THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS

THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Charles Higham, Judith Cameron, Nigel Chang, Cristina Castillo4, Sian Halcrow5, Dougald O Reilly6, Fiona Petchey7 and Louise Shewan8 Deartment of Anthroology and Archaeology, University of Otago, charles.higham@otago.ac.nz School of History, Culture and Language, Australian National University, judith.cameron@anu.edu.au School of Arts & Social Sciences, James Cook University, nigel.chang@gmail.com 4 Institute of Archaeology, University College, London, criscastillo@mac.com 5 Deartment of Deartment of Anatomy, Otago School of Medical Sciences, University of Otago, sian.halcrow@otago.ac.nz 6 School of Archaeology and Anthroology, Australian National University, dougald.oreilly@anu.edu.au 7 Radiocarbon Dating Laboratory, University of Waikato, fetchey@waikato.ac.nz 8 Monash University, Melbourne Australia and University of Warwick, Coventry, UK, louise.shewan@monash.edu ABSTRACT Non Ban Jak is a large, moated site located in the uer Mun Valley, Northeast Thailand. Excavations over three seasons in -4 have revealed a sequence of occuation that covers the final stage of the local Iron Age. The site is enclosed by two broad moats and banks, and comrises an eastern and a western mound searated by a lower intervening area. The first season oened an 8 by 8 m square on the eastern mound, while the second and third seasons uncovered art of the low terrain rising into the western mound, encomassing an area of 5 by m. The former revealed a sequence of industrial, residential and mortuary activity that involved the construction of houses, kiln firing of ceramic vessels and the interment of the dead within residences. The latter involved four hases of a late Iron Age cemetery, which again incororated house floors and wall foundations, as well as further evidence for ceramic manufacture. The excavation sheds light on a late Iron Age town occuied at the threshold of state formation. INTRODUCTION Non Ban Jak is a large, moated site in Amhoe Non Sung, Nakhon Ratchasima Province. It is located at latitude 5 5.89 and longitude.99 (Figure ). The mound is oval, with a maximum length of 6 m and width of 7 m. Identified during a rogramme of site surveys in 996, the only revious excavation involved sectioning the two moats and banks to investigate their form and date (McGrath and Boyd ). Eleven C4 determinations were obtained from charcoal and shell excavated from the inner and outer banks and both moats (Table ). These indicate that the construction eriod falls in the 4th-5th centuries AD. The site comrises a western and an eastern elevated mound with a deression between them. During the three seasons of excavations, an 8 by 8 metre square was oened on the eastern mound, and an area of 5 by metres was excavated in an area which began in the deression and then rose onto the margin of the western mound (Figure ). Figure : The uer Mun Valley study area, showing the location of the Iron Age moated sites and those mentioned in the text. JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4): -4

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Table : Radiocarbon determinations from occuation and mortuary contexts. Lab code (Wk-) Material dated C ( ) 4 C age BP calibrated age (AD) 68.% rob. AA surf, feature 8887 Filoaludina -.7 ±. ± 5 65-69 AA -/surf 4, feature 8888 Filoaludina -. ±. 6 ± 5 66-77 BB surf 4, feature 8889 Pseudodon -.5 ±. 57 ± 5 4-54 BB surf 4, feature 6 889 Filoaludina -7. ±. 6 ± 5 65-76 AA surf 4, feature 889 Pseudodon -6.8 ±. 54 ± 5 5-6 AA surf 4, feature 889 Filoaludina -7.4 ±. ± 5 77-87 AA surf 4, feature 889 Filoaludina -4. ±. 65 ± 5 9-5 BB surf A, feature 8 8894 Filoaludina -4.8 ±. ± 5 66-77 BB surf A, feature 8895 Pseudodon -5.9 ±. 477 ± 5 56-6 BB surf A, feature 8896 Pseudodon -6.9 ±. 546 ± 5 4-56 BB surf A, feature 4 8897 Filoaludina -. ±. 566 ± 5 4-54 BB surf A, feature 8898 Pseudodon -7.9 ±. 554 ± 5 4-55 AA - feature 8899 Filoaludina -5. ±. 96 ± 5 67-77 BB surf 4, feature 89 Pseudodon -.5 ±. 5 ± 5 54-6 BB surf 5, feature 89 Filoaludina -5.9 ±. 584 ± 5 4-54 AA surf 5, feature 89 Pseudodon -.7 ±. 48 ± 5 55-6 Samle source Western Mound layer Figure : The lan of Non Ban Jak, showing the location of the two excavation squares. (Courtesy Dr N. Chang). THE CHRONOLOGY An additional 4 radiocarbon dates have been obtained during the first three seasons at Non Ban Jak (Table ). All were calibrated using Intcal curves (Reimer et al. ) using the rogram OxCal v4. (Bronk Ramsey ). Those given in the text are reorted at 68.% robability. Table : Radiocarbon determinations for the moats and banks. Location Lab. code Material C4 age BP calibrated age (AD) 68.% rob. Inner channel Wk 649 charcoal 568±57 5-5 Inner channel OZD 67 shell 76±4 9-54 Outer channel Wk 749 shell 575±56 8-54 Outer channel OZE shell 69±7 4-54 Outer channel OZD 67 charcoal 6± 4-54 Outer bank OZE shell modern Outer bank OZE 4 shell 66±7 9-54 Outer bank OZE 99 shell 6± 6-4 Outer bank OZE 5 shell 6±9 4-55 Outer bank OZE 6 shell 65±48 5-9 Outer bank OZE 7 shell 68±5 4-55 layer

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Z surf, feature 4 896 Filoaludina -4. ±. 57 ± 5 Western Mound burials 5-6 layer Burial MP 6 894 Pseudodon -7. ±. 65 ± 5 4-54 AA surf, feature 89 Pila -6.8 ±. 557 ± 5 4-55 Burial MP 88 89 Pseudodon -6.8 ±. 68 ± 5 9-5 BB surf, feature 7 894 Filoaludina -7.6 ±. 54 ± 5 4-56 Burial MP 85 89 Pseudodon -6.8 ±. 68 ± 5 4-4 AA surf, feature 8 895 Pseudodon -5. ±. 587 ± 5 4-54 Burial MP 4 89 eggshell -9. ±. 77 ± 5 5-4 AA surf, feature 896 Pseudodon -5.7 ±. 6 ± 5 8-5 Burial MP 49 899 Pila esmei -8.5 ±. 559 ± 5 4-55 AA surf, feature 5 897 Filoaludina -.7 ±. 59 ± 5 4-54 Burial MP 8 895 Pseudodon -7.8 ±. 688 ± 5-4 AA surf 5, feature 898 Pseudodon -7.8 ±. 656 ± 6 5-4 surf: surface, MP: mortuary hase AA surf 5, feature 6 899 Filoaludina -5.6 ±. 58 ± 5 4-54 AA surf 6, feature 89 Pseudodon -4. ±. 7 ± 5 6-9 Z -4 feature 897 Filoaludina -4.8 ±. 58 ± 5 4-54 AA 4 surf, feature 89 Filoaludina -. ±. 766 ± 5 - AA 4 surf, feature 89 Pila -.7 ±. 58 ± 5 4-54 AA 4 surf 4, feature 5 89 Pseudodon -7. ±. 66 ± 5 5-4 Y surf, feature 898 Pseudodon -.8 ±. 58 ± 6 4-54 A 6 surf, feature 894 Pseudodon -6.9 ±. 649 ± 5 5-4 B 5 surf 7, feature OxA788 rice -4.6 579 ± 6 4-54 A 5-A feature 6 OxA788 rice -.66 64 ± 9-5 A 5- feature OxA788 rice -4.6 6 ± 4 4-54 The radiocarbon determinations for the cultural sequence on the mound itself come from three sources. The first, involves carbonized rice from two occuation contexts in the basal art of the eastern mound, where a fire had destroyed a building. Rice grains were also dated from the rake-out of a kiln. These indicate that this art of the site was occuied in the 5th century AD (Table ). The sequence in the western mound was dated using in-situ freshwater shellfish recovered from discrete middens. The stratigrahic record here was divisible into four layers (hases), with radiocarbon determinations coming from each. To refine our chronological interretation we have utilized Bayesian statistical methods whereby 4C ages are constrained by this rior stratigrahic information. The model we develoed for Non Ban Jak is shown in Figure. It consists of a series of hases each searated by a transitional boundary (i.e., contiguous multihase model) (Bronk Ramsey 9a). Within this model the internal consistency of the calibrated dates has been tested using an outlier detection method that enables a robabilistic measure of the degree to which samles aear to be outliers, and then calculates an offset relative to the context within which it is found (Bronk Ramsey et al. ). In this instance we have used a General t-tye model with rior outlier robability of 5%. This allows outliers to be either too young or too old (Bronk Ramsey 9b) and enables us to down weigh their influence in the model. Within the western mound sequence there were four major outliers (osterior outlier robabilities >7% comared with the rior of.5%, including three of >96% which are not included in the model results 96% of the time and therefore do not influence the model) and a further two with outlier robabilities of ~% (i.e. 4 times more likely to be outliers). Three of these outliers come from Layer, and two of which aear to be of similar age to material in Layer below suggesting intermixing layer 4 Eastern Mound

