The Contribution of Metallurgical Studies to South-East Asian Archaeology

Author(s): Anna Bennett

Source: World Archaeology, Vol. 20, No. 3, Archaeometallurgy (Feb., 1989), pp. 329-351

Published by: Taylor & Francis, Ltd.

Introduction

The development of metallurgy undoubtedly played a key role in the cultural history of South-East Asia. Metals greatly increased the efficiency of agricultural and hunting activities, and the uneven distribution of the raw materials used in metal production gave rise in this region, as elsewhere, to considerable variations in economic wealth, power and status.

Evidence has accumulated since the first discovery of the Dongson site some sixty years ago, in North Vietnam, of a bronze technology which developed out of the Phung Nguyen neolithic culture of the Red River Valley, and extended forward to the brilliant Dongson culture of the Iron Age from about 500 BC (Ha Van Tan 1980; Pham Minh Huyen 1979) (Figure 1). There is a similarity of artefact forms, of design motifs and of settlement location from the Neolithic to the Iron Age, and it is at least possible that the knowledge of bronze technology developed in Vietnam independently of influences beyond the range of south China and the Indochinese region.

In Thailand, which is separated from the lower reaches of the Red River Valley by the Annamite mountains, the earliest metal-age sites which have so far been discovered are on the Khorat Plateau in the north-east of the country (Figure 1). The conflicting chronologies attributed to these sites have given rise to much disagreement between archaeologists about the evolution of metallurgy in the region and about any influence upon this from external sources (Bayard 1979, 1986; Higham 1984; Loofs 1983). Excavations at Non Nok Tha (NNT) (Bayard 1971, 1984), Ban Chiang (BC) (Suthiragsa 1979; Gorman and Charoengwongsa 1979) and Ban Na Di (BND) (Higham and Kijngam 1984), (Figure 1) have produced much new information during the past twenty years, but there is still disagreement about chronology. Nevertheless, there can be little doubt that, by the first half of the second millennium BC, there existed a sophisticated bronze-working tradition, including alloying and casting, spanning at least 1,500 years before the appearance of iron. At Ban Chiang and Non Nok Tha (Figure 1) a neolithic culture preceded the appearance of bronze, but no evidence has so far come to light of an intervening copper age or of experimental stages in bronze working. Moreover, there is no evidence of copper smelting at these sites, and one of the many unsolved questions in Thai archaeology concerns the source of the metal used by the early bronze smiths. There are no copper ores on the Khorat Plateau and it is possible that the techniques of bronze working were imported, together with the raw materials, from some as yet unidentified site.

Figure 1

The author has been concerned with an area of central Thailand, south-west of the Khorat Plateau, which contains both copper and iron mineral deposits. In recent years, increasing evidence has been found of ancient mining and smelting activities (Natapintu 1984a, 1984b, 1985, 1987; Pigott, Natapintu and Theetiparivata n.d.) in the neighbourhood of Lopburi town (Figures 1 and 2), which have provided the basis for the present archaeometallurgical studies, including the location of ore sources, ancient mining methods, the proximity of the mines to production sites, the processes involved in extracting metals from their ores, the scale of production and the extent of trade in ore, metal, and finished artefacts.

Types of ore used and possible ore sources

Near the top of Khao Phu Kha mountain (KPK) (Figure 2), which is inside the Lopburi air force base, is a large man-made cave with galleries, the walls of which are stained green. X-ray diffraction analysis of ore samples, collected in 1984 from an exposed wall, indicated that malachite (CuC03.Cu(OH)2) and chrysocolla (CuSi03.2H20) are among the minerals present. The ore often contained 10 per cent copper; the remaining minerals, known as gangue, are composed of the oxides of silicon, iron and calcium in the ratio 4:2:3, with small amounts of aluminium oxide (Table 2).

Figure 2

Plate 1

In the same valley as this mine are three extensive mounds of slag, which may be where the ore from the mine was smelted. The smelting site of Non Pawai (NP) covers an area of approximately 50,000 square metres, Non Mak La (NML) and Nil Kam Haeng (NKH) are smaller, although they are of the same order of magnitude. These, together with two other large mounds of slag at Tha Khe (TK) and Khao Sam Yoi (KSY) (Figure 2), were visited in 1984 and 1985 in order to assemble samples of past metallurgical activities.

