Introduction

1The advent and spread of metallurgy in the Caucasus mark a new step in pyrotechnology and the exploitation of the natural resources located in the highlands. The technological innovations that came into play are rooted in earlier fire manipulations that resulted in the irreversible alterations of raw materials. But early metallurgy probably developed together with another innovation: the vertical mobility of the herds, also known as transhumance, which must have played a role in the discovery of copper-ore deposits in mountainous areas. In point of fact, the Lesser Caucasus hosts some of the largest ore-containing mountains of the ancient world (Chernykh 1992). The metalliferous specificity of this region includes rich polymetallic deposits, and of course many copper-oxide or sulfuric zones (fig. 1). The metallurgical remains recently discovered on the Late Chalcolithic sites of Ovçular Tepesi and Zirinçlik, which comprise tools and copper-ore fragments, suggest that extractive metallurgy and cast metal technology were introduced in Nakhchivan (an autonomous province of Azerbaijan) as early as the second half of the 5th millennium BC (Gailhard et al. 2017).

Fig. 1 – Physical map of the Caucasus (M. Kurtubadze).

2It is thus probably no coincidence that major changes in settlement patterns occurred in the Caucasus valleys during the Chalcolithic and the Early Bronze Age as the grassland belt expanded in the mountains (Mocci et al. 2005, pp. 203‑205). Human settlements started to multiply in the uplands and the piedmonts of the Lesser Caucasus (Uçan Ağıl, Zirinçlik, Godedzor, Chobareti, Gegharot…), at the same time as human activities diversified into mining, metallurgy, and the probable production of charcoal.1 These changes occurred together with the development of agro-pastoralism in high elevation, as shown by the presence of cultivating practices next to herding at Godedzor (Palumbi et al., this volume) and long-distance exchanges, as shown by the distribution of ceramics, obsidian and lapis lazuli (Bobokhyan et al. 2014, p. 294; Chataigner et al. 2010, pp. 401‑402; Palumbi et al., this volume).

3Thus, drawing on the metallurgical remains that were recently discovered on the Late Chalcolithic sites of Ovçular Tepesi, Uçan Ağıl and Zirinçlik, this paper will discuss the appearance of metallurgy in Nakhchivan in a global context marked by many changes, in settlement patterns, technical skills, and subsistence strategies. Data pertaining to the Early Bronze Age levels of Kültepe I, Ovçular Tepesi and Uçan Ağıl will also be included in order to discuss the characteristics of South Caucasian early copper metallurgy from a wider angle. The archaeological and the technical aspects of the metal production over a 2500‑year time span, from the Chalcolithic to the Early Bronze Age, will be addressed in an attempt to provide a comprehensive study of the chaîne opératoire involved in this metal production.

Nakhchivan and the sites under study

4The autonomous Republic of Nakhchivan, a province of Azerbaijan, is located along an east-western axis that follows the Araxes valley. This region has an easy access to the Lesser Caucasus towards the north through several river valleys, while the plains of the Urmiah basin in Iran are accessible to the south via the Culfa (Djolfa) pass: Nakhchivan thus lies in a strategic position at the crossroads between cardinal routes (fig. 1).

5This study will be carried out through the analysis of the metallurgical remains attested on four settlements: Ovçular Tepesi, Uçan Ağıl, Zirinçlik and Kültepe I (fig. 2), which are located respectively in western Nakhchivan (Ovçular) near Şarur, and in central Nakhchivan (Uçan Ağıl, Zirinçlik and Kültepe I) in the Sirab basin.

Fig. 2 – Location of the sites under study (MBA; map background: O. Barge, CNRS).