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure : Bayesian age model for the Non Ban Jak sequence roduced using OxCal 4. (Bronk Ramsey,, 9a, b). The lighter shaded distributions reresent the unmodelled calibrated dates, and the darker outlines reresent the results after Bayesian modelling. 4

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure 4: Summary of the transitional dates between layers. may have occurred. Desite these outliers, the Non Ban Jak model convergence values generated by the OxCal MCMC algorithms were uniformly high (>98.4), and therefore indicate that the model is robust (low values indicate many different incomatible solutions to the model) (Bronk Ramsey 9a). Dates for boundaries between layers are roduced as art of the modeling and rovide temoral constraints for each layer (Figure 4). These indicate that the initial occuation at this art of the site began in the rd-4th century AD (AD -8). The transition into layer took lace around AD 46-5, while the final hase of occuation ended around AD 7-8. The third series of dates come from freshwater bivalves shells laced as mortuary offerings with the dead. These determinations conform to mortuary activity during the 4th, 5th and early 6th centuries AD (Table ) at the same time as occuation activities in the eastern and western mounds. Figure 5: The north section of the eastern mound excavation. 5

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure 6: The surface of layer 5., western half of the square, showing two suerimosed buildings. A is the eastern wall of the later structure. B is the eastern wall of the earlier building, which was laid out on a different orientation. C is the clay floor bearing what are robably bamboo imressions. D is a ceramic kiln laid out on the same orientation as the earlier of the two structures. E is an infant burial cut through the floor, and F is the kitchen area. In the foreground, one can see the charcoal, ash and burnt daub of the floors and collased walls of the earlier building. The radiocarbon dates lace this structure in the 4th-5th centuries AD. 6

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure 7: The surface of layer 5-, showing the lan of a residential structure. Note the infant burial cut through a floor. The internal divisions of the structure are clearly visible. The structure dates to the 4th-5th centuries AD. THE EASTERN MOUND The sequence on the eastern mound comrised six layers which reached a deth of 4.5 m below datum (Figure 5). The following summary of the sequence begins with the basal layer 6 and roceeds through to the last rehistoric occuation hase and the final eriod of occuation, which is dated to the eriod of the Dvaravati state on the basis of the ceramic remains. The excavation recognised six major layers, which were divided into sits aroxi- mately cm thick. Distinctive features, such as hearths, walls, floors, kilns, grave cuts, ostholes and its were assigned an unique feature number, and excavated searately to retain all finds in their cultural context. The nomenclature adoted below reflects this excavation rocedure. Thus layer 6- refers to the second sit in layer 6, and lans are normally taken of features identified on the surface of a new sit, or of features such as house lans or burials found within a articular sit. 7

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Layer 6 Layer 6- was identified on the basis of a change in the colour to a yellow, natural-looking sandy material. However, there were many ostholes cut into this layer. The base of a circular kiln for firing ottery vessels was found to have severed a lower concentration of otsherds ringed by a line of ash. This hints that there may be a second kiln beyond the area excavated. The surface of layer 6- has two rincial features. A dee cutting was found, originating at the edge of the kiln cut down to this context, and extending to the west. lt contained faunal remains and broken otsherds, and was embedded into the natural laterite substrate and the water table. In the northwest corner of the excavated area, a concentration of occuation material was found which was again thick, and extended below the water table. The surface of layer 5- contained the remains of a further building that had been destroyed by fire. This building immediately underlies that described in layer 5, but was laid out on a slightly different orientation (Figure 6). The southern art of this structure comrised clay wall foundations, within which there were thick deosits of burnt material including charred structural timber and rice grains. In the southeastern corner of the building, there was an incinerated clay floor, on which lay a ottery vessel, two hearth stones and a large quantity of carbonised rice grains. The most southerly art of this building had surviving burnt and blackened wooden foundations. Proceeding north, a thick layer of charcoal and burnt daub was found over the original clay floor. The daub incororated the rofiles of wattle suorts, and this is interreted as art of a collased wall overlying blackened occuation material laden with rice grains, The clay floor lis u against the edge of the walls. Burial, an infant, was cut through this floor as well as the collased wall, meaning that it could date to the building found above. Figure 6 shows the basal remains of the structure first seen in 5-. In the southeast corner, there is a surviving area of burnt clay floor, incororating an uright ceramic vessel, and many carbonised rice grains. The northwest art of this building comrised a series of arallel, semicircular imressions on the clay floor, which end abrutly and in unison, as if they had reached the edge of an internal room within the building. Further clay floors, one burnt, were found at the northeastern corner of this building. There is a second structure to the east, laid out on the same orientation. This was not well reserved, although some evidence for burning is seen within the area enclosed by walls (Figure 7). Burial, another infant, was found within the confines of this structure. Between these two buildings, and on the same orientation, we found an oval structure containing broken ceramic vessels and much red daub liberally temered with rice straw, together with rice grains, ash and charcoal. On excavation, this feature turned out to have been a ceramic kiln that still contained a large ottery vessel in its base (Figure ). The quantity of collased daub suggested that this kiln had been enclosed by a domed roof. Under the buildings identified in layer 5-/, there was evidence for occuation in the form of disersed hearths, areas of burning, its and ostholes. In the southeastern quadrant, a hearth was surrounded by three broken ottery vessels. The surface of layer 5-5 has as its main feature, a straight line of charcoal in the southeastern quadrant. At the time of excavation, this was hard to interret. However, as we robed deeer into the mound, we were to find that it reresents the tomost remains of the rake out from a ceramic kiln. Other finds in this art of the site include a concentration of otsherds, and just to the north, many rice grains. Other than these finds, the layer included several ostholes, some its and an area of burning in the southwestern art of the square. Layer 5-6 was dominated by a large concentration of burnt daub, charcoal and large ottery sherds. It is flanked Layer 5 Layer 5 was recognized in the northeastern corner of the square on the basis of an extensive lens of dark occuation on to of a clay floor. The layer incororated the structural remains of houses, orientated a few degrees to the west of due north, as well as the remains of two kilns for firing ceramic vessels (Figures 6-8). The clay floor that signalled the uermost context in layer 5 contained much charcoal, reflecting the destruction of a building in a conflagration (Figure 6). Adjacent lay a wall on an orientation about west of north. To the south, lay a series of its two of which contained relatively intact Phimai Black Iron Age ceramic vessels. There was overall, much evidence for burning, including carbonised rice grains. Two radiocarbon dates derived from carbonised rice grains in this building are 9-5 and 4-54 cal. AD (Table ). The western art of the square was dominated by the foundations of a second structure. It had clay and laterite outer walls incororating ostholes that robably suorted the house walls (Figure 7). Within, a series of thin subdivisions of laterite divided the structure into rectangular areas, which are too small to have been rooms. They may have been the foundations for internal divisions that artitioned the structure. Particularly in the northern art of this building, the narrow artitions were marked by ostholes which again, may have suorted furniture or have acted as subdivisions. The interior contained clay or laterite floors which had been subjected to considerable heat, robably when the structure was burnt. These floors contained concentrations of carbonised rice and broken ceramic vessels. The skeleton of a rat was found adjacent to the southern exterior wall. The structure as a whole measured.4 by 6.4 m, although interior artition walls roceeded south and beyond the excavation square. A laterite floor in the centre of the largest interior room sealed an oval it that contained the infant burial. Since this grave also cut through an earlier laterite floor, it must date to the eriod when the building was in use. To the west of this building there is a clay wall foundation lying on the same orientation. 8