Only a few small fragments of gangue, discarded from copper ore, were found at the smelting sites of Non Pawai and Non Mak La, which suggests that ore dressing was carried out elsewhere. However, the presence of cube-shaped stones made from local volcanic rock, with small circular depressions at the centre of each face (Plate 1), suggests that some final ore crushing and/or grinding took place at the smelting sites. Ceramic smelting debris, derived from slag-lined reaction vessels and two-piece moulds, remain at some of these sites as process witnesses, but the clearest evidence of the ores and the processes used comes from the chemical and petrographic analysis of the slag.

Chemical analyses determined by X-ray fluorescence spectroscopy (Table 1) showed that the slags found at the sites were essentially composed of iron silicates, as is generally the case with ancient copper smelting slags.

A comparison of the chemical composition of the slag with that of the discarded gangue minerals found at the smelting sites, and with the siliceous material at the worked out copper mine of KPK, indicated that iron ore was added to flux the charge (Table 2). Pieces of haematite were found at some of the smelting sites and this ore would have been readily available from the iron mine at Khao Tap Kwai (KTK), which is still being worked today (Figure 2). Although most of the deposit was made up of iron oxide, with only 2-3 per cent Si02 and 1-2 per cent A1203, malachite (CuCo3.Cu(OH)2), azurite (2CuC03.Cu(OH)2) and chrysocolla (CuSi03.2H20) have

Table 1

Table 2

Plate 2

all been identified in the output from this mine. Copper minerals seem to have been deposited within fissures in the dense iron oxide, and analysis of samples of such deposits collected from Khao Tap Kwai indicated up to 28 per cent copper oxide (Table 2). It is therefore possible that some of the copper ores leing smelted derived from such an iron rich source and would have been at least partially self fluxing.

The presence of sulphur in the slag (Table 1) indicated that some sulphidic ores were being smelted. There was no evidence of roasting prior to smelting, and the low proportion of sulphur detected therefore suggested that these ores were partially weathered or that they were mixed with oxide ores, such as those available from Khao Phu Kha mine (KPK) or from a deposit at Khao Phra Bat Noi (KPBN) (Figure 2). This deposit consists of malachite and azurite in two subparallel brecciated zones in the granite, which is capped by a limestone roof pendant (Brown 1953). Analysis of the gangue associated with this ore is given in Table 2.

To complement the bulk chemical analysis, polished sections of slag samples were examined under a metallographic microscope and an electron probe microanalyser. Numerous tiny copper metal inclusions, known as prills, too small to have been retrieved by crushing the slag, were observed. In addition, inclusions of copper sulphides with varying proportions of iron, known as matte, permeated the slag, and their presence confirmed the use of some sulphur-containing ores (Plate 2). Purer Cu2S inclusions might have suggested the use of bronchantite (Cu4(S04)(OH)6), but the

Table 3

Figure 3

Plate 5

Table 4

more iron rich mattes in the slags analysed (Table 3) probably derived from the use of chalcopyrite (CuFeS2) or bornite (Cu5FeS4).

During smelting, as chalcopyrite ore, mixed with a copper oxide charge, passed through the oxidising zone of the reaction vessel, some sulphur would have been lost from the system, as sulphur dioxide gas, giving rise to a metal-enriched sulphide phase. As can be seen from the phase diagram for the Cu-Fe-S system (Figure 3), liquid mattes cover the entire range from the composition of the original ore to Cu2S and FeS. This has given rise to inclusions with varying ratios of Cu:Fe:S (Table 3). Examination of metallographic sections of the slag samples showed that the matte inclusions were typically spherical (Plate 2) and consisted of a light blue-grey phase and a darker blue phase. This variation within the inclusions can be seen clearly in sample HPNP20 (Plate 3) and was the result of a structural alteration which occurred during cooling of the solid solution (MacLean et al. 1972).

Plate 4

Plate 5,6

Arsenic was detected by electron microprobe analysis, in the matte inclusions and in the copper prills suspended in the slag (Table 4). Electron microprobe analyses indicated that the copper prills also contained between 1 per cent and 3.5 per cent iron. Highly reflective iron dendrites were sometimes observed in the copper prills (Plate 4) and, like the arsenic, this probably originated from the ore. Such metal would have been too brittle for the manufacture of large hammered objects, although it would have been suitable for the casting of small objects.

