Complementary use of X-ray methods to study ancient production remains and metals from Northern Portugal

Research article Received: 11 December 2013 Revised: 19 February 2014 Accepted: 4 March 2014 Published online in Wiley Online Library (wileyonlinelibrary.com) DOI 10.1002/xrs.2541 Complementary use of X-ray methods to study ancient production remains and metals from Northern Portugal Pedro Valério,a* M. Fátima Araújoa and Rui J.C. Silvab Archaeological artefacts recovered at Castanheiro do Vento (Northern Portugal) were characterised by integrating macro and micro-energy dispersive X-ray fluorescence spectrometry (EDXRF) and scanning electron microscopy with X-ray microanalysis. The collection includes metallurgical remains (ceramic crucibles, a metallic nodule and a vitrified fragment) and metals (tools and ornaments) whose chronology spans from the Chalcolithic to the Roman Age. The study of production remains was able to identify distinct copper-based metallurgical operations including the smelting of copper ores, the melting of copper and tin and/or the melting of bronze scrap. Micro-EDXRF identified copper and arsenical copper tools as well as bronze and leaded bronze ornaments. The composition of tools (Cu with varying As contents: 0.46–3.6%) reveals an incipient technology, typical of the Chalcolithic till the Middle Bronze Age. On the contrary, ornaments are composed by different alloys – low tin bronze (4.8% Sn), high tin bronze (14.9% Sn) and high tin-leaded bronze (16.5% Sn and 2.4% Pb) evidencing technological and economic choices that clearly indicate a late period such as the Roman Age. In conclusion, this multiproxy approach was able to study those ancient artefacts with a minimum impact on their archaeological and museological significance while providing important answers to the interpretation of the archaeological settlement and to better understand the metallurgical evolution in the Portuguese territory. Copyright © 2014 John Wiley & Sons, Ltd. Introduction Metals have played a central role in daily life of ancient societies, sometimes being the driving force behind significant cultural and economic amendments. Their study is therefore fundamental not only to understand the technology behind the production of artefacts but also to better comprehend the interactions of ancient societies. An interesting collection of such materials was recovered during several archaeological campaigns at the site of Castanheiro do Vento (Vila Nova de Foz Côa, Northern Portugal).[1] Some of the best preserved structures (houses and defensive walls) were radiocarbon dated establishing a wide occupation period of 2900–1500 cal BC. Other structures were dated to about 800–300 cal BC showing the reutilisation of the prehistoric ruins during the Iron Age. Unfortunately, the metals and related production remains could not be directly associated with dated contexts. Moreover, the typological features of some of the metals clearly evidence a late chronology such as the Roman Age. It was therefore important to establish the composition of metals and the origin of production remains because the comparison with known metallurgical data can ascertain a broad chronology to the materials and allows an improved interpretation of the archaeological settlement. An integrated approach mostly on the basis of X-ray methods was then selected to study those unique cultural artefacts. Conventional energy dispersive X-ray fluorescence spectrometry (EDXRF) provided an efficient preliminary characterisation, mostly because of its multi-elemental, accurate and fast nature. However, due to the relatively low penetration of the X-rays, it is strongly affected by the superficial corrosion layer usually present on archaeological metals. Micro-EDXRF analyses have added a vital feature to the study, as it became possible to X-Ray Spectrom. (2014) interact with only a very small area of the artefact (i.e. only a minute area was cleaned for quantitative analysis). Moreover, scanning electron microscopy with X-ray microanalysis (SEM-EDS) allowed the characterisation of selected artefacts to an even smaller scale, identifying different phases and alteration products. The efficiency of the manufacture of metals can also be indicative of their chronology and use, and thus, it was