A new interpretative approach to the chemistry of copper-alloy objects: source, recycling and technology.

Bray, P.J. ; Pollard, A.M.

[ILLUSTRATION OMITTED]

Introduction

Archaeometallurgy has often been criticised for not considering the human and environmental reality behind the chemical analyses--for being far too concerned with documenting technology, at the expense of considering the social context of the manufacture and use of metal. This paper shows how a new procedure uses the chemical composition of bronze objects in north-western Europe to deduce their histories; not only where they originated, but how they were made by observing how often the bronze had been recycled and reworked.

The conceptual basis of the model

It has long been accepted that the composition of copper-alloy objects is, in part at least, dictated by the mineral assemblage of the ore from which it is smelted (e.g. Friedman et al. 1966). Thus, native copper, oxide ores, and sulfide ores will give rise to copper alloys with distinctive proportions of other elements, depending on the exact mineralogy of the ore and the smelting process(es) involved. The highly influential Studien zu den Anfangen der Metallurgie (SAM) project (Junghans et al. 1960, 1968, 1974) systematised this approach by producing a series of decision trees based on concentration levels of trace elements in order to group objects by chemical variation. Northover (1980) modified this by noting the presence of key elements when creating his A to H alphabetic system to denote metal types. More recently this work continues through the linking of chemistry, geology, isotopes, archaeological objects and mines (Ambert 1991; Ambert & Barge-Mahieu 1991; Krause & Pernicka 1996; Ixer & Budd 1998; Needham 2002; Ixer & Pattrick 2003; Niederschlag et al. 2003).

A clear consensus has emerged through this and similar work that low levels (rarely exceeding 5%) of arsenic (As), antimony (Sb), silver (Ag) and nickel (Ni) show characteristic variation in prehistoric copper objects, and that similarities in these patterns of variation can often be interpreted as indicating a common ore source. This composition will obviously then be influenced by the precise ore beneficiation and smelting processes applied. We argue that, additionally, the final element signature of a given object can be affected by a range of post-smelting processes that are central to how people actually used metal: melting and casting, smithing, recycling and alloying.

[FIGURE 1 OMITTED]

Re-melting copper alloys results in a loss of certain of the included elements through oxidation, particularly arsenic and antimony (McKerrell & Tylecote 1972; merkel 1982; Pernicka 1999; Earl & Adriaens 2000), a process illustrated by an Ellingham diagram (Figure 1). This plots, amongst other information, the affinity of an element for oxygen at different temperatures, and is the foundation of modern chemical metallurgy (Beeley 2001). Alongside the underlying driving force of oxygen affinity, the precise rate of oxidation is also affected by the degree of agitation of the molten metal, mutual solubilities, activities, volatilisation and partial pressures (Merkel 1982: 30; Beeley 2001: 31). These properties are exploited by modern industry in order to oxygen-refine (i.e. remove unwanted levels of other metals) the copper produced by smelting (Copper Development Association 2011). Consequently there is an extensive literature on oxidation effects in molten metals that is consistent with the archaeological experience: a bronze object will be depleted in certain of its metals depending on how many times it has been reheated, and to what temperature (e.g. Hampton et al. 1965; Charles 1980; Pickles 1998; Beeley 2001: 497; Tanahashi et al. 2005; Lee et al. 2009).

Summarising the causes of chemical variation, Pernicka (1999) concludes that a very limited range of trace elements are directly related to the provenance of the ore (most significantly gold [Au], silver [Ag], bismuth [Bi], iridium [Ir] and nickel [Ni]), whilst a larger number of the others (including tin [Sn], providing it is present at less than c. 1% [i.e. has not been deliberately added], zinc [Zn] and lead [Pb] if less than 5%, plus arsenic [As], cadmium [Cd], cobalt [Co], indium [In] and mercury [Hg]) are indicators of either technology or provenance. Merkel (1982: 287) refers to the "confusion" that differential oxidative losses could cause. However, if we shift our theoretical focus somewhat, these potential chemical changes under melting could present us with an extremely powerful interpretative tool.

Collating existing datasets

Determining the significance of variations in composition requires a large database, such as has now been accumulated for the Early Bronze Age (EBA) in the British Isles (Gowland 1906; Coghlan & Cook 1953; Case 1954; Coghlan & Case 1957; Britton 1963; Junghans et al. 1968; Coghlan 1970, 1979; Rutland & Coghlan 1972; Needham 1983; Rohl & Needham 1998; Northover pers. comm.). The tally of chemical analyses for the British and Irish EBA is 2129 objects (Bray 2009).