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) on either side with more charcoal. At the northeastern end of this feature, there is a greater dominance of red clay daub liberally temered with rice straw. There are several nearly comlete Phimai Black ots north and south of this feature, and many large broken otsherds. Two infant jar burials flanked this concentration of daub and otsherds. As was discovered in due course, this area turned out to lie over a circular kiln for firing ceramic vessels. There is also a concentration of rice grains on this burnt area that rovided a 4C determination of 4-54 cal. AD (Table ). The rest of the excavated area contained a scatter of ostholes. of white rice, and beyond further, a hearth, ash, and charcoal. A second concentration of burnt clay and ash was found along the western edge of the excavated area. There are also a number of ostholes. Below it lay a circular enclosed kiln for firing vessels of variable form. At least eight were found in situ (Figures 8, ). The kiln was fashioned from clay liberally temered with rice straw. A channel for the introduction or ouflow of air was found. The quantity of charcoal mixed with the daub and lying outside the circular kiln, which had a diameter of. m, suggests that the heat was alied from wood or charcoal itself. A socketed iron imlement with wings, one of which had been broken in antiquity, accomanied the ots. Why this should be so is hard to trace. The asymmetric rofile of this artefact, and the resence of wings, suggested to the local rice farmers that it was a loughshare rather than a sade. An almost identical artefact was found as a mortuary offering at Noen U-Loke during Iron Age (Connelly 7:45). The structure of the kiln, quantity of charcoal, liberal resence of rice chaff and straw and resence of a ossible chimney or tuyère to allow the ingress or escae of air from the interior, are all recisely mirrored in the kilns now used to convert wood into charcoal. Burial, an adult, was just resent in the southeastern corner of the square. It dates to the late Iron Age. In the northwestern art of the square, there is a large discrete area that had been subjected to heat. Two smaller similar contexts were found in the southern art of the excavation. A ossible clay wall on a north to south axis was found near the southeast edge of the square. The lowest context in layer 5 comrised the disturbed remains of a structure reresented by a grey clay floor, art of a burnt wall and several ostholes. Layer 4 Layer four contrasts with its redecessor by a distinctly yellowish colour and an increase in the amount of occuation material lying on the surface. There is also a marked increase in the number of ash-filled hearths. The surface of layer 4- is dominated by the two building hases indicated by suerimosed wall foundations on the same orientation. Essentially it was ossible to recognize three or four chambers. The three on the western art of the square contain very oorly reserved fragments of clay flooring, and there are two hearth areas together with a ossible internal buttress or furnishing also of clay. The middle and southern rooms were cut into by a later rectangular it. The base of wall three, which lies on the eastern edge of the excavation, is still resent. Concentrations of ottery and darker occuation lenses are found along the southern art of square Layer 4- was dominated by two rectangular rooms in the western half of the excavated area. It is evident that the walls of these rooms were rebuilt over time. The northern room had within it on the southern wall, an internal buttress, which might have been an item of furniture, and two hearths. Figure 8: The kiln in layer 5. Above, rior to excavating the wall collase and below, fully oened. Pottery vessels remained within, together with a socketed iron loughshare one wing of which had been fractured in antiquity. Scale cm. With the removal of the overburden of daub, charcoal and ash from the kiln rake out, we encountered a lens of large sherds being an accumulation of broken ots, or wasters, from firing. At this juncture, we also defined the circular edge of the kiln after removing a quantity of broken daub. To the west of the rake out, there was a lens 9

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure 9: The surface of layer 4-7, showing mounds of laterite, clay lined furnaces and in section, much evidence for heat. Scale 5 cm. Layer 4- incororates the clay base of a wall within which there is a grey clay floor. Postholes lie just outside the wall and were robably art of the structure. The wall roceeds in a westerly direction, and contains to the south a deosit of decayed laterite, an area of burning, concentrations and ottery and animal bones. The northwestern quadrant contains the lower art of a clay buttress belonging to a building that had already been removed. A hearth was found to the west of this buttress, and floors within it to the north. Layer 4-4 contains the laterite foundation walls for robably two structures on an identical alignment as later buildings. The southern building incororated what might well have been a cooking area. This comrises a series of hearths constrained within thick mounds of laterite. There are two comlete ottery vessels, and the articulating limb bones of a bovid. To the north, there is a double set of arallel walls, albeit oorly reserved, and art of a clay floor. There is a small number of ostholes, three its, and one or two concentrations of Phimai Black otsherds. Layer 4-5 incororated in the southeastern quadrant, a laterite wall foundation at a deth of.69 mbd lying on the same orientation and below the later structures in this art of the excavation. A row of three ostholes lying alongside and to the west of the wall was robably art of the same structure. To the southwest lay a concentration of clay-lined hearths or areas of burnt clay. A singular feature of the southern edge of the square, was a set of two thick mounds of decayed laterite within which lay a further hearth and concentration of ash. A grey clay wall foundation again on the same orientation as its successors was identified in the southwestern quadrant. Further north, there were more hearths and a clay wall associated with further suerimosed floors. Two sets of lid-to-lid Phimai Black ots were found on the same east to west alignment of this wall. Layer 4-6 contains the foundations for a laterite wall at a deth of.74-.78 mbd. It lies under and on the same orientation as the walls above it in layers and. The northern edge of the square also contains the badly reserved remains of a clay floor. In the middle of the square, there is a dense concentration of Phimai Black otsherds, lying west of the wall foundation, as well as a hearth. A marked feature of the southern art of this area, is a concentration of five clay-lined hearths, filled with ash and on occasion bone and otsherds. The southwestern corner is an area subjected to considerable burning. There are two raised laterite accumulations surrounding a rich charcoal deosit. The southwest quadrant of layer 4-7 incororated, as did the layer above, evidence for the construction of a series of clay-lined furnaces. The resence of iron slag and deosits of laterite suggest that these were emloyed for smithing, or alternatively, ossibly for smelting. The analysis of the slags and laterite are not yet initiated (Figure 9).

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure : The surface of layer 4:8 looking to the southwest. The configuration of a new structure is clearly seen. Figure : Layer southeastern room, viewed from west. A. hearth, B. suerimosed floors, C. suerimosed walls. Scale 5 cm. A burnt red clay floor lies to the northeast of this concentration. In the northwestern quadrant, a mutilated clay floor lies adjacent to a clay wall foundation, suggesting that there was a residential comlex in this art of the site. A laterite wall foundation runs a few degrees west of due north in the central art of the square, and there are also several ostholes that were dug from this deth. The surface of layer 4-8 contained, in the southern art of the square, the first indications of a house structure, seen in the configuration of wall foundations (Figure ). These were made of clay and laterite. A badly disturbed laterite floor lay within the rectangle formed by the southeastern corner of the room of this house, over which lay a circular hearth of comacted ash and flecks of char-

