Madurai and the Tamil temple-city

If we return to Madurai to search for a historical context for the possible destruction of the Pandyan temple then the most likely moment would seem to be the late thirteenth to fourteenth century, the period following the great Chola Empire's disintegration in the 1280s following the death of Rājendra III in 1279. Over the subsequent century, a series of incursions into the Tamil country by the Hoysalas of southern Karnataka and others were followed by raids by the Khaljī and Tughluq sultanate from Delhi in 1310–11, 1318 and 1323. The conquest of Madurai by Jalāl al-Dīn Ḥasan Shāh in 1327 and his establishment of the Madura Sultanate (or Sultanate of Ma‘bar) independent from Delhi in 1333 displaced the last Pandyan rulers of Madurai further south.Footnote ² Within a brief period all of the old polities of southern India disappeared. This long period of political upheaval in the Tamil country continued from the 1360s, when the newly-founded and expansionist Vijayanagara Empire with its capital in northern Karnataka extended its rule over northern Tamilnadu under the leadership of Kampaṇa, eldest son of Bukka I, who then moved south overthrowing the Madura Sultanate. In some literary rhetoric, such as the fourteenth-century Sanskrit poem (kāvya) Madhurāvijaya, the royal general Kampaṇa and the Vijayanagara armies were presented as liberators, restoring order and temples to worship in those places where it had been disrupted.Footnote ³ The same is true of a few temple inscriptions from the period from 1361 to 1374. One example, on the maṇḍapa wall before the rock-cut cave of the Kakōḷanātha temple in the hill at Tirukkalakkudi near Tiruppattur east of Madurai, refers to the disorder caused by the Muslim Sultanate and its conquest by Kampaṇa: “The times were Tulukkan … Kampaṇa-Udaiyar came, destroyed the Tulukkan, established orderly government throughout the country and appointed many chiefs (nāyakkanmār) for inspection and supervision in order that worship in all temples might be revived regularly as of old”.Footnote ⁴

Some temples were certainly looted in the fourteenth century, and damage to the architectural fabric undoubtedly took place, but this was far from a widespread and sustained policy by any of these groups, Turk or Hindu, Tughluq or Hoysala.Footnote ⁵ Invading armies seemed very rarely, if ever, to have targetted the structure of the temple, but their presence put pressure on the local populace that resulted in the neglect of the temple and caused its gradual deterioration. (I would like to thank Leslie Orr for discussing this with me.) In common with all Indic conquerors, the Sultanate raids on south India were strategic in deciding which temples to target, focusing on the wealthy pilgrimage temples or those temples that were parts of claims to sovereignty, including those at Srirangam, Chidambaram and Madurai.Footnote ⁶ The Kūṭal Aḻakar temple in the heart of the Pandyan capital may have been considered such a worthy target, thus explaining why no trace of a pre-sixteenth-century structure has survived. But both Vijayanagara and the Sultanate courts had ideological reasons to exaggerate the degree of temple disruption and destruction. For the Sultanates it emphasized the success of their conquests and the acquisition of loot; for the Vijayanagara warrior-leaders it justified their interventions in Tamil affairs.

Yet even before pursuing a critical examination of temples in their political context, the assumption amongst some art historians has been that a temple was sacrosanct, given that it is the residence, or even the body, of God. Tamil temples are comparatively unusual across South Asia in having long building histories with repeated additions and expansions, at some sites over a millennium or more, to create the characteristic temple-cities of the region. South Indian temples have previously been described as developing only outwards in tree-ring or onion-skin fashion from the inviolate sacred main shrine, with the vimāṇa diminishing in size from the twelfth century onwards relative to the gopuras that are built on all four sides and increase in size towards the outer walls. This development of the rows of carefully aligned gopuras away from the central vimāṇa seems to embody the cosmogonic process that is often considered to be at the symbolic heart of all Hindu temples: the expansion outward from the sacred centre to the four quarters. The greater size of gopuras may also be the result of centuries of royal patronage: as the main shrine cannot be added to or replaced, so later rulers competed with their predecessors to build ever larger and grander structures around the outside of the temple. In this model of the Tamil temple, the earliest portions are understood to be those towards the centre with successive and later rings around the core.

This is evident in some of the early scholarly writings on the Tamil temple and has been repeatedly reiterated to account for their great scale. The first scholar to attempt a detailed study of the south Indian temple, and indeed the history of Indian architecture, was James Fergusson (1808–86). In the 1840s in the second major publication, Picturesque Illustrations of Ancient Architecture in Hindostan, of his later prolific academic career he wrote:

Such a conception of the south Indian temple was reiterated and more widely circulated in his later publications, The Illustrated Handbook of Architecture (1855) and the much-cited History of Indian and Eastern Architecture (1876, revised edition 1910).

Following extensive annual surveys of southern India over two decades from 1882, and the acquisition of an intimate knowledge of the region's architectural heritage and its many temples, the head archaeologist of the Madras Presidency Alexander Rea wrote of Srirangam in 1904 that: “As with the majority of the great Dravidian temples, the chief central shrine or vimana is of much greater age than the surrounding buildings. The shrine has been a small one, which has acquired sanctity by some means or other. Successive kings in later ages, have added the outer courts and mandapams”.Footnote ⁸ This view of the sanctity of the sometimes humble or “artistically insignificant” main shrine that preserved “the atmosphere of pristine holiness with which they were originally endowed” is evident in Percy Brown's account of the “Pandyan period” in the twelfth and thirteenth centuries in his influential and widely read study of Indian architecture. Noting a perceived shift in architectural production from the sanctuary to the outlying portions of the temple, he wrote: “Religious emotion with regard to such edifices had however to find some form of expression, and it did so by exalting their environment, surrounding them by high walls to emphasise their sanctity, and making the entrances to the enclosures thus formed into gateways of imposing size and rich appearance.” (See figure 2).Footnote ⁹

