The term ‘Dvāravatī’

The initial source of the name ‘Dvāravatī’ was found in Chinese annals in the late nineteenth century.Footnote ⁹ By the mid-twentieth century this was supplemented by the discovery of a number of inscriptions at sites in central Thailand. The Sanskrit term ‘Dvāravatī’ literally means ‘which has gates’, with Phasook Indrawooth, for instance, suggesting that this refers to characteristics of this culture's urban planning.Footnote ¹⁰

The first historical evidence for the term appears in the seventh century. Two Chinese monks, Xuanzang and Yijing, both refer to a country in Southeast Asia called Duoluobodi and Duheloubodi respectively.Footnote ¹¹ In the nineteenth century, it was proposed that these toponyms were the transliteration into Chinese of the Sanskrit ‘Dvāravatī’.Footnote ¹² Further evidence from Chinese sources comes from the recording of three diplomatic missions sent by a polity named Dvāravatī to China in 638, 640, and 649 CE.Footnote ¹³ The Chinese evidence is significant from two angles. First, the sending of diplomatic embassies to China clearly indicates that at this time there was a political entity that identified itself using the name Dvāravatī and that it was actively seeking ‘international’ recognition. Second, the presence of Chinese monks in Southeast Asia during the period illustrates that Buddhist pilgrims were actively travelling around the region.

Archaeological evidence for the term Dvāravatī initially came from the discovery of two silver coins/medallions excavated at Nern Hin near Phra Pathom Chedi, Nakhon Pathom in 1943 (fig. 1).Footnote ¹⁴ Each of the coins bears an inscription in Sanskrit in Pallava script that reads ‘śrīdvāravatīśvarapuṇya’. Coedès translated this as ‘meritorious act of the King of Dvāravatī’.Footnote ¹⁵ Since then, several other medallions with inscriptions including the term ‘Dvāravatī’ have been found at sites throughout central Thailand such as Khu Bua and U Thong (fig. 1).Footnote ¹⁶

Figure 1. Map of central and northeast Thailand showing the main Dvāravatī sites (© Stephen A. Murphy)

This led to a consensus among scholars that the Dvāravatī political entity was based in the Chao Phraya Basin of central Thailand with its centre most likely located at either Nakhon Pathom or U Thong (fig. 1). Excavations and survey work carried out throughout Thailand over the past forty years have led to more robust definitions of what Dvāravatī material culture primarily consists of. Although objects show some degree of variation between sites and regions, their overall similarity allows them to be viewed as a relatively unified material culture tradition. Objects include several distinctive types of earthenware pottery, with carinated bowls, high-fired orange ware, line and wave ware, relatively high-fired matte-surfaced semi-fine ware bowls, and to a lesser extent kendi, being the most common diagnostic Dvāravatī types (fig. 2). Dvāravatī ceramics incorporate forms and decorative techniques from both local and South Asian traditions and usually consist of open-fired earthenware. Cord or mat-wrapped paddle impressions are common, with line and wave incising or carved stamp impressions also recorded at most sites.Footnote ¹⁷ There is also a distinctive range of moulded clay tablets/sealings, terracotta oil lamps, glass, stone and gold jewellery and iron tools.Footnote ¹⁸ Objects such as large finger-marked bricks, glass, stone and metal jewellery, and silver coins are characteristic of Dvāravatī assemblages, but also occur at contemporaneous sites throughout Southeast Asia.

Figure 2. Common diagnostic Dvaravati pottery types excavated at Kamphaeng Saen. Clockwise from bottom left; sherd with line and wave designs; fragment of carinated vessel found in context dating to the 5th–7th century; fragment of a relatively high-fired matte-surfaced semi-fineware bowl found in context dating to the 5th–6th century; cord-marked sherd typical of Dvaravati pottery but also found throughout Southeast Asia during this period; finger/fingernail-impressed sherd; sherd with ‘hanging triangles’ design (all photographs courtesy of Matthew D. Gallon)

The process of urbanisation has also been studied as an indicator of Dvāravatī culture. Quaritch Wales, for instance, attempted to create a chronology of their evolution over time from more irregular-shaped plans, which he saw as growing out of Iron Age precedents, to more regular plans influenced by incoming Indic ideas of urban planning.Footnote ¹⁹ However, Gallon has pointed out that the chronology of these moats is uncertain at most sites. Therefore it is not possible at this point to develop a clear timeline for their development.Footnote ²⁰ Gallon identifies six site types ranging from irregular to rectangular, but does not see a linear development. Instead they coexist and overlap in time.Footnote ²¹ Most, however, are either semi-rectangular or rectangular in shape, with some such as Si Thep having an extension added at a later point in time. Other consistent features include religious architecture in brick, laterite or a combination of both. Large-scale stupas are present at sites such as Nakhon Pathom and Si Thep while other monastic buildings such as assembly and ordination halls have been identified at Mueang Sema and Mueang Fa Daed.Footnote ²² The presence of these moats and earthen banks, coupled with religious architecture in stone, laterite or brick are therefore taken as key indicators of Dvāravatī culture.

