The first millenium in Thailand

The first millenium in Thailand

Buddha in meditation, from Khok Pip district, Prachinburi. Prachinburi National Museum.

The first millenium in Thailand

The time in the tenth century is a period within which developments can be meticulously followed in China, in most of India, and even in neighboring Cambodia, thanks to the abundance there of both dated inscriptions and sacred lithic architectural complexes. Region to region, they varied in nature. Knowledge of later Thai history promotes a focus on central Thailand and on the political entity Dvaravati in the seventh and eighth centuries, especially at the sites of Nakhon Pathom and U Thong. There is much to be said both for such a concentration and for a more multifocal viewpoint, in which each region is allowed to speak for itself throughout the period as well as when it seems to have the most to say (the Mae Klong basin, at Khu Bua, in the second half of the seventh century; the Mun basin, at “Prakhon Chai,” around 800; the Chi basin, at Fa Daet, in the ninth century, and so forth). Given the present state of understanding, a survey of art and architecture must make of the material what it will, and methodological consistency is not achievable.

The first millenium in Thailand

Dharmacakra, from U Thong. Detail. U Thong National Museum.

Indianization
If indeed by the end of the first millennium B. C. there were enclaves with stronger cultural and economic ties to a milieu of international traders than to local political structures, the problem remains of establishing the relationship of such enclaves to the less internationally-directed communities. Archaeology may establish whether the distinction is valid and, if it is, contrast the material culture of the two societies. What it may not be able to do is to clarify the nature of ideological exchange. At precisely what point does the local ruler endeavor to bring into his entourage a foreign advisor who promotes the adoption of Indian religions and titles? What are the points of congruence between native beliefs and the foreign religion? Or is it the leaders of the enclave who “go native”?
An ivory comb excavated at Chansen (Takhli district, Nakhon Sawan) in central Thailand (fig. 8) makes an interesting case study, especially when seen against the background of the bronze bowls of Ban Don Ta Phet (fig. 7). Piriya Krairiksh once considered the comb a local product but dated it to the fifth century, not unreasonably seeing in it connections with the murals of Ajanta.
To such views, the excavator, Bennet Bronson, responded that the radiocarbon dates from the layer in which the comb was found make a date later than A. D. 300-350 impossible and that “the comb is in fact more securely dated than the great majority of early Indian objects.” Indeed, it seems that the motifs on the comb can be paralleled at Amaravati and that a third-century date is entirely possible. If the culture of South and Southeast Asia is viewed as an encompassing entity, the engraved design on the comb can be understood as a development of that on certain incised bowls. Within Thailand itself, however, the comb seems to represent something of a cultural loss. The unexpected merger at Don Ta Phet of Chinese, Dong Son, and Indian elements (fig. 7 and pl. 2) is extraordinary; if the art of the region had continued to develop in the same direction, both its independent qualities and its generic ties to China would be far more evident. The auspicious symbols on the comb, on the other hand (sun, vase, cornucopia, snvatsa, parasol, conch, flywhisk, moon), belong entirely to the Indian tradition.
Something of the world of which the comb was a part can be reconstructed from archaeological excavations and written records. Chinese sources tell of the kingdom of Fu-nan and permit the history of the kingdom to be divided into two periods. The first period extends from the first century A. D.—from the time of the legendary marriage of an Indian named Kaundinya to a local princess—to the first half of the fifth century. Within the first period, historical evidence of a relationship between Funan and the peninsula (and possibly central Thailand) is furnished by a passage describing conquests in the early third century by the Funanese king called in Chinese Fan Shih-man. In the first half of the fifth century there was a “second Indianization” and a second Kaundinya, said to be an Indian brahman who arrived in Fu-nan by way of P’an-p’an, a state on the peninsula. This second period lasted until the rise and expansion of a new state in the late sixth century and early decades of the seventh. In the standard view, Fu-nan was a Mekong Delta state with a cosmopolitan seaport at Oc-eo and an interior capital at the foot of the mountain Ba Phnom—the place later Khmer incriptions called Vyadhapura.

The first millenium in Thailand

Figure 8. Drawing of the design on the ivory comb excavated at Chansen. Fine Arts Department of Thailand.

Phase II is that of the ivory comb. No evidence for inhumation burials was found, and the people—unlike those of Ban Don Ta Phet—may have already cremated their dead as a result of influences from India. Some of the pottery may have Indian connections; some is comb-incised, like that of Oc-eo and the Plain of Jars.
Both Phases III and IV at Chansen are characterized by artifacts similar to those found at Oc-eo and at another central plains site, U Thong. Earthenware stamps which may have been used for printing designs on fabrics; small bronze bells decorated with filigree spirals; gold jewelry of various kinds, stone bivalve moulds for making this jewelry; and a type of coin or medal decorated with a trisula-like design and known to archaeologists as ‘Fu-nan’ coins despite their probable Burmese origin.” Bronson wrote:
Such an impression is reinforced when we consider the number of objects in Chansen III and IV which seem to be actual imports from overseas. About one-seventh of one per cent of the site’s surface was excavated; the objects were found widely distributed through numerous separate trenches whose locations were chosen randomly with respect to the protohistoric strata. Yet those strata produced two Chinese artifacts, one possibly Burmese artefact, several objects which might be from Oc-eo, and no fewer than eight metallic blackware bowls, at least two of which are close enough in paste to the Hanbantota examples to have actually come from Ceylon. The indicated volume of extra-regional trade is very large for a site as undistinguished and remote as Chansen.
The picture is a complex one. There certainly were Indian immigrants in Thailand, if not necessarily at Chansen, and pottery considerably more Indian in character than that at Chansen has been found elsewhere in Southeast Asia. Indian elements are found at various levels, and there was an evident desire of an elite for foreign luxury goods. The presence of cloth stamps suggests a waning interest in supporting local traditions of cloth manufacture, which may have involved the ikat technique. At the same time, change extended beyond the interest of elites in exotic luxury items; if this had not been the case, the practice of cremation (presumably) would not have been adopted.
If, on the other hand, it is assumed that the archaeological record at Chansen is the upcountry reflection of a truly Buddhist culture in an urban community closer to the coast, then weight can be given to the evidence that has been uncovered.

The first millenium in Thailand

Figure 9. Shrine 25 at Muang Si Mahosot, elevation and plan.

Funanese Art
In A. D. 503, following the ascension of the Liang dynasty in Nanking, Fu-nan sent a tribute mission. Among the objects brought by the envoys was a coral image of the Buddha. In the same decade Funanese monks in Nanking were busy translating Buddhist texts, among them a version of the Asoka legend (T. 2043), a Prajnaparamita in 700 lines (T. 233), and other Mahayana sutras. On another tribute mission in 519, the envoys reported that in Fu-nan there was hair of the Buddha twelve feet in length. The Liang emperor Wu (r. 502-49), a fervent Buddhist, sent an emissary of his own to acquire the hair. If the temple he built in the 520s had not subsequently burned, the nature of the Buddhist art at his court would be less of a mystery. Such Indian motifs as the predella of dancers and musicians.
Buddhism was not the exclusive religion of Fu-nan, as is evident from the Chinese reports, from inscriptions, and from material remains. Most relevant to works found in Thailand is a group of stone images of Visnu; in each the deity holds the conch held with his lower left hand against his hip. Although the latest of these images could date from the first half of the seventh century, the earliest may have been carved in the fifth century or even before. In the central plains, numerous small objects of Funanese type have been found, especially at U Thong, but little or nothing in the way of large-scale sulpture. Stone images of Visnu that may be grouped with the Funanese sculptures have, on the other hand, been found on the peninsula (pl. 3). The “second Kaundinya” had ruled P’an-p’an, a peninsular state, before ascending the Fu-nan throne in the first half of the fifth century, and so there was a shared heritage. Much of Fu-nan’s Indian trade, furthermore, seems to have crossed the peninsula, where the people may have been Mon speakers, as that is the language of a sixth- seventh-century inscription preserved at Nakhon Si Thammarat.
An earlier history is suggested by inscriptions found a few miles to the south. The Bukit Meriam inscription contains the Buddhist jye dharma formula (“all phenomena proceed from a cause . . .”), followed by a second verse beginning ajnanacciyate karma, “through ignorance karma is accumulated.” Another inscription, on which there is an engraving of a stupa, contains the second of the two stanzas and a passage with the name of the sea captain Buddhagupta of Rak-tamrttaka, who evidently hoped that erection of the stone would help effect a successful voyage. This Raktamrttaka may have been that identified as the site of Rajbadidanga in the Murshidabad district of West Bengal, where in the seventh century a Buddhist monastery was visited by HsUan-tsang. Although it has not been identified with a Chinese toponym, another peninsular site, the trading emporium of Khlong Thom in Krabi province, was apparently issuing its own coinage in the course of the first half of the first millennium; the small gold coins, modeled on a silver prototype associated with Pegu in Burma, had a conch on the obverse, a snvatsa on the reverse, and were used for exchange. A number of stone seals have been found at the site, bearing Pallava-type letters.

The first millenium in Thailand

Figure 10. Sketches of ceramic shards from Chansen.

The Funanese and peninsular images of Visnu fall more readily into a group than do images of the Buddha. The Visnu from Chaiya (pl. 3) is a chunky and severely frontal image, different from Indian prototypes because the sculptor was not particularly interested in rounded countours or swelling volumes. Yet it is neither boring nor inept. The interesting relationships of the layered accoutrements are the two-dimensional ones, as can be seen in the incised curves of the dhoti, the heavy verticle pleat, and the sway of the sash, leading into the conch. These and other elements interact in a way that pulls the eye from one part to another and then back again.
Nevertheless it is probable that the work was produced after rather than before 400 and that a northern Indian model played a role. Some of the motifs are very early in type. But the necklace and the placement of two belt-ends on the left thigh are more satisfactorily paralleled in Gupta images of around 400 or later. A fifth- or even sixth-century date seems probable. The other three peninsular images of the group, all found in Nakhon Si Thammarat province, do not particularly help to narrow the date. One, with a small spoked aureole behind the head, would be an alternate response to an Indian model of about the same time; in it, there is a greater interest in bulging flesh. A second has a foliate-patterned mitre that is proportionately very tall, and the lowered right arm holds an attribute.
A stone image of the type, heavily damaged, has been placed in the fifth century, and the full, rounded volumes make such a date possible. Its provenience is unknown, but a stone head with somewhat similar characteristics was discovered in Sichon district, Nakhon Si Thammarat. The date of bronze images of the type has long been a matter of speculation Key examples are an image from Sungai Kolok, probably imported, and another from Nakhon Pathom that may have been locally made. The postures of the attendant figures place the votive tablets in the seventh or eighth century—implying that the Buddha with Amaravati-type robe (in stone and bronze as well as in clay) was a fixed iconographic type that persisted over a number of centuries.
Sacred structures of brick must have been erected in the sixth century, but stronger evidence for their existence comes from terracotta architectural elements and stone Sivalinga than from the ruins of any single site. Parallels for certain motifs, such as the hair tied with a ribbon at the top of the head, can be found in fourth-or fifth-century Champa.

The first millenium in Thailand

Figure 11. Dvaravati town plans. (a) Khu Bua. (b) Muang Bon, Nakhon Sawan. (c) Nakhon Pathom.

One important stone image of the Buddha from Muang Si Mahosot can be assigned to the sixth century (pl. 4). The Buddha is carved in high relief against a flat back slab. His feet are loosely crossed at the ankles, in a fashion associated with the Amaravati tradition (and seen in a stucco from U Thong), and he is flanked by a pair of engraved stupas, reminiscent in shape to the one that appears on the mariner Buddhagupta’s inscription. One of the most intriguing qualities of this image is the way the modeling of the face and body, with their rounded but rather schematic volumes, evokes Chinese Buddhist sculpture of the second half of the sixth century. This is a reminder that a significant component of Southeast Asian Buddhist culture arose from the passage of monks between India and China.
The Cakravartin Dynasty
Sui Wen-ti’s troops entered Nanking in 589, and the Ch’en fell. After centuries of division, China was again ruled by a single emperor. But in 590 there were revolts in Hanoi and Canton, and established trading patterns must have been disrupted. It was at this very time that a new dynasty came to power in Cambodia. In Chinese annals the name Fu-nan disappears; Cambodia became known as Chen-la.
Relations with China were avoided until 616 or 617, shortly following the accession of Isanavarman. Before then two kings ruled, the first being Bhavavarman, the second his brother or cousin
Mahendravarman, the two being grandsons of a figure called the Cakravartin (K. 978) or Sarvabhauma (K. 496). The inscriptions of the two kings have for the most part been found in locations that seem to have been outposts of the kingdom, along its borders, or at the sites of military campaigns. Describing himself as a conqueror, Bhavavarman set up a linga at Banteay Neang in Battambang province (K. 213), and an inscription found at Si Thep mentions his accession (K. 978). The brother or cousin Citrasena, as he was known before his accession, was meanwhile busy establishing lingas in Kratie province along the Mekong (K. 116, K. 122) and south of the Mun River in Thailand’s Nakhon Ratchasima province (K. 514), and into the Chi watershed as far north as Khon Kaen (K. 1102).
Somasarman, brother-in-law of Bhavavarman set up an inscription there (Veal Kantel, K. 359), and in this region a number of small sanctuaries have been found. It is the lintels that provide the most evidence regarding the style of the period. These have a double arc, as do their Indian prototypes, and although none of them can be positively identified with the sanctuary at which Somasarman set up his inscription, they are older than the lintels of Sambor Prei Kuk, the earliest of which date from the reign of Mahendravarman’s son and successor Isanavarman.
Lintels of the “Thala Borivat” style have also been found in Thailand, three in Chanthaburi province (pl. 5A) and one or two in Ubon. The two locations can be seen as standing at the ends of an arc running from the northern to the western frontier of the kingdom of Mahendravarman and Bhavavarman, a frontier otherwise marked by their inscriptions. In Chanthaburi, inscriptions of Isanavar- man were found (K. 502, 503), but perhaps the lintels predate his accession in 616. If that is the case, Chanthaburi may have served as a seaport for the empire of Bhavavarman and Mahendravarman.