Figure 2. Aerial view of the Bhaktavatsala temple, Tirukkalukundram

But the detailed examination of temples’ development does not support such a model. The main shrine is not necessarily the oldest part of the temple as in many cases the vimāṇa has been replaced – sometimes several times – defying the assumption that it is inviolate. Neither has the development of the temple taken place in a consistently outward manner from a single core shrine. The conception of the vimāṇa with its adjuncts as the core of the sacred site can also be challenged, for the “subsidiary” deities may have been or may become “central” ones. For example, Padma Kaimal has argued that the early eighth-century Kailāsanātha temple at Kancipuram should be seen as “at least two contemporary and conjoined temples of equal significance, one to Śiva and one to goddesses. The prākāra articulates the female principle as it protects, enfolds, and circumscribes the male principle embodied in the vimāna”. Footnote ¹⁰ The early evidence for rock-cut sites is suggestive, with several adjacent shrines dedicated to up to five deities with no clear sense of spatial or ritual priority. The late eighth-century cave within the Skanda temple at Tirupparankundram has five shrines, dedicated to Śiva, Viṣṇu, Durgā, Skanda and Gaṇeśa, but from the fifteenth century the ritual emphasis was placed on Skanda, despite this shrine being to one side of the spatially central, and apparently pre-eminent, Durgā. Many other Tamil cave-temples from the sixth to ninth centuries have multiple shrines rather than a single ritual focus on a central deity, and with Śaiva, Vaiṣṇava and Jaina deities co-existing.Footnote ¹¹ An alternative model to the development of the south Indian temple as growing in onion-skin or tree-ring fashion, layer upon layer, would then be a more dynamic one of oscillating development. In Śaiva Siddhānta cosmology, the dominant form of Śaivism in Tamilnadu, individual ritual performance re-enacts the oscillation of the universe between emission and reabsorption.Footnote ¹² The historical dynamics of a temple's architectural development seem to embody this process, with an oscillation between expansion of the temple boundaries followed by the infilling of the open areas between the prākāra walls.

There is therefore no need to invoke violent destruction – or indeed religious change – for the remaking of a temple in sixteenth-century Madurai or indeed elsewhere in the Tamil region. Nor is there much justification for regarding the architectural fabric of a temple as sacrosanct and inviolate to the temple's devotional community. The rebuilding of the Kūṭal Aḻakar temple in the later sixteenth century was instead one element in a wider remaking of Madurai as an imperial and sacred centre under Nayaka rule.

The Nayaka remaking of the Tamil past

The Nayaka period in the later sixteenth and seventeenth centuries was an active era for temple construction and renovation all across the Tamil country. New temples were founded and existing temples were substantially expanded with additional subsidiary shrines, columned halls (maṇḍapas), corridors, and walled enclosures (prākāra) entered through towering, pyramidal gateways (gopuras). But many temples' main shrines, both Śaiva and Vaiṣṇava, were also completely rebuilt at this time: examples include the Ādi Keśava Perumāḷ at Sriperumbudur,Footnote ¹³ the Kaḷḷapirāṉ at Srivaikuntam,Footnote ¹⁴ the Naṭarāja at Chidambaram,Footnote ¹⁵ the Bhaktavatsala at Tirukkalukundram, the Kāḷamēkaperumāḷ temple at Tirumohur, the Vedanātha temple at Tiruvatavur, the Ādinātha temple at Alvar Tirunagari and the Bhū Varāha at Srimushnam. All of these temples' vimāṇas are good examples of Nayaka-period architectural design, to be seen alongside Madurai's Aḻakar temple. Though the sites on which these temples were built are much older, given their status in the pilgrimage geography of south India as places visited by the Tamil poet-saints, the Vaiṣṇava Āḻvārs and the Śaiva Nāyaṉmār in the sixth to ninth centuries, the main shrine's design is clearly Nayaka-period. The renovated (or more accurately, replaced) shrine does not replicate the older language of architecture: the design is not ninth or tenth century. However, the site on which the temple is built is exactly the same and the temple may have almost the same dimensions. The presence of displaced pre-Nayaka inscriptions within the temple or adjacent buildings in a clearly earlier architectural style further emphasizes that the Nayaka-period structure seen today is not the earliest foundation at the site.

Several examples from across the Tamil country may illustrate this Nayaka-period replacement of an earlier vimāṇa. The Ādinātha temple at Alvar Tirunagari on the river Tamraparni in southern Tamilnadu is located at the birthplace of Nammāḻvār, the most famous and revered of the āḻvārs. The present temple faces east and within two prākāras are two adjacent shrines: on the main axis aligned with the gopuras is a shrine dedicated to Ādinātha (Viṣṇu), and to its north a shrine to Nammāḻvār. Pandyan inscriptions on the north and south walls of the Nammāḻvār shrine indicate that there was clearly a temple on the site by the early thirteenth century.Footnote ¹⁶ However, even a brief examination of the main Ādinātha shrine (Figure 3) demonstrates that if there was an earlier vimāṇa here nothing now remains: the contrast in design between the plain mahāmaṇḍapa and the exquisite detail in architectural design and relief sculpture of the present Ādinātha's vimāṇa and ardhamaṇḍapa is striking. There are no inscriptions on this later building.

