From the iron age to Angkor: new light on the origins of a state.

Higham, Charles

[ILLUSTRATION OMITTED]

Introduction

The moated sites of north-eastern Thailand were first brought to international attention by Peter Williams-Hunt in an article published in Antiquity in 1950 (Williams-Hunt 1950). Having taken a comprehensive series of aerial photographs just after the Second World War, he prepared a map showing a dense concentration of settlements then assumed to be prehistoric, but was unable to propose their date and context for lack of any information. Sensibly, he wrote that: "[t]he total number of archaeological remains is enormous. In some areas there is scarcely a square mile in which there are not at least three or four sites of varying size and date. Detailed analysis from the air will take many months and there is the work of a lifetime on the ground" (Williams-Hunt 1950: 32). In his concluding remarks, he also stressed that: "[f]urther comment would be futile. The excavators spade alone will provide the final answer" (Williams-Hunt 1950: 35). Half a lifetime after his words were published, Amphan Kijngam and I summarised our fieldwork on some of these sites again m Antiquity,

and my re-reading of that paper reveals how vital it is to have sufficient data within a reliable chronological framework. We concluded that: "[l]ike William-Hunts consideration three decades ago, this is but a way station on the road to a proper understanding of Thai prehistory. We are loath to offer firm conclusions until further fieldwork has strengthened our data base" (Higham & Kijngam 1982: 108).

I feel that after a further 30 years of spadework, much of it within moated settlements, another review is timely. Since 1998, I have co-directed with Thai colleagues excavations at four moated sites in a study area located in the upper reaches of the Mun River (Figure 1). Our results dovetail with and are best reviewed in conjunction with two advances in related disciplines that provide new social and geospatial contexts within which to assess the implications of the new excavated data. The first is Michael Vickery s reinterpretation of the so-called Chenla period inscriptions (Vickery 1998). Chenla is a name derived from Chinese historical records often used to describe an essentially protohistoric period dated between AD 550 and 800 that followed seamlessly from late prehistory. The need to incorporate his detailed analyses of the social changes that took place during those vital centuries in any review of the transition to the state of Angkor is a reminder to avoid Yoffee's stricture that "[t]he Mesopotamian historians myopia is, of course, matched by the prehistorian's tendency to fold his or her tent and steal away at the dawn or history" (Yoffee 2005: 199). The second is the startling new information provided by remote sensing in general, and the lidar surveys at Angkor in particular, that have revealed the magnitude of the early urban complex of Mahendraparvata on the Kulen Plateau, which was probably inspired by Jayavarman II, the founding king of the first Angkorian dynasty in about AD 800 (Evans et al. 2013).

Fresh insights into the origins of social inequality also encourage a new synthesis of the articulation between late prehistory and the origins of the early states. Thus Yoffee (2012), with reference to Mesopotamia, has stressed the need to identify emergent properties in the former that can, after a long gestation, rapidly give birth to city states. These include increasing populations, long-distance trade in exotic materials and the formation of interaction spheres which stimulated new avenues for the appropriation of valuables and the gaining of prestige. To these, we can add the importance of agency, the role played by resolved and charismatic leaders in promulgating social change through, for example, innovations in warfare, the attraction of supporters, agricultural intensification, significant changes in ideology and control over the deployment of surpluses (Flannery & Marcus 2012).

