed
tion is appliedto the shapesof moldingsand detailsas much over and through the pilasters;at the Kasivisvesvara,the
as to the overalldesign, so that its surfacecharacter,as well niche has usurpedthe place of the forwardmostprojection
as its concept, is unique. This is in addition to the relief of the sila-aedicule from which it grows, and its super-
carvings,dwarfs,and monkey groups that adorn the tem- structurethrustsup throughthe kapotaand into the parapet.
ple exterior. The west vimana follows the now recessive (It is in the temples of the Hoysalasthat centralwall shrines
notion of the top tala,here the thirdone, takingthe form of eventuallyjumpright out and become fully roundedminor
an alpavimana,though the crowningkutais almost entirely vimanas,only just touching the wall from which they have
lost (Figure23). In the first two talasthe wall has five pro- emerged. By this stage the spreading eaves have merged
jections: modified staggered sla/-aedicules at the center, into a continuous canopy,dividing the wall into two hori-
kuta-aediculesat the corners,and kuta-stambhas in between, zontal zones.)
of which the kutas are a special, Nagara-inspired form. The central Sekhariwall shrine of the Kasivisvesvara
These are grouped in with the central element, tied at the appearsat a smaller scale in the second tala, the repeated
level of the kapota(which is rather like an overhanging forms caught in the emanatoryrush of a nasi-cascade,the
chadyacanopy) and plinth (with its upapitha,or subbase), lower "naszs" now trilobate.Enough of the top talaremains
where there are no interveningrecesses.22 to show that this cascadeoriginatedin the crowning dome

TEMPLES OF KARNATAKA 193

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 Figure 23 Kasivisvesvaratemple,
Lakkundi,west vimana,c. mid eleventh
century

of the vimana.The double-nasiideais now appliedat a small stambhas.Double ndsisand centralwall shrinesin each tala
scalein the plinth,wherenasisin the plinthkapotagive forth form a cascadingspine, echoed in the narrowkutas of the
toranas(archways). kuta-stambhas.
The experimentationof the eleventh centurydoes not Within this composition, subtle changes of detail
lead to a fixed standard,but there comes a moment when give dramaticresults. For example, in the sikhara mold-
innovationin the compositionof vimanasbecomesno more ing, the layer of stone that is modulated to form the kuta
significantthan innovationin the overalllayoutof the tem- domes and ia-droofs, instead of stopping at nasis and sadl
ple and surroundingsand, at the other end of the scale, in gables, is carriedthrough, strengthening horizontal con-
the nuances of small detail. The Mahadevatemple, Ittagi tinuity at the expense of the definition of roof images.
(A.D. 1116), is no mere derivativework, but rather,accord- More than ever, a sense of productive fusion is achieved,
ing to a dedicatoryinscription, an "emperoramong tem- an oceanic continuum of horizontal strata where the
ples" (Figures 18e, 24). Yet its vimanacomposition is not familiarimages emerge and dissolve. The effect was evi-
original, closely following the Amrtesvara temple at dently noticed and admired. For example, while clearly
Annigeri.The central element is a double-staggereddsal- built by a differentworkshop,the Tarakesvara,Hangal (c.
aedicule,the place of its middle projectionin the wall zone mid twelfth century),an impressivework of the feudatory
(in the mannerof the Kasivisvesvara) takenby an extralarge Kadambadynasty,follows Ittagi in these small details as
wall shrine, while the intermediate projections are kuta- well as in overall form.

194 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 Figure 24 Mahadevatemple, Ittagi,1116 (a later
additionhas replaced the originaldome)

The most radical transformation of the Karnata It maybe thatimpetustowardstellateformscameabout

