See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/285815957

Chemical analyses in archaeometallurgy: a view on the Iberian peninsula

Article · January 2009

CITATIONS READS 20 221

2 authors:

Roland K. Gauss Ernst Pernicka EIT Raw Materials Curt Engelhorn Zentrum Archäometrie and University of Heidelberg

24 PUBLICATIONS 670 CITATIONS 361 PUBLICATIONS 9,273 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Bronze Age planoconvex copper ingots from the Saalach and Salzach Valley ("Salzburger Gusskuchenprojekt") View project

Bronze Age tin - Tin isotopes and the sources of Bronze Age tin in the old World (ERC project) View project

All content following this page was uploaded by Ernst Pernicka on 02 February 2016.

The user has requested enhancement of the downloaded file.

Contents

Ben Roberts – Tobias L. Kienlin Foreword ...... 9

Caroline Jackson Of Barbara...... 11

Christian Strahm Die Begegnung mit Barbara Ottaway: Erinnerungen an die Impulse für die frühen akademischen Studien ...... 16

Publications of Barbara S. Ottaway...... 18

I. Metals and Societies

Christopher P. Thornton Archaeometallurgy: Evidence of a Paradigm Shift?...... 25

Martin Bartelheim Elites and Metals in the Central European Early Bronze Age...... 34

Rüdiger Krause Bronze Age Copper Production in the Alps: Organisation and Social Hierarchies in Mining Communities ...... 47

Tobias L. Kienlin – Thomas Stöllner Singen Copper, Alpine Settlement and Early Bronze Age Mining: Is There a Need for Elites and Strongholds? ...... 67

Emma C. Wager Mining Ore and Making People: Re-thinking Notions of Gender and Age in Bronze Age Mining Communities...... 105

Christian Strahm – Andreas Hauptmann The Metallurgical Developmental Phases in the Old World ...... 116

Ben Roberts Origins, Transmission and Traditions: Analysing Early Metal in Western Europe...... 129

Benoît Mille – Laurent Carozza Moving into the Metal Ages: The Social Importance of Metal at the End of the Neolithic Period in France...... 143 Contents

Dirk Brandherm The Social Context of Early Bronze Age Metalworking in Iberia: Evidence from the Burial Record ...... 172

John Bintliff Is the Essence of Innovative Archaeology a Technology for the Unconscious?...... 181

II. Aspects of Copper and Bronze Age Metallurgy

Dušan Borić Absolute Dating of Metallurgical Innovations in the Vinča Culture of the Balkans...... 191

Nikolaus Boroffka Simple Technology: Casting Moulds for Axe-adzes ...... 246

Tobias L. Kienlin – Ernst Pernicka Aspects of the Production of Copper Age Jászladány Type Axes ...... 258

Mark Pearce How Much Metal was there in Circulation in Copper Age Italy? ...... 277

Paul Ambert – Valentina Figueroa-Larre – Jean-Louis Guendon – Veronika Klemm – Marie Laroche – Salvador Rovira – Christian Strahm The Copper Mines of Cabrières (Hérault) in Southern France and the Chalcolithic Metallurgy ...... 285

Roland Müller – Ernst Pernicka Chemical Analyses in Archaeometallurgy: A View on the Iberian Peninsula...... 296

Susan La Niece – Caroline Cartwright Bronze Age Gold Lock-rings with Cores of Wax and Wood...... 307

Trevor Cowie – Brendan O’Connor Some Early Bronze Age Stone Moulds from Scotland ...... 313

Viktoria Kiss The Life Cycle of Middle Bronze Age Bronze Artefacts from the Western Part of the Carpathian Basin...... 328

Elka Duberow – Ernst Pernicka – Alexandra Krenn-Leeb Eastern Alps or Western Carpathians: Early Bronze Age Metal within the Wieselburg Culture. . . 336

Marianne Mödlinger – Gerhard Trnka Herstellungstechnische Untersuchungen an Riegseeschwertern aus Ostösterreich ...... 350

Barbara Horejs Metalworkers at the Çukuriçi Höyük? An Early Bronze Age Mould and a “Near Eastern Weight” from Western Anatolia ...... 358 Contents

Vincent C. Pigott “Luristan Bronzes” and the Development of Metallurgy in the West-Central Zagros, Iran. . . . . 369

Quanyu Wang – Jianjun Mei Some Observations on Recent Studies of Bronze Casting Technology in Ancient China...... 383

III. Approaches to Early Metallurgy

Walter Fasnacht 7000 Years of Trial and Error in Copper Metallurgy – in One Experimental Life...... 395

Caroline Jackson Experimental Archaeology and Education: Theory without Practice is Empty; Practice without Theory is Blind...... 400

Salvador Rovira – Ignacio Montero-Ruiz – Martina Renzi Experimental Co-smelting to Copper-tin Alloys...... 407

Julia Heeb Thinking Through Technology – An Experimental Approach to the Copper Axes from Southeastern Europe...... 415

Colin Merrony – Bryan Hanks – Roger Doonan Seeking the Process: The Application of Geophysical Survey on some Early Mining and Metalworking Sites...... 421

IV. Studies in Historical Metallurgy

Alessandra Giumlia-Mair – Péter Gaboda – Hedvig Györy – Irén Vozil Two Statuettes with ḥmty km Inlays in the Fine Arts Museum Budapest...... 433

Nerantzis Nerantzis Using Mills to Refine Metals:Iron Smelting Technology of the Transitional Byzantine to Ottoman Period in Macedonia, Greece...... 443

Paul T. Craddock Perceptions and Reality: The Fall and Rise of the Indian Mining and Metal Industry...... 453