The prehistoric Iron Age: 420 BC-AD 600

The chronological backbone to this review therefore has six phases. The first four cover the prehistoric Iron Age, followed by the period of Chenla and, finally, the first dynasty of Angkor (Figure 2). Of the four moated sites opened by excavation, Ban Non Wat has the longest sequence, divided into two Neolithic, six Bronze Age and three Iron Age (IA) phases. Noen U-Loke has all four Iron Age phases, while Non Muang Kao was occupied during IA3-4 and Non Ban Jak only during the final phase of the Iron Age. IA1 contexts at Ban Non Wat (420-100 BC) were dominated by an inhumation cemetery that extended east from a series of Late Bronze Age graves. The sample is large: 57 male, 34 female, 18 adults where the sex could not be identified and 33 infant burials. Many mortuary offerings, including the form of the pottery vessels, continued virtually unchanged. The differences were seen in the first iron offerings, together with very rare items of glass, carnelian and agate jewellery (Higham 2011). The transition from the Late Bronze Age cemetery into the Early Iron Age has been radiocarbon dated on the basis of bivalve shells placed with the dead. The Bayesian analysis places this transition in about 420 BC (Higham & Higham 2009). These exotic ornaments reflect the fact that from at least the fourth century BC, and possibly earlier, coastal tracts of Southeast Asia were directly involved in a maritime exchange network that linked China with India and westward to the Mediterranean world. Indian craftspeople are thought to have settled in coastal port towns like Khao Sam Kaeo, and begun manufacturing hard stone and glass beads to meet the requirements of local elites (Bellina-Pryce & Silapanth 2006). Some of their output was exchanged into inland sites such as those in the Upper Mun Valley. It was, therefore, during the Iron Age that Southeast Asia participated in a new and extensive interaction sphere. The demand for new and exotic goods is seen in the mortuary record and was, presumably, accompanied by increased local production by participating communities.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

The earliest iron artefacts included large socketed spears; bimetallic spears with a bronze socket and iron blade; tool kits of knives, awls and hoes; and bangles. There was a rise in the quantity of bronze ornaments found with IA1, and for the first time some of the bangles and anklets were cast by the lost-wax method. Feasting took place, to judge from the many fish remains found in mortuary vessels, and the bones of cattle, pig and water buffalo. Occupation contexts included butchering floors containing stone anvils, iron knives and fractured animal bones. There are two groups of graves at IA1 Ban Non Wat, one with the dead orientated with the head to the north, the other to the south. Neither group, nor any individuals, stood out on the basis of prestigious mortuary offerings (Higham 2012). Only six LAI burials were found at Noen U-Loke. Again, we find evidence for heavy iron spears, iron ornaments and hoes, but there was no exotic stone or glass jewellery. However, bronze and iron torcs, tiger- and pig-canine pendants and shell ear ornaments distinguish a group which included one leper (Tayles & Buckley 2004; Higham et al. 2007).

Iron Age 2 (100 BC-AD 200) is represented by two clusters of inhumation burials at Noen U-Loke. There are several developments. The dead were now interred in graves filled with rice. The first glass, carnelian and agate jewellery was found at this site, and young pigs were interred with the dead. No iron objects were found, and pottery vessels were very rare. It was with IA3 at Noen U-Loke (AD 200-400) that we encounter what Yoffee (2005, 2012) would probably call 'emergent properties'. There were four distinct clusters of burials, each containing the graves of men, women, infants and children interred in rice-filled graves. Three clusters contained one or two particularly wealthy individuals, and some infants were also richly endowed with mortuary offerings. After removing the eggshell-thin ceramic vessels from the skeleton of one man, we found that he wore an unparalleled quantity of bronze ornaments: 3 belts, 150 bangles, 67 finger rings and 4 toe rings. Silver coils covered in gold were worn in his ear lobes, and an iron knife lay by his left hand. Agate and glass jewellery was also worn (Higham et al. 2007: 214-15). A woman in another cluster wore a necklace of agate and gold beads, a silver earring and toe ring, 64 bronze finger rings, 9 toe rings and 38 bangles (Higham et al. 2007: 177). Cluster C included a man with 4 bronze belts, 124 finger rings, 33 toe rings and 20 bangles (Higham et al. 2007: 198). Even an infant less than a year old wore 12 bronze bangles, 5 rings and 7 anklets. However, the fourth cluster was markedly poorer, and the dead were accompanied by a disproportionate number of spindle whorls.