Figure 3. Vimāṇa of the Ādinātha temple at Alvar Tirunagari

Near Madurai, the Kāḷamēkaperumāḷ temple at Tirumohur is one of the divyadeśas; Nammāḻvār celebrated Viṣṇu here in his seated, lying and standing postures (āsana, śāyana, sthānaka). Although it is certainly an earlier foundation, for Pandyan inscriptions dating to the twelfth–thirteenth centuries are located on the interior of the first prākāra wall, the temple seen today was rebuilt in the sixteenth century and Viṣṇu is now standing only.Footnote ¹⁷ This later date is indicated by the design of the main shrine, its attached maṇḍapas, and the large sculptures of Rāma, Sīta, Lakṣmaṇa, Rati and Kāma at the entrance, and by the two recorded inscriptions dating to c. 1541 on the walls of the maṇḍapa. At nearby Tiruvatavur, the later Śaiva poet-saint Māṇikkavācakar's birthplace, the Vedanātha (Tirumaṟaināta) temple was a Pandyan-period foundation, as several thirteenth-century inscriptions on the second prākāra wall suggest. But the vimāṇa was entirely replaced in the early seventeenth century, though its earlier adjoining mahāmaṇḍapa and the surrounding prākāra walls were not.Footnote ¹⁸

Further north at Srimushnam (Tirumuttam), west of the well-known pilgrim town of Chidambaram, is the temple dedicated to Viṣṇu as Bhū-Varāha (Figure 4). The Viṣṇu temple faces west and directly behind it is another east-facing temple, dedicated to Śiva as Nityeśvara (or Nittīśvara). The Bhū-Varāha temple is of regional importance within southern India, however, for it is one of the eight “self-manifested” (svayamvyakta) Vaiṣṇava sites, which also include Srirangam in central Tamilnadu and Tirumalai (Tiruvenkatam) over the border to the north in modern Andhra Pradesh. Today Srimushnam is sometimes considered to be among the 108 divyadeśas, the list of sites of great pilgrimage importance to the Śrīvaiṣṇava community of Tamil Vaiṣṇavism. However, none of the Āḻvārs, whose poetry is considered canonical within the Śrīvaiṣṇava tradition, praised the site; this is a requirement for being considered one of the pre-eminent sites.Footnote ¹⁹ Srimushnam is instead one of the smaller group of Śrīvaiṣṇava abhimānasthalas, “esteemed places”. Although some scholars have used the poetry of the Āḻvārs and Nāyaṉmār to suggest the antiquity of a temple, there is no clear evidence that all the places of which they sang the praises were defined by a shrine or temple structure when they visited them. The surviving material fabric and associated inscriptions give a much clearer indication of a site's historical development. The building seen today at Srimushnam is essentially a good example of a late Vijayanagara or Nayaka-period temple, the great building age in southern India that defines the sacred landscape encountered today. The scale and layout of the whole temple, together with the design of the gopuras, the columns of the interior halls and indeed the main shrine and its attached maṇḍapas all clearly date the structure to the later sixteenth or seventeenth century (Figure 5). This is made clear by comparing the temple to the Śiva temple directly behind it (Figure 6).

Figure 4. Western gopura and the main entrance to the Bhū-Varāha temple, Srimushnam

Figure 5. Vimāṇa of the Bhū-Varāha temple, Srimushnam (© American Institute of Indian Studies)

Figure 6. Nityeśvara temple, Srimushnam

For both temples the epigraphic evidence suggests the phases of their development, and indicates the interests of various dynastic patrons across a period of five hundred years at this site. Despite the greater religious significance of the Bhū-Varāha temple, the earliest inscriptions in Srimushnam are all on the Nityeśvara temple. The foundations of the two temples (around 1100, or the immediately preceding decades) are suggested by two inscriptions on the walls of the Nityeśvara's vimāṇa dated to the Chola king Kulōttunga I's thirtieth and thirty-first years, recording grants of villages to temples of both Tirumuṭṭamuṭaiya-Mahādeva and Śrīvarāha Āḻvār by the king.Footnote ²⁰ A slightly earlier foundation date for the Bhū-Varāha temple may be suggested by the fragmentary inscription built into the second prākāra wall of the Nityeśvara temple dated to Vīrarājendra year 6 (1070) that seems to record a gift of land to Varāhadeva.Footnote ²¹ Upon its foundation (c. 1100) the Nityeśvara was quite modest, but expanded to its present extent, with two prākāras entered through a single gopura and with a separate goddess shrine dedicated to Pirahannāyaki on the north side of the outer enclosure, in the mid- to late fifteenth century.Footnote ²² Notable too amongst the inscriptions on the Nityeśvara temple are the references to the adjacent Viṣṇu temple both on the two temples' supposed foundations in the final decades of the eleventh century, and in the fifteenth century. There was clearly a Varāha temple in Srimushnam from around 1100 to at least the mid-fifteenth century; yet almost nothing remains of this earlier temple and the present structure may be substantially dated to the last quarter of the sixteenth century. In contrast to the chronological range of inscriptions by a host of donors across four-hundred years recorded on the walls of the Nityeśvara temple, fourteen of the sixteen recorded inscriptions from the Bhū-Varāha temple, all of which are on the first prākāra walls, date to the period from 1582 to 1600, making it clear that this temple was rebuilt by Koṇḍama Nāyaka, son of Vaiyappa Kṛṣṇappa Nāyaka of Gingee in the 1580s.Footnote ²³ Though there is no declaration of the temple's reconstruction, this is rarely a feature of Tamil temples in this period or indeed in all of the previous centuries; temples with clear, explicit foundation dates are the exception not the norm. A second objection that might be proffered is that the presumed earlier temple's vimāṇa could not be wholly replaced in the late sixteenth century because of the supposed sanctity of the earlier structure. But this example, and both earlier and later examples, demonstrates that it is the site that is sacred and not the structures built upon it and so earlier stone temples have been swept away and entirely replaced by pious renovators.

