Introductory Essay – III

III. DIVERGENT DEVELOPMENT: KARNATAKA

The divergence between Karnata Dravida and Khmer temples can be traced through their development of the similar, square vimana found at Bhitargaon and prasat at Sambor Prei Kuk over the succeeding seven centuries. The analyses of the site plans on these two websites demonstrate that both traditions employed a similar design tool, a vastupurusha mandala, to regulate the “spatial extension” or “emanation” of their shrines.8 Although based on the same cosmogenic principles, Karnataka and Khmer architects seem to have used these grids at different scales which emphasize different moments in the cycle of emanation. While Khmer builders used mandalas to project vast concentric terraces or tiers around a central prasat which remained largely “invariant” or unchanged, Karnataka sthapatis employed these grids principally to determine the intricate projections and recesses of a vimana’s interior and exterior walls or devakostas.  Since these were determined by the brahmapadas occupied by the garbhagriha’s altar, their outline could change greatly but their overall dimensions were constrained; hence emanation was principally expressed through the emergence of more complex, more faceted forms from the initial square.

Karnata Dravida “Architectural Language”                        

Prof Adam Hardy’s pathbreaking studies, already acknowledged, decipher the lexicon and grammar of what he terms a Karnata Dravida9 “architectural language” comprised of shared motifs, practices and beliefs enabling the elaboration of the basic vimana found at Bhitargaon and Deogarh into the complex forms seen in later temples such as the Virupaksa, Pattadakal (745 – 754,) Kasivisvesvara, Lakkundi (1084, 1173) and Isvara, Arsikere (TAQ 1220.) Prof. Hardy distinguishes eight specific syntactic operations10 which Karnataka sthapatis deployed for the “aedicular expansion” of their vimanas: 1) projection, emergence, 2) staggering, telescoping, 3) splitting, 4) bursting walls and aedicules, 5) “expanding repetition,” repetition of aedicules of increasing size on each tala, 6) “progressive multiplication,” multiplication of the number and type of aedicules on descending talas, 7) pictorial or iconic representation and 8) non-orthogonal, gyration or rotation. Together they express a dynamic of emergence shared with Hindu cosmogenesis, displaying both emanation and, simultaneously, the force behind and shattering it. These two websites attempt to apply Prof. Hardy’s powerful heuristic, doubltless often maladroitly and inappositely, in their analyses of one hundred Karnataka and forty Khmer “temple mountains.”  A steady progression of new aedicular types was introduced over the seven centuries identified by Prof. Hardy as constituting a Karnata Dravida lineage or tradition, spanning Early Chalukya (347-574,) Later Chalukya (974 -1189) and Hoysala Dynasties (c. 1117-1343,) up to the Islamic incursions of the early 14th Century. The three, initial alpa vimana or simple shrine types – kuta, panjara and shalawere single- , double- and triple-staggered, stellated, hybridized, fused, positioned diagonally and radially to the temple’s walls and placed on columns or stambhas.They wereeven combined into multi-aedicular forms like the vimanas of which they were a part, for example, as Bhumija stambhas and the many fanciful inventions for wall shrines.

The greatest aedicular variation may have been achieved at an often over-looked, late Hoysala temple the Lakshminarasimha, Bhadravati (1260.) Its prastara (parapet, hara or first tala) includes: 1) stellated, octagonal kuta aedicules positioned, radially or corner-out, 2) irregular kuta-shala stambha hybrids (found at only one other temple and a wall shrine at a third)11, 3) panjara stambhas, 4) shala-panjara stambha hybrids and 5) intermediate diagonal kuta stambhas. This temple is also notable as the only Hoysala foundation, apart from the consummate Kesava, Somanathapura (1258 – 1265,) whose three kutas (here vimanas) and eastern porch could be said to form an equal-armed cross. 



The Gavaksha As Architectural Trope

The gavaksha has already been mentioned as one of the most widely used tropes or “figures of space” in Karnata Dravida architectural rhetoric where it plays a central role in the “aedicular expansion” of the vimana. 1) It can form the horseshoe or ogee vault of a panjara aedicule, equivalent to the gable end of a shala, where it frames an opening into itself and, by implication, into the cella behind it. 2) It can also become a window or a tympanum in which the head of a god or an aedicule like it appears. 3) As a dormer, a gavaksha splits the shikhara (here, an aedicule’s dome, roof or gable of kutas and shalas.4) It also recurs at regular intervals along the kapota eave moldings both on a vimana’s superstructure and plinth. 5) As a harantara panjara aedicule, it protrudes through the galleries, (cloisters or colonnades of arpita shrines,) linking each tala’s hara (necklace) of aedicules. 6) Finally, the emanative potential of the split gavaksha makes possible the interlocking gavaksha meshes covering the pagas (vertical strips)and urushringas (subsidiary towers or peaks) of the shikharas of Nagara or northern Indian Latina and Sekhari mulaprasadas, as well as some of the early temples in Karnataka at Pattadakal.