One man in cluster C was accompanied by a socketed iron artefact which we have published as being a spade (Connelly 2007: 435). It has wings and, viewed from the side, it is asymmetric. An identical specimen was found in a kiln dating to IA4 at the site of Non Ban Jak in November 2012. One of the wings on this specimen was broken, and it might have been placed in the kiln to heat it before forging a repair. The asymmetric form, including the socket, make both unlikely to have been used as spades, and we now interpret these rare artefacts as ploughshares (Figure 3). This application of iron to agricultural implements continued into IA4 (AD 400-600), where at Noen U-Loke, the dead, while markedly poorer than during the preceding phase, were now accompanied by sickles. One young man belonging to this final phase of the Iron Age had been killed when an arrowhead severed his spine.

Our knowledge of this later Iron Age is not solely dependent on the cemeteries. Excavations across the moats and banks that ring each site have dated their construction to IA3-4. These are, by any measure, substantial engineering works. The five concentric moats at Noen U-Loke extend over a distance of 200m. Just one of the two moats at Non Ban Jak is 50m wide, and the banks, labelled ramparts by Williams-Hunt (1950), are still impressively high. Each site was located by a river or stream that was diverted into the moats. An examination of aerial photos has not only traced the course of the former river patterns that linked the moated sites, but also linear features congruent with and emerging from the moats and banks. While examination of these on the ground is for the future, they look like water capture and distribution facilities. This topic has been exhaustively examined in Iron Age settlements north-west of Angkor where Hawken (2011) has mapped extensive prehistoric field boundaries overlain by the regular rice-field banks of the later Chenla and Angkorian periods.

[FIGURE 3 OMITTED]

The area enclosed by the moats in the study area varies from 4.6 to 29.5ha, but there are much larger sites in the Mun Valley catchment. Non Dua in Roi Et Province covers 109ha, with a population in the thousands. This site commands an extensive deposit of salt which was exploited during the Iron Age and the Upper Mun sites are also associated with the small mounds that accumulated as a result of the local technique of extracting salt.

With excavations away from the cemetery areas of these sites, we have a growing impression of their constituent parts. At Noen U-Loke, the eastern edge of the settlement had been used for forging iron. The recovery of moulds for casting bronzes makes it clear that there were local ateliers for the specialised production of ornaments. Spindle whorls and the impressions of fabric on iron and bronze artefacts indicate a weaving industry. Salt was manufactured (Rivett & Higham 2007). At Ban Non Wat, Nigel Chang (pers. comm.) has identified the presence of a pound to confine cattle and water buffalo, clearly seen in the impressions of numerous hoof prints. Superimposed clay plastered floors and wall foundations were encountered in the small excavated area at Non Muang Kao (O'Reilly 1997). A much larger excavation at Non Ban Jak revealed a sequence of IA4 houses. The lowest had been destroyed in a conflagration. The clay floors were partially baked by the heat, such that hearths survived with pottery vessels still in place, along with dense concentrations of carbonised rice grains (Figure 4). Being better preserved, later structures revealed solidly constructed, broad clay walls and floors, the former incorporating postholes for wall timbers infilled with wattle and daub. One room had ritual connotations, with lidded vessels placed in each corner, and three burials within (Figure 5). Indeed, there appears to have been a custom of interring an infant under the floor of a newly constructed house. The Iron Age was also a period of increasing aridity in an area already prone to drought during the vagaries of the monsoon (King et al. 2013). The foundation of new and large settlements at Non Muang Kao and Non Ban Jak during the later Iron Age suggests that the population was growing at this juncture.

[FIGURE 4 OMITTED]

The last 30 years of research in the Upper Mun Valley have transformed our understanding of these Iron Age sites. The tightly dated Iron Age cultural sequence reveals a concordance of changes with profound social implications. These begin with the mortuary record. New and exotic valuables were obtained through exchange, and were interred with elite individuals. Food was incorporated into burial rites: rice, fish, cattle, pig and water buffalo, it is argued, reflect sumptuary feasting. As Hayden (2009) has emphasised, feasting provides the opportunity to exhibit status and gain alliances. Salt was extracted on an industrial scale and presumably made available for exchange transactions. Labour was galvanised to construct massive water storage and control measures. Specialist smiths made ploughshares, hoes and sickles as well as spears and arrowheads. The relative quantities of gastropod shellfish species at Noen U-Loke reveal a sharp change in the later Iron Age. Pila ampullacea, which is today a coloniser of fixed rice fields, replaced Filopaludina, a species adapted to streams with running water (McCaw 2007: 520). It was also at this juncture that Wohlfarth et al. (2012) have identified increasing aridity due to a weakening of the monsoon.