Sacred landscape in stone

This rebuilding of the vimāṇa is not new to the Nayaka period; the long view of the Tamil temple over 1,400 years shows many moments of remaking of the temple's centre, the shrine to the presiding deity. Such a practice is clear from the better-studied temples of the ninth and tenth centuries in the Chola region of the Kaveri delta. It is often stated that the earliest temples in South Asia were built in perishable materials, such as wood or brick, and were in some cases later replaced in more durable and higher-status stone. This is evident from the depiction of wooden architecture in stone in the forms of Buddhist stupas at Sanchi and Amaravati from the first century and the depiction of architecture in the railings and gateways' reliefs, for example, or the stone temples of coastal Kanara in the sixteenth century that are clearly based on wooden models. One of the earliest known stone monuments in the Tamil region is the rock-cut cave at Mandagappattu sixty kilometres south of Kancipuram dated to the late sixth century. The inscription produced on its foundation is often cited as inaugurating the stone tradition of architecture in the region. In this Mahendra Pallava I (c. 580–630) proclaims:

The seventh-century rathas at Mamallapuram, the well-studied location of the earliest substantial body of architecture and sculpture in Tamilnadu, have been understood as monolithic renderings of the brick, thatch and timber temples that have not survived. That more substantial structures continued to be built of brick in the Tamil region for many centuries after the widespread introduction of stone is clear from inscriptions from at least as early as the tenth century that mention the degradation and conversion of the brick structure to a stone one.

Rebuilding may be evident from the stone fabric itself and from the location and content of inscriptions on the temple. The modification of a temple may be suggested from the surviving fabric, for example, when the immoveable liṅga is placed at a lower level than usual within later enclosing walls, or the lowest base mouldings rise above the praṇāla, the waterspout that channels ablutions over the liṅga outside. Sometimes inscriptions were recopied or the earlier stone slabs reused in the later, renovated, temples. A few inscriptions are even explicit about the re-engraving of these land grants and endowments. Among the many examples that could be cited from the late ninth and tenth centuries are several temples rebuilt under the patronage of the great Chola queen Sembiyan Mahādevī (active 941–1001), better known for her patronage of over ten new temples such as at the eponymous village named after her, at Konerirajapuram and at Vriddachalam.Footnote ²⁵ Before the erection of the stone Tirukkotiśvara temple at Tirukkodikaval near Kumbakonam, the inscribed stone slabs of the earlier endowments that were found scattered all over the place were, on Sembiyan Mahādevī's orders, to be re-engraved on the walls of the newly built temple according to an inscription dated to Uttama Chola year 11 (c. 980). The old inscriptions were then discarded as useless.Footnote ²⁶ The Āpatsahāyēśvara temple at nearby Aduturai was similarly rebuilt in stone by Sembiyan Mahādevī in Uttama Chola year 16 (c. 985) according to an inscription which states that the previous grants to the temple were re-engraved.Footnote ²⁷ This renovation from brick to stone continued into the twelfth century, and no doubt later as inscriptions suggest, to the extent that very few wholly brick temples from the seventh to twelfth centuries remain, usually under thick layers of plaster as, for example, at the early eighth-century Pallava Tiruvīraṭṭāneśvara temple at Tiruvadigai.Footnote ²⁸ An inscription dated to the fortieth year of Kulōttunga I (c. 1110) on the south wall of the maṇḍapa of the Bilvanātheśvara temple at Tiruvaikavur, ten miles north-west of Kumbakonam, clearly states that the vimāṇa and ardhamaṇḍapa of the old brick temple was dilapidated and removed, and then replaced in stone.Footnote ²⁹ Brick was used as a building material until recently, but usually for the superstructure of vimāṇas and gopuras and not their base and walls. A notable exception are the temples built in the Kaveri region in the Maratha-period from the eighteenth to early nineteenth centuries which make extensive use of brick for walls and indeed vaulted roofs.

In other examples, rebuilding can be assumed to have taken place because older inscriptions appear re-engraved on a new building; in some instances it states that the inscriptions have been recopied, but not all. The Vaidyanātha temple at Tirumalavadi, a late ninth-century temple dating to the reign of Āditya I (r. 871–907), was renovated in stone in c. 1013 under royal order with the copying of the old inscriptions.Footnote ³⁰ Further repairs and the re-engraving of old inscriptions was mentioned again in 1026 under Rājendra I.Footnote ³¹ The Vaikuṇṭha Perumāḷ temple (Aḻakiya Pallava Vinnakar) at Tiruvennainallur was similarly pulled down and renovated in Kōpperunjinga year 11 (1254) and on this occasion earlier records dated to Kulōttunga III year 29 (1207), Rājarāja II year 12 (1158) and Kulōttunga III year 35 (1213) were re-engraved on the north and south walls of the vimāṇa.Footnote ³² At the Śivakkoḻundiśvara temple at Sivankudal, however, mixed and damaged inscriptions dated across Kulōttunga I's long reign from c. 1070 to 1120 on the south and west walls of the vimāṇa Footnote ³³ and one dated to the third year of Rājendra I (c. 1015/17) on a columnFootnote ³⁴ are located alongside three intact and ordered inscriptions of Vikrama Chola years 9, 10, 11 (r. 1118–35) all on the same vimāṇa's walls, suggesting that an eleventh-century temple was rebuilt c. 1120–28.Footnote ³⁵

In other instances rebuilding of the central shrine has clearly taken place because older inscriptions remain on the new temple – not on the main shrine but on, for example, the adjoining maṇḍapa, the first gopura or prākāra wall. This means that the central shrine may not be the oldest structure in the temple as a whole, a point emphasized by the potential stylistic discontinuity between newer vimāṇa and older ardhamaṇḍapa. At the Tiruccadaimudi temple at Tiruchennampundi, Pallava inscriptions dated to the reign of Nandivarman III (c. 846–69) on the entrance and columns are located before a later vimāṇa dated to the reign of Parāntaka I (907–55), suggesting that an earlier (perhaps brick) shrine has been replaced but leaving the adjoining structures in front.Footnote ³⁶ The main shrine of the Bhaktavatsala temple at Tirukkalukundram was rebuilt in the late sixteenth or seventeenth century, as its design suggests, but replaced an earlier structure as indicated by misplaced inscriptions dating to the early twelfth and thirteenth centuries reinserted into the central shrine's walls.Footnote ³⁷ Inscriptions on a completely different temple may also suggest the earlier presence of a temple on an adjacent site that has since been entirely replaced, such as at Srimushnam discussed above.