Dravidavimanais its metamorphosisinto the stellateform. throughcontactwithNagaratraditions-with the muchear-
This is commonly associatedwith the Hoysalas,who tried lier stellatetemples of DakshinaKosala24or contemporary
out many variants. However, it is in the Later Calukya stellate Bhumijatemples of central India. It is true that a
"mainstream" thatstellatetemplesfirstappearin Karnataka, growingfascinationwith diagonalityin primarycomponents
towardthe end of the eleventh century,at Savadi,Kalagi, and in details did not predatethe introductionof stellate
Konnur,and Dambal.Of these the most ambitious,andthe aedicules, notably those in the Sudi Joda Kalasa temple.
only one to survivevirtuallyintact,is the Doddabasappaat However,it is possibleto conceiveof the star-shapedvimana
Dambal(earlytwelfthcentury),wherethe hall as well as the springingfrom an organicdevelopmentof the regionaltra-
vimanaare stellate. The vimanaplan is based on a square dition,asthe apogeeof the tendencytowarda morepervasive
rotated about its center to create a star with twenty-four use of identicalaediculetypes:as forcespushout diagonally,
points, each point locating the salient angle of a kuta- the bubbleof centralemphasisfinallyburstsand everything
aedicule (Figure 25). Six further,identicalstories rise to a settlesinto the descendingspiralof the uniformstellateplan.
dome of identical stellate plan. Between the primaryele- In any case, the Doddabasappais manifestlya neplusultra.
ments, reentrant projections, located on plan by rotated ElsewhereI have arguedthat, without completelyviolating
aretreatedas deeplyembeddedkuta-
equilateraltriangles,23 the natureof the architecturallanguage-for example,with-
stambhas. out makingthe kuta-aediculesabsurdlytall and thin-this is

TEMPLES OF KARNATAKA 195

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 Figure 25 Doddabasappatemple, Dambal,
earlytwelfth century

the formwith the maximumnumberof pointsto the starand lished, its resurgent temple architecturestems from the
the maximumnumberof talas.25 Dravidatraditionof Tamil Nadu, with only faint Karnata
Any further evolution would have brought complete resonances.
fusion into a new unit, a groovedcone, which under differ-
ent circumstancesmight have become the seed of a new
Conclusion: "Vesara"-Invention or
cycle of growth and proliferation.This was preciselywhat
happenedwith Nagaratraditions,the multiaedicularproto- Inevitability?
Nagara fusing into the unitaryLatina,26which eventually We maynow returnto the questionof whether"Vesara"or
branched into the increasingly composite Bhumija and "late KarnataDravida"is the more suitable term for the
Sekhari.In the two centuriesafterthe Doddabasappa,archi- temples of medieval Karnataka.Though terminology
tecturalinvention in Karnatakais channeled,for example, shouldnot be overemphasized,it is confusing,whateverthe
into the planningof elaborateopen halls and the forms of traditional texts may have meant by the term, to make
theirnumerouspillars.But vimana designmaintainsits vari- "Vesara" equivalentto "Nagara"and"Dravida." Nagaraand
ety, even if the sense of evolutionary direction is not as Dravidaare two architectural"languages"in that they pro-
markedas before. Permutationsof stellate and semistellate vide a "vocabulary," a familyof relatedforms, a kit of parts
forms are explored,as well as variousmeans of achievinga (primarilya paletteof aedicularcomponentsand a rangeof
Nagara/Dravidahybrid.27 moldings), that can be put together in variousways. Dis-
It is only the later works of the Hoysalas that give a tinct "modes,"or organizingprinciples,are developedfor
sense of sterility. Early in the fourteenth century, as the the Nagara: Latina, Valabhi, Sekhari,Bhumija.28Could
whole Deccan fell prey to Muslim invasions,the Karnata "Vesara," then, be definedas a mode of Dravida?Not really,
Dravida tradition, already stagnating, dried up. When, since, in transformingthe Dravid.avocabulary,the way of
around1336, the Hindu empireof Vijayanagara was estab- combiningthe elementsis not fundamentallychanged.The