[FIGURE 5 OMITTED]

In Southeast Asia, rice is the engine that drives the state. Its cultivation is greatly expanded through the employment of animal traction to plough within fixed fields. This parcelling of land opens the door to the preferential ownership of this most valuable of assets in sustaining the community. The value of land holdings could also be increased if they lay in the path of irrigation canals fed by the moats/reservoirs around the settlements. As Rousseau wrote, "[t]he true founder of civil society was the first man who, having enclosed a piece of land, thought of saying 'this is mine', and came across people simple enough to believe him" (Cole 1913: 207). The social as well as economic significance of ploughing in place of hoe agriculture has been explored in depth by Goody (1976). With reference to the Near East and Europe, he noted that "a man's potential area of cultivation during one season could be increased from about 6 acres to a theoretical possibility of some 60 acres" (Goody 1976: 25). This, he argues, had a profound impact on interpersonal relations, leading to the distinction between lord and serf, landlord and tenant. The essence of the rice agriculture in the Mun Valley is that rainfall is retained where it falls on bunded fields, and allowed to percolate slowly through the fields so that blue-green algae can sustain the growing plants. The traditional use of a single buffalo to haul a plough in the Mun Valley has brought under cultivation extensive tracts of land that vary in value relative to the ease with which water can be introduced into the fields during periods of the wet season when rainfall falters. Animal traction is used first to turn soil softened by early rains, and then to break up the clods with a harrow to produce a creamy soil into which rice can be transplanted from seed beds.

Whatever other purposes might have been served by the moats, the presence of deep water up to 200m wide would have aided defence at a time when the concentration of sites and the proliferation of iron weaponry indicate the rise of conflict. Again, if there was an emerging trend to private ownership, then those with preferred access to the water in the moats/reservoirs would have been able to corner the supply of fish. It is concluded that these emergent properties reflect a rise in the status and power of social elites during the later Iron Age. Most fortunately, in north-east Thailand and Cambodia, it is possible to follow the sequel through the settlements and associated texts dated from the sixth to the eighth centuries.

Protohistory: the micro city states of Chenla

Many Iron Age settlements akin to those in the Upper Mun Valley are also found south of the Dang Raek range. Prei Khmeng, Baksei Chamkrong and Lovea, which is ringed by a double moat, lie within the extent of treater Angkor. Phum Snay, 60km to the west, has yielded burials with ceramic vessels, iron weaponry and exotic ornaments that would have been familiar to Iron Age communities in the Mun Valley (O'Reilly et al. 2006). The point is that the cultural changes identified in the Thai moated sites were not isolated. The protohistoric sequel, the Chenla period 550-800 AD, is illuminated by three sources of information: archaeological, epigraphic and documentary.

The archaeological sites of protohistoric Chenla present a complete contrast to the Iron Age moated settlements, most of which were by now abandoned. Parmentier (1927) has published a detailed description of sites, many of which have since suffered looting or marked deterioration. They are dominated by brick temples, in which a shrine chamber was entered through a portal embellished with a stone lintel decorated with religious Hindu themes. These have a wide distribution that extends north of the Dang Raek escarpment into the Mun Valley (Figure 6). The largest and best-documented centre was known as Isanapura, the city of Isanavarman. It is dominated by three walled precincts containing brick temples and sunken pools. The inscriptions record a dynasty of kings, including Isanavarman himself, who ruled there during the early seventh century. The site also includes a large rectangular reservoir, and rice field boundaries lie beyond the limits of this centre (Shimoda 2010). A description of the court of this period has survived in a compilation by the Chinese historian Ma Duanlin (1883), who mentioned a palace, guards, a large populace and regular royal audiences.