A miscast arrowhead (Plate 5) excavated 30cm below the ground at the smelting site of Non Mak La was found to have been manufactured in unalloyed copper which contained up to 0.8 per cent arsenic and approximately 4 per cent iron. The metal thickness of the arrowhead, which still has the flash marks from casting, was only 0.2mm. The electron microprobe analyses indicated that the composition of this arrowhead was not homogeneous (Table 5).

A similar composition was recorded for a small unfinished axe head, 5.5cm in length,

Table 5

Table 6

and for several small discs of copper, approximately 7cm in diameter, which were found at the ancient smelting site of Nil Kam Haeng (NKH) at the base of the copper bearing mountain Khao Phu Kha (KPK) (Figure 2). Examination of the microstructure of the axe head (Plate 7) and of some of the discs of copper indicated that the metal was permeated by small matte inclusions and was similar in composition to the copper prills retained in the slag samples.

Macrostructure of the slags

Although only a very small amount of copper metal was recovered from any of the smelting sites investigated, the large quantities of slag indicated that the area around Lopburi had been a large industrial complex and that smelting had been carried out on a large scale. Excavation of a surface area of 5 x 4 metres at Non Mak La produced over 3,500kg. of slag. Most of this slag was recovered from between 10 and 20cm below ground level and consisted of broken fragments, but there were also irregularly shaped plates of slag (Plate 8). The largest plates weighed over 1.2kg, had diameters in excess of 15cm and were 5.5cm thick. The upper surfaces, which were a deep red-brown colour, exhibited a flow structure of elongated ridges or strings of formerly viscous material. The other surface often had clay material adhering to it and pieces of soil were incorporated into the slag. Fingers of slag were also recovered. No stratification was observed within any of the slag cakes and each cake appeared to represent the pouring from a single smelt. Most of the cakes had been broken up, possibly in order to retrieve any trapped copper metal prills. Freshly fractured slags ranged from a glassy texture and a blue black colour to a greyer more crystalline texture. Green staining and bright green spots of corrosion were sometimes visible. A small number of reddish-brown porous lumps of slag which derived from iron smelting were also found.

At the smelting site of Non Pawai, the surface was littered with slag cakes and small fragments. The plano-convex slag cakes, which were more regular in size than the plates from Non Mak La, varied in size from 13 x 11cm to 19 x 15cm and were

Plate 7

between 3 and 7.5cm. thick. They had often been broken into quarters. Contraction marks, induced by rapid cooling, were sometimes observed on the surface of the slag and impressions of charcoal or wood were occasionally seen. The underside of the cakes had a rough texture, and clay material had often become incorporated in the slag. In a few instances, small pieces of porous ceramic, characteristic of the reaction vessels found at the site, had become incorporated into the underside of the slag cake. Most of the slag cakes seem to have originated from a single pouring, although a few of the cakes appear to have resulted from a second flow of slag solidifying on top of the original. Freshly fractured edges showed that the slag was dense and dark grey in colour with a crystalline structure. Copper coloured metal and lustrous grey prills were sometimes visible under a hand lens.

Microstructure of the slags

Petrographic examination (Plate 9) and electron microprobe analyses of the copper smelting slags from Non Mak La and Non Pawai (Table 7) showed that they were assemblages of olivine crystals (sample 19 - analysis no.l) in a glassy matrix (sample 19 - analysis no. 6) or, in the case of some of the slags from Non Mak La, no crystalline phase was present (sample 2). Just as in iron-rich slags from modern pyrometallurgical processes (McLellan 1930; Faber 1954) the most commonly occurring olivine in these ancient slags was fayalite (Fe2Si04). The fayalite had formed lath-like crystals and in some of the laths fractionation of elements had resulted in an outer zone of Ca rich olivine (CaFeSi04) and a nuclear zone which was comparatively rich in Mg. This phenomenon, due to conditions during crystallisation, can be explained by the miscibility gap which exists in the presence of magnesium (Hauptmann 1985:50). The

Plate 8

glass matrix of these fayalite-rich slags had a pyroxene composition (sample 19 -analysis 1), which was similar to that of the non-crystalline slags (sample 2). However, in these glassy slags, the magnesium and manganese were present in the glass instead of being concentrated in the olivine phase.