The first millenium in Thailand

Figure 12. Phra Phothisat Cave, Saraburi rubbing of the figures.

Sculptural evidence pertaining to the period is less firmly established; no images can be connected with either an inscription or a lintel. A seventh-century inscription found at Robang Romeah (K. 151), not far from Sambor Prei Kuk, tells of the establishment of an image of Visnu during the reign of Bhavavarman, in A. D. 598, but the image has not been identified. At any rate it is images of Visnu that are of primary concern, and images showing him wearing a long robe. One key work is the Visnu of Kompong Cham Kau; because it was found on a Mekong tributary northeast of the site of Thala Borivat it must bear some relationship to developments there. Two other images were found in Kandal province: in the Visnu of Tuol Chuk, the curves of the tightened muscles present an imposingly dramatic effect, and the horse-headed “Visnu-Kalki” of Kuk Trap is notable for its long hips and the bulky, squared-off gathering of cloth between the legs.
Isanavarman came to the throne before A. D. 616/17 and reigned until after 627—perhaps the mid-630s. A number of the structures at Sambor Prei Kuk—his presumed capital of Isanapura—must date from this period. Examples of the Sambor style in the western flank of the kingdom are three lintels found at Prasat Khao N§i (Aranya- prathet district, Prachinburi) and a fine female torso now in Bangkok. These works may date from the 630s, when the region appears to have become independent, at the time of or a little before two local inscriptions of A. D. 637 and 639 (K. 506, 505, the latter Buddhist).
In fact it is not possible to arrange the images chronologically with much conviction, or even to be sure whether it is the presence or absence of certain motifs (the diagonal sash, the knotted belt), certain technical features (reserves of stone between mitre and upper hands), or qualities of modeling that should be taken as the determining characteristics. As the peninsular images and the ones from Muang Si Mahosot are carved from a gray sandstone (probably a graywacke) indistinguishable in appearance, furthermore, it is not certain how many of the images were carved in the regions where they were found. Not all the images need date from around 600; some—especially those with a diagonal sash and knotted belt—may have been carved considerably later in the seventh century. But the distribution of the type must be the result of political, cultural, and commercial ties that were in existence at the beginning of the century.

The first millenium in Thailand

Figure 14. Conjectural reconstruction, the dharmacakra and associated pillar found at site 11, U Thong.

The Muang Si Mahosot image illustrated here (pl. 6) was excavated in 1975 and so was not included in Pierre Dupont’s study of the Visnu images. It is a sculpture of power and delicacy—power as the result of the solidity of the lower part of the body, delicacy as the result of a sensuous delight in making palpable certain aspects of the anatomy, as in the neck. This is the only Visnu of the type with earrings; gold ones were probably attached to the others. It is also the only Visnu to be found at the site of a shrine with significant architectural features. This was a rectangular laterite platform (fig. 9), probably with stairs on the east, and with six wooden columns to support a roof. The form of the sanctuary may not be recoverable, but the evidence presented by the excavators suggests something unique: a platform for offerings on the eastern wall of the sanctuary and an entrance only on the back side. Sanctuaries of comparable form were excavated in Beikthano, Burma.
The largest and most powerful of the peninsular images is the one from Takua Pa (pl. 7). In a sense it is a development of the potentialities of a Gupta image—just as was the Chaiya Visnu (pl. 3)— but in a very different direction. The rounded, well-fed volumes of the bodies of Indian gods become here something more full of tension and more dependent on bone and muscle. The tension is due not merely to the anatomy itself but to structural organization. It is apparent, for instance, that the thick central pleat has an important role in supporting the image; it is this role that gives to the central vertical a concentration of force that allows the taut curves of calves and thighs to play off it so dynamically. Pierre Dupont put this image at the very head of a long-robed series, putting it back—for him— in the sixth century. For Stanley J. O’Connor, on the other hand, the Visnu was the product of a “development in which local preoccupations were expressed by opening up the image so that it stands freely and actively in space,” and hence no earlier than 650 and perhaps more than a hundred years later.
It is apparent that this area flourished in the seventh century, and its history is probably traceable back into the Fu-nan period, to the sixth century and earlier. Long-robed Visnu images have been discovered here as well as numerous Sivalingas, which are similar to ones found in Cambodia. Inscriptions that may belong to this period also suggest links with Cambodia, both on palaeographic and cultural grounds.
Another group of images that should be placed in a comparable timespan is that depicting Ganesa, who was worshiped as an independent deity (“Mahaganapati”—leader of Siva’s troops) in the Phnom Da region, according to an inscription of 611 (K. 557). At least some of the Ganesa images should probably be pushed back into a somewhat earlier period, in the second half of the sixth century. The images have been found at disparate sites, from Vietnam and Cambodia to the peninsula, and in a variety of postures—some seated with legs crossed at the ankles, some in royal ease, others standing. Petrological analysis might reveal more clearly where the Ganesas were carved and whether the pattern of their distribution should be tied to religious proselytization or to a political network.
No group of Buddha images of the early seventh century exists that supports the same degree of patronage or spread of related stylistic ideals as do the Visnu images or the Sivalingas. It is apparent, however, that by the seventh century the Amaravati image type was no longer the sole or chief model and that sources ultimately Gupta in origin were playing a role. One seventh-century stone image of the seated Buddha discovered in Sathing Phra district is closely related to Cambodian examples. In the small image of unknown provenance that Pierre Dupont put at the head of his discussion of Dvaravatl bronzes, the right hand is lowered in vara-mudra and stocky proportions are combined with a slight sway. This is a type that can be seen in Nepal, in a Burmese bronze, and in a bronze from Kedah with bared right shoulder; it was associated with the Buddha Dipankara, first of Sakyamuni’s Buddha predecessors, who appears in such a pose at Nalanda. A bronze discovered at Phong Tuk seems to exemplify a rather different sort of image, one that can be attached to the “Sandalwood” Buddha lineage important in the Far East: the looping folds of the Buddha’s robe are indicated by incised lines, and in the left hand of the standing Buddha, which is extended at about waist height, no robe end appears.
King Isanavarman sent missions to China again in 623 and 628. In 635, P’an-p’an returned to the scene with its final mission. Isanavar- man may still have been alive in 637, but in 638 the rash of missions was quite likely a response to the king’s death and to an altered balance of power. This was not only the year of missions from states in northwestern Cambodia, it marks the first appearance to the scene of Kedah (Chia-cha) and Dvaravatl (To-ho-lo). The implication is that in the two decades prior to this year Cambodia had dominated the political and economic order of the region. In the middle decades of the seventh century new influences arrived, and new styles developed across the map.

The first millenium in Thailand

Figure 15. Groundplans of peninsular shrines. (a) Phra Barommathat, Chaiya. (b) Wat Ko, Chaiya.

The Appearance of Dvaravati
In A. D. 638, 640, and 649 a Southeast Asian state known as To- ho-lo or To-lo-po-ti sent tribute to China, where the T’ang dynasty had begun its rule a generation previously. If this were the only evidence preserved, To-ho-lo would be as disembodied a ghost as are many other ancient toponyms. HsUan-tsang and I-ching, two Chinese monks who traveled in India in the seventh century also recorded the existence of To-lo-po-ti, however, and even in the nineteenth century it was recognized that the local name that lay behind the Chinese transcription was Dvaravati.
Confirmation of the existence and location of Dvaravati eventually came in the form of two silver medals found in Nakhon Pathom. The medals are inscribed with the words sridvdravatisvarapunya, “the meritorious work (or works) of the lord of Sri Dvaravati.” Possibly the meritorious work was the stupa or stupas under which the medals were buried. Additional medals with the same legend have turned up in Lopburi, Chainat, and Uthai Thani provinces. One medal, of unknown provenance, is an indication of the state’s farflung contacts: it has the same words written in Kharoshthi script, which was used in northwestern India, rather than in the Pallava script ordinarily used throughout Southeast Asia. There is no doubt, therefore, that a state calling itself Dvaravati wielded power in the seventh century.
What territories the political entity Dvaravati ruled, and for how long, may never be known. It is entirely possible that the findspots of the silver medals are an indication of the limits of the kingdom. This would mean that Lopburi (with its own medals, marked Lavapura) was—at least for a period—a separate polity. Dvaravatl also issued silver coins, which have a conch on one face, a “temple” enclosing a vajra on the other. The oldest, found at Nakhon Pathom, were evidently intended for use as foundation deposits. A later type with a conch design on both faces was used for trading purposes; hundreds have been recovered at U Thong. Dvaravatl can also be defined culturally, in high art, by such forms as standing images of the Buddha (pl. 11), dharmacakra, or wheels of the law (fig. 14, pl. 5B), and brick stupas with stucco or terracotta ornament. Everyday artifacts, too, can be defined as Dvaravatl, and in the period around the seventh and eighth centuries, both a cord-marked everyday pottery and various fine wares characterize many Dvaravatl sites. Among the fine wares are vessels with stamped designs in the body (fig. 10)—designs similar in character to those found carved in stone or formed in stucco (pl. 13AB). Another Dvaravatl (and pre-Dvaravati) ceramic shape is the spouted vessel, an Indian form introduced into central Thailand and the Northeast (and also China) in the early centuries A. D. For a period of time there was a widespread culture with a degree of homogeneity in its grander and simpler artifacts. There is also a typical townplan (fig. 11b), first established in the pre-Dvaravatl proto-historic period, with circular earthenworks and moat, and sometimes a fair number of religious foundations outside the small central area. Essentially the type is the same in central Thailand and the Northeast. The Chansen moat was built in the “Dvaravati” phase (phase V), after ca. 600. As many as sixty- one towns from the Dvaravati or Lopburi periods have been spotted in the Chao Phraya basin alone, and some 1200 sites (not all necessarily of the Dvaravati period) can be identified in aerial photographs of Thailand as a whole. It is the largest and most important cities (Nakhon Pathom, Khu Bua) in which a somewhat different plan is found (figs. 11a, 11c).
The political borders of Dvaravatl are not known, but the name can be used in a cultural sense. An alternative is “Mon.” There is no doubt that Dvaravatl people spoke Mon, for that is the only vernacular language recorded. To use the word Mon instead of Dvaravatl is awkward, however, for it implies that linguistic and stylistic borders overlie one another. The name Dvaravatl is a real one, and the concept of Dvaravatl culture is a legitimate one, but the story of the rise and fall of Dvaravatl art is elusive. Scholarship is strewn with wildly contradictory opinions. The reasons are simple: not a single work of Dvaravatl art can be dated with any exactitude. The uncertainty of the chronology has made it difficult to deal with readings of Dvaravatl character that depend on views of the development of its art. Dvaravatl is of “great interest,” thought H. G. Quaritch Wales, primarily because of its “lack of any progressive evolution, either through elaboration from simpler forms or through the pressure of a local genius.” But questions about what constitutes a “progressive evolution,” or about whether “local genius” can even be detected in the absence of both outside stimulus and favorable material conditions are beside the point so long as the nature of the development of Dvaravatl art has not been firmly established.

Trends in Seventh-Century Art
The rise of Dvaravatl is merely one of several diverse developments of the seventh century, characterized in general by new influences from India, the absence of any single dominant center, and the increasing importance of Buddhism. In Cambodia a new, intrusive style of architectural decor appeared, that called Prei Kmeng, and a regional school of sculpture developed (it is here argued), that of
Phnom Da. On the peninsula the cult of Visnu does not seem to have totally disappeared, but a small number of sculptures of Visnu, in their way, reflect the arrival of new influences. In one work, a limestone image said to come from Sathing Phra, Visnu’s discus and conch are attached to the edges of the circular nimbus, a feature perhaps derived from the Deccan, and the flattened chest and curved profile formed by the inside of the arms reflect a new aesthetic out- look. The sculptural traditions of Muang Si Mahosot were carried to Si Thep, and a graywacke sculpture of a seated Aiyanar from Si Thep indicates some of the transformations: in this case the anatomical concerns of the Muang Si Mahosot sculptors (as can be seen in pl. 6) have been applied to the creation of the southern Indian deity whose task it is to guard water tanks (and the Aiyanar is said to have been found in one of Si Thep’s ponds). The cult suggests the presence of immigrants. A Mon-language inscription at Khao Wong Cave near Lopburi mentions the placename Anuradhapura. At Isurumuni, in the Sri Lankan Anuradhapura, Aiyanar made an approximately contemporary appearance in the mid-seventh cen- tury. Another artifact suggesting a close link with Sri Lanka—though of uncertain date (possibly 761, the purported date of inscription K. 997)—is a pair of Buddha footprints at Sa Morokot near Muang Si Mahosot. They are carved in laterite, bear cakra on the soles but no other symbols, and extend 3.5 meters in length.

The first millenium in Thailand

Figure 16. Plan of the stone markers at Ban Hin Tang, Nakhon Ratchasima.