The archaeological hardware of these vital centuries is provided by the surviving temples, reservoirs and rice fields, but the social software has to be teased out of the surviving inscriptions. Carved onto stone stelae, these were inscribed in Sanskrit and old Khmer languages. Nearly all relate to the foundation and administration of a temple. They regularly refer to a ruler or the title and name of a local grandee associated with the temple foundation and its maintenance. The Khmer text includes information on rice fields, their boundaries, donations of surplus products to the temple, and the number and duties of individuals assigned to its support.

[FIGURE 6 OMITTED]

The inscriptions reflect a social organisation with a vrah kamraten an at its apex. This Khmer title may be translated as ruler, or king, but it was also applied to gods, suggesting that the king was at least semi divine. Personal names of rulers ended with the Sanskrit title -varman. Hence the name Isanavarman means protege of Siva or Sivas shield. There was no single state or ruler at this time, but rather a series of competing micro city states. Some inscriptions record a local hegemon appointed over a dependent centre by a king. Others name a local ruler but no such dependency.

Vickery (1998) has extracted from the available texts the divisions of the population below these rulers. Pon was a title accorded a person of high status, who exercised authority over temples and their sustaining populace. It is important to appreciate that the temple was not just an institution for the worship of an indigenous god or a Hindu deity, but it fulfilled the role of a community ritual and economic centre, to which donations of land, workers and domestic stock including cattle and buffalo were directed. The personnel included dancers, singers and officiants, as well as weavers, spinners, leaf sewers, potters and field workers. The surpluses of cloth, precious metals, ceramics and foodstuffs including rice were available for the pon to deploy to maintain the non-productive section of the community, exchange with other elite leaders, and accumulate for such social purposes as feasting. These texts describe the pon as being in charge of water reservoirs, which are often cited when designating rice field boundaries. Thus a web of social and economic characteristics identified in the Chenla texts resonate with the late prehistoric Iron Age: elite individuals, weavers, potters and smiths, as well as bounded rice fields and water control.

Conflict has also been identified in the Late Iron Age, and this too is documented in a text that describes how King Jayavarman I ('victorious shield') went to war in the autumn, when his enemy s moats were dry. There is a series of inscriptions by a man named Citrasena, later King Mahendravarman (shield of the great Indra), recording his campaign up the course of the Mun River, one which might well be linked with the abandonment of Late Iron Age sites in that region and absorption of the population into new centres. Although the concentration of inscriptions and temple sites lies south of the Dang Raek escarpment, the interest shown by Citrasena in the rich rice lands of the Mun floodplain is given archaeological credibility by the presence of brick temples, and a number of poorly dated but relevant texts. At the later Angkorian centre of Phimai, Iron Age occupation has been found under the main temple, but the construction of the latter entailed the destruction of an earlier brick shrine. Phanom Wan, a second Angkorian temple complex in the Mun Valley, overlies an Iron Age cemetery and the remains of a brick structure (Talbot & Janthed 2001). Jacques (1989) has reviewed inscriptions from Phimai and Wat Ban Song Puay which name a king, a dynastic sequence, and two 'kingdoms', Sambuka and Sri Canasa.

These surviving texts are probably the tip of the iceberg, for many other instances of conflict will not have been recorded. Thus, these three protohistoric centuries witnessed the extremely rapid rise of ruling elites claiming divine titles, the establishment of hereditary inequality on an intense scale, and warfare designed to take territory and place trusted followers into new administrative centres. Interpretation of this period as one characterised by warfare and the rise and fall of competing microstates is supported by the records of Tang Dynasty tribute missions from Southeast Asia (Smith 1979). During the first half of the seventh century, 20 microstates sent 41 missions. The next 50 years saw 28 missions from 14 microstates, and in the first half of the eighth century, the number of named polities fell to 5, and the number of missions rose to 29. One inference is that the number of independent microstates fell through a process of conquest and consolidation. Some of these have been identified as being located in north-west Cambodia by Wolters (1974).

