Journal of Archaeological Science: Reports 16 (2017) 248–257 Contents lists available at ScienceDirect Journal of Archaeological Science: Reports journal homepage: www.elsevier.com/locate/jasrep Bronze Age Caucasian metalwork: Alloy choice and combination a,⁎ Marianne Mödlinger , Benjamin Sabatini a b b MARK IRAMAT-CRP2A - UMR 5060 CNRS, Université Bordeaux Montaigne, Maison de l'archéologie, Esplanade des Antilles, 33607 Pessac, France Department of Materials Science & Engineering, Massachusetts Institute of Technology, 13-5065, 77 Massachusetts Avenue, Cambridge, MA 02139-4307, USA A R T I C L E I N F O A B S T R A C T Keywords: Bronze Age Greater Caucasus Metallurgy Daggers Lead isotope analyses Chemical composition The chemical composition and microstructure of eleven Bronze Age Caucasian daggers from North OssetiaAlana, Russia were studied in order to establish a baseline for metallurgy and alloy production in the region. They have been housed in the Natural History Museum, Vienna since the 1880′s. The assemblage comprises arsenical bronzes characterized by high concentrations of arsenic (2–10 wt%), tin bronzes with high tin (10–17 wt%), and several ternary Cu-As-Sn dagger blades. The chemical composition of the dagger blades was analysed with EDXS and XRF. Furthermore, metallographic analyses and lead isotope analyses were carried out. Two of the arsenical bronze blades showed extreme γ-phase segregation along their surfaces and grain boundaries. Two tin-bronze dagger blades, containing high amounts of eutectoid, prevented the measurements of the hardness of the eutectoid. One dagger combined a tin bronze blade and arsenical bronze hilt. Lead isotope analyses of selected daggers indicate a close relation to copper ore sources in the Greater Caucasus and Armenia. 1. Introduction The aim of this study is to document the microstructure, chemical composition, and lead isotopic signatures of several copper-based daggers from North Ossetia Alana, Russia. These artefacts are listed as having belonged to the Koban culture (phase I) by the Stuttgart-based Studien zu den Anfängen der Metallurgie (SAM) project (Junghans et al. 1968–1974; Krause, 2003), but have no other extant information. Until now, no detailed characterisation of metal objects from the eponymous find spot of Koban has been produced. Those chosen for study were reported by the SAM project as having high As content, ranging between 2 and 5 wt% As. Presented at the 5th Allrussian Congress of Archaeology, September 8–11, 1881 in Tbilisi, Georgia, these daggers and other bronzes from Koban (Fig. 1) were sold to the Naturhistorisches Hofmuseum in Vienna, and other museums, including the Berlin museum and MAN in St. Germain-en-Laye. The daggers studied originated from the Natural History Museum, Vienna, Austria, where they have been housed since the late 19th century, having been purchased by F. Heger, the director of the anthropological and ethnographic department between 1881 and 1893 (Fig. 2). Although purchased, the daggers are known to have been plundered by landowner C. D. Kanukov beginning in 1869 from the Verchni Koban and Chmi cemeteries, both located in the Republic of North OssetiaAlania, Russia. Of the more than 1000 graves (see Heinrich 2006/7, 126), most of the metal artefacts were sold to various museums throughout Europe devoid of context and detailed description. The grave goods associated with the daggers, and gender of the interred, are, however, listed in Table 1 as they appeared on the parcel listing that accompanied the daggers to the museum. Record of the parcel listing comes from the museum inventory book. As a result of their uncertain procurement, much archaeological evidence of the grave sites has been lost, however, several excavations were undertaken by G. D. Filimonov in 1869, V. B. Antonovič in 1871, E. Chantre in 1881, and F. Heger himself during his last visit to the site in 1891. Artefacts sent to the Natural History Museum, Vienna, including those studied in this work, arrived in several parcels between 1883 and 1888 with the following information (after Heinrich 2006/7, 139): 1) Parcel XVIII, 1883: prehistoric finds from Koban and Chmi, ‘excavated’ by C. D. Kanukov; the acquisition of the finds was paid by the consul general of Peru, L. Schiffmann (inv.no.s 41.226–41.441). 2) Parcel XIX, 1883: prehistoric finds from Koban and Chmi, ‘excavated’ by C. D. Kanukov; the acquisition of the finds was paid by the consul general of Peru, L. Schiffmann (inv.no.s 41.442–42.154). 3) Parcel I, 1884: prehistoric finds from Koban (inv.no.s 42.155–42.361). 4) Parcel VI, 1884: prehistoric finds from Koban (inv.no.s 42.362–42.465). 