Also belonging to the Muang Si Mahosot-Si Thep tradition and equally indicative of significant cultural change is the relief at Phra Phothisat (formerly Phra Ngam) Cave (Kaeng Khoi district, Saraburi), along a route that might be taken between central and northeastern Thailand. Approximately life-sized figures were carved in low relief on a wall (fig. 12). On the left sits the Buddha, his right hand performing a gesture of exposition, his left hand holding an end of his robe. Beside the Buddha is a seated divinity, usually identified as Siva, with flaring hair and a rosary in his left hand. Beside him is a standing Visnu. Held high are the discus and conch broken off in the Muang Si Mahosot image (pl. 6), but the lower hands are crossed, in a gesture of respect—a gesture made by a naga king depicted on a Sambor Prei Kuk lintel. The mitre seems to have some ornament at the bottom and so is not, therefore, exactly like the ones seen in pls. 6 and 7. The figure wears a long robe, but it is treated as diaphanous. Farther to the right is an adorant, his hair in a topknot, and two celestials. The choice of figures is unique, and one straightforward explanation—that the Buddha is merely preaching to the gods—leaves questions about the composition of the figures and about why the carving should have been executed in a cave. There must be other, local themes at work—having to do with royal identification with the Buddha, with the relationship of the Buddha to the Brahmanical gods, and with the cave as a place of birth. A seventh-century Cambodian lintel from Vat Eng Khna shows a king being anointed, and the similarity of pose and throne suggest that the Buddha here is a kind of surrogate for a king. Perhaps the Buddha in the cave is somehow to be understood as the offspring of the attending Hindu gods. Although many of the figures ultimately have sources in the Buddhist caves of western India, the asymmetrical composition evokes certain schemes at Mahabalipuram of the first half of the seventh century. Other features also suggest a relatively early date, if not in the seventh century then possibly even earlier: the fact that the Buddha holds a robe end (a trait that soon passed out of fashion); and the flying figures, related to ones found in low relief in terracotta at Muang Si Mahosot and U Thong.
Stylistic developments within Cambodia in the middle decades of the century left significant traces in the border areas. Toward the end of Isanavarman’s reign or after his death the size of the kingdom was reduced. In A. D. 639 four principalities in northwestern Cambodia sent tribute to China. One of these, Chiu-mi (in the far northwest of the Cambodia of the period) remained independent— or else regained its independence a second time—for together with Fu-na, which may have been in the Aranyaprathet region, it sent tribute to China again in 671. Isanavarman’s successor Bhavavarman apparently conducted a military campaign in the region; in an inscription (K. 1142) from the Aranyaprathet district, he claimed to have have conquered Sambuka—a name there are reasons to believe could refer to Dvaravatl itself (Th. 16). At the top of the inscription are lotus-like designs that are a new element in the decorative vocabulary, and similar motifs appear on two Prei Kmeng-style lintels at the nearby Prast Khao N§i (where there are also lintels in the earlier Sambor style). A third Prei Kmeng-style lintel was found at Wat Thong Thua in Chanthaburi. At the ends of a fourth, from Prasat Ban Noi, Wattana Nakhon district, Prachinburi, are outward- facing hamsas—a unique occurrence. In another area altogether a sanctuary of the period still stands at Prasat Phum Phon in Sangkha district, Surin. Near the brick sanctuary tower is a plainer brick cell that apparently had no superstructure; its exquisite lintel has been despoliated. The last related works are now at Wat Sa Kfeo in Ubon province. One is a large stone overdoor, with ornament in relief in eight different zones. Parts of the decor resemble the scrolling vegetation on the Vat Phu inscription (K. 367) of Bhavavarman’s successor Jayavarman I (reigned before 657-after 681). The other work is a stone pedestal incorporating bands of foliage and floral motifs rather like those on Bhavavarman’s inscription.
The exact date of the sculptures assigned to the “style of Phnom Da” is an issue long unresolved. An inscription from after A. D. 1200 (K. 549) connects the Visnu of Phnom Da—and by inference related works from the region of ancient Vyadhapura—with King Rudravarman of Fu-nan (r. 514-after 539). This interesting belief is the basis for the traditional date. Jean Boisselier’s doubts about the validity of a sixth-century date slipped into print in the 1960s, but he never presented his case in full. By and large, the Phnom Da style can be seen as a regional tradition made possible, in part, by the absence of a strong central monarch in the years between the death of Isanavarman (628?) and the accession of Jayavarman I (before 657). One scheme would place the high-relief Krsna Govard- hana, with its echoes of the long-robed Visnu tradition, around 620; the Visnu of Phnom Da and related sculptures around 640; and the Harihara of Asram Maha Rosei—its coiffure and the bulge of flesh below the navel indicating connections with Sri Lankan or Southern Indian traditions—around 660. This Harihara, the two sanctuaries Asram Maha Rosei and Kuk Preah Theat, the lintel at Sambor Prei Kuk N-22 (like those at the just-named sanctuaries, it has a double arc), and the Sambor inscription of Jayavarman I (K. 439) would form a contemporaneous cluster. The elegant Sanskrit inscription of Jayavarman’s court physician (K. 53) provides evidence that a Brahman from Chidambaram in southern India lived in the delta 109 region.
Outside influences evidently played a role in the rise both of the Prei Kmeng and the middle and later phases of the Phnom Da styles. The same must be said about one of the most significant developments of the second half of the seventh century, the spread of Buddhism. One Cambodian Buddhist inscription (Vat Prei Veal, K. 49, from the Ba Phnom area) mentions Jayavarman I and was set up in A. D. 664. Another (K. 163), on the northernmost of three towers at Ampil Rolum (between Sambor Prei Kuk and the Tonle Sap) records gifts to the Buddha, Maitreya, and Avalokitesvara, images of whom must have been placed in the three sanctuaries. This triad is the one the Chinese Buddhist pilgrim HsUan-tsang had observed at the Mahabodhi temple during his stay in India (629-45). Although the inscription is undated, the towers probably belong to the late seventh century. In both inscriptions Buddhist monks are given the title pu cah an, “our elder lord.”

The first millenium in Thailand

Figure 17. Sketches of kumbha stupas. (a) Boundary stone at Hin Khon, Pak Thong Chai district, Nakhon Ratchasima. (b) Boundary stone from near Muang Bon, Nakhon Sawan. c. Terracotta kumbha from Thap Chumphon, Nakhon Sawan, with inscription in characters of the ninth-tenth centuries. In Mon, “This kyak puna [meritorious offering] was made by the ancestors near the vihara.” In Pali: ye dhamma … (d) Bronze reliquary from Na Dun district, Maha Sarakham. (e) Silver sheet with repousse design, from Kantharawichai district, Maha Sarakham. (f) Boundary stone, Phnom Kulen. (g) Boundary stone, from Ban Tat Thong, Yasothon.

Early Dvaravatl
Apparent in early Dvaravatl works are outside connections of two main sorts—one with Cambodian art, as it had developed by the midseventh century, the other with an imprecisely defined peninsular and Indian (or perhaps Sri Lankan) strain. The Cambodian aspect tends to characterize Hmayana work in stone (especially wheels of the law) at Nakhon Pathom. The peninsular aspect is more apparent in the evidently Mahayana terracottas of Khu Bua. The most elusive element in the Dvaravatl background is the role of local tradition.
Robert L. Brown’s exhaustive study of the motifs found on wheels of the law (pl. 5B, fig. 14) suggested that all the motifs, though ultimately Indian in origin, passed through Cambodia before reaching Dvaravatl. It was not the very earliest Cambodian art, like that on the lintel (pl. 5A), however, on which all motifs could be found, only that of the Prei Kmeng style, formed by the middle of the seventh century. It is even possible to point to historical evidence, as Brown did, to explain the connection. A goldplate inscription found at U Thong (K. 964) names King Harsavarman, grandson of Isana- varman; King Harsavarman gave a kosa or mask to a Sivalinga. If this Isanavarman is identified as King Isanavarman of Cambodia (r. mid-610s-?628), his grandson may have been responsible for bringing Cambodian artisans into the Dvaravati territories. Something of this process might be observed by comparing pls. 5A, 5B, and 9A. The Wat Thpng Thua lintel fragment (pl. 5A) of the late sixth or early seventh century can be seen as providing the pattern found on the Dvaravati dharmacakra of pl. 5B.
The Garuda on the lintel provides another connection, one with a Dvaravati Garuda who supports a Buddha and Bodhisattva- like attendants (pl. 9A). Belly, chest strap, earrings, and hands (with nagas or lotus stems) are similar. Although the Garuda on the lintel doesn’t wear a big-bead necklace, a sprite on a Sambor-style Prasat Khao N§i lintel does. The move of a Garuda from a lintel to a stele on which it serves as the Buddha’s vehicle would appear to have required a bold intervention. A possible prototype for this sort of Dvaravati stele is a sculpture of Visnu discovered at Srikshetra, Burma. The substitution of a Buddha for a Visnu, like the Phra Phothisat Cave relief, evokes a world in which adherents of the Buddhist and Hindu divinities either battled one another or gave and took without paying attention to customary distinctions. Buddhist thought does make a place for magical powers akin to Visnu’s: the technical term is bala narayana or Visnu power. An inscription in Chaiyaphum province calls the king—and by implication the Buddha— a sky-traveler (svarrabaha, K. 404), and perhaps related ideas are involved. In the later steles the Garuda was replaced by a less easily recognizable monstrous bird, called Banaspati in Thai, as the result of a modern borrowing (via Dutch scholarship) from Indonesian. The individual elements in the early Buddha-on-Garuda sculptures may be derived from Khmer art, but at the same time a new icono- graphic type has been invented, possibly as the result of a stimulus of which a trace has been preserved in Srikshetra.
The stone wheels of the law (dharmacakra)—which will be discussed in the following section—were set up outside stupas, as had been the case in India. Votive tablets from Khu Bua and the southern site of Khuan Saranrom, however, suggest that at least in that milieu smaller dharmacakra were placed beside Buddha images. A small stone dharmacakra discovered in Yarang district, Pattani, suggests the possibility of an intriguing connection. A bronze wheel was uncovered at Rajbadidanga in Bengal—thought to be the Raktamrttika (Red Earth Land) from which the captain Buddhagupta of the Kedah inscription had sailed. There was also a Red Earth Land (“Ch’ih- t’u”) on the peninsula, known to have been situated somewhere south of Lang-ya-hsiu or Langkasuka and therefore probably south of Yarang district but conceivably in it. Terracotta architectural elements such as false windows found at the Yarang sites exhibit cultural connections with Dvaravatl.
The Yarang sites fall into three groups, at Ban Prawae, Ban Jalae, and Ban Wat. A brick sanctuary at Ban Jalae (“BJ 13”) has an interior circumambulation path (like that of the stone sanctuary of Asram Maha Rosei in Cambodia, for which a date around 660 was proposed above). In the cella a number of small terracotta stupas and votive tablets were deposited. Among the votive tablets are those showing single stupas, groups of three stupas, and the Buddha seated between two stupas, all inscribed in southern Indian characters with either the ye dharma saying (the “Buddhist creed”) or the words khasamanaya nirodha marge ye va (“on the path to annihilation, in accordance with the method”), in Sanskrit. On epigraphical grounds, the tablets can be dated to the first half of the seventh century. In Dvaravati, the Buddhist creed appears frequently, though written with Pali rather than Sanskrit spellings, and the theme of the Buddha between two stupas can be seen, for instance, in the Muang Si Mahosot stele, pl. 4.

The first millenium in Thailand

Fragment of a lintel, from Chanthaburi province. National Museum, Bangkok.

The Chinese evidence concerning Ch’ih-t’u is significant, for it does suggest the presence of a culture that in certain ways may have anticipated Dvaravati. It has long been realized that two Ch’ih-t’u titles rendered in Sanskrit as nayaka and (adhi)pati reappear on the pedestal of a seventh-century Dvaravati image of the standing Buddha (Th. 16). Ch’ih-t’u seems to have stood politically and culturally somewhat outside the orbit of the early seventh-century principalities with Visnu cults. After the accession of the Chinese emperor Yang-ti in 604—perhaps through parricide—there began a period of crazed military adventurism. Aware of the riches that lay to the south, in 605 Yang-ti sacked the capital of Champa, bringing the destruction of more than a thousand Buddhist texts. Ch’ih-t’u welcomed the Chinese in 608 and immediately sent a return mission back to China; two other missions followed in 609 and 610. Thereafter Ch’ih-t’u vanished entirely, never to be heard of again.
It is possible to advance different explanations for Yang-ti’s lack of interest in the major powers of P’an-p’an and Chen-la. Perhaps the fortunes of these states were at such a low ebb that they were not worth bothering with. It was just that Yang- ti wanted to bypass this axis and go where he was welcome. A decade later both states did send missions, P’an-p’an in about 616, Ch’en-la in 616-17. But the brief interest in the Red Earth Land has fortunately meant that a valuable account of the state has survived. On each of the triple gates of the city were “paintings of spirits in flight, bodhisattvas and other immortals.” The buildings in the royal palace consisted of “multiple pavilions with the doors on the northern side” and in front of the king’s couch was a recumbent golden ox. After the king’s body was cremated, his ashes were placed in a golden jar and deposited in a temple. The king’s father, a Buddhist, had abdicated “so that he could preach the Word,” and it was “the custom to worship the Buddha.” But “greater respect is paid to the brahmans.”
A beautiful votive tablet (pl. 8A), of a type found in Ratchaburi and elsewhere, also provides evidence for stressing Dvaravatl’s links with the peninsula. Long thought to depict the Great Miracle at Sravasti—with the lotus created by the Naga kings, and the gods assembled to the side of the Buddha in the moment before the emanation of multiple Buddha figures—the tablet might possibly have some other subject altogether. It may date from the second half of the seventh century. In the upper corners are two gods in discs— perhaps the sun and moon—as found in undoubted Dvaravatl representations of the Great Miracle. The asymmetrical arrangement, the flowing curves of the body profiles, the varying scale of the figures, the textured ground from which figures emerge at an angle characterize mid-seventh-century Mahabali- puram, especially the Varahamandapam. Perhaps the depiction of the sun and moon can be traced to the same source.
Two of the oldest Buddhist Dvaravatl structures are stupa 40 at Khu Bua (fig. 13b) and—at the very center of the largest Dvaravatl city, Nakhon Pathom (fig. 11c)—Chedi Chunla Pathon (fig. 13a). Only the lower parts of these two brick stupas remain. The buildings had superimposed stories, each apparently with its own characteristic schema and dominant features. Because the culminating domes have long since disappeared, it is difficult to get a complete sense of the relationship of one story to another. At both Khu Bua 40 and Chedi Chunla Pathon, re-entrant angles are an important element. The general organization recalls some stupa bases among the ruins of Sarnath but does not seem to be identical to anything there. The insets or reentrant angles must be imagined, in part, as an expressive feature, because of the way they helped create strong contrasts of dark and light in the strong sun.
Of the many terracotta figures excavated at stupa 40 at Khu Bua in 1961, only two appear here (pls. 9B, 10). The head (pl. 9B) must have originally been attached to a figure like that in pl. 10 and so also should be considered as representing a guardian. The broad mouth and lowered lids suggest a fathomless control and calm. This head numbers among the earliest works of Dvaravatl art. (The pair of terracotta figures from Khu Bua commonly identified as Bodhisattvas— one with an antelope skin over the left shoulder, and holding a flask—must date from the same time, if facial modeling can be taken as the primary characteristic.) 9B head recalls forms found at Elephanta in the sixth century. Comprising the headdress is a variety of elements that Dvaravatl was not to retain, turning instead to an increasingly limited vocabulary.
The various fragments from Khu Bua were not fitted together until the mid-1960s, and the standing figure long resisted completion because no one had expected that proportions of the lower part of the body to be so drawn out. The pose and proportions recall Mahabalipuram to some extent. This guardian is not exactly like any of the standing figures on the two votive tablets seen in pls. 8A and 8B, for their rhythms are a little more regular. But the sensibility at work is not so different. The tendency in the Khu Bua guardian for the left side of the torso to become a segment of an arc can also be found in the tablet with many figures (pl. 8A), and body movement in both cases is essentially two-dimensional, parallel to the background surface. Indeed, the overall original effect of Khu Bua stupa 40 must have been rather like that of the votive tablet.