Based on this evidence, archaeologists have identified a central area of Dvāravatī material culture in the Chao Phraya Basin that spread out to encompass the Khorat Plateau of northeast Thailand and to a lesser extent Laos, and peninsular Thailand.Footnote ²³ However, differences and similarities in material culture do not always follow linguistic, ethnic, or political boundaries. The extent of the Dvāravatī material culture, like the art style, is generally accepted to be wider than that of the political entity. It is also possible that this distribution reflects areas of influence or indirect control. Moreover, a common material culture does not necessarily imply a common or homogenous ethnic group.

There is considerable disagreement as to the ethnicity of and language(s) spoken by the inhabitants of Dvāravatī. The most favoured theory is that the majority of people in central Thailand, at least, were ethnically Mon and spoke Mon language. Today, they are based primarily in Mon State in Lower Myanmar. However, Gérard Diffloth has argued that the Nyah Kur people, still present to a very limited extent today in central and northeast Thailand, are the direct linguistic descendants of the Mon of the Dvāravatī period.Footnote ²⁴ However, the epigraphic evidence only indicates that Mon was used for religious purposes and the question remains as to whether it was the dominant vernacular language. As Nicolas Revire has pointed out, in central and northeast Thailand, donative/dedicatory inscriptions were written in either Mon or Khmer while citation inscriptions were in Pali or Sanskrit.Footnote ²⁵

The Dvāravatī art style

The most widely recognised use of the label ‘Dvāravatī’ is in reference to a style of art and architecture. The concept of a Dvāravatī art style originated in 1926 with the publication of Tamnan phraphuttha chedi by Prince Damrong Rajanubhab as part of his formulation and classification of art styles in Thailand.Footnote ²⁶ This work was later translated into English in an abridged form.Footnote ²⁷ This classification system was espoused by Coedès in his guide to the National Museum Bangkok published two years later.Footnote ²⁸ The way the objects were displayed and curated in the museum also followed this classification and by the time the guide was published in 1928 this system had essentially become entrenched in the scholarship of both Thai and international scholarship alike.Footnote ²⁹ The classification of a Dvāravatī art style was further developed by scholars such as Pierre Dupont, M.C. Subhadradis Diskul,Footnote ³⁰ Jean Boisselier and Hiram Woodward, among others.Footnote ³¹ The only challenge to date of this system has been by the Thai art historian Piriya Krairiksh who argued that the art of Thailand should be classified by regional styles and religious affiliations.Footnote ³² This system is also problematic, however, not least because in many instances it is difficult to ascertain the precise religious affiliation of an artefact or sculpture.Footnote ³³

It is with the concept of Dvāravatī art that we encounter a key problem in attempts to define a Dvāravatī period. This concept, is in a sense, a case of academically placing the cart before the horse. Scholars first defined the art style and then extrapolated an historical period from it, as opposed to the other way around. Therefore, from the outset the boundaries of the Dvāravatī period have been set, and essentially restricted by the limits of this art style, even when there has been archaeological evidence suggesting a revision of these dates. Furthermore, the concept of Dvāravatī has become closely associated, or arguably inseparable, from the emergence of Buddhism, and by extension, Buddhist art in the region. However, there is a need to reassess this association. The emergence of Dvāravatī and the emergence of Buddhism are not one and the same thing.Footnote ³⁴ They are interrelated, however, as shown in the overview of the features of Dvāravatī art below.

In 1939 to 1940 Pierre Dupont and the Fine Arts Department of Thailand (FAD) carried out excavations in Nakhon Pathom. Out of this came the posthumous 1959 publication, L'archéologie mône de Dvāravatī. It discussed the excavations and their results, as well as a detailed analysis of Dvāravatī art. This analysis laid the foundations upon which all subsequent scholarship on Dvāravatī art was built. Dupont devised a relative chronology from Dvāravatī Buddha images based on criteria such as the evolution of the monastic dress, the lotus pedestal, facial features, the curls of hair, the ushnisha and the appearance of the lotus bud on top of it.

Some of the most striking characteristics of Dvāravatī sculpture are the posture, the gestures and the drapery (fig. 3). The Buddha's robe is usually depicted in a smooth, highly stylised fashion and has a diaphanous quality to it. It falls in a ‘U-shaped’ design terminating just above the ankles, a characteristic shared with Gupta images of India. The Dvāravatī art style has one unique iconographic expression, that of the double vitarka mudra. While a single vitarka mudra is relatively widespread in Southeast Asian art, it is rare in South Asia.Footnote ³⁵ However, the depiction of a standing Buddha image with both hands raised in this mudra is one found primarily in Dvāravatī art and then continues to be employed in some examples of Lopburi period art.