5) Parcel XVIII, 1884: finds from Chmi, excavated by I. Dolbežev in Wladikawkas in 1883. The content of seven graves is a donation Corresponding author. E-mail addresses: marianne.modlinger@u-bordeaux-montaigne.fr (M. Mödlinger), bsabatin@mit.edu (B. Sabatini). ⁎⁎ http://dx.doi.org/10.1016/j.jasrep.2017.10.018 Received 21 March 2017; Received in revised form 21 September 2017; Accepted 7 October 2017 2352-409X/ © 2017 Elsevier Ltd. All rights reserved. Journal of Archaeological Science: Reports 16 (2017) 248–257 M. Mödlinger, B. Sabatini Fig. 1. – Verchni Koban and Chmi cemeteries located in the Republic of North Ossetia-Alania, Russia. Fig. 2. – Studied Bronze Age daggers. Daggers A – I likely originated from the Koban cemetery, and J and K from Chmi. Samples were taken from the locations indicated with an arrow. The OES analyses from the SAM database are indicated by circles. However, the sampling location of the OES analyses could not be identified on daggers E and I. The samples taken from the blade of dagger J for OES analyses were not published in the SAM database. Despite their insecure find associations, the daggers can be approximately dated. Dagger B is dated to the Early Bronze Age, while dagger J is associated with the Meskhetian Culture of southwestern Georgia. Dagger J and K are dated to the beginning of the Late Bronze from the Anthropologische Gesellschaft (F. Zwicklitz) (inv.no.s 42.466–42.553). 6) Parcel I, 1888: old finds from the Caucasus, which were bought from S. Kulov in Wladikawkas, February 1888 (inv.no.s 42.554–42.637). 249 Journal of Archaeological Science: Reports 16 (2017) 248–257 Microstructural characterisation of the daggers, and chemical analyses, were performed on cross-sections taken at the locations shown in Fig. 2. Further chemical analyses were carried out on drilling samples taken with a 1 mm drill bit. Each sample was mounted in cold epoxy resin and polished using 400-1200 SiC papers, and then by diamond suspension paste of up to 0.25 μm granulometry. The samples were characterized using optical microscopy in both bright and dark field, chemically analysed using EDXS, and then etched for metallographic examination using NH4OH + H2O2 (according to the ASTM E407 no. 44 recipe). Chemical analyses of the polished cross-sections were also conducted using X-ray fluorescence (XRF). These analyses were compared to past results by Optical Emission Spectrometry (OES) given in the SAM database, and current ones obtained by EDXS. The analytical methodology employed by the SAM study, and the condition of the samples are unknown. In order to completely characterize the daggers, micro-hardness measurements and Pb-isotope analyses were carried out on select daggers. Chemical analyses were performed using a JEOL JSM-6460LV SEM with an Oxford Instruments SDD XMax 20 under high vacuum, and an ARL Quant'X (Thermo Fisher Scientific) XRF at 28 kV (with Pd filter), and 50 kV (with Cu filter). The SEM was calibrated using the internally provided software database standards library, as well as certified pure standards of Si, Ca, etc. for the quantitative analyses. The results shown are the mathematical average of 6–10 spectra of approximately 200 × 600 μm taken for 60 s each. The element concentrations for each averaged analytical campaign were classified as follows: alloying (wt % > 1), minor (1 > wt% > 0.3), and trace (wt% < 0.3). Compositional analyses by SEM and XRF were taken at the same locations. The XRF results are the average of two measurements of the whole surface of the sample, which were quantified using the manufacturer's software. Micro-hardness of the daggers, as well as Pb-isotope concentrations, were also measured. Micro-hardness was taken at 10 g for 10 s (μHV10,10) for daggers B and E. Lead isotope analyses were conducted on the hilt of dagger A, and several of the dagger's blades (A–D, F, and J). One sample from each dagger, samples that were first studied metallographically, were later analysed for isotopes at the Curt-EngelhornZentrum Archäometrie GmbH (CEZA), Mannheim, Germany using a Neptune Plus HR-MC-ICP-MS. These results were compared to scantly available lead isotopic signatures for the region. The hilt of dagger A and several blades (daggers A–D, F, and J) were selected for isotopic analysis based on their chemical composition. Artefacts of different composition were chosen in order to identify and document possible differences in their signatures, and any signs of metal recycling and mixing. One should note that Cu-As-Sn alloys are generally considered the result of recycling in this region (Kunze, 2012), and, consequently, arsenical bronzes with no or small amounts of Sn (the hilt of dagger A, blades of daggers B and C, and blade of dagger D) were selected for analysis. Tin bronzes with no detected As (the blades of daggers A and J) and a ternary bronze (blade of dagger F) were also selected. Lead isotopes give indication of the age of the ore used in the manufacture of a copper-based artifact. Knowing the age of the ore, this information can aid in identifying its geographic origin and thus the source of the metal used to manufacture the daggers. This methodological approach to provenancing metal finds is well-recognized and powerful, however is often more useful in eliminating possible ore sources than definitively identifying them (e.g. Pernicka, 1987). Though significant copper bearing minerals exist in the Greater Caucasus', the ones presented by Kunze et al. (2017), and here by the 1 2 X 1 1 1 1 Belt buckle, mace head (quartzite) bronze button, 4 short, zylindric bronze spirals Tweezers Knife, 2× jewellery, bronze sheets Neck ring, tweezers Earring, neckring, 2 finger rings, bronze sheets 1 X 2 2 2 1 3 1 1 2 1 31 9 1 2 1 X 2 1 1 3 3 (C) (D) (E) (F) (G) (H) (I) (J) 41.389 41.631 41.788 41.803 41.927 41.928 42.186 42.466 42.623 (K) Koban, grave XI (adult, male) Koban, grave XVI (juvenil male) Koban, grave XXIX (adult, male) Koban, grave XXX (juvenil, male?) Koban (?) Koban (?) Koban, grave III (adult, male) Chmi (grave I) Chmi (?) 