also assessed by reflected light microscopy and Vickers microhardness testing. Overall, this study highlights the importance of the complementary use of X-ray methods to extract significant evidences from archaeological artefacts while giving a better understanding about the metallurgical evolution in the Portuguese territory. Materials and methods Castanheiro do Vento is a monumental enclosure located at the top of a small hill and contains several subcircular structures bounded by three lines of defensive walls. Metals recovered there comprise tools and ornaments. The first type includes three * Correspondence to: Pedro Valério, Centro de Ciências e Tecnologias Nucleares, Instituto Superior Técnico, Universidade de Lisboa, Estrada Nacional 10 (km 139,7), 2695-066 Bobadela LRS, Lisboa, Portugal. E-mail: pvalerio@ctn.ist.utl.pt a Centro de Ciências e Tecnologias Nucleares, Instituto Superior Técnico, Universidade de Lisboa, Estrada Nacional 10 (km 139,7), 2695-066 Bobadela, Lisboa, Portugal b CENIMAT/I3N, Departamento de Ciência dos Materiais, Faculdade de Ciências e Tecnologia, FCT, Universidade Nova de Lisboa, 2829-516 Monte de Caparica, Portugal Copyright © 2014 John Wiley & Sons, Ltd. P. Valério, M. F. Araújo and R. J. Cordeiro Silva awls (M1, M2 and M3), a chisel (M4) and a ‘blade’ (M5), whereas the ornaments comprise an annular fibula (M6), a penannular fibula (M7) and two decorated plaques (M8 and M9) (Fig. 1). There is also a small fragment (M10) with an unknown functionality. These tools have rudimentary typologies that were used by prehistoric communities since the Chalcolithic. Fibulae were used to fasten clothes, and the annular types were very common during the 1st–4th centuries AD due to their robustness and simple design.[2] Penannular fibulae have been disseminated since the 1st century AD, but it is impossible to ascertain a precise chronology due to the huge variety of types. The exemplar from Castanheiro do Vento has a hinge assembly on reverse with remains from an iron catch plate. Finally, the decorated plaques exhibit an incised geometric ornamentation on the front and four rivet holes indicating the attachment to a support material, such as leather or wood. Production remains from Castanheiro do Vento include three ceramic crucibles (R1, R2 and R3) with similar fabrics and relatively thin walls (~1 cm). Crucibles inner surfaces exhibit a thin slag layer and several greenish nodules of variable dimension (microscopic up to about 1 mm). Additionally, there is a roundshaped metallic nodule (R4, ~0.5 cm diameter) and a vitrified fragment (R5, ~3 cm length) resembling pumice stone due to its very porous texture. Archaeological materials were first analysed by EDXRF without any sample preparation. Metals M1, M2, M3, M7 and M9 were polished on a small surface area (∅ ~3–5 mm) with a manual drill and diamond pastes of progressively finer grit size (15–1 μm). Small sections from metals M4, M5, M10 and metallic nodule R4 were cut with a jewellers saw, mounted in epoxy resin and polished with silicon carbide papers (1000, 2500 and 4000 grit sizes) and diamond pastes (3 and 1 μm). Prepared surfaces were analysed by micro-EDXRF and observed by reflected light microscopy. Additionally, sections of metals M4 and M5 mounted in epoxy resin were submitted to Vickers microhardness testing. SEM-EDS analyses were made in sections from crucibles R1 and R3 cut with a rotary saw, polished with silicon carbide papers (1000, 2500 and 4000 grit sizes) and enclosed with carbon conductive bridge to prevent charge accumulation (metallic nodule R4 was also analysed by SEM-EDS). Energy dispersive X-ray fluorescence spectrometry analyses were performed in a Kevex 771 spectrometer (Kevex Instruments, USA) equipped with an Rh X-ray tube, selectable secondary targets and a liquid nitrogen cooled Si(Li) detector with a resolution of 165 eV at 5.9 keV. X-ray tube, secondary target, sample tray and detector have an orthogonal geometry design (triaxial geometry). Moreover, the secondary target changer allows selecting the monochromatic radiation that efficiently excites the sample elements, thus improving the detection limits (Fig. 2). X-ray beam irradiates a nearly circular sample area with a diameter of about 2.5 cm. Gd secondary target (57 kV of tube voltage, 