Figure 3. Three examples of Dvaravati sculpture. Clockwise from bottom left: bronze Buddha from U Thong National Museum with hands in the double vitarka mudra gesture; Detail of a Buddha image from the National Museum Bangkok showing the characteristic facial features such as the broad face with thick lips and thick heavy hair curls; Buddha from the Phra Pathom Chedi National Museum showing the classic Dvaravati U-shaped robe (photographs by Stephen A. Murphy)

Carved stone dharmacakras (wheels of the law), are another hallmark of Dvāravatī sculpture (fig. 4). They combine the South Asian religious and political symbol of the wheel with locally derived decorative motifs.Footnote ³⁶ These objects are often found associated with square-based brick chedi that bear stucco decoration.

Figure 4. Three sections of a Dvāravatī dharmacakra: Clockwise from bottom left, the socle, the cakra, the stambha (these three objects are from separate dharmacakra and were not found as a set) (photographs by Stephen A. Murphy)

The geographical extent of the Dvāravatī art style extends beyond the boundaries of the political entity/entities and material culture. However, the boundaries of all three of these phenomena are somewhat ephemeral and not clearly defined or agreed upon. For instance, Dvāravatī style sculpture occurs in peninsular Thailand as far south as the sites of Chaiya, Nakhon Si Thammarat, and Yarang.Footnote ³⁷

To the northeast of the Chao Phraya Basin, Dvāravatī sculpture and motifs have been found at sites such as Mueang Fa Daed (Kalasin province) and Vientiane province of modern-day Laos.Footnote ³⁸ There are variations between these two regions, however. One of the most notable is the popularity of carved sema stones (Buddhist boundary markers) on the Khorat Plateau. The carvings of scenes from the Jatakas, the Life of the Buddha, and motifs such as dharmacakras, stupas and stupa-kumbhas that are found on these stones clearly incorporate Dvāravatī style imagery; however, the stones themselves are relatively rare outside of the northeastern region.Footnote ³⁹ In the north of Thailand, Dvāravatī style art has been found at the ancient city of Haripunjaya, modern-day Lamphun (Lamphun province).

The wide geographic distribution of the Dvāravatī art style has in the past resulted in some scholars equating it with the boundaries of a conceived Dvāravatī kingdom or at the very least, Dvāravatī political presence.Footnote ⁴⁰ However, it seems more likely that the distribution of Dvāravatī art objects represents the spread of artistic and religious ideas. As monks, Brahmins and artisans travelled throughout the region, they could have carried Buddhism and Brahmanism and subsequently the Dvāravatī art style across political and even cultural boundaries.

Returning to the dating of the Dvāravatī art style, on comparative grounds with India it is hard to argue for it beginning any earlier than the fifth century. At Sarnath during this time a distinctive school of Buddhist sculpture had developed which often depicted the Buddha in a granting boons posture (varada mudra), with diaphanous robes, and body depicted in a flexed posture.Footnote ⁴¹ We see this form transmitted to Dvāravatī in a number of surviving examples found in Thailand.Footnote ⁴² Art historians have generally dated the Dvāravatī pieces to the sixth or seventh centuries onwards due to stylistic features such as the use of incised lines to indicate the Buddha's garments and also considerations such as the time it would take for the art style to spread from one region to another.Footnote ⁴³ Pala influence from Bengal is also detectable in certain Dvāravatī pieces and thus represents a ‘middle phase’ in the art form's evolution with these examples dating to the eighth century.Footnote ⁴⁴ A late phase is detectable by Khmer influence and dates from c. ninth to tenth centuries.

A review of the surviving epigraphic evidence concurs with a sixth to tenth century date range for a Dvāravatī period, with the majority of inscriptions attributed to the eighth to ninth centuries.Footnote ⁴⁵ They can be found on a variety of objects including clay tablets, cave walls, sema stones, dharmacakras and Buddha images. Revire lists five Buddha images with inscriptions, one dating approximately to the seventh century, three to the seventh to eighth centuries, and one to the eighth to ninth centuries on paleographic grounds.Footnote ⁴⁶ Given that the two forms of evidence, stylistic and epigraphic, both point to these pieces being no earlier than the sixth century, with the majority falling between the seventh to eighth, it is hard to argue that Dvāravatī art should be dated earlier than this.

There is limited sculptural evidence from the fifth to sixth centuries, but it is from the south of Thailand, an area not generally associated with Dvāravatī. Furthermore, it is Brahmanical in nature — the earliest known example being a Vishnu image from Chaiya, Surat Thani province, generally dated to the late fifth to early sixth centuries.Footnote ⁴⁷ A number of other Vishnus from the region, including a group from Nakhon Si Thammarat, are also dated to the fifth to sixth centuries.Footnote ⁴⁸ This corpus does at least illustrate that sculpture in the round had arrived in the region by the fifth century. However, unlike Dvāravatī pieces, the earliest examples are still very ‘Gupta-esque’, showing little sign of indigenous traits. It is only from the seventh century onwards that these Brahmanical sculptures begin to manifest a particularly Southeast Asian aesthetic.