6The Sirab basin is itself part of the Araxes drainage catchment. It is characterized by numerous sources of water with a high mineral content. The region of Sirab is located between the winter plains of the Urmiah basin in Iran and the pastures of the Upper‑Karabagh in the Zangezur mountains. Its strategic situation, along with the presence of copper ores, is undoubtedly one of the main reasons that attracted so many Chalcolithic groups during the 5th and 4th millennia BCE. In point of fact, a concentration of presumably mobile sites belonging to the Chalcolithic and the Early Bronze Age has been found in the vicinity of the Sirab and Qahab rivers along an east-west axis that may correspond to cattle tracks, as suggested by present-day pastoral routes (Bakhshaliyev, Novruzov 2010; Bakhshaliyev, this volume). Three of these sites (Şorsu, Uçan Ağıl and Zirinçlik) are being currently excavated in the framework of the Mission archéologique du bassin de l’Araxe: available preliminary data certainly confirms the identification of these settlements as mobile camps. Two of them have yielded evidence of metallurgical activities and will be analysed here accordingly. Early Bronze Age occupation levels are attested on all these sites, but three of them (Ovçular Tepesi, Uçan Ağıl and Zirinçlik) also contain Chalcolithic layers.

Kültepe I

7The site of Kültepe I is located on the left bank of the Naxçivan Çay (river), 10 km north from Nakhchivan city (fig. 2, 3). This site was first excavated during the Soviet period between 1951 and 1964 under the supervision of O.A. Abibullayev; excavations were resumed in 2012 by a French-Azerbaijani team directed by V. Bakhshaliyev and C. Marro, while the archaeological assemblages brought to light by the Soviet excavations, among which metallic objects and related metallurgical remains, are currently being re-examined in the framework of the Shelby-White foundation Kültepe I project, led by R. Berthon. This work has produced new ¹⁴C readings, according to which the Kura-Araxes layers of Kültepe I should be dated to the 3360‑2350 BC time span.

Fig. 3 – General view of Kültepe I in 2006 from the south-west (MBA, C. Marro).

Ovçular Tepesi

8Ovçular Tepesi is located on the left bank of the Arpaçay river in western Nakhchivan (fig. 2, 4). The occupation sequence of the site has been divided into four phases, two of which (Phases I and II) belong to the Late Chalcolithic: Phase I (ca 4400‑4350 BC) and Phase II (ca 4350‑3900 BC). After a 600‑year gap in the middle of the 4th millennium BC (ca 3900‑3300 BC), it was occupied again during the Early Bronze Age until ca 2400 BC (Marro, Bakhshaliyev, Ashurov 2011; Marro, Bakhshaliyev, Berthon 2014, pp. 132‑135).

Fig. 4 – Ovçular Tepesi (MBA, O. Barge).

9The Late Chalcolithic (4400‑3900 BC, Phase I and II) levels of Ovçular Tepesi have yielded an important corpus of copper objects related to the production of metal: fragments of nozzles, crucible remains and small artefacts are attested. As far as the Kura-Araxes period is concerned, the quantity of metallurgical artefacts is fairly small considering the area that was excavated. This may be due to the fact that the Early Bronze Age Kura-Araxes (EBKA) village was on the whole poorly preserved. The small EBKA metal corpus nevertheless makes possible interesting comparisons with earlier productions.

10It should be noted that Ovçular Tepesi is close to Gümüşlü, a silver-lead mine that overlooks the gorge of the Arpaçay river near Yaycı: a piece of galena (lead sulphide) was found at Ovçular in the Late Chalcolithic levels beside copper-ore fragments. The copper-ore deposits, on the other hand, are located fairly far from Ovçular, in central and eastern Nakhchivan.

Uçan Ağıl

11Uçan Ağıl is a small campsite located on a promontory, at the confluence between the Sirab river and a small wadi (fig. 5). This site was occupied during two periods, dated respectively to the Chalcolithic and the Early Bronze Age. In 2017, however, the earliest occupation layer in Chantier 1 has yielded a number of potsherds that recall the Later Neolithic assemblage of Kültepe I (Level 2, see Marro et al. 2019). The Kura-Araxes occupation has been dated through four ¹⁴C readings to the 3100‑2400 BC cal. period, while the earliest Chalcolithic level has been attributed to the 4800‑4600 BC time span (2 sigmas). Copper ore and a fragment of pure copper have been found in a filling layer (US 2045) very close to a fireplace (US 2041), associated with Kura-Araxes pottery, where some charcoal was sampled for radiocarbon dating.

Fig. 5 – Uçan Ağıl (alt. ca 1,200 m) from the north: this flat site is located along a modern sheep and goat track (MBA, C. Marro).