1 (toy) 1 1 1 1 41.268 (A) 41.283 (B) Koban, grave IV (juvenil, male?) Koban, grave V (juvenil) 1 Further finds Ceramics Necklace Spirals Temple ring Bracelet Appliques Beads Pin Fibula Pendant Belt plate Axe 2. Methodology Dagger inv. no. (in brackets letter for identification) Associated finds Age at the middle of the 2nd mill. BC. All other daggers are dated to the Middle and Late Bronze Age. Find spot Table 1 – Alleged associated finds and find spots of the studied daggers. Both the sex and age of the interred are uncertain. ‘X’ indicates the presence of specific types of objects where the exact number is unknown (information as from the inventory book of the Natural History Museum Vienna). M. Mödlinger, B. Sabatini 250 Journal of Archaeological Science: Reports 16 (2017) 248–257 M. Mödlinger, B. Sabatini Table 2 – Chemical composition of the analysed Koban- and Chmi-daggers. Apart from three drilling samples (dagger A hilt, dagger E hilt and rivet), all EDXS analyses were carried out on freshly polished and un-etched metal surfaces. Corrosion zones were avoided throughout, which prevented the analysis of dagger K. The reported OES analyses were retrieved from Krause (2003). Dagger Source Detail Cu A SAM EDXS XRFa EDXS XRF SAM EDXS XRFa SAM EDXS XRF SAM EDXS XRF SAM Hilt Hilt Hilt Blade Blade Blade Blade Blade Blade Blade Blade Blade Blade Blade Blade (upper part) Blade (lower part) Rivet Hilt Blade Blade Blade blade blade Blade Blade Blade Blade Blade Rivet Blade Blade Blade Blade 94.34 90.15 88 89.31 91.02 96.11 93.98 86.85 94.79 93.23 90.34 94.64 93.33 93.11 92.03 B C D E EDXS F G H I J K a b SAM EDXS XRFa SAM EDXS XRFa SAM EDXS SAM EDXS SAM EDXS XRFb SAM EDXS 82.09 Mn S Fe Co Ni Zn As 5 9.85 11.2 0 < 0.005 < 0.05 < 0.01 0.06 < 0.1 < 0.005 < 0.05 < 0.01 0.03 0.01 < 0.1 < 0.005 < 0.05 < 0.01 0.04 0.01 < 0.1 < 0.005 < 0.05 < 0.01 0.02 0.02 < 0.1 < 0.05 < 0.01 0.05 0.03 < 0.1 ### < 0.005 0.08 3.80 6.02 12.37 5.00 6.77 9.13 3.50 4.78 5.05 3.70 Se Ag Cd 0.08 0.10 0.10 0.02 < 0.1 0.03 0.08 < 0.1 0.03 0.03 0.04 0.01 < 0.1 4.15 4.48 3.00 4.45 3.82 3.00 5.82 5.50 2.60 2.90 2.20 1.75 2.70 0.50 2.50 Sb 0.07 0.3 tr. 0.16 0.55 0.07 0.06 < 0.005 #### < 0.005 < 0.005 0.09 0.02 < 0.01 < 0.005 0.36 0.01 < 0.005 0.06 0.04 < 0.005 0.19 0.04 tr. 0.30 0.11 < 0.005 0.08 0.06 tr. 0.08 2.50 0.34 0.14 tr. 0.23 1.40 1.30 1.13 1.00 16.94 0.97 2.72 2.54 0.82 0.91 0.82 0.38 0.70 0.43 4.00 4.63 3.80 5.17 1.15 8.72 8.76 0.03 0.91 0.67 0.28 tr. 0.25 0.55 0.93 0.58 0.19 0.40 1.95 1.83 0.04 0.57 0.13 0.03 < 0.005 < 0.005 tr. 89.00 ### 1.53 89.38 ### 1.53 95.23 93.99 tr. 93.90 < 0.005 < 0.05 < 0.01 95.57 91.86 92.54 < 0.005 < 0.05 < 0.01 92.63 91.91 ### 90.99 90.38 ### 95.58 90.52 ### 89.98 < 0.005 < 0.05 < 0.01 97.29 Too corroded for chemical analyses Sn 0.03 < 0.005 0.25 0.05 < 0.005 < 0.005 0.22 0.07 < 0.005 0.08 0.03 < 0.005 0.24 0.07 < 0.01 0.053 10.13 8.66 n.d. Te Pb Bi 0.1 n.d. < 0.005 0.15 < 0.01 < 0.005 0.07 n.d. < 0.01 n.d. < 0.005 0.05 0.01 < 0.01 n.d. < 0.005 0.09 0.34 tr. 0.42 0.62 < 0.01 n.d. < 0.005 < 0.005 < 0.005 < 0.005 tr. 0.53 0.64 0.97 0.42 0.69 0.79 0.42 tr. 0.95 0.72 0.46 0.42 0.04 < 0.01 0.01 0.01 < 0.01 0.01 < 0.01 0.01 n.d. 0.01 < 0.01 0.03 Low intensity. Presence of corrosion. blades. Consequently, their analyses may have showed higher amounts of Sb, Ni, Bi, and Pb than by EDXS, since these elements are usually enriched on the surface. Interestingly, the amounts of As are usually higher according to EDXS, except for dagger I because As was not enriched in the corrosion layer (see also Fig. 3c). In most of the cases, the EDXS and XRF analyses are in good agreement for alloying elements (e.g. As, Sn, Sb, and Pb), except for the amount of As in the blades of daggers B and C, and in general for dagger E. The difference can be explained by the presence of γ-phase close to the surface, and especially along the edges of the blades of daggers B and C. The OES analyses of the same daggers were taken from the center of the blade where there is less γ-phase present. Coupled together, these methodological differences led to higher amounts of As reported by XRF compared to either EDXS or OES. Dagger B′s blade, for example, is reported by OES to contain 3.8 wt% As, 6 wt% with EDXS, and 12.4 wt% with XRF. Differences in sample preparation and analytical method aside, a vast discrepancy exists between the analyses for the blade of dagger E. While EDXS did not show any significant amounts of As, OES showed almost 4 wt%. The SAM data also showed about 2.5 wt% of Sn, which is low compared to the EDXS analyses at almost 17 wt%. The latter results are confirmed by the microstructure of the dagger's blade, which is indeed full of Cu-Sn eutectoid. The discrepancy in composition is the result of a common ‘restoration’ practice in the 19th and early 20th century of combining parts of artefacts if they seemingly belonged together (or fit together). The part of the blade analysed by OES was at the upper portion of the blade, and by EDXS at the lower. From the current authors, are the only ones reported in the southeast. As a result, the only geographically relevant dataset is the one recently reported for Armenian copper ores (Meliksetyan and Pernicka, 2010). The results of our lead isotope analyses are compared to these data. In addition to the scarcity of comparable ore sources, the applicability of artifact lead isotopic signatures must be considered in light of both pre- and post- metal production activities. To be useful they ideally should not have been the product of recycling and mixing of differently sourced metal, nor the mixing of several ores pre- metal production. The chemical and lead isotopic signatures of potential ore sources should also not be common to several others, and an ore source should ideally have a homogenous lead isotopic signature across the entire ore body (e.g. Pernicka, 1987; Sabatini, 2015). 3. Chemical composition The chemical composition of the daggers, according to the analytical methods mentioned above, is reported in Table 2. The relative purity of the alloys, having few reported elements when measured using EDXS, is due to the detection limit of the instrument (0.1 wt%); however low amounts of Ni in Late Bronze Age Caucasian copper alloys are typical (Japaridze et al., 2007; Kunze, 2012). The same limitation of EDXS also accounts for the lack of reported Bi and Ag, which are often present in the SAM data (Krause, 2003). Comparatively higher concentrations of trace elements in the SAM data are also likely due to their sampling strategy. Each 2 mm sample drilling included the corrosion layer on both sides of the dagger's 251 Journal of Archaeological Science: Reports 16 (2017) 248–257 M. Mödlinger, B. Sabatini Fig. 3. – Selected SEM images of dagger blades B, C, D, E, and F. a) Dagger B (arsenical bronze): the light grey γ-phase precipitated on the surface in a thick band (partly covered by corrosion), and at the grain boundaries in thinner bands. At the grain boundaries growing corrosion was noted (small to big black zones). b) Dagger B (arsenical bronze): the darker α-phase contains about 4.5 wt% As, and the light grey γ-phase about 30 wt%. Sb is enriched in the corrosion. c) Dagger C (arsenical bronze): the light grey band between sound metal and corrosion (dark grey) consists of γ-phase. There was no As detected in the outer, darker zone of corrosion. d) Dagger D (arsenical bronze): dark inclusions consist of CuS with c. 21 wt% S. White inclusions consist mainly of Cu-Pb with c. 20–22 wt% Pb, and other elements such as Sb, Sn, and Ag with each up to 1.5 wt%. e) Dagger F (ternary bronze): 1 and 2 are Sn crystals with c. 90 wt% Sn, 3.7 wt% Sb, 0.9 wt% Ta, and 7 wt% Cu; 3 is a Pb-inclusion with c. 65 wt% Pb, 19 wt% As, 15 wt% Cu, and 1 wt% Sn; 4 is γ-phase (As-Cu system) with c. 3 wt% Sb. f) Dagger E: soldered on tin bronze top with eutectoid up to 3.4 wt% Sb. The Sb tends to enrich more in the δ-phase. present in either minor or trace amounts (0.01–0.95 wt%) according to OES and XRF analysis. soldering, and the vast differences in chemical composition between the top and bottom halves, it is clear that two different blades were soldered together. The microstructure of the adjoined dagger top reveals significant amounts of (α + δ) Cu-Sn eutectoid, where Sb is enriched, and no signs of Pb, which is typically seen as white spots (Fig. 3f). Furthermore, some analytical discrepancies between analytical campaigns may have arisen due to sample preparation. Drilling samples taken for analysis in the SAM project from the hilts of the daggers may have contained small amounts of the blades, as was certainly the case for dagger A (this resulted in significantly lower amounts of As measured by OES, since the blade of dagger A contains almost no As). Interestingly, none of the reported OES analyses contain Fe or S. These elements were detected by EDXS, and S was detected in half of the daggers (D, E, F, H, I, and J). Iron, meanwhile, was only detected in the three drilling samples of one rivet and the hilt of dagger E (that also contain up to 1.7 wt% S), which were taken for EDXS analyses. It should be noted that Fe likely did not derive from the drill bit, since the drilling sample from dagger A contains no Fe. Low concentrations of Fe were also noted by XRF (< 0.05 wt%), however this analytical method is typically imprecise. It is also not uncommon for ancient copper alloys to contain low amounts of Fe. In all but one dagger (dagger B), Pb is 4. Microstructure The microstructure of all of the daggers, whether arsenical or tin bronze, or a mixture (Cu-As-Sn), show corrosion and inhomogeneous αsolid solution within their dendritic structures (so-called ‘ghost-structure’) that are visible through differences in their copper concentrations. The blade of dagger K also shows cracking, which was likely caused by tension during deformation (Fig. 4d), indicating not enough or too late