Northeast Thailand before Angkor: Evidencefrom an Archaeological Excavation at the PrasatH in Phimai A e A o A SARAH TALBOT AND CHUTIMA JANTHED ALMOST A THOUSAND YEARS AGO, the walled city of Phimai (ancient Vimaya- pura) (Siribhadra and Moore 1997:232) was a major center of the polity of Angkor, which dominated much of mainland Southeast Asia from the ninth to the fifteenth centuries A.D. The Khmer empire is best known for the vast temple complex of Angkor Wat in the Tonle Sap region of Cambodia, although it also left a rich legacy of similar temples throughout Cambodia, Laos, and Thailand (Aymonier 1901; Briggs 1951; Freeman 1996; Siribhadra and Moore 1997). An extensive network of roads and resthouses linked the provincial centers and temples. One such road covered the 225 km from Angkor Wat to the most important Khmer temple in Thailand, the Prasat Hin Phimai (Figs. 1, 2). Con- struction of the Prasat began during the reign of the Angkorian king Jayavarman VI (A.D. 1080-1107), whose family had ruled for several generations at Mahi- dharapura, perhaps Phimai itself (Jacques 1996:147; Pichard 1976:1). The name Mahidharapura has been given to the subsequent dynasty of Khmer kings, which included Suryavarman II (A.D. 1113-1150), responsible for the construction of Angkor Wat, and Jayavarman VII (A.D. 1181-1219), the last great king of Ang- kor, whose image can still be seen at Phimai today. Khmer overlords had in fact shown interest in the region since the sixth century A.D. Several inscriptions refer to Citrasena-Mahendravarman's (c. A.D. 550-611) military victories beyond the Dang Raek Range, including one found at Phimai itself (K.1106) (Higham and Thosarat, 1998:194; Jacques 1989:17; Vickery 1998:75). However, while the consecration ofJayavarman II as monarch in A.D. 802 marked the start of the Angkorian period, it was not until some two centuries later that the region north of the Dang Raek Range was integrated into Angkor. From the reign of Rajendravarman II (A.D. 944-968) some control began to be exerted in the northeast (Siribhadra and Moore 1997:31), but most Khmer temples in Thailand were built after A.D. 1000. Although Jayavarman VII (A.D. 1181-1219) constructed the network of roads, resthouses, and hospitals that Sarah Talbot is with the Anthropology Department, University of Otago, Dunedin, New Zealand, and Chutimajanthed is with the Royal Thai Fine Arts Department, Phimai, Thailand. Asian Perspeaims, Vol. 40, No. 2 D 2002 byU nivceity of Hawai'i Press. I80 ASIAN PERSPECTIVES e 40(2) * FALL 200I Fig. 1. Sites discussed in the text. linked provincial centers to Angkor, his death saw the empire fragment. Prayers were offered at the Prasat Hin Phimai upon the death of Indravarman III in A.D. 1243 (Briggs 1951:238), but the late thirteenth century effectively saw the end of the Angkorian control in what is now northeast Thailand. NORTHEAST THAILAND BEFORE ANGKOR Northeast Thailand in the centuries immediately prior to Angkor has traditionally been described as lacking social complexity and subject to influences from two competing kingdoms-Dvaravati, a Mon-Buddhist culture in central Thailand, and Chenla, a Khmer-Hindu culture in the Kompong Thon region of Cambodia (see, for example, Quaritch Wales 1969). The very nature of the ostensible Dvara- vati and Chenla "kingdoms" have more recently been reconsidered (Brown 1996; TALBOT AND JANTHED * NORTHEAST THAILAND BEFORE ANGKOR I8I Fig. 2. The Prasat Hin Phimai, northeast Thailand. Diskul 1979; Glover 1980; Jacques 1979; Mudar 1999; Smith 1979; Vickery 1998), and the Mon-Khmer distinction in Northeast Thailand itself seems some- thing of a false one, at least before A.D. 1000 (Keyes 1974; Siribhadra and Moore 1997:25). Archaeological excavations at prehistoric sites such as Noen U-Loke have also shown that independent local communities were increasing in social complexity during the Iron Age (c. 500 B.C.