The use of such a composite dataset appears at first sight dangerous because of the time span over which different research projects have been providing data, each employing a range of analytical techniques. However, a series of comparative studies have found no serious contradictions in the published chemical results of the major European research projects for the elements of interest (Muller & Pernicka 2009; see also Bray 2009: ch. 2). There is also some safety in numbers. By using averages over very large datasets constructed from several different research groups and methods, we are less susceptible to errors inherent in any one analysis.

It is not now possible to reliably estimate the accuracy and precision of many of the individual analyses in the database. We can, however, calculate the standard error of the mean for each of the defined groups, which allows us to test the significance of the proposed differences. The patterns described below are all shown to be statistically significant using the Mann-Whitney U-test, and therefore to indicate real trends within the data. Of course, this does not negate the need for the ideas presented here to be tested using data produced to modern analytical standards. Ongoing work in many laboratories continues to produce new data and test the precision and accuracy of historical datasets, but in order to get the size of dataset we need, we are currently still reliant on data produced up to 100 years ago.

Identifying and understanding fine structure in chemical composition There are a number of factors which can give rise to systematic variations in the concentration levels of minor and trace elements in ancient copper alloys, including:

(i) The mineralogy of the copper ore source (native, oxidised, sulfidic).

(ii) Partitioning of trace elements during the smelt between metal and slag phases, plus volatilisation during smelting, and also contamination from various sources such as the crucible.

(iii) The deliberate addition of other metals, such as tin or lead, to the finished metal, or co-smelting of different ores to produce alloys directly. These practices are distinct from the deliberate targeting of poly-metallic ore sources that will naturally create alloys in the smelt.

(iv) Changes in chemistry through melting, re-melting and working of the metal.

(v) Potential mixing of signals through recycling of objects.

(vi) Post-depositional weathering and corrosion.

We are concentrating here on those factors which arise post-smelting, factors (iv) and (v).

In order to apply the principle of differential oxidative losses encoding the history of copper-alloy objects, we need to know the broad starting composition of the metal, governed by the ore source, and a chronological framework, ideally linked to absolute dates, for the metal artefacts that were produced. Figure 2 and Table 1 summarise the framework developed for British (EBA) material, mainly through the work of Burgess (1980), Needham (1996) and Needham et al. (1997, 2010) in linking object typologies, hoards, burial assemblages, ceramic sequences and absolute dates.

It is widely accepted that, in the earliest stages of the Irish and British EBA, a single source produced much of the copper in circulation. The Ross Island Bronze Age mine, Co. Kerry, Ireland, has been linked with a distinctive EBA copper-arsenic-antimony-silver artefact composition (approximately 96.2% Cu, 2% As, 1% Sb, 0.8% Ag). This As-Sb-Ag combination is so ubiquitous that independent chemical analyses have consistently identified it, successively naming it as Class 1 metal (Coghlan & Case 1957), B2 in SAM 1 (Junghans et al. 1960), E11 in SAM 2 (Junghans et al. 1968), and 'A' metal (Northover 1980). Excavation of the mine site has produced evidence of a Beaker period working site, structures associated with metalworking and evidence that two ore bodies on the site, the Blue Hole and Western Mine, were being exploited in antiquity. Archaeological work indicates that mining in the Blue Hole seems to have ceased around 2100 BC, while production in the Western Mine may have continued for a further 200 years. Work appears to have been abandoned due to flooding, ore having been removed down to levels allowed by the rising water (O'Brien 2004: 572). The tennantite-tetrahedrite ([Cu.sub.12][As.sub.4][S.sub.13]- [Cu.sub.12][Sb.sub.4][S.sub.13)] series (with smaller amounts of chalcopyrite, CuFe[S.sub.2]) makes Ross Island the only known site in Britain and Ireland capable of producing ores that could make the arsenic- silver-antimony 'A' metal (Ixer & Budd 1998; Ixer 1999; Ixer & Pattrick 2003; O'Brien 2004). Smelting experiments have been carried out at the site using ores from the Western Mine. O'Brien and Hannam roasted tennantite-rich copper ore and then conducted a carbon reduction smelt in a simple pit furnace using hand bellows. This produced prills of copper, embedded in lumps of slaggy matter, which were chemically indistinguishable from EBA examples of 'A' metal (O'Brien 2004: 532).