This renovation of temples from the tenth to twelfth centuries took place in the context of the expansion of Chola authority over the wider Tamil region, and resulted in the sacred landscape of the Kaveri delta, celebrated in the poetry of the Āḻvārs and Nāyaṉmār, acquiring firm, stone foundations. Given that we have few details of the brick temples that preceded the stone ones, their potentially innovative character remains in terms of the material alone. In the later Nayaka period stone undoubtedly replaced stone, but on this occasion there is clear evidence for the updating of the design: sixteenth–seventeenth century renovations do not replicate the design of the buildings they replaced. Unlike in the earlier era, inscriptions on the replacement shrines built in the Nayaka period rarely draw attention to the “new” building.Footnote ³⁸ Only occasionally are renovations indicated in inscriptions from the fifteenth to seventeenth centuries: at Alagarkoyil, for example, the main shrine was renovated “from upāna to stūpi” (bottom to top) in 1464, but in this instance the renovation does not appear to have radically transformed the eleventh-century building with its unusual circular vimāṇa, or the fifteenth-century renovation may be deliberately archaic in design.Footnote ³⁹ Even if the visually-adept modern observer may recognize distinct period traits, innovation in the architectural tradition is cloaked not celebrated.Footnote ⁴⁰

“Pious vandalism”: remaking the Tamil temple in colonial south India

The remaking of Tamil temples, not just by adding new structures to the periphery but by the wholesale replacement of the main shrine, has continued into the more recent past. One striking era of temple construction and renovation was between the 1870s and 1920s when many of the major Śaiva pilgrimage temples of the Tamil region received the architectural patronage of the Nakarattar (Nattukkottai Chettiar) community. These included the five “elemental” temples to the liṅgas of air, earth, fire, water and ether (ākāśa) – the Kāḷahastīśvara at Kalahasti, the Ēkambareśvara in Kancipuram, the Aruṇācaleśvara at Tiruvannamalai, the Jambukeśvara at Tiruvannaikka on Srirangam island and the Naṭarāja at Chidambaram – in addition to the Mīnākṣī-Sundareśvara in Madurai, the Kuṟṟalanātha at Kuttralam, and the Rāmaliṅgeśvara at Ramesvaram. At least thirty further Śaiva temples from across the region were also renovated in this period, especially if they were a pāṭal peṟṟa talam, “a place sung by the saints”.Footnote ⁴¹ At some of these temples, Chettiar patronage resulted in new corridors or maṇḍapas within or enclosing existing structures, but at many of these sites the existing shrines, halls, corridors and walls, up to a thousand years old, were wholly replaced leaving no trace of the earlier temple.

Their extensive patronage was noted by the colonial authorities at the time – and often deplored, with the different architectural priorities and notions of an archaeological heritage, that had been developing both in Europe and in India from the middle of the nineteenth century. The annual reports of the officers of the Archaeological Survey in Madras Presidency repeatedly show concerns over the renovation of temples by the Nattukkottai Chettiars. In 1902, the Government epigraphist Eugene Hultzsch wrote that: “In the course of these ‘repairs’ they have totally destroyed the[se] … shrines with every one of their inscriptions. … What the Musalmans did not destroy is being demolished by pious Hindus!”Footnote ⁴² In 1905, Hultzsch's colleague Alexander Rea wrote in his annual archaeological report: “Has it even occurred to those who wish to … reconstruct a new temple with new materials, and new associations, and a depreciation of the sacredness which attached itself to the old building … [that this sanctity will need] … to be reacquired, by the newly erected one?”Footnote ⁴³ John Marshall, the Director General of the Archaeological Survey of India, remarked in 1908 that “the ruthless demolition of ancient and historic shrines in Southern India, at the hands of the Nāttukōttai Chetties, is a subject which has been alluded to more than once in these reports. Short of taking new powers by legislation to interfere in the matter, the Government has done all it could to put a stop to the evil, but so far its efforts have borne little fruit…”, and he goes on to report the destruction of three more temples that year.Footnote ⁴⁴ Jean-Philipe Vogel, the Acting Director-General in 1911, expressed a similar if moderately understanding sentiment of the Nattukkottai Chettiars' renovations, describing them as “pious vandalism”.Footnote ⁴⁵

Two examples of Nakarattar-sponsored temple renovations will illustrate the modern continuity of the practice of vimāṇas being entirely replaced. Just under 150 inscriptions dating from before 1300 have been recorded from the Mahāliṅgasvāmi temple at Tiruvidaimarudur near Kumbakonam at the heart of the Kaveri river delta, a considerable number in comparison with many other contemporary temples testifying to its early importance (Figure 7).Footnote ⁴⁶ The earliest dates to around 911 (Parakēsarivarman (Parāntaka I) year 4) and two-thirds of those recorded are on the vimāṇa or its attached maṇḍapas dating from between the early tenth and early twelfth centuries. A further forty are on the first or second prākāra walls, the earliest dating to c. 1096. Only three are recorded from the gopuras: one on the second gopura on the east dated to c. 1223, and another on the third east gopura dated to c. 1286.Footnote ⁴⁷ All this suggests to the historically-minded visitor that the temple is a tenth-century Chola foundation expanded over the course of the following centuries into a vast temple complex with three enclosures entered from the east through three aligned gopuras, and with a goddess shrine in its own prākāra on the south. It suggests that this temple would be a good comparison with the similarly late Chola-period Kampahareśvara temple at Tribhuvanam a mile away.

