Academia.eduAcademia.edu
21 Journal of Southeast Asian Studies, 34 (1), pp 21-39 February 2003. Printed in the United Kingdom. © 2003 The National University of Singapore As in Heaven, So on Earth: The Politics of Vi≥◊u, ∆iva and Harihara Images in Preangkorian Khmer Civilisation Paul A. Lavy Analysis of the earliest sculpture and epigraphy of Southeast Asia reveals contrasting geographic patterns regarding the worship of Hindu deities. During the seventh century, efforts to consolidate political authority by Khmer rulers led to the deployment of Harihara, a god that embodied multiple conceptions of power and could serve as a ready statement of political and religious unification. The mitred four-armed Vi≤§u images of early Southeast Asia have been the focus of a number of scholarly studies.1 Most of these have dealt primarily with chronology and stylistic and iconographic relationships with Indian art rather than the cultural context of the images themselves. In a recent article that attempts to address all of these issues, Nadine Dalsheimer and Pierre-Yves Manguin argue that the popularity of Vi≤§u in ancient Southeast Asia – and particularly in Preangkorian Khmer civilisation (generally defined as pre-ninth-century CE) – was due to the role played by Vai≤§avism in Southeast Asian trading networks.2 Their emphasis on trade as an explanation for the diffusion of Indian religions and art into Southeast Asia is a restatement of long-asserted arguments that are open to considerable doubt. While trade may partially account for the distribution of the mitred Vi≤§u images in the coastal areas of Southeast Asia, it does not adequately explain the popularity of these images, the significance of Vi≤§u in Southeast Paul Lavy is a Ph.D. candidate in Art History at the University of California at Los Angeles and Instructor of Asian Art History at the Pennsylvania State University, University Park. His e-mail contact is plavy@hotmail.com A draft of this article was originally presented at ‘Crossroads and Commodification: A Symposium on Southeast Asian Art History’, University of Michigan, 25-6 March 2000. I would like to thank Robert L. Brown for the valuable criticism he provided over the course of several versions of this paper; Ketkanda Jaturongkachoke for assistance with Thai-language sources and for her generous hospitality, support and friendship while I was conducting fieldwork in Thailand; and Kanika Mak for her insightful critiques. I would also like to acknowledge the two anonymous readers for their helpful and encouraging comments. 1 The most extensive and important study of the early Vi≤§u images is Stanley J. O’Connor, Hindu gods of peninsular Siam (Ascona: Artibus Asiae, 1972). Other important studies include Pierre Dupont,‘Vi≤§u mitrés de l’Indochine occidentale’, Bulletin de l’École Française d’Extrême-Orient [henceforth BEFEO], 41 (1941): 233-54; Jean Boisselier, ‘Le Vi≤§u de Tjibuaja (Java Occidental) et la statuaire du Sud-Est Asiatique’, Artibus Asiae, 22, 3 (1959): 210-26; idem., The heritage of Thai sculpture (New York: Weatherhill, 1975), pp. 71, 97101; and Robert L. Brown, ‘Indian art transformed: The earliest sculptural styles of Southeast Asia’, in Panels of the VIIth World Sanskrit Conference, ed. Johannes Bronkhorst (Leiden: Brill, 1992), pp. 40-53. 2 Nadine Dalsheimer and Pierre-Yves Manguin, ‘Vi≤§u mitrés et réseaux marchands en Asie du Sud-Est: Nouvelles données archéologiques sur le Ier millénaire apr. J.-C.’, BEFEO, 85 (1998): 87-116. 22  .  Asian cultures or the fact that the tradition persisted among the Khmer for many centuries. This article argues that the popularity of Vi≤§u and other Brahmanical deities was linked to patterns of political authority and that the Southeast Asian ruling elite, whether kings or chiefs, utilised images of the gods with these considerations in mind.3 The deities Vi≤§u and ∫iva embodied two different conceptions of sovereignty (or leadership in general), and images of these deities were employed to exploit these contrasting notions according to location and styles of rule. These practices are perhaps best understood through analysis of a third case: Harihara, a composite deity generally characterised in ancient Indian and Khmer art by a strict bilateral division between the proper left side with the attributes of Vi≤§u (Hari) and the proper right with the attributes of ∫iva (Hara).4 Khmer sculptures of Harihara invariably have four arms and a vertical demarcation of the head into two ‘half-faces’ so that the right side of the head is piled high with ∫iva’s elaborately tangled locks (ja∑¡muku∑a) and the left side is covered by Vi≤§u’s tall cylindrical mitre (kir#tamuku∑a). By unifying ∫iva and Vi≤§u in one anthropomorphic form, Khmer images of Harihara served as a divine analogue for the concentration of the two forms of royal power. Harihara is commonly interpreted, however, as a syncretic deity that brought about the rapprochement of two allegedly ‘rival’ Hindu sects, ∫aivism and Vai≤§avism.5 This explanation is over-simplistic and dubious, particularly if applied to the ancient Khmer; there is no evidence from the Preangkorian period, for example, to indicate hostilities or competition between various ‘exclusive’ sects of Hinduism.6 The popularity of Harihara 3 Michael Vickery has argued that ‘kings in fact play a relatively small role in the pre-Angkor corpus, which is dominated by other ruling class figures’. According to Vickery, approximately 65 Preangkorian inscriptions, of the total corpus of about 140, include a name that is usually associated with a king or queen. He argues that at least until the reign of Jayavarman I (ca. 657-81?), so-called Khmer ‘kings’ may more accurately be referred to as ‘chiefs’ and that they may have held primarily ritual rather than political functions; Michael Vickery, Society, economics, and politics in Pre-Angkor Cambodia: The 7th–8th centuries (Tokyo: The Centre for East Asian Cultural Studies for Unesco, the Toyo Bunko, 1998), pp. 84-92, 177-89, 321-4, 366-9; quotation from p. 322. 4 The composite deity commonly referred to as ‘Harihara’ is mentioned in Preangkorian Sanskrit and Khmer inscriptions under a variety of epithets including ∫a¶kara-N¡r¡ya§a, ∫a¶kara-Acyuta, Hari∫a¶kara and ∫ambhu-Vi≤§u. In Sanskrit ‘Hara’ is an epithet of ∫iva that can be translated as ‘Seizer’ or ‘Destroyer’. ‘Hari’, an epithet for an aspect of Vi≤§u (but in the ancient Indian context also used for other gods, including ∫iva and Indra) means ‘yellow, reddish brown, or green’, though as Gösta Liebert points out, it is often ‘hardly correctly’ translated as ‘the remover of sorrow’ or ‘he who gives joy’; Gösta Liebert, Iconographic dictionary of the Indian religions: Hinduism – Buddhism – Jainism (Leiden: Brill, 1976), p. 100, and M. A. Dhaky, ‘Harihara in Cambodian inscriptions and hieratic art’, in Madhua: Recent researches in Indian archaeology and art history, ed. M. S. Nagaraja Rao (Delhi: Agam Kala Prakashan, 1981), pp. 263-6. 5 See, for example, Lawrence Palmer Briggs, ‘The syncretism of religions in Southeast Asia, especially in the Khmer Empire’, Journal of the American Oriental Society, 71, 4 (1951): 230-49; and Marguerite E. Adiceam, ‘Les images de ∫iva dans l’Inde du sud V. – Harihara’, Arts Asiatiques, 13 (1966): 83-98. A different view that emphasises Vi≤§u’s hierarchically inferior position on the proper left side of Harihara is in Raju Kalidos, ‘V¡mac¡ra Vi≤§u in Hindu iconography: A problem in sociological values’, East and West, 44, 2-4 (1994): 275-91. 6 There is also no evidence in the Preangkorian context to suggest an interest in an androgynous form of Harihara that united ∫iva in male form with a female aspect of Vi≤§u (Mohin#). The Mohin# myths that appear in the Mah¡bh¡rata and several of the Pur¡§as are often used, somewhat problematically, to explain the origin of Indian Harihara images; Wendy Doniger O’Flaherty, Women, androgynes, and other mythical beasts (Chicago: University of Chicago Press, 1980), pp. 326-30. In Preangkorian art, depictions of androgynous deities are rare. I am, moreover, unaware of any half-male/half-female images, particularly among early Indian or Southeast Asian sculpture, that can be identified as Harihara with any certainty. ≠Ω, ∞      23 in Preangkorian Cambodia seems instead to have been related to specific historical and political circumstances among the Khmer during the seventh and early eighth centuries. Rulers based in northern Cambodia, where the style of rule was linked to ∫iva, were trying to assert and/or maintain control over coastal areas to the south, where Vi≤§u had been the traditional symbol of royal power. These northern rulers consequently employed an icon that represented the union of both deities and the concurrent conceptions of authority represented by each, in order to symbolise and legitimise their own territorial and political aspirations. None of this is meant to minimise the importance of religious considerations in ancient mainland Southeast Asia; rather than placing religion at the service of politics, I intend to underscore the inseparability of the two in Preangkorian society and sovereignty. I agree with Melford Spiro, who has been critical of interpretations of ‘religion as almost exclusively an instrument in the political struggle for power and prestige.’ It is an historical fact, he argues, that religion can be used for political and economic gain, but manipulation of religion must be distinguished from actual religious behaviour and beliefs.7 The earliest evidence (fifth and sixth centuries) Throughout Southeast Asia, the earliest and most important archaeological and epigraphic evidence for Indian influence relates to the worship of Vi≤§u and his various avatars.8 From his first known appearance in Southeast Asia (ca. 450) in the Ci-arutön rock inscription, associated with the polity of T¡rum¡ in western Java, Vi≤§u was linked to kingship and to territorial control and expansion. The inscription, accompanied by a pair of carved footprints, refers to P∞rnavarman, the ruler of T¡rum¡, and compares his footprints to those of Vi≤§u (Vi≤§upada). In the Jambu rock inscription, also associated with T¡rum¡ and bearing similar footprints, P∞rnavarman, ‘the unequalled lord of men’, As Liebert points out, it is virtually impossible to prove that androgynous images represent Harihara and not other deities like Ardhan¡r#Ωvara, who combines ∫iva and ∫akti in one bodily form; Liebert, Iconographic dictionary, p. 101. For more on Mohin# and Ardhan¡r#Ωvara, see Raju Kalidos, ‘Vi≤§u’s Mohin# incarnation: An iconographical and sexological study’, East and West, 36, 1-3 (1986): 183-204; Ellen Goldberg, The lord who is half woman: Ardhan¡r#Ωvara in Indian and feminist perspective (Albany: State University of New York Press, 2002); and Kamaleswar Bhattacharya, Les religions brahmaniques dans l’ancien Cambodge d’après l’épigraphie et l’iconographie (Paris: École Française d’Extrême-Orient, 1961), pp. 92-3. 7 Melford Spiro, ‘Religion: Problems of definition and explanation’, in Anthropological approaches to the study of religion, ed. Michael Banton (London: Tavistock Publications, 1966), pp. 105-6. 8 The literature dealing with the nature and mechanisms of ‘Indianization’ in Southeast Asia is extensive. Some useful general and theoretical studies include: I. W. Mabbett, ‘The “Indianization” of Southeast Asia (I): Reflections on the prehistoric sources’, Journal of Southeast Asian Studies [henceforth JSEAS], 8, 1 (1977): 1-14; idem., ‘The “Indianisation” of Southeast Asia (II): Reflections on the historical sources’, JSEAS, 7, 2 (1977): 143-61; idem., ‘The “Indianization” of Mainland Southeast Asia: A reappraisal’, in Living a life in accord with Dhamma: Papers in honor of Professor Jean Boisselier on his eightieth birthday, ed. Natasha Eilenberg et al. (Bangkok: Silpakorn University, 1997), pp. 342-55; Adhir Chakravarti, ‘Indianization of South East Asia – A re-examination’, Journal of Ancient Indian History, 15, 1-2 (1985-6): 229-61; Hermann Kulke, ‘Indian colonies, Indianization or cultural convergence?: Reflections on the changing image of India’s role in South-East Asia’, in Onderzoek in Zuidoost-Azië: Agenda’s voor de jaren negentig, ed. Henk Schulte-Nordholt (Leiden: Vakgroep Talen en Culturen van Zuidoost-Azië en Oceanië, Rijksuniversiteit te Leiden, 1990), pp. 8-32; and Monica L. Smith, ‘“Indianization” from the Indian point of view: Trade and cultural contacts with Southeast Asia in the early first millennium C.E.’, Journal of the Economic and Social History of the Orient, 42, 1 (1999): 1-26. 24  .  is said to have resisted numerous foes; his footprints were ‘ever dexterous in destroying hostile towns’. Although Vi≤§u is not mentioned in the Jambu inscription, its similarity to the Ci-arutön text with its reference to Vi≤§upada clearly indicates that P∞rnavarman was employing the same symbolism in both cases, namely that the king’s/Vi≤§u's footprints were meant to indicate the imposition of the king’s authority. They were set up to literally mark out his realm, or at least the areas he sought to dominate.9 It is possible that these types of inscriptions indicate zones of conflict rather than demarcating regions under firm control. The reason for invoking Vi≤§u is nevertheless clear, as Vi≤§upada were often associated with the three strides by which he is said to have traversed and conquered the entire universe. The linking of royal authority with Vi≤§u and his footprints occurs again in the inscription of Tháp M‹·i (K.5), now in southern Vietnam, which Coedès dates to the second half of the fifth century on the basis of the form of the script.10 According to this inscription, Gu§avarman, the son of a king (perhaps of the polity known to us as ‘Funan’, probably located in the lower Mekong River delta and/or around the Gulf of Siam), controlled a ‘realm wrested from the mud’.11 It further commemorates the foundation of a sanctuary named Chakratirthasvamin which contained Vi≤§upada. Coedès contrasts the nature of the footprint-associated inscriptions of Gu§avarman and P∞rnavarman: ‘[w]hereas the footprints of P∞rnavarman in Java perhaps marked, as has been said, the 9 J. Ph. Vogel, ‘The earliest Sanskrit inscriptions of Java’, in Publicaties van den Oudheidkundige Dienst in Nederlandsch-Indië, 1 (1925): 15-35; quotation from p. 25. See also George Coedès, The Indianized states of Southeast Asia, ed. Walter F. Vella, trans. Susan Brown Cowing (Honolulu: University of Hawai‘i Press, 1968), pp. 53-4, W. F. Stutterheim, ‘De voetafdrukken van Purnawarman’, Bijdragen tot de Taal-, Land- en Volkenkunde van Nederlandsch-Indië, 89 (1932): 288; and Hariani Santiko, ‘The religion of King P∞rnavarman of T¡rum¡nagara’, in Fruits of inspiration: Studies in honour of Prof. J.G. de Casparis, ed. Marijke J. Klokke and Karel R. Van Kooij (Groningen: Egbert Forsten, 2001), pp. 423-33. 10 George Coedès, ‘Deux inscriptions sanskrites du Fou-nan’, BEFEO, 31 (1931): 1-8, and Coedès, Indianized states, p. 60. The ‘K’ designation refers to the official inventory of inscriptions associated with Cambodia (whether in Sanskrit or Khmer). 11 The Chinese terms ‘Funan’ and ‘Zhenla’ (or ‘Chenla’) are found in records of embassies and ‘tributebearing’ or ‘trade’ missions from polities in Southeast Asia to the Chinese imperial court. While these terms undoubtedly refer to actual polities and may be Chinese transliterations of ancient Southeast Asian place-names, it is virtually impossible with current knowledge to link them precisely to geographical locations or local toponyms. As Claude Jacques has argued, these terms should be abandoned in favour of locally attested place-names found in the inscriptions; Claude Jacques, ‘“Funan”, “Zhenla”: The reality concealed by these Chinese views of Indochina’, in Early South East Asia: Essays in archaeology, history and historical geography, ed. Ralph B. Smith and William Watson (New York: Oxford University Press, 1979), pp. 371-9. Charles Higham suggests, however, that terms like ‘Funan’ and ‘Zhenla’ can be useful, but only if they are ‘understood to mean an area’ and not specific polities; Charles Higham, The archaeology of mainland Southeast Asia: From 10,000 BCE to the fall of Angkor (Cambridge: Cambridge University Press, 1989), p. 255. The nature of these two polities has been the subject of much recent rethinking. In addition to the sources just mentioned, some important examples include Yoshiaki Ishizawa, ‘Chinese chronicles of 1st-5th-century AD Funan, southern Cambodia’, in South East Asia & China: Art, interaction & commerce, ed. Rosemary Scott and John Guy (London: University of London, Percival David Foundation of Chinese Art, 1995), pp. 11-31; Claude Jacques, ‘China and ancient Khmer history’, in the same volume, pp. 32-40; idem., ‘Le pays khmer avant Angkor’, Journal des Savants (Jan.–Feb. 1986): 59-95; idem., ‘Funan, Zhenla, Srîvijaya’, in Les apports de l’archéologie à la connaissance des anciens états en Thaïlande, 3e Symposium franco-thaï, 11-13 December 1995 (Bangkok: Silpakorn University, 1995), pp. 14-23; Michael Vickery, ‘What and where was Chenla?’, Recherches nouvelles sur le Cambodge, ed. François Bizot (Paris: École Française d’Extrême-Orient, 1994), pp. 197-212; and idem., Society, economics, and politics, pp. 18-23, 3347. ≠Ω, ∞      25 taking possession of a country after a military conquest, these prints of Vishnu mark a peaceful conquest, after drainage and partial raising of embankments...’ While this is an interesting suggestion, it should be noted that the Tugu rock inscription of western Java credits P∞rnavarman with the construction of two canals.12 Although the Tugu inscription contains no references to Vi≤§u or footprints, it indicates that Gu§avarman and P∞rnavarman viewed their authority in similar terms and associated the imposition of authority with the control of water through physical alteration of the lands in question. Another inscription (K.875), from the southern part of Takèo Province in Cambodia and also attributed by Coedès to the polity of Funan, refers to a queen named Kulaprabh¡vat# and her foundation of a hermitage apparently dedicated to Vi≤§u. Thus both of these inscriptions, from the coastal areas of southern Cambodia and Vietnam usually associated with the polity of Funan, invoke Vi≤§u in the context of the exercise of royal authority. The case of Kulaprabh¡vat#, who is identified in K.875 as a r¡jñ# (queen), reminds us that in ancient Southeast Asia women were able to wield political authority and participate in practices that served to express the ‘extraordinary qualities’ associated with ‘people of prowess’.13 Just as the Vi≤§u-oriented inscriptions are among the oldest extant epigraphy in Southeast Asia, the same deity accounts for the earliest known anthropomorphic images from the region. These four-armed mitred Vi≤§u images, most of which probably date from between the fifth and seventh (or perhaps eighth) centuries, have been found throughout coastal Southeast Asia: in peninsular Thailand, in western Java, at the site of Kota Kapur in the province of South Sumatra (island of Bangka), and in the Mekong River valley of present-day southern Cambodia and Vietnam. It is the latter group, consisting of over thirty sculptures, that is most relevant to this paper. In contrast, only one pre-ninth-century image of Vi≤§u, the Vi≤§u of Kompong Cham Kau (Stung Treng Province), is known to have come from northern Cambodia.14 The overwhelming prevalence of Vi≤§u-centred foundations in the southern Indochinese Peninsula during the fifth and sixth centuries was matched in contemporary northern Cambodia by dedications directed primarily towards ∫iva. During the Preangkorian period, he was depicted almost exclusively in the phallic form of the li¶ga. Unlike the common four-armed image of Vi≤§u, there is perhaps only a single 12 Quotation from Coedès, Indianized states, p. 60; Vogel, ‘Earliest Sanskrit inscriptions’, pp. 28-34. 