[FIGURE 2 OMITTED]

From considering the distribution of elements within the whole archaeological assemblage of Ross Island-type copper (777 artefacts, listed in detail in Bray 2009). Needham (2002) defines the limits of the composition as:

As >0.24%

Sb >0.1%

Ag >0.1%

Ni <0.09%

This gives the boundaries for the identification of Ross Island metal used in this paper.

Given the dating of the mine, in Metalwork Assemblage (MA) typo-chronological terms (Needham 1996) we would expect to see Ross Island metal being very prevalent in MAs 1, 2 and 3, with smaller levels present in MA 4 (see Table 1 for corresponding date ranges). Table 1 shows precisely this pattern, with the Ross Island metal signature dominating the beginning of EBA metallurgy, and then sharply dropping away as the mine ceases production at around MA4. Using the most recent definition of MAs (Needham 1996; Needham et al. 1997), Ross Island copper accounts for over a third of all analysed copper and bronze objects from pre-Middle Bronze Age Britain and Ireland. This level leaps to over 70% for the earliest periods (MA 1-2). These figures were established by gathering together for the first time the complete scientific dataset for EBA British and Irish copper alloys (Bray 2009; these data will shortly be available via the RLAHA website).

The tennantite-tetrahedrite ores at Ross Island can therefore be seen as the likely point source for a huge and readily identifiable proportion of EBA British and Irish copper. As O'Brien (2004: 539) notes, the composition of objects made from Ross Island copper is "remarkably consistent" over time. However, buried within an apparently unchanging triad of As-Sb-Ag levels is a wealth of information linked to a human scale of technology--patterning that is created by differing levels of oxidative loss.

Technological fine structure

Figure 3 shows the average arsenic level in a series of artefact types. The average arsenic level for copper halberds (1.93%, 121 artefacts) is higher than that for the contemporary axes (1.19%, 260 artefacts) produced from the same Ross Island metal. Using the non-parametric Mann-Whitney U-test this difference is statistically significant at the 99% confidence level. This can be explained by the use of a different procedure in manufacture: halberds were cast in closed two-piece moulds, while axes at this time were made in open moulds (Schmidt & Burgess 1981). The metal in the open axe castings was exposed to hot oxidising conditions for longer, which led to a slightly greater average loss of arsenic. Alongside casting technique, the data also suggest that halberds were very rarely melted and re-cast, a possibility that is highly significant for understanding their social role. This shows that changes in the average level of arsenic in a large number of artefacts from a common ore source can be related to an underlying technological process, which in turn is indicative of particular patterns of social behaviour. This effect has been suggested before (Northover 1999) to explain the composition of particular artefacts, but has not been demonstrated empirically across large sections of the EBA dataset.

[FIGURE 3 OMITTED]

There is a second technological echo present in Figure 3, which is caused by alloying copper with tin. The slight fall in arsenic from the copper to bronze version of each artefact type is not a dilution effect. Removing the tin percentage from the total and re-normalising the data to 100% has compensated for the addition of tin. The added-tin versions of Ross Island metal axes (288) and halberds (8) have slightly lower average arsenic levels than their copper equivalents (copper halberds 1.93%; tin-bronze halberds 1.82%; copper axes 1.19%; tin-bronze axes 1.07%), showing that on average the copper has been molten for longer due to the additional alloying process. Using the Mann-Whitney U-test for both axes and halberds, these differences are statistically significant at the 95% confidence level.

Geographical patterning

If we allow for the systematic differences between artefact types and only compare like with like, important geographical trends also emerge from the data. These can be used to interpret the route and mechanism for the flow of metal between regions. These two factors are central to understanding the movement of copper and bronze around Britain and Ireland during the EBA (Needham 1998). Figure 4 shows the regional groupings used in this study. The areas aim to correspond with common archaeological groupings, while the slight differences in size are a result of including a significant and even number of analysed objects within each region.

[FIGURE 5 OMITTED]

Figure 5 shows the data for the average levels of arsenic, antimony and silver in EBA axes made from Ross Island metal from a number of regions in Britain and Ireland. All of the axes are from Metalwork Assemblage 3 (c. 2200-2000 BC), a period during which the Ross Island mine was still active. The bold x-axis represents the overall average for each element in all the artefacts, and each region's average has been plotted as an offset from this figure. Ireland shows on average a small relative enrichment compared to the overall average, Scotland shows a small relative depletion, whilst eastern England is lowest on this scale. The falls from Ireland and Scotland down to eastern England, for arsenic and antimony, are statistically significant at the 95% confidence level, using the Mann-Whitney U-test.