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS coal. There were also concentrations of otsherds within and just beyond this room, and the base of a ottery vessel. The red clay floor to the east of this chamber was first uncovered in layer 4-6. A wall running arallel with the clay wall in the centre of the square is found in the southeast corner. Figure : Lower layer, looking south. The grave cuts for burials 8 and 9 can be seen on the surface of the floor within the ossible mortuary chamber. Burial 7 has been exosed. A rectangular clay floor can be seen on the surface of :8 in the distance, cut by a rectangular it. A lane between two buildings runs north-south on the left of the hotograh before turning a right angle. Probably 6th century AD. Layer Layer was a thin occuation deosit under layer structures. On removing the walls of the layer burial chamber and the associated buildings, a further set of walls on the same orientation were encountered. These enclosed in the southwestern quadrant, a solid clay floor containing a rectangular cutting of mottled fill. The rectangle was surrounded on the western and southern sides with rows of ostholes. This looked very like a further grave cut, but when oened, it was found to contain no such remains. The wall to the north of this floor had been cut into with the creation of the grave for burial 7. There is a further structure on the eastern edge of the excavation. It comrises a wall foundation of white clay with a lining of grey clay. Within, there is a series of suerimosed grey or red clay floors (Figure ), and in the southeastern corner, two hearths. A third hearth filled with grey ash was found towards the north of the room. It is evident that this room was occuied over such a eriod of time that reflooring was eriodically necessary. Layer The uer contexts in layer incororated concentrations of Phimai Black otsherds, indicating continued late Iron Age occuation. Burial, an infant, was found in the southern edge of the excavation. It was associated with a small number of glass beads. Historic eriod its were cut down into uer layer, as well as some ostholes. Layer -4 revealed the surface of wall foundations in the southwestern quadrant. Lying adjacent to the northsouth wall, was burial. The orientation of this erson was with the head to the south. Just to the north, burial was encountered in the base of a historic eriod it. It is assumed that this it dug through and disturbed an Iron Age burial. In the northeast quadrant, there were three sets of li to li Phimai Black vessels. The two in close conjunction were found south of a dee cut within which we encountered burial 6. A white clay floor dominated the eastern art of the area oened. Layer -5 included further clay walls and the intervening town lane between two structures. The bases of several its iminged on these walls. One of these had roba-

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure : The mortuary chamber, looking south. Note the surviving art of a floor sealing the cut for burial 7, and the clear grave cuts for burials 8 and 9. The central row of ostholes were robably to suort the roof. Probably 6 th century AD. bly disturbed an earlier adult burial, the bones from which were scattered within the it (burial ). This it also contained the cranium and uer bones of an infant (burial 4). The walkway in the southwestern quadrant contained the limb bones of a further infant (burial 5). In the northeastern quadrant, there are two sets of li to li Phimai Black ots lay at a higher level and to the south of burial 6. This child was interred on a north to south axis, with the head to the north. It was associated with two Phimai Black ottery vessels, an iron earring and an iron sickle by the right hand. The surface of layer -6 incororated the to of the wall foundations and a buttress for the two structures the floors of which were to be uncovered at a lower level. At this juncture, the gas between the northwestern and southeastern walls were designated walkways or town lanes. The bases of three large its dug down during the historic eriod are seen in the northwestern quadrant. Lower layer incororated a series of walls divided by lanes. The interior of the room in the northwest quadrant contained a lastered clay floor through which a rectangular it had been cut. Three burials lay within this chamber, all on the same orientation as the north to south walls (Figures -). One is an adult, and there is also one child and an infant. The adult burial lies within two arallel rows of large ostholes. A hearth is found in the southern half of this room. Li to li Phimai Black ottery vessels were laced in three of the corners of this cham- ber, with a fourth set also in the southeast corner. The northeastern corner was disturbed in antiquity, and no such vessels survived there. There are also some broken ots in the southern art of the chamber. This room is searated from the building to the southeast by a lane that, unlike the rooms themselves, contained occuation debris dominated by broken Iron Age otsherds. No further burials were found in these rooms, but a set of hearths was resent in the southeastern corner room. It is ossible that this comlex reresents a domestic dwelling while the chamber to the northwest was emloyed for interring the dead. An alternative interretation could be that both buildings were occuied, and the burials were cut down through the clay floor after they were abandoned. This seems unlikely given the identical orientation and regular lacement of the graves relative to the room that contains them, and the ossible ritual nature of the recisely located ottery vessels in at least three corners of the room. Moreover, a clay floor or ossibly a collased wall was found artially to cover art of the grave of burial 7 (Figure ). In this resect, it is articularly noted the exenditure of energy devoted to wall construction during this eriod (Figure 4). The foundation is at least 5 cm wide, and comrises white solid clay, of the sort that would have been available during the excavation of the moats to form the retaining banks that ring the site. The interior edge of the wall contains a line of ostholes for timbers that

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS would have suorted the roof. A narrow channel running between the osts and the wall may well have been the foundation for the interior lining of the chamber. The floor abutted the wall and one of the sets of li to li Phimai Black ots was lodged hard against the latter, sitting on the surface of the clay floor through which the grave of burial 8, located in this vicinity, had been cut. Figure 4: The eastern wall of the mortuary chamber. Note its marked breadth, structural ost foundations, and the li to li set of ottery vessels adjacent to the wall. Scale 5 cm. Probably 6th century AD. Layer Layer incororates a series of its, containing broken ottery vessels and animal bones: large deer, cattle and water buffalo. There are also several concentrations of burnt clay. These might well have come from wattle and daub structures. In the southeast corner of the square, there as an extensive shell midden. There are also several ostholes. It is evident that Non Ban Jak was occuied during the early historic eriod for the broken ottery vessels are similar to those recovered from Dvaravati sites in Central Thailand (Figure 5). Figure 5a & b: The historic eriod ceramic vessels THE EASTERN MOUND BURIALS The excavation of Non Ban Jak encountered a residential quarter of the site. However, several burials were identified, three of which were found within a chamber in one of the buildings in circumstances which suggest that this was a ritual room designated for burial. This stems from the fact that the infant burial was located in one corner of the room, and burial 7, an adult, was artially covered by 4

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure 6: Burial 7, a woman interred in a layer room. a floor or ossibly collased wall. It is also the case that the room in question, unlike rooms in the southeastern quarter, which had many suerimosed clay floors, had only two or three thick white clay floors. Two of the lower residences also contained rooms in which an infant grave had been cut through a floor. The best reserved 5

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure 7: Burial 9. was centrally laced within a room, suggesting that it was associated with the building rather than a later burial fortuitously interred into an unrelated structure. Burial comrises the bones of an infant in layer -, Burial 5, a second infant, was found within a lane in layer -4. Burial 4, a further infant, was found at the base of an historic eriod it in layer -4. It is likely that the it damaged but did not comletely disturb this infant burial when it was dug. Burial 8 is the comlete grave of an infant, found in the southeastern corner of the mortuary chamber which also contained the remains of burials 7 and 9. No mortuary offerings were found in this grave. Burial 7 was found at a deth of.6 m below datum with the head orientated slightly to the northwest (Figure 6). The grave lay within a mortuary chamber, ositioned between two rows of large ostholes. The body was found 6

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) on the back, about 4 cm below the floor through which the grave was cut. Personal ornaments comrised five bronze rings on the right hand, a bronze bangle on each wrist, a lead earring and a bimetallic ring on the chest. Three ottery vessels were resent, a Phimai Black bowl by the right hand, a large cord-marked vessel by the right ankle and a further fragmentary vessel by the left ankle. An iron knife blade had been laced by the right elbow. Burial 9, a child, was found within the same mortuary chamber and on the same orientation, at a deth of. m below datum. The head was found within a ottery vessel, and the feet in another. This child wore a bronze bangle on each wrist and bronze earrings (Figure 7). Burial 6, another child, was found interred with a large ceramic vessel, an iron earring and iron sickle. Figure 8: The southern section at squares AA and BB in the western mound. THE WESTERN MOUND The stratigrahy of the excavation that straddled the low ground and the rise u to the western mound varied remarkably. There were no structures and virtually no features in the former art of the square, until the laterite natural substrate was encountered at a deth of about m below datum. At this oint, human burials were found either lying on the laterite, or slightly excavated into it. However, with the rising ground to the west, a series of walls and floors were encountered in a comlex sequence encomassed within four layers (Figure 8). The relationshi between human graves and the structures has underwritten the recognition of four mortuary hases, the earliest of which corresonds to those found resting on the natural laterite. The occuation of this art of the site began in the 4th century AD and lasted for about -4 years. This eriod, and the associated material culture, corresonds to the fourth and final hase of the Iron Age at Noen U-Loke (Higham et al. 7) and extending into the eriod of the Dvaravati state. area, burial being an extended, suine inhumation. With layer -, the eastern art of the excavated area included three infant jar burials, and some comlete ots. Proceeding to the west, a large ceramic kiln was oened. Mortuary hase (MP) burials were now encountered in the uer contexts of layer 4. Burial 8 was found covered by a clay floor. A row of ostholes on the same orientation as the walls ran north south across the square, and in the northwestern corner, burial 5 was found oriented uniquely from west to east. This basal layer accumulated directly over the natural substrate. MP burials concentrate in the eastern art of the excavated area. There seem to be two grous searated by a four metre ga. The many infant jar burials during this eriod suggest a remarkable degree of infant mortality. Layer The transition from layer to is barely noticeable. Again all the activity is confined to the western art of the excavated area with a system of walls and floors, through which burials were inserted. There is also evidence for occuation in the form of its and concentrations of ottery sherds. Burial 7 was cut through a floor, at the corner of a clay wall. Burials 76 and 77 lie within the clear grave cut through the floor in AA (Figure ). Further walls and floors aear in BB and. With layer -, burial 49 was cut into the edge of a clay wall. Further west, a series of burnt floors suggest that a fire had taken lace rior to the interment of burials 76-8 through a floor. Slightly deeer in layer -4, concen- Layers -4 The uer contexts of layer contained the remains of occuation, including clay floors along with comlete ceramic vessels and scatters of ash and concentrations of otsherds. Infant and adult burials were cut down to this layer (Figure 9). Proceeding into layer, it was found that domestic activity was concentrated in the western art of the excavated area, with clay floors under concentrations of ottery, ash and stones. Burial 84, an adult, was found with no cranium. Two infant burials were found in the eastern 7