Inscriptional evidence and stylistic dating make it very difficult to push the date of the Dvāravatī art style back before the sixth century. However, archaeological evidence now discussed does build a case for seeing a proto-Dvāravatī culture starting from at least the fifth century. Thus from this evidence we may begin to argue for a transition period that starts before the physical manifestation of a sculptural and epigraphic tradition.

Archaeological evidence

Scholarship over the past four decades or so, particularly by Thai archaeologists, has led to the discovery of numerous Dvāravatī sites within the Chao Phraya Basin. Indrawooth for instance records over thirty sites while Gallon lists almost fifty.Footnote ⁴⁹ Karen Mudar's study looks at sixty-three sites based on Supanjaya and Vanasin's survey work.Footnote ⁵⁰ However, at present many of these sites have not been investigated in great detail. Despite this, the data does reflect the extensiveness of this culture in the region. Eight sites in particular have seen sustained archaeological research (see fig. 5) and are discussed in greater detail below. This allows for a more concise concept of proto-Dvāravatī to be developed.

Figure 5. Dvāravatī: Main sites discussed and proposed chronologies

Unlike Barram and Glover's definition which argues that the sculpture and the cultural assemblage of Dvāravatī as a whole be re-dated to earlier centuries (viz. first to fourth CE), my proposed definition of proto-Dvāravatī differs somewhat. I argue that the fully-fledged sculptural tradition cannot be dated earlier than the sixth century. Likewise, the full gamut of Dvāravatī pottery does not seem to appear in the proto-Dvāravatī period. Kendi, in the context of central Thailand for instance, do not appear to be found before the sixth century. Furthermore, in the fourth to fifth centuries some prehistoric forms still persist in the material record. Proto-Dvāravatī, as I define it, is very much a transitional phase out of which develops the Dvāravatī period that follows.

‘Proto-Dvāravatī’ in this definition also indicates a society at a ‘proto-state’ level of development, whereas ‘Dvāravatī’ would represent a ‘state-level’ society. Most archaeological definitions in this regard revolve around centralised control, particularly in terms of economic, social and religious activities.Footnote ⁵¹ These in turn predicate the existence of specialised officials such as administrators, a religious caste (in this case Buddhist monks and Brahmins) and artisans and craftspeople. Archaeologists attempt to ascertain these characteristics by identifying artefacts in the material record that represent these activities. These can range from pottery distribution, to seals and sealings, public building projects and site-size hierarchies. In terms of Dvāravatī this evidence can been seen in the construction of Buddhist and Brahmanical religious monuments, the sculptural tradition in bronze and stone reached by the sixth to seventh century onwards, common pottery types being distributed throughout central Thailand and the creation of large-scale moated sites.Footnote ⁵² It is significant that in this region, all of these characteristics begin to appear from the fourth to fifth centuries onwards. The proto-Dvāravatī period, it is argued herein, is also the timeframe where incipient ‘proto-states’ emerge.

A recently published article by Fiorella Rispoli et al. has established a cultural sequence for the Lopburi region of central Thailand and provides a welcome case study in tracing the rise of social complexity and early states.Footnote ⁵³ Their work indicates that there was an acceleration in the growth of social complexity from the late first millennium BCE. This coincides with the appearance of objects of Indian origin in the material record such as glass, carnelian and agate beads, terracotta ear lobe ornaments and ‘skin rubbers’.Footnote ⁵⁴ This begins to occur in what the authors classify as Iron Age Period 2 (c.200–500/600 CE). They argue that during this period exposure and interaction with India intensified and the process of localisation consequently increased. Like the sites discussed in more detail below, their research clearly illustrates continuity as opposed to rupture in the cultural development of their study area and that the transitions to early statehood began taking place in the third to sixth centuries.

Coming back to Mudar's study, it looks at moated sites within central Thailand based on Supajanya and Vanasin's surveys.Footnote ⁵⁵ The latter's work has been highly influential in Dvāravatī studies due to their proposal that the coastline stretched much further inland during the Dvāravatī period. They argued that Dvāravatī moated sites would have been located in close proximity to this ancient shoreline. However, a recent study by Trongjai Hutangkura has illustrated that Supajanya and Vanasin overestimated the height of the mean sea level.Footnote ⁵⁶ What they argued was the coastline has in fact been shown to be a floodplain.Footnote ⁵⁷ Despite this, Mudar's work remains valid, as her argument does not depend upon the location of the aforementioned coastline. However, the revisions to its position should be kept in mind while reviewing her work.

By measuring their size and agricultural catchment area, Mudar calculates the approximate population of each site and the amount of food each could produce.Footnote ⁵⁸ She then looks at which sites have population levels larger than their food production levels. She concludes that sites that could not produce a sufficient surplus would have had to obtain it from elsewhere, most likely through agricultural tribute.