[FIGURE 6 OMITTED]

Figure 5 can be interpreted as representing the flow of Ross Island metal away from the Irish source, and the increased frequency of re-melting as distance from the source increases. Due to its relative remoteness from Ross Island, the As-Sb-Ag copper in eastern England is more likely to have passed through a longer chain of melting events in order to end up in eastern England, and in its final shape. Each of these events, such as re-casting to new locally-preferred shapes, or smithing by a new owner, will have reduced the levels of antimony and arsenic within the axe. The significant number of moulds found in non-metalliferous areas supports this view of regional re-casting (Hodges 1959, 1960), as does the rise of regional artefact shapes (Schmidt & Burgess 1981; Needham 1983). Welsh axes are not included in Figure 5, as people in that region were uniquely moving away from depositing objects made of Ross Island metal, due in part to the beginnings of local extractive metallurgy (Timberlake 2002, 2003).

On average, Ross Island metal axes eventually deposited in Ireland are more likely to have moved a much shorter distance, through fewer hands and with perhaps less social need for remodelling. In his theoretical treatment of metal movement and recycling, Needham (1998: 277) argues that: "The average number of re-castings within each region, per unit weight of metal, is however a crucial variable in the modelling exercise." Assessing the relative losses of arsenic, antimony and silver due to melting begins to provide this crucial variable, allowing some insights into many central aspects of EBA metallurgy.

Chronological patterns

Table 1 shows dramatic changes in the volume of Ross Island metal that was being deposited over the course of the EBA. This variation is closely linked with differing trading patterns across the study area, and the abandonment of the Ross Island mines beginning at around 2100 BC for the Blue Hole. These changes are also reflected in the average composition of Ross Island metal artefacts with time, plotted in Figure 6. These values have been adjusted to compensate for the addition of tin if necessary (see above). The chronological differences here conflate the technological effects described above (losses due to melting during casting and tin alloying) and also the increased recycling of existing old Ross Island metal objects in later periods. The relatively high levels of arsenic seen in the halberd data is caused by their closed two-piece moulds preventing oxidative loss during casting. The difference between arsenic levels for MA 1/2 to 4/5 objects and MA 1/2 to MA 6 artefacts are both significant at the 99% confidence level using the Mann-Whitney U-test.

The MA 6 (c. 1800-1600 BC) Ross Island objects were produced using metal that must have been smelted around 500 years earlier, and we can see their correspondingly low arsenic, antimony and silver levels as a result of extensive recycling. Again, if we did not have access to the actual production site, the data presented in Figure 6 would still allow us to infer that later artefacts of Ross Island type were more commonly recycled than earlier examples, and to a more intense degree. This shift in the volume and nature of recycling is revealing of the changing interaction between people and metal. Over the course of the EBA there was a gradually increasing recognition, social acceptance and exploitation of metal's mutability (Bray 2009, in press).

Diagnostic chemical metallurgy: interpreting chemical datasets

The particular properties of the constituent trace elements of EBA copper-alloy objects can also be used diagnostically to interpret chemical groups that are either difficult to explain or are not currently linked to known ore sources. An example of the first type is a group of 66 EBA copper-alloy artefacts that have a strange chemical signature of mostly copper and tin, but also low levels of silver (between 0.1 and 1%), with other elements almost absent (below 0.1%). In Ireland this composition makes up large amounts of the assemblage in the second half of the EBA (c. 2000-1500 BC), comprising around 10% of flanged axes for example (Bray 2009).

Under the simple paradigm of connecting chemistry to provenance, this chemical group is difficult to interpret, as there are no clear links to a possible source of ore. Allowing for shifts in the chemical composition as a result of post-smelt alteration, however, gives this group an obvious identity as being heavily recycled Ross Island metal. The more vulnerable arsenic and antimony have been oxidised away and only the more noble silver remains, as predicted by the Ellingham diagram and experimental results (Figure 1; Merkel 1982; Beeley 2001). The increasing frequency of this depleted 'silver only' copper type towards the end of the EBA supports the possibility of its Ross Island origins. The mine closed around 2000 BC, and old metal was therefore increasingly recycled over the next five centuries to leave objects with this strange composition.