Zirinçlik

12Zirinçlik is a small campsite dated to the Chalcolithic period. It is located to the north-east of Kültepe I on the slopes of the Sarıdağ (fig. 6). At Zirinçlik, the occupation layers date back to the end of the 5th and the beginning of the 4th millennium. Judging by the pottery and the ¹⁴C dates, the occupation of this site is probably restricted to the Late Chalcolithic; interestingly enough, some pottery with Kura-Araxes overtones is also attested, but no typical Kura-Araxes shapes have been found. Two radiocarbon readings place this site within the 4240‑3690 BC cal. time span (2 sigmas).

Fig. 6 – The site of Zirinçlik (alt. ca 1,300 m) viewed from the east (MBA, C. Marro).

13The copper mine of Misdağı 1 is located only 3.5 km from the site. This copper deposit is characterized by its malachite and cuprite components, which have close parallels with some of the copper ores (malachite) found at Zirinçlik. Three mould fragments were collected from the surface, among which a double-flat-axe mould is particularly noteworthy (fig. 7).

Fig. 7 – Double-ceramic-axe mould from Zirinçlik, Late Chalcolithic (MBA, N. Gailhard).

14Thus, the four sites under study are divided between valley settlements (Ovçular Tepesi and Kültepe I) and small campsites (Uçan Ağıl and Zirinçlik) that are located in the piedmonts of the Lesser Caucasus: this is a unique opportunity for studying the relationships between mountain and valley settlements and their respective roles in the rise of extractive metallurgy, especially as some of them are connected to long-distance exchange networks that include eastern Anatolia, western Iran and the Mesopotamian world: these networks combine into an interaction sphere that we have called the “Chaff-Faced Ware oikumenè” (Marro 2010).

The material

General overview of the metallurgical evidence from Nakhchivan

15Altogether, 217 metal objects from Kültepe 1, Ovçular Tepesi, Zirinçlik and Uçan Ağıl, together with copper-ore fragments from the same sites, have been collected. Most of them have been sampled for chemical, and a smaller number for lead isotope, analyses. Some ore specimens collected from several copper deposits located in eastern Nakhchivan have also been added to this assemblage. A metallographic microscope examination was carried out to determine the technics used by the coppersmiths. These studies, which have been partly published (Gailhard et al. 2017), are still in progress.

16Three categories of remains were collected all together: copper-based metal objects, metal ores from mines or archaeological sites, and clay artefacts related to the practice of metallurgy (crucibles, nozzles and moulds) (tab. 1). All the objects date to the Chalcolithic or the Early Bronze Age, and were collected between 2006 and 2017, except for the artefacts retrieved during the Soviet excavations of Kültepe I in the 1950s.

Kültepe I

17The material from Kültepe related to metallurgy must be examined with caution: the practice of making beads with minerals such as malachite or azurite is also attested during the Neolithic period on this site, even in the earliest levels, but this has little to do with copper metallurgy. Since no Chalcolithic occupation layers are attested at Kültepe I2, early metallurgical practices and the production of metal from copper ore can only be studied through the material brought to light in the Kura-Araxes occupation levels. Only 20 of a 34‑piece assemblage are clearly related to metal production and result from high-temperature processes (fig. 8). The Early Bronze Age corpus includes needles, a ring, a blade, an awl, a spearhead and several mould fragments.

Fig. 8 – A mould fragment and a Kura-Araxes spearhead from Kültepe 1, material from the Soviet excavations, stored at the National Museum of History in Baku, Azerbaijan (MBA, N. Gailhard).

Ovçular Tepesi

18At Ovçular Tepesi, 158 items connected to metallurgy were found. The bulk of the finds come from the Chalcolithic levels (87 artefacts), the rest has been dated to the Early Bronze Age or cannot be dated with certainty. A total of 93 samples were analysed in Bochum. They have been divided in three categories: metal objects, ore fragments and ceramics used in metallurgical processes.

Metal objects

19There are four types of metal artefacts:

• small tools like awls, needles, a nail, blades and a hook, which constitute the majority of the finds (79 pieces);
• large tools illustrated by axes: these are the largest metal objects discovered at Ovçular; six examples are attested, among which flat axes, an axe-hammer and an adze-axe (for more information on the typological parallels, see Gailhard et al. 2017);
• jewellery and ornaments: 13 artefacts;
• undefined artefacts: about 50 pieces that were mostly found through sediment wet-sieving processes.