In an attempt to show possible centralised control, from which we could infer the incipient stages of state-level societies, Mudar organises the sites into a settlement hierarchy.Footnote ⁵⁹ At 602 ha, Nakhon Pathom is the only site to fit within the top bracket of what she defines as a ‘primary centre’. Two other sites, Suphanburi and Praaksrigacha fall into the second category. Mudar further organises the sites into an administrative hierarchy, with Nakhon Pathom again being the only site to fall within the top bracket, classified as a ‘supra-regional centre’.Footnote ⁶⁰ U Thong falls into the third level as a ‘district centre’ which tempers arguments that were put forward in the past that it was the capital of a Dvāravatī kingdom.Footnote ⁶¹

Mudar's study shows interdependence between the sites of central Thailand during the Dvāravatī period. She argues that in the early stages of the Dvāravatī period there may have been smaller-scale competing polities in the Chao Phraya Basin, but that the archaeological evidence at present is not robust enough to ascertain whether this represented any form of unified control or state-level society.Footnote ⁶² However, she does argues for the existence of a centralised authority by the end of the Dvāravatī period based around Nakhon Pathom from c. the ninth to tenth century onwards. However, political power and influence could have shifted over time, and this may explain both U Thong's apparent prominence in the seventh century and its later decline.Footnote ⁶³

It appears from Mudar's settlement hierarchy analysis that there was no centralised Dvāravatī state before at least the seventh century and probably much later. In terms of state development therefore it seems favourable to envision the fourth to sixth centuries CE as a proto-Dvāravatī period. While these sites were most likely interacting with each other, they also appear to be developing largely independent of each other. It is only in the seventh century onwards that we begin to see epigraphic references for kings and settlement hierarchies that begin to point towards a more centralised state.

It seems appropriate to start the review of archaeological sites with Chansen for two reasons in particular. One, it represents the first systematic archaeological excavation of a Dvāravatī site. Two, Bronson's attribution of the earlier layers of this site as ‘Funan’-related is a key issue in the argument over the dating of the ‘proto-Dvāravatī period’ and has been called into question.Footnote ⁶⁴

The site of Chansen, about 30 km north of Lopburi, was excavated between 1968 and 1969 by a joint team from the National Museum of Thailand and University of Pennsylvania Museum (fig. 1).Footnote ⁶⁵ Based on the radiocarbon and thermoluminescence dates, Bronson deduced five phases of occupation: phases II, III, and IV (spanning the first to seventh centuries) represent a proto-historic period, with fully-fledged state development occurring during the transition between phases IV and V, that is, the seventh century.Footnote ⁶⁶ He classified phase III (200–450/500 CE) as ‘Funan’-related, and appeared to see it as the initial phase of early state development. In this, Bronson's interpretation falls in line with the conventional dating of the Dvāravatī period to the seventh century onwards. However, as Barram and Glover have argued, the label Funan for the early phases dating to the third to late fifth/early sixth centuries is less than ideal given the fact that the material actually reflects more similarities with Dvāravatī cultural assemblages. However, Bronson did not have the benefit of the data sets that we have available to us today and could not have known that the material he labelled as Funan was more likely that of an incipient Dvāravatī culture.

Chureekamol Onsuwan Eyre's study of hierarchical and heterarchical social systems in the eastern Chao Phraya Valley included Chansen as one of the twenty-five sites she surveyed in an area of 255 ha.Footnote ⁶⁷ Her research and review of previous surveys in the region indicates that rank size analysis did not indicate the existence of site hierarchy during the period of 200 BCE–300 CE.Footnote ⁶⁸ However, by c. the fifth century CE her results show a strong trend towards a hierarchical settlement pattern around the site of Chansen indicative of a two-level hierarchy.Footnote ⁶⁹ The two-level hierarchy she identifies is clearly indicative of the formative stages of state-level society.

New evidence for a proto-Dvāravatī period has been recently uncovered at excavations at Nakhon Pathom. This site represents one of the largest and most highly developed urban centres during the Dvāravatī period.Footnote ⁷⁰ It covered an area of roughly 3,700 by 2,000 m (740 ha). Situated on the west bank of the Ta Chin River, it is about 60 km directly west of modern-day Bangkok. Two large stupas were discovered at the centre of the settlement: the first, known today as Chula Pathon Chedi, was excavated in 1940 by Dupont, while the second, known as Phra Pathon Chedi, has recently been excavated by FAD.Footnote ⁷¹ Dupont's expedition also uncovered a further stupa, Wat Phra Men, located approximately 1 km outside the moat to the southwest.