List of Contributors...... 465 296

Roland Müller – Ernst Pernicka

Chemical Analyses in Archaeometallurgy: A View on the Iberian Peninsula

Abstract from different sources is comparable. This paper aims The paper focuses on the issue of comparability of analyti- to discuss the major analytical series of Iberian cop- rd cal data produced in archaeometallurgical research on the per based artefacts dating from the early 3 to the be- st Iberian Peninsula during the last 120 years. Indeed, the ginning of the 1 millennium BC (Copper and Bronze results produced by the different laboratories are largely Ages), focusing on the issue of data comparability. comparable with each other. Although a large-scale overall comparison of the major quantitative analytical series is still lacking, it is very promising that the conclusions drawn Chemical analyses of copper-based from these different kinds of analytical data do not contra- dict with each other. A small-scale evaluation of the repro- objects from prehistoric Iberian ducibility and accuracy of the trace element analyses run contexts by the ‘Studien zu den Anfängen der Metallurgie’ (SAM) As pointed out above, the usefulness of examining project revealed that they fit very well into the good overall prehistoric metal artefacts chemically was recognized picture of SAM data quality found by previous comparative very early. However, the research questions behind studies. such analyses have varied greatly. Chemical analyses may be conducted for instance in order to identify the composition of a metal object and learn about the material properties of the alloy. Furthermore, due to Introduction the apparent impartial nature of scientific analysis, archaeologists have attempted to develop unbiased Chemical analyses of metal artefacts have a long his- artefact classifications based on systematic trace tory in archaeological research. Indeed, the first ‘sys- ele­ment studies (e.g. Junghans et al. 1954; Butler/ tematic’ analyses date back to the formation period van der Waals 1966).2 In addition, the trace element of scientific archaeology, i.e. to the early 19th century. signatures of copper artefacts compared with those Scholars such as Klaproth (1815) and Hünefeld and of slags and ores may allow the identification of the Picht (1827) studied prehistoric metal artefacts using geological provenance of the metal. Obviously, only wet chemical methods, in order to learn about their chemical analysis allows comparison of objects that material properties.1 Chemical analyses have also – although they may derive from the same kind of played a crucial role in archaeological studies on pre- technological process – are of completely different historic Iberia. For example, the Portuguese scholar shape. From chemical analyses of metal objects one Estacio da Veiga (1889: 98–101, 117) pointed out that may also learn more about the metal technology of a the earliest metal used on the Iberian Peninsula was certain period, such as testing whether artefact types made of pure copper and that the earliest Iberian cop- correlate with a certain type of bronze composition. per using cultures still stood in a Neolithic tradition. Finally, chemical examinations are used to study the Thus, he coined the term ‘Copper Age’ for southern corrosion processes of metal objects. Portugal and Iberia in general. During the last two centuries, various analytical Since then, there have been several attempts techniques have been applied to prehistoric Iberian to characterise prehistoric Iberian metal artefacts metal objects with these research objectives. The chemically by various research groups. Today, the analyses cited by Estacio da Veiga (1889: 98) were investigator is often confronted with the problem of undertaken by Mr. Witnich, a chemist who exam- locating all the data that is widely distributed in the ined flat axes from prehistoric graves of the Odemira literature. Even more problematic is whether the data district (southern Portugal), which had been discov-

1 See discussion in Pernicka (1986) and Schmitz (2004: 2 See also discussion in Härke (1978) and Pernicka (1990: 135–142); see also Trigger (2000: 73). 76–99). Chemical Analyses in Archaeometallurgy: A View on the Iberian Peninsula 297

Fig 1: Artefacts from prehistoric sites of southeast Spain, which were analysed by Henri and Louis Siret. They found that the earliest metal objects from southeast Spain had similar shapes as regular stone tools and that they often lacked tin as the crucial element for producing tin bronze (after Siret/Siret 1887: plate XII). ered by Silva Ribeiro. The analyses were ordered by “to determine the nature of some metallic instru- Emil Cartailhac, who published a book on prehistoric ments”, and second, “to verify the existence of a lo- Spain and Portugal (Carthailac 1886). Obviously, in cal [i.e. Portuguese] copper or bronze metallurgy” (do this case the analyses were used to identify the com- Paço 1955: 33). The analyses revealed that the assem- position of the metal. The same applies to most of the blage of Vila Nova de São Pedro comprises artefacts comprehensive analyses of Copper and Bronze Age made of copper as well as tin bronze. Due to the ab- artefacts from sites in southeast Spain published by sence of tin in all copper droplets analysed it was sug- Henri and Luis Siret (fig. 1). They calculated for in- gested that tin bronze was not processed at the site stance the relative amounts by weight of the main (do Paço 1955: 33; see also Müller/Soares 2008). Al- Bronze Age metals, i.e. of copper, silver, tin (as part though the quality of the report was not comparable of bronzes), and gold (Siret/Siret 1887: 408–414). The with modern standards (e.g. only few artefacts were analyses published by the Siret brothers were ac- analysed, arsenic was not determined, the methodo- complished with various institutions and different logical background was not described at all etc.), the researchers including themselves. Thus the quality of idea of applying chemical analyses for archaeological the data varies greatly (Montero 1994: 45–46; Harri- research questions was innovative. Even more so, do son/Craddock 1981: 158). The pioneering work of the Paço clearly argued for invasive and even destructive Siret brothers was neither followed nor decisively de- chemical analyses of metal objects as they would offer veloped further by other researchers of the late 19th valuable insights for understanding prehistoric trade and early 20th century. routes and ancient metallurgical technology (do Paço In 1955, Afonso do Paço published chemical ana­ 1955: 32). This notion of a scientific approach to ar- lyses of some metal objects found at the Copper Age chaeology offered a perfect background for the SAM3 fortificated site of Vila Nova de São Pedro (Portugal), archaeometallurgical project in Portugal (Junghans which he had excavated. They still stand in the same et al. 1960; 1968; 1974).4 tradition as the Siret analyses: they were conducted by the National Laboratory of Civil Engineering in 1951 and 1953, and focused on the determination of 3 SAM: Studien zu den Anfängen der Metallurgie (Studies on the the major and minor elements of the metal artefacts. Beginnings of Metallurgy). The research goals were clearly formulated, i.e. first, 4 See also Sangmeister (1995; 2005; in press) 298 Roland Müller – Ernst Pernicka