196 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
stellate"Vesara" could more properlybe calleda new mode, providing continuity behind the ingenuity. But when it
but even here the way of putting the partstogether is still a came to particulartemple designs, it was not fixed forms
matter of piling them up pyramidally. that were transmitted,but a way of creating. This way of
As a classification,then, the term "Vesara"should be creating, which can be understood when a proper way of
used with caution.Is it useful for its connotationof hybrid- seeing is found, demanded improvisation. Architectural
ity? It has alreadybeen pointed out that the hybridity in invention was not something exceptional;it was essential.
question is not the same as the deliberate mixtures that Where standardtypes of individualvimanacompositiondid
Karnataarchitectsenjoyedfrom earlytimes. It should now emerge, mainly among humbler temples, these types did
be clear that the apparentlyNagara characteristicsof the not remain static but evolved with everythingelse, at least
form-the staggered plan, the radial continuity and cen- in the KarnataDravidatradition.
tralspine, the barelysteppedprofile-were not introduced The question, then, is not whether the temple archi-
suddenly, borrowed ready-made, but evolved. These tects were inventive,but in whatway.Their creationscould
aspects emerged through tendencies that had already be virtuosicand, as inscriptionsshow, could evoke admira-
begun before Nagara temples arrivedat a staggered plan. tion and amazement.But their experimentswere not indi-
Nagara temples may have been a stimulus, but develop- vidualisticor arbitrary.Viewed in a broad perspective,the
ments were not interruptedin the Rastrakutaperiod,when way of designing was such that all the exploration of the
no Nagara temples were built in Karnataka. Karnata potential combinations and permutationsof the medium
Dravida temples are reminiscent of Nagara ones because led in a particular direction. Increasingly, the forms
they evolved in the same way. If "Vesara"implies deliber- expresseda dynamicsense of "centrifugalgrowth,"and the
ate hybridizationat a particulartime, then the term is best compositionsbecame dense and complex,proliferatingyet
discarded. merging in a "cyclicalevolution."
It is impossible to identify a moment when temples The question of why the artists created in this way is
ceased to be "Dravida"and became "Vesara,"and once best understoodin relationto culture,or worldview,or way
"Vesara"had been achieved, nobody continued to build of thinking. Whether or not "all forms of Indian art"31
"Dravida."The traditionmust be studiedas a whole to see express "cosmic manifestation," the great traditions of
that the eleventh-centuryforms arose from a way of think- Indiantemple architectureembodyin their formalpatterns
ing that had been transformingvimanadesigns for the pre- the intuitionsof realityalso articulatedin philosophies,the-
vious four centuries. There was no sudden "conceptual ologies, and mythologies. Not merely through association
shift."It is not true that "VesaraarchitectschangedDravida or metaphor,but throughtheir actualcompositionalstruc-
architecture from a tradition to be preserved and repro- ture, the temples make concrete a sense of how one
duced to a problem to be explored,"29because Dravida becomes many,and many returnto one.
architecturewas never static.
Can the later monuments be predicted by the earlier
ones?To claim that they were predestinedwould be ridicu- Glossary of Architectural Terms
lous; yet, given an architecturallanguage with inherent alpavimana"Minorvimana,"unitaryform, equivalentto a
properties,coupled with a propensityfor a particularkind single "kuta-aedicule,"or occasionally to a "sa-l
of evolution, certain developments do bring an uncanny aedicule,"rectangularor apsidal
sense of inevitability.The unfolding traditionimprints its haraChain of pavilionsin an entablatureor parapet
stagesin the unfoldingtemple patterns.Vimanaformspro- kapotaRoll cornice molding, representingthatchedeaves
liferate downwardinto furthervimanaforms. The single- kutaDomed pavilion,formingthe superstructurein a "kuta-
staggeredsala-aediculeblossomsinto the double-staggered, aedicule"
the nas into the doublenasi.It is not purelywith the wisdom kuta-stambha Pillar form with a kutaon top
of hindsight that we are not surprisedby the arrival,at the nasiHorseshoe arch motif
Mallikarjunatemple at Kuruvatti(Figure 18f), of a triple- pafjara "Cage,"pavilionwith a ndszroof (equivalentto the
staggeredsa/-aedicule with a triple nasi. gable end of a saila),forming the superstructurein a
Architecturallanguages,their basic elements and cer- "panijara-aedicule"
tain ways of arranging them, were indeed revered and sild Barrel-roofedpavilion,forming the superstructureof a
passed down through the tradition.30These languages, "is/d-aedicule,"normallyrectangularbut occasionally
Nagara and Dravida, with their differentkinds of "aedicu- apsidalat one end
lar composition," were the actual medium of expression, sikharaSpire-likesuperstructurein Nagaraarchitecture,or