annealing. Significant deformation was also noted in the blades of daggers A, F, H, J, K, and the soldered-on top of dagger E (see also Fig. 4b–c). Deformed equiaxed grains with twins and strain lines are visible. The corrosion usually followed the Cu-rich areas of the αsolid solution. Minor and trace elements were mainly found in inclusions and along the grain boundaries. Deformed equiaxed grains with twins and strain lines are visible. The corrosion usually followed the Curich areas of the α-solid solution. Minor and trace elements were mainly found in inclusions and along the grain boundaries. Due to greater electrochemical potential, corrosion preferentially 252 Journal of Archaeological Science: Reports 16 (2017) 248–257 M. Mödlinger, B. Sabatini Fig. 4. – Microstructure of the blades of daggers G, H, J, and K (all ternary Cu-As-Sn bronzes). a) Dagger G: corrosion outlines the grain boundaries of the inhomogeneous α-solid solution, and is significantly increased at the grain boundaries. b) Dagger H: etched with NH4OH + H2O2. The Curich zones of the inhomogeneous α-solid solution are corroded. In the Cu-poor zones severely deformed grains with strain lines and twins can be seen. c) Dagger J: the microstructure is clearly visible even without etching. Corrosion outlines not only the grain boundaries, but also twins and strain lines. d) Dagger K: the dagger was too corroded to carry out chemical analyses. A fragile fracture is shown in the picture. It was caused during deformation of the blade that increased its fragility due to the many CuO-inclusions. Fig. 5. – Microstructure of the blade of dagger B (arsenical bronze). Note the vast amount of light grey γ-phase, which precipitates all along the surface (thicker bands) and along the grain boundaries (thinner bands). The dark spots are corrosion, which start at grain boundaries and especially at their crossing. Inclusions in all of the daggers showed low amounts of deformation, which is unusual. However, since most of the inclusions contain significant amounts of Pb, and are not of the more common variety of Cu2 − xFexS known from Central European Fe- and S-rich tin-bronzes (Mödlinger and Piccardo, 2013), they cannot be used to interpret the total amount of deformation applied to the blade. After each annealing process the inclusions would have returned to a globular form, so only information about the last annealing was preserved. The advantage of SEM-EDXS analyses is that they allow one to see how, and in which form, minor and trace elements and different phases are distributed in the microstructure. The composition of the inclusions varied widely between each blade: the blade of dagger D showed both Cu-S and Cu-Pb inclusions (the latter containing also Ag, Sn, and Sb) (Fig. 3d); dagger G showed Pb-inclusions containing about 2.6 wt% Sn (2 wt% more than in the α-solid solution), 18 wt% As, and only 14.5 wt % Cu; dagger H, Cu-Pb-inclusions with slightly more Sn and Sb, and slightly less As than in the α-solid solution. The blade of dagger I showed a variety of vastly different inclusions, including: Cu-Pb- begins at the intersection of grain boundaries. On the surface, azurite and malachite were visually identified as the main corrosion products with copper oxides such as tenorite and cuprite between them and close to the surface. The blade cores showed indications of the original surface metal in the form of thin layers of copper carbonates. The corrosion layer was especially thick on dagger K. Some of the alloying elements, but also minor elements, are significantly enriched in the corrosion, as was observed for Sb and Sn on the blade of dagger I. Two arsenical bronze daggers, daggers B and C, showed significant amounts of γ-phase, which precipitated along the grain boundaries and formed thin band-like structures starting from the edge of the blade and extending to their centers (Fig. 5). These γ-phase bands are also found on the surface of the daggers in thicker bands, and are covered by corrosion in some zones (Fig. 3a–c). They are not, however, evidence of the inverse segregation of arsenic (Mödlinger and Sabatini, 2016). Dagger C showed blotted γ-phase around the intersection of the grain boundaries. It contained about 1 wt% Sb in the bands of γ-phase as measured by EDXS, which was comparatively 0.2 wt% by XRF. 253 Journal of Archaeological Science: Reports 16 (2017) 248–257 M. Mödlinger, B. Sabatini Fig. 6. – Microstructure of dagger blade D (arsenical bronze blade) and the soldered on top of dagger E (tin bronze blade), and its micro-hardness values. a) Dagger D: The grain boundaries are outlined by corrosion. Moreover, corrosion is severe around the inclusions and where grain boundaries cross. b) Dagger E: Even without etching the microstructure is clearly visible. Corrosion outlines not only the grain boundaries, but also twins and strain lines. c) Dagger E: The (α + δ) eutectoid of the Cu-Sn system is clearly visible. d) Dagger E: micro-hardness measurements (μHV10,10). The imprints were made in zones with different amounts of alloying elements in the inhomogeneous α-solid solution and on the eutectoid. Fig. 7. – Microstructure