In 1954, Edward Sangmeister, specializing in the BC were made of the ‘younger’ copper containing sig- Iberian Copper Age as well as having been one of the nificant levels of arsenic, antimony, and silver (Sang- leading members of the SAM project, began his work meister 1995: 72–76). at the German Archaeological Institute Madrid. Tra­ In 1981, Harrison and Craddock published about velling to various Portuguese and Spanish museums one hundred analyses of “Bronze Age Metalwork and research institutions, he was in an excellent po- from the Iberian Peninsula” stored at the British Mu- sition for meeting local colleagues and for sampling seum (Harrison/Craddock 1981). The basic aim was step by step a large number of artefacts (Sangmeister to characterise the assemblage in terms of its trace in press). The aim of the project was to analyze quan- element patterns and thus be able to discuss tech- titatively a large number of prehistoric Eurasian cop- nological aspects of the use of metal in Copper and per and bronze artefacts in order to find composition- Bronze Age Iberia. Atomic absorption spectrometry ally similar groups. From their distribution in time was applied to analyse drill shavings of 10 to 20 mg.9 and space the research group hoped to find copper Among the most important findings of this study was types that would be characteristic for certain regions a correlation between artefact type and arsenic con- and/or periods (Junghans et al. 1960; 1968; 1974).5 The tent of the early metalwork. They observed that flat analytical method used was optical emission spec- axes tend to consist of pure copper whereas Palmela troscopy (today one would rather refer to it as atomic points and daggers tend to show higher concentra- emission spectrometry) that allowed high sample tions of arsenic (Harrison/Craddock 1981: 161–162). throughput requiring relatively small samples, which They also confirmed that arsenical copper dominated were obtained from the bulk metal by drilling. The the metal assemblage throughout the Argaric Bronze project accomplished more than 22,000 analyses of Age (ibid: 164). prehistoric copper-based artefacts, of which almost In the early 1980s, a large-scale analytical program 1,700 derived from the Iberian Peninsula. After a pe- called ‘Proyecto de Arqueometalurgia’ (PA) was initiat- riod of initial enthusiasm, the data produced by the ed by a group of Spanish archaeologists and scientists, project as well as the way they were classified and in- which has had a major impact on our present day un- terpreted were widely criticised (e.g. for the Iberian derstanding of early Iberian metallurgy (e.g. Rovira et Peninsula: Montero 1994: 46; see discussion below).6 al. 1997; Delibes/Montero 1999; Rovira/Gómez 2003). Nevertheless, several conclusions of major histori- More than 10,000 analyses of metal objects were pro- cal relevance were distilled from the data, e.g. that duced, of which about 2,000 belong to the Copper and pure copper and later arsenical copper dominate the Early Bronze Ages (Rovira 2002). The objects were sur- metal assemblages of Copper and Early Bronze Age face cleaned in one area and then analysed using X-ray Iberia (Junghans et al. 1968: 127). A large part of the fluorescence (XRF) (Rovira et al. 1997: 7). The project objects analysed originated from south central Por- has provided a comprehensive overview of Copper, tugal (1,012 artefacts dating to the 3rd and first half of Bronze, and Iron Age metalwork of Spain. Key conclu- the 2nd millennium BC). Later Sangmeister developed sions drawn are for example that the arsenical cop- a classification system using a combination of three per dominating the 3rd and 2nd millennium BC in fact two-dimensional element diagrams; i.e. diagrams represents a ‘natural alloy’, i.e. the arsenic most likely showing the contents of As-Ni, As-Bi, and Sb-Ag and derived from the processed ores. Slag, ore, and cruci- defined regional groups. The lead and tin contents of ble studies furthermore showed that a simple crucible the artefacts were discussed separately (Sangmeister smelting technology was primarily applied, which was 1995: 39–43; cf. Sangmeister 2005). Based on a typo- based on the exploitation of rich oxidic ores (Rovira/ logical chronology he concluded that i) pure copper Ambert 2002; Müller et al. 2004). Metallographic stud- was primarily used during the early Copper Age, ii) ies have documented the working cycles of the metal that arsenical copper with low concentrations of oth- diachronically (Rovira/Gómez 2003). er elements7 was used predominantly throughout the In 2004, a research project was launched by the Copper Age, and iii) that arsenical copper with high- German Archaeological Institute in cooperation with er concentrations of antimony, silver, and/or nickel the Institute of Archaeometry, University of Mining tended to occur in late Copper Age artefact types. and Metallurgy Freiberg (Saxony) to characterise Thus, none of the 24 tanged daggers8 of the south chemically and mineralogically archaeometallurgical central Portuguese data set was made of pure copper finds from Zambujal and other Chalcolithic sites of and nearly all tin bronzes of the early 2nd millennium the Portuguese Estremadura.10 104 quantitative XRF