-A.D. 500) (Higham and Thosarat 2000). The volume of prehistoric ceramics and artifacts recovered in earlier exca- vations at Phimai indicated that it was more than a simple village even in pre- history, perhaps due to its strategic location on the main trade route between the Khorat Plateau and the Chao Phraya Delta (Bronson 1979:327). The centuries from the end of prehistory to the splendors of Angkor have been described as "the Dark Ages of Isan" (Rogers 1996:51). However; there is evi- dence that the people of this period shared important new Indian-influenced practices with their neighbors to the west and south, particularly the use of in- scriptions and the construction of religious buildings in permanent materials. The inscriptions recording the "exploratory probes" (Vickery 1998:79) of the sixth- century overlords from the south have already been mentioned, but a number of other inscriptions refer to protohistoric kings, several Buddhist, who ruled in the region (Brown 1996; Higham and Thosarat 1998:194-195,Jacques 1989). A succession of brick pre-Angkorian teniple sites has been linked to the politi- cal consolidation of the Khmer in Cambodia, Sambor Prei Kuk being the most famous (Benisti 1970; Boisselier 1955, 1966; Jacques 1996; Parmentier 1927; Siri- bhadra and Moore 1997:26). In contrast, very few Khmer-style temples were built in Thailand before the tenth century (Charernsupkul 1981; Siribhadra and I82 ASIAN PERSPECTIVES * 40(2) FALL 2001 Moore 1997:31). For example, the oldest known Khmer structure in Thailand in good condition is the seventh-century A.D. brick and sandstone Prasat Phum Phon (Parmentier 1927:232; Siribhadra and Moore 1997:85). However, at least some later stone temples were built on the sites of earlier brick structures. For example, a series of brick temples underlie the sandstone temple of the Prasat Phanom Rung, and date back to at least the seventh or eighth centuries (Higham and Thosarat 1998:200). Recent excavations at Prasat Phanom Wan revealed prehistoric remains, including Iron Age burials (Buranrak 2000; Higham and Thosarat 1998, fig. 310; Phongdam 1997). During restoration, the monument was dismantled, and a square 160-cm-thick brick structure filled with soil was found just beneath the central tower. The bricks for this foundation may have been removed from an earlier and adjacent brick structure (Thosarat pers. comm.). Previous investigations at the Prasat Hin Phimai also recorded the pres- ence of bricks beneath the central sanctuary (Pichard 1976: pl. X). Other early buildings in the northeast relate to the Mon-Dvaravati tradition of Buddhist architecture, particularly brick sthpas or cledis at sites such as Nakhon Pathom, U-Thong, and Ku Bua in central Thailand. Similar brick architecture in northeast Thailand includes the remains of an uibosoth (monastery) at Kantarawichai (Diskul 1979) and brick religious structures at Ban Prasat (FAD 1992). Dvaravati- style stutpa bases occur at the site of Muang Fa Daet, which is more famous for its large seniias, or boundary stones, carved with figurative Buddhist scenes (Boisselier 1972; Diskul 1956). While nonfigurative seinia appear as early as the seventh cen- tury, the pictorial seinias date to the ninth and perhaps as late as the eleventh cen- tury (Brown 1996; Krairiksh 1974:57). Muang Sema is another particularly im- portant Dvaravati site in the northeast, with a large reclining sandstone Buddha figure, dhanziacakra ("wheel-of-the-law"), and many brick structures and seinas. Recent excavations at Muang Sema recovered a Dvaravati cultural layer some 100-cm thick between layers containing Iron Age and Khmer ceramics (Thosarat pers. comm.). THE 1998 PRASAT HIN PHIMAI EXCAVATION The Origins of Angkor Project (hereafter OAP) is a joint project of the Anthro- pology Department of the University of Otago, New Zealand, and the Royal Thai Fine Arts Department. It has investigated a number of prehistoric sites in the region, including the Bronze Age cemetery of Ban Lum Khao and the Iron Age sites of Non Muang Kao and Noen U-Loke (Boyd et al. 1999; Habberfield-Short 1999; Higham and Thosarat 1998, 2000; O'Reilly 1999; Rivett 1999). In 1998, the OAP conducted an excavation at the Prasat Hin Phimai to investigate the underlying sequence and examine the brick structure previously reported in more detail. Most previous work at the Prasat concentrated on restoration, although some excavation was done within the temple compound (Table 1). On 5 January 1998, a 4-by-7.