TEMPLES OF KARNATAKA 197

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
 roof-likemolding (dome of kuta,etc.) in DrIvidaarchi- define new norms for temple structuresthrough practice itself." Sinha
tecture develops his argumentin relation to a number of Later Calukyatemples
illustrativeof a phase of experimentand inventiveness.Largelythe same
stambhaPillar or pillarimage
examplesare used in the presentarticle,so that the interpretationscan be
talaTier or story
compared.
vimanaThe "shrineproper"of a Dravidatemple-sanctum 7. In exploringarchitecturalpracticeon the basisof the monuments,Sinha
and superstructure admitsthat he does not "meanto claiminsight into the minds of the archi-
tects"(ibid., 382). Less modestly,I would claimexactlythat insight, at least
vyalamalaBandof leonine mythicalbeasts (vyalas);molding
used in the base and superstructureof a Dravidatem- to the extent that an appropriateway of seeing is a matterof graspingthe
composition,the conceptualstructure,andthusthe way in which the archi-
ple, representinga line of joist ends tect thought about and createdthe form. This much can be deducedfrom
the buildings themselves, in which the formal structureis contained. At
the level of form, if not content, culturaldistanceis not prohibitive.Given
a few examples,a musicalMartianmight unravelthe structureof a fugue.
Notes For Indian temples the task is easierbecausethey are not purelyabstract.
1. Michael Meister has extensivelyanalyzedtemple plans and their rela- More thanwith most architectures,here we havea representationaldimen-
tionship to mandalagrids, e.g., "Measurementand Proportion in Hindu sion, an architecturalimagery,that is at the origin of the formalstructure.
Temple Architecture,"Interdisciplinary ScienceReviews10, no. 3 (1985): 8. These ideasseem to be gainingacceptance,to varyingdegrees.For exam-
248-257. ple, Vidya Dehejia includes the broad argumentin IndianArt (London,
2. If, at least since the nineteenth century,architecturalhistory has been 1997), 146 and fig. 100. Gerard Foekema uses the aedicularanalysisin
sensitiveto evolutionarypatternsin architecturalform, an honorabletra- HoysalaArchitecture, 2 vols. (New Delhi, 1994),but steersclearof expressed
dition of skepticismtowardevolutionarychronologiesdatesbackat least to movement,evolutionarytendencies,and symbolism.
Geoffrey Scott's debunkingof the "BiologicalFallacy,"seen as valuing a For KapilaVatsyayan,on the other hand, the symbolismis the self-
building'splace in a patternof developmentratherthan on its own terms evident part. In her introductionto Concepts of Time,AncientandModern
(TheArchitecture of Humanism[London, 1914]).Alongsidethis mistrustof (New Delhi, 1996), xxix, referringto my paper "Time in Indian Temple
diachronicgrandschemes has been a critiqueof attemptsto explain,syn- Architecture"(354-372), she writes:
chronically,the creationsof a given time or culturein termsof a worldview While art historians will no doubt debate the structural analysis of Dr. Adam
or of a "spiritof the age," and to claim that meaning could be embodied
Hardy... there will be littledisagreement with his cautiouslystated conclusionthat
througharchitectureas a reflectionof such a worldview.Skepticismof both "the dynamicformalstructureof Indiantemples shows irresistibleanalogies with
kinds of all-embracingschemes,which are often seen as Hegelian, is most
certain metaphysicalideas recurrentin Indianthought:of the manifestationin tran-
cogentlyexemplifiedby ErnstGombrich,for example,in TheSenseof Order
sient, finite multiplicityof a timeless, limitless, undifferentiatedyet all pervading
(Oxford,1979).The writingsof the philosopherKarlPopper have formed