If we incorporate agency into the transformation from competing city states to unification matching, for example, that of Sargon of Agade, then King Jayavarman II is the prime candidate. Very little is known of him, the most informative inscription from the site of Sdok Kak Thom being set up on 8 February 1053. The consensus is that Jayavarman was a charismatic leader whose initial royal centre was located at or near Banteay Prei Nokor on the left bank of the Mekong River. With his followers, he moved upriver before heading for the vicinity of Angkor. He was engaged in several battles, and settled his loyal followers on conquered land. In a mystical religious ceremony traditionally dated to AD 802, he was consecrated the cakravartin, or supreme leader on earth, at a temple complex on the Kulen upland north of Angkor. It is here that the results of the recent lidar surveys have made a critical contribution to archaeological research on this seminal development. By penetrating the jungle on the Kulen, lidar surveys have identified the layout of a city that is quite possibly Mahendraparvata, the first capital of Jayavarman (Evans et al. 2013). Temples were henceforth constructed on an ever-increasing scale, and were dedicated to the king often syncretised with the god Siva in the form of an erect stone phallus known as a linga. Some rulers were aligned with Visnu, while Jayavarman VII portrayed himself as a bodhisattva. Shrines were erected to honour deified ancestors. Although the succession to the highest office was at least thrice accompanied by changes in dynasty and endemic warfare, the state of Angkor endured for a further 750 years.

Conclusions

Evaluation of the rapid transition from the Late Iron Age into the kingdom of Angkor is sharpened by incorporating comparative instances of state formation. This continues to be a topic with no theoretical consensus. Thus Flannery and Marcus (2012) have focused on the foundation and both the common and individual characteristics of first-generation states in Mesoamerica, Mesopotamia and Egypt, liberally employing comparative information from our cognate discipline, social anthropology. Their volume stresses the value of identifying studies that "could be most readily used to interpret the archaeological evidence ... that either captured an important moment of social change or made explicit the logic of inequality" (Flannery & Marcus 2012: xii). Their choice of early states for detailed examination was determined by those providing 'evidence of residences, public buildings, ritual features, or burials that show some aspects of inequality" (Flannery & Marcus 2012: xiii). None of their examples lies east of Mesopotamia.

These and related sequences have encouraged Flannery (1999:15) to list instructions for agents bent on founding a state: be a charismatic and ambitious male, secure loyal followers, adapt your society's ideology in your favour, innovate to provide an efficient subsistence economy, identify and exploit new military methods to defeat and subdue rivals, and place trusted kin or supporters to rule new territories. Documenting this list archaeologically must be dovetailed into identifying emergent properties and tracing the rise of powerful elites. With the later Iron Age, these have been identified through the banks that ringed the sites to retain and control the flow of water. Smiths fashioned heavy iron ploughshares and sickles. At Lovea in Cambodia, rice field boundaries radiated out from the moats. The division and improvement of land and increased production occurred as elites at Noen U-Loke were being interred in graves filled with rice, along with outstanding sets of exotic ornaments. Smiths also forged iron arrowheads and heavy spears. Some settlements grew to be much larger than others. Valued cattle and water buffalo were protected in corrals within the moats, and substantial houses were constructed in the residential quarters of the moated towns. It is suggested that this was a period of formative social change involving the emergence of powerful leaders rooted in hereditary inequality.

Acknowledgements

I wish to thank my co-directors Amphan Kijngam and Rachanie Thosarat for their invaluable contribution to the Origins of Angkor research project. Funding for the excavation of Noen U-Loke, Ban Non Wat and Non Muang Kao was provided by the Marsden Fund and Earthwatch and its Research Corps. The current fieldwork at Non Ban Jak is funded by the Australian Research Council (DPI10101997) through a grant to Dr D. O'Reilly and Dr L. Shewan. Documenting the age and sex of the human burials from the Upper Mun Valley excavations has been undertaken by Associate Professor N. Tayles, Dr K. Domett and Dr S. Halcrow. I am most grateful for the constructive comments from two referees.