The unassuming gavaksha dormer or nasi also plays an unexpectedly dramatic role in splitting the entire face of a Karnata Dravida vimana; Prof. Hardy cites the example of the western (Shiva) vimana of the Kasivisvesvara temple at Lakkundi. There the vyalas atop the central panjara and shala aedicules of the remaining 2nd and 1st talas emit a second vyala which not only splits the original nasi but spews its own gavaksha crashing through the prastaras of the lower talas. These, in turn, are split into three lobes by large Nagara aedicules, which can be read as jyotirlingas, extensions of the vyalas’ tongues, pendants or replicas of the vimana itself and hence Mt. Meru. They form what appears as a continuous “nasi cascade” down the vimana’s faceterminating in a large wall shrine bursting through the devakosta of the garbhagriha. In all these cases, the panjara’s and gavaksha’s distinctive horseshoe arch underlines their common ancestry in the chaitya cave temple which becomes even more explicit on the Kasivisvesvara’s eastern (Suryanarayana) vimana.12

A split gavaksha may also have been a source for the three- and five-lobed floral garlands or torana arches spanning the split pilasters of the wall shrines of Karnata Dravida temples. They appear on Khmer lintels and around the pediments above them, their reptilian connotations emphasized by the gaping jaws of the makaras, crocodiles or sea monsters,  which terminate them. A torana arch is also the basis for the pyatthat, (cf.Sans.> prasada palace, temples,) the pediments13 above the doors and windows of many of the three thousand shrines at Bagan in Myanmar – which itself constitutes a third “temple mountain” tradition. All these forms could be seen as homologous, in the biological sense of bearing structural traces of their common origin; hence they are related not just by analogy but ancestry. A gavaksha not only enacts the transformative dynamic of emergence and reabsorption, it functions as an engine of that emanation.

Aedicular Density

As the walls of a Karnata Dravida vimana begin to bulge and acquire more rathas, (pagas, projections, redents, facets)through thetransformational syntax of “aedicular expansion,” they stretch itsoriginal square footprint until it approximates a shallow, stepped octagon. This eventually leads to quasi- and fully-stellate plans whose points are positioned by inscribing a circle around the vimana square and then “staggering” that square at 90°, 45°, 22.5°, 15° intervals. This limits the “aedicular expansion” of the original vimana to its diagonal or 1.41 times the length of its square’s side.

The fully stellate Doddabasappa temple at Dambal (1226) has 24 major kuta points (technically, a “stellated icositetragon,”) around a larger circle, as well as 24 intermediate diagonal kuta stambha points positioned by staggering an equilateral triangle around a smaller, inner circle, each aligned on radii separated by 7.5°. These 48 points ascend up all seven talas, giving the superstructure the profile of a truncated cone (or nuclear cooling tower) and the base room for 120 wall shrines (without subtracting those “submerged” inside its antarala and sukanasa.) The resulting mural surface is so folded or fractured that the individual aedicules and their facets are subordinated to an overall sense of superabundance.

Dvikuta, trikuta, chaturkuta and pancakuta temples, that is, with two, three, four and five kutas, (vimanas, peaks, shrines) clustered around a shared mandapa, became common under the Kannadiga Hoysala Empire (c.1117 – 1343.) Temples such as the Kesava, Somanathapura, display a tendency towards fusion, a quasi “four-way” or “cruciform” symmetry with an extended mandapa (rangamandapa, mukhamandapa, ardhamandapa,)as their fourth arm. Thus, the addition of vimanas can be seen as having as much a centripetal as centrifugal impact on the design’s unity.  Similarly, the rotation implicit in a stellate vimana and the radial, rather than strictly orthogonal or diagonal, alignment of its aedicules contributes to a sense of motion, while emphasizing the gravitational pull of the invisible (niskala) center around which its points were staggered. Such aedicules do not simply seem to emerge but spin around their centers, as well as the vimana’s unseen center, an architectural equivalent to Dervishes’ semazen or “whirling dance.”