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

Using fine structure within superficially similar chemical groups can also help interpret copper types not currently connected to geological sources. A large number of EBA artefacts (121) contain a significant amount of arsenic (>0.1%), but no other elements aside from copper and tin. This is around 6% of the total EBA assemblage but it is mostly found in objects from MA 4 onwards (c. 2000 BC onwards; see Table 1). This 'arsenic only' alloy is central to a re-evaluation of metal origin and movement during the EBA. Variations in the average arsenic percentage composition within the British 'arsenic only' material (Figure 7) show increased levels towards south-west England. On average it loses arsenic as it gets re-cast after exchange into locally appropriate and current shapes; with metal deposited in eastern England having the lowest arsenic levels. As Buddet al. (2000) have demonstrated, small-scale extraction of copper was occurring in the English south-west during the EBA. However, placing the detailed chemical structure of British 'arsenic only' copper in a wider context suggests a more intriguing possibility. Figure 8 shows the distribution of arsenic within the axes of EBA Britain and Ireland. The clear pattern emerges that British material is shifted to the left-hand 'depleted' side of the graph, compared to the Irish 'arsenic only' metal. Irish material, on average, had undergone fewer melting events and, working from the model established above, is more likely to be closer to the beginning of the exchange path.

[FIGURE 9 OMITTED]

The coverage and depth of the European dataset produced by the SAM and later projects, and made available by Krause (2003), allows us to follow the arsenic depletion pattern further. Through plotting the percentage presence of 'arsenic only' EBA artefacts for each region of Europe, a clear focus of deposition of this copper type emerges on the Iberian peninsular (Figure 9). Allied to this, the precise chemical pattern also indicates that Iberia is a strong candidate as the source of the British and Irish material. The average level of arsenic in the Krause (2003) database for EBA material (SAM Period code 3) in SAM regions 15 and 11 (the Iberian Atlantic coast) is 2.39% (87 artefacts); for Atlantic France (SAM regions 21 and 23), 1.15% (118 artefacts); Irish artefacts contain just 0.75% arsenic (29 artefacts). Patterns of chemical change upon high temperature processing can therefore detail ancient technology on a continental scale. Rather than just look for correlations between composition and ores that produce the correct copper type, this model of provenance links the chemistry to the technology and character of prehistoric metal exchange and the remodelling of objects.

This brief case study needs further development, in particular the application of lead isotope analysis, lab-based reconstructions of the melting and mixing of early alloys, and a collation of further sets of analytical data and archaeological context. However, even at this stage we can begin to sketch links between chemical metallurgy and the models of EBA contact along the Atlantic sea routes. Detailed, empirical, understanding of long-distance exchange networks of copper-alloy could play a useful part in delineating these systems.

Conclusions

Evidence from experimental archaeology and modern process metallurgy shows unequivocally that the element composition of a copper alloy will vary systematically as a result of periods spent in the liquid state. We suggest that these properties offer an additional interpretative tool when considering the average composition of groups of copper alloy artefacts, one which gives an indication of the number of times objects within that group, on average, have been re-melted. The final measured composition of an object is the result of a series of cumulative processes, the starting point for which is created by the mineralogy of the ore source, but the end point being defined by the life history of that particular object.

Whilst retaining the empirical underpinnings of process metallurgy, this paper aims to move the reading of archaeometallurgical chemical data closer to mainstream archaeological interpretations (e.g. Budd & Taylor 1995; Ottaway 2002). People's choices created technological, geographical and chronological patterning in the composition of the artefact as found. For the first time we can begin to put empirically-derived constraints on processes such as recycling and metal exchange. Identities can be given to groups of objects that have no obvious connection to a mineralogical source. As the chemical processes involved are universal, these ideas can apply to any copper-alloy using society. More fundamentally important, though, is that we can begin to see the people acting behind the numbers.

Acknowledgements

We gratefully acknowledge the support given by The Queen's College, Oxford, in the form of a Hastings Senior Scholarship to the first author, and the subsequent support of the John Fell Fund (University of Oxford) and the Leverhulme Trust (Grant F/08 622/D). We are also grateful for a number of comments on a previous version of this paper.

Received: 26 August 2011; Accepted: 1 November 2011; Revised: 23 January 2012

References

AMBERT, P. 1991. L'emergence de la metallurgie chalcolithique dans midi de la France. Archeologie en Languedoc 10/11: 51-58.

AMBERT, P. & H. BARGE-MAHIEU. 1991. Les mines prehistoriques de Cabrieres (Herault), in J.-P. Mohen & C. Eluere (ed.) Decouverte du metal: 259-77. Paris: Picard.

ANDERSON, C.T. 1930. The heat capacities of arsenic, arsenic trioxide, and arsenic pentoxide at low temperatures. Journal of the American Chemical Society 52: 2296-300.

BEELEY, P. 2001. Foundry technology. Oxford: Butterworth-Heinemann.