Extensive remains such as Buddha images, dharmacakras, and Pali and Mon inscriptions have been found at this site. This fact, along with its size and location, have led many to conclude that it was a capital city of the Dvāravatī polity, particularly from the eighth century onwards.Footnote ⁷² However, after Dupont, no systematic archaeological research was carried out here until 1981, when Indrawooth carried out excavations in Phra Prathon subdistrict.Footnote ⁷³ She unearthed pottery, bronze ornaments, iron tools, spindle whorls and glass beads. With the one radiocarbon date she acquired being inconclusive, she relied instead on pottery and art historical analysis, concluding that the artefacts dated from the eighth to ninth centuries.Footnote ⁷⁴

Further excavations were undertaken at the site of Hor-Ek in 2009 by Saritpong Khunsong as part of his doctoral research.Footnote ⁷⁵ This site is located about 500 m to the northwest of Phra Pathon Chedi. Khunsong's excavations show it was inhabited from the third to eleventh centuries with some of the pottery in the earlier layers showing similarity with material excavated by Bronson at Chansen. This included some burnished ware similar to Phimai Black and which Bronson had dated to the ‘late Funan phase’ (500–600 CE). Other pottery types such as a painted potsherd with a sun pattern resembled material excavated in Phum Snay in Cambodia radiocarbon-dated to the fifth century.Footnote ⁷⁶

Khunsong identifies three phases of occupation. The first dates from around the third to sixth centuries and is contemporary with Bronson's Chansen phase III dated to 250–500 CE. The material record shows similarity with Oc Eo and Angkor Borei in the Mekong Delta of Vietnam and Cambodia respectively. It appears that Nakhon Pathom was part of the maritime trade network active at this period. Based on the material record, Khunsong et al. suggest that this phase should be referred to as ‘Pre’ or ‘Proto-Dvāravatī’.Footnote ⁷⁷ The second phase spans the seventh to eighth centuries and conforms to the more conventional date range of the Dvāravatī period. The third phase runs from the eighth to eleventh centuries and reflects a later Dvāravatī phase.

Khunsong's research indicates that Nakhon Pathom was in existence long before the conventionally accepted dates for the Dvāravatī period of the sixth to ninth centuries. The occupation sequence begins from the third century onwards and by the fourth to fifth centuries appears to have trade contacts with Oc Eo and the ‘Funan’ culture. It seems inappropriate to refer to material dating from the fourth to fifth century as a ‘Funan Period’, as the ceramic evidence from Nakhon Pathom appears to show that there was trade contact only, not a form of overlordship which was presupposed by earlier scholarship.Footnote ⁷⁸ Also as Barram and Glover have indicated, since the 1960s considerably more work has been done, and it is clear that the material that both Bronson and Boisselier (see below) refer to as ‘Funanese’ is in fact much more similar to Dvāravatī material.Footnote ⁷⁹

U Thong, in Suphanburi province, central Thailand, is another substantial Dvāravatī period moated site and shows considerable evidence for both Buddhist and Brahmanical practices (fig. 1).Footnote ⁸⁰ It too has been touted as an important Dvāravatī centre or capital, particularly during the seventh century.Footnote ⁸¹ Excavations at this site were conducted by Boisselier under FAD's auspices in the 1960s which mainly focused on investigating stupas.Footnote ⁸² From 1966 to 1970 systematic excavations were carried out by William Watson and Helmut Loofs which produced five radiocarbon dates.Footnote ⁸³ In 2010, excavations consisting of three test pits were carried out by San Thaiyanonda as part of his Master's thesis for Silpakorn University.Footnote ⁸⁴

Barram and Glover's reappraisal of Dvāravatī dating focus particularly on Watson and Loofs’ excavations due to them obtaining radiocarbon dates. By processing an additional five radiocarbon dates and analysing the pottery types found, Barram and Glover compared them with the material from Chansen. They concluded that pottery, such as those with wave and line decoration, spouted vessels and carinated wares as well as clay lamps, are all indicative of Dvāravatī period wares. These artefacts begin to appear in layers datable to the first through fourth centuries.Footnote ⁸⁵

Thaiyanonda's 2010 excavations of three test pits add additional data to the chronological sequence at U Thong. In test pit 1 on the rampart of the moat he too uncovered pottery sherds with wavy line decoration. By comparison with excavated examples recovered from the Pontian River in Pahang, Malaysia, and Go-Hang and Canh Den in Vietnam, he dates it to the fourth to fifth centuries.Footnote ⁸⁶ Furthermore, a potsherd with a Pallava inscription appears to date to the mid-sixth century and gives clear indication of Indic influence at the site by this period. Test pits 2 and 3 produced evidence of occupation from the more conventional date ranges of the seventh to ninth centuries. There was also evidence of a late Dvāravatī phase of the ninth to eleventh centuries. Thaiyanonda concludes that both the moat and connections with Indic culture date to at least the mid-fourth century CE. Gallon has argued that moat construction was one means of public works instigated by more centralised authorities and is an indicator for early state development.Footnote ⁸⁷ That it appears during the fourth century at U Thong lends weight to the argument that we are seeing ‘proto-state’ level activity during the proto-Dvāravatī period.