of the blade of dagger I (ternary Cu-As-Sn bronze) after etching with NH4OH + H2O2. a) and b) Remnants of the dendritic structure are still visible, as well as inclusions of different compositions. c) The dendritic structure is clearly visible and not significantly deformed. The alloy did not receive much deformation in the sampled area. d) Hardness measurements (μHV10,10). The imprints were made in zones with different amounts of alloying elements in the inhomogeneous alpha solid solution. showing various types of inclusions, such as Pb-Cu-As (c. 60 wt% Pb and small amounts of Sb, Sn, and Cl), the blade also contains small amounts of Cu and Sn crystals (Fig. 3e). These crystals are of particular importance since they show that the Sn was not completely molten when the alloy was cast and may be the result of the last-minute addition of cassiterite or another Sn mineral to molten copper or a recycled tin bronze (e.g. Rovira et al., 2009; Farci et al., 2017). There are inclusions with about 85 wt% Pb; Cu-Pb-inclusions with c. 18 wt% Pb, and slightly less As, Sn, and Sb than in the α-solid solution; Cu-Sninclusions with almost 10 wt% Sn, 4 wt% Sb, and 2 wt% As; CuS-inclusions; and finally, Cu-S-inclusions with Sb and Sn. The blade of dagger J showed two types of inclusions: one with Cu-Pb and some Sn; and, the other with Cu-Pb with Sn and Fe. The most intriguing inclusions were visible in dagger F. Apart from 254 Journal of Archaeological Science: Reports 16 (2017) 248–257 M. Mödlinger, B. Sabatini Fig. 8. – Lead isotope ratios of the analysed daggers, and of copper ores from the south-eastern Great Caucasus (Kunze et al., 2017). The data from Armenia is taken from Meliksetyan and Pernicka, 2010, 51–52, Tab. 5–6. c) shows details of a), while d shows details of b). also small Cu crystals still visible in the matrix, pointing more likely to a last-minute addition of Cu to the melt rather than a recrystallisation of Cu afterwards. Finally, the temperature that the daggers were annealed was not consistent nor necessarily very high, since there is eutectoid in daggers A and E (tin bronze), and F (ternary bronze). The blade of dagger A and the top of dagger E show (α + δ) eutectoid of the Cu-Sn system. Of particular importance is that the δ-phase contains significant amounts of Sb of up to 3.4 wt% (dagger A). The corrosion is concentrated on the α-solid solution, and sparsely on the δ-phase as well as zones inside the eutectoid that seem to be pure copper (dagger E; see Fig. 6c–d). One can also clearly see the progression of corrosion around the more corrosionresistant inclusions and grain boundaries (Fig. 6a), as well as the copper-rich zones of the α-solid solution, twins, and strain lines (Fig. 6b). Interestingly, Cu or Cu-rich zones were often found as ‘bands’ inside the eutectoid (Fig. 6c). 5. Micro-hardness Several micro-hardness measurements (μHV10,10) were carried out in different Cu-rich and Cu-poor zones, and on the (α + γ) eutectic of dagger's B and (α + δ) eutectoid of dagger E (i.e. the top of dagger E) (Fig. 6d). Daggers B is considerably higher in arsenic than the rest of the studied set (c. 6 wt%). Micro-hardness measurements of dagger B showed higher values for the γ- than α-phase; however, these values should be considered relative since the bands of γ-phase were too thin to be certain that only the bands themselves were tested and not the underlying α-phase. The measurements on dagger E showed lower hardness values in the Cu-rich than in the Cu-poor zones, and much higher in the eutectoid (about 450 ± 30 μHV). Fig. 7d clearly shows the impact the amount of As and Sn has on the hardness of the alloy (dagger I; ternary bronze). Even though the blade of dagger I was annealed and cold hardened several times, the alloy did not homogenize, indicating that annealing times were rather short at lower temperatures. The micro-hardness measurements were around 140–170 μHV 255 Journal of Archaeological Science: Reports 16 (2017) 248–257 M. Mödlinger, B. Sabatini The Sn crystals in dagger F can provide valuable information regarding the origin of the metal through the identification of their minor and trace elements. Significant to dagger F, and its Sn crystals, is the detection of c. 1 wt% Ta, which was not found in any other dagger. Hosking (1982) notes that ‘intensely red-pale coloured cassiterites containing Ta and possibly Nb in the lattice, are restricted to the west [Southeastern Asian Tin] belt and are paramagnetic, whilst those that are brownpale colour pleochroic, a phenomenon probably also due to Nb/Ta or W in the lattice, occur in both belts and may be ferromagnetic’. One of the two crystals also contains 0.4 wt% Fe, and both of them contain 3.5–4 wt% Sb (dagger F; see Fig. 3e). However, no Zr or Nb was detected, as one might expect to find if the Sn originated from cassiterite or pegmatite minerals. Finally, the lead isotope signatures of all but one dagger match well with those from Georgian copper ore sources overlapping with those from Armenia. The daggers were, apart from dagger J, which only overlaps with the Armenian copper ores, most likely made of copper ore from the Greater