The Dharmacakra
More than three dozen of the typical Dvaravati dharmacakras or wheels of the law are known. Most were made of a stone sometimes identified as limestone, the great majority at Nakhon Pathom. A detail of one of the earliest dharmacakras, found at stupa 2 at U Thong, appears in pl. 5B, and fig. 14 is a sketch showing the supposed original appearance of the dharmacakra found with socle and column outside stupa 11 at U Thong. Many, though probably not all the Dvaravatl dharmacakras were originally raised high on columns in similar fashion. Near the site of Muang U Taphao (Manorom district, Chainat), a round brick base, seven meters in diameter, supported an octagonal pillar surmounted by a wheel. Votive tablets suggest that there was an alternative function—placing a dharmacakra beside a Buddha image. But it was the dharmacakra high on a column, near a stupa, that had ancient Indian roots. Later examples have been found in Godaveri district, Andhradesa, and at Sanchi. The fifth-century example at Sanchi, a copy of the Mauryan type, may have been a direct progenitor of the Dvaravatl wheels.
Pali inscriptions found on the wheels themselves, or on the socle or the supporting column, provide insights into the meaning of the wheel. As suggested by the presence of small stone-carved deer that may have been placed on the socles, the wheels symbolize the content of the first sermon at the Deer Park at Sarnath. There are four noble truths and three aspects of knowledge (nana, Skt. jnana) about the truths: that this is a truth (saccananam), that it ought to be perfectly known (kiccananam), and that it is perfectly known (katananam). Or, in the translation by Peter Skilling:
Insight into into truth, task, and accomplishment each performed four times make up the three turnings and twelve aspects (dvadasakara) that are the wheel of the dhamma of the Great Sage.
The wheels, also, as might be expected, have a solar aspect. From Sap Champa in Lopburi province have come fragments of a pillar inscribed with various passages from Buddhist texts. One verse (the Buddha-udana-gatha) makes clear the solar connection:
Truly, when things [dhamma, “phenomena,” here including the Four Noble Truths] grow plain to the ardent mediating brahman, Routing the host of Mara does he stand Like as the sun when lighting up the sky.
The inscription suggests a relationship between the proper understanding of reality and the physical nature of the sun. At the base of two or three of the wheels there is a figure of Surya, further strengthening the solar connections.
That there also existed dharmacakras which flanked Buddha images is indicated by votive tablets, such as ones found in Phunphin district, Surat Thani, on which the Buddha, seated in mediation, has on his right a small wheel, raised on a platform to the height of his head; on the other side is a stupa. A tablet with a similar arrangement was discovered inside stupa 1 at Khu Bua. One interpretation would be that the wheel is solar, the stupa lunar, and the Buddha more brilliant than either. The study of Dvaravati (and other Southeast Asian) coinage makes it clear that the pairing of the sun and the moon was a widespread theme. There were probably local genealogical associations, having to do with sun and moon dynasties. A sun and moon connection is not certain, however, and there are also votive tablets with two dharmacakras, and others with two stupas.
As mentioned above (p. 62), one of the discoveries at Ban Prawae in Yarang was a small stone wheel with eight spokes, placed on a rectangular socle over a bull capital, the whole less than half a foot high. The Ban Prawae wheel has been dated as early as the fifth century. Whether or not it is so old, it does have a chance of having been intended for placement beside a Buddha image. What has not been established is whether such a placement ought to be distinguished doctrinally from placement beside a stupa and from an emphasis on a Pali-language dvadasakara.’4’2 Another wheel fragment found in the south, at Nakhon Si Thammarat, is made of terracotta; conceivably it, too, was set up beside an image rather than a stupa.
Robert L. Brown studied the patterns on the stone dharmacakras. There are three main motifs used on the felly bands—the lozenge and circle (as in pl. 5B), the volutes and circle, and the rinceau. He believed that all three were dependent upon the Khmer seventh- century repertory rather than directly upon Indian patterns. The earliest wheels, which tend to be the ones carved in the round, are found, interestingly enough at disparate sites—U Thong, Nakhon Pathom, Lopburi, Si Thep, and Chaiya. These wheels have various similar characteristics, but they are not uniform. Nevertheless, at some point in time in the seventh century there must have been significant political or religious intercourse among all these sites. Most of the later wheels, the eighth-century ones, have been found in Nakhon Pathom, which became either religiously or politically iso- lated—or perhaps both. It is true, as Robert Brown has observed (following the lead of Quaritch Wales) that the development is away from “the organic and complicated to the geometric and simplified.” The repertory of motifs grew smaller, and their character changed, it would be safe to say, in other media as well. But there need not have been a qualitative decline in art as a whole; it may have been a matter of creative energies being channeled elsewhere than upon the dharmacakra.
Brown’s intensive study of the dharmacakra has led to a number of intriguing questions. One is whether there was some indigenous concept that the wheel of the law usurped or embodied. On one hand, the solar aspect of the wheel might have associations with royalty; wheels might allude to the notion that a king is a sun on earth, or that his family is descended from the sun. On the other hand, the connection could be more mystical and allude to shamanistic powers of actually visiting the sun (or moon), as found in Taoism and— it has been proposed—incorporated into Borobudur. Other questions relate to the political implications of the distribution of the earliest and the latest dharmacakra. Quite possibly the links among principalities established by the dharmacakra in about the middle decades of the seventh century, in a period when regional divisiveness beset Cambodia, paralleled a linkage evidenced earlier in the seventh century by the long-robed Visnu images. And in both cases, petrological analysis might clarify the question of the extent to which the images and wheels were distributed from a central point. The later wheels, nearly all found in Nakhon Pathom and unlikely to have been made any later than the eighth century, open the way to speculation as to why the quality of workmanship declined and why at some point production ceased altogether.

First Sermon Socle
When the Buddha gave his first sermon—the expounding of the wheel of the law—he had for an audience the pancavaggiya, five companions from his years as an ascetic. At the end of the sermon, the Buddha proclaimed that one of the five had attained understanding. This disciple was ordained and then the other four. Later the Buddha presented a second sermon, proclaiming that no self or soul can be found among the constituents of the person. An important relief found in Nakhon Pathom (pl. 12) appears to show the pancavaggiya before ordination on the Buddha’s left, after ordination on his right. This relief comprises one face of a stone block that evidently originally served as a socle for a dharmacakra. It corresponds to one of the three elements forming the ensemble that appears in fig. 14— to the pedestal placed between the pillar and the lotiform base of the wheel. It is more of a substitute for this form, however, than a development. Although two somewhat related socles exist, they have depictions only of a monster mask, a lion, and a narasimha.m
This socle may be assigned a date of convenience of around 700. Internal evidence supports such a date, for comparison of the ruby- and-diamond motif of the border with the early dharmacakras (pl. 5B) suggests the passage of some decades. On the socle this pattern is somewhat shallower, perhaps somewhat more schematic, perhaps more refined. Evidence of a different sort comes from Cambodia and from East Asia, and in both cases this evidence suggests that the period around 700 was one in which fresh outside influences were reaching Southeast Asia. In Cambodia the style of Kompong Preah succeeded that of Prei Kmeng in about the early eighth century. The lintels of two dated temples, Phum Prasat (A. D. 706) and Preah Theat Kvan Pir (A. D. 716), mark the change. The new lintel type was surely intrusive, and perhaps outside influences were at work in the Dvaravatl territories at about the same time. One motif that according to Mireille Benisti had a limited life span in Khmer art around the beginning of the eighth century was the bilo-bate oval, and this motif appears also on the socle, as the ruby element in the border, replacing the earlier open flower. The presence of the bilobate oval supports a date of circa 700.
As in the early Phra Phothisat Cave relief (fig. 12), the Buddha on the socle (pl. 12) is seated in “European” fashion, the forward edge of his robe hanging down from above his knees; beneath is a undergarment folded in such a way that it forms a panel in the front. The end of the Buddha’s shawl lies pleated over his shoulder. The Buddha in the cave relief held a robe end in his left hand; this Buddha has placed his left hand in his lap. The posture is one also seen in a group of votive tablets found on the peninsula and in Nakhon Sawan province. The composition is very different from that of the votive tablet seen in pl. 8A. Instead of asymmetry, there is balance; instead of a graded scale, a uniform one; instead of isolated figures, ones grouped to form an audience. There is a sophisticated use of texture and of varied height of relief. In sum, the organization seems so different from that of the votive tablet that it is necessary to think of two entirely different modes. One cannot have developed from the other. There is much in the relief that looks forward to compositional schemes found at Borobudur at the end of the eighth century: the sense of entourage or audience, the individualization of the figures, the sharp demarcation between sitting and standing figures. These themes were not developed at Nakhon Pathom, and so there must have been a counterpart in painting established elsewhere in Southeast Asia, one that became the basis for the composition of the Borobudur panels.
The relief must at least have had local precursors, perhaps Buddhist counterparts of the seventh-century Khmer lintel of Vat Eng Khna, with its row of standing figures. At the same time, however, there are some Far Eastern works of art that suggest either that the Dvaravati relief was inspired by a Chinese painting or that a Chinese pilgrim passed through Southeast Asia with an irrecoverable Indian model. The posture of the Buddha, with left hand in lap, for instance, is attested in one of the reliefs from the Pao-ch’ing-ssu Temple in Si-an of 703 or 704, in Japanese tiles from a temple founded in 703, and in a Nara-period embroidered tapestry, one of the National Treasures of Japan. The tapestry depicts Sakymuni preaching, and in the foreground are arranged the disciples. This tapestry also includes cloud space dividers, counterparts of the forms over the heads of the disciples in the relief. With no surviving Dvaravatl painting on cloth, and without precise dates, the relationship of the relief to its Central Asian, Chinese, and Japanese counterparts cannot be fully understood. But it seems probable that there was at least some Chinese role in its genesis, and in the years around 700.

The Standing Image in Stone
As quintessentially Dvaravatl as the wheels of the law are the stone images of the standing Buddha. An example now in Seattle appears in pl. 11. The picture of Dvaravatl sculpture presented so far is a disjointed one. The first sermon socle cannot be a development of the terracottas from stupa 40 at Khu Bua. They belong to different schools and—possibly—different sects. Where does an image like that of pl. 11 fit in?
Although the lower arms of this image have been broken off, as is the case with other early examples, both hands surely performed vitarka-mudra—as had probably been the case in the older Buddha- on-Garuda (pl. 9A). The double gesture became established sometime in the seventh century, for reasons that have never been established. It was accompanied by the symmetrical treatment of the lower part of the Buddha’s robe. In an alternative Dvaravatl type, much fewer in number, the right hand is lowered in vara-mudra and a left hand holds the robe, in accordance with the Dipankara tradition. The double vitarka was not universal, however; sometimes the fingers of the right hand are lowered against the palm in a gesture of beckoning (katakahasta- or ahuya-mudra).
Pierre Dupont made an exhaustive study of the Dvaravatl stone images known to him, and he classified them in groups of images that were made either about the same time or sequentially. The
Seattle image (not discussed by Dupont) might be placed with Group B images, falling a little later than those Dupont (no doubt correctly) considered the oldest. It appears to belong to a stage in which there was a high consciousness of facial proportions, of the way lips and eyes should be properly curved and sharply outlined. The Buddha would postdate the Khu Bua terracottas, but it is less easy to place it in relationship to the first sermon socle (pl. 12), for which a date of circa 700 has been proposed. If somewhat earlier, then the facially differentiated disciples on the socle might be understood as a loosening of restraints. If the Seattle image is somewhat later, then the facial proportions would be a refinement of the vigorous explorations seen in these same disciples, and the image, as a work of about the first quarter of the eighth century, would parallel in its degree of stylization an approximate Cambodian contemporary, the Harihara of Prasat Andet.
At the same time, there needs to be explained the kinship of this and other Dvaravatl Buddha images with Indian sculptures of the fifth-century Sarnath style, as is apparent in the curves of the eyes and the flexion of the lips. There is no evidence that the Sarnath style was established in Southeast Asia in the fifth century and somehow remained alive, nor is it likely that a mere internal refinement of either the Khu Bua terracottas or of such earlier stone images as the Buddha on Garuda (pl. 9A) accidentally gave to the face of the Seattle Buddha a Sarnath look. What is more probable is that the seventh-century revival of interest in the Sarnath type—a revival evident in the stuccos of Site III at Nalanda—was in the second half of the seventh century carried to Dvaravatl. Such a hypothesis suggests a role for the Chinese pilgrims who in the late seventh century traveled between Nalanda and China, with stops in Southeast Asia along the way. As a result of such circumstances a memorable image was created, in which the relationships of eyebrow to eyes to lips have an exquisitely refined subtlety.

The first millenium in Thailand

Figure 18. Octagonal stupas ground plans. (a.) U Thong, site 13. (b.) Fa Daet, site 10.