BRAY, P.J. 2009. Exploring the social basis of technology: re-analysing regional archaeometric studies of the first copper and tin-bronze use in Britain and Ireland. Unpublished PhD dissertation, University of Oxford.

--In press. Before [sup.29]Cu became copper, tracing the recognition and invention of metalleity in Britain and Ireland during the third millennium BC, in M. Allen (ed.) Is there a British Chalcolithic? People, place and polity in the later 3rd millennium (Prehistoric Society research paper 4). Oakville (CT): The Prehistoric Society.

BRITTON, D. 1963. Traditions of metalworking in the later Neolithic and Early Bronze Age of Britain: Part 1. Proceedings of the Prehistoric Society 29: 258-325.

BUDD, P. & T. TAYLOR. 1995. The fairie smith meets the bronze industry: magic versus science in the interpretation of prehistoric metal-making. World Archaeology 27: 133-43.

BUDD, P., R. HAGGERTY, R.A. IXER, B. SCAIFE & R.G. THOMAS. 2000. Copper deposits in south-west England identified as a source of Copper Age metalwork. Available at: http://goodprovenance.com/provenance.html (accessed 17 April 2012).

BURGESS, C. 1980. The age of Stonehenge. London: Phoenix Press

CASE, H.J. 1954. Studies of Irish and British early copper artefacts: second series. Man 54:18-27.

CHARLES, J.A. 1980. Recycling effects on the composition of non-ferrous metals. Philosophical Transactions of the Royal Society of London Series A 295: 57-68.

COGHLAN, H.H. 1970. British and Irish Bronze Age implements in the Borough of Newbury Museum. Newbury: Newbury Museum.

--1979. Analyses of Bronze Age artefacts from Irish museums. Historical Metallurgy 13: 98-105.

COGHLAN, H.H. & H.J. CASE. 1957. Early metallurgy of copper in Ireland and Britain. Proceedings of the Prehistoric Society 23:91-123.

COGHLAN, H.H. & M. COOK. 1953. Studies of British and Irish Celts: first series. Man 53:97-101. Copper Development Association. 2011. Annual data 2011. Copper: brass: bronze. Copper supply and consumption 1990-2010. Available at: http://www.copper.org/resources/ market-data/pdfs/annual_data.pdf (accessed 17 April 2012).

EARL, B. & A. ADRIAENS. 2000. Initial experiments on arsenical bronze production. Journal of the Minerals, Metals and Materials Society 52:14-16.

FRIEDMAN, A.M., M. CONWAY, M. KASTNER, J. MILSTED, D. METTA, P.R. FIELDS & E. OLSEN. 1966. Copper artifacts: correlation with some types of copper ores. Science 152: 1504-506.

GOWLAND, W. 1906. Presidential address: copper and its alloys in prehistoric times. Journal of the Anthropological Institute of Great Britain and Ireland 36:11-38.

HAMPTON, D.F.G., P. BENNETT, D. BROWN, R. LANCASTER, J.L. RICE, A.L. SHARP; H.A. STEPHENS & H.T. BIDWELL. 1965. Metal losses in copper-base alloys. The British Foundryman 58: 225-40.

HODGES, H.W.M. 1959. The Bronze Age moulds of the British Isles, Part 1: Scotland and northern England--moulds of stone and clay. Sibrium 4: 129-37.

--1960. The Bronze Age moulds of the British Isles, Part 2: England and Wales--moulds of stone and bronze. Sibrium 5:153-62.

IXER, R.A. 1999. The role of ore geology and ores in the archaeological provenancing of metals, in S.M.M. Young, A.M. Pollard, P. Budd & R.A. Ixer (ed.) Metals in antiquity (British Archaeological Reports international series 792): 43-52. Oxford: Archaeopress.

IXER, R.A. & P. BUDD. 1998. The mineralogy of Bronze Age copper ores from the British Isles: implications for the composition of early metalwork. Oxford Journal of Archaeology 17:15-41.

IXER, R.A. & R.A.D. PATTRICK. 2003. Copper-arsenic ores and Bronze Age mining and metallurgy with special reference to the British Isles, in P.T. Craddock & J. Lang (ed.) Mining and metal production through the ages: 9-20. London: British Museum Press.

JUNGHANS, S., E. SANGMEISTER & M. SCHRODER. 1960. Metallanalysen kupferzeitlicher und furhbronzezeitlicher Bodenfunde aus Europa (Studien zu den Anfangen der Metallurgie 1). Berlin: Gebr. Mann.