Copper ores

20The Ovçular Tepesi assemblage includes several pieces of copper ore, which suggest that extractive metallurgy was carried out on the site itself. Copper ore was found in seven different loci, five of which are dated to the Late Chalcolithic (Loci 6468, 8009, 8022, 22038 and 23063), while the other two come from unreliable contexts (Loci 6530 and 14239). They may date either to the Late Chalcolithic or to the Early Bronze Age.

21The context where they were found are generally located near or in the houses. One of them comes from Locus 6468, which is associated to a fire place and a floor. Several ore fragments were found in Locus 8009 (11 pieces) and Locus 8022 (3 pieces). One of the pieces from 8022 is in fact a slag or semi-product, which will be detailed below.

Ceramic tools

22Nozzles are artefacts linked to metallurgical activities. They have an essential function in the process of smelting or melting because, stuck to the tip of the bellows, they provide artificial air for the furnace, which makes high temperatures possible.

23A nozzle found in Locus 5292 is fairly well-preserved (fig. 9): its bottom part is broken, but the small hole is intact. Made of fired clay, its external surface is reddish and carefully burnished. This type is conical in shape, with an inner channel that is almost rectilinear (for a complete typology of nozzles, see Gailhard 2009, p. 73, pl. 45).

Fig. 9 – A nozzle with a conical shape (Locus 5292); its inner channel is almost rectilinear (MBA, N. Gailhard).

24The stratigraphic context of Locus 5292 is fairly clear, but again, this artefact comes from a filling layer belonging to House 11‑1, which straddles squares 5 and 11. Next to House 11‑1, two copper artefacts (a pin and a knife blade in a very bad condition) were found in another Late Chalcolithic house (House 5‑2), close to a large quantity of ash and dried-dung coal. Some kind of metallurgical activities was certainly conducted in House 5‑2, but it is not clear whether the nozzle was connected to the work carried out in this house or in House 11‑1. At any rate, it should be dated to Phase II.

25A second nozzle fragment was found in Locus 22003, close to the surface, but this time, only the fore part was preserved; it shows signs of heavy use, with its inner surface deformed by high heat. Its heavily vitrified surface shows that this tool was in use during smelting operations when it broke. Its state of preservation is not good enough to say anything about its original length, but interestingly, its diameter is similar to that of nozzle 5292 – its bad state of preservation, however, is not surprising considering the function of such objects. At any rate, this nozzle is another piece of evidence that indicates the practice of metallurgy at Ovçular, even if it cannot be dated with certainty: it comes from a filling layer which, in this particular case, is not connected with any building.

26Other traces of metallurgy include two fragments of crucibles. One was found near the western baulk of square 8. It was found in the filling layer of a room (Locus 8087) dating to the Late Chalcolithic (fig. 10). The inner surface is characterized by a fairly thick, glassy slag spilt. The exterior surface is red-brown and chaff-tempered with macroscopic mineral inclusions.

Fig. 10 – Fragment of a crucible (Locus 8087) covered by a thin layer of slag inside (4249/13); this fragment measures 44 mm in length, 37 mm in width, with a thickness of 15 mm (MBA, N. Gailhard).

27This fragment is probably part of a sub-cylindrical object with a rounded base identified as Type 2 (Gailhard 2009, pp. 88‑95, pl. 65). This kind of crucible is often deeper than wide. Similar examples were found in Period VI B2 at Arslantepe together with chalcopyrite ore (Palmieri et al. 1999). In term of absolute dating, the Period VI B2 in Arslantepe covers the 3100‑2900 BC time span3, which corresponds to the Early Bronze I (East Anatolian chronology). Several other examples are documented in eastern and south-eastern Anatolia, for instance at Hacınebi (Özbal, Adriaens, Earl 2000, p. 62), Tepecik (Esin 1987, pp. 71, 77), Norşuntepe (Hauptmann 1982), Cüdeyde (Braidwood, Braidwood 1960, p. 269), and Nevalı Çori (Hauptmann et al. 1993, p. 548). Chronologically, most of these crucibles are dated to the 3rd millennium BC, apart from those from Tepecik (second half of the 4th millennium BC) and Hacınebi (Hauptmann et al. 1993, p. 548).