Chedi Chunla Pathon and its Reliefs
The stupa of Chunla Pathon in Nakhon Pathom was excavated by Pierre Dupont in 1939-40, and the story of its successive renovations became a cornerstone in his study of Dvaravati art and architecture, published in 1959. In 1968, it was discovered that Dupont had stopped excavating just before reaching the most interesting part of the monument, namely a series of stucco panels around the plinth (fig. 13a, pl. 13AB). Subsequently the archaeological evidence was used by Jean Boisselier and Piriya Krairiksh to reach very different conclusions about the chronology and nature of Dvaravati art.
Boisselier proposed that the stucco panels were made no earlier than the late eighth century and were the product of a renascence in Dvaravati art, due to influence from Srivijaya, the maritime Buddhist kingdom with a capital at Palembang in Sumatra. Piriya Krairiksh, who addressed the subject matter of the reliefs in a monograph published in 1974 and then the monument as a whole in his dissertation of 1975, initially favored a much earlier date, no later than the middle of the seventh century. It might appear that a structure built in successive states, with one skin covering another, would offer clear and unambiguous evidence regarding the development of Dvaravati art. Unfortunately, the archaeological evidence can be interpreted in divergent ways. Nevertheless, there are good reasons for dating the most significant aspect of the monument—its stucco reliefs—to sometime in a period running from the later decades of the seventh century into the eighth century.
Pl. 13A shows a part of the southwest face of the monument in the course of the 1968 excavations. The reliefs are now on display in the museum in Nakhon Pathom. Despite what Dupont thought (fig. 13a), it is not entirely certain that the monument ever had quite the appearance shown in the plate. On top of the plinth, reaching out as far as the reliefs, was built a new story (having moldings at the top with a rhythm somewhat like those behind the Buddha’s shoulders on the first sermon socle, pl. 12). But this new story turned the projecting staircases on the middle of each face into false stairs, and Piriya Krairiksh suggested that the reliefs (and the stairs) may have been covered when the new story was built.
The stucco reliefs on the seventy-two panels may not belong to the original fabric, for on the northeast side there are four terracotta panels and one with both terracotta and stucco, perhaps indications that once terracotta reliefs covered the entire perimeter of the monument. Given the proportions of the surviving terracotta reliefs, it is thought that they belong to an earlier period, not merely an initial phase of a single sculptural campaign. At least one of the terracotta reliefs depicts a scene from the story of Maitrakanyaka, a tale found in various Sanskrit avadana collections. Among the stucco reliefs are single figures, such as a kinnara (pl. 13B), in which the stylistic concerns are akin to those seen in the stone Buddha in pl. 11—the shape and modeling of the eyebrows, the newly discovered ideal facial shape, and the lively interaction among the different outlines. The other panels are devoted to narrative scenes, both from canonical jatakas and from jatakas and avadanas known only in Sanskrit texts. It is not yet clear whether this is an indication of the presence of a Buddhist sect different from that responsible for the wheels of the law, with their Pali inscriptions.
Some sense of the general direction taken by Dvaravati art in the course of the eighth century, and a context for Chedi Chunla Pathon, can be acquired by looking at the Great Miracle relief now at Wat Suthat in Bangkok (pl. 14). The man in the left foreground, with his hand before his chest, is King Prasenajit, and on the opposite side are six figures representing the Buddha’s six heretical opponents, the fat naked one being Purana Kasyapa the brahman, who is converted upon seeing the miracle. Celestials behind clouds stand at the level of the Buddha’s throne, and above are the magical twin appearances. At the sides, the standing Buddhas reach out to touch discs— probably the sun and the moon. Most of this accords with the Divyavadana account of the Great Miracle, and the pairing of King Prasenajit with Purana Kasyapa is attested in one of the fifth-century Sarnath steles depicting scenes from the life of the Buddha. A frieze separates the Great Miracle from an upper register in which the Buddha is shown seated in the same fashion as below. Here he is evidently preaching in Tavatimsa heaven to his mother—the female figure seated below the throne to our left.
When this relief is compared to the first sermon socle (pl. 12), it can be seen that the postures of the figures and the composition are comparable. The subtleties of the first sermon relief—its textural distinctions, its variations in height of relief—are absent in the Wat Suthat relief, however, suggesting for it a date somewhat later.
Compositional elements may be similar, for the six heretical opponents in the Great Miracle relief parallel the five disciples on the socle, but there is a loss of sophistication. The Wat Suthat relief makes a cosmological division, with a terrestrial event below, a celestial one above. The incorporation of the frieze inevitably evokes architectural organization, everything above such a plinth in a temple or a stupa symbolically representing a heavenly realm.
The frieze in the relief is divided into symmetrically arranged panels with emblematic single figures. Some of the panels at Chedi Chunla Pathon are of this sort, the ones at the corners and in the middle; the rest are narrative. It may be that a movement toward a preference for single figures can be seen in a broader perspective. A key monument built at Nalanda in the second half of the seventh century was Site No. 2, a temple having a plinth decorated with stone panels, most of them with single figures in keyhole niches. The terracotta figures at temples further east—Antichak (Vikramaslla), Paharpur, Mainamati—are related in type, but seem to be a little later. It cannot be said that at Chedi Chunla Pathon a preference for single emblematic figures—such as are found at these monuments—replaced an interest in narrative. The interest in narrative continued.
Stuccos from site 10 in Khu Bua itself appear later in date than those of the Fa Tho Cave or Chedi Chunla Pathon, and composition and architectural context are obscure. By the time stupa 10 was built at Khu Bua, the religious climate had changed from the time of the terracotta Mahayana Buddhist figures at site 40. The biggest structure at Khu Bua, Wat Khlong, a rectangular platform near the center of the city (fig. 11a), with dimensions of 66 X 22 meters and a lower story of moldings and an upper one of engaged columns, for an extant height of seven-eight meters, may belong to the end of the Mahayana phase. At Khu Bua itself, on the other hand, the culture became more like that at Nakhon Pathom, and it may even be that the political situation in the eighth century was different from that in the seventh. The monument at which was uncovered the largest number of relief sculptures—after the terracottas of stupa 40—is stupa. Among the stuccos from stupa 10, the best known depict five members of an all-girl orchestra; the figures constitute a group of attendants or participants in a scene—and hence must have formed part of a composition somewhat like that of the first sermon socle (pl. 12) and, like the socle, anticipated the compositional schemes of the Borobudur reliefs. Illustrated here (pl. 15) is a single seated figure, a continuation of the bias detected at Chedi Chunla Pathon (pl. 13). The elements in the crown are simpler than those found at Chedi Chunla Pathon, and the reduction to simple triangles is a tendency also evident in the Wat Suthat relief (pl. 14). Inside the triangular elements, instead of a variety of forms (as there had once been at Khu Bua, pl. 9B), there is a reduced vocabulary—in the direction of tendrils with hooked ends. The facial modeling is more vapid than that of the Chunla Pathon stuccos and does not reveal the same concern for the interrelationship of curves evident, for instance, in pl. 13B. Very probably the same could once have been said about the figure’s relationship to its original frame or to surrounding figures.

Snvijaya and the Peninsula
It has been seen how the long-robed sandstone Visnu images and Sivalingas of the peninsula must be connected with the culture of Cambodia’s Cakravartin dynasty in a period extending from the late sixth century until 638 and perhaps beyond. One Visnu image, possibly of limestone, it was suggested (p. 55), provides evidence of a shift in outlook, perhaps the result of new Indian connections in about the third quarter of the seventh century. The subsequent period in the history of the peninsula may be thought of as lasting until 775, the year of the “Ligor” inscription. It was a predominantly Buddhist period, and could be called the “Srlvijaya” period, but the material evidence is much slimmer than is that for Dvaravatl. Some of this evidence has already been mentioned. Other works of art, archaeological discoveries, historical sources, and inscriptions provide a very slender basis for a cohesive account.
Political developments that may have started in the 630s brought the demise of hitherto important states like P’an-p’an, which sent its final mission in 635. The last quarter of the seventh century was the period of the rise of the kingdom of Srlvijaya. The Chinese monk I-ching arrived in Srlvijaya (Fo-shih) in Sumatra in 671, and in 673 he embarked for Tamralipti in India. The return voyage occurred in 685. Some time after that, but before I-ching’s return to China in 695, Kedah apparently became a dependency of Snvijaya. Sumatran inscriptions of the 680s record events connected with the kingdom’s rise to power. A Chinese envoy visited Snvijaya in 683, and it sent missions to China in 702, 716, 728, and 742 (but not again until 904).
I-ching’s account of his own journey and the brief biographies he wrote of the other Chinese pilgrims provide information about the monastic networks of the second half of the seventh century. Nalanda was by far the most significant Indian center. The chief Indian seaports were Tamralipti and Nagapattinam, and so there were opportunities for intercourse with southern as well as northern Indian Buddhists.
The role Chinese and other monks may have had in bringing ideas to Dvaravati has been touched upon, but the archaeological evidence from Kedah itself pertaining to I-ching’s time is somewhat limited. There are a number of simple undecorated sanctuaries along the Bujang River but little in the way of sculpture. In 1976 there was uncovered a cruciform structure that can be compared to Wat Phra Men in Nakhon Pathom, and in 1977 an octagonal monument was excavated that may or may not be related to the octagonal-base Dvaravati stupas. The most concrete evidence regarding the nature of the Buddhism practiced in Kedah is a small stone bar engraved with a verse from the Sagaramatipariprccha (T. 400).
The last Srivijaya mission took place in 742, and the middle decades of the eighth century were the period in which the Sailen- dra dynasty of central Java was becoming increasingly powerful; its control was effected by 778, and then and in the coming decades it was responsible for the great Buddhist monuments of central Java. In 767 the Javanese state Ho-ling sent a mission for the first time in nearly a century; Ho-ling continued to send missions until 818; and subsequently the Javanese kingdom in contact with China was recorded as She-po. The last concrete document concerning Srivijaya in this period is the 775 inscription (Th. 23) known as that of Wat Sema Myang or Ligor—though there is a body of opinion that holds (no doubt correctly) that it came from Chaiya, not Nakhon Si Thammarat. This inscription records the establishment by the king of Srivijaya of three brick shrines containing images of Avalokitesvara, the Buddha, and Vajrapani, as well as three stupas. On the other face of the stele is an unfinished later inscription that may have been carved at the order of the Sailendra king Balaputra, who was exiled from Java to Snvijaya before 860. As Balaputra’s maternal grandfather is thought to have been the king of Snvijaya responsible for the three shrines, it would have been appropriate for him to record a donation there. How large a state this king of Snvijaya ruled in 775 is not known; a good assumption is that it was somewhat smaller than it had been in the late 600s or early 700s and that the rise of Java was accompanied by the waning of Snvijaya power. Unfortunately the three shrines of the foundation have never been identified, although it is possible that one of them is the shrine of Wat Kfeo in Chaiya (fig. 15b), which may have been subsequently modified.
Significant stone sculptures from this period are few indeed; bronze images are more numerous but hard to assign to a particular locality. Votive tablets and archaeological evidence of monumental structures help to fill in the picture somewhat. It cannot yet be demonstrated that the workshops responsible for the Visnu images at some point turned to the production of Buddhist statuary. The Lokesvara from Chaiya that appears in pl. 16 is carved from the same sort of grayish sandstone (graywacke?) as are the Visnu images, which makes continuity conceivable. (It again raises problems in regard to identifying the place of manufacture of all the sandstone sculptures.) The Chaiya Lokesvara and the Takua Pa Visnu (pl. 7) exhibit a comparable interest in making the figure exist in space, and it is possible to imagine the later sculptor smoothing and relaxing the Visnu’s bold muscles, its taut erect body coaxed into a relaxed and svelte flexion.
The Chaiya Lokesvara must be put into some sort of relationship to the Khu Bua stupa 40 Bodhisattva: the antelope skin over the shoulder, the knotted belt, the flexed posture all suggest that the two works owe at least something to a common model. There must also be a connection between this Lokesvara and another from Chaiya—a smaller one, carved from what might be limestone, and probably of a different iconographic type. Because of facial modeling and coiffure, the latter can be compared to one of the Phnom Da style sculptures of Cambodia (a comparison made by George Credes in 1928). These connections suggest a date in about the third quarter of the seventh century.
Probably the right hand of the Chaiya Lokesvara was lowered in vara-mudra, and the left was raised, holding a lotus. It is a posture found in one of the Phnom Da-style images. The reserves on the Bodhisattva’s left side evidently reached out to support the lotus stem while the one on the right may have joined an attendant figure. The Bodhisattva to the Buddha’s right on the votive tablet, pl. 8A, suggests the appearance. This is the type the Sadhanamala, the eleventh- century collection of invocations, calls the Lokanatha. Nalanda images with this posture and an intact lotus stem have been placed in the eighth century. Related stone-carving techniques can be seen in a Chinese Bodhisattva dated A. D. 706. The Indian and Chinese examples suggest that monks returning from Nalanda may have helped inspire the Chaiya Lokesvara, and raise the possibility of a date for it closer to the year 700 than to 650.
The “Lokanatha” was not the most common type of standing image. Among bronzes a number have survived of a type of Lokesvara which at Ajanta is shown with a right hand raised, holding a rosary, and the left hand holding both a flask and the lotus stem. One source may have been the Deccan or Southern India, as exemplified by the Krishna Valley bronze in the British Museum. A second possible source is Sri Lanka. Unfortunately there are too few bronzes, and too little can be said about their place of manufacture, to reach firm conclusions about the relationship between the Nalanda and the southern Indian strains in the standing images of Lokesvara from Thailand.
The peninsula is archaeologically rich, but in excavated Buddhist monuments it is poor in comparison with central Thailand. Four brick terraces on the hill of Khao Kha in Sichon district, Nakhon Si Thammarat, may date from no earlier than the mid-seventh century, but they were Hindu. The Yarang sites described above (pp. 62-63) no doubt continued to flourish. They have yielded architectural elements (such as diaper work) that are paralleled at Chedi Chunla Pathon.
A second area of importance was the region around Chaiya. Trading activities in the eighth-tenth centuries are well attested by the quantities of T’ang ceramics uncovered at Laem Pho. Phunphin district, south of the town of Chaiya, has yielded one of the large sandstone Visnus in Bangkok, two smaller Visnu images, and the torso of what may be a Lokesvara, and at the Surat Thani psychiatric hospital, which lies within the same district, is an important Buddhist site, Khuan Saranrom, where a large cache of votive tablets was found. The stupa base excavated there is somewhat similar in its outline to the platform at Phong Tuk (Kanchanaburi province) and may date from the eighth century. By and large the votive tablets represent types earlier discovered elsewhere. There are tablets with an Amaravati-type robe adjustment, similar to pl. 8B; a variant in which the central standing Buddha has more of a Dvaravati look; tablets of a sort discussed in connection with the first sermon socle (pl. 12), with a European-seated Buddha, left hand in lap, his right hand raised; ones with a mediating Buddha flanked by dhar- macakra and stupa (as mentioned in the section on dharmacakra, p. 69); tablets with a single Buddha meditating or in Maravijaya; and small, bell-shaped votive stupas. A similar group of votive tablets was uncovered in the sanctuary BJ 13 at Yarang.
Two additional types of votive tablet found on the peninsula provide important evidence of religious developments. One depicts eight Bodhisattvas, the other the twelve-armed Avalokitesvara. The explanation for this is likely to be chronological, if these two Mahayana types did not appear until some time in the eighth century, later than the time of the Khuan Saranrom and Yarang groups. Tablets with a central
Buddha in dharmacakra-mudra, surrounded by eight seated figures, have been found in Phunphin district, and the subject matter must be the eight Bodhisattvas of the Astamandalaka-sutra.2006 As the type has an Indian prototype, and its key-hold niche connects it to the seventh- century reliefs at Site No. 2, Nalanda, it can be dated either to the late seventh century, the time of the Chinese pilgrims, or—preferably—the first half of the eighth, at least before the return of the presumed carrier of the Astamandalaka-sutra, the translator Amoghavajra, to China in 746. These tablets have the creed in northern Indian or Nagan characters. The second sort of votive tablet depicts the twelve-armed Bodhisattva Avalokitesvara, sometimes erect, some- times—especially at Malaysian sites—standing with a pronounced sway. Again there is a northern Indian connection, for a stele depicting this rare iconographic form and dating from about the early eighth century was discovered at Nalanda in 1971.
The votive tablets may give a false impression of actual workshop production on the peninsula in the eighth century. One truly indigenous product is likely to be the sandstone Lokesvara from Wat Phra Barommathat, Chaiya, illustrated in pl. 17. This is the work of a sculptor who made no attempt to emulate the gracefulness of works in other media (or earlier sculptures in stone); he lumped together the elements below the waist to ensure structural support. Some of the qualities of this statue can be traced back to stylistic tendencies established earlier—the curve to the left arm, the heavy rounded shoulders. But the jewelry is of a later type and has widespread correspondences; the armlets are a little like those in eighth-century Dvaravati stucco (pl. 15), the cummerbund like that on ninth-century Cham bronze images of Avalokitesvara, and the necklace pendants like those seen in Javanese sculpture. The tiger skin around the waist, a kind of royal cushion associated with the god Siva, appears in both earlier and later sculpture. This stone Lokesvara might well date from a period as early as the third quarter of the eighth century, but as it is an isolated object, its position cannot be ascertained with certainty. If this dating is correct, the beautiful bronze Lokesvara from Chaiya must be considered an imported object.