5 See also Sangmeister (1995) and Müller (in press). 9 For more methodological details see Hughes et al. (1976). 6 See also discussion in Härke (1978) and Pernicka (1984; 1987: 10 The research project was initiated by Martin Bartelheim, 626–628; 1990: 79–99). Michael Kunst, Hermann Parzinger, and Ernst Pernicka. It 7 This copper type is called by Needham (2002: 107) ‘arsenic- is financed by the German Research Council (DFG) and the only copper’. German National Academic Foundation (Studienstiftung 8 In the Iberian Peninsula tanged daggers are found within the des Deutschen Volkes). After the Institute of Archaeometry socio-cultural context of the Bell Beaker phenomenon, which in Freiberg was closed in 2004, the Curt-Engelhorn-Centre flourished in the second half of the 3rd millennium BC. for Archaeometry, a research institute associated with the Chemical Analyses in Archaeometallurgy: A View on the Iberian Peninsula 299 and neutron activation analyses (NAA) of copper ob- be combined in order to draw conclusions of histori- jects – mainly scrap metal – as well as 104 NAA of ores cal relevance. Indeed, they are largely comparable and slags were conducted and combined with lead as the previous summary of the history of research isotope analyses and mineralogical analyses of slags, has already shown. The question is rather on what crucibles, and ores (Müller in press).11 The trace ele- discrimination level one compares the different ana- ment analyses were carried out on drill shavings or, if lytical results. The early wet-chemical analyses dis- the object was too small, on freshly cut metallic sur- cussed by Estacio da Veiga, the Siret brothers and do faces of the bulk metal.12 The investigations focused Paço as well as non-destructive XRF surface analyses primarily on i) identifying metal production remains aimed to identify the metal composition. Some of the within Zambujal and the other settlement contexts; conclusions that can be drawn from these analyses ii) finding the raw material sources exploited by the have been discussed above; they neither contradict early copper metallurgy; iii) reconstructing the in- each other nor the quantitative data of other analyti- novation process of copper metallurgy at Zambujal cal series. and the other Estremadurian sites (Müller in press).13 The comparison of the available quantitative data Some of the results which are based on trace element is more complex. So far a detailed, large-scale com- analyses will be discussed in section three. parison has not been made of the individual data pro- Next to these major analytical series, very few duced by the four major analytical series: ‘Studien zu small-scale analytical projects have been published, den Anfängen der Metallurgie’ (SAM), ‘British Muse- mostly focusing on the assemblage of a particular site um’ (BM), ‘Proyecto de Arqueometalurgia’ (PA), ‘Zam- (see for example discussion in Harrison/Craddock bujal Project’. First of all, there are methodological 1981: 158; Montero et al. 1990: 9–12). This includes a constraints: whereas the analyses by optical emission number of semi-quantitative analyses of Portuguese and atomic absorption spectrometry of the SAM and copper objects, which were mostly carried out us- BM projects were examining drill shavings of the bulk ing non-destructive X-ray fluorescence analysis of metal, the XRF analyses of the PA project were con- corroded artefact surfaces (Gonçalves 1989: 479–481; ducted on polished surfaces of artefacts. However, the Soares et al. 1996; Sousa et al. 2004). As an example, surface of a metal artefact may have a different com- examinations of the artefacts found at the Copper position than its bulk, due to segregation pro­cesses Age sites of Santa Justa and João Marques (Algarve, (Budd/Ottaway 1991: 136–138; Rovira et al. 1997: 7). Portugal) showed that they were made of pure and ar- The XRF and neutron activation analyses run by the senical copper. In a few cases tin was also detected at Zambujal project were carried out on drill shaving, low concentrations (Gonçalves 1989: 479–481). Hence, too; or, if the object was too small, on entire objects, generally speaking, the two assemblages fit well into freshly cut and polished. Thus they should represent the overall picture of a Chalcolithic south central Por- the composition of the bulk metal. The varying re- tuguese pure and arsenical copper industry. The same producibilities and detection limits of the different applies to artefacts found at south Portuguese sites analytical methods are further problems that have to such as Porto Torrão, Porto Mourão, and Três Moin- be considered. Finally, the metal of an artefact may hos (Soares et al. 1996; see also Soares et al. 1994). be inhomogeneous, due to segregations or inclusions, Twenty-two neutron activation analyses of copper which naturally would have an influence on evaluat- objects from Leceia are different in that they allow a ing the reproducibility of trace element analyses. quantitative evaluation of the bulk metal (Cardoso/ A systematic comparison of the SAM data and the Guerra 1997/98; Müller/Cardoso 2008). neutron activation analyses of the Zambujal project show that most of the results correlate over the range of two orders of magnitude (fig. 2). This was also ob- Comparing different data sets served in previous comparisons with analyses of the SAM project from other regions (fig. 3) (cf. Ottaway Clearly, there are numerous chemical data sets of 1982: 260; Pernicka 1990: 84; see also Chase 1974). prehistoric copper-based artefacts from the Iberian Most of the Portuguese Chalcolithic copper analysed Peninsula. The question arises whether the differ- is quite pure except for arsenic. Thus, apart from this ent data sets are comparable and whether they can element, the comparison is constrained to concentra- tions below 1,000 ppm (i.e. 0.1 %). The values for sil- ver show the largest differences (fig. 2). Indeed, this University of Tübingen, was founded, which became the new home of the project and some of its members (http://www. was also observed by Pernicka (1984: fig. 3), especially cez-archaeometrie.de/de/). below concentrations of 400 ppm silver. It seems that 11 See also Müller et al. (2007), Müller/Soares (2008) and Müller/Cardoso (2008). at low concentrations the silver signal may be inter- 12 For details on analytical methods see Müller (in press); see fered by another element. Nevertheless, the overall also Lutz/Pernicka (1996; on XRF), Müller/Cardoso (2008; on reproducibility of the SAM data is sufficient for ar- XRF) and Kuleff/Pernicka (1995; on NAA). 13 See also Müller et al. (2007), Müller/Cardoso (2008) and tefact classifications. The results of the reference Müller/Soares (2008). alloys that were used in Stuttgart for quantification 300 Roland Müller – Ernst Pernicka