Figure 7. Exterior of first prākāra wall, Mahāliṅgasvāmi temple at Tiruvidaimarudur

However, those seeking the tenth-century, early Chola-period shrine and its many inscriptions at the temple's heart will be disappointed to find that the core of the temple was built from around 1907. The entire first prākāra was demolished under Nakarattar patronage from December 1906 and there is no trace on the new building of the inscriptions copied in 1894 and then, as a matter of urgency, between June 1907 and July 1908.Footnote ⁴⁸ In contrast to the earlier recopying of endowments onto the new walls of the temple, the long legal history of the temple's – the deity's – possessions were discarded. But this was not the first time the temple had been subject to such radical renovation. In around 973 the Mahāliṅgasvāmi temple was rebuilt in stone, replacing an earlier brick structure from the late ninth century.Footnote ⁴⁹ According to this earlier dated inscription, all inscribed stones relating to endowments and kept in underground cellars prior to the rebuilding of the temple in stone were to be re-engraved on the walls of the new temple. Those recorded a century ago as dating before 973 are from this earlier brick temple. Fortunately, ten photographs of the late tenth-century stone building were taken by the officers of the Archaeological Survey before the temple's demolition, and show the temple's design before the total renovation in the years up to 1907 (Figure 8).Footnote ⁵⁰ Although it is around 930 years old there is nothing to suggest that the temple was so dilapidated that it required total replacement; around one hundred ninth- and tenth-century temples have survived in the Kaveri region.

Figure 8. Tenth-century vimāṇa of the Mahāliṅgasvāmi temple until the renovation from 1907 (© British Library Board, Photo 1008/9 (2551))

The entire inner of the temple – the most sacred part – was replaced. The new vimāṇa and its attached maṇḍapas are very similar to sixteenth- or seventeenth-century Nayaka design with crisp even mouldings; but the high polish to the stone and the quite naturalistic reliefs of flowers in pots on the columns indicate the later date. The entire inner prākāra is roofed over, in contrast to the earlier building; composite columns on a raised platform line a corridor for circumambulation all round. For some reason the new inner prākāra is smaller than before, leaving a gap between the first gopura on the east and the doorway into the darker, enclosed interior. For pioneering art historians of the Chola period, such as Douglas Barrett in the 1970s, the tenth-century modification of this sacred site is renovation – in the twentieth century the same practice is destruction “by misguided patronage”.Footnote ⁵¹

Whilst the ASI were hastily recording the inscriptions from the walls of the Śiva temple at Tiruvidaimarudur, they were alarmed to hear of another planned renovation of a major Chola temple fifteen miles away at Mayuram (Mayavaram). A modest sixteen inscriptions were recorded and four photographs taken from the Mayuranātha (Śiva) temple in 1907 and 1911 (Figure 9), before the inner sections of the temple were totally replaced. Though less informative than the substantial epigraphic corpus from Tiruvidaimarudur, there is sufficient data to indicate that the foundation of the temple was again in the tenth century, with a prākāra wall added by the late twelfth century and a separate shrine for the goddess Tirupaḷḷiyaṟai Nācciyar by 1229.Footnote ⁵² But almost all of this was swept away in the years prior to the mahākumbabiṣekam (reconsecration) in 1928, an event recorded in an inscription on the new first prākāra's gopura. The temple's design is very similar to the “neo-Nayaka” main shrine at Tiruvidaimarudur – there is clear stylistic uniformity amongst the Chettiar-sponsored temples dating to the early twentieth century – and one recent author has mistakenly attributed this temple to the Nayaka period on the basis of its design.Footnote ⁵³

Figure 9. Mayuranātha temple, Mayuram

At Mayuram the textual connection with the temple's past may have been erased with the discarding of the inscriptions, but the temple's sculpture maintained a material connection with the earlier histories of the temple and the sacred site. Just before the entrance to the main Mayuranātha shrine, and also before the goddess shrine to Apayāmpikai in its own prākāra to the north, are two pairs of standing portrait images, each about 130 cm high, of the temple's two early twentieth-century patrons. Inscriptions beneath the pair before the goddess shrine identify them as Al. Vīravīrappa Chettiar, who is joined by a smaller figure of his wife, and their son Al. Vīrapetta Chettiar. Like the temple's architectural design, these sculptures hark back to a visual model of approximately life-size three-dimensional images of donors, depicted standing with their hands together in añjalimudrā established as a significant feature of Nayaka-period architectural sculpture in the Tamil region.Footnote ⁵⁴

Furthermore, the earlier temple's complete set of nine niche images and two huge dvārapālas are also reused rather than discarded, and have been inserted into the new fabric, presumably in the same places. Douglas Barrett considers them to be amongst the masterpieces of stone sculpture from the late tenth century, and compares them with the celebrated bronzes from Konerirajapuram dated c. 969–977.Footnote ⁵⁵ The iconographic programme is typical for temples built in the Kaveri region in the late tenth century, with three niches (devakoṣṭha), rather than one, on either side of the ardhamaṇḍapa which, together with the three on the vimāṇa make nine in all. The sequence, clockwise, is: south side of the ardhamaṇḍapa, Gaṇeśa, Naṭarāja (Ānandatāṇḍava), standing Śiva embracing Uma (Āliṅganacandraśekhara); clockwise on the vimāṇa are Dakṣināmūrti, standing Śiva and Brahmā; and on the north side of the ardhamaṇḍapa are Gaṅgādhara, Durgā, Bhikṣātana (Figure 10).Footnote ⁵⁶ Each image is between 130 and 150 cm in height and they are clearly stylistically distinct from other modern temple sculpture. The visual disjunction is also evident from the size of the niches, which are slightly wider than the original tenth-century ones; the whole temple is probably modestly larger in plan.