A final, tantalising piece of possible evidence comes in the form of a fragmentary terracotta relief moulding of three Buddhist monks with alms bowls, today housed in the U Thong National Museum (fig. 6). Stylistically it has been dated to the Nagarjunakonda school of sculpture. This would give it a date range of third to fourth century CE.Footnote ⁸⁸ It most likely formed part of the façade of a stupa, and has led a number of scholars to conclude that the associated monument must also date to the third to fourth century.Footnote ⁸⁹ This would place Buddhism arriving at U Thong some two to three hundred years before the ‘official’ start date of the Dvāravatī period. It would also put it more in line with the arrival of Buddhism at the Pyu sites in Myanmar (see Stargardt, this vol.). However, while both Boisselier and Indrawooth suggest Buddhism may have been present from the third century onwards, neither seemed to consider the ramifications of this in terms of the classification of the Dvāravatī period.Footnote ⁹⁰ However, as Gallon points out, the age of the monument in question needs to be ascertained by absolute dating methods before any firm conclusions can be reached.Footnote ⁹¹ Even so, this fragment does add to a growing body of evidence for reconsidering the date ranges for the early arrival of Buddhism.

Figure 6. Fragmentary terracotta relief moulding of three Buddhist monks with alms bowls, today housed in the U Thong National Museum (photograph by Stephen A. Murphy)

Barram and Glover's reappraisal of Watson and Loofs’ excavations indicate that U Thong was occupied as early as the first century CE with evidence for Dvāravatī type pottery dating to between the first and fourth centuries. Thaiyanonda's excavations indicate Dvāravatī-style pottery and Indic influence from the fourth century onwards.

The lack of systematic excavations at Dvāravatī period sites has been further addressed recently by Gallon's doctoral research at the moated site of Kamphaeng Saen (fig. 1).Footnote ⁹² His dissertation sets out to gain a better understanding of urbanisation during the Dvāravatī period using the site of Kamphaeng Saen as a case study. Situated roughly equidistant between Nakhon Pathom and U Thong, it represents a smaller-scale site and may have functioned as a satellite settlement of its two larger neighbours. As with its larger counterparts it has clear evidence for the adoption of Buddhism. The discovery of a dharmacakra and its inscribed socleFootnote ⁹³ as well as three Buddha images are indicative of this.Footnote ⁹⁴ Of most significance to this discussion are the results of eight AMS radiocarbon dates obtained from controlled excavations which Gallon undertook in 2009 and 2010.Footnote ⁹⁵ They indicate that the site was occupied from the early fifth century and declined a few centuries later. The most intensive phase of occupation was from the fifth–sixth centuries to the mid-seventh century, with the site being abandoned by the ninth century.Footnote ⁹⁶

Gallon's research has cast new light on the debate regarding the Dvāravatī period. He points out that material culture from the stratigraphic layers dating to the fifth to sixth centuries has Dvāravatī-style ceramics, saddle querns and beads, but is also missing a number of other markers of Dvāravatī culture such as earthenware with stamped impressions as well as blue-green glazed Persian wares.Footnote ⁹⁷ He therefore prefers the labels ‘Early Dvāravatī’ or ‘Proto-Dvāravatī’ for the fifth to sixth century layers. Based on stylistic dating, he also notes that it is only in the seventh to eighth centuries that we see significant investment in Buddhist art and architecture at the site.Footnote ⁹⁸ This latter observation highlights the point stressed throughout this article that the characteristic art and architecture of Dvāravatī only becomes visible in the archaeological record from the sixth to seventh centuries onwards.

At two sites excavated over the past ten years, Phromtin Tai and Kheedkhin, further evidence has emerged for continuity between late prehistory and the Dvāravatī period (fig. 1). However, in both cases a well-defined proto-Dvāravatī period was not clearly identified.

The moated site of Phromtin Tai about 20 km northeast of Lopburi town has been subject to archaeological investigations led by Thanik Lertcharnrit from 2004 to 2013 with future seasons planned.Footnote ⁹⁹ The site shows evidence for occupation from the Bronze Age through to the Dvāravatī period. Lertcharnrit has interpreted these levels through a combination of radiocarbon dating and analysis of the cultural material discovered. He identifies a clear Dvāravatī occupation layer from the sixth to ninth centuries. He also took five radiocarbon dates from the Iron Age layer, however the calibrated date ranges vary widely.Footnote ¹⁰⁰ He assigns the Iron Age layer to 500 BCE–500 CE and states categorically that there was no break in occupation between this layer and that of Dvāravatī.Footnote ¹⁰¹