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS which was interred abutting the eastern side of the wall. Many ash sreads further reflect burning. In BB, burial 8 was located. This individual wore two bronze belts matched at Noen U-Loke during Iron Age. Layer According to the villagers, the to of the site in this area has been levelled with a bulldozer. Hence the late Iron Age burials are found very close to the resent surface. The western art of the excavated area, barely 5 cm below the surface, had a number of concentrations of otsherds, shellfish, animal bone and some comlete vessels, indicating that this art of the site had been occuied during the late Iron Age. MP 4 burials were encountered in the uermost context of this layer. In Y, there is an concentration of burnt clay. Scatters of concentrated otsherds and shellfish as well as some ostholes suggest that occuation also took lace in this art of the site. With layer -, there is evidence in the western art of the excavated area for structures reresented by clay walls and floors. MP4 burials continued at the level, together with concentrations of otsherds and shells. An area of burnt clay is found in Y. Summary There was evidently a concentration of activity in the western art of the excavated area but nothing to the east. Structural remains comrise wall foundations and clay floors. The former have ostholes embedded in them, resumably for the walls, while walls also terminated with a large block of stone. Burial 7 was cut through a floor within a building, but the configuration of the walls to the west suggest the resence of a lane within the settlement. Burial 46 was also cut into a clay structure. There are concentrations of otsherds, that in AA being within the room containing burial 7. A wall in AA suggests that structures continue in a southerly direction to the interior of the settlement. THE WESTERN MOUND: THE MORTUARY SEQUENCE Mortuary hase The burials of MP were disosed in two grous, one centrally in the excavated area and the other at the eastern end (Figure 9, 4). Thirty-two graves were identified of which 6 involve infants most of whom were interred in lidded ceramic vessels. Of the adults, there were three females, four males and of undetermined sex. If a reflection of the real situation, the number of infant burials is one of the highest recorded in rehistoric Southeast Asia. The adults were buried suine, on a north to south axis, with a reference for the head directed to the north. One adult was interred in a flexed osition and was robably a secondary burial. Mortuary offerings were sarse. Indeed comared with the wealth of the contemorary graves at Noen U-Loke, this grou of individuals, who might well reresent the initial settlers of Non Ban Jak, were imoverished. One or two ottery vessels were Figures 9-: Western mound burial locations in Mortuary Phase (to left), (to right), (bottom left) and 4 (bottom right) trations of ash and otsherds, ostholes and its suggest that at this juncture, the western art of the site had do mestic occuation. In AA, a clay wall was identified on a recise north to south orientation. Layer -5 incororated a clay floor associated with the north to south wall in AA, together with infant burial 8, 8

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure : Burials 76 and 77 of Mortuary Phase were cut through a floor, on the same orientation as the walls of a room. found with most of the adult dead, with just two forms reresented: a cord-marked globular vessel and an oen bowl. There were no iron mortuary goods. One man wore 6 bronze bangles, and one or two other bangles, finger and toe-rings of bronze were also recovered. Other ornaments were equally sarse; one man wore an agate and a shell endant, and another robable male, a silver coil on each ear. The flexed burial was associated with a collection of red glass beads. Animal bone offerings were rare, and confined to ig bones. All but two of the infants were interred in lidded ceramic vessels, most of the forms being aralleled at Noen U-Loke and Ban Non Wat during Iron Age 4. The two other infants were interred like the adults, suine, and with the head orientated to the north. Mortuary offerings were dominated by ornaments, although never in the abundance seen at contemorary Noen U-Loke. The infant in burial wore seven bronze bangles, and other bronzes were found in the form of a ring, an earring and anklets. The infant in jar burial wore three bronze belts. Glass beads were found with eight of the infants. There is a long tradition in the uer Mun Valley sites from the Neolithic onwards for the association of bivalve shells with the dead, and this was continued with four of the MP infants. Burial 59 also wore two Anadara shells ierced for susension. This is a marine secies, indicating long distance exchange. The only MP iron offerings come from burial in the form of two bangles. Table : The finds associated with intact adult burials of mortuary hase 9 Burial 9 4 5 4 64 Sex F M M F M -?M Age A A A MA OA A MA Orientation of head N N N N N S N Pots Bronze bangle 6 Bronze finger ring Bronze toe ring Bronze anklet Silver ear siral Stone adze Glass beads Agate endant Shell endant Sindle whorl Pig bones

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Table 4: The finds associated with mortuary hase infants Burial 6 7 8 6 7 9 4 48 54 55 56 59 6 Pottery vessel Bivalve shell Bronze bangle 7 6 4 Bronze anklet Bronze ring Bronze necklace Bronze earring Bimetallic ring Iron bangle marine shell endant Glass beads : resent egg, again matched at Iron Age Noen U-Loke and Ban Non Wat. Two cowrie shells were associated with burial 9. Other offerings comrised a bronze anklet, bimetallic rings, an Anadara marine shell, one bead of carnelian and a sindle whorl. Mortuary hase Burials of MP concentrated at the western art of the excavated area (Figures, 5). There were five male, three female and four other adult graves, together with infants. The roortion of infants, 64% of the samle, exceeded even that of the receding hase. The ascrition of burials to MP was based on their being sealed by later floors, and being at a higher level and different orientation to MP graves. Since there are no suerositions, however, the recedence of MP remains under review. The orientation was now with the head either northwest or southeast, with one male being interred with the head to the east. Mortuary offerings with the adults continued to involve globular cord-marked vessels and oen bowls. There were also a few bronze bangles and rings. The first iron sickle and iron knife were identified, as well as a bimetallic iron and bronze ring of a size frequently found at Noen U-Loke. Four of the five intact adult burials contained at least one agate bead, and there were also bivalve shell offerings, beads and sindle whorls (Table 5). Fourteen of the infants were found in mortuary jars, which were often lidded. The slightly older infants were buried like adults, suine and on the same orientation. Infants were accomanied by a more varied set of grave goods then the adults, and some could be described as being moderately wealthy. Thus burial 99 wore nine gold beads. Burial 86 wore ten agate beads, glass beads, four bronze bangles and a bronze ring. There were also two lums of clay. Clay was regularly found in the late Bronze and early Iron Age graves at Ban Non Wat, and could indicate either an interest in ottery making or dying fabric. A large enclosed kiln was found in the same art of the site as this burial. Burial 4 contained a bird s Mortuary hase Burials of the third mortuary hase were linked through their common association with walls and floors which are interreted as residential buildings. The graves are set out in rows, on an identical orientation as the walls, and were often found to have been cut through the floors within. The roortion of infants fell to 4% of the samle. Of the adults, there was a reonderance of males, with eight against just two females. There was also one adult of unknown sex. There was a singular uniformity about the adult mortuary ritual (Figures, 6). Graves were laid out on a northwest to southeast axis. Associated ceramic vessels continued to comrise globular cord marked ots and oen bowls. Most were accomanied by an iron sickle and an iron knife. The former had been hafted with a wooden handle, traces of which have survived. Knives had been contained in a sheath fashioned from broad, woven fibre of some sort (Figure 8). To a background of uniformity, there were some individual characteristics. A young man in burial 6 wore a gold ring. Burial 49 had the only bivalve shell. The woman in burial 75 wore bronze finger rings. Bronze ornaments were more abundant than in revious hases and one man, burial 8, wore two bronze belts with a circular cross section identical in form to those found in two male Iron Age graves at Noen U-Loke (Figure 6). Burial 68, another male, might

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure 4: Burial, an adult male, was found at a deth of.89 m, with the body orientated on a north to south axis. The dead erson wore a bronze bangle on each wrist, and around the neck, an agate and a shell endant. A ottery vessel might have been associated, and it is found to the right of the cranium. Glass beads were located in the neck area and were resumably strung as a necklace. Beyond the excavated area, a ig s cranium and a bronze toe ring were recovered.