Srivijaya, Muang Si Mahosot, Si Thep
The true lineage of large-scale sandstone sculpture depicting figures with limbs projecting into space—the lineage of pls. 6, 7, and 16— does not lead to this peninsular Lokesvara, pl. 17, but to the sculpture of Si Thep (pl. 19). It will be seen that it is possible to give shape to certain late seventh-century and eighth-century developments in farflung places by paying attention to parallels in the insular world.
Mention was made above (p. 87) of votive tablets found on the peninsula depicting the twelve-armed Lokesvara standing in a flexed position—a rare iconographical type in favor for a relatively brief period of time in the eighth century. The double incisions on the Bodhisattva’s robe suggest a connection with Sri Lankan practice. The pedestal, consisting of a deeply contracted pair of rising and falling petals supported by a flaring molding and slightly rounded foot, has a profile that can be seen on other Dvaravati bronze images of the eighth century, including images of the Buddha. In the period of the sandstone Visnus in the seventh century, relations between Muang Si Mahosot (the chief site in Prachinburi province) and the peninsula were close. Whether the links in Buddhist practices suggested by this twelve-armed Lokesvara were of the same nature cannot be known.
There is a Chinese toponym, “T’o-yUan,” which might be identifiable as Muang Si Mahosot. It sent tribute to China in 644 and 647 but sometime later became a dependency of Dvaravati. This fits in at least partly with archaeological evidence; no early dharma- cakra has been found in Muang Si Mahosot, for instance, but there is a later one, suggesting an initial period of religious independence from Dvaravati. It is also the case that the sandstone sculpture tradition of the city was subsequently developed not at Muang Si Mahosot itself but at Si Thep, just as might be expected if a Dvaravati kingdom centered further west disrupted Muang Si Ma- hosot’s own workshops. The statue chosen here to exemplify the Brahmanical sculpture of Si Thep is an image of Visnu (pl. 19). Comparison with the earlier Visnu from Muang Si Mahosot (pl. 6) makes it clear that here is a continuous tradition, concerned with the same anatomical features in chest and shoulders, a tradition that was pushed in certain directions by outside influences. The Asia Society sculpture of the southern Indian deity Aiyanar, found in Si Thep but even closer in modeling than is the Si Thep Visnu to the Visnu of pl. 6, suggests—as indicated above (p. 55)—that immigrants from Sri Lanka could have been responsible for stimulating the developments.
The Si Thep Visnu of pl. 19 can be no earlier than the second half of the seventh century and presumably was made at some point in the eighth century. It now seems that after its early Brahmanical activities at the time of Mahendravarman in the late sixth century, the ancient city of Si Thep went through a Buddhist phase. Among the evidence is a group of wheels of the law, considered as dating no later than the earliest wheels from central Thailand.
When the twelve-armed Lokesvara (pl. 18), the stone Visnu (pl. 19), and a gold plaque said to have come from Si Thep (pl. 20) are looked at together, it can be seen that, despite the differences of medium and iconography, the artists had similar concerns. In each case, the pedestal defines a section of space that when extended provides an imaginary envelope for the figure—one that can be stretched, even pierced, but continues to provide a foil for the undulations of the body. In all three works the position of the feet is of great importance, for it is the opening chord in this counterplay between body and imaginary envelope. Such formal concerns were not unknown in an earlier period—the sculptors responsible for the Khu Bua terracotta guardian, pl. 10, and the Phra Phothisat Cave relief, fig. 12, were aware of them—but pls. 18, 19, and 20 reveal a common outlook, and therefore historical linkages, within a certain timespan. Perhaps, though, some of these stylistic tendencies were more tied to Mahayana Buddhist and Brahmanical traditions than to the Hmayana milieu of central Thailand.
It is not only the bronze twelve-armed Lokesvara which (on account of its connections with the votive tablets) has ties to insular developments, for the gold plaque (pl. 20) does too. A silver twelve-armed Lokesvara found in central Java must, on stylistic grounds, be assigned a pre-Sailendra (pre-778) date. It shares with the twelve-armed Lokesvara of pl. 18 the tiger skin around the waist, the pelt over the shoulder, and the long robe with double incisions, but is less attenuated a figure. A related object, also pre-Sailendra in date, is a gold plaque depicting Visnu found at Gemuruh in central Java in 1903. The Si Thep and Gemuruh plaques are evidently descended from similar Pallava models, to which the Javanese example is probably more faithful than is the one from Si Thep. The attendant figure on the latter cannot be given a name, for instance; he is unlikely to personify one of the Visnu’s attributes, since all four are clearly shown in the god’s hands. The parallel figure on the Javanese plaque, however, has a clear identity; he is a humanoid Garuda, with wings and snake-like naga in his hands.
Two images of the standing Buddha provide evidence that the sculptors who produced the monumental Brahmanical figures also created some Buddhist works—ones in which the facial modeling is reminiscent of that of the Chedi Chunla Pathon stuccos while the pose and modeling of the body share qualities with the pl. 19 Visnu. Somewhat later—it would seem—after the demand for freestanding Brahmanical sculptures waned, another Buddhist phase began, one exemplified by the heads of Buddhas and Bodhisattvas from the walls of Thamorat Cave fifteen kilometers west of the city. The relief figures at the cave consist almost entirely of images of the standing Buddha, but there are three Bodhisattvas, one of them four-armed and wearing a short loin cloth, which has been identified as the source for one of the removed heads, a Maitreya with tiered headdress.
These figures, in turn, provide a link to the bronze sculptures of the Prakhon Chai group. A votive tablet fragment found at Si Thep with an inscription in both Sanskrit and Chinese probably dates from the same time as the cave sculptures. Evidently the city was a cosmopolitan center in the period around the late eighth century.

Srlvijaya, Central Thailand
As outlined above (p. 82-83), Snvijaya power waned in the middle decades of the eighth century, just as Central Java was in the ascendency. Snvijaya missions to China stopped after 742, and in 768 the Javanese state of Ho-ling sent a mission for the first time in nearly a century. The peninsular inscription of 775 (Th. 23), which records on one face the establishment by the king of Snvijaya of three brick shrines, is best understood as the product of a kingdom with a reach considerably less than had been the case a half century previously.
Among the sources for central Javanese art of the Sailendra period were earlier Javanese art (like the Gemuruh gold plaque) and “Snvi- jayan” art—as far as the latter can be defined from finds in Sumatra or on the peninsula, or inferred from such Dvaravatr works as the first sermon socle (pl. 12). There were other new stimuli, however, most importantly from Bengal. Sailendra control of central Java was achieved by A. D. 778. In one of the earliest Sailendra foundations of 782—thought to be that of Chandi Sewu—there is mention of a teacher from Gaudi, or Bengal. It is the Bengal connection which is significant for Dvaravatr chronology, for it is possible to isolate a group of Bengali-style bronzes found in Thailand (including pl. 21).
These, in turn lead toward the identification of other objects in which there is a Bengali element. There may have been a moment—say around the 770s or 780s—at which the impact of Bengali influences was especially marked.
One of the Bengali-type bronzes, found in Khong district, Nakhon Ratchasima, appears in pl. 21. Another bronze belonging to the group was found in Kosum Phisai district, Maha Sarakham, and a third was acquired in Lopburi. The Bodhisattva in pl. 21, a unique form of Manjusri, has massively rounded shoulders, upon which are clustered the little beads that form armlets, necklace, and ringlets. The conical headdress is adorned with rosettes and the figure of a Buddha in meditation; its profile is similar to that seen on one of the attendant figures on a plaque discovered at Chedi Chunla Pathon (pl. 23A), and various sorts of conical headdresses in later Dvaravati art can be traced back to such a source.
The twelve-armed Avalokitesvara from Prachinburi province (pl. 18) and the Si Thep gold plaque (pl. 20) provide evidence of connections with the peninsula and Java prior to the time of the Bengali impact—perhaps at a moment before Srivijaya power had begun to diminish in the middle of the eighth century. In central Thailand too, there are features that have a Srivijaya character and make an appearance sometime in the eighth century—or perhaps the ninth (whether prior to, at the same time as, or later than the moment of the Bengali impact it is not always possible to determine). One such trait is the presence of smaller stupas at the corners of a brick stupa, such as at Chedi Chunla Pathon in its third state (fig. 13c) and sites 2 and 9 in U Thong. The much rebuilt Phra Barommathat at Chaiya (fig. 15a)—although a shrine, not a stupa—has related structures at the corners, and may provide evidence of the spread of the convention. Perhaps this building in its original state was one of the sources upon which Sailendra-period architects drew, beginning in the 770s, rather than the implantation of a Javanese architectural form in the Sailendra period. Another new type was the octagonal stupa, seen in Kedah and at site 13 in U Thong (fig. 18a). In this case, although the concept may have spread in the eighth century, the Dvaravatr example at U Thong seems no earlier than the ninth century.
At the U Thong octagonal stupa an important bronze image of the Buddha was excavated in 1963. Sixty-five centimeters in height and cast in two pieces, it is presumably older than or contemporary to the stupa. The type is that seen in pl. 22—right shoulder bare, a slight sway to the body, the right hand exhibiting vitarka-mudra, the left hand forward in a kind of vara-mudra that may be no more than a vestige of a robe-holding gesture. This is what the left hand does in the earliest Southeast Asian examples of images with this pose, found in Cambodia. In early Dvaravatr the type was avoided, and almost all the surviving examples were made of bronze. Compared to the standard-type stone image in Seattle (pl. 11), there is a relaxation of tension: the eyes are similarly double-curved and have a ridge or lip around them, but the curve is less severe; a similar observation could be made about the mouth. The facial shape has subtly altered; it is somewhat less round and more heart-shaped. Two features also characterize terracotta high-relief images of the Buddha found at U Thong (especially site 5) and belonging to the period of monuments with corner stupas: one is the broadly pleated shawl (samghati) that lies on the Buddha’s left shoulder; the other is the gem that surmounts the usnlsa. This gem may be a Bengali-inspired trait. Despite these connections with the town of U Thong, it is by no means certain that the bronze in pl. 22 was cast in central Thailand; the modeling of the brows, for instance, is rather like that seen in a Maitreya from the Northeast (pl. 26). A number of bronzes belonging to this iconographic type were, in fact, made in northeastern Thailand.
Evidence of some of these developments can be seen at Chedi Chunla Pathon in its third and final state (fig. 13c). The terrace platform, with its reliefs (pl. 13), was covered over and upon its upper surface four small corner stupas were built. The standing Buddha images of the original niches of the main body of the monument were replaced with images of naga-protected and pendant-leg Buddhas. A group of four repousse plaques uncovered by Pierre Dupont must have been deposited when the monument was enlarged. One is illustrated in pl. 23A. The Buddha sits on a rectangular throne, his right hand in a teaching gesture. His face and those of the attendants seem schematized in the direction of an inverted triangle, somewhat in the manner of the Khu Bua stucco of pl. 15. The figure on the left has a conical coiffure ornamented with small elements, and a fold of his waist cloth hangs low between his legs. Also associated with State III are decorated bricks, some of which have simple lozenges in reliefs. In a painted one, the foliate forms of early Dvaravati ornament have been turned into stems with hooks at the end—a tendency observed in the triangular ornaments on the crown of the Khu Bua stucco, pl. 15.
The seated Buddhas and attendants seen in pl. 23A and on the other plaques bear comparison with figures on the votive tablets of the type seen in pl. 23B. Here is another Great Miracle, simplified in form when compared to the great relief (pl. 14), and with significant substitutions—instead of the pendant-leg Buddha, for instance, here is a cross-legged one. There is no interest in indicating a ground- plane, no overlapping of figures, no cloud space-dividers (in the manner of the first sermon relief, pl. 12). Instead, each figure is isolated and provided with a kind of platform in space. Some elements derive from the Wat Suthat relief and can be understood as simplifications of it, but outside influence traceable to Bengal seems to be at work at the same time. One of the plaques found at the Triple Gem monastery in Mainamati of about the late seventh century has a ground line that provides support for a kneeling adorant and creates the same sort of shallow stage for isolated figures. There may also be a connection between the conical coiffure seen both in the plaque (pl. 23A) and in the tablet (pl. 23B) and that on the bronze Manjusrr (pl. 21).
Pls. 24 and 25 illustrate two additional objects that can be discussed in the context of a middle-Dvaravatr period, in the eighth and ninth centuries. The stucco head in pl. 24 may be identified as having come from the site of Khu Bua (possibly monument 31). The diadem, an ornamented band with central floral medallion, is different from that seen in earlier Dvaravatr stuccos or terracottas and is more like that in the Bengali-style Manjusrr (pl. 21). Parallels can be found on Srivijayan bronzes of apparent pre-Sailendra date, including one from the Palembang area. In this head there is much that is classically Dvaravati in feeling—the shape of the face, the curve of the eyes—but it has been combined with accoutrements that reveal connections with a larger arena of eighth- and ninth- century art.
By the late eighth and ninth centuries Muang Si Mahosot seems to have fallen fully within the cultural orbit of Dvaravati, but the grayish sandstone (or graywacke) of the older local tradition continued to be used. The naga-protected Buddha illustrated in pl. 25, attributable to eastern Thailand (although brought to Bangkok from Ayutthaya), could belong to the second half of the eighth century. The concept of the mask is not new; it appears on a dharmacakra socle at Nakhon Pathom. This mask may have at least some similarities to the those having no lower jaw on the Javanese-influenced Khmer lintels of the Kulen style, however, pushing the date into the ninth century. The facial shape is somewhat like that seen in the bronze standing Buddha, pl. 22, with marked contraction from temples to chin, and the theme of the naga-protected Buddha can be linked to the appearance of the type at Chedi Chunla Pathon in its third state.
The situation at Muang Si Mahosot in the eighth century is illuminated by a lengthy inscription discovered there—if, that is, the inscription (K. 997) really dates from A. D. 761, and the nearly completely effaced numerals at the head of the text were once 683. A Pali portion consists of praise of the Triple Gem (from the Telakatagatha) while the Khmer portion names the author (Vuddhasira, possibly a monk), dates the inscription to the cyclical year of the ox (the earliest instance of cyclical-year dating known in Thailand), and appears to refer to the establishment of a footprint, possibly the one that was uncovered at Sa Morokot in 1986 (above, p. 55). Here is evidence of a Dvaravati Theravada in apparent close communication with Sri Lanka, as well as of the simultaneous use of the Khmer language.
All these works of art raise questions about the role of Java and Javanese art within Thailand following the rise of Sailendra power in the 770s. The southern Cham kingdom of Panduranga, in which Buddhist traditions were strong, suffered Javanese raids in 774 and 787. Perhaps Dvaravatr experienced similar attacks. A few decades later, Champa itself was acting as an expansionist power, raiding Cambodia in the 810s. It was perhaps at this moment that the Cham-style temple Prasat Damrei Krap was built at Phnom Kulen. By and large, classical Javanese art of the late eighth and first half of the ninth centuries did not have a positive impact upon Dvaravatr. Looked at this way, the celebrated bronze torso of Avalokitesvara discovered at Chaiya must be considered an import from Java, and one that did not produce local progeny because at the time there were no active bronze workshops in the area in a position to follow its inspiring lead. The works discussed in this section may either predate or postdate possible raids from Java in the 770s. If they predate such raids, then perhaps the period around 800 was a fallow one for the Dvaravatr kingdom of central Thailand. If they postdate the 770s, then the art of Dvaravatr around 800 can be looked upon as making use of elements from Bengal, on one hand, and on Srrvi- jayan features that had spread decades previously, on the other. Armies from Nan-chao may have entered the region in the 830s; if they had an effect on Dvaravatr, that might have been yet another blow.