Fig. 2: Comparison of analyses by optical emission spectrometry (OES) of the SAM project with neutron activation analyses (NAA) of the Zambujal Project; 33 replicate analyses of the same objects from south central Portugal. Most of the time, sample material left over from the SAM project was used. However, in few cases samples had to be taken again from the objects. The arrows indicate the detection limit of the OES. Concentrations are given in ppm.

demonstrate that under ideal measurement condi- can certainly be used as published (cf. Müller et al. tions even better reproducibility and accuracy was 2007: 18; Müller in press). obtainable (fig. 3, open circles). Other interferences Apart from this direct comparison of replicate were detected with antimony: First, there seems to analyses of the same objects, one may compare the be a small but systematic difference between emis- conclusions drawn from the quantitative evaluation sion spectroscopy and NAA, and second, a few analy- of the different analytical series. It is very encourag- ses are dramatically erroneous as indicated by filled ing to see that the results of the SAM project, those squares in figure 3. This was later verified by check- published by the British Museum, as well as those ing the original photographic plates that were used produced in Madrid (Proyecto de Arqueometalurgia) as detectors and which are now stored in the Curt- reveal that Palmela points and tanged daggers tend Engelhorn-Centre for Archaeometry in Mannheim. to show systematically higher arsenic contents than At the beginning of the SAM project an instrument flat axes and awls (see respectively: Müller et al. 2007: with lower spectral resolution was used and, there- 19; Harrison/Craddock 1981: 161–162; Rovira 2005: fore, two adjacent lines of antimony and iron could 507). It seems that copper with varying arsenic con- not be completely resolved. This was later corrected tents was extracted by smelting arsenic rich copper by re-determination of the concentrations of arsenic, oxide ores. The metal produced could then be sorted antimony and bismuth (Junghans et al. 1974). Since according to specific properties such as colour and Iberian Chalcolithic and Bronze Age copper-based hardness. Certain metal types could thus be reserved artefacts do not contain significant amounts of iron, for certain artefact types (cf. Gale et al. 1985; Müller this interference is negligible so that the SAM data et al. 2007; Müller in press). Chemical Analyses in Archaeometallurgy: A View on the Iberian Peninsula 301

Fig. 3: Comparison of the results of optical emission spectroscopy (SAM project) with neutron activation and atomic absorption analyses (Max- Planck-Institute for Nuclear physics Heidelberg; after Pernicka 1990: 84). The filled circles represent prehistoric artefacts sampled by the SAM project. The open circles correspond to comparative analyses of standard materials. The four outliers found in the Sb diagram are due to interfer- ences from Fe. Concentrations are given in percent. 302 Roland Müller – Ernst Pernicka

Fig. 4: Dendrogram of the cluster analysis sorting the As-, Sb-, Ni-, and Ag-signatures of 435 analysed artefacts found at Zambujal, Penedo, Fórnea, Leceia, and VNSP (Chalcolithic Estremadurian sites). The assessment of the agglomeration distance plot suggested the formation of 14 cluster. However, the four material groups defined by Sangmeister (1995) and one outlier are found at the dashed line; from right to left: outlier, pure copper, arsenical copper, arsenical copper with Sb and Ag, and arsenical copper with Ni (+Sb, Ag).

Another relevant comparison refers to the at- senic, antimony, nickel, and silver were considered. tempt of relating trace element patterns of copper The key problem of using cluster analysis is to decide artefacts with chronological sequences. As pointed on how many clusters should be used to divide a data out above, Sangmeister (1995) argued for a develop- set, which reveals the interpretative character of this ment in the use of different types of copper in Copper classification method. If it is structured too finely, Age Portugal. His argument was based on the evalu- small archaeologically as well as analytically non-re- ation of the composition of chronologically signifi- producible groups may be formed. If there are too few cant artefact types. If Sangmeisters material groups groups, the classification may produce trivial results really are historically relevant, one would suppose to of no relevance. One way to approach this problem is find these in other equivalent data sets as well. Thus, to produce an ‘Agglomeration Distance Plot’, showing a new data set of 435 analyses was grouped using the average distances between the different ‘objects’ cluster analysis (fig. 4) (Müller in press; cf. Ottaway of consideration, which are – in this case – defined 1974). The data base included the existing SAM data by the arsenic, antimony, nickel, and silver concen- of Zambujal artefacts (337 analyses) as well as new re- sults of neutron activation analyses of copper objects SAM data was withdrawn. 17 of the 98 neutron activation from Zambujal, Penedo, Fórnea, Leceia and Vila Nova analyses failed due to technical problems. The results of the de São Pedro, which are all Chalcolithic settlements XRF analyses were used in this case instead (on comparabi­ of the Estremadura (98 analyses).14 The elements ar- lity of XRF and NAA measurements see: Lutz/Pernicka 1996; Müller/Cardoso 2008; Müller in press). Log concentrations of selected trace element concentrations were classified using average-link cluster analysis. Element concentrations that 14 The 98 analyses contained 29 replicate analyses of Zambujal were below the detection limit were given artificial values of artefacts sampled by the SAM project. In this case, the older factor of 10 smaller than the respective detection limit. Chemical Analyses in Archaeometallurgy: A View on the Iberian Peninsula 303

Partial overlaps exist, because the diagrams represent only two dimensional sections of a four dimensional space. In order to test Sangmeisters idea of a chrono- logical significance of the individual metal types, the material groups found along the dashed line were re- lated to the Zambujal stratigraphy (fig. 6), i.e. all the Zambujal signatures of the studied data set were re- lated to the occupation phase of the settlement. The result is in good agreement with the observations by Sangmeister in that the arsenic-only copper domi- nates clearly and that the percentage of pure copper occurs is slightly higher in the earlier phases of the settlement. The only difference is that arsenical cop- per with higher concentrations in antimony, silver and/or nickel (representing the two material groups of ‘impure’ arsenical copper) occurs in low quanti- ties throughout the occupation at Zambujal; an in- crease in the use of this – according to Sangmeister – ‘younger’ arsenical copper does not seem to be identifiable. Whether this is a site-specific phenom- enon or whether this is due to the smaller size of the data set remains unanswered at present. The study reveals that the principal material groups of Chalcolithic south central Portuguese cop- per as defined by Sangmeister can also be found using cluster analysis of modified data sets including new analytical data. The chronological significance of these groups referring to the Zambujal stratigraphy is similar but slightly different to the one proposed by Sangmeister, which was based on a more general Portuguese Chalcolithic artefact typology.