unity;of the identity of this oceanic infinitudewith the all-containinginfinitesimal
a basisfor a critiqueof evolutionismby Gombrichand others;see Popper's
point;of finite things as fleeting transmutationsof the infinite,momentarilydiffer-
ThePovertyofHistoricism (London, 1961).All of these viewswould be skep-
entiated,then sinkingback into unity,in unendingcycles of growth and decay."This
tical towardthis understandingof Indian temple architecture,which nev-
is true not only of Indianarchitecture,but of all forms of Indianart.
erthelessis borne out, I believe, by the temples themselves.
3. See AdamHardy,IndianTemple Architecture:
FormandTransformation, the Ajay Sinha has recently reviewedmy IndianTempleArchitecture in Artibus
KarnataDravidatradition,7th to 13th centuries(New Delhi, 1995). This is Asiae63, no. 3/4 (1999):358-361. With reservations,he acknowledgesthe
based on my Ph.D. dissertation,"The KarnataDravidaTradition:Devel- argumentsfor dynamism,for a particularkind of evolution, and for the
opmentof IndianTempleArchitecturein Karnataka,7th to 13thCenturies," symbolism of manifestation. Oddly, he seems not to notice the idea of
Councilfor NationalAcademicAwards(U.K., 1991).The other recentspe- "aedicularcomposition,"the basis for understandingthe others, and does
cialistworkson the LaterCalukyaphaseof this tradition,focusedon in this not see the differencebetween "kuta"and "kufta-aedicule" (op. cit., 359).
article, are M. A. Dhaky, Indian TempleFormsin Karna.taInscriptions and 9. Heinrich Zimmer perceived in Indian sculpture a "phenomenon of
Architecture (New Delhi, 1977),M. A. Dhaky,ed., Encyclopedia ofIndianTem- expandingform,"expressingthe conceptof a universeconstantlyin the flux
pleArchitecture, vol. 1, part 3 (New Delhi, 1996), and a forthcomingbook of creation and dissolution;see Mythsand Symbolsin IndianArt and Civi-
by AjayJ. Sinha, based on his Ph.D. dissertation,"Originalityand Origi- lization(New York,1946), 130-136.
nation of VesaraArchitecture"(Universityof Pennsylvania,1993);see also 10. Stella Kramrischfirst describedthe temple as a "monumentof mani-
n. 6. festation";see TheHinduTemple(Calcutta,1946), 165.
4. See Hardy,IndianTemple 310-316: Appendixon "The Rela-
Architecture, 11. Through the structureof interconnectedaediculesit became possible
tionship between Karnata Dravida and Tamil Dravida,"a comparison for the vimana,the house for the god, to expressthe idea of a divinepalace
mainlyof EarlyCalukyaand PallavaDravidatemples. with many mansions,each a microcosmof the whole, just as the deity has
5. See Hardy, Indian TempleArchitecture,chap.10, 296-304; also idem, many aspects and manifestations.A hierarchyof divinities (of the kind
"HybridTemples in Karnataka,"in First UnderHeaven:TheArt of Asia, ascribedto the vastu-purusa-mandala,the sacreddiagramrituallytracedout
FourthHali Annual(London, 1997), 26-43. beforeconstructionof the temple)is thussymbolized,whetheror not sculp-
6. AjayJ.Sinha,"Architectural Inventionin SacredStructures,the Case of turalimagesof the gods are shelteredwithin the aedicules.
VesaraTemplesof SouthernIndia,"JSAH55, no. 4 (1996): 382-399; quo- 12. In his reviewof IndianTemple Sinhacallsmy formalanaly-
Architecture,
tationsp. 382. Sinha'smain point is that "architectsnegotiate architectural sis a "structural-linguistic
approach"(ArtibusAsiae 63, no. 3/4 [1999]:358).
traditionsas well as religiousideas in the act of creatingthe formaldetails The analysisis based on the architecture,however,the linguistic analogy
of theirmonuments.Farfromreproducingcanonsfixedin sacredtexts,they being an illuminatingbut very loose one.