The Northeast
In Chinese eyes, Cambodia was divided during the eighth century into “Land” and “Water” Chen-la. A trip to the capital of Land Chen- la, or Wen-tan, is described in Chinese sources; it lay in modern Laos or northeastern Thailand. Wen-tan sent tribute to China along with Chen-la in 711 and 717; it subsequently sent missions of its own in 753-54, 771, and, for the last time, 799. It remained powerful; that is to say, practically until the time of the consecration ceremony of 802 (according to a later inscription) held for Jayavarman II on Mt. Kulen northeast of the future site of Angkor, a ceremony culminating a generation-long period of territorial aggrandizement.
There is no reason why the art of eighth-century (and even later) northeastern Thailand should not be called “Wen-tan,” despite the ignorance regarding both the location of the capital and the extent of its territories. It may be as legitimate a name as Dvaravati. “Java”— the kingdom remembered as Cambodia’s enemy in the time of Jayavarman II (K. 956)—may also be a legitimate name; it is one that has persisted in regional tradition as a name for Luang Pra- bang. The words Dvaravati (in a cultural sense) or Mon have been used instead, however. There are sufficient debts to and similarities to the art of the central plains to make the name Dvaravati an acceptable one. Numerous Mon-language inscriptions indicate that Mon was the primary vernacular language, at least north of the Mun. Stone inscriptions in both Khmer and Sanskrit, meanwhile, attest to a cultural indebtedness to Cambodia. What above all distinguishes the Northeast is the importance given to Buddhist boundary stones (sima or sema).
Though the art of the Northeast is today less well known than that of the central plains, eventually it will be possible to tell the story of its development in some detail. There are enough boundary stones to permit a history that would extend from the eighth century into the eleventh, and there is also a handful of inscriptions. Two inscriptions provide the name of a kingdom—Canasa (K. 400) or Canasa (K. 949), and others give names of kings. But Canasa may or may not have been Wen-tan. The B§ Ika inscription (K. 400)—one of the inscriptions to mention Canasa—was found at Muang Sema, an ancient town in Sung Noen district of Nakhon Ratchasima. Paleographically it may date back to the seventh cen- tury. (On the other side is an inscription of A. D. 868, recording the establishment of a linga and mentioning that the area was then depopulated.)
A wheel of the law was found at Muang Sema, and so it was in the religious orbit of Dvaravatr. Although this wheel has some unique features (like a monster mask), Robert Brown’s analysis of motifs suggests that it should be placed relatively late in the sequence—after the first sermon socle (pl. 12) and so probably in the eighth century. At Muang Sema, at the site known as Ban Hin Tang, there is also a configuration of boundary stones that are just unfinished slabs of red sandstone, some big and thick, others more slender (fig. 16). A group—as reported in 1975—surrounded a low mound. The chronological relationship of such unfinished stones to the finished ones has not been determined, but it may be surmised that both prehistoric unfinished pillars and the historical sima were memorials to deceased ancestors. Mon-language inscriptions on some of the sima appear to call the erection of the stone a meritorious activity, the fruit of which should go to a relative. At Ban Hin Tang, not all the rocks are in a circle, and here as elsewhere there is some question about whether they necessarily enclosed a building. At the same time, inscriptions identify the stones as sima, although in one such case, the inscription pertains to a single stone, not a set (K. 981).
The earliest figured boundary stones also provide some evidence of associated beliefs. One of the earliest slma must be the one sketched in fig. 17b—not found in the Northeast but near the site of Muang Bon (Phayuha Khiri district, Nakhon Sawan), a small town probably established sometime in the seventh century (fig. 11b). At the bottom, in the low-relief depiction, is a stand, then a pot, or kumbha, and then other elements to which it is hard to give names; for the sake of convenience, they might be called a stalk and a finial. Such pots apparently served as burial urns, and near stupa number 13 at Muang Bon the remains of one were found. Such a practice recalls the secondary burials that were carried out in northeastern Thailand in late prehistoric times (see p. 16-17) and that have also have been found at Fa Daet, a site rich in boundary stones of about the ninth century. There were also many burial urns placed in and around sacred structures at Beikthano in Burma. Furthermore, at Thap Chumphon in the same province of Nakhon Sawan, two later (ca. eighth-ninth century) terracotta kumbha-type stupas have been found (fig. 17c), bearing the ye dhamma formula and, on one, a reference to ancestors. If these kumbha-type stupas served as urns, they suggest that the Muang Bon stele itself can be said to have had a kind of memorial character.
The kumbha stupa became an important element in the art of the Northeast. It is found not only depicted on slmas and on small sheets of silver (fig. 17e) but, three-dimensionally, in bronze (fig. 17d). Such a container was used as a reliquary and placed under the plain laterite base of what was evidently once a stupa. In each case there is a variation upon the units found at Muang Bon (fig. 17b): stand, kumbha, stalk, and finial. The silver sheet (fig. 17e), one of sixty-six found in a jar beneath a laterite platform for a Buddha image at a site in Kantharawichai district belongs to the tradition of the Si Thep plaque (pl. 20) and can be dated to the eighth or ninth century— as should the bronze reliquary from Na Dun (fig. 17d), about fifty kilometers south of Kantharawichai (where the silver plaques were found). The more slender, attenuated—and presumably later—type of kumbha-stupa (fig. 17g) need not postdate the ninth century (p. 111 below). If the Muang Bon stele was itself the source for all the variations, then an important route of intercourse between the central plains and the Northeast must have passed through Si Thep and Kaset Sombun district, Chaiyaphum.
Close together in Kaset Sombun district are a number of different sites with sima, on some of which are important inscriptions. But these incriptions, together with others of the period found at
Kumphawapi district, Udon Thani (K. 981-83), Kuchinarai, Kalasin (K. 511), and at Hin Khon (K. 388-89) in Pak Thong Chai district, Nakhon Ratchasima, do not reveal as much as might be hoped about religious orientation. According to one Sanskrit inscription from Kaset Sombun (K. 404), Cudamam, a high-ranking lady or queen, was ornamented by dharma-filled wisdom (prajna)—a Mahayana quality. Another inscription, concerning an acarya (“master”) named Candraditya (“moon sun”), mentions the Abhidharma, possibly suggesting a Hmayana rather than a Mahayana orientation (K. 965). This inscription is one of the few in Northern Indian letters, as is another Kaset Sombun slma text that dates from 991 and mentions a sugatapratimavuddhaslma, evidently an image of the Buddha set up within a set of boundary stones. One of the Kumphawapi district inscriptions (K. 981) states that a monk (bhiksu) honored by Brahmans erected a stone (sila) which functions as a slma. The Hin Khon inscription was set up by a prince-turned-monk (rajabhiksu) who gave four slma of the best stone (K. 388), along with even more substantial donations. An old sketch of a sima at Hin Khon from a group of six pairs appears as fig. 17 a. It is possible that originally the form at the top was more kumbha-shaped.
This slma culture spread to Phnom Kulen in Cambodia, where there are two mounds surrounded by eight pairs of sima, forming perfect rectangles. On a number of these, kumbha-stupas appear on one face (as in fig. 17f), wheels of the law mounted on kumbhas on the other, in the same sort of pairing remarked upon in votive tablets (above, p. 69). The progressive development of the design of kumbha stupas places the Phnom Kulen slmas prior to the time ofJayavarman’s 802 ceremony; they must mark the southernmost extent of Land Chen-la or Wen-tan cultural penetration. The kingdom’s last mission to China occurred in 799, and an inscription of 791 (K. 244), in praise of Lokesvara, has been found in the Angkor region—evidence of the presence of Buddhists. After completing his conquests, it can be surmised, Jayavarman chose as ceremonial site a place already sacralized by religious activities.