Conclusions Fig. 5: The material groups of the cluster analysis plotted in ele- In summary, this work has shown that the results of ment diagrams (cf. fig. 4). Pure copper can be recognized clearly in the As vs. Sb diagram. The other groups show varying Sb contents; different analytical series of prehistoric Iberian cop- yet, they can be discriminated by looking at the Ag-Ni-diagram. per-based artefacts are comparable with each other. The white circular data point represents an ‘exotic’ metal: arseni- The key problem is to identify the appropriate level cal copper with traces of Sb, Ag, Ni, and even Sn. The linear ar- of discrimination, when different data sets are com- rangements of some data reflect the varying detection limits of OES, NAA, and XRF analyses; in this case, an assumed value of pared. The conclusions drawn from early qualitative factor 10 below the respective detection limit was used. Error bars analyses are neither contradicted by later ones nor of ±20 % are shown as a general orientation for the uncertainty by quantitative analyses produced during the last of the analyses; the differences between the laboratories may be decades. Nevertheless, although first comparative larger in individual cases (see figs. 2 and 3). Concentrations are given in percent. examinations are very promising, it is of fundamen- tal importance to conduct a large-scale systematic comparison of the quantitative data published by trations of each copper artefact. The agglomeration the major projects, i.e. ‘Studien zu den Anfängen der distance plot led to the formation of 14 clusters. How- Metallurgie’ (SAM), the ‘British Museum’ (BM), the ever, for the current discussion on evaluating Sang- ‘Proyecto de Arqueometalurgia’ (PA), and the ‘Zam- meisters hypothesis it is remarkable to see that his bujal Project’. Such interlaboratory comparisons are groups emerge, if the dendrogram is cut off at the very rare in archaeometallurgy (e.g. Oddy 1972; Chase dashed line, i.e. on a less resolved level (fig. 4). The 1974), but necessary to allow better evaluation of data four groups and one outlier are also clearly recogniz- quality and to find the appropriate discrimination able in diagrams of elemental concentrations (fig. 5). level for artefact classification. 304 Roland Müller – Ernst Pernicka

Fig. 6: a. Diagram showing the distribution of the four major types of copper in relation to the Zambujal stratigraphy and the number of analyses conducted on the material from each occupation phase. The groups of arsenical copper with Sb and Ag and arsenical copper with Ni (+Sb, Ag) were merged together for this graph (cf. Sangmeister 1995: 73). Some of the finds could not be associated clearly with a certain occupation phase; thus, the level of accuracy in allocating objects to certain phases varies. b. The Zambujal occupation phases as reflected by the stratigraphy of the excavation area VX, located between fortification lines 1 and 2. Traces of the last occupation phase – phase 5 – were found on top of a phase 4 destruction layer; they are not shown here. Photo: H. Schubart 1972, German Archaeological Institute. Chemical Analyses in Archaeometallurgy: A View on the Iberian Peninsula 305