Small amounts of a white angular phase in the crystalline slags were identified as magnetite, a spinel which forms cubic crystals. Some A1203 substitution had occurred (sample 19 - analysis no. 13) and a hercynite core (FeO.Al203) could be detected. Delafossite (CuFe02) was occasionally observed, precipitating from the magnetite (Plate 9).

The mineral assemblages of the slags analysed indicated that the ancient metallurgists at Non Mak La and Non Pawai were able to control their smelting operation adequately. They were operating at sufficiently high temperatures to afford an efficient

Table 7

Figure 4

separation of slag and metal and were able to avoid large quantities of magnetite in the slag, which would have increased its viscosity and would have tended to hinder the settlement of copper metal prills and sulphide particles. Bulk chemical analysis of the slag showed only 0.9 per cent to 2.6 per cent copper (Table 1) and from the graph (Figure 4) it can be seen that the amount of copper retained in the slags was proportional to the viscosity.

Classification of slags

A theoretical classification of the slags analysed was achieved by normative calculation. This method is widely used in the field of geology and its application to slag investigation was initiated by Bachmann (1977, 1980) and developed by Keesmann, Bachmann and Hauptmann (1982) and by Kresten and Serning (1983). Having predicted the minerals formed from equilibrium crystallisation, the amount of each mineral phase in the slag was calculated from the cations in the bulk chemical

Figure 5,6

composition (Table 1). The theoretical mineral assemblages of the slags were then plotted into phase diagrams which are part of the quaternary system CaO-FeO-Al203-Si02. This quaternary system can be graphically represented as a regular tetrahedron with ternary systems as external faces or internal planes (Figure 5). In order to include the additional oxides present in the slags (Table 1), and in accordance with the solid solutions in the slag phases identified by the electron microprobe (Table 6), the quaternary system was extended by MnO, MgO, K20, Na20. Using the phase diagrams (Figures 6 and 7) it can be seen that the initial phases to crystallise, represented by the CaO-Si02-FeO system, are essentially olivines with

Figure 7

Figure 7

a fayalitic composition, although some of the slags from Non Mak La have a pyroxene composition. A more marked difference between the slags from Non Mak La and Non Pawai is visible in the final stages of solidification and while this phase in the slags from Non Mak La has a composition approximating to anorthite (Figure 6), the corresponding phase in the slags from Non Pawai approximates to gehlenite (Figure 7).

Reaction vessels and associated slag linings

Large numbers of fragments derived from thick walled, organically tempered reaction vessels/crucibles were recovered at the smelting sites of Non Mak La and Non Pawai. Examination of their freshly fractured edges indicated that while the ceramic fabric on the external surface of the vessel was red, as a result of oxidising conditions, the fabric within was grey from reducing conditions. A layer of slag 0.1 to 0.5cm thick, often full of small gas bubbles, lined the inner surface of the ceramic fragments. There were spots of green corrosion on the slag and, in some instances, a thin layer of green corrosion was visible between the slag and the ceramic fabric. The size of the vessels from which the fragments derived varied; the internal diameter of one of the larger vessels was estimated at 24cm at the widest point, tapering sharply towards a flat base, while the smallest diameter recorded was 12cm. The thickness of the bases varied between 4 and 5cm and the wall thickness near the rim was between 0.5cm and 1cm. The ceramic had often been eaten away during the reaction, sometimes reducing the original wall thickness by half. For example, one particular base fragment with an original thickness of 3.6cm, had been attacked by the slag, and was reduced in parts to

Plate 9

1.6cm. The slag sometimes covered the rims, and solidified drops of slag were seen on the exterior of the fragments. The slag over the rim may have resulted from bubbling over during smelting or from the pouring out of the slag at the end of the smelting process. The crucible would have to have been lifted and tilted so that the slag could be poured out, and a small amount would inevitably run back down the inside walls of the vessels, giving rise to the slag lining in which gas bubbles can be observed.

No evidence of the re-use of these crucibles was observed, either by the naked eye or by petrographic thin section. Under a binocular microscope at ten times magnification, rice and quartz temper was readily visible in the freshly fractured edges. Small laterite nodules and pieces of epidote (Theetiparivata pers.comm.) were also observed, and it seems that the metallurgists manufactured their reaction vessels from local clay containing laterites, and that they probably added broken up gangue from ore dressing waste.