13 Coedès, Indianized states, p. 60; idem., ‘A new inscription from Fu-nan’, Journal of the Greater India Society, 4 (1937): 117-21. On women as ‘people of prowess’, see O. W. Wolters, History, culture, and region in Southeast Asian perspectives, rev. edn (Ithaca: Cornell University Southeast Asia Program, 1999), pp. 169-70. 14 I am restricting this discussion to images published before the 1970s and the subsequent flooding of the antiquities market with fake and unprovenanced Khmer sculpture, including numerous depictions of Vi≤§u and Harihara. Aside from the important ethical issues involved with publishing such images that are almost certainly either looted or fake, unprovenanced images are of very limited use to a study such as this which depends to a great deal on the original geographic placement and distribution of sculpture and epigraphy. Although the knowledgeable French art dealer Jean-Michel Beurdeley estimated in 1992 that 80-90 per cent of the Khmer art that has come onto the market in recent years has been modern forgeries or fakes, I am not excluding the possibility that important genuine pieces of Preangkorian sculpture may have recently come to light and may now reside in museums or private collections outside of the well-known museum holdings in Cambodia, Thailand and France. 26  .  anthropomorphic image of ∫iva prior to the late ninth-century style of Preah Ko.15 Otherwise, ∫iva is only encountered on the right half of Harihara images or in the form of the li¶ga, both of which have a wide geographical distribution in Preangkorian Khmer art. Louis Malleret, in his archaeological study of the Mekong Delta, includes a number of ∫iva li¶gas from that region. He employs a typology in which the naturalistic li¶gas that more closely resemble human anatomy are placed in the Preangkorian period while the abstract or geometric li¶gas are assigned to the Angkorian period (post-802 CE). This is controversial, however, and, while a range of the fifth to the eighth or ninth centuries is reasonable, it is by no means certain. Even if some of the li¶gas have dates as early as Malleret suggests, the Vi≤§u images in the south far outnumber them. Early (fifth- and sixth-century) epigraphic references to ∫iva in the south are similarly lacking.16 Northern Cambodia, northeastern Thailand and southern Laos, on the other hand, provide clearer evidence of li¶ga dedications during the fifth and sixth centuries. A number of brief inscriptions dating to this period and usually attributed to the so-called ‘Dangrek [mountain range] Chieftains’, record the installation of li¶gas. The earliest of these inscriptions is probably that of Vat Luong Kau (K.365) in southern Laos, which Coedès dates to the second half of the fifth century.17 In addition, there are at least twelve sixth-century inscriptions (dated by paleography) associated with a chief/king named Citrasena (also known as Mahendravarman) and three others that mention Bhavavarman (I), all of which record ∫iva-oriented foundations – usually li¶gas but also images of the bull Nandin. Pierre Dupont noticed that the erection of these li¶gas coincided with the taking ‘possession of the soil’ and he suggested that the foundations were specifically intended to demarcate territorial control.18 Coedès, following Dupont, argued that ‘[s]ince these li¶gas and images were set up on the occasion of “conquest of the whole country”, we can conclude 15 In his magisterial study of Preangkorian sculpture, Dupont suggests three possibilities: a ‘∫iva’(?) head from Angkor Borei, the ‘∫iva’ of Kompong Cham Kau, and the ∫iva of Trapeang Phong found in Roluos near Angkor; only the identification of the last one is secure. The earliest Khmer images of ∫iva have two arms, as opposed to Vi≤§u’s four; Dupont, Statuaire préangkorienne, pp. 56-7, 119-21; pl. XIB, XXB, XLIIA. 16 Louis Malleret, L’archéologie du Delta du Mékong, vol. 1 (Paris: École Française d’Extrême-Orient, 1959), pp. 377-88. Following Malleret’s typology, Stanley J. O’Connor discusses the li¶gas from Peninsular Thailand in ‘Si Chon: An early settlement in peninsular Thailand’, Journal of the Siam Society [henceforth JSS], 56, 1 (1968): 14-16 and ‘Some early Siva lingas in Nakhon Si Thammarat, peninsular Siam’, JSS, 71, 1-2 (1983): 1-5. Robert Brown summarises the complexities of developing a chronology for li¶gas in ‘Indian art transformed’, pp. 47-9. As Vickery points out, ‘only three or four inscriptions have survived from what might have been pre-7th century Funan in the south of Cambodia and Vietnam’ (Society, economics, and politics, p. 37). Two of these are discussed above. 17 On the basis of its script, Coedès considered K.365 to be contemporary with Gu§avarman’s inscription from roughly the second half of the fifth century; Georges Coedès, ‘Nouvelles données sur les origines du royaume khmèr. La stèle de Vat Luong Kau, près du Vat Phu’, BEFEO, 48, 1 (1956): 209-20; see also Claude Jacques, ‘Notes sur l’inscription de la stèle de Vat Luong Kau’, Journal Asiatique, 250, 2 (1962): 249-56. K.365 refers to a mah¡r¡j¡dhir¡j¡ (‘a great supreme king of kings’) named Dev¡nika, ‘who “came from a distant country” to be ‘installed in supreme royal power . . . by the grace of ∫r# Li¶gaparvata, the mountain of Wat Phu’ (Vickery, Society, economics, and politics, p. 73). Another ‘mountain of the li¶ga' may have been located in southern Vietnam, but it is not mentioned until much later, in a twelfth-century inscription from Phnom Svan (K.418); ibid., p. 38. 18 Dupont, Statuaire préangkorienne, pp. 75-6. For an overview and list of these inscriptions, see Vickery, Society, economics, and politics, pp. 74-5; the inscriptions found in eastern and northeastern Thailand are enumerated in M. C. Subhadradis Diskul, ‘The newly discovered inscription of Mahendravarman in Northeastern Thailand’, in Eilenberg et al. ed., Living a life, pp. 131-5. ≠Ω, ∞      27 that Mahendravarman followed the expansionist policies of his predecessor [Bhavavarman I]’. Michael Vickery, however, while agreeing that these ‘“Dangrek chieftains” were indeed conquerors, or at least would be conquerors’, has recently argued that these inscriptions ‘should be regarded as records of exploratory probes rather than enduring conquests, with little, if any, permanent effect… [and not as] “delimiting” any kingdom’.19 Whatever the case may be, what is significant here is that these rulers consciously selected ∫iva li¶gas to indicate their presence and potential authority, whereas, as we have already seen, contemporary rulers in the south opted for images of Vi≤§u. The seventh and eighth centuries During the seventh century, the pattern of distribution of images became decidedly more mixed. This was probably linked to the ever-increasing rise of power inland in what is now northern Cambodia. At the centre of these developments, the deity Harihara rose to prominence. As mentioned above, anthropomorphic images of ∫iva were extremely rare during the Preangkorian period; they only began to appear in appreciable numbers in Khmer art in the late ninth century, during the reigns of Indravarman I (877–ca. 886) and his son YaΩovarman I (889–ca. 915). Anthropomorphic images of Vi≤§u and Harihara, by contrast, were relatively common during the this period. Unlike the mitred Vi≤§u images, however, there are no known Southeast Asian sculptures of Harihara that can be convincingly dated to the sixth century or earlier. Both depictions of, and inscriptions referring to, Harihara probably appeared for the first time in Southeast Asia during the seventh century.20 Furthermore, unlike the wide distribution of the mitred Vi≤§us, early Harihara images in Southeast Asian were confined to Khmer art. Harihara seems to have had its greatest appeal in Preangkorian Khmer culture and it achieved this degree of popularity in no other region of Southeast Asia or India, or indeed at any time in later Khmer history – though the epigraphic and art historical evidence from Cambodia indicates that Harihara continued to be worshipped through the thirteenth century, albeit in an ever-diminishing capacity. As a result of Khmer influence, several interesting bronze images of Harihara were made in the Thai kingdom of Sukhothai during the thirteenth and fourteenth centuries, but by this time the worship of Hindu deities had clearly taken a back seat to Buddhism in mainland Southeast Asia. In other Hindu-influenced areas of Southeast Asia, Harihara is either completely absent from the historical record, as at Si Thep (Thailand)21 and in Burma; extremely rare, as in Champa22; or else dates from a much later period, as in Java. 19 Coedès, Indianized states, pp. 68-9; Vickery, Society, economics, and politics, p. 79. 20 I am aware of the recent article by Emma Bunker in which she places one Harihara image in the fifth century. While her discussion is interesting, I am, for reasons already stated, excluding recently published unprovenanced pieces. See Emma C. Bunker, ‘Harihara images of the Pre-Angkor period in Cambodia’, Arts of Asia, 31, 2 (2001): 91-107. 21 Virginia Dofflemyer, ‘The ancient city of Si Thep: A study of the extant Brahmanical sculpture (5th10th centuries)’ (Ph.D. diss., University of Southern California, 1982), p. 138, n. 11. However, as one of my readers reminded me, because Si Thep has not been completely excavated (particularly the Klang Nok area) and looting has been a problem at the site, we cannot be certain that Harihara is indeed ‘totally absent’. 22 Harihara, under the epithet ∫a¶kara-N¡r¡ya§a, is mentioned only once in Cham epigraphy: Face B of the Glai Lamov Stele (C.24/801 CE) commemorates the foundation of an image of ∫a¶kara-N¡r¡ya§a; see M. Abel Bergaigne, Inscriptions sanscrites du Cambodge (Paris: Imprimerie Nationale, 1893), pp. 227, 230 (no. 23); Jean Boisselier, La statuaire du Champa: Recherches sur les cultes et l’iconographie (Paris: École Française d’Extrême-Orient, 1963), p. 420, note for p. 70. 