Figure 10. Niche sculpture of Naṭarāja from the Mayuranātha temple, Mayuram

At nearby Tiruvidaimarudur, and at most other Chettiar renovations, the surviving niche images seem to have been discarded. The reuse of the sculpture at Mayuram is unusual, for many modern temples simply omit such images and have empty niches, perhaps from a sense of economy rather than any lack of skill in the production of figural sculpture.Footnote ⁵⁷ The magnificent series of twenty-five sculptures of deities on columns all around the Nandi, the balipīṭha and the dhvajastambha in Madurai's Mīnākṣi-Sundareśvara temple all date to the 1870s, part of the Nattukkottai Chettiar-sponsored renovation of the temple before the 1878 mahākumbabiṣekam, are good examples of the high quality of this period's sculpture. Upon visiting Trichy in 1908, John Marshall remarked of the Chettiar renovations being carried out in Uraiyur that “The sculptured pillars and stones are of excellent workmanship, far superior to much of the older work, and they show that the old art has never entirely died out. It is certainly evident that much can still be done with the stone workers here, if they are properly supervised”.Footnote ⁵⁸

Early twentieth-century Tamil temple renovations may then have indirectly led not only to the institutional collection of inscriptions as a matter of urgency before they disappeared, but also to a growing supply of sculpture for local or international museums – or indeed gallery salerooms – as interest in the medieval sculpture of south India grew in precisely this period.Footnote ⁵⁹ Inscriptions were the bedrock on which historians of pre-modern south India in the colonial period established dynastic chronologies of south India. The renovations that led to the rapid recording of so many inscriptions on stone walls also threw up other finds. In 1904, repairs to the Vaṭāraṇyeśvara temple at the village of Tiruvalangadu initiated once again by Nattukkottai Chettiars resulted in the discovery of thirty-one massive copper-plates weighing over 200 pounds, which gave a more detailed and exhaustive genealogy of the Cholas up to the reign of Rājendra than previously established.Footnote ⁶⁰

In 1908 the Kūṭal Aḻakar temple approached at the outset was once again subject to substantive renovation, when the goddess shrine was entirely demolished and subsequently rebuilt; two other nearby temples in Madurai, the Madana Gopālaswami and Immaiyil Nanmaitaruvar, were also renovated in the same decade.Footnote ⁶¹ The only Nayaka temple structure outside of Tamilnadu today is in the Philadelphia Museum of Art, purchased in Madurai in 1912 and acquired by the museum in 1919, an open maṇḍapa with magnificent architectural sculpture from the recently renovated Madana Gōpālaswami temple.Footnote ⁶² The renovation of these temples in the early twentieth century continued a process of architectural change and remaking of the sacred site of Madurai that spreads over a thousand years.

Landscape, history and the Tamil temple

The repeated replacement of temples' vimāṇas over the course of the past thousand years can be seen as one more episode in the history of a site, from the transformation of a sacred stone, tree or megalith into a temple of brick, timber or thatch, and its recreation in stone – once or perhaps several times. The renovation in the tenth century and later that resulted in the creation of a stone Chola-period temple and the loss of the ancient brick one may be considered acceptable historic practice to some scholars, while the modern replication of such temple replacement is viewed as objectionable destruction. The expectation of those responsible for building and using the temple would appear to be that it is not a fixed and finished structure, but one that will need to be added to, modified and constantly re-made. The Tamil term for such renovations is tiruppaṇi, literally “sacred work”, with no suggestion of rebuilding or radical change, though there has been a varying degree of self-consciousness or expression of these processes.Footnote ⁶³ What is clear from an examination of the material evidence is that some architectural historians and archaeologists have underestimated the degree to which many Tamil temple sites have been completely remodelled by their own worshipping communities. Samuel K. Parker's study of the many “unfinished” monuments at Mamallapuram is important in this regard, for he notes that describing a temple in such terms suggests a simple binary with “finished”. Instead he suggests that there are degrees of completeness suggesting “the continuous unfolding of the temple's (and deities') potential for further physical manifestations”.Footnote ⁶⁴ The creation of temples and images, he argues, is an extended act of sacrifice, with the remnants of sacrifices being the empowered, creative seeds of further developments. “Temporally, the same creative logic that locates generative power in the remnant of the sacrifice is also expressed in the indefinitely incomplete structure of temple complexes that continue to grow, decay, be renewed, and grow again, from one generation to the next”.Footnote ⁶⁵

But what became of the consecrated icon within the vimāṇas that were repeatedly being remade? Temple icons are known to have been stolen or looted, as Richard Davis has notably discussed, but may too have needed replacing when worn out, broken or fallen down; Davis cites medieval Śaiva texts such as Somaśambhupaddhati and Mayamata as listing the ways in which icons might become degraded and be in need of ritual recovery.Footnote ⁶⁶ The radical renovations of temples may have taken place when an icon was in need of such reconsecration. But it is also evident from the architecture itself that a new vimāṇa might be built around an established and immoveable icon. The later fabric around an icon and pīṭha sunk below floor level testifies to this, as seen for example at the Paraśurāmēśvara temple at Gudimallam with the famous five foot high first-century bceliṅga reached down steps within a temple rebuilt in stone in the early twelfth century.Footnote ⁶⁷ Some of the Nakarattar renovations of Śaiva temples appear to have retained the existing icons unmoved. When Marshall visited the new temple being built at Uraiyur in 1908 he remarked that: “The only apparent portion of the ancient building left in position is the linga of the central shrine, which is enclosed in a sort of sentry box”.Footnote ⁶⁸ It was the icon's sanctity that was critical to the devout rather than the shrine enclosing it, as the colonial officials imagined.

There is therefore no need to explain temple renovations as a consequence of the destruction wrought by an aggressive external agency, often understood to be Muslims and historicized to the troubled long fourteenth century. Temples were certainly looted and sacked, a common Indic political practice, but the evidence for widespread aggressive temple destruction is not forthcoming in south India. Where temples have been destroyed, the culprits may be Portuguese or Dutch as much as “Turk” or Muslim, and the material traces of such destruction have long been swept away by the “pious vandals”, the Nattukkottai Chettiars, and others in the early twentieth century. The absence of any early temple remains in the Chennai suburb of Mylapore, the site of the Kapalīśvara temple, may be explained with reference to the Portuguese occupation and possible destruction of the site in the sixteenth and seventeenth centuries, but the temple was thoroughly renovated around 1900 with the east gopura completed in 1906.Footnote ⁶⁹ The Dutch raid and occupation of the Skanda (Murukan) seashore temple at Tirucendur in 1646–48 may have similarly resulted in the destruction of the early sandstone temple suggested by the presence of ninth-century Pandyan inscriptions.Footnote ⁷⁰ Anything that had survived this destructive Dutch raid would have been swept away by the combined action of a further three centuries of coastal erosion and the rebuilding of the temple on a massive scale between 1848 and the temple's reconsecration in 1941 under the leadership of a group of three ascetics.