References

BELLINA-PRYCE, B. & P. SILAPANTH. 2006. Weaving cultural identities on trans-Asiatic networks: Upper Thai-Malay Peninsula--an early socio-political landscape. Bulletin de l'Ecole Francaise d'Extreme-Orient 93: 257-94.

COLE, G.D.H. (ed.). 1913. The social contract and. discourses by Jean Jacques Rousseau. London: J.M. Dent.

CONNELLY, R. 2007. The iron and bimetallic artefacts, in C.F.W. Higham, A. Kijngam & S. Talbot (ed.) The origins of the civilization of Angkor. Volume 2. The excavation of Noen U-Loke and Non Muang Kao: 431-46. Bangkok: Fine Arts Department.

EVANS, D.H., R.J. FLETCHER, C. POTTIER, J.-B. CHEVANCE, D. SOUTIF, BOUN SUYTAN, I. SOKRITHY, E. DARITH, T. TIN, K. SAMNANG, C. CROMARTY, S. DE GREEF, K. HANUS, P. BATY, R. KUSZINGER, I. SHIMODA & G. BOORNAZIAN. 2013. Uncovering archaeological landscapes at Angkor using lidar. Proceedings of the National Academy of Sciences of the USA 110:12595-600.

FLANNERY, K. 1999. Process and agency in early state formation. Cambridge Archaeological Journal 9(1): 3-21. http://dx.doi.org/10.1017/ S0959774300015183

FLANNERY, K. & J. MARCUS. 2012. The creation of inequality. How our prehistoric ancestors set the stage for monarchy, slavery and empire. Cambridge (MA): Harvard University Press, http://dx.aoi.org/ 10.4159/harvard.9780674064973

GOODY, J. 1976. Production and reproduction. Cambridge: Cambridge University Press.

Hawken, S. 2011. Metropolis of ricefields: a topographic classification of a dispersed urban complex. Unpublished PhD dissertation, University of Sydney.

HAYDEN, B. 2009. Funeral feasts: why are they so important. Cambridge Archaeology Journal 19(1): 29-52. http://dx.doi.org/10.1017/ S095977430900002X

HIGHAM, C.F.W. 2011. The Iron Age of the Mun Valley, Thailand. The Antiquaries Journal 1-44. http://dx.doi.org/10.1017/S0003581511000114

--2012. The Iron Age: synthesis, in C.F.W. Higham & A. Kijngam (ed.) The origins of the civilization of Angkor. Volume 6. The excavation of Ban Non Wat. Part 4: the Iron Age, summary and conclusions: 331-70. Bangkok: Fine Arts Department.

HIGHAM, C.F.W. & T.F.G. HIGHAM. 2009. A new chronological framework for prehistoric Southeast Asia, based on a Bayesian model from Ban Non Wat. Antiquity 83: 125-44.

HIGHAM, C.F.W. & A. KIJNGAM, 1982. Irregular earthworks in N.E. Thailand: new insight. Antiquity 66:102-10.

HIGHAM, C.F.W., A. KIJNGAM & S. TALBOT (ed.). 2007. The origins of the civilization of Angkor. Volume 2. The excavation of Noen U-Loke and Non Muang Kao. Bangkok: Fine Arts Department.

JACQUES, C. 1989. The Khmers in Thailand: what the inscriptions inform us. SPAFA Digest 16-24.

KING, C.L., R.A. BENTLEY, N. TAYLES, U. STRAND VIDARSDOTTIR, G. NOWELL & C.G.

MACPHERSON. 2013. Moving peoples, changing diets: isotopic differences highlight migration and subsistence changes in the Upper Mun River Valley, Thailand. Journal of Archaeological Science 4?: 1681-88. http://dx.doi.org/10.10l6/j.jas. 2012.11.013

MA DUANLIN 1883. Ethnograthie des peuples Strangers a la Chine II: Meridionaux. Translated by Le Marquis d'Hervey de Saint Denys. Geneve: H. Georg.