28  .  A terminus post quem of the early seventh century for the first appearance of Harihara in Khmer culture requires some discussion as it must be seen in the light of a very important recent revision of the generally accepted chronology of Preangkorian art. It is also interesting to note the often overlooked yet pivotal position held by one particular Harihara image in the sequence of Preangkorian sculpture. Establishing a chronology for the corpus of Harihara and Vi≤§u images and for Preangkorian sculpture in general is one of the most complicated and controversial problems in the field of Southeast Asian art history. This is due primarily to the fact that most Preangkorian statuary is free-standing and free-floating and consequently cannot be linked with specific sanctuaries or inscriptions. In the most comprehensive and important study to date, Dupont pointed out the lack of Preangkorian art that meets what he called a ‘double equation’, meaning the simultaneous survival of a sanctuary, its central image and an inscription referring specifically to that image. At the time he was writing, only two known works of Preangkorian sculpture could definitely be placed in the context of a temple: a sculpture of Brahm¡ found inside Sambor Prei Kuk sanctuary N22 and the well-known Harihara found inside Prasat Andet. In neither case, however, do inscriptions on the site mention these particular images.23 Rita Régnier subsequently published a sculpture of Harihara that she convincingly argues meets Dupont’s criteria for a ‘double equation’. The head and body were found separately in the debris of the sanctuary of Prasat Phum Prasat, which can be dated to the beginning of the eighth century not only on the basis of its architectural style, but also by an inscription on the south doorjamb of the main entrance (K.145/706 CE) that records offerings made to the god ∫a¶kara-N¡r¡ya§a (≈Harihara), presumably the very image enshrined in the sanctuary.24 Thus, it is a Harihara made ca. 706 that is probably the most securely dated piece of Preangkorian sculpture. Nowhere are the difficulties of establishing a reliable chronology for Khmer sculpture more pronounced than in the dating of what has often been considered the earliest style, that of ‘Phnom Da’, which takes its name from a hilltop site in southeastern Cambodia. Of particular importance for this discussion is the well-known Harihara of Asram Maha Rosei, generally believed to be one of the outstanding examples of this style. Dupont divided what he considered to be the earliest statuary from the area of Phnom Da and Angkor Borei into two parts, ‘Styles A and B’, to which he added a number of ‘prolongements’ in order to include pieces that were either slightly later or that were found outside of present-day southern Cambodia. The Harihara of Asram Maha Rosei was assigned to Phnom Da Style A, which he dated specifically to the reign of Rudravarman (ca. 514-39). Another Harihara, which now survives in two pieces, was assigned to the subsequent period of Style B, dated to the second half of the sixth century and perhaps 23 Dupont, Statuaire préangkorienne, p. 214; see pls. XXA and XXXIIIA–XXXIVA. In cases where sculpture has been found in the vicinity of a temple (and even in these two cases in which the images were found inside the cella), it is difficult to know if they were actually made at the same time as the sanctuary, or whether they were subsequently brought there. 24 Rita Régnier, ‘Note sur l’évolution du chignon (ja∑¡) dans la statuaire préangkorienne: Réflexions à propos d’une tête de Harihara provenant de Prasat Phum Prasat, au Musée National de Phnom Penh’, Arts Asiatiques, 14 (1966): 32-4. Régnier (pp. 32-3) assigns the image to the style of Kompong Preah, dated by Dupont to the first half of the eighth century; Dupont, Statuaire préangkorienne, p. 180. For the inscription, see Inscriptions du Cambodge [henceforth IC], ed. George Coedès (Paris: E. de Boccard, 193766), vol. 6, p. 72; all inscriptions in this collection are numbered according to the system mentioned above. ≠Ω, ∞      29 extending through the early years of the seventh.25 The preeminent scholar of Khmer sculpture, Jean Boisselier, was suspicious of the early date accorded to the style of Phnom Da by Dupont. Despite a few comments here and there, however, he never fully articulated an alternative in print, tending to accept Dupont’s chronology in his best-known publications. Other scholars also generally followed Dupont until recently, when Boisselier’s suggestion of a later date for the Phnom Da style seems to have crept into numerous publications without comment or citation.26 It was not until Nancy Dowling’s article advocating a mid-seventh-century date that any scholar provided a cogent argument explaining why such a revision is in fact appropriate. A chronological adjustment of this magnitude has major implications for the study of Khmer art. According to Dowling, the new date ‘shortens by 100 years the chronology for early Cambodian sculpture. Of further importance, no gap of 100 years now separates the Phnom Da style from the early to mid-seventh century date for the Sambor style.’ I would go even further and suggest that, in addition to compressing the duration for the development of Khmer sculpture, a mid-seventh-century date for the style of Phnom Da would make it contemporary or, even more likely, slightly later than the style of Sambor, which is much more firmly anchored in the early to midseventh century.27 These are critical points for the present discussion since the Harihara of Asram Maha Rosei should no longer be seen as the earliest Preangkorian image of that deity and since the first appearance of Harihara in Khmer art, on the basis of the available evidence, seems thus to have been tied to early to mid-seventh-century political developments at Sambor Prei Kuk. Under ∞Ω¡navarman I (ca. 616-37), Sambor Prei Kuk, after which the style of Sambor is named, became a dominant regional power centre. Located in the valley of the Stung Sen River in the Bassac region of northern Cambodia (province of Kompong Thom), it 25 Dupont, Statuaire préangkorienne, pp. 25-42, pl. IIA-B, and Sculpture of Angkor and ancient Cambodia: Millennium of glory, ed. Helen Jessup and Thierry Zéphir (Washington, DC: National Gallery of Art, 1997), pp. 164-5 (no. 16). For the dating of the Phnom Da Style, see Dupont, pp. 22-5, 42, 48; the discussion by Coedès upon which Dupont based his arguments is in IC, vol. 2, pp. 155-6. According to Dupont (pp. 113, 150), there may have been some overlap between the end of the Style of Phnom Da and the beginnings of the next style, the ‘Style of Sambor’, which he thought probably occupied the first half of the seventh century, beginning sometime between 610/615 and 630. The head of the broken statue is in the Musée Guimet, Paris and the body is in the National Museum, Phnom Penh; Dupont, pl. VIIIB and XIA. 26 Boisselier expressed doubts regarding Dupont’s chronology in his ‘La statuaire préangkorienne et Pierre Dupont’, Arts Asiatiques, 6, 1 (1959): 61-2; idem., Trends in Khmer art, trans. Natasha Eilenberg and Melvin Elliott (Ithaca, NY: Cornell University Southeast Asia Program, 1989), p. 27; and personal communication with Boisselier cited in Kamaleswar Bhattacharya, ‘Hari Kambujendra’, Artibus Asiae, 27, 1/2 (1964): 78, n. 25. Towards the end of his career, however, Boisselier seems to have strengthened his resolve that the style of Phnom Da must date much later than Dupont thought. In a comparatively recent Italian publication, he placed the Phnom Da images in the late seventh or eighth century and argued that the style of Sambor Prei Kuk constitutes the earliest Preangkorian art; Boisselier, Il Sud-Est asiatico (Turin: Storia Universale dell’Arte, 1986), pp. 27-8. 27 Nancy H. Dowling, ‘A new date for the Phnom Da images and its implications for early Cambodia’, Asian Perspectives, 38, 1 (1999): 59. Dowling’s re-dating of the style of Phnom Da is based on new observations regarding a jewelled band motif consisting of alternating ovals and rectangles (known as ‘la bande à chatons’) and a double-lotus base, which probably date to ca. 650 and ca. 616-35 respectively. The revision may also shift some of the so-called ‘statuaire du Tchen-La’ – assigned by Dupont to the styles of Prei Kmeng (seventh century) and Prasat Andet (mid-seventh to early eighth centuries) – to the eighth century, but this remains to be argued. 30  .  