Prakhon Chai
In 1964 several dozen metal sculptures were unearthed somewhere in Buriram province and quickly entered the international art market, where they became known as the Prakhon Chai bronzes. They came, in fact, from Plai Bat Hill, which straddles Prakhon Chai and Lahan Sai districts and is the site of two tenth- or early eleventh- century temples, Prasat Plai Bat (1) and, further west, near the findspot, Prasat Plai Bat (2), in Lahan Sai district. One of these bronzes appears here as pl. 26. Its strengths have been much admired: the grace of the posture; the proportions and placement of the long legs; the precisely thought-out but seemingly casual relationship between the folds of the loin cloth and the belt that holds it up; the fleshy quality to the torso; the aristocratic elegance to the curve of the fingers; the vigorous rhythm of the strands of hair. The illustrated bronze is an image of the Bodhisattva Maitreya, and, although no proper sets of three have been identified, worship was probably centered upon the Avalokitesvara-Buddha-Maitreya triad (cf. p. 60).
Some of the qualities of the Rockefeller collection Maitreya can only be explained by reference to eighth-century Cambodian developments that have not been described here; the mustache, for instance, probably has as predecessors those on the Harihara of Prasat Andet and the Avalokitesvara in the Philadelphia Museum of Art, both probably of the first half of the eighth century. But the Maitreya needs also to be seen against the background of the Prachinburi twelve-armed Avalokitesvara (pl. 18) and the Si Thep Visnu (pl. 19). Part of the power of this bronze derives from the fact that it suppresses the more emphatic curves and rhythms of these works. As a result the body seems less self-absorbed in its own rhythmic flow and more capable of impinging upon the viewer’s space. These works, along with the Cambodian ones already mentioned, go far to explain the ancestry of the Maitreya, and yet some other element seems to be present, too.
This other factor may be the impact of Bengal. The standing figures in the Chedi Chunla Pathon plaque (pl. 23A) and the Great Miracle tablet (pl. 23B) have angular presence that distinguishes them from the smooth-flowing and languid grace of the Avalokitesvara and the Visnu (pls. 18, 19). Quite possibly it is the influence of such objects which accounts for the rather different poses that can be seen in the Prakhon Chai bronzes. At the same time, an image like the Manjusri of pl. 21, which represents another aspect of the Bengal connection and could have been in the region as early as the second half of the eighth century, might have stimulated the exploitation of what, in contrast to the earlier sculptures, are fuller volumes and softer modeling.
These observations may shed some light on the ancestry of the Rockefeller Maitreya, but they illuminate only dimly the position of the Prakhon Chai bronze culture as a whole, and the nature and location of the center where they were made. Technically, the sculptures of Prakhon Chai type are characterized by a relatively high amount of tin, in the range of 14 to 20%, but only after more and more bronzes are subjected to analysis will the geographical range of the workshop tradition become clearer. About eighteen kilometers northwest of Plai Bat Hill, the findspot, is the hill of Phu Phra Angkhan (Nang Rong district), where has been discovered a set of stone boundary stones, each with a standing figure. They are in worn condition, and the faces have been restored, but the figures carry lotus stems, and it seems probable that they represent Bodhisattvas. At least one small Prakhon Chai-type bronze, a Maitreya, has a skirt adjustment—with cloth flaring out to the side— similar to that on the boundary stone figures. But this image is likely to be later than most of the Prakhon Chai sculptures. This skirt adjustment and the rendering of the feet seen on the boundary stones, furthermore, resemble those on the slma at Muang Fa Daet (pl. 27 and below), which depict jatakas and are unlikely to be earlier than the ninth century. The Phu Phra Angkhan slma, therefore, would appear to postdate the majority of the Prakhon Chai bronzes and cannot be used as evidence for the nature of the setting in which they arose. The Mahayana Buddhist community responsible for the slma evidently drew on the skills of craftsmen of disparate backgrounds following the retreat of Wen-tan, in the decades after the re-establishment of Cambodia under the leadership ofJayavarman II. The Mahayana and Buddhist boundary stone culture that had earlier spread to Phnom Kulen and the Angkor area withdrew as Jayavarman encroached upon its territories and was consecrated at Phnom Kulen itself in 802.
Three other sites can be mentioned in connection with the Prakhon Chai bronzes. The first, about 50 kilometers northwest of Prasat Plai Bat, is Muang Ban Fai, where a group of three bronzes of related style were uncovered. A town with a moat, a flattened circle in plan, Muang Ban Fai has the plan of hundreds of other sites in the Northeast and in central Thailand (the type of fig. 11b). No slma have been reported in Muang Ban Fai, but a Dvaravati-period naga- protected Buddha was found there. At a second site, Ban Tanot (Non Sung district, Nakhon Ratchasima), at which was found a giant bronze Bodhisattva head in Prakhon Chai style, there are also uncarved sima—though these may or may not have been set up by the people who worshiped the Bodhisattva. The most distant site is Si Thep, where on the walls of the nearby Thamorat Cave are carved images of the standing Buddha, accompanied by a smaller number of Bodhisattvas. These figures are probably somewhat older than the Prakhon Chai bronzes.
Geographically and chronologically, an important Sanskrit inscription of A. D. 829 (from Kham Thale So district, Nakhon Ratchasima, Nm 38) ought to shed light on these developments, but it can be interpreted in various ways. It is strongly Brahmanical in character and concerns the foundation of an image of the god Harihara at a mountain asrama. But it goes on to describe the establishment at another location of an image of the sugata—conceivably the Buddha. It may be that by 829 a new sort of Cambodian culture had been implanted in the region, but it is more probable that the Wentan/Prakhon Chai culture was still strong. One way to reconcile the written and material evidence would be to stress the small bronze sculptures of meditating hermits who seem simultaneously to be Bodhisattvas and Hindu ascetics. One such bronze depicts a bearded figure with a small Buddha image in front of his jata. Rather similar figures excavated at Si Thep are unbearded and have Maitreya’s stupa on their head.

Sites in the Central Chi Region
Characteristic artifacts include not only Buddhist boundary stones but also giant images of the reclining Buddha. Two lie in the central Chi area, and a third stands on Phu Wiang Mountain in the western hills (pl. 28). It is 3.75 meters long. In the same district (Chum Phae) have been found two inscriptions, one mentioning a punyaksetra, here meaning (suggested Credes) a place of pilgrimage (K. 985). The reclining Buddha, at the top of a mountain, requires an exhausting walk to reach and must itself have been such a place. In its facial modeling and arrangement of curls, and because of the presence of a small gem surmounting the usnlsa, the Phu Wiang image is reminiscent of the naga-protected Buddha with the mask (pl. 25). In its present state it is rather less refined, however, and suggestive of what could be accomplished some distance away from a metropolitan center.
Muang Fa Daet (Kamalasai district, Kalasin) was a flourishing center in the central Chi region. Best known for its figured boundary stones (pl. 27), Fa Daet contained a number of monasteries, and excavations uncovered the bases of stupas, various types of votive tablet, and stucco fragments. The inhabitants may have continued to carry out secondary burials, in urns, into the Buddhist period. Phra That Ya Khfi, one of the largest stupas (fig. 18b), had an octagonal base—a form found also at U Thong. Excavated on the western and southern sides of Phra That Ya Khfi were two boundary stones, one with scenes identified as the Kulavaka-jataka, the other (pl. 27) depicting the Sarabhanga-jataka The religious orientation at Fa Daet appears to have been Theravada. At Wat Non Sila in Khon Kaen province a somewhat later slma, one decorated with an undifferentiated swordlike spire, has a Mon-language inscription that gives personal names, states that the stone constitutes a merit winning dedication, and mentions the coming of the Buddha of the future, Maitreya.
There are various reasons for dating such boundary stones to the ninth century. The development of kumbha-stupas (fig. 17) provides clues to the chronology. The slender attenuated type produced at Ban Tat Thong in Yasothon province (fig. 17g) presumably belongs to the end of the sequence. At the same site is an ablution spout or pranala having at least some connections with ninth-century Khmer style. (And one boundary-stone-shaped inscription found in Yasothon province records a foundation of A. D. 889.) Among the figured boundary stones, those with two or three large-scale figures may be earlier than those that like pl. 27 have tiers of smaller figures. Examples of such larger-figured steles can be found at Phnom Kulen and in Chaiyaphum and Udon Thani provinces. The development of the kumbha-stupas can be understood as a progressive stylization. The changes undergone by the figured sima, on the other hand, might have been a response to an exterior model. The Sarabhanga- jataka stone (pl. 27) might be compared with the votive tablet of pl. 23B. In both instances appear figures of about the same scale, against a plain background, each (with one exception on the boundary stone) in its own space—figures whose vertical superimposition is partly but not wholly translatable into the dimension of depth. Some of the motifs match up as well—the conical coiffure composed of many small units, the Buddha with legs crossed at the ankles. In another, better-known Fa Daet boundary stone a similar sort of offering table appears beneath the Buddha. There are suggestions in the stele of an older form of composition: the cloud barriers echo those of the first sermon socle (pl. 12), as does the single instance of overlapping figures. It would appear that an older compositional format has been transformed by the influence of one newly introduced—a format of which the votive tablet of pl. 23B provides an example from another region.
The Sarabhanga-jataka is the story of the Bodhisattva’s existence as Jotipala, whose skill with the bow and arrow allowed him to succeed in a test, as he defended himself from the arrows shot by royal archers. Other jatakas depicted on the boundary stones of Fa Daet include the Maha-ummagga (546), Khandahala (542), Vidhurapandita (545), Sama (540), and Vessantara (547)—all from the final ten lives known as the Dasajati. Everything depicted could have come—and probably did come—from canonical Pali sources. Pl. 27 and the other Fa Daet slma provide evidence of a Theravada art of about the ninth century, an art of vigor and originality that owes something to both older Northeastern traditions and more recent influences.

Late Dvaravatl: the Cham and Khmer Connections
It has already been pointed out that after the Javanese raids on the Cham kingdom of Panduranga in 774 and 787, there is little substantive evidence regarding the role of Java in the political affairs of the mainland and that there are few indications that local workshops ever endeavored to imitate the classic Javanese styles. A group of sculptures in southern Indian style provides yet another indication of a degree of political and cultural weakness. At Takua Pa, a Tamil inscription erected by a merchant gild called the Manikkaraman has been dated to the time of the Chola king Nandivarman III, who reigned in about the middle of the ninth century. A group of Brahmanical sculptures found at Takua Pa—presumably imported images—probably dates from the same time. The inscription and the images may well be an indication of a power vacuum on the peninsula in the mid-ninth century, when Java itself was in fact still quite strong. There is no body of local work that can be considered a response to the Takua Pa images. Meanwhile, the exiled Sailen- dra king Balaputra, who appears to have resided in Chaiya in the mid-ninth century (see above, p. 83), did not—according to this sce- nario—stimulate a large-scale adoption of Javanese culture. Yet, at the same time, judging from the evidence of Chinese ceramics at Laem Pho and at Ko Kho Khao on the west coast, commercial activities flourished.
On the peninsula, connections with Cambodian developments can only be picked up later, in the tenth century. The kingdom with which cultural relations were most important in the ninth century was apparently Champa. That may place it in the same period as the Takua Pa sculptures—a period in which the peninsula was evidently subject to intrusion and commercial exploitation. The Cham connection may be pushed both backward and forward in time, however. The ground plan of the sanctuary is similar to that of Chandi Kalasan—a central Javanese temple founded by the Sailendras in 778 (but thought to have been subsequently modified). A proposal that may resolve some of these discrepancies is that Wat K£o is in reality one of the monuments erected by the king of Srivijaya in 775 but was subsequently modified. In the later period—the last decades of the ninth century and the tenth—may also be placed the stuccos of Khfiha Sawan Cave in Surat Thani. The stuccos include a pen- dant-leg Buddha with other figures in an architectural setting.
Did the Cham “connection” also help shape the later art of the central plains? After excavating at Phong Tuk in 1927, George Credes noted that a golden flower discovered there had a Cham counter- part. A late-Dvaravat! terracotta demon suggests a relationship with Cham art (pl. 29A). Although the diadem can be considered an elaboration of that in the Philadelphia stucco head (pl. 24), it is unlikely that the busy, curving outlines of the leaves—as on the medallions or at the summit of the head—would have developed without the spread of a type of pattern-making apparent in the vermiculous decor at the late ninth-century Cham temple of Dong-du’o’ng.
A somewhat similar facial modeling and outline to the mustache characterize the silver image illustrated in pl. 30, but the matter of a Cham connection is less straightforward. Although it is not known where it was found, this image is more probably a work of the central plains than of either the peninsula or the Northeast. It may have no single feature that need be explained by Cham influence. Some of its features—like the modeling of the face—should be understood against the background of much older works, such as the first sermon socle (pl. 12). At the same time, however, it seems possible that Cham art served as a stimulus for the rejuvenation that seems to have taken place in the ninth or tenth century, and that without this stimulus, the straight-sided face and particular curve to the jointed eyebrows seen in pl. 30 would not have come into being.
The central pleat visible between the legs has long been recognized as a late Dvaravatl trait. A somewhat related pedestal, on which the petals also are in the process of unfolding away from the base (possibly to be connected with a type seen in T’ang China) characterizes a tenth-century Khmer-style eleven-headed Avalokitesvara found in Songkhla province. The silver Buddha must be somewhat older. Behind its flamey aureole lie such nimbuses as the one on the Manjusri of pl. 21. The form of the flames, however, is characteristically late Dvaravatl. These flattened and incised hooked leaves are descended from elements in the seventh-century repertory, and they can be seen in an (apparently) earlier form on the diadem of the stucco head illustrated in pl. 24. The study of dharmacakras indicates how seventh-century patterns became increasingly geometricized but not how, in other media, flattened and incised hooked leaves were treated with increasing vigor. Such leaves were an important feature of the stucco architectural decoration at Phong Tuk and at U Thong. The Cambodian connection cannot be said to have been one brought about by the imposition of the newly formed Angkorian style upon Dvaravati. Instead, it must have been a matter of a give and take of the sort of which there may be earlier instances, as in the mask at the base of the naga-protected Buddha (pl. 25).
Here is a compositional dynamism and an interest in imparting a nervous energy to the outlines of the forms. Yet another outside influence seems to be at work—quite possibly Pala art. The character of the lower part of the tablet is roughly akin to that of the late ninth-century relief fragment at the Bakong in Roluos, Cambodia, though it lacks the latter’s pattern and texture. Perhaps artists in the two regions were making use of the same sorts of foreign art in different ways.
Another work in which a connection with Pala art may be detectable is the Buddha image from Buriram province in pl. 31. Taken by itself, the face of this image, with its curvilinear, ridged eyes might be considered a Dvaravatl work of the seventh or eighth century. It may be distinguished, however, from traditional older image types on account of the position of the legs (vajrasana) and the right hand (Maravijaya), the rounded, fleshy, modeling of the hand, and the treatment of the mantle (samghatl), which hangs down nearly to the level of the elbows. A date around the late ninth century and a Pala source of around that time are possibilities; if this is the case, then the work may just precede the expansion of the Khmer kingdom into the Mun basin during the time of King Yasovarman (r. 889-900).
In the course of the tenth century the relationship between the Dvaravatl lands and the kingdom of Cambodia changed substantially. There are several different stories, some better understood than others, and all interlocking: the fate of the local traditions in the shadow of Angkorian political and cultural expansion; the creation of a provincial Khmer Buddhist art, and development of a cosmopolitan Buddhist art, culminating in the building of the Bayon by Jayavarman VII at the end of the twelfth century.
The loss of artifacts and the absence of documentation make the story of Dvaravatl art hard to tell and obscure some truths that may have a universal relevance: that there is a torch that is passed from generation to generation (or, to put it differently and perhaps more correctly, artists recognize the beacons of the past and use them to light their own fires); yet only rarely is a momentum sustained in a single place or in a single medium for more than a few generations; and outside stimulus is crucial. Therefore, to concentrate on a single spot or a single medium in Dvaravatl or kindred art and to be disappointed by the degree of creativity is to have had unrealistic expectations. The torch wanders from spot to spot, and if its course cannot always be traced, that does not mean that it has died out. As the art of these centuries comes to be better understood, the paths of the torch will be more easily and exactly followed.