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure 5. Burial 96 of mortuary hase contained the skeleton of old adult, robable male, with the head oriented SSE. The skull was found at a deth of.58 m below datum. There were few grave goods. A ottery vessel on the abdomen had been cut through when a large osthole was driven through the grave. There was an iron sickle alongside the right uer arm, bronze rings on each hand and a bivalve shell inside the left inste. Three glass beads were also associated.

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure 6: Burial 8 was a mid to old male of Mortuary Phase, found at a deth of.m below datum with the head oriented NNW. It is the richest interment of mortuary hase and was sealed under a clay floor. He was accomanied by six ottery vessels disosed over the uer body. He wore two bronze belts, of a form aralleled at Noen U-Loke, where two men wore three and four such belts resectively. Other bronzes included three earrings and one bimetallic ring of bronze and iron. Two bivalve shells had been laced with the corse, one by the right ankle and the other by the left hand. An iron sickle was found over the uer body, and an iron knife blade on the uer chest. The radiocarbon determination for a bivalve shell from this burial laces it in the 4 th century AD. According to the dimensions of his belt, this man had a waist measurement of about 6 cm (5 ). well have been a otter, since he was interred with a lum of clay and a burnishing stone. Unfortunately, burial 46 had been badly disturbed and only the foot bones were intact. This man wore 4 bronze anklets. Glass beads were extremely rare, and animal bone offerings also notable for their near absence. Pseudomorhic remains of wo-

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Table 5: The finds associated with intact adult burials of mortuary hase Burial 57 84 96 5 8 Sex F F?M M M Age Old Old Old Old Old Orientation N NW SE E SE Pots Bronze bangle 4 Bronze finger ring 6 Bronze ring Iron sickle Iron knife Iron fragment Bimetallic ring Glass bead Agate bead Bivalve shell Sindle whorl : resent ven fabric on some of the bronze ornaments suggest that the dead were either interred wearing clothing, or were covered by a shroud (Figure 8). Two of the infants of MP were buried with the head in one ceramic vessel, and the feet in another. This is matched by a burial on the eastern mound, and by burial 9 at Noen U-Loke. The ottery vessels at both sites are identical in form. Burial 7, a child aged about -4 years at death, was interred like the adults, suine with the head to the southeast. The two associated ots again conform to the norm for this site, a globular cord-marked form, and an oen bowl. This grou of infants had few associated offerings. Eight intact graves shared five bronze rings, two earrings, one toe ring and two bronze bangles. There was an iron oint, two bimetallic rings and an iron ring. Unlike revious infants, glass beads were absent. Table 6: The finds associated with the infant burials from mortuary hase Burial number 79 5 8 86 9 9 9 94 95 97 99 4 7 Pottery vessel 5 Bronze bangle 4 9 9 Bronze ring 8 Bronze earring Bronze anklet Bronze fragment Bimetallic ring 6 Bivalve shell Marine shell Cowrie shell Egg Carnelian beads Glass beads Agate beads Gold beads 9 Sindle whorl Clay : resent 4

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Table 7. The burials associated with intact adults burials of mortuary hase Burial 49 6 68 7 7 75 76 77 8 8 Sex Male?Male Male Male Male Female Female Male Male Male Age Mid adult Young adult Adult Mid adult Young-mid adult Adult Mid-old adult Mid-old adult Mid adult Mid- old Orientation N S NW SE NW SE NW SE SE NW Pots 4 6 Bronze earring Bronze bangle 6 4 9 Bronze finger ring 6 5 Bronze ring Bronze belt Iron sickle Iron knife Bimetallic ring Gold ring Red ochre 4 Clay Burnishing stone Bivalve shell Animal bone Glass beads : resent ground surface, some have been disturbed. There are two male, two female and two adult graves, together with the remains of six infants. Infants thus comrise 5% of the samle. The definition for MP4 graves, is that they overlie the clay walls and floors associated with the receding hase. The mortuary ritual continued virtually unaltered. The orientation remained the same save for one adult found with the head ointing to the south. The same range of ceramic vessels were found, although some of the oen bowls now had a footed base. Burial 65 included an iron sickle, and iron knives were found with three of the four intact adults. Two were also interred with sindle whorls. Bronzes were rare, with just five finger rings. Burial 69 wore a gold coiled earring. Two of the four infants were buried with the head and feet within ottery vessels of the same form as those associated with MP. The infant in burial 66 wore an ornament of glass beads, but other mortuary goods were rare, comrising two bronze bangles, seven bronze ear rings, and iron bangle, iron oint and a clay counter. This set of burials belongs to the very end of the rehistoric eriod as currently known in the uer Mun Valley with Table 8: The finds with mortuary hase infants Burial 4 5 47 58 7 74 78 8 Pottery vessel 4 Bronze ring 5 Bronze ear ring Bronze toe ring Bronze bangle Bimetallic ring Iron oint Pig bone Iron ring Mortuary hase 4 The burials of MP4 are confined to the western art of the excavated area (Figure ). Being so close to the modern 5

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure 7: The distribution of features in layers :4 to : showing the burial of MP individuals within residential buildings.every likelihood that they overla with the establishment of the Dvaravati civilization in Central Thailand. 6

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Table 9: The finds associated with intact adult burials of mortuary hase 4 Sex?female Age mid adult adult mid adult old adult NW S NW SE Pots 4 Bronze finger ring 4 The distribution of burials in lower layer and uer layer (Figure 7) takes the form of two rows. When the configuration of walls and floors is suerimosed, a attern emerges whereby the burials align with the walls, and cut through floors. The relative deths of the floors and the walls against that of the graves suggest that the burials were located within these structures. Moreover, the floors contain concentrations of ottery sherds and faunal remains suggesting that they were occuied as houses. Hence, there is a serious likelihood that as in the first season of excavations at this site, the Iron Age inhabitants followed the ractice of residential burial. Iron sickle Table : The finds associated with infants of mortuary hase 4 Iron knife Iron ring Gold earring Red ochre Sindle whorl Clay Shell bead Burial Orientation 8 45 65 69 female male The western burials: summary The nature of the mortuary rituals at Non Ban Jak requires a detailed assessment of the relationshis between human burials on the one hand, and the structural evidence on the other. The latter comrises the foundations of walls and the associated floors. Walls were made from clay and wooden osts, the latter robably being the framework of studs to suort wattle and daub walls. The floors were made of white or red clay, and some show evidence for burning. These walls and floors were also associated with evidence for occuation in the form of broken ottery vessels, faunal remains, concentrations of shellfish and clay-lined hearths. In the season, one hearth area within a residence was associated with large quantities of carbonised rice. Due to the rebuilding of houses over the foundations of earlier structures, the latter were often disturbed in rehistory, and the remains survive in a badly damaged condition.table 7: The finds associated with intact adults burials of mortuary hase This is seen just over the surface of the hard, laterite natural substrate in square AA, where wall fragments lie at a deth of.97 to. m below datum. Burial 8 overlies these, the base of the grave lying at a deth of.87 m. The burial is intact, and was laid out on the same orientation of the walls found at a higher level. In turn, this burial was covered by a clay floor and fragmentary wall foundations, the former barely 4 cm above the human bones. It is considered unlikely that this conjunction is fortuitous, and more likely that the covering of the grave with a floor was a deliberate act reflecting interment within a residential or ossibly a dedicated mortuary structure. Burial 7 9 66 7 9 Pottery vessels 4 Bronze bangle Bronze ear ring 7 Iron bangle Iron oint Glass beads Clay counter THE MORTUARY OFFERINGS Ceramic vessels The ottery vessels in adult burials from the successive mortuary hases show little variation over time (Figure 8). The commonest form was a globular vessel with a restricted neck and everted rim. These are very similar to the village cooking ots seen today, and one found with burial 84 of MP contained carbonised rice grains. The second common form is an oen bowl. These all fall into the Phimai Black tradition, the interiors being decorated with designs imarted with a burnishing stone. With MPs and 4, some of these bowls were given a edestalled base. The regular conjunction in burials of the two forms suggests that one contained food, and the other was used in life as a latter for eating from. The vessels from infant burials fall into two grous. Large, lidded ots were used to contain the bodies. Cordmarked globular ots and oen bowls were often laced with the dead infant (Figure 9). The form of most burial jars was similar during MPs and, while those from the two later hases differed. MPs and also included different forms for containing dead infants. One from MP has a restricted neck and flared rim, while a articularly fine vessel from MP was burnished all over the outer surface. Iron and bronze artefacts Iron at Non Ban Jak was forged into tools and ornaments, but there is very little evidence for weaonry. The con- 7