About 40 km to the south in Saraburi province lies the site of Kheedkhin. It is a small-scale Dvāravatī site of only approximately 620 by 620 m. Notable finds include the usual pottery assemblage of carinated, spouted and stamped ware as well as a terracotta lion image and a fragment of a dwarf (gana) image.Footnote ¹⁰² Excavations undertaken at the site between 2007 and 2008 by Silpakorn University revealed evidence for Iron Age occupation from c.500 BCE onwards. Radiocarbon dates obtained from the excavation give some indication of dating. However, there are wide ranges in dates, sometimes as large as ±690 years.Footnote ¹⁰³

Khunsong therefore primarily used relative dating techniques to establish the chronology.Footnote ¹⁰⁴ He divides the occupation into two ranges, the prehistoric period from between 2500–1600 Before Present (BP) (550 BCE–350 CE) and the Dvāravatī period dating between 1400–1200 BP (550–750 CE) or the sixth to eighth centuries.Footnote ¹⁰⁵ This appears problematic as according to these dates there is a gap of two hundred years between the late prehistoric period and the Dvāravatī period which goes unexplained in the article. However, from the Thai language site report it is clear that there is continuity at the site and that these above date ranges indicate the timeframes which can be clearly bracketed. The individual stratigraphy of the test trenches show that there is continuity between layersFootnote ¹⁰⁶ and it appears more a case that the excavators were unable to clearly establish a relative or absolute dating for the two-hundred-year gap c.350–550 CE with the diagnostic finds available to them. Furthermore, they state that the material record from test trenches excavated in the ramparts shows that the site was occupied, or at least in use, in late prehistory before the moat was built and continued to be occupied in the Dvāravatī period.Footnote ¹⁰⁷

At both Kheedkhin and Phromtin Tai there is clear evidence for habitation from late prehistory into the Dvāravatī period from the sixth century onwards. While the material record and dating techniques at both sites did not allow for a more refined dating of the sequence of the site, neither show any signs of abandonment and appear to have transitioned from late-prehistory to Dvāravatī somewhere in the fourth to fifth centuries. It seems that there was a proto-Dvāravatī phase at these sites, but it is not clearly detectable in the stratigraphy.

Turning to the northeast, the two largest sites in the region, Mueang Sema and Mueang Fa Daed, also show evidence for proto-Dvāravatī phases (fig. 1). Mueang Sema, located on the Lam Ta Khong River, part of the Mun River system, is in Sung Noen district, Nakhon Ratchasima province.Footnote ¹⁰⁸ It covers an area of over 150 haFootnote ¹⁰⁹ and has been divided into three phases of occupation.Footnote ¹¹⁰ The earliest phase dates from the fourth to fifth centuries and has been classified as late prehistoric by the excavator.Footnote ¹¹¹ The second phase dates from the sixth to ninth centuries and has been classified as Dvāravatī period owing to the pottery record and clear evidence for Buddhism during this time. The third phase, from the ninth to twelfth centuries, shows evidence of growing Khmer influence as indicated by the presence of Buriram ceramics and Song Dynasty ware.

These three phases fall neatly into the normal parameters of the established Thai dating system. However, a closer look at this chronology and the material record indicates that it is not as clear-cut as it seems. For instance, one particular group of ridge pottery found in the earliest stratigraphic layers (fourth–fifth century) is also found in layers III and IV at Chansen.Footnote ¹¹² Furthermore, Khemica Wangsuk points out that this pottery type is also found in the second phase and evolves into a pottery type that is commonly used in the Dvāravatī period of the sixth to ninth centuries. Another group of earthenware has also been found in the layers representing phase 1 and are common to the Pasak region of Thailand. They were also recorded at Chansen in layer IV, which dated to between the mid-fourth to sixth centuries.Footnote ¹¹³

Analysis of the pottery record between phases one and two at Mueang Sema shows there is clear continuity between them. This along with comparisons with Chansen indicate that it is more appropriate to refer to the fourth to fifth century phase at Mueang Sema as proto-Dvāravatī. When we also consider that there is no evidence for early phases of prehistory in the stratigraphic sequence at the site, the term late-prehistory also appears to be an inappropriate designation at this site.

The largest moated site in the Khorat Plateau is Mueang Fa Daed in Kalasin province. Measuring about 170 ha in total, it also has the most substantial evidence of Buddhist material remains in the region.Footnote ¹¹⁴ Excavations carried out at the site in 1991 by Silpakorn University revealed a continuous occupation sequence stretching back into the late prehistoric period, c.300 BCE.Footnote ¹¹⁵ This was followed by what the excavation reports refer to as a proto-historic period. Based on radiocarbon dates and the pottery record, it spanned around the second to sixth centuries CE. After this we have a historic period which dates from the seventh to eleventh century and is associated with Dvāravatī culture. At this site the excavators have chosen to deviate from the standard periodisation. By recognising a proto-historic period they add weight to the argument that the fourth to sixth centuries were a transition phase between late prehistoric and historical (viz. Dvāravatī) societies.