1.0 mA of current intensity, 300 s of live time and Ta collimator) was used for elements with higher K-edges (e.g. Ag, Sn or Sb), while others having lower K or L-edges (e.g. Fe-K, As-K or Pb-L) were excited with the radiation obtained from an Ag secondary target (35 kV, 0.5 mA, 300 s and Ag collimator). Quantification uses experimental calibration factors calculated with reference material Phosphor Bronze 551 (British Chemical Standards, BCS) and relies on fundamental parameters to account for matrix effects. Quantification limits (10 × background1/2/sensitivity) were calculated with BCS Phosphor Bronze 551 being 0.05% Sn, 0.10% Ni, 0.10% Pb, 0.10% As and 0.05% Fe. A detailed description of the equipment and analytical procedure has been previously published.[3] Figure 1. Tools and ornaments from Castanheiro do Vento: awls (M1, M2 and M3), chisel (M4), ‘blade’ (M5), annular fibula (M6), penannular fibula (M7, with scheme of hinge mechanism: (a) pin and (b) remains of iron catch plate) and decorated plaque (M9) (with indication of sampling area; R, sampling on reverse). wileyonlinelibrary.com/journal/xrs Copyright © 2014 John Wiley & Sons, Ltd. X-Ray Spectrom. (2014) Elemental analysis of slag and metal by EDXRF, micro-EDXRF and SEM-EDS Figure 2. Schematic representation of spectrometers Kevex 771 EDXRF and ArtTAX Pro micro-EDXRF. Prepared surfaces were analysed by micro-EDXRF using an ArtTAX Pro spectrometer (Bruker, Germany) equipped with a Mo X-ray tube and a silicon-drift electrothermally cooled detector with a resolution of 160 eV at 5.9 keV. Focusing polycapillary lens are able to produce a nearly circular area of primary radiation with a diameter of about 70 μm.[4] The area to be analysed is selected by a charge-coupled device (CCD) camera and the optical triangulation of three positioning diodes (Fig. 2). Analyses were made with 40 kV of tube voltage, 0.5 mA of current intensity and 100 s of live time. Samples were analysed on three spots to account for possible heterogeneities. Calibration involved the WinAxil software comprising fundamental parameters and experimental factors calculated with BCS Phosphor Bronze 551. Accuracy was estimated with BCS Phosphor Bronze 552 and Industries de la Fonderie (IDLF) 5 (Paris, France) (Table 1). Quantification limits from elements of interest on Cu–Sn matrix are 0.5% Sn, 0.10% Ni, 0.10% Pb, 0.10% As and 0.05% Fe. Micro-EDXRF and EDXRF show analogous sensitivities for most elements except for Sn. The worse sensitivity of micro-EDXRF comes from the lower fluorescence yield of the Sn-Lα energy peak compared with the Sn-Kα energy peak used in EDXRF. Scanning electron microscopy with X-ray microanalysis was made with a Zeiss DSM 962 electron microscope (Zeiss, Germany) with secondary electron and backscattered electron imaging modes. Elemental analyses involved an Oxford Instruments INCAx-sight spectrometer (Oxford Instruments, United Kingdom) with a Si(Li) detector (133 eV at 5.9 keV) having an ultrathin window to detect low Z elements such as oxygen and carbon. Experimental conditions were 20 kV of accelerating voltage, ~3 A of filament current, 70 μA of emission current and 25 mm of working distance. Pure compounds commonly present in phases and/or alterations products of archaeological copper-based artefacts (e.g. cuprite and malachite) were analysed to infer about their presence in studied artefacts. Identification was based on the comparison of the main lines in EDS spectra. Reflected light microscopy observations were carried out with a Leica DMI 5000 M microscope (Leica Microsystems, Germany). Samples were etched with an aqueous ferric chloride solution. Vickers microhardness testing was made in a Zwick-Roell Indentec equipment (Zwick Roell AG, United Kingdom) using an indentation of 0.2 kg for 10 s. To obtain a relative standard deviation less than 5% , at least three indentations were made on each sample. Results and discussion Production remains Crucibles were first analysed by EDXRF because the large analysis area (∅~2.5 cm) of the Kevex spectrometer allows a fast preliminary characterisation of this type of highly heterogeneous samples. Analyses of inner and outer surfaces from crucibles showed that the slags present inside these crucibles are enriched in Cu, As and Sn (Fig. 3). The significant