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure 8: Ceramic vessels from adult burials. centration of iron slag in the eastern mound indicates local ironworking but it is too early to be sure whether this reresents smelting or smithing. The issue of the local exloitation or otherwise of laterite ore also needs further study (Cawte and Boyd ; Pryce and Nataintu 9). Tools are dominated by the airing of a socketed sickle with a ridged knife (Figure ), both of which recur in identical form in the later contexts at Noen U-Loke. Socketed hoes, which were found in the earlier Iron Age at Ban Non Wat, are absent. However, it is known that large socketed loughshares were in use during this eri- 8

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure 9: Ceramic vessels from infant burials. -od at Non Ban Jak following the discovery of one in a kiln in the eastern mound excavation. There is one sickle in MP graves, and they became regular offerings in those of MPs -4. They are so consistently resent in MP-4 that it seems highly likely that this social grou was closely involved in rice cultivation. Rings made of bronze and iron were absent from MP and 4 graves, but common with MP -, when at least 5 were recovered from the intact graves of the former, and seven with the latter. Two iron bangles were found with a MP infant, an iron ring with MP and a further bangle in MP4. 9

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure : Iron sickles and knives from Non Ban Jak burials. Bronze was emloyed for a variety of ornaments, articularly bangles and rings for the fingers, ears and toes. The man in MP burial 8 wore two belts, and the MP infant burial wore three, each having the same tye of fastening device as was seen on those at Noen U-Loke (Figure ). Intact burials from MP contained 48 bangles, six anklets, three belts, and a single ear, finger and toe ring. For MP, there were 8 bangles, three earrings, seven finger rings and rings of unknown location on the body. There were also many bronze ornaments in MP burials: bangles, ear rings, 45 finger rings, two belts and a toe ring. The numbers fell away, albeit with fewer overall burials, in MP4, with two bangles, seven ear rings and five finger rings. These figures ale before the

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure : Bronze ornaments from Non Ban Jak burials. Non mortuary ottery vessels Pottery vessels were recovered from three non-mortuary contexts. The first, a regular characteristic for the Iron Age of the uer Mun Valley, is the lacement of lidded bowls in ritual contexts, such as the corners of a room in which burials were located. Comlete ots were also found in kilns, or associated with one of the kilns as wasters. The third context is urely domestic, for examle in a wealth in bronze jewellery associated with IA and IA 4 individuals at Noen U-Loke. The same situation obtains for other ornaments. While the range is similar to later burials at Noen U-Loke, including gold, silver, carnelian, agate and glass, the quantities are far less. Two unusual ornaments from Non Ban Jak are the marine shells ierced for susension, a gold finger ring and gold siraliform earring (Figure ).

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure. Ornaments from Non Ban Jak burials. A. agate endant, burial EP; B. shell endant, burial EP; C. silver ear coils, burial 64 EP; D. agate beads, burial 96, MP; E. Anadara shell endants, burial 59, MP; F. glass beads, burial MP; G. cowrie shell, burial 9 MP; H. carnelian bead, burial 95 MP; I. agate beads, burial 86, MP; J. gold beads, burial 99 MP; K. gold finger ring, burial 6, MP; L. shell bead, burial 69 MP4. burnt and abandoned kitchen, or its located just outside a residence. Pottery vessels from the two kilns in the eastern square have been reconstructed. The circular kiln contained eight ots (Figure ). One of these was regularly used in the first two mortuary hases for interring infants. All the other forms were also emloyed as mortuary offerings, and broken secimens were also recovered from ritual and occuation contexts. The smaller of the eastern mound kilns was oval in shae, and contained a single very large vessel (Figure 4). While not identical, it does resemble the mortuary ot containing burial which was located on the western mound. Several ots were also reconstructed from those found as kiln rake out from the larger of the two eastern mound kilns. Again, they are similar in form to those found in graves as mortuary offerings. To judge from the quantity of collased clay daub covering each kiln, it is suggested that the kilns themselves were enclosed by a domed body constructed of daub liberally temered with rice straw. Two of the kilns had been emloyed to fire a single, very large vessel. The other was still filled with ots of different forms. The closed nature of the kilns could well exlain how the tyical late Iron Age ots were so often fired in a reducing atmoshere. We do now know of any other Iron Age kilns in the Mun Valley, and certainly the firing technolo- gy differs markedly from that seen in most otting villages today. Ritual deosits comrising a lidded ot were first identified at Non Muang Kao and Noen U-Loke. At Non Ban Jak, their relationshi to mortuary areas is clearly defined. These are very finely decorated with attern burnished designs on the interior (Figure 5). Three sets were found in the surviving three corners of the chamber found on the eastern mound with three burials within (Figures -). Others have been found in the vicinity of buildings and burials, albeit in the case of the latter, not within the grave itself. A selection of the ots from occuation contexts that could be reconstructed reveals similarity in form to those laced with the dead. One vessel, cat. 69 in figure 6, was found in the kitchen area of a house on the eastern mound, and was clearly used for cooking. THE SPINDLE WHORLS AND TEXTILES Most of the 5 sindle whorls recovered were made from fired clay, with a few manufactured from bone or turtle caraace. They were used to extend the length and tensile strength of fibres. Five burials (B6, B8, B64, B, B5) contained whorls as mortuary offerings. The rest came from occuation contexts. The resence of whorls in all hases of the site reveals technological continuity

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure : The ceramic vessels found in the kiln B 5:6 feature. from the Bronze Age contexts seen at Ban Non Wat through to the late Iron Age. However, the relatively small number of such secialised tools at Non Ban Jak might suggest small-scale household roduction of textiles. Over a thousand whorls were recovered from Ban Non Wat, only 8 km to the east. If this is not a function of the greater area excavated, it could be seen as the result of more intense, secialized manufacture of cloth. The energy invested in cloth roduction at Non Ban Jak is more comarable with that seen during the final Iron Age at

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure 4: The ceramic vessel found inside the kiln in B 5 surface of feature. Noen U-Loke. The functional attributes of the tools from both these sites are very similar. Cat. 5 is the most striking whorl in the Non Ban Jak assemblage (Figure 7A). This atyical whorl tye, in the form of a door knob, comes from a late context, dated to the 6th to the 8th century AD. While not widely aralleled in Southeast Asia, it is also seen in the late Iron Age buri- als at Tha Kae in Central Thailand. Whorls of this tye first occur at South Indian archaeological sites on rehistoric trade routes where they were found with iron sindles (Cameron ). This exotic form oints to Indian contact. It is also far from coincidental that basic flat disc whorls from the uer layers are areciably lighter than those from earlier sequences, and would have been better 4