MCCAW, M. 2007. The faunal remains: results and conclusion, in C.F.W. Higham, A. Kijngam & S. Talbot (ed.) The origins of the civilization of Angkor. Volume 2. The excavation of Noen U-Loke and Non Muang Kao: 513-20. Bangkok: Fine Arts Department.

O'REILLY, D.J.W. 1997. The discovery of clay-lined floors at an Iron Age site in Thailand--preliminary observations from Non Muang Kao, Nakhon Ratchasima Province. Journal of the Siam Society 85: 133-50. http://dx.doi.org/10.1353/asi.2006.0024

O'REILLY, D., A. VON DEN DRIESCH & V. VOEUN. 2006. Archaeology and archaeozoology of Phum Snay: an Iron Age cemetery in northwest Cambodia. Asian Perspectives 45:188-211.

PARMENTIER, H. 1927. L'Art Khmer primitif (Publications de l'Ecole Francaise d'Extreme-Orient XXI-XXII). Paris & Bruxelles: G. Vanoest.

RIVETT, P. & C.F.W. HIGHAM. 2007. The archaeology of salt production in C.F.W. Higham, A. Kijngam & S. Talbot (ed.) The origins of the civilization of Angkor. Volume 2. The excavation of Noen U-Loke and Non Muam Kao: 589-93. Bangkok: Fine Arts Department.

SHIMODA, I. 2010. Study on the ancient Khmer city Isanapura. Unpublished PhD dissertation, Waseda University, Tokyo (in Japanese).

SMITH, R.B. 1979. South East Asia in the seventh and eighth centuries, in R.B. Smith & W. Watson (ed.) Early South East Asia: 443-56. Oxford: Oxford University Press.

TALBOT, S. & C. JANTHED. 2001. Northeast Thailand before Angkor: evidence from excavation at the Prasat Hin Phimai. Asian Perspectives 40: 179-94. http://dx.doi.org/10.1353/asi.2001.0027

TAYLES, N.G. & H.R. BUCKLEY. 2004. Leprosy and tuberculosis in Iron Age Southeast Asia? American Journal of Physical Anthropology 125: 239-56. http://dx.doi.org/10.1002/ajpa.10378

VICKERY, M. 1998. Society, economics and politics in pre-Angkor Cambodia. Tokyo: Centre for East Asian Cultural Studies for UNESCO.

WILLIAMS-HUNT, P.1950. Irregular earthworks in Eastern Siam: an air survey. Antiquity 24: 30-37.

WOHLFARTH, B., W. KLUBSEANG, S. INTHONGKAEW, S.C. FRITZ, M. BLAAUW, P.J. REIMER, A. CHABANGBON, L. LOWEMARK & S. CHAWCHAI. 2012. Holocene environmental changes in northeast Thailand as reconstructed from a tropical wetland. Global and Planetary Change 92-93: 148-61. http://dx.doi.org/10.1016/j.gloplacha. 2012.05.008

WOLTERS, O.W. 1974. North-western Cambodia in the seventh century. Bulletin of the School of Oriental and African Studies 37(2): 355-84. http://dx. doi.org/10.1017/S0041977X00136298

YOFFEE, N. 2005. Myths of the archaic state. Cambridge: Cambridge University Press, http://dx.doi.org/ 10.1017/CBO9780511489662

--2012. Deep pasts: interconnections and comparative history in the ancient world, in D. Dorthrop (ed.) A companion to world history. 156-70. Oxford: Wiley-Blackwell. http://dx.doi.org/10.1002/ 9781118305492

Received: 18 July 2013; Accepted: 18 September 2013; Revised: 19 September 2013

Charles Higham, Department of Anthropology and Archaeology, University of Otago, PO Box 56, Dunedin 9054, New Zealand (Email: charles.higham@otago.ac.nz)