--1968. Kupfer und Bronze in der fruhen Metallzeit Europas (Studien zu den Anfangen der Metallurgie 2). Berlin: Gebr. Mann.

--1974. Kupfer und Bronze in derfruhen Metallzeit Europas (Studien zu den Anfangen der Metallurgie 4). Berlin: Gebr. Mann.

KILLICK, D. 2001. Science, speculation and the origins of extractive metallurgy, in D.R. Brothwell & A.M. Pollard (ed.) Handbook of archaeological sciences: 483-92. Chichester: Wiley.

KRAUSE, R. 2003. Studien zur kupfer- und fruhbronzezeitlichen Metallurgie zwischen Karpatenbecken und Ostsee (Vorgeschichtliche Forschungen 24). Rahden: Leidorf.

KRAUSE, R. & E. PERNICKA. 1996. Das neue Stuttgarter Metallanalysenprojekt 'SMAP'. Archaologisches Nachrichtenblatt 1: 274-91.

LEE, Y.Y., H.W. TSENG, Y.H. HSIAO & C.Y. LIU. 2009. Surface oxidation of molten Sn (Ag, Ni, In, Cu) alloys. Journal of the Minerals, Metals and Materials Society 61: 52-58.

MERKEL, J.F. 1982. Reconstruction of Bronze Age copper smelting, experiments based on archaeological evidence from Timna, Israel. Unpublished PhD dissertation, Institute of Archaeology, University of London.

McKERRELL, H. & R.F. TYLECOTE. 1972. Working of copper-arsenic alloys in the Early Bronze Age and the effect on the determination of provenance. Proceedings of the Prehistoric Society 38:209-18.

MOLLER, R. & E. PERNICKA. 2009. Chemical analyses in archaeometallurgy: a view on the Iberian peninsula, in T.L. Kienlin & B. Roberts (ed.) Metals and societies. Studies in honour of Barbara S. Ottaway: 296-306. Bonn: Rudolf Habeh.

NEEDHAM, S.P. 1983. The Early Bronze Age axeheads of central and southern England. Unpublished PhD dissertation, University College, Cardiff.

--1996. Chronology and periodisation in the British Bronze Age, in K. Ran&borg (ed.) Absolute chronology: archaeological Europe 2500-500 BC (Acta archaeologica 67): 121-40.

--1998. Modelling the flow of metal in the Bronze Age, in C. Mordant, M. Pernot & V. Rychner (ed.) L'atelier du bronzier en Europe du [XX.sup.e] au [VIII.sup.e] siele avant notre ere, III: production, circulation et consommation du bronze: 285-307. Paris: CTHS.

--2002. Analytical implications for Beaker metallurgy in northwest Europe, in M. Bartelheim, E. Pernicka & R. Krause (ed.) Die Anfange der Metallurgie in der alten Welt: 99-133. Rahden: Marie Leidorf.

NEEDHAM, S.P., C. BRONK RAMSEY, D. COOMBS, C. CARTWRIGHT & P. PETTITT. 1997. An independent chronology for British Bronze Age metalwork: the results of the Oxford radiocarbon accelerator programme. Archaeological Journal 154:55-107.

NEEDHAM, S., M. PARKER PEARSON, A. TYLER, M. RICHARDS & M. JAY. 2010. A first 'Wessex 1' date from Wessex. Antiquity 84: 363-73.

NIEDERSHCHLAG, E., E. PERNICKA, T. SEIFERT & M. BARTELHEIM. 2003. The determination of lead isotope ratios by multiple collector ICP-MS: a case study of Early Bronze Age artefacts and their possible relation with ore deposits of the Erzgebirge. Archaeometry 45:61-100.

NORTHOVER, P. 1980. The analysis of Welsh Bronze Age metalwork, in H. Savory (ed.) Guide catalogue of the Bronze Age collections: 229-43. Cardiff: National Museum of Wales.

--1999. The earliest metalworking in southern Britain, in A. Hauptmann, E. Pernicka, T. Rehren & U. Yalcin (ed.) The beginnings of metallurgy (Anschnitt Beiheft 9): 211-26. Bochum: Deutsches Bergbau-Museum.

O'BRIEN, W. 2004. Ross Island. Mining, metal and society in early Ireland. Galway: National University of Ireland.

OTTAWAY, B.S. 2002. Towards interpretive archaeometallurgy, in M. Bartelheim, E. Pernicka & R. Krause (ed.) Die Anfange der Metallurgie in der alten Welt:. 7-13. Rahden: Marie Leidorf.