28Another fragment of crucible was found in Locus 6332, but it is too small for reconstructing its original form. The inner surface shows heavy traces of heating with small prills of coppers. More investigation will be carried out on this ceramic through microscopic examination.

Zirinçlik

29The discovery of artefacts and ore connected to copper metallurgy on this small mobile camp is particularly significant for tracing back the relationships between pastoralists and metallurgists. They also give clues for reconstructing ore-circulation networks through valley and mountain sites during the Late Chalcolithic. A total of nine artefacts connected with metallurgy was collected at Zirinçlik, which cast some light on the chaîne opératoire. The stratigraphy of the site, together with the presence of copper ores, some slag and a few mould fragments, clearly shows that metallurgical practices were carried out by mobile groups living in the mountains. Their exact relationships with valley groups are not known yet, but some relevant information may be given by the isotopic analyses of domestic animal remains that are currently in progress (Berthon et al., this volume).

30The three moulds found in Zirinçlik are made of clay. One is a double-axe mould for casting operation; other moulds are too fragmented for us to reconstruct their exact shapes. All are open moulds. A similar discovery was made at Alkhantepe, where several moulds and crucibles were found together (Akhundov 2014, pp. 11‑16; Stöllner 2016, p. 214). This shows that the practise of smelting and mould-casting in the Caucasus was also performed on small mobile sites.

Uçan Ağıl

31The practise of metallurgy at Uçan Ağıl is illustrated by a piece of pure copper (US 2045), which is quite unusual (fig. 11). Judging by the form, the size and the first macroscopic studies, this piece must be considered as an ingot (without any specific form) or as some prepared raw material ready to be melted for the production of copper or bronze objects. This piece of pure copper is extremely dense and heavy: it weighs 166 g for a size of 5x2 cm. It proved almost impossible to cut when we tried to sample it for chemical analysis. The dating of this ingot is still uncertain: it was found in a filling layer, close to an Early Bronze Age hearth.

Fig. 11 – A piece of pure copper from Uçan Ağıl (US 2045); at first, we thought to have found a piece of copper ore, but after some test, it turned out to be a chunk of pure copper (MBA, N. Gailhard).

Analytical methods

32A selection of metal objects from the four excavations were prepared for analysis at the Deutsches Bergbau‑Museum Bochum (DBM) in Germany, with a view to determining the provenance of the raw material. X‑ray diffraction (XRD) was applied to identify the mineral content of the ores and Inductively Coupled Plasma Mass‑Spectrometry (ICP‑MS) was used to analyse the chemical bulk composition of both the ores and the metal finds. All this work was conducted by the team of the DBM; the results were studied by N. Gailhard and M. Bode. The Lead Isotope Analysis (LIA) of the ores and the metal artefacts was carried out at the Institut für Geowissenschaften of the Goethe‑Universität (Frankfurt am Main), under the direction of Prof. Dr S. Klein.

Results

33The results of the analyses of the copper-rich minerals found in the archaeological layers of the four sites are presented below, together with the ores from the outcrops of copper deposits sampled in Nakhchivan between 2007 and 2016, and a selection of copper-based objects.

Copper-bearing rocks

34The bulk of the copper collected from the archaeological sites under study and the different copper deposits sampled in Nakhchivan are copper-oxide and carbonate minerals. Depending on their type, they are generally green or blue; malachite and cuprite are the most common copper minerals attested so far. As far as the archaeological sites are concerned, this may be explained by the periods under study, the Chalcolithic and the Early Bronze Age, when these copper ores were mostly used.

35According to the XRD analyses so far carried out (tab. 2), the copper ores found at Ovçular Tepesi and Zirinçlik are all oxide (malachite or cuprite). As for Kültepe I, the copper minerals come from the Neolithic levels excavated between 2012 and 2016. These are small pieces of blue mineral found in Hearth P2‑005, Locus B0020 and Locus E0189. The latter was analysed by XRD; it turned out to be azurite. The analysis of the other samples is still in progress: they seem to be linked to the bead industry predating the copper-smelting activities rather than to real metallurgy. The ore sampled from Misdağı 1 and 2 contained cuprite and malachite for Misdağı 1 and only cuprite for Misdağı 2.