is a large complex with epigraphic evidence of occupation from at least the seventh through the tenth centuries. The identification of Sambor Prei Kuk with ∞Ω¡navarman’s capital, ∞Ω¡napura – a name that appears in various forms in both Chinese texts and local inscriptions – is based on several inscriptions found on the site that mention ∞Ω¡navarman in conjunction with extravagant dedications made there. Three groups of monuments, as well as a number of isolated sanctuaries, comprise the site. The bulk of the monuments in the northern and southern groups probably date to the first three quarters of the seventh century, whether to the reign of ∞Ω¡navarman I or those of his successors, Bhavavarman II (ca. late 630s–50s?) and Jayavarman I (ca. 657–81).28 Three important images were found in the northern group: the aforementioned Brahm¡ from sanctuary N22; an image of Durg¡ Mahi≤¡suramardin#, parts of which were found in sanctuaries N1 and N9; and a fragmented image of Harihara found in sixty pieces in front of sanctuary N10. Unlike the other Preangkorian structures at Sambor Prei Kuk, temple N22 has a lintel in the style of Prei Kmeng. Consequently, Dupont dates the Brahm¡ from N22 a little later in the seventh century than the other two images, which form the core of his ‘Style of Sambor’. Precise dating of the Durg¡ and Harihara images presents a problem because of the wide range of dates found in the inscriptions of the north group of Sambor Prei Kuk. Dupont and Bosselier, however, date them to the first half of the seventh century, a reasonable attribution given the references to ∞Ω¡navarman and (probably) to Bhavavarman II in the epigraphy of the northern group.29 If we accept Dupont and Boisselier’s dating of these images, they would precede the 28 Two almost identical inscriptions from the southern group, one from the door jamb of the east door of the external enclosure (K.440) and the other from the mandapa of tower S2 (K.442), contain lengthy panegyrics to King ∞Ω¡navarman and commemorate the erection of a gold image of PrahasiteΩvara, the ‘lord of laughter’, presumably an epithet of ∫iva. They go on to mention several other images (pratima) enshrined by the king, including four (in stanza XXXI in K.440) of ∫iva, Nandin, Harihara (?), and another whose name is effaced. Subsequent stanzas (XXXII-XXXIV) also list a gold li¶ga, an image of Brahm¡, another ∫iva, a ∫iva Nataraja, a Sarasvat# and a silver Nandin. Apparently none of these images has been found (IC, vol. 4, p. 11). The southern group (Group S) seems to be the earliest of the three and shows the most internal consistency. The central group (Group C) was probably built towards the end of the seventh century. The northern group (Group N) consists of monuments dedicated over a long period of time, from perhaps the sixth century well into the Angkorian period. The earliest structure at Sambor Prei Kuk probably belongs to the north group (N17); ibid., pp. 3-35. See also Dupont, Statuaire préangkorienne, p. 83; Lawrence Palmer Briggs, The ancient Khmer Empire, Transactions of the American Philosophical Society, n.s. 41, 1 (Philadelphia: American Philosophical Society, 1951), p.74; Jacques Dumarçay and Michael Smithies, Cultural sites of Burma, Thailand, and Cambodia (Kuala Lumpur: Oxford University Press, 1995), pp. 73-9; and Jacques Dumarçay and Pascal Royère, Cambodian architecture eighth to thirteenth centuries, trans. Michael Smithies (Leiden: Brill, 2001), pp. 39-44. Philippe Stern has dated the central temple of Group C to the ninth century; Philippe Stern, ‘Le style du Kulên (décor architectural et statuaire)’, BEFEO, 38 (1938): 139-41. The historical circumstances surrounding Bhavavarman II are complex and mysterious; for discussion, see Dupont, Statuaire préangkorienne, pp. 89-93; Vickery, ‘What and where was Chenla?’, p. 203, n. 12; idem., ‘What to do about The Khmers’, JSEAS, 27, 6 (1996): 392; idem., Society, economics, and politics, pp. 21-5, 328-34, 340-4. The two inscriptions that clearly refer to Bhavavarman II are K.79 and K.1150. Jacques devotes a lengthy discussion to these inscriptions and the light they shed on ∞Ω¡navarman and Bhavavarman II in ‘Pays khmer avant Angkor’, pp. 79-82. 29 On the dating see Dupont, Statuaire préangkorienne, pp. 150-5, and Jean Boisselier, ‘Une statue féminine inédite du style de Sambor’, Arts Asiatiques, 2, 1 (1955): 18-34. It is worth noting that N1 was the central shrine of the group and N9 and N10 were corner sanctuaries flanking it. In other words, N1, N9 and N10, and presumably the Harihara and Durg¡ found in or near these sanctuaries, occupied the same central platform. If the northern group was built in a series of phases beginning in the early seventh ≠Ω, ∞      31 revised middle to late seventh-century date for the Phnom Da images. However, if we allow for the possibility that the images in the Sambor style may date to the reigns of Bhavavarman II or Jayavarman I, then there is a distinct possibility that these two styles may have existed simultaneously. Given the likelihood that the political pattern was one of multiple autonomous or semi-autonomous principalities, sometimes called ma§∂alas, it would make sense for there to have been coexisting regional artistic styles rather than the often presumed straightforward linear stylistic development. Indeed, the styles of Phnom Da and Sambor Prei Kuk probably represent two distinct and coexisting ‘schools’ of seventh-century Khmer art, in the south of present-day Cambodia and to the north in the vicinity of Kompong Thom respectively.30 As for Harihara images, the redating of the style of Phnom Da would place at least three large free-standing sculptures of the deity in the seventh century, probably between ca. 620-80: Sambor Prei Kuk, Asram Maha Rosei and the Phnom Da style B. Four other images of Harihara can be attributed to the style of Prasat Andet, a designation first used by Dupont to refer to a number of Preangkorian sculptures that he dated from the middle of the seventh to the beginning of the eighth centuries. Boisselier has since argued that the Prasat Andet style should be pushed back to the first half of the eighth century.31 This would make the images roughly contemporary with two other images of Harihara assigned to the style of Kompong Preah (first half of the eighth century) by Dupont and Régnier. A second Harihara from Trapeang Phong probably constitutes the latest extant century, it would make sense that the central shrines would be early; however, it is also possible that they were rebuilt and new images were installed, but based on the available epigraphic evidence, this probably could not have happened after the reign of Jayavarman I. It is therefore highly unlikely that these images date later than the third quarter of the seventh century. 30 Zéphir, perhaps following Boisselier, has advocated the use of the more flexible term ‘school’ over style; Thierry Zéphir, ‘khmer art’, in Maud Girard-Geslan et al., Art of Southeast Asia, trans. J. A. Underwood (New York: Harry N. Abrams, 1997), p.161 Boisselier, Heritage of Thai sculpture, p. 63. It must be pointed out that Dupont did allow for overlap between his stylistic groups (see, for example, Dupont, Statuaire préangkorienne, pp. 51, 113, 156, 167). Some important studies of Southeast Asian polity formation and development that have informed my discussion here include Robert Heine-Geldern, Conceptions of state and kingship in Southeast Asia (Ithaca: Cornell University Southeast Asia Program Data Paper, 1956); Wolters, History, culture, and region, pp. 27-40; Higham, Archaeology of mainland Southeast Asia; Hermann Kulke, ‘The early and the imperial kingdom in Southeast Asian history’, in Southeast Asia in the 9th to 14th centuries, ed. David G. Marr and Anthony C. Milner (Singapore: Institute of Southeast Asian Studies, 1986), pp. 1-22; Jan Wisseman Christie, ‘Negara, mandala, and despotic state: Images of early Java’, in the same volume, pp. 65-94; idem., ‘State formation and early maritime Southeast Asia: A consideration of the theories and data’, Bijdragen tot de Taal-, Land- en Volkenkunde, 151 (1995): 235-88. 31 Dupont, Statuaire préangkorienne, pp. 166-79 and pl. 33A, 34A, 35A-B. The four images that Dupont included in the transitional period of Prasat Andet are the type-image, the Harihara of Prasat Andet (Kompong Thom Province, the heart of the realm of ∞Ω¡navarman and his successors); that of Kompong Speu (southern Cambodia); that of Trapeang Phong (Siem Reap Province); and a head of Harihara that Louis Malleret found along the approach to the sanctuary of Linh-s◊n t3 at the site of Phnom Bathê in present-day southern Vietnam. See Malleret, Archéologie du Delta du Mékong, vol. 1 (text), pp. 409-10 and vol. 1 (plates), pl. LXXXVIIb. Boisselier’s comprehensive study of Khmer sculpture published in 1955 did not recognise the Prasat Andet style at all; instead, he subsumed it under his third style, combining those of Prei Kmeng and Kompong Preah (Boisselier, La statuaire khmère, p. 15). In his handbook of Khmer art and archaeology published in 1966, he recognised the Prasat Andet style but argued that it was probably contemporary with the beginnings of the Kompong Preah style; Boisselier, Le Cambodge, manuel d’archéologie d’Extreme-Orient, première partie Asie du Sud-Est, tome I (Paris: A. et J. Picard, 1966), p. 242. In a subsequent article, he refined the stylistic chronology of Preangkorian sculpture by pushing back the style of Prasat Andet to the first half of the eighth century, arguing that it began around 700; idem., ‘The AvalokiteΩvara in the Museum’s Wilstach Collection’, Bulletin of the Philadelphia Museum of Art, 77, 333 (1981): 18-21. 32  .  Preangkorian sculpture of Harihara.32 Thus, in summary, at least ten Harihara images survive from the Preangkorian period (compared to only two known images during the entire Angkorian period). The earliest extant images of Harihara date to the reign of ∞Ω¡navarman (in the early seventh century) while the two latest known examples date to during or just after the reign of Indravarman I (in the late 800s).33 The period between and including these reigns, therefore, corresponds to the likely range for all known Khmer Harihara images. Of the total corpus of approximately twenty-three inscriptions that contain definite or probable references to Harihara (under a variety of epithets), at least four belong to the twentyyear reign of ∞Ω¡navarman I; nine others probably also date to the seventh and eighth centuries. The other eleven span the period from 881 to the late thirteenth or early fourteenth centuries. In other words, more references to, and images of, Harihara appear during a two-century span of the Preangkorian period than occur over the course of the six centuries of the Angkorian period that follow. The ‘politics’ of gods: ≥Ω¡navarman and Khmer kingship of the Preangkorian period The Preangkorian epigraphic references to Harihara provide little explanation for the unusual popularity this deity enjoyed at that time. Most simply commemorate the foundation of images and temples or record, like an inventory, gifts made in honour of the god. A few, however, do provide other types of information. Examined in their larger