-m unit was laid out along the axis of the Prasat, just to the west of the central sanctuary (garbhagra) and 1 m south of a square excavated in 1997 by Chutimajanthed (n.d.). The datum was set on the corner of the bottom step of the west entrance to the sanctuary, approximately 0.3 m above the surface of the ground. Temple construction during the Angkorian period involved the deliberate deposition of layers of fill, which can clearly be seen in the stratigraphy of the site (Figs. 3, 4) (cf. Nakao 1992: fig. 2; Pichard 1976). As a re- 0 X .2 0 4)4 .i -X * sO .4 t 0 0~~~ 0z *o0 0a 44, ~ 4) -4 ° As : -O0 ) -I~ 0- >..4=U3O i>a0_ 0.0 g4a *0 ._ 0 W> 44 ' .= z 04t) ,I 0 c 0 n 4° 2 N C\~~~~~~~~ V) - = 0 o En.2 :Q -} -0-0 X\ u z f.4 .:01. E N 4) -N U = 1- -41 .4 oi' ,1 08 4) t tnZaV 9_ Qu0- X , 0 " 0 a' 0 tn a- s Uz, u ¢\ >t i C\a -2\'=2'a 2- 2C- c, .-05 0 FVW44 .14 z 0 00 0 0 0 0~ 0 ~ ~0 ~ ~0 ~~~~0 0 0 0 0~~~~~~~~~-~~~~~~~~~~~~~~~~~~~~~ V)4 I- O2a 0 0 020 .4 002 0 J. 002 0 . -U) 0020 J. LtS- a04 . O02a 4S 0a004 . 4 sxt43X 4-~ J~°Q ~ ~u =- t& n0 4 4); t t } tu 4) en tb t S 0, 0e 4)0 O 41 4H) t44 4 H4 H) H H 4H) m4) H H H 0 Z H 0 0 0 0 .4 0 0~~~~' 0 0 1-10 0 r. r~~~~~~~~~~~~. >~~~~~~~ 0 o 0 >-'0 >- >2 ~~~a 4 ) .4~ ~)~ ~~~~~~1~ ~~~ < x% I xl n . -m, . V) ,_ a' 5 ' \ .. '\. 'a0' - 4) d a' 4) I .': XI> CZ Ca\ t ' LI =\. = \C 0\WO I \0 5 4o \-O E UO\ C0 \ o 0 - 0So 1-. 4) >U< U. 3 3Z. 4O4S 0. 4a..'-O 0 0.4. - r. 0. -. o . P. v) q i7) 184 ASIAN PERSPECTIVES * 40(2) *FALL 2001 4M- Fig. 3. The north section of the 1998 excavation showing layers of sand and charcoal associated with construction of the temple. Key: IA, pale yellow, 2.5yr 8/4, disturbed; IB, dark grayish brown, 2.5yr 4/2, includes pit contents; IC, pale red, 7.5yr 5/4, includes fragments of red sandstone; 2A, reddish yel- low, 7.5yr 7/6, coarse sand, very disturbed in places; 3A, brown, IOyr 5/3, silty, sandy clay, includes many potsherds; 4A, very pale brown, IOyr 8/3, sand, includes clumps of pale gray clay; 4B, pale brown, IOyr 6/3, sand; 4C, strong brown, 7.5yr 5/8, sand; 6A, dark grayish brown, 2.5yr 4/2, clay, in- creasingly sandy toward bottom. sult, stratigraphy was complex, with much disturbance and redeposition, although major depositional events were marked. Samples were taken by the OAP for more detailed geomorphological analysis. The following section describes the stratigra- phy. Layer One. This -was a dark, mottled, and disturbed layer. Red sandstone pieces and eroded laterite lenses probably resulted from construction of the two late twelfth-century A.D. towers nearby. .The layer contained bronze fragments, Chinese blue-on-white porcelain, animal bone, a small gold ritual plaque, "olive- brown" (Munsell 2.5 4/3) glazed Khmer ceramics, and a ballalee (lotus-bud ceramic roof decoration) (Talbot 2001 :116). In recent times, the area was cleared during restoration. TALBOT AND JANTHED * NORTHEAST THAILAND BEFORE ANGKOR Fig. 4. The east section of the 1998 excavation showing the brick feature. Key: IA, pale yellow, 2.5yr 8/4, disturbed; IB, darkl grayish brown, 2.5yr 4/2, includes pit contents; 1C, pale red, 7.5yr 5/ 4, includes fragments of red sandstone; 2A, reddish yellow, 7.5yr 7/6, coarse sand, very disturbed in places; 3A, brown, IOyr 5/3, silty, sandy clay, includes many potsherds; 4A, very pale brown, IOyr 8/ 3, sand, includes clumps of pale gray clay; 4B, pale brown, IOyr 6/3, sand; 4C, strong brown, 7.5yr 5/8, sand; 5A, red, 2.5yr 5/8-2.5yr 6/8, bricks in dark grayish brown, 2.5yr 4/2, clay; 6A, darkl grayish brown, 2.5yr 4/2, clay, increasingly sandy towards bottom. Layer Two. This layer of coarse, soft sand was mixed and disturbed, particu- larly on the side closest to the temple. It seems that riverine sand was collected and deposited, as a single event, apparently to level the ground and perhaps pro- vide drainage during the monsoon. Layers of compacted sand were also described in Pichard's (1976) report on the reconstruction of the temple. The