Ceramic tools

36The corpus is divided into three categories: nozzles, crucibles and moulds. Two nozzle and two crucible fragments were found at Ovçular Tepesi. Otherwise, three mould fragments come from Kültepe I, three from Zirinçlik and another one from Ovçular Tepesi. We may note that the mould from Kültepe I is dated to the Early Bronze Age Kura-Araxes period, in contrast to those from Ovçular Tepesi and Zirinçlik, which are dated to the Late Chalcolithic period.

37Nozzles have an important function in smelting and melting operation during metallurgical processes. These technical tools create the necessary conditions for elevating the temperature of the furnace, which is necessary for the reduction of the ore or the manufacturing of metal objects. The nozzles from Nakhchivan are very similar to those found throughout the Caucasus on Kura-Araxes sites, for example in the metallurgical workshop of Amiranis Gora (Chubinishvili 1963, pp. 94‑103, tab. XXIV, 3), which must have been used during the 4th millennium BC (Kavtaradze 1999, pp. 73‑74; Chernykh, Avilova, Orlovskaja 2000, pp. 74‑75). Clay nozzles were also found at Kültepe II, Baba Derviş II and Mişarçay (Makhmudov, Munchaev, Narimanov 1968, p. 19, fig. 4.2). The examples from Baba Derviş and Mişarçay are similar in shape and size to nozzle 5292 from Ovçular Tepesi.

38Crucible 8087 from Ovçular Tepesi reveals interesting information on its use. After careful examination, we concluded that it was heated from above, which resulted in the heavy vitrification of the inside surface, but no alteration of the exterior surface. The practice of heating the ore from above is often attested on crucibles belonging to the Chalcolithic and the Bronze Age (Thornton, Rehren 2009, p. 2710). The crucible from Ovçular is a good example of this technique, which shows that the practice of ore-smelting from a crucible was performed on that site at least during the Late Chalcolithic (Gailhard et al. 2017, pp. 12‑13).

39The preliminary study of the moulds of Zirinçlik and Kültepe I shows that we are faced with two types of moulds: ceramic open moulds in the case of Zirinçlik (fig. 7), ceramic open moulds (fig. 8), and stone or ceramic bivalve moulds in the case of Kültepe I (fig. 12).

Fig. 12 – Moulds from Kültepe I both made of stone; left: an open mould; right: a bivalve axe-mould (MBA, N. Gailhard).

Slags

40Very few slags are attested in the corpus: one piece comes from Ovçular Tepesi and there might be another one from Zirinçlik. The slag from Zirinçlik is still under study. But the example from Ovçular Tepesi (Locus 8009‑6) is very interesting because it contains information of past pyrometallurgical activities, showing the components that contributed to its formation. It was found together with malachite ores. These pieces of malachite do not seem to have been used, contrary to slag 8009‑6. We are here in front of some slag or the semi-product of a smelting process. Its appearance and the phase assemblage, observable by electronic microscope, differ from the rest of malachite ores (fig. 13). The presence of cuprite (Cu₂O) and delafossite (CuFeO₂), as shown by the XRD analysis, indicates a semi-product obtained through a limited copper-ore smelting process. The presence of a small inclusion of a Cu‑rich slag may be the first outcome of the smelting operation, perhaps in a crucible, in a slightly reducing atmosphere (fig. 13).

Fig. 13 – Semi-product from Ovçular Tepesi with copper prills (US 8009‑6); left: small prills of copper inside the matrix; right: detail of a copper prill from the same slag, this piece may have been obtained after a smelting process in an open crucible (MBA, N. Gailhard).

41This semi-product may be considered as the result of extractive metallurgy carried out at Ovçular Tepesi. It could result from co-smelting with polymetallic ores (Rostoker, Dvorak 1991). Comparable conclusions were reached by Thornton when he examined two pieces of slag from Tepe Hissar, which have been dated to ca 3600 BC (Thornton 2009, p. 314; Thornton, Rehren, Pigott 2009, pp. 311‑312).

Object composition and structure

42The chemical analyses were carried out by Dr M. Bode. Only some of these analyses will be discussed here for the purpose of this paper: the data related to the axes of Ovçular Tepesi and the Kura-Araxes metal objects from Kültepe I (tab. 3).