Acknowledgements Junghans et al. 1954 S. Junghans/H. Klein/E. Scheufele, Untersuchungen zur We would like to thank Dr. Thorsten Schifer who con- Kupfer- und Frühbronzezeit Süddeutschlands. Bericht der Römisch-Germanischen Kommission 34, 1951/53 (1954), ducted the neutron activation analyses of the Zambu- 77–114. jal project and assisted in producing figure 4. We are Junghans et al. 1960 also very grateful to Dr. Michael Kunst, who provided S. Junghans/E. Sangmeister/M. Schröder, Metallanalysen figure 6b. kupferzeitlicher und frühbronzezeitlicher Bodenfunde aus Europa. Studien zu den Anfängen der Metallurgie. Berlin: Verlag Gebr. Mann 1960. Junghans et al. 1968 S. Junghans/E. Sangmeister/M. Schröder, Kupfer und Bronze in der frühen Metallzeit Europas 1–3. Studien zu References den Anfängen der Metallurgie. Berlin: Verlag Gebr. Mann 1968. Budd/Ottaway 1991 Junghans et al. 1974 P. Budd/B. S. Ottaway, The Properties of Arsenical Copper S. Junghans/E. Sangmeister/M. Schröder, Kupfer und Alloys: Implications for the Development of Eneolithic Met- Bronze in der frühen Metallzeit Europa. Studien zu den allurgy. In: P. Budd/B. Chapman/C. Jackson/R. Janaway/B. Anfängen der Metallurgie 4. Berlin: Verlag Gebr. Mann S. Ottaway (eds.), Archaeological Sciences 1989. Oxbow 1974. Monograph 9. Oxford: Oxbow 1991, 132–142. Klaproth 1815 Butler/van der Waals 1966 M. H. Klaproth, Beyträge zur chemischen Kenntniß der J. J. Butler/J. D. van der Waals, Bell Beakers and Early Met- Mineralkörper (1795 bis 1815), 6 Bände. Berlin: Rottmann alworking. Palaeohistoria 12, 1966, 41–139. 1815. Cardoso/Guerra 1997/98 Kuleff/Pernicka 1995 J. L. Cardoso/M.F. Guerra, Análises químicas não destru- I. Kulef/E. Pernicka, On the INAA of Native Copper: Some tivas do espólio metálico do povoado pré-histórico de Le- Methodological Considerations. Journal of Radioanalytical ceia, Oeiras e seu significado no quadro da intensificação and Nuclear Chemistry 191, 1995, 143–159. económica calcolítica da Estremadura Portuguesa. Estudos Lutz/Pernicka 1996 Arqueológicos de Oeiras 7, 1997/98, 61–87. J. Lutz/E. Pernicka, Energy Dispersive X-ray Fluorescence Cartailhac 1886 Analysis of Ancient Copper Alloys: Empirical Values for E. Carthailac, Les Áges prehistoriques de l’Espagne et du Precision and Accuracy. Archaeometry 38, 1996, 313–323. Portugal. Paris: Reinwald Librairie 1886. Montero 1994 Chase 1974 I. Montero, El origen de la metalurgia en el sudeste de la W. T. Chase, Comparative Analysis of Archaeological Bron­ Penínbsula Ibérica. Almería: Instituto de Estudios Alm- zes. In: C. W. Beck (ed.), Archaeological Chemistry. Advanc- erienses 1994. es in Chemistry Series 138. Washington: American Chemi- Montero et al. 1990 cal Society 1974, 148–185. I. Montero/S. Rodríguez/J. Rojas, Arqueometalurgia de la Delibes/Montero 1999 Provincia de Toledo: Minería y recursos minerales de co- G. Delibes/I. Montero (eds.), Las primeras etapas metalúr- bre. Toledo: Servicio de Arqueología 1989. gicas en la Peninsula Iberica II. Estudios regionales. Ma- Müller in press drid: Instituto Ortega y Gasset 1999. R. Müller, Zambujal und die Anfänge der Metallurgie in Estacio da Veiga 1889 der Estremadura (Portugal). Technologie der Kupferge- S. P. M. Estacio da Veiga, Antiguidades Monumentaes do winnung, Herkunft des Metalls und soziokulturelle Be- Algarve. Tempos Prehistoricos. Volume I. Lisbon: Imprensa deutung der Innovation. Iberia Archaeologica 15,2. Mainz: Nacional 1889. Zabern. Gale et al. 1985 Müller et al. 2004 N. H. Gale/A. Papastamataki/Z. A. Stos-Gale/K. Leonis, Co- R. Müller/Th. Rehren/S. Rovira, Almizaraque and the Ear- pper Sources and Copper Metallurgy in the Aegean Bronze ly Copper Metallurgy of Southeast Spain: New Data. Ma- Age. In: P. T. Craddock/M. J. Hughes (eds.), Furnaces and drider Mitteilungen 45, 2004, 33–56. Smelting Technology in Antiquity. British Museum Occa- Müller et al. 2007 sional Papers 48. London: British Museum 1985, 81–93. R. Müller/G. Goldenberg/M. Kunst/M. Bartelheim/E. Per- Gonçalves 1989 nicka, Zambujal and the Beginnings of Metallurgy in South- V. Gonçalves, Megalitismo e metalurgia do Alto Algave ern Portugal. In: S. La Niece/D. Hook/P. T. Craddock (eds.), Oriental: uma aproximação integrada. Lisbon: Imprensa Metals and Mines. Studies in Archaeometallurgy. Selected Nacional 1989. papers from the conference ‘Metallurgy: a Touchstone for Härke 1978 Cross-cultural Interaction’. British Museum 28–30. May H. Härke, Probleme der optischen Emissionsspektralanaly- 2004. London: Archetype Publications 2007, 15–26. se in der Urgeschichtsforschung. Prähistorische Zeitschrift Müller/Cardoso 2008 53, 1978, 165–276. R. Müller/J. L. Cardoso, The Origin and Use of Copper at Harrison/Craddock 1981 the Chalcolithic Fortification of Leceia, Portugal. Madrider R. J. Harrison/P. T. Craddock, A Study of the Bronze Age Mitteilungen 49, 2008. Metalwork from the Iberian Peninsula in the British Mu- Müller/Soares 2008 seum. Ampurias 43, 1981, 113–179. R. Müller/A. M. M. Soares, Traces of Early Copper Produc- Hughes et al. 1976 tion at the Chalcolithic Fortification of Vila Nova de São M. J. Hughes/M. R. Cowell/P. T. Craddock, Atomic Absorp- Pedro, Portugal. Madrider Mitteilungen 49, 2008. tion Techniques in Archaeology. Archaeometry 18, 1976, Needham 2002 19–38. S. Needham, Analytical Implications for Beaker Metallurgy Hünefeld/Picht 1827 in North-West Europe. In: M. Bartelheim/E. Pernicka/R. L. Hünefeld/F. Picht, Rügens metallische Denkmäler der Krause (eds.), Die Anfänge der Metallurgie in der alten Vorzeit, vorzugsweise chemisch bearbeitet, und als Beytrag Welt. Forschungen zur Archäometrie und Altertumswis- zur vaterländischen Altertumskunde. Leipzig: Voss 1827. senschaft 1. Rahden/Westf.: VML 2002, 99–133. 306 Roland Müller – Ernst Pernicka