A few ceramic fragments were found which differed markedly from those described above. These fragments, which were derived from artefacts with a larger diameter than the crucibles, were characterized by their very thick walls and their bright orange colour, indicative of oxidising conditions. None of these fragments contained slag, and their lack of exposure to a reducing atmosphere indicated that they had no direct contact with the smelting operation. The internal diameter of these fragments, (the most complete example: internal diameter 16cm external diameter 26cm) suggests that the large-diameter vessel would have fitted snugly around the reaction vessel and that it might have acted as an insulating collar or chimney.

Projection of the bulk chemical analyses of slag samples, removed from the reaction vessels, onto phase diagrams indicated a very similar chemical composition to that of the slag cakes - suggesting that the slag derived from the same process (Figures 6 and 7). However, petrographic and electron microprobe analysis revealed that the microstructure of these slags, unlike that of samples removed from the cakes of slag,

Plate 10

contained appreciable amounts of magnetite and were therefore produced in a more oxidising atmosphere (Plate 10). This may be due to air having been sucked into the hot reaction vessel as the molten slag was poured out.

Copper artefacts and moulds

Although no copper ingots were recovered from the smelting sites of Non Mak La and Non Pawai, discs of copper which were about 7cm in diameter and only a few millimetres thick (Plate 6), were recovered from the smelting sites of Tha Khe and Nil Kam Haeng in the same area. These often had a layer of slag adhering to their surfaces. The fact that this slag was similar in chemical and mineralogical composition to the smelting slag, suggests that metal was being poured directly from the reaction vessel into a mould, and that some slag had accidentally followed. Ceramic moulds, for casting a shape corresponding to these copper discs (Plate 6), were found in the collection belonging to the temple of Wat Tung Singto (WTST) (Figure 2). Although no ingots were found at Non Pawai, a few fragments of shallow moulds similar to those in the collection at Wat Tung Singto were found. However, the majority of the ingot moulds at Non Pawai were smaller than these and were for casting ingots of a different shape. Some had a simple bowl shape with a flat base, while others had a pedestal more than half the total height of the mould (Plate 11). This pedestal would have allowed the mould to be secured in a rack or in the ground, while the molten metal was being poured into it. The cups were not of a standard size. The largest were approximately 5cm in diameter, 5cm in height and 4cm deep, and the smallest were 4cm in diameter, 3.5cm in height and 2cm deep. A few of the cups had been distorted in shape, suggesting that the ceramic may not have been fired before handling. The cups often had a small chip broken off the rim, which may have facilitated the removal of the ingot. Green spots of corrosion were sometimes observed on the inside of the cup moulds. All the examples retrieved from the 1985 survey were analysed by X-ray

Plate 11

fluorescence spectroscopy, and copper was detected inside the cup in every case. However, no ingot corresponding to the shape of these moulds has yet been recovered from this site. A few pieces of slag which exactly correspond to the shape of the small cup moulds were recovered (Piggot 1986 pers. comm.) and this further supports the idea that copper was being poured directly from the reaction vessels into the ingot moulds.

Figure 8,9,10

Although most of the moulds were for casting metal ingots, there were also some fragments of two-piece terracotta moulds, for casting artefacts. Similar bivalve moulds have also been recovered from a number of other sites in the area (Ho 1982) and they seem to have been used to cast very thin objects (Figures 8,9,10). The two-piece mould fragments recovered as surface finds rarely had any indentation on the inner surface, and presumably there must have been depressions corresponding to the shapes of the objects to be cast in the other halves of the moulds. In some cases the shapes of the objects to be cast could be seen as darkened areas caused by the molten metal burning the top half of the mould (Figure 10). The fragments of two-piece moulds sometimes had incised markings on the outside, which may have served to match the two halves or may have indicated the type of object being cast.

Discussion

Archaeological evidence, which is only just beginning to emerge, indicates that there were many smelting sites in the area around the town of Lopburi. Evidence so far suggests that the ancient metallurgists were smelting locally available sulphur-containing ores. There is no evidence that the ore was roasted prior to smelting, and microscopic examination of prepared sections has indicated that copper artefacts, ingots and prills were permeated by small matte inclusions. Although a few very thin unalloyed copper artefacts, such as the arrowhead and the axehead, were being cast at the smelting sites, the bulk of the metal appears to have been used to produce small raw copper ingots, which may have been further refined at the artefact production sites with which they were traded.