historic and geographic context, these inscriptions offer clues as to the function and symbolism of images of Vi≤§u, ∫iva and Harihara in Preangkorian civilisation. In many cases Harihara is mentioned in association with other gods. This can be seen in several inscriptions that belong to the reign of ∞Ω¡navarman I, two of which mention Harihara dedications made in association with various forms of ∫iva (in each case including a li¶ga) and one of which is associated with a li¶ga and an image of Vi≤§u.34 Two other approximately contemporary inscriptions also suggest a mixed context. It is clear from these and other inscriptions left by ∞Ω¡navarman and his 32 Dupont dated it to the eighth century, but this attribution is made solely on the basis of what he sees as its ‘extreme mediocrity’ (Dupont, Statuaire préangkorienne, p. 187, pl. 42B). The aforementioned Harihara of Prasat Phum Prasat (Province of Kompong Thom), dated by inscription to 706 CE, and the Harihara head from Vat Prasat are now in the Musée Guimet de Lyon; a photograph of the latter can be found in Plate XLA. 33 The two Angkorian period images of Harihara are (1) the Harihara of Bakong and (2) a head of Harihara that has been alternately dated to the styles of Bakheng (ca. 893-925) and Pre Rup (ca. 947-65). See Jean Boisselier,‘Le Harihara de Bakon’, BEFEO, 46, 1 (1952): 253-6; idem.,‘Une tête angkorienne de Harihara’, Arts Asiatiques, 44 (1989): 44-9; and Jessup and Zéphir ed., Sculpture of Angkor and ancient Cambodia, pp. 206-7. Several lesser-known pieces could potentially be added to the ten Preangkorian sandstone images of Harihara mentioned here; these are discussed individually in my dissertation (in progress). 34 K.22, K.440, K.926/624; see IC, vol. 3, pp. 143-7; vol. 4, pp. 5-11; vol. 5, pp. 20-2. Two inscriptions from Thma Krê (K.926 and K.927) refer to ∫a¶kara-N¡r¡ya§a (Harihara); both date to the seventh century and mark the location of a sanctuary dedicated to Harihara. K.926, carrying a date of 624 (during the reign of ∞Ω¡navarman), refers to the erection of an image of Harihara and then enumerates gifts of men, women, cows and ricefields to be held in common by ∫a¶kara-N¡r¡ya§a (Harihara) and a god named Suvar§ali¶ga (a ∫iva li¶ga made of gold). The undated stele of Vat Po (K.22) mentions ∞Ω¡navarman and records dedications made by a muni (sage or ascetic) named ∞Ω¡nadatta, said to have erected a Harihara and several other images, including a li¶ga and a Vi≤§u. Based on these inscriptions it is clear that worship of ∫iva and deities associated with him was an important component of the reign of ∞Ω¡navarman. ≠Ω, ∞      33 immediate successors that worship of ∫iva prevailed at Sambor Prei Kuk during the seventh century.35 Harihara, usually mentioned in association with ∫iva, was an important component of his dedications; Vi≤§u, on the other hand, is mentioned only infrequently. These ∫iva-oriented dedications are the opposite of the tendency in the south, where dedications are dominated by Vi≤§u images, with very few epigraphic references to ∫iva and some ∫iva li¶gas that are difficult to date. How can we explain this distinct geographic pattern and the importance of Harihara images during the middle to late seventh century? The idea of the equivalence of king and god is a frequent feature of royal panegyrics in the epigraphy of Cambodia (and throughout the Indic world) and is stated repeatedly in the inscriptions associated with ∞Ω¡navarman. The stele of Vat Po (K.22) clearly places Harihara and ∞Ω¡navarman in an analogous or parallel relationship. The first stanza is an invocation to the victorious Hara and Acyuta, who unite for the good of other beings. The second stanza similarly praises ∞Ω¡navarman as triumphant, for he carries the earth like the serpent ∫e≤a. Here, both Harihara and ∞Ω¡navarman are represented simultaneously as conquerors and protectors. Inscription K.440 from Sambor Prei Kuk describes the god PrahasiteΩvara as victorious and follows with a panegyric to ∞Ω¡navarman that emphasises his physical beauty, good actions and military exploits. Similarly, K.80 describes ∫iva and ∞Ω¡navarman as victorious and then compares them as masters of the earth. A tenth-century inscription from Koh Ker states that ∫iva composed a portion of the king; this is reminiscent of similar statements regarding Vi≤§u.36 Jan Gonda has analysed the close relationship in Indian kingship between the ruler and Vi≤§u, both of whom are protectors of the world who defend their followers and ‘punish the wicked’. Basing his discussion on Vedic texts, the epics, and the Pur¡§as (many of which were known among the ancient Khmer), Gonda points out that ‘by identifying himself with Vi≤§u the king is able to conquer the worlds’. The ruler is even said to consist of a ‘portion’ of Vi≤§u. Kamaleswar Bhattacharya uses this same evidence to argue that Vi≤§u was identified with Khmer kings and that the eight-armed image from Phnom Da represents a king in the guise of Vi≤§u as guardian of the eight regions of the cosmos.37 Vi≤§u, when not represented in the form of one of his avatars, was usually depicted in Khmer art as a world sovereign or cakravartin, wearing the royal mitre and an elaborately folded sampot (a garment consisting of a rectangular cloth wrapped around the waist and tied in front). The attributes held in Vi≤§u's four hands are constant In every case cited, Harihara is mentioned in association with a ∫iva li¶ga. However, the stele of Vat Po (K.22), with its mention of a Vi≤§u image, does seem to imply a more mixed context. 35 K.21/ca. 639, K.107; see IC, vol. 5, pp. 5-6; vol. 6, pp. 38-9; Auguste Barth, Inscriptions sanscrites du Cambodge (Paris: Imprimerie Nationale, 1885), pp. 21-6. Lines 8-13 of K.21 (from Poñ¡ Hòr) record the dedication of two li¶gas, a Durg¡ and, although it is not entirely clear from the context, probably an image of ∫ambhu-Vi≤§u (≈Harihara) as well. Ten lines later an image of Vi≤§u Trailokyas¡ra is also mentioned. The date of this inscription is uncertain (the Khmer portion of the inscription originally began with a date that has not survived). Like K.21, K.107 ( found at Prah Th#t Khnai Van) suggests a mixed pattern of worship where Harihara was concerned. It lists gifts of lakes or ponds (pi¶) to several different deities, including Svayambh∞ (Brahm¡), ∫a¶kara-N¡r¡ya§a and J¡yadeva. This undated inscription has been assigned by Coedès to the Preangkorian period, probably the seventh century, on the basis of its orthography; other examples are in IC, pp. 4-5. 36 IC, vol. 3, pp. 143-7 (K.22); vol. 4, pp. 7, 9 (K.440); vol. 6, pp. 3-4 (K.80); vol. 1, pp. 58, 61 (Koh Ker). 37 Gonda, Aspects of early Vi≤§uism, pp. 164-7; Bhattacharya, ‘Hari Kambujendra’, pp. 72-8. 34  .  throughout the duration of Khmer art – the conch (or Ωankha, upper left), the disc (or cakra, upper right), the earth ball (or bh∞mi, lower right), and the mace or club (or gad¡, lower left). The mace and cakra were power symbols; the conch seemed to embody fertility (Gonda notes that it resembles the vulva) and may have functioned as an apotropaic emblem; and the ball, representing the earth, probably referred to Vi≤§u's powers of creation and preservation, duties that were also expected of earthly sovereigns.38 It is also likely that Vi≤§u, a deity that developed as a synthesis of apotheosized Indian heroes, appealed strongly to indigenous Southeast Asian religious traditions centred around ancestor worship.39 ∫iva, however, was even more intimately tied to Khmer kingship than Vi≤§u. According to Paul Mus, ∫iva became assimilated into indigenous chthonic cults in India and Southeast Asia. Li¶gas, which represented the union of the king and ∫iva, were an expression of ‘old territorial rituals in which the materialisation of the god of the soil, in the person of a dynastic ancestor, expressed the contract, defined in time and space, of the group with its territory’.40 Coedès has written extensively on the nature of Khmer ‘personal cults’ in which kings and members of the royal family were equated with deities.41 This is a complicated subject and cannot be dealt with fully here; suffice it to say, however, that ∫iva li¶gas seem to have become established as a specifically royal emblem during the Preangkorian period and in 802 Jayavarman II is supposed to have performed a ritual that preserved his royal power or essence in a li¶ga that was later known as the ‘devar¡ja’.42 O. W. Wolters has, moreover, argued that an important aspect of ‘Khmer 38 Gonda, Aspects of early Vi≤§uism, pp. 96-104; Stanley J. O’Connor has identified this particular configuration as the Jan¡rdana or V¡sudeva murti (Hindu gods of peninsular Siam, p. 31). Robert Brown has analysed the stylistic and iconographic development of the early Southeast Asian Vi≤§u images in ‘The early Vi≤§u images from Southeast Asia and their Indian relationships’, paper presented at ‘Crossroads and commodification: A symposium on Southeast Asian art history’, University of Michigan, 25-26 March 2000. 39 This development of Vi≤§u from Indian heroes has been discussed chiefly in the context of early Indian images of Vi≤§u, Kr≤§a and V¡sudeva; see Herbert Härtel, ‘Archaeological evidence on the early . V¡sudeva worship’, in Orientalia Iosephi Tucci memoriae dicata, ed. Gherardo Gnoli and Lionello Lanciotti (Rome: Istituto Italiano per il Medio ed Estremo Oriente, 1985-8), vol. 2, pp. 573-87; Doris Srinivasan, ‘Early Vai≤§ava imagery: Caturvyuha and variant forms’, Archives of Asian Art, 32 (1979): 39-54; idem., ‘Vai≤§ava art and iconography at Mathura’, in Mathura, the Cultural Heritage, ed. Doris M. Srinivasan (New Delhi: American Institute of Indian Studies, 1989), pp. 381-92; idem., Many heads, arms and eyes – origin, meaning and form of multiplicity in Indian art (Leiden: Brill, 1997), pp. 211-59. Brown, ‘Early Vi≤§u images', briefly deals with these issues in the Southeast Asian context, arguing that the famous ‘Chaiya Vi≤§u', usually considered the earliest Vi≤§u image in Southeast Asia, is not Vi≤§u at all, but rather V¡sudeva-Kr≤§a. . 