Oddy 1972 tional: Société Préhistorique France, Memoire 37. Rennes: W. A. Oddy, The Analysis of Gold Coins – a Comparison of Pôle éditorial archéologique de l’Ouest 2005, 19–25. Results obtained by non-destructive Methods. Archaeo­ Sangmeister in press metry 14, 1972, 109–117. E. Sangmeister, My path to Zambujal. In: M. Kunst (ed.), Ottaway 1974 Bell-Beakers. Symbols of a 5000-year-old cultural commu- B. Ottaway, Cluster Analysis of Impurity Patterns in Amor- nity in Europe. Exhibition catalogue, Museum Torres Ve- ico-British daggers. Archaeometry 16, 1974, 221–231. dras. Torres Vedras: Câmara Municipal. Ottaway 1982 Sangmeister/Schubart 1981 B. Ottaway, Earliest Copper Artifacts of the Northalpine Re- E. Sangmeister/H. Schubart, Zambujal. Die Grabungen gion: their Analysis and Evaluation. Schriften des Seminar 1964 bis 1973. Madrider Beiträge. Mainz: Zabern 1981. für Urgeschichte 7. Bern: Seminar für Urgeschichte 1982. Schmitz 2004 do Paço 1955 A. Schmitz, Typologische, chronologische und paläomet- A. do Paço, El Castro de Vila Nova de S. Pedro (VII). Con- allurgische Untersuchungen zu den frühkupferzeitlichen siderações sobre o problema da metalurgía. Zephyrus 6, Kupferflachbeilen und Kupfermeißeln in Alteuropa. Dis- 1955, 27–40. sertation. Saarbrücken: Philosophische Fakultät der Uni- Pernicka 1984 versität des Saarlandes 2004. E. Pernicka, Instrumentelle Multi-Elementanalyse archäo- Siret/Siret 1887 logischer Kupfer- und Bronzeartefakte: Ein Methodenver- H. Siret/L. Siret, Les premiers âges du métal dans le Sud- gleich. Jahrbuch des Römisch-Germanischen Zentralmuse- Est de l’Espagne. Antwerb: Charles Peeters 1887. ums Mainz 31, 1984, 517–531. Soares et al. 1994 Pernicka 1986 A. M. M. Soares/M. d. F. Araujo/J. M. P. Cabral, Vestígios da E. Pernicka, Archäometrie: 200 Jahre Anwendung analyti- práctica de metalurgia em povoados Calcolíticos da bacia scher Methoden in der Archäometrie. Kultur & Technik 3, do Guadiana, entre o Ardila e o Chança. In: J. Campos/J. 1986, 180–191. Perez/F. Gomez (eds.), Arqueología en el entorno del Bajo Pernicka 1987 Guadiana. Actas del encuentro internacional de arqueo- E. Pernicka, Erzlagerstätten in der Ägäis und ihre Ausbeu- logia del suroeste. Huelva: Universidad de Huelva 1994, tung im Altertum. Geochemische Untersuchungen zur 165–200. Herkunftsbestimmung archäologischer Metallobjekte. Soares et al. 1996 Jahrbuch des Römisch-Germanischen Zentralmuseums A. M. M. Soares/M. d. F. Araujo/L. Alves/M. T. Ferraz, Ves- Mainz 34, 1987, 607–714. tígios metallúrgicos em contextos do Calcolítico e da Idade Pernicka 1990 do Bronze no sul de Portugal. In: M. J. Maciel (ed.), Mis- E. Pernicka, Gewinnung und Verbreitung der Metalle in cellanea em Homenagem ao Professor Bairrão Oleiro. Lis- prähistorischer Zeit. Jahrbuch des Römisch-Germanischen boa: Colibri 1996, 553–579. Zentralmuseums 37, 1990, 21–129. Sousa et al. 2004 Rovira 2002 A. C. Sousa/P. Valério/M. d. F. Araújo, Metalurgia antiga do S. Rovira, Early Slags and Smelting By-Products of Cop- Penedo do Lexim (Mafra): Calcolítico e Idade do Bronze. per Metallurgy in Spain. In: M. Bartelheim/E. Pernicka/R. Revista Portuguesa de Arqueologia 7, 2004, 97–117. Krause (eds.), Die Anfänge der Metallurgie in der alten Trigger 2000 Welt. Forschungen zur Archäometrie und Altertumswis- B. Trigger, A History of Archaeological Thought. Cam- senschaft 1. Rahden/Westf.: VML 2002, 83–98. bridge: Cambridge University Press 2000. Rovira 2005 S. Rovira, Tecnología metalúrgica Campaniforme en la Península Ibérica. Coladas, moldeado y tratamientos postfundición. In: M. A. Rojo-Guerra/R. Garrido-Pena/I. García-Martínez de Lagrán (eds.), El Campaniforme en la Península Ibérica y su Contexto Europeo. Valladolid: Junta de Castilla y León 2005, 495–521. Rovira et al. 1997 S. Rovira/I. Montero/S. Consuegra, Las primeras etapas metalúrgicas en la Península Ibérica, I. Análisis de materi- ales. Madrid: Instituto Ortega y Gasset 1997. Rovira/Ambert 2002 S. Rovira/P. Ambert, Les céramiques à réduire le minerai de cuivre: une technique métallurgique utilisée en Ibérie, son extension en France méridionale. Bulletin de la Société Préhistorique Française 99, 2002, 105–126. Rovira/Gómez 2003 S. Rovira/P. Gómez, Estudios Metalográficos. Las primeras etapas metalúrgicas en la Península Ibérica. Madrid: Insti- tuto Ortega y Gasset 2003. Sangmeister 1995 E. Sangmeister, Zambujal. Kupferfunde aus den Grabungen 1964 bis 1973. In: E. Sangmeister/M. d.l. C. Jiménez Gómez, Zambujal. Kupferfunde aus den Grabungen 1964 bis 1973. Los Amuletos de las Campaños 1964 hasta 1973. Madrider Beiträge 5. Mainz: Zabern 1995, 1–153. Sangmeister 2005 E. Sangmeister, Les débuts de la métallurgie dans le sud- ouest de l’Europe: l’apport de l’etude des analyses métal- lographiques. In: P. Ambert/J. Vaquer (eds.), La première métallurgie en France et dans les pays limitrophes. Carcas- sonne, 28–30 Septembre 2002. Actes du Colloque Interna-

View publication stats