sand con- tained fragments of roof tile (Talbot 2001:115-116), bronze, white porcelain, and pottery. Several large il-defined features extending from above contained Angkorian ceramics, roof tiles, small sandstone fragments, a few worn sherds, and fill similar to Layer One. Layer Three. Beneath the sand lay dark silty and sandy clay typical of the region's flood plain. The surface sloped down approximately 0.2 m from the northeast over the width of the excavation. Shallow areas of charcoal and a series of postholes were visible. Layer Three produced more ceramic material than all other layers combined, and roof tiles, animal bone, an orange glass bead, a metal spout, and iron slag were also recovered. Layer Four. This layer comprised a thick deposit of sand, regularly interspersed with charcoal layers. While the top surface of the sand was level, the bottom sur- face sloped steeply down from the east. This layer lay just above the water table and as the sand became increasingly damp during the excavation, a 1-m secondary baulk was left along the west and south to preserve the walls. Use of a bore indi- cated that the sand below was at least 2-m thick along the western (unexcavated) side of the unit. Most of the (very little) cultural material from Layer Four was I86 ASIAN PERSPECTIVES * 40(2) * FALL 200I found within the charcoal deposits. The same charcoal layers were also found in the 1997 excavation, only 1 m to the north. A charcoal sample from that excava- tion was radiocarbon dated and calibrated at the University of Waikato to A.D. 966-1020 (1-sigma). Given the possibilities of inbuilt age, this date is appropri- ate, considering that the initial construction of the Prasat Hin Phimai temple is thought to have taken place during the reign ofJayavarman VI (A.D. 1080-1107) (Jacques 1996: 149). Layer Five. Beneath the sand layer, along the eastern baulk, lay naturally deposited clay over 3-m thick. Embedded at the top was a thick sloping feature of haphazardly arranged bricks and brick fragments (Fig. 5). Few ceramic sherds were recovered. There was evidence for burning in the dark charcoal-rich clay directly above the bricks. This feature is presumed to be part of the brick struc- ture previously reported to lie beneath the central sanctuary (Pichard 1976: pl. X; Solheim 1970). Pichard (1976:22) described it as a square sump (ini puisard carre), which was concentric to the tower but on a slightly different axis. According to Pichard (1976) its sloping internal walls each measured 3.60 m across at the top 5m N Fig. 5. Plan of the brick fea- ture uncovered during the 1998 excavation. TALBOT AND JANTHED * NORTHEAST THAILAND BEFORE ANGKOR 187 and were constructed of bricks of variable size, averaging 7 by 17 by 32 cm. This structure resembled that found beneath the central sanctuary at nearby Prasat Phanom Wan (Buranrak 2000). Layer Six. This layer of clay extended down to the bottom of the excavation. Below the brick feature, the water table was reached and due to the need to pump water, excavation was limited to a 1-m trench along the eastern wall, beneath the bricks and close to the sanctuary. Water flowed, through the sandy lower levels of clay extremely rapidly. Many eroded prehistoric sherds were found, forming a link to Iron Age occupation at sites such as Noen U-Loke. Excavation continued to 4.70 m below datum when the eastern baulk began to collapse and the exca- vation was halted on 25 February 1998. At this point completion was already being considered, as the deposit had become increasingly sandy, with natural lat- erite nodules and few sherds. BRICKS FROM THE PRASAT HIN PHIMAI EXCAVATION A total of 230 kg of bricks was recovered during the 1998 excavation, all, with few exceptions, from the brick feature in Layer 5. The "red" (Munsell 2.5yr 5/8- 2.5yr 6/8) bricks were heavily rice-tempered with occasional laterite or even pot- sherd inclusions, unevenly surfaced, and not especially durable. Only eleven were recovered in which a measurement could be made of the original length, breadth, and width. The bricks were roughly square, with lengths and breadths