The composition of the samples of copper slags recovered from the smelting sites of Non Mak La and Tha Khe have been found to be very similar. The presence of small amounts of iron slag (Figure 6), identified during the present research, and of a few iron artefacts found at the sites, suggest an Iron Age context. The copper slags from the smelting site of Non Pawai were found to be different in composition and, since no evidence of the use of iron has been found, this site is thought to be earlier. The copper smelting technology, which involved the use of crucibles from which the molten copper metal was poured directly into ingot moulds, appears to have been similar at all the sites investigated. These sites date from a time when the ancient metalworkers in central Thailand appear to have had no access to tin for the production of bronze and were casting small thin objects in unalloyed copper, in which small quantities of arsenic were sometimes present. The arsenic was probably derived from the smelting of arsenical sulphides and, whether or not the use of such ores was deliberate, the presence of arsenic at levels of up to 5.3 per cent is significant. Quantities of 1 per cent to 2 per cent increase the hardness of the metal, which can be cold worked to a harder material. In an area which has shown no evidence of the use of tin at this time, the production of arsenical bronzes would have been advantageous.

At Tha Khe, metalworking appears to have been abandoned by the later phase, which is characterised by ceramics dating from the sixth to the eleventh centuries AD (Bhumathon 1984). It is possible that, at about this time, copper smelting was transferred to the site of Khao Sam Yoi (KSY) (Figure 2), which has produced pottery from the Dvaravati and Sukhothai periods. The composition of the slag samples from KSY was found to be similar to that of samples from Tha Khe.

Bronson has pointed out that central Thailand seems backward 'when compared with the showy and expansive societies of the north-east some 2000 years before' (Bronson 1979:321). Although there is abundant evidence for neolithic cultures in the area, some with developed inter-regional trade networks (Ho 1984; Watson 1979) bronze, as far as is known at present, appeared only in an Iron Age context. The present study has, however, shown that in prehistory there was well established, large scale copper production operating efficiently in the area, and using a technology which has not been encountered elsewhere. The number of skeletons found amongst metalworking debris, apparently buried without ceremony, is remarkable. The explanation of this might be the use of slave labour.

Throughout the whole of Thailand there appears to have been a very marked cultural transformation in the mid-first millennium BC. As Higham has pointed out, 'Before 500 BC or thereabouts, settlements were of about the same size, and I suspect autonomous. After that date, there were primate centres associated with craft specialisation, and intensified production' (Higham 1984:252). At this time, when the neolithic settlements in central Thailand entered the metal-age and engaged in mass production of copper metal, there is the first evidence of the use of domestic water buffaloes in South-East Asia, and also the use of iron (Bennett 1982), an intensification in the production of bronze objects, and the development of long distance trade (Glover in press). This is the period when Non Nok Tha was abandoned, when Ban Nadi appears to have been reoccupied, after a short period of abandonment. These changes were accompanied by significant developments in bronze technology (Stech and Maddin 1967; Rajpitak 1983), for example, the introduction of high tin bronzes (Rajpitak and Seeley 1979) and the production of the finest Dongson drums by the people of the Red River Valley (Nguyen Duy Hinh 1984; Pham Minh Huyen et al. 1987).

The lack of an identifiable bronze age in central Thailand, and the sophistication of the copper production, which had developed by the end of the first millennium BC, suggests that metallurgical knowledge had been brought into the region. At the period of discontinuity noted by Higham, there had been a disruption of metalworking in the north-east and it is possible that the causes of this produced a migration of metalworkers in search of new sources of raw materials.

Acknowledgements

This work was carried out under a post-graduate scholarship from the British Academy. Additional financial support was provided by London University. I am most grateful to my supervisors Dr I. C. Glover and Dr N. Seeley for their help and encouragement throughout this study. The material was kindly loaned by the Fine Arts Department of Thailand. Bhuthorn Bhumathon (Lopburi National Museum), Surapol Natapintu (Central Thailand Archaeological Project), Colonel Monchai (Central Air Force Base,Lopburi), Dr V. Piggot (University of Pennsylvania), and Udom Theetiparivata (Department of Mineral Resources, Thailand), provided invaluable assistance during periods of fieldwork. Help with the operation of analytical equipment was given by Ian Young (University College London), Kurt Kunzmann and his staff (Degussa, Frankfurt), Hennie Niebhaus (Bochum University), and Andreas Ludvi (Bochum Mining Museum). I also thank Professor H. G. Bachmann (Degussa Precious Metals Refinery, Frankfurt), Dr A. Hauptmann (Bochum Mining Museum), Dr R. Clough, and Professor Keesmann (Mainz University) for many useful discussions. I am especially grateful to Professor Bachmann for reviewing the manuscript.