PERNICKA, E. 1999. Trace element fingerprinting of ancient copper: a guide to technology or provenance, in S.M.M. Young, A.M. Pollard, P. Budd & R.A. Ixer (ed.) Metals in antiquity (British Archaeological Reports international series 792): 163-71. Oxford: Archaeopress.

PICKLES, C.A. 1998. Selective oxidation of copper from liquid copper-silver alloys. Metallurgical and Materials Transactions B 29: 39-51.

ROHL, B. & S.P. NEEDHAM. 1998. The circulation of metal in the British Bronze Age: the application of lead isotope analysis. London: British Museum.

RUTLAND, R.A. & H.H. COGHLAN. 1972. Bronze Age flat axes from Berkshire. Berkshire Archaeological Journal 66: 45-59.

SCHMIDT, P.K. & C.B. BURGESS. 1981. The axes of Scotland and northern England. Munich: C.H. Beck'sche Verlagsbuchhandlung.

TANAHASHI, M, T. FUJINAGA, S. ZHIJIAN, K. TAKEDA, S. HONG YONG & C. YAMAUCHI. 2005. Effects of coexisting oxygen and antimony in molten copper on rate of arsenic elimination from the copper phase by the use of [Na.sub.2]C[O.sub.3] slag. Materials Transactions 46:2180-89.

TIMBERLAKE, S. 2002. Ancient prospection for metals and modern prospection for ancient mines--the evidence for Bronze Age mining within the British Isles, in M. Bartleheim, E. Pernicka & R. Krause (ed.) Die Anfange der Metallurgie in der alten Welt:. 328-57. Rahden: Marie Leidorf.

--2003. Excavations on Copa Hill, Cwmystwyth (1986-1999); an Early Bronze Age copper mine within the uplands of central Wales (British Archaeological Reports British series 348). Oxford: Archaeopress.

P.J. Bray & A.M. Pollard *

* Research Laboratory for Archaeology and the History of Art, University of Oxford, Dyson Perrins Building, South Parks Road, Oxford OX1 3QJ, UK (Email: peter.bray@rlaha.ox.ac.uk; mark.pollard@rlaha.ox.ac.uk)

Table 1. Approximate date ranges of Needham's metalwork assemblages
(after Needham 1996; Needham et al. 2010). Also the percentage of the
British and Irish Early Bronze Age copper-alloy artefact assemblage
formed by Ross Island copper (Bray 2009).

 Needham's
Approximate Metalwork
date range Assemblage (MA)

Mid 2nd millennium to 2200 BC 1 and 2
2200 BC to 2000 BC 3
2000 BC to 1700 BC 4 and 5
1700 BC to 1500 BC 6
All pre-Middle Bronze Age copper
(Total number of analysed EBA copper-alloy objects)

 Total no. of analysed
Approximate objects with an
date range assigned MA

Mid 2nd millennium to 2200 BC 349
2200 BC to 2000 BC 402
2000 BC to 1700 BC 552
1700 BC to 1500 BC 370
All pre-Middle Bronze Age copper 1673
(Total number of analysed EBA copper-alloy objects) (2129)

 No. of Ross
Approximate Island copper Ross Island
date range objects % of total

Mid 2nd millennium to 2200 BC 254 72.8
2200 BC to 2000 BC 246 61.2
2000 BC to 1700 BC 52 9.4
1700 BC to 1500 BC 17 4.6
All pre-Middle Bronze Age copper 569 34.0
(Total number of analysed EBA copper-alloy objects)

Figure 4. Regional divisions used in this paper and the location
of the Early Bronze Age mine at Ross Island, Co. Kerry, Ireland.

 Number of
 artefacts
 examined
 Name Components scientifically

1 North and Scotland, Cumbria, Durham, 265
 Scotland Humberside, Lancashire,
 Northumberland, Yorkshire

2 Central Bedfordshire, Berkshire, 197
 England Buckinghamshire, Chesire,
 Derbyshire, Greater london,
 Hampshire, Herefordshire,
 Isle of Wight, Lincolnshire,
 Northamptonshire,
 Nottinghamshire,
 Oxfordshire, Shropshire,
 Staffordshire, Surrey, Sussex

3 Wales Wales 116

4 South-west) Avon, Cornwall, Devon, 161
 England Dorset, Gloucestershire,
 Somerset, Wiltshire

5 Eastern Cambridgeshire, Kent,
 England Norfolk, Suffolk 138
 Ulster 208

7 Leinster Leinster 162

8 Munster Munster 138

9 Connacht Connacht 99