The Karnata Dravida vimana could thus be said to expand its mural surfaces – and with them the opportunities for additional facets, aedicules and sculptural embellishment – not by expanding its dimensions but primarily through crenellating, folding, projecting and redenting its walls. This has a net effect of thickening or concentrating the vimana’s “aedicular density,” in marked contrast with Khmer “temple mountains’” expansive spacing of their prasataedicules.” Thus, the walls of Karnata Dravida vimanas express the contradiction or dichotomy implicit in their own “aedicular expansion” by both “emerging” and “bursting boundaries,” simultaneously revealing themselves and revealing the force driving them apart. This density and dynamism can make stone temples appear to jitter, pulsate, even breathe, as the beholder confronts the sheer profusion of unfolding details, a prolix, grandiloquent architectural rhetoric which sometimes threatens to drown out its particulars. This might, however, be the point: the emanations are lost in the emanating, the stellate form in the scintillating energy field of the non-manifest.

The Vimana: Linear Expansion

Karnataka shrines, apart from the Early Chalukyan royal precinct at Pattadakal (c.675-754,) the literally monolithic Kailasa temple (756-773) at Ellora and the Hoysala dynastic complexes at Belur and Halebidu (1117/1121,) were not royal foundations but sponsored by military leaders, court officials and local magnates. These endowments’ comparative modesty might also have been influenced by, what could be described as, the “Fouquet Effect,” the imprudence of a display of over-weening architectural ambition. For their part, Kannadiga kings may have been reluctant to appear to favor any sect in their multi-confessional domains; once a temple was established, however, these rulers often granted generous agraharas, villages entailedfor the support of brahmanic scholarswho also attended to a community’s ritual needs.

Despite the fulsome panegyrics to these worthies on dedicatory stelae, it would have been impious for them to pretend to the divinity of a Khmer devaraja. It is hard to imagine an Indian potentate sequestering himself at the top of a shikhara to commune with his divine relations, even had that been architecturally possible – anymore than a mortal would claim to have scaled Mt. Meru in the face of the six oceans and mountain ranges intended to prevent him. Therefore, it is not surprising that Karnata Dravida templesare resolutely earthbound and the intersection of heaven and earth at their peaks merely symbolic.

The epigraphic record is strangely reticent about these temples’ liturgical, as opposed to self-promotional, purposes and it would be “presentist” to impute today’s Hindu practices to such a remote past. To some extent, the temples must be left to “speak for themselves” – with the caveat that “architectural languages” permit extreme latitude in their translation (as these two websites demonstrate too frequently.) It seems clear, however, that the limited size of Karnata Dravida vimanas played a decisive role in how they could be expanded or extended. Early Chalukyan temples, such as the Upper and Malegitti Shivalaya at Badami, and Galaganatha and Virupaksa, at Pattadakal, were sandhara, that is, with pradakshina pathas wrapped around the garbhagrihas inside their vimanas.  The weight of a superstructure large enough to cover these internal passageways may have led to such designs falling into desuetude, since Later (Western, Kalyani) Chalukya temples were almost exclusively nirandhara, without provision for interior parikrama. Hoysala shrines often placed their vimanas on a jagati or platform lined with narrative bas relief and six or more adhisthana friezes for viewing during dakshina (south-facing, right-handed, solar) circumambulation.

Since there was no place in these vimanas for worshippers to assemble, perform purificatory rites or honor ancillary aspects of the deity, only a doorway through which to experience darshan, antaralas (vestibules,) gudhamandapas (enclosed mandapas,) open or screened sabhamandapas (assembly halls) and rangamandapas (pavilions with a stage,) as well as ardhamandapas (half mandapas) and mukhamandapas (outward-facing mandapas or porches,) were attached to the east of the garbhagriha. Karnata Dravida temples, as well as Nagara temples, such as the celebrated Kandariya Mahadeva at Khajuraho and Lingaraja at Bhubaneshwar, might therefore be described as “linearly” or “additively” expanded, with well-defined, “directional,” east-west liturgical paths admitting few deviations. This contrasts with the “concentric” or “cruciform” expansion of Khmer “temple mountains” with their satellite prasats and connecting cloisters.

The gudhamandapas and rangamandapas of most Karnata Dravida temples, for example, the Mahadeva, Itagi and the Isvara, Arsikere, hold roughly the same number of people which suggests the majority of visitors to these shrines could eventually expect to reach their inner sanctums. At the same time, the constrained spaces and close columniation of these halls would militate against formal processions and elaborate public rituals, encouraging instead a continual flow of worshippers and contributing to the “casualness” sometimes attributed to Hindu puja or worship. The small size of Karnata Dravida temples implies they were primarily intended for the daily use of local communities and the resident brahmins, although groups of these shrines, such as those at Mahakuta and formerly, at Somanathapura and Talakadu, could at certain times of year become important pilgrimage destinations.