References

Bachmann, H. G. 1978. Schlacken: Indikatoren archaometallurgischer Prozesse. In Mineralische Rohstoffe als kulturistorische Informationsquelle (ed. H. W. Hennicke). Hagen: Verlag Verein Deutscher Emailfachleute, pp. 66-103.

Bachmann, H. G. 1980. Early copper smelting techniques in Sinai and in the Negev as deduced from slag investigations. In Scientific Studies in Early Mining and Extractive Metallurgy (ed. P. T. Craddock). British Museum Occasional Paper No. 20, pp. 103-4.

Bayard, D. T. 1971. Non Nok Tha. The 1968 Excavation: Procedure, Stratigraphy, and a Summary of the Evidence. University of Otago Studies in Prehistoric Anthropology 4.

Bayard, D. T. 1979. The chronology of prehistoric metallurgy in north-east Thailand: Silabhumi or Samrddhabhumi? In Early South East Asia (eds R. B. Smith and W. P. Watson). Oxford: Oxford University Press, pp. 15-32.

Bayard, D. T. 1984. Rank and wealth at Non Nok Tha: the mortuary evidence. In Southeast Asian Archaeology at the XV Pacific Science Congress (ed. D. Bayard). University of Otago Studies in Prehistoric Archaeology, pp. 87-122.

Bayard, D. T. 1986. Chronology, evolution and diffusion in the later southeast Asian cultural sequence: some comments on Higham's recent revision. Bulletin of the Indo-Pacific Prehistory Association, 7: 118-40.

Bennett, A. T. 1982. Iron artefacts from Thailand. BSc thesis, University of London (unpublished).

Bhumathon, B. 1984. Certain archaeological artifacts from the 1983 excavation at Ban Tha Kae. In The Archaeological Site of Ban Tha Kae (ed. B. Bhumathon). Bangkok: Fine Arts Department, (in Thai), pp. 13-19.

Bronson, R. B. 1979. The late prehistory and early history of central Thailand, with special reference to Chansen. In Early South East Asia (eds R. B. Smith and W. P. Watson). Oxford: Oxford University Press, pp. 315-26.

Brown, G. F., Biravas, S., Charaljavanaphet, J., Jalichandna, N., Johnston, W. D., Stresthaputra, V., and Taylor, G. C. 1953. Geological reconnaissance of the mineral deposits of Thailand: US Geological Survey Bulletin 1984; Geological Survey Memoir 1. Bangkok: Royal Department of Mines.

Faber, W. 1954. Mikroskopie der Metallhuttenschlacken. In Handbuch der Mikroskopie in der Technik. Band ii, Teil 2 (ed. H. Freund). Frankfurt am Main: Umschau Verlag, pp. 521-94.

Glover, I. C. In press. Early trade between India and Southeast Asia - a link in the development of a world trading system. Occasional paper. University of Hull, Centre for Southeast Asian Studies.

Ha Van Tan, 1980. Nouvelles recherches prehistoriques et protohistoriques au Vietnam. Bulletin de I'Ecole Frangaise d'Extreme Orient LXVIII: 113-54

Hauptmann, A. 1985. 5000 Jahre Kupfer in Oman. Der Anschnitt. Beih 4. Veoffentlichungen aus dem Deutschen Bergbau-Museum Bochum. Nr.33

Higham, C. F. W. 1984. The Ban Chiang culture in wider perspective. Mortimer Wheeler Archaeological lecture. Proceedings of the British Academy 69:229-61

Higham, C. F. W. and Amphan Kijngam (eds) 1984. Prehistoric investigations in Northeastern Thailand. BAR International Series 231 (i-iii). Oxford: British Archaeological Reports.

Ho, Chu Mei. 1982. The final report on archaeological survey in Lopburi province June 1982-July 1983. Institute of Archaeology, University of London (unpublished).

Ho, Chui Mei. 1984. The Pottery of Kok Charoen and its farther context, PhD thesis, University of London (unpublished).

Keesmann, I., Bachmann, H. G. and Hauptmann, A. 1982. Norm-Berechnungsverfahren fur eisenreiche Schlacken. Fortschr.Miner. 60, Beih.l : 110-11.