Architectural historians are well aware that buildings made from ephemeral materials are missing from the archaeological record, but it is also evident that seemingly permanent stone structures have similarly been swept away by their own worshipping communities, and not only in the recent past. Striking too is the volume of brick temple architecture that has disappeared from the Tamil region. Sembiyan Mahādevī is lauded for the patronage of such wonderful stone temples in the tenth century with amongst the finest stone and bronze sculpture created, but rarely is she criticized for destroying brick temples. The absence of brick temple architecture in the Tamil region is all the more remarkable when compared with contemporary South-East Asia, where fine brick temples with detailed ornament have survived in great numbers in Cambodia and Champa (coastal Vietnam) from as early as the sixth through to the ninth centuries.

The sustained Tamil impetus to wholesale temple renovation may also be partly responsible for the limited number of surviving structural monuments from the far south in Pandyanadu before the twelfth, or even sixteenth, centuries. The region is well known for the many substantial rock-cut caves with monumental sculpture, but though ruled over by the Pandyans from their capital at Madurai from the sixth to the early fourteenth centuries as contemporaries of the Cholas there are very few surviving structural temples from this period in Pandyanadu compared with the Kaveri region. No temple remains in Madurai predate the sixteenth century and in the surrounding area to the east and south, the number of surviving structural temples, especially major temple complexes with multiple gopuras, dating from before the twelfth century are few and far between. Pandyanadu has long been more sparsely populated than the well-irrigated and agriculturally fertile Kaveri delta. There are also fewer of the widely known sacred sites in Pandyanadu: 190 of the 269 shrines celebrated in the Śaiva poetry of Appar, Cundarar and Campantar are in the Chola country compared with fourteen in Pandyanadu;Footnote ⁷¹ of the ninety-five Vaiṣṇava divyadeśas that are in Tamilnadu, only twenty are in Pandyanadu. One might also argue that the brunt of the destruction wrought by the fourteenth-century Sultanate was on southern Tamilnadu, given that the short-lived capital was in Madurai or nearby.Footnote ⁷² But an additional factor may have been that the lesser-known temples in the southern districts of the Tamil country were more substantially remade in the period from the sixteenth to the nineteenth centuries, during the rule of the Madurai Nayakas and their Maravar successors in Sivagangai and Ramnad, in comparison with their counterparts further north in the expanding sphere of the Marathas and then the Nawab of Arcot. The Nattukkottai Chettiars also built new temples in their Chettinadu homeland south of Pudukkottai in this period – both additions to much earlier cave-temples, as at Pillaiyarpatti, and complete replacements for others.

The rebuilding and expansion of a temple also emphasizes the importance of place, the site on which the temple is built: the site is sacred, not the architecture. This is clear from the examples of temples on old sacred sites that were wholly replaced in the Nayaka period in the highly elaborate form described above. The importance of place is evident in the site-specific names of deities: Raṅganātha is not simply a form of Viṣṇu but Viṣṇu on Srirangam Island. As Leslie Orr has noted, contemporary inscriptions often simply name male deities as the “Lord of such-and-such a place”, with no clear sectarian reference to either Śiva or Viṣṇu (or indeed a Jain Tīrthaṅkara or a Buddha).Footnote ⁷³ At Srimushnam, for example, Viṣṇu as Bhū-Varāha is named Tirumuṭṭamuṭaiya-nāyaṉār (“the Lord of Tirumuṭṭam”) in inscriptions. Contemporary literature also bears witness to a growing awareness of the importance of individual sacred sites. The Nayaka period was marked by political instability and substantial temple construction, both in numbers and size. An element of this expansion was the composition of site-related mythological literature, the sthalapurāṇas. David Shulman has noted that Śaiva sthalapurāṇas typically attempt to elevate the particular site to great antiquity and mythological importance.Footnote ⁷⁴ The sacred site is dehistoricized in this literature and set in mythological time; rarely is there any mention of architecture. Though undoubtedly preceded by oral literature and now lost texts, the bulk of the sthalapurāṇas known today were written down exactly when so many temples were being expanded or founded, in the sixteenth and seventeenth centuries.Footnote ⁷⁵

From a pan-Indian perspective, repeated renovations of sacred sites have resulted in little early archaeological material, despite their attested antiquity and importance. At Varanasi, for example, no substantial temple dating before the late eighteenth century remains intact without substantive reconstruction or modification. Even if much of the absence of early Hindu temple architecture in Varanasi may be a consequence of its location in the region of Sultanate and then Mughal rule, it is clear that the site has remained sacred throughout the repeated remaking of the temples that define the sacred landscape of the city. Similarly striking is the absence of any archaeological trace at some Jain sites of architecture or inscriptions pre-dating the nineteenth century, such as Sammeda in Bihar where twenty of the twenty-four tīrthaṅkaras attained liberation, despite the great antiquity of these sacred places in Jain literature.

The issue of temple renovation in Tamilnadu draws attention to the differing perspectives of scholars and historians, and the worshipping communities of the temples themselves over the millennia of their sacred lives. The scholarly tendency to privilege the “original” fabric, the temple's founding, the artists' or patrons' intention, and the academic taste for the oldest stone monuments contrasts with the “living” character of these sacred monuments. The clash between the temple as monument and the temple as a sacred site was an acutely felt issue in the late nineteenth and early twentieth centuries at the outset of detailed archaeological surveys of South India's monuments, and coincided with numerous temple “renovations” by the Nakarattar community. The European conception of “sacred architecture” as inviolate came up against the local stress on the devotional value of the sacred Tamil land itself and not the “stone books” of the past built upon it for over a thousand years.