Caucasus'. According to the prior and produced chemical, microstructural, and isotopic data, there is no indication of recycling with non-localized copper ores. Table 3 - Measured lead isotope ratios. The analyses were carried out with a Neptune Plus HRMC-ICP-MS at the Curt-Engelhorn-Zentrum Archäometrie GmbH (CEZA), Mannheim, Germany. The precision of measurement is less than ± 0.001% for ratios with 208Pb and 206Pb in the denominator and for 207Pb/204Pb, and up to ± 0.02% for 207Pb/206Pb. For 208Pb/204 Pb it is up to ± 0.06%. Sample weight: 100 ppb. Dagger Object 208Pb/ 206Pb 207Pb/ 206Pb 208Pb/ 204Pb 207Pb/ 204Pb 206Pb/ 204Pb A A B C D F G J 2.0900 2.0821 2.0857 2.0847 2.0839 2.0856 2.0860 2.0659 0.84957 0.84470 0.84574 0.84577 0.84518 0.84608 0.84630 0.83425 38.414 38.499 38.543 38.486 38.504 38.504 38.513 38.818 15.615 15.619 15.629 15.614 15.616 15.620 15.625 15.675 18.380 18.490 18.480 18.462 18.477 18.462 18.463 18.790 Hilt (CuAs) Blade (CuSn) Blade (CuAs) Blade (CuAs) Blade (CuAs + tr. Sn) Blade (CuAsSn) Blade (CuAs + 0.7% Sn) Blade (CuSn) for Cu-rich zones, and above 200 μHV for Cu-poor zones (i.e. zones with higher amounts of As and Sn). 6. Lead isotope analyses 8. Conclusion As shown in Fig. 8, the isotopic signatures for Georgian ores and the daggers overlap well with the Armenian ore sources. Apart from the blade of dagger J, the ratios for the daggers are concentrated in a very narrow field at 208Pb/204Pb vs. 206Pb/204Pb at 38.5 and 18.5, respectively (Table 3). The same is true for the 208Pb/206Pb vs. 207Pb/ 206Pb at 2.85 and 0.845, respectively. The ratios for the ores are instead concentrated at 208Pb/204Pb vs. 206Pb/204Pb at 38.40–38.45 and 18.40, respectively, and for 208Pb/206Pb vs. 207Pb/206Pb at 2.95 and 0.850. The isotopic signature of dagger A overlaps with the Georgian isotopic values from the Stori valley (Kunze et al., 2017). The other daggers, apart from dagger J, however, are very close to the Georgian ore signatures. Until disproven by additional isotopic analyses of ores in the region, all daggers, apart from dagger J, which seems to only overlap with the Armenian copper isotopic ratios, were likely made from Caucasian copper ore. Due to the lack of ore analyses from the Greater Caucasus', no further relationship between the daggers and the source for their metal can be postulated. Also, according to the available data, there were no indications of recycling of differently sourced metal or ore. Eleven Bronze Age dagger blades from Koban and Chmi, North Ossetia-Alana, Russia, were chemically analysed by EDXS, and their microstructures characterized. The dagger blades were made of either arsenical bronze with up to ~7 wt% As (daggers B-D), of tin bronze with 10 or 17 wt% Sn (dagger A, and soldered on top of dagger E), or of a ternary Cu-As-Sn alloy (upper part of the blade of dagger E, and daggers F–K). The alloys characteristic have low amounts of Ni, Ag, Pb, S, and Fe. It should be noted that Ni and Ag were usually present in amounts below the detection limit of EDXS. Trace elements were mainly detected in the inclusions or along grain boundaries. Antimony is in ten out of eleven of the blades, and in four daggers (daggers D, E, G, and I) at around 1–2 wt%. The blades were analysed by the SAM project, however the current analyses showed several discrepancies in chemical composition. These differences can be explained by the inclusion of corrosion products in the OES analyses, and the presence of γphase on the surfaces of the daggers (blades of daggers B and D). In two instances the chemical composition of the dagger hilts were analysed and found to contain almost 10 wt% As (dagger A) with a blade containing 10 wt% Sn, while the hilt of dagger F was made of a ternary Cu-As-Sn alloy. These material choice applications suggest that differences in the alloys were recognized, and that each was used intentionally during the manufacture of the daggers. All of the dagger blades have been annealed and cold worked. Two dagger blades (dagger A, and the top of dagger E) show Cu-Sn (α + δ) eutectoid, indicating low temperature or short-term annealing. The amount of deformation applied during working is not connected to the chemical composition and none of the blades are homogenous. Their dendritic structures differ visibly throughout based on the amount of alloying elements present in the α-solid solution. This difference in composition also results in different micro-hardness values in Cu-rich and Cu-poor zones in the alloys. The blade of dagger F showed Sn crystals, indicating that tin was added when the copper was not completely molten. These untouched Sn crystals can help identify the metals origin, since they contain significant amounts of Ta (c. 1 wt%). One of the two analysed crystals also contained 0.4 wt% Fe, and both have 3.5–4 wt% Sb. Due to scarcity of lead isotope analyses of copper minerals from the Greater Caucasus', no relationship between the daggers and ore sources, other than from the Greater Caucasus', can be drawn. However, a close relationship with Armenian ores is likely. According to the available data, no indication of recycling with non-local ore seemed to have occurred. These results support the interpretation that Sn and As, in some form, was intentionally added to the local copper. 