contents of Cu and Sn in crucible R1 are secure evidence that this vessel was used to produce bronze. However, the minor amounts of tin in crucibles R2 and R3 are not conclusive to determine the production of copper or bronze because the use of copper ores with small amounts of tin has sometimes been attested in the archaeological record from Iberia (see for instance[5]). Pre-Roman bronze in Iberia was produced in ceramic crucibles by one of the following methods: (1) the direct reduction of copper and tin ores (co-smelting), (2) the cementation of metallic copper with tin ore (mostly cassiterite) or (3) the melting of metallic copper with tin. Alternatively, bronze scrap could be melted to obtain a new bronze ingot. To better investigate the usage of the crucibles from Castanheiro do Vento, prepared samples were further analysed by SEM-EDS. A large metallic nodule (0.5 cm length) inside the slag from crucible R1 revealed an α–Cu matrix with abundant Cu2O eutectic (Fig. 4). This eutectic indicates a high concentration of oxygen in molten copper, which is common when a deoxidizing element, such as tin or arsenic, is absent from the metallic bath. The lack of tin in this copper nodule, despite being present in the slag (EDXRF analysis has identified Sn), shows that the alloy was being produced from distinct sources of copper and tin instead of bronze scrap. Table 1. Micro-EDXRF analysis of BCS 552 and IDLF 5 (average ± SD of three analyses; values in %) BCS 552 Cu Sn Pb As Fe Ni Zn Certified Obtained Accuracy 87.7 88.1 ± 0.2 0.4% 9.78 10.0 ± 0.2 2% 0.63 0.62 ± 0.05 2% n.d. n.d. — 0.10 0.11 ± 0.02 10% 0.56 0.63 ± 0.02 12% 0.35 0.49 ± 0.01 40% Cu Sn Pb As Sb Ni Zn 68.5 69.1 ± 0.3 0.9% 19.9 19.8 ± 0.4 0.5% 1.42 1.46 ± 0.09 3% 5.75 5.70 ± 0.08 0.9% 2.23 2.19 ± 0.16 2% 0.67 0.57 ± 0.16 15% 0.94 0.90 ± 0.08 4% IDLF 5 Certified Obtained Accuracy EDXRF, energy dispersive X-ray fluorescence spectrometry; BCS, British Chemical Standards; SD, standard deviation; n.d., not detected. X-Ray Spectrom. (2014) Copyright © 2014 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/xrs P. Valério, M. F. Araújo and R. J. Cordeiro Silva Figure 3. EDXRF spectra of inner and outer surfaces of crucibles R1, R2 and R3 from Castanheiro do Vento (Gd secondary target excitation, 57 kV, 1.0 mA and 300 s; dotted circle corresponds approximately to analysed area; crucible length is 5.6, 4.2 and 3.6 cm, respectively). Figure 4. EDS spectra of copper (α-Cu), Cu2O eutectic, copper (oxidised) and slag in crucibles R1 and R3 from Castanheiro do Vento (SEM-BSE images showing (a) metallic nodule from crucible R1 and (b) slag from crucible R3). Moreover, the thin slag layer inside crucible R1 suggests a reduced interaction of the metal and oxide melts with the crucible fabrics. This feature is consistent with a melting operation because the reduction of metallic ores originates a more complex and heterogeneous slag layer. Consequently, analytical evidences indicate that crucible R1 was most likely used to produce a bronze alloy by the melting of metallic copper and tin. The archaeological record from the Iberian Peninsula suggests that the usage of this method of bronze production began during the Iron Age.[6] Scanning electron microscopy with X-ray microanalysis analyses of slag in crucible R3 identified an aluminium silicate matrix with numerous Cu nodules of variable dimension (Fig. 4). The absence of certain compounds that usually formed during the smelting of copper ores (e.g. relic minerals, copper sulphides, magnetite and delafossite[7]) suggests a melting operation. In this case, the reduced Sn content established by EDXRF analysis could wileyonlinelibrary.com/journal/xrs point to the melting of a low tin bronze. Nevertheless, it must be taken into account that the thermochemical behaviour of metals is not straightforward, depending upon temperature, oxygen pressure and time of operation of the process, plus volatility, free energy of oxidation and reactivity of different metals with charcoal and crucible material.