40 Paul Mus, India seen from the east: Indian and indigenous cults in Champa, ed. Ian W. Mabbett and David P. Chandler, trans. Ian W. Mabbett (Clayton: Monash University Centre of Southeast Asian Studies, 1975), pp. 43-9. Although Mus was writing about the Cham, his words are no less applicable to the Preangkorian Khmer situation. 41 See, for instance, George Coedès, Angkor: An introduction (Hong Kong: Oxford University Press, 1963), pp. 22-33. 42 The term devar¡ja as a particular ritual or ‘cult’ only occurs once in the entire corpus of Khmer epigraphy (K.235/CE1052), although there may be minor allusions to it in other Angkorian inscriptions; Kamaleswar Bhattacharya, Recherches sur le vocabulaire des inscriptions sanscrites du Cambodge (Paris: École Française d’Extrême-Orient, 1991), pp. 54-5. At any rate, the devar¡ja is not relevant to a discussion of the Preangkorian period. Some important studies of the devar¡ja include Ian W. Mabbett, ‘Devar¡ja’, Journal of Southeast Asian History, 10 (1969): 202-23; Hermann Kulke, The Devar¡ja cult, trans. ≠Ω, ∞      35 “Hinduism”’ during the seventh century was asceticism as it related to the worship of ∫iva, the arch-ascetic. Michael Vickery has been highly critical of the concept of ‘Khmer Hinduism’ as conceived primarily by Wolters but also developed by George Coedès and Kamaleswar Bhattacharya. In particular, he contends that Wolters’ reliance on Sanskrit inscriptions (to the exclusion of those in Old Khmer) led him to exaggerate the importance of ∫aivite devotionalism (bhakti) and asceticism in his formulation of ‘Khmer Hinduism’. Vickery argues that these concepts are ‘in fact mentioned in very few contexts, which may be little more than formalistic phraseology preceding the important organisational details of the Khmer text’.43 While it may be true that the term bhakti is rare, the epigraphic evidence adduced by Wolters and by Bhattacharya combined with the vast corpus of large iconic stone images indicate that, at least for the Khmer elite, Hindu deities played a central role in not only their religion, but also perhaps in a perception that they were participating in a larger ‘Hindu world’.44 Wolters does not use this term to refer to ‘the actual world of regional polities in India’ or to a self-consciously ‘Hindu’ identity, but rather to describe a self-perception that the Khmer were participating in ‘the world of the gods and heroes as it is set forth in Indian sacred literature and perhaps, above all, in the Mah¡bh¡rata…’. It should also be pointed out in this regard that there is epigraphic evidence to indicate that the Mah¡bh¡rata, R¡m¡yana and the Pur¡§as were well known among the Khmer elite by the seventh century CE. Likewise, Brahmins – be they locals or Indians – versed in the Vedas, Upavedas and the Vedangas are mentioned in an inscription from southern Vietnam that probably dates to the late fifth century.45 Simply because the references to Hindu deities in Sanskrit inscriptions can be identified as ‘formalistic phraseology’ does not mean that they were meaningless for the patron, nor does it render them useless to the scholar. The conscious choice of which deities to invoke, and for what reasons, reveals something of the patron’s ideals and aspirations with regard to the supernatural realm, I. W. Mabbett (Ithaca: Cornell University Southeast Asia Program Data Paper, 1978); and Hiram W. Woodward, Jr, ‘Practice and belief in ancient Cambodia: Claude Jacques’ Angkor and the Devar¡ja question’, JSEAS, 32, 2 (2001): 249-61. 43 O. W. Wolters, ‘Khmer “Hinduism” in the seventh century’, in Smith and Watson ed., Early South East Asia, pp. 427-42. Vickery – quite rightly, it seems to me – questions Wolters’ emphasis on ‘deathwishes’ among early Khmer kings (Society, economics, and politics, pp. 170-1). See also Coedès, Indianized states, pp. 14-35; Bhattacharya, Religions brahmaniques; idem., ‘Religious speculations in ancient Cambodia’, in R.C. Majumdar felicitation volume, ed. Himansu Bhusan Sarkar (Calcutta: Firma K. L. Mukhopadhyay, 1970), pp. 78-97; idem., ‘The religions of ancient Cambodia’, in Jessup and Zéphir ed., Sculpture of Angkor and ancient Cambodia, pp. 34-52. 44 Wolters, ‘Khmer “Hinduism”’, pp. 431-4, 440; idem., History, culture, and region, pp. 109-10; Kamaleswar Bhattacharya, ‘The Pâñcarâtra sect in ancient Cambodia’, Journal of the Greater India Society, 14, 2 (1955): 111-6; idem., ‘Secte des Pâsupata dans l’ancien Cambodge’, Journal Asiatique, 243, 2 (1955): 479-90. Related issues are discussed in Sheldon Pollock, ‘The Sanskrit cosmopolis, 300-1300: Transculturation, vernacularization, and the question of ideology’, in Ideology and status of Sanskrit: Contributions to the history of the Sanskrit language, ed. Jan E. M. Houben (Leiden: Brill, 1996), pp. 197-247. 45 Wolters, ‘Khmer “Hinduism”’, pp. 437-8. See Coedès, Indianized states, pp. 60, 74; Thierry Zéphir, ‘The progress of Rama: The Ramayana in Khmer bas-reliefs of the Angkor period’, in Silk and stone: The art of Asia, ed. Jill Tilden (London: Hali Publications Ltd., 1996), p. 83. See also the inscription from Vãl Kantél (K.359) published in Barth, Inscriptions sanscrites, p. 28 (no. 4); George Coedès, ‘Une inscription du sixième siècle çaka’, BEFEO, 11 (1911): 393-6; idem., ‘Deux inscriptions sanskrites’, pp. 5-7, st. 9. 36  .  even if such decisions do not necessarily reflect widespread practices at all levels of society. Having said this, in arguing that Hinduism among the Khmer was a religious experience that was as comprehensible to peasants as it was to the elite, Wolters may indeed be overstating its relevance to Khmer society as a whole. Information about the common people and their religious beliefs and practices is scarce and it is difficult to disentangle so-called ‘indigenous’ or ‘folk’ elements from imported Indian ideas. Nidhi Aeusriwongse has argued that at the village level, worship was not so much oriented around the universal gods of Hinduism (Siva, Vi≤§u, etc.), which may have had a stronger appeal for Khmer royalty and elite, but rather around local deities that were syncretised with these gods to varying degrees from place to place. It was, he suggests, a common belief in ancestral spirits – as well as these syncretic local deities – at all levels of Khmer society that generated some measure of solidarity. Wolters also alludes to a synthesis of Hinduism and local Khmer beliefs that prevented a ‘religious wedge’ from being driven between the elite and the general population. His use of the term ‘Khmer Hinduism’ is meant to underscore the fact that ‘“Hinduism” in Cambodia was not just the Indian variety transplanted.’46 Aside from this reservation – that Wolters may exaggerate the degree to which Sanskritic Hinduism was a ‘popular’ religion among the early Khmer – I maintain, in contrast to Vickery, that Wolters’ characterisation of seventh-century ‘Khmer Hinduism’ remains a valid and useful concept. Based on a number of inscriptions that attest to the ideal, if not always the actual practice, of asceticism among the Khmer elite and royal advisors, Wolters concludes that the king’s ‘abnormal powers of leadership’ were in part dependent on ‘his ascetic efforts in devotion to ∫iva’. ∞Ω¡navarman I, for example, is said to have ‘taken pleasure in the company of sages’ and his son Bhavavarman II is at one point described as ‘possessing unshakable self-control as a result of his austerities’.47 Thus ∫iva, like Vi≤§u, embodied characteristics that were integral to the Khmer concept of sovereignty. Despite the absence of anthropomorphic ∫iva images in early Khmer art, the ascetic nature of ∫iva was nevertheless indicated on sculptures of Harihara, particularly on the image of Asram Maha Rosei with its obvious ascetic’s hairstyle and tiger-skin (upon which ∫iva is said to have meditated). Indeed, Harihara may have represented an ideal 46 Wolters, ‘Khmer “Hinduism”’, pp. 437-8; the quotation is from Vickery, Society, economics, and politics, p. 170. Nidhi Aeusriwongse’s perspective is found in his ‘Devaraja cult and Khmer kingship at Angkor’, in Explorations in early Southeast Asian history: The origin of Southeast Asian statecraft, ed. Kenneth R. Hall and John K. Whitmore (Ann Arbor: University of Michigan, 1976), pp. 107-48. See also Mus, India seen from the east; Vickery, Society, economics, and politics, pp. 139-74; Robert L. Brown, The Dv¡ravat# wheels of the Law and the Indianization of South East Asia (Leiden: Brill, 1996), pp. 183-8. To properly and fully assess the validity of the concept of ‘Khmer Hinduism’, however, it is necessary to examine these various arguments in relation to the larger literature on Hinduism. Important recent studies that tackle similar issues in the Indian context can be found in Hinduism reconsidered, ed. Günther-Dietz Sontheimer and Hermann Kulke, 2nd edn (New Delhi: Manohar, 1997); Robert Eric Frykenberg, ‘Construction of Hinduism at the nexus of history and religion’, Journal of Interdisciplinary History, 23, 3 (1993): 523-50; Heinrich von Stietencron, ‘Religious configurations in pre-Muslim India and the modern concept of Hinduism’, in Representing Hinduism: The construction of religious traditions and national identity, ed. Vasudha Dalmia and Heinrich von Stietencron (New Delhi: Sage Publications, 1995), pp. 51-81; and Wilhelm Halbfass, ‘The idea of the Veda and the identity of Hinduism’, in Tradition and reflection: Explorations in Indian thought, ed. Wilhelm Halbfass (Albany: State University of New York Press, 1991), pp. 1-22. 