within 3 cm for over half. The majority measured 7 cm in thickness, with a range of 6-8 cm. Most had a length of 18-25 cm, with four measuring 22 cm. Average dimensions were 7.8 by 17.3 by 20.8 cm and a single brick weighed up to 5 kg. Twenty-nine otherwise indistinguishable bricks were patterned on one of the large faces with two sets of two or three straight diagonal grooves crossing at the. center (Fig. 6). One fragment had a curved set of lines and three had parallel lines :~~~~~~ 0 E t t @i ,|, '01S4' 0 ,t';:: ' i Fig. 6. Brick marked with diago- nal finger marks (Cat. 155) (7 by I88 ASIAN PERSPECTIVES * 40(2) FALL 2001 near the edge. The grooves were made before firing and appear to be finger marks. Such bricks were randomly distributed throughout the feature. In Myanmar, similar finger-marked bricks were part of a cultural tradition coinciding with the construction of Mon and Pyu walled cities in the early cen- turies A.D.-"... the finger-strokes marking the ancient bricks remain[ing] as a surprisingly intimate record of a long ago human gesture" (Aung Myint and Moore 1991:101; Aung Thaw 1968: fig. 8). Such finger-marked bricks were found at Thaton, where seimas similar to those found in the northeast of Thailand indicate contact between the two regions in the late first millennium A.D. (Aung Myint and Moore 1991:93). This distinctive patterning has also been found on bricks from structures at Dvaravati sites in central and southeast Thailand. At U-Thong, such bricks were found at a chedi (Site No. 15), whose stucco ornamentation recalls that of ninth- century Khmer art (Kulen and Preah Ko styles), and at a iniandapa (Site No. 21), which apparently originally had a roof reminiscent of the pre-Angkorian temple of Kuk Preah Theat (Boisselier 1972:32-33; Parmentier 1927:197). Similar bricks were also found at Chedi Wat Khlong at Ku Bua (Site No. 18) (Boisselier 1972:33, figs. 40 and 41) and at the site of Si Mahosot (Pisnupong 1992, 1993). The finger-marked patterns were not in themselves decorative but aided the ad- herence of a wet clay mortar, and have been used to date religious brick structures to about the end of the eighth century A.D. (Boisselier 1972:32-33). Other evidence also suggests that the Prasat Hin Phimai was built at a site that had been sacred for some time. Early last century, a Dvaravati style dhlarnracakra ("wheel-of-the-law") and Buddha image were found together at the temple (Siribhadra and Moore 1997:229). Carved on a stone re-used as a doorjamb in the temple wall is an eighth-century inscription (K.1000) describing a Buddhist king called Sauryavarman (Sawrya = valour, might), and a stele carved with homages to both Shiva and the Buddha dates to the same period (Jacques 1989:19, 1996: 149). In a similar fashion to Prasat Phanom Wan, the bricks recovered at Prasat Hin Phimai were apparently reused from an earlier structure (Buranrak 2000; Thosarat pers. comm.). The period of the seventh to ninth centuries is an appro- priate date for the Prasat Hlin Phimai bricks, considering the radiocarbon date obtained from the overlying charcoal (966-1020 A.D., plus inbuilt age). THE CERAMIC SEQUENCE AT THE PRASAT HIN PHIMAI Just over 50 kg of potsherds were recovered from the 1998 excavation, with Layer Three containing over 70 percent (by weight) of all sherds at the site. Many deposits were mixed, and redeposited prehistoric sherds were found in all layers. However, despite the fragmented nature of the assemblage and the mixed nature of the deposits, ceramic analysis distinguished some diachronic trends. Phimai Phase One (Layer Six). Dense deposits of redeposited Iron Age sherds suggest that the site of the Prasat Hin Phimai was indeed important in prehistoric times. Comparison with the existing ceramic sequence in the region (Welch and McNeill 1988-1989), suggests that while the site appears to have been occupied by the mid-first millennium B.C., the main prehistoric ceramic assemblage dates to the late first mil}ennium B.C. and the early centuries A.D. The earliest, red-slipped, sherds probably belong to the Prasat phase (c. 600-200 B.C.). However, most