[8] Micro-EDXRF analysis of the metallic nodule R4 identified a copper matrix with 0.96% As and low amounts of Fe (<0.05%). Additionally, it shows a coarse microstructure with numerous and large pores indicating a strong degassing (Fig. 5). This type of microstructure is typical of smelting nodules slowly cooled inside the crucible. SEM-EDS analysis identified cuprite and malachite inside some of those pores, in this case, representing secondary alteration products formed during the burial period (Fig. 5). SEM-EDS analysis also identified very small Pb-rich inclusions (<5 μm) containing Se and Mn among the α–Cu matrix. These elements are usually present in archaeological copper artefacts Copyright © 2014 John Wiley & Sons, Ltd. X-Ray Spectrom. (2014) Elemental analysis of slag and metal by EDXRF, micro-EDXRF and SEM-EDS Figure 5. EDS spectra of copper matrix (α–Cu), secondary cuprite (cup) and malachite (mal), plus Pb-rich inclusion in nodule R4 from Castanheiro do Vento (copper in Pb-rich inclusion is mostly from the surrounding α–Cu matrix; bottom right, SEM-BSE image showing degassing pore filled with secondary alteration products; nodule length is 0.9 mm). at very low levels. For instance, Chalcolithic coppers from Eastern Mediterranean region have Se in the range of 10 μg/g or even lower.[9] Ores and fluxes are capable of introducing Mn, but it is mostly accumulated in the slag.[10] The copper nodule R4 should therefore be a product from a primary operation involving copper ores, as this type of elements is commonly absent in metal from latter metallurgical processes, such as refining, melting or casting. Moreover, the low amount of iron suggests a poor reducing atmosphere during the smelting, which is typical of pre-Phoenician metallurgical operations.[11] Energy dispersive X-ray fluorescence spectrometry analyses of the vitrified fragment R5 did not found traces of metals apart from Fe. Moreover, this fragment has a very porous morphology that is similar to the bloom, often called sponge iron, obtained by the smelting of Fe minerals with the bloomery process.[11] Metals Preliminary EDXRF analyses have established that awls, chisel, blade and fragment are mostly composed of Cu, whereas the major elements in fibulae and plaques are Cu and Sn. The typology, decoration and conservation state of the two decorated plaques are identical. Moreover, both EDXRF spectra are similar, and so they probably share the same alloy composition. Thus, only the plaque M9 was subjected to the cleaning procedure necessary for the following microanalyses. The cleaned metal surfaces of tools and ornaments were analysed by micro-EDXRF being composed of copper, copper with arsenic, bronze or leaded bronze (Fig. 6). Tools and fragment are composed of copper with variable arsenic contents (0.46–3.6%) and very low iron amounts (Table 2). The low iron content of copper-based artefacts results from X-Ray Spectrom. (2014) excessively high pO2 during smelting, and it can be used to identify the pre-Phoenician metallurgy from this region.[12] Moreover, studies concerning the ancient copper-based metallurgy in this western end of Iberia have shown that copper and arsenical copper alloys (As > 2%) are typical of the Chalcolithic to Middle Bronze Age (3000–1200 BC). Some examples can be found in the Middle Bronze Age necropolis of Monte da Cabida 3 and Torre Velha 3[13] and in Chalcolithic fortified settlements, such as Leceia,[14] Vila Nova de São Pedro[15] and Zambujal.[16] Additionally, it is remarkable to find out that at Castanheiro do Vento, as well as on these settlements, elongated awls present higher arsenic contents. The manufacturing procedure and hardness of artefacts do not seem to vary significantly with the arsenic content. Therefore, the selection of alloys for specific typologies probably would be made because of the silvery colour of alloys rich in arsenic. The composition of these plain tools from Castanheiro do Vento relates well with the prehistoric occupation of the site during 2900–1500 cal BC. This early chronology also agrees with the manufacturing procedures identified by reflected light microscopy. Hammering and annealing operations were used to give the final shape to the artefact but have failed in improving its mechanical properties. For instance, the blade’s cutting edge presents a low hardness increase (body, 93 ± 8 HV0.2; cutting edge, 111 ± 2 HV0.2) despite having being sharpened by deformation (i.e. having smaller grains). Another example is the low hardness of the chisel (65 ± 2 HV0.2), a type of