Kresten, P. and Serning, I. 1983. The calculation of normative constituents from the chemical analyses of ancient slags. Jerncontorets Forskning 25, Serie H:l-25.

Krivsky, A. and Schuhmann, R. 1957. Thermodynamics of the Cu-Fe-S System at Matte Smelting Temperatures. Journal of Metals, 209:981-8.

Loofs, W. 1983. The development and spread of metallurgy in Southeast Asia: a review of the present evidence. Journal of Southeast Asian Studies, 14(1): 1-11.

MacLean, W. H. Cabri, L. J. and Gill, J. E. 1972 Exsolution products in heated chalcopyrite. Canadian Journal of Earth Sciences, 9:1305-17.

McLellean, R. D. 1930. A petrographic study of lead and copper furnace slags. Transactions of the American Institution of Mining Metallurgy and Engineering, 91: 244-64.

Natapintu, S. 1984a. Ancient settlement at Ban Tha Khae in Lopburi. Muang Boran Journal, 10 (4):41-8

Natapintu, S. 1984b. Archaeological evidence from Ban Tha Kae and some notes on ancient settlements in lower Central Thailand. In The Archaeological Site of Ban Tha Kae (ed. B. Bhumathon). Bangkok: Fine Arts Department, (in Thai), pp. 1-12.

Natapintu, S. 1985. Ancient Bronze Metallurgy in Lop Buri. Art and Culture, 6(8):94-9 (in Thai).

Nguyen Duy Hinh. 1985. Bronze drums in Vietnam. Vietnam Social Sciences, 3:59-72.

Pham Minh Huyen. 1979. Various phases of the development of primitive metallurgy in Viet Nam. In Southeast Asian Archaeology at the XV Pacific Science Congress. The origins of agriculture, metallurgy, and the state in Mainland Southeast Asia, (ed. D. Bayard). University of Otago Studies in Prehistoric Anthropology 16, pp. 173-82.

Pham Minh Huyen, Nguyen Van Huyen and Trinh Sinh. 1987. Dong Son bronze drums. Nha Xuat Ban Khoa Hoc Xa Hoi. Ha Noi (in Vietnamese).

Piggot, V. Natapintu, S. and Theetiparivata, U. nd. An abstract presented in a lecture and discussion on Prehistoric Copper Bronze Production in Thailand Archaeometallurgy Project at Bangkok National Museum, on April 23, 1986.

Rajpitak, W. and Seeley, N. J. 1979. The bronze bowls from Ban Don Ta Phet, Thailand: an enigma of prehistoric metallurgy. World Archaeology, 11:26-31.

The contribution of metallurgical studies to South-East Asian archaeology 351

Rajpitak, W. 1983. The development of copper alloy metallurgy in Thailand in the pre-Buddhist period, with special reference to high tin bronzes. PhD thesis, University of London (unpublished).

Stech, W. H. and Maddin, R. 1967. The Techniques of the Early Thai Metalsmith, Expedition 18(4):38-47.

Suthiragsa, Nikom. 1979. The Ban Chiang Culture, In Early South East Asia (eds R. B. Smith and W. P. Watson). Oxford: Oxford University Press, pp. 42-52.

Watson, W. 1979. Kok Charoen and Early Metal Age of Central Thailand. In Early South East Asia (eds R. B. Smith and W. P. Watson). Oxford: Oxford University Press, pp. 53-62.

White, J. 1982. Ban Chiang: the discovery of a lost Bronze Age. Philadelphia: University of Pennsylvania Museum.

White, J. 1986. A revision of the chronology of Ban Chiang and its implications for the prehistory of Northeast Thailand. PhD thesis, University of Pennsylvania, Philadelphia (unpublished).

Abstract

Bennett, Anna

The contribution of metallurgical studies to South-East Asian archaeology

There is evidence of large scale copper production in the Lopburi area of Central Thailand. Prehistoric mining and smelting activities were identified during fieldwork in 1985 and 1987. Subsequent analysis of finds from these sites, including ore, slag and artefacts, has been used to indicate the level of sophistication of the metallurgy, employing a technology not previously encountered elsewhere. Archaeometallurgical studies have important implications for South-East Asia, where there has been much disagreement between archaeologists about the evolution of metallurgy in the region and about any influence upon this from external sources.