7. Discussion Only four of the eleven Bronze Age dagger blades are not ternary Cu-As-Sn alloys (daggers A, B, C, and D). These blades consist of arsenical (daggers B, C, and D) or tin (dagger A) bronze. Dagger A is unique in the assemblage in that the hilt contains ~9.8 wt% As and no Sn, while the blade contains 10.1 wt% Sn and no As. The selection and use of these alloys, especially for dagger A, suggests intentionality in metal use. Apart from dagger E, which comprises two different alloys soldered together in the modern era, all of the others studied are consistently Cu-As, Cu-Sn, or Cu-As-Sn. The overall chemical composition, corrosion structure, and identified inclusions of dagger E indicate that the soldered on piece is ancient. As for more precise dating of the daggers and their relationship to the graves they originated, no interpretation can be made at this time due to the circumstances of their procurement. Objects made of ternary Cu-As-Sn alloys with > 1 wt% Sn and As are known from the Late Bronze Age site of Udabno, Georgia (one pin) (Kunze, 2012), the Middle Bronze Age objects from Atskuri, Georgia (three daggers and one ring), and the Late Bronze Age Colkhian bronzes reported by Japaridze et al. (2007) (one dagger, one axe). The finds from Udabno, which contain ≤1 wt% Sn and As, suggest heavy recycling of old arsenical bronzes with tin bronzes, resulting in ternary alloys (Kunze 2012, 96, Fig. 5.6). 256 Journal of Archaeological Science: Reports 16 (2017) 248–257 M. Mödlinger, B. Sabatini (II-I millenia BC). GFMS Journal 1, 5–11. Junghans, S., Sangmeister, E., Schröder, M., 1968–1974. Kupfer und Bronze in der frühen Metallzeit Europas. In: Berlin, Gebr. Mann, (4 volumes). Krause, R., 2003. Studien zur kupfer- und frühbronzezeitlichen Metallurgie zwischen Karpatenbecken und Ostsee. Rahden/Westfalen, Leidorf. Kunze, R., 2012. Interdisziplinäre Studien zu den Kleinfunden der Siedlungen Udabno I–III (Ostgeorgien). Unpublished PhD, University of Tübingen. Kunze, R., Rödl, T., Mödlinger, M., Rödl, T., 2017. Kaukasisches Kupfer in der Bronzezeit: Gewinnung und Einfluss. In: Prähistorische Zeitschrift 2/2, (in press). Meliksetyan, C., Pernicka, E., 2010. Geochemical characterisation of Armenian Early Bronze Age metal artefacts and their relation to copper ores. In: Hansen, S., Hauptmann, A., Motzenbäcker, I., Pernicka, E. (Eds.), Von Majkop bis Trialeti. Gewinnung und Verbreitung von Metallen und Obsidian in Kaukasien im 4.-2. Jt. v. Chr., Beiträge des Internationalen Symposiums in Berlin vom 1.–3. Juni 2006. Kolloquien zur Vor- und Frühgeschichte 13 (Bonn 2010), pp. 41–58. Mödlinger, M., Piccardo, P., 2013. Manufacture of Eastern European decorative discs from 1200 BC. Archaeol. Anthropol. Sci. 5 (4), 299–309. http://dx.doi.org/10.1007/ s12520-012-0111-6. Mödlinger, M., Sabatini, B., 2016. A re-evaluation of inverse segregation in prehistoric Cu-As objects. J. Archaeol. Sci. 74, 60–74 (DOI 10.1016/j.jas.2016.08.005). Pernicka, E., 1987. Erzlagerstätten in der Ägäis und ihre Ausbeutung im Altertum: Geochemische Untersuchungen zur Herkunftsbestimmung archäologischer Metallobjekte. Jahrbuch des Römisch-Germanischen-Zentralmuseums Mainz 34 (2), 607–714. Rovira, S., Montero-Ruiz, I., Renzi, M., 2009. Experimental co-smelting of copper-tin alloys. In: Kienlin, T.L., Roberts, B. (Eds.), Metals and Societies. Studies in Honour of Barbara S. Ottaway. Universitätsforschungen zur prähistorischen Archäologie, Bonn, pp. 407–414. Sabatini, B., 2015. The As-Cu-Ni system: a chemical thermodynamic model for ancient recycling. JOM 67, 2984–2992. http://dx.doi.org/10.1007/s11837-015-1593-3. Acknowledgments The authors would like to acknowledge the financial support provided by the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie Actions, grant agreement no. 656244. The XRF and lead isotope analyses were funded by the Fritz Thyssen Stiftung (project number Az. 20.15.0.084AA) as part of a bigger project on copper mineral study in Eastern Georgia by René Kunze, Tübingen, and Marianne Mödlinger. The authors are very grateful to Armanda Barbangelo, Università degli Studi di Genova, Italy, for the micro-hardness measurements and fruitful discussions regarding microstructure. References Farci, C., Martinón-Torres, M., González Álvarez, D., 2017. Bronze production in the Iron Age of the Iberian Peninsula: the case of El Castru, Vigaña (Asturias, NW Spain). J. Archaeol. Sci. Rep. 11, 338–351 (doi 10.1016/j.jasrep.2016.12.009). Heinrich, A., 2006/2007. Franz Hegers Reisen und Ausgrabungen im Kaukasus und die Entstehung der ‘Sammlung Kaukasischer Alterthümer’ im Naturhistorischen Museum in Wien. In: Mitteilungen der Anthropologischen Gesellschaft in Wien 136/137, 2006/2007, pp. 107–143. Hosking, K.F.G., 1982. The nature and significance of the Pleochroism of the Cassiterites of the Southeast Asian tin belt, ore. Genesis 2, 753–759. http://dx.doi.org/10.1007/ 978-3-642-68344-2_70. Japaridze, I., Badzoshvili, T., Hauptmann, A., 2007. Bronze Metal treatment in Georgia 257