instrument whose usage would certainly benefit from a higher toughness. Contrary to those prehistoric copper tools, the ornaments from Castanheiro do Vento are composed by distinct bronze alloys (Table 2). The impurity pattern varies and includes Pb, As, Fe and Ni. The annular fibula has a low Sn content (4.8%) indicating the usage of a poor bronze alloy to manufacture this simple and inexpensive ornament. Compositions of the disc fibula and Copyright © 2014 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/xrs P. Valério, M. F. Araújo and R. J. Cordeiro Silva Figure 6. Micro-EDXRF spectra of awls (M1 and M3), penannular fibula (M7) and decorated plaque (M9) from Castanheiro do Vento. Table 2. Micro-EDXRF results of metals from Castanheiro do Vento (average ± SD of three analyses; values in %) Artefact M1 M2 M3 M4 M5 M6 M7 M9 M10 Type Reference Cu Sn Pb As Fe Awl Awl Awl Chisel ‘Blade’ Annular fibula Penannular fibula Decorated plaque Fragment Q115.43/C3 Q95.53/C3 Q113.32/C3 Q84.24/C3 Q112.44/Sup Q108.39/C3 Q101.25/C3 Q111.43/C3 Q87.28/C3 96.4 ± 0.5 99.1 ± 0.1 99.5 ± 0.1 98.1 ± 0.1 98.7 ± 0.2 94.8 ± 0.4 81.0 ± 1.1 84.6 ± 0.1 98.7 ± 0.2 n.d. n.d. n.d. n.d. n.d. 4.8 ± 0.3 16.5 ± 0.9 14.9 ± 0.2 n.d. n.d. n.d. n.d. n.d. n.d. 0.32 ± 0.09 2.4 ± 0.2 0.49 ± 0.04 n.d. 3.6 ± 0.4 0.81 ± 0.07 0.46 ± 0.12 1.9 ± 0.1 1.3 ± 0.2 <0.10 n.d. n.d. 1.3 ± 0.2 <0.05 <0.05 <0.05 <0.05 <0.05 <0.05 0.08 ± 0.01 0.09 ± 0.01 <0.05 Ni n.d. n.d. n.d. n.d. n.d. n.d. n.d. <0.10 n.d. EDXRF, energy dispersive X-ray fluorescence spectrometry; SD, standard deviation; n.d., not detected. decorated plaque also show a deliberate selection of alloys. The higher Pb amount of disc fibula (2.4%) results in a better castability, necessary to the proper casting of this relatively large but very thin artefact (reflected light microscopy observation indicates that the fibula was left as-cast, so there was no post-casting thickness reduction). The high tin content of the disc fibula and decorated plaque (16.5% and 14.9%, respectively) also contributes to a higher castability. Nevertheless, the effect of tin wileyonlinelibrary.com/journal/xrs on colour was probably the pursued property. Recent research[17] established that a bronze with ~16% Sn has a golden colour, which will be certainly valued for prestige ornaments. Bronze artefacts are common in Northern Portugal since the Late Bronze Age (1200–800 BC),[18,19] but leaded bronzes only appear during the Early Iron Age (800–500 BC).[12] The Roman Age metallurgy in this region is still poorly studied, but other areas also in the periphery of the Roman Empire, such as Copyright © 2014 John Wiley & Sons, Ltd. X-Ray Spectrom. (2014) Elemental analysis of slag and metal by EDXRF, micro-EDXRF and SEM-EDS Britain[20] or NW Africa,[21] show a diversification of alloys with the use of brass and gunmetal apart from copper and bronze. However, those ornaments from Castanheiro do Vento were all made with bronze indicating a conservative usage of this alloy to manufacture clothing artefacts and prestige items. Conclusions The present work established the importance of using a multiproxy approach based on non-invasive and microanalytical methods to characterise archaeological artefacts. Complementary X-ray methods, namely, EDXRF, micro-EDXRF and SEM-EDS, were associated with microstructural and hardness data to address relations between elemental composition, manufacture, functionality, technology and chronology of ancient metals and production remains. In this particular case, the set of artefacts recovered from the archaeological site of Castanheiro do Vento has a long chronology, enabling an insight of more than 3 millennia of metallurgy in the Northern Portuguese territory. The interpretation of the use of crucibles and other productions remains is always challenging, in this case being hindered by the absence of a known chronology. However, with the exception of a small vitrified fragment, all crucibles and metallic nodule were related with the copper-based metallurgy involving distinct metallurgical operations such as the smelting of copper ores, the melting of copper and tin and/or the melting of bronze scrap. Moreover, production remains present evidences of having been subjected to poor reducing atmosphere, a typical feature of pre-Phoenician