47 Wolters, ‘Khmer “Hinduism”’, pp. 431-2; IC, vol. 5, p. 26; vol. 2, p. 70. ≠Ω, ∞      37 form of king who encompassed in one unified form two very different conceptions of sovereignty that were associated individually with Vi≤§u and ∫iva. That this symbolism persisted to some degree through the Angkorian period, while at the same time images of Harihara seem to have fallen out of favour, is attested by two inscriptions that suggest that the king was himself composed of the two halves or attributes of Harihara.48 A concrete link is in fact made between ∞Ω¡navarman and Harihara in the dated Sanskrit inscription of Vat Chakret (K.60/626-7) from the vicinity of Ba Phnom in southern Cambodia. It records the consecration of a statue of Harihara by a local ruler or general under ∞Ω¡navarman who erected the image to increase his glory and commemorate a military victory over the village or town of T¡mrapura. The inscription is clear that this ‘master [iΩvara] of T¡mrapura’ had long since reduced three other areas to ‘ornaments on his feet’.49 In other words, there is a clear connection between the erection of an image of Harihara and the imposition of political authority in the name, or at least invoking the name, of the king. As we have already seen, this idea can perhaps be seen earlier in the erection of li¶gas by Bhavavarman I and Citrasena or Mahendravarman, who seem to have made these foundations in association with military ventures. Thus, like other Khmer images (li¶gas and Vi≤§us), sculptures of Harihara – particularly during the seventh century and the reign of ∞Ω¡navarman – seem to have represented the king and the presence of his authority. This is not to say that images of Harihara were ‘portraits’ or physical likenesses of Khmer kings.50 Rather, they served as divine analogues for the concentration of royal power – a power that was legitimated, sanctified and maintained through this very association with the gods. It should be borne in mind that elite-sponsored ‘religious’ foundations were an important means of consolidating control over an area and, consequently, were instrumental in the development of centralised kingdoms in early Southeast Asia.51 Temples were often the focal points of settlements and they served as important centres of education (for the elite), the arts and the redistribution of local agricultural production. Thus they served cultural and economic functions that brought some degree of stability to a society that tended to fragment as a result of shifting allegiances and competition between regional leaders.52 48 The foundation stele of Pre Rup (K.806) from the reign of Rajendravarman II (944-968) and the stele of Prasat Crun (at Angkor Thom) from the late thirteenth or early fourteenth century (K.287); IC, vol. 1, p. 131 and vol. 4, pp. 239, 246. 49 Barth, Inscriptions sanscrites, pp. 38-42 (no. 6). The image is mentioned twice, once as Hari-∫a¶kara and again as Hari and ∫ambhu united in one body. Vickery argues that the ‘three areas’ were all in southern Cambodia near the location of K.60; Vickery, Society, economics, and politics, pp. 336, 409. 50 This claim is often made for the Harihara of Prasat Andet. Portrait sculpture does appear in later Khmer art; see George Coedès, ‘Les statues du roi khmèr Jayavarman VII’, Académie des Inscriptions et Belles-Lettres (July–October 1958): 218-26; idem., ‘Le portrait dans l’art khmèr’, Arts Asiatiques, 7, 3 (1960): 179-98; idem., Angkor: An introduction, pp. 22-33, 99; and Jeanine Auboyer, ‘Trois portraits du roi Jayavarman VII’, Arts Asiatiques, 6, 1 (1959): 70-4. See also Jan Fontein, The sculpture of Indonesia (Washington, DC: National Gallery of Art, 1990), pp. 552-5. 51 Kulke, ‘Early and the imperial kingdom’, pp. 13-15; Kenneth R. Hall, ‘Economic history of early Southeast Asia’, in The Cambridge history of Southeast Asia, ed. Nicholas Tarling, vol. 1 (Cambridge: Cambridge University Press, 1992), pp. 229-30; idem., Maritime trade and state development in early Southeast Asia (Honolulu: University of Hawai‘i Press, 1985), pp. 137-8; Brown, Dv¡ravat# wheels, p. 195. 52 Hall, Maritime trade and state development, pp. 136-68. 38  .  Conclusion The dedication of images of Vi≤§u, ∫iva and Harihara in early Southeast Asia cannot be adequately explained as a by-product of trade. When situated within their historical context, it becomes clear that at least among the Khmer, these deities were consciously adopted, employed and worshipped by the elite in a particular ‘style of rule’.53 During the fifth and sixth centuries the two conceptions of kingship rooted in the symbolism of Vi≤§u and ∫iva were associated with different regions of the Khmer lands, the south and north respectively. This general pattern persisted during the seventh century, though in diminishing – or increasingly mixed or integrated – form. Numerous important images of Vi≤§u and his avatars were erected during the middle to late seventh or even early eighth centuries at Phnom Da in southern Cambodia. Meanwhile, in the north, the dedications of ∞Ω¡navarman were predominantly oriented towards ∫iva in the form of li¶gas and, at his capital (∞Ω¡napura), under the particular epithets of PrahasiteΩvara and Gambh#reΩvara. ∞Ω¡navarman's interest in ∫iva follows that of his predecessors in northcentral Cambodia; his father, Mahendravarman and uncle, Bhavavarman I were the ‘Dangrek chieftains’ known to have dedicated numerous li¶gas in what are today northern Cambodia and adjacent areas of Thailand and Laos. Vickery emphasises the peaceful succession and continuity of this dynastic line which, beginning with Bhavavarman I, seems to have maintained its core realm in the area around Sambor Prei Kuk in what is now Kompong Thom.54 It is also during the period of ∞Ω¡navarman and his immediate successors, however, that the first images and epigraphical mention of Harihara appear. As noted above, at least four of these inscriptions belong to the roughly twenty-year reign of ∞Ω¡navarman I. Likewise, at least three images of Harihara would seem to date to the seventh century, with what is probably the earliest extant Khmer image of Harihara – that of Sambor Prei Kuk – firmly associated with ∞Ω¡navarman's northern capital at ∞Ω¡napura. It is probably no coincidence that the only two inscriptions to mention dedications of all three deities (Vi≤§u, ∫iva and Harihara) – those of Vat Po (K.22) and Poñ¡ Hòr (K.21) – most likely date to the 620s and the 630s respectively or, in other words, to the reigns of ∞Ω¡navarman and (in the case of the latter) perhaps to the reign of his son and successor Bhavavarman II. This sudden interest in Harihara during the middle of the seventh century corresponds to the political interests or, perhaps more accurately, the territorial aspirations of ∞Ω¡navarman, Bhavavarman II and Jayavarman I, all of whom seem to have maintained similar realms with control strongly held in Kompong Thom (and Prey Veng) but ‘decreasingly exerted toward the south and southwest where local elites merely evoked [their] suzerainty while maintaining their own local authority’. The distribution of the inscriptions of ∞Ω¡navarman and Jayavarman indicate, as Vickery makes clear, that both rulers endeavoured ‘to maintain administrative control over certain coastal areas which would have been ports of Funan’. He even suggests that ∞Ω¡navarman ‘maintained’ 53 See Brown’s discussion of Preangkorian kingship in Dv¡ravat# wheels, p. 196. Trade-related arguments from a much broader perspective are in the pioneering work of Frederik Bosch, ‘The problem of the Hindu colonisation of Indonesia’, in his Selected studies in Indonesian archaeology (The Hague: Martinus Nijhoff, 1961), pp. 10-11; and Jacob van Leur, ‘On early Asian trade’, in his Indonesian trade and society (The Hague : W. Van Hoeve, 1955), p. 107. 54 Vickery, Society, economics, and politics, pp. 330-6, 339. ‘Gambh#reΩvara’ is also attested at Ak Yom near Angkor (K.749/674) and at Ba Phnom (K.53/667) (p. 150). ≠Ω, ∞      39 Funan and was himself responsible for some of the missions to China that are recorded as having taken place during the period corresponding to his reign.55 The inscriptions of Vat Po and Poñ¡ Hòr were located in the far south and are reflective of ∞Ω¡navarman and Bhavavarman II’s efforts to impose their authority in these areas. In order to do so they invoked not only ∫iva, the deity associated with their capital, but also Vi≤§u, a deity that was much more popular in the region they were seeking to control. It should also be remembered that it was this area along the coast that gave rise to many of the early images of Vi≤§u, apparently the first Indian deity to be represented anthropomorphically in stone in Southeast Asia. It was probably the four-armed mitred image of Vi≤§u that served as the template for the new image of Harihara. This may explain the fact that the iconography of Vi≤§u tends to be dominant in Khmer images of Harihara, a deity that seems to have arisen in an initially more ∫iva-oriented context at ∞Ω¡napura. (Similarly, throughout Indian history, Harihara was most often employed in a ∫aivite context.) Harihara, then, served as a visual expression of the integration of varying regional styles of rule rooted in the symbolism and power of ∫iva and Vi≤§u. This would also explain the relatively large numbers of Harihara images that appeared throughout Cambodia during the second half of the seventh century and the first half of the eighth, a period that on the basis of ‘steady investment in art and architecture’, Vickery convincingly characterises as one of consolidation, ‘political stability and continuing wealth accumulation’.56 All of this was built on foundations initially laid by ∞Ω¡navarman and revealed by his extensive building programme and large corpus of inscriptions, both all the more remarkable for being the earliest securely dated Khmer material. 55 Ibid., pp. 337, 342-3, 350; quotations from pp. 337 (decreasingly exerted) and 339 (Funan). The location of Jayavarman I’s capital remains unknown, but Vickery thinks it highly unlikely that he ruled from ∞Ω¡napura; his discussion of the issues and possibilities is on pp. 350-6. 56 Ibid., pp. 390-2.