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure 5: The ceramic vessels from occuation contexts. Cat. 77-8 A 5: F; cat. 47 B 5: F; cats. 79-8 B 5:; cat. 89 X : F; cats. - A 4:8 F; cats. 75-6 A 5:A building B; cat. 87 B 5:4, cat. 84 X : F; cat AA 4: F6; cat. 4 AA 4:4 F7; cat 55 X :4 F. These vessels were commonly found as airs, li to li, laced in strategic arts of buildings 5

HIGHAM ET AL.: THE EXCAVATION OF NON BAN JAK, NORTHEAST THAILAND - A REPORT ON THE FIRST THREE SEASONS Figure 6: The ceramic vessels from occuation contexts Cat. B 5: F; cat. 44 B 5: F; cat. 87 B 5:6 F; cat. 88, cat. 8 B 6: F; cat. 8 B 6: F; cat. 9 B 5: F; cat. 69 B 5:, floor of building B; cat. 7 A 5:. Cat. 69 is art of a stove found on the burnt floor of building B. cat. 85 B 5:6 F, cat 8 B 4:9 feature ; cat. 4 B 5: feature. 6

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4)) Figure 7: Sindle whorls from Non Ban Jak. A. cat. 5, A -4, B. cat. 7 C. cat. 7, B -, D. cat 485, burial 8, E. cat. 488, burial 8, F. cat. 489, burial 8, G. cat., B -, H. cat 67, burial 64, I cat. 8, B 6-, J. cat 5, A 4-. Scale: diameter 7A is cm. adated to sinning lighter fibres robably cotton (Gossyium s.), which was also introduced from the Indian subcontinent during this eriod. The whorl thus joins many other instances of Indian insiration that had been roceeding for centuries before the occuation of Non Ban Jak, much of it surely through Southeast Asian intermediaries. Mineralised fabric seudomorhs as well as coarsely woven scabbards have been identified in the mortuary contexts (Figure 8). Their material and structural comosition range from simle laited bamboo knife scabbard (cat. 655, B.69) to a coarse : tabby weave fabric (cat. 8 B.76) and a very finely woven war-faced weave on an iron knife also in B76. The resence of textile seudomorhs on only one side of certain artefacts suggests the use shrouds or clothing. Those found on both sides of the metal tools and bangles (cat. 788) indicate that the artefacts were robably wraed in fabric unless ost burial movement occurred. cuation areas including stoves and hearths (context nos., 6, 69). Of articular interest was the collased kiln which yielded, tuyères, daub and raked out waste roduct. Sub-samles from these contexts were taken. The daub is robably a mixture of clay and rice straw (context no. ). There are samles containing mostly comressed rice husk and some grains and are related to the collased kiln (context nos. 88, 4, 54). The inhabitants at Non Ban Jak were relying on rice as art of their subsistence regime, but were also using the rice rocessing waste in other activities such as kiln construction, signifying a ready suly of rice roduction byroducts. The lant remains assemblage shows 8% of the total is made u of rice lant arts including rice caryoses, husk and sikelet bases (Figure 9). The examination of the rice sikelet base rachilla scars show redominantly domesticated rice. The abscission scars in 96% of the rice sikelet bases (n=98) had the tyical irregular and gouged out deressions found in domesticated rice. This is not surrising considering the age of the site, since the earliest evidence for domesticated rice in Thailand dates to -5 BC at Khok Phanom Di (Thomson 996), and other Thai sites that have yielded rice remains dating to the Bronze and Iron Ages have also demonstrated redominantly domesticated rice (Castillo ). The weed assemblage at Non Ban Jak is of considerable interest. Although the contexts examined did not yield abundant wild and weedy secies (% of total lant remains), one articular context (NBJ ) contained sever- THE BOTANICAL REMAINS Preliminary archaeobotanical results came from fourteen contexts at Non Ban Jak. An average of thirty-six lant arts er litre were recovered in the contexts analysed (Table ). All samles contained rice lant arts and in half of the contexts, rice was the only taxon found. Initial visual insection of the samles showed rich deosits of rice grains in secific contexts related to oc- 7

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) Figure 8: Fabric and a woven scabbard. To: cat. 799, a knife from burial 76, middle: cat. 8, a sickle from burial 76 and bottom: cat. 655, a knife from burial 69. 8

JOURNAL OF INDO-PACIFIC ARCHAEOLOGY 4 (4) with the centre of distribution in Southeast Asia (Flora of China). There are several Fimbristylis secies reorted as weeds of rice in Thailand, and many are found in wetland fields. However, there are also secies that occur in uland rice systems. More work will be undertaken to identify the Fimbristylis seeds to secies to determine the cultivation ractices engaged by the eole of Non Ban Jak. Another context (A -9 feature ) also contained the secies Chara zeylanica, a green algae from the Characeae family, which is reorted as a weed of rice in Thailand articularly in translanted, dry-seeded and wet-seeded rice (Moody 989). The weed assemblage from Non Ban Jak is very different from that analysed in Bronze Age 5 Iron Age Ban Non Wat (ca. 79- BC), and the Metal Age sites Khao Sam Kaeo and Phu Khao Thong (ca. 4 BC). The main difference lies in the distinct absence of the dryland weed secies Acmella aniculata, which is the redominant weed in the above mentioned sites; but also the aearance of a new suite of weeds, including Fimbristylis s. and Chara zeylanica, which u to now has not been reorted in any other site in Thailand. Figure 9: Breakdown of lant remains into main grous. SUMMARY AND CONCLUSIONS The two areas excavated have furnished radiocarbon determinations indicating that Non Ban Jak was a settlement initially occuied during the final hase of the Iron Age in the uer Mun Valley. The oened areas on the eastern and the western mounds have revealed a matching record involving burial within houses, together with industrial activity. The material culture and the mortuary rituals closely match those identified during IA hase 4 at the nearby site of Noen U-Loke. The three or four centuries in question are vital in the transition from rehistory into early historic kingdoms, for it was during this eriod that inscritions reveal the establishment of states at, for examle Wat Phu, and a military exedition u the Mun Valley led by Citrasena, the future king Mahendravarman (Vickery 998). Indirect contact with India has a long history in the uer Mun Valley, evidenced in the earliest Iron Age cemetery at Ban Non Wat from at least the 4 th century BC in the form of glass, carnelian and agate jewellery. A most intriguing issue raised by the new evidence for an overla of at least two centuries between the late rehistoric occuation of Non Ban Jak and the establishment of cities and temles in the lower Mun area at, for examle, Kurukṣetra (Lorillard 4). The lack of any evidence, as yet, of further Indian influence in the form of religious motifs, texts, or seals calls into question the degree to which the new states of Chenla enetrated into the late Iron Age communities of the uer Mun Valley. In this context, it is stressed that the final occuation of Non Ban Jak involved its containing ottery vessels that aear similar in form and finish to those of the Dvaravati states of Central Thailand. These rovide a shar contrast to the redominantly black, burnished ceramic reertoire of the late Iron Age Phimai Black tradition as may be seen in comaring figures 5 and 8. The establishment of a new moated settlement late in the Iron Age hints at a growing oulation, and it is consid- Figure 4. Fimbristylis s. and Chara zeylanica seeds. Table Summary statistics of the botanical datasets from flotation in Non Ban Jak season. Number of Secimens 856 Number of Identified Secimens 858 No. of samles 4 Plant arts er litre MEAN 6 Plant arts er litre MIN Plant arts er litre MAX 89 No. of taxa MODE No. of taxa MIN No. of taxa MAX 4 Based on samles fully sorted (n=4) Based on samles fully sorted (n=4); excludes unidentified lant remains, arenchyma and modern lant arts with the centre of distribution in Southeast Asia (Flora of China). There are several Fimbristylis secies reorted as weeds of rice in Thailand, and many are found in wetland fields. However, there are also secies that occur in uland rice systems. More work will be undertaken to identify the Fimbristylis seeds to secies to determine the cultivation ractices engaged by the eole of Non Ban Jak. Another context (A -9 feature ) also contained the secies Chara zeylanica, a green algae from the Charace al secies from the genus Fimbristylis (Figure 4). The genus Fimbristylis is made u of more than secies 9