metallurgical operations. Copper and arsenical copper rudimentary tools are representative of an incipient metal technology, which perfectly fits into the prehistoric occupation (2900–1500 cal BC) of the fortified settlement. An elongated awl richer in arsenic, such as similar types from other Chalcolithic settlements, suggests the selection of raw materials, probably related with the silvery colour of arsenical coppers. The low hardness of the blade’s cutting edge and a chisel also points to an incipient prehistoric metallurgy, whose artefact manufacture does not always follow the mechanical requirements. Ornaments show quite different elemental compositions including low tin (annular fibula), high tin (decorated plaque) and high tin-leaded bronze (penannular fibula). The usage of different bronze alloys for distinct typologies is quite clear, comprising both technological and economic issues. So it is natural to find an ordinary Roman annular fibula made with an inexpensive alloy or a larger and more prestigious penannular fibula composed of a golden alloy with improved castability. These technological, economic and cultural relations point to more developed and complex societies with a rather later chronology. X-Ray Spectrom. (2014) Acknowledgements The work was carried out in the framework of the project ‘Early Metallurgy in the Portuguese Territory, EarlyMetal’ (PTDC/HIS/ ARQ/110442/2008) financed by the Portuguese Science Foundation. CENIMAT/I3N acknowledges the funding through the Strategic Project - LA 25 - 2013-2014 (PEst-C/CTM/LA0025/2013). The authors acknowledge the Department of Conservation and Restoration (Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa) for the use of the micro-EDXRF spectrometer. Ana M. Vale and Vítor O. Jorge (Faculdade de Letras, Universidade do Porto) are also acknowledged for allowing the study of the artefacts collection. References [1] A. M. Vale, in Archaeology and the Politics of Vision in a Post-Modern Context (Eds: V. O. Jorge, J. Thomas), Cambridge Scholars Publishing, New Castle, 2008, pp. 186–208. [2] S. Ponte, Corpus Signorum das Fíbulas Proto-Históricas e Romanas de Portugal, Caleidoscópio, Casal de Cambra, 2006. [3] P. Valério, M. F. Araújo, A. Canha. Nucl. Instr. Meth. B 2007, 263, 477. [4] H. Bronk, S. Rohrs, A. Bjeoumikhov, N. Langhoff, J. Schmalz, R. Wedell, H. E. Gorny, A. Herold, U. Waldschlager. Fresen. J. Anal. Chem. 2001, 371, 307. [5] S. Rovira, I. Montero, in The Problem of Early Tin (Eds: A. Giumlia-Mair, F. Lo Schiavo), Archaeopress, Oxford, 2003, pp. 15–22. [6] S. Rovira, in Avances en Arqueometría 2005 (Eds: J. Molera, J. Farjas, P. Roura, T. Pradell), Universitat de Girona, Girona, 2007, pp. 21–35. [7] P. Valério, A. M. M. Soares, R. J. C. Silva, M. F. Araújo, P. Rebelo, N. Neto, R. Santos, T. Fontes. J. Archaeol. Sci. 2013, 40, 439. [8] T. Kearns, M. Martinón-Torres, T. Rehren. Historical Metallurgy 2010, 44, 48. [9] A. Hauptman, The Archaeometallurgy of Copper, Berlin, Springer, 2007. [10] R. F. Tylecote, H. A. Ghaznavi, P. J. Boydell. J. Archaeol. Sci. 1977, 4, 305. [11] P. T. Craddock, Early Metal Mining and Production, The University Press, Cambridge, 1995. [12] P. Valério, R. J. C. Silva, M. F. Araújo, A. M. M. Soares, L. Barros. Mater Charact 2012, 67, 74. [13] P. Valério, A. M. M. Soares, M. F. Araújo, R. J. C. Silva, E. Porfírio, M. Serra. J. Archaeol. Sci. 2014, 42, 68. [14] R. Müller, J. L. Cardoso. Madr. Mitteilungen 2008, 49, 64. [15] F. Pereira, R. J. C. Silva, A. M. M. Soares, M. F. Araújo. J. Archaeol. Sci. 2013, 40, 2045. [16] R. Müller, G. Goldenberg, M. Bartelheim, M. Kunst, E. Pernicka, in Metals and Mines. Studies in Archaeometallurgy (Eds: S. L. Niece, D. Hook, P. Craddock), Archetype, London, 2007, pp. 15–26. [17] J. -L. Fang, G. McDonnell, T. Rehren. Historical Metallurgy 2011, 45, 52. [18] E. Figueiredo, P. Valério, M. F. Araújo, R. J. C. Silva, A. M. M. Soares. X-Ray Spectrom. 2011, 40, 325. [19] P. C. Gutiérrez-Neira, A. Zucchiatti, I. Montero-Ruiz, R. Vilaça, C. Bottaini, M. Gener, A. Climent-Font. Nucl. Instr. Meth. B 2011, 269, 3082. [20] D. Dungworth. J. Archaeol. Sci. 1997, 24, 901. [21] E. Gliozzo, W. Kockelmann, L. Bartoli, R. H. Tykot. Nucl. Instr. Meth. B 2011, 269, 277. Copyright © 2014 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/xrs