198 JSAH / 60:2, JUNE 2001

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions
13. Hardy,Appendix(see n. 4). bhadra(centralprojection)at Kukkanurcorrespondsto that of the sanctum.
14. For a restoreddrawingof the Takht-i-Bahimonastery,see AdamHardy, This does not become the rule, thoughby this time the sanctumwidthdoes
"The Indian Subcontinent,"chap. 26 in Banister Fletcher, A Historyof regulatethe size of the principalstructuralbaysthroughoutthe temple.
Architecture,20th ed. (London, 1996), 752, Fig. E, or Hardy,IndianTemple 22. Sinha'sterm "bhadracluster"(ibid., 390) is meaningfulhere, where the
Architecture,408, fig. 11d. central and intermediate elements in a five-projection wall are bound
15. Examplesof the apsidalformof alpavimana, possiblyconsiderablyolder together, but the term does not seem relevantto normal five-projection
than the Aihole shrine but with the same molding sequence in the super- schemes.
structure,are found at Ter (Maharashtra) and Cherzala(AndhraPradesh), 23. Ratherthan,"Thewallshowsa columnwedgedbetweenthe karna piers.
both brickbuildingsand probablyoriginallyBuddhist. The column is createdwithin the boundariesof the rotatedsquareby sim-
16. The Mahakutesvara is the temple that Sinha,not recognizingthe con- ply carvingbackthe recessesaroundit";Sinhare the Savaditemple, ibid.,
stantdevelopmentof Dravida,chooses to illustratethe "typicalfeaturesof 394.
a Dravidatemple,"in contrastto "Vesara"("Architectural Invention,"384, 24. Such as the temple at Dhobini, illustratedby Sinha,ibid., 395, fig. 16.
fig. 3). 25. Hardy,IndianTemple 176-178 and fig. 28.
Architecture,
17. This is a version of the double-staggeredsada,the "nestedsala" that 26. The Dhobini temple (see n. 24) is againan example.
Sinha sees as essential to an eleventh-century "conceptualshift" (ibid., 27. Nagara and Dravidaelements often appearside-by-side,but the most
384-385). thoroughgoing synthesis of the two "languages"is achieved where the
18. All of this is lost if one fails to see aedicularityand movement: "At vocabularyof one is put together according to a mode of organization
Mahakuta,the hara just above the sanctumwall has a nested madhyasdld belonging to the other.At full scale this is best exemplifiedby the extraor-
[centralsaia];the one above has an unmodifiedDrvid a sala; and the top- dinarySomesvaratemple at Lakshmeshvara, in which Dravidaelementsare
most has only moldingssupportingthe hara without the stringof architec- arrangedaccordingto the compositionalprinciplesof the complexSekhari
turalmotifs that characterizeother stories. In effect, therefore,Mahakuta mode of Nagara. See n. 5.
seems to show only two fully articulatedstories below the dome" (ibid., 28. For a pictorial chart showing a basic typology of Indian temples, see
386). The staggeredsai/ain the firsttalais only mistakenfor the earlierform AdamHardy,"The Indian Subcontinent,"chap. 26 in Fletcher,Historyof
if the correspondingpilastersin the wall below are ignored;and this con- Architecture,767 (see n. 14). In relationto the classificationof "language"
voluted description of the top tala arises from not recognizing the alpa and "mode"suggestedhere, the term "style"is probablybest left to denote
vimana form,in which the dome and the moldingsbelow it constitutea big the idiosyncrasiesof particularschools or workshops."Dravidastyle"and
kuta. "Nagarastyle"have alwaysbeen as confusingas "Doric style"and "Ionic
19. Sinha (ibid., 391) refersto this vimana as a "standardization of Vesara style."
architecture,"though this precise design is not repeated.It should also be 29. Sinha, "ArchitecturalInvention,"396.
noted both that architecturalinventivenessremainsa prerequisitefor any 30. A systemof proportionswas certainlyamongthe rulesthatwere passed
importanttemple, and that, as in any Indian temple-building tradition,a down, but this is not yet understood;see Meister,"Measurementand Pro-
scaleof options,in termsof the degreeof complexity(numberof projections portion"(see n. 1).
and levels),is alwaysavailableto suit shrinesof differentdegreesof impor- 31. Quoted from Vatsyayan,Concepts of Time(see n. 8).
tance.
20. Lionel D. Barnett, "Inscriptions of Sudi," EpigraphiaIndica 15
(1919-1920): 85-94; Sinha,ibid., 382, 393, and n. 6; Hardy,IndianTemple Illustration Credits
Architecture,60 and 209 n. 4. Other relevantinscriptionsare discussedby Drawingsby author
Dhakyin IndianTempleForms(see n. 3). Figures8, 14. Photographsby author
21. Sinha("Architectural Invention,"384) points out that the width of the Figures9, 19, 20, 23, 24, 25. Photographsby GerardFoekema

TEMPLES OF KARNATAKA 199

This content downloaded from 141.233.160.21 on Wed, 02 Dec 2015 20:28:44 UTC
All use subject to JSTOR Terms and Conditions