Tab. 3 – Chemical composition of the axes from Ovçular Tepesi together with some Kura-Araxes objects from Kültepe I; the chemical elements were obtained by ICP‑MS; values are given in parts per million (ppm) except for As and Cu, which are given in weight‑% (wt.‑%); the following elements, which are below the detection limit (Au<2 ppm, Se<7 ppm, Te<10 ppm), are not listed in the table.

43The axes from Ovçular Tepesi are particularly important because of their early date (ca 4320‑4040 BC). Two of the flat axes (Loci 17002/2 and 5007) are made of pure, unalloyed copper. Arsenic, nickel or lead are insignificant in these examples. By contrast, the three other axes display a high quantity of copper associated with a notable content of arsenic, between 1 and 2% (tab. 3, lines 1, 3, 4). The axe-hammer n°17002/3 contains 1.95% arsenic, which is comparable to the arsenic content in the flat axe from 12079. The second flat axe from Locus 17002 (17002/1) has 1.05% of arsenic. We may venture to call this an ‘“accidental” alloy (Yener 2000, p. 29) or “dirty copper” as A. Hauptman called it, that is alloy resulting from the deliberate choice of ore types, but not from the artificial addition of arsenic to the copper ore.

44At Kültepe I, the composition of the metal objects retrieved from Kura-Araxes contexts is somewhat different. The percentage of arsenic is more important (<2%). This fits well with the average content of arsenic found in the artefacts from other Kura-Araxes sites (Courcier 2014, p. 640). This suggests a possible change in technical processes or different sources of copper ores with arsenic content. But another hypothesis can also be raised, i.e. the possibility of variations in the content of arsenic on different sites, as suggested by the earrings (Locus 11285) found in a Kura-Araxes tomb at Ovçular Tepesi, which contain a lower content of arsenic (0.74% only). Variations in arsenic content may of course also be linked to the object’s function (jewellery, tools). Studies of large corpuses of metal objects have indeed revealed a deliberate choice in the concentration of arsenic depending on the type of objects: at Ikiztepe in Anatolia, where more than 370 artefacts dated to the Early Bronze Age have been analysed, arsenic concentrations are higher in weapons than in other groups (Özbal et al. 2002, p. 42; Gailhard 2009, pp. 81‑83). We can observe a similar phenomenon at Arslantepe, where differences in arsenic content have been observed between swords, spearheads and other objects (Caneva, Palmieri 1983, pp. 643‑644; Gailhard 2009, pp. 83‑85).

45Metallographic analyses carried out on several objects from Ovçular Tepesi shows that the practice of open-cast moulding is attested on this site; a practice that has also been evidenced at Zirinçlik and Kültepe I, through the analysis of the moulds.

46Annealing, hot and cold hammering and the use of copper sheets were also known practices, as illustrated by two well-preserved, spiral-shaped earrings (fig. 14). They were found in a Kura-Araxes tomb (Grave 11285) from Ovçular Tepesi. One of the earrings is broken in three pieces but the other shows that the spiral was crafted with a single copper piece.

Fig. 14 – Ovçular Tepesi, earrings from a Kura-Araxes tomb, US 11285 (MBA, N. Gailhard).

47We proceeded to a 10‑second etching with ammonium persulfate solution (fig. 15). A different structure between the surface and the centre appeared, indicating important hammering on the surface. The grain size is smaller at the periphery than at the centre. This also characterizes hot annealing work with recrystallization, as can be seen on the following image with a x20 magnification (fig. 15, right photo).

Fig. 15 ‑ View of the microstructure with traces of hammering on the surface (upper photo) and annealing (lower photo), as shown by the presence of recrystallized and twinned grains (Scoot 1991, p. 8; MBA).

Provenance analysis

48Different approaches were combined to determine the possible provenance of the ores and the objects found on the four archaeological sites. Typological, geochemical and isotope analyses were carried out on the metal objects, on the ore from archaeological sites, as well as the ore sampled from local deposits. An important point is the presence of molybdenum (Mo) inside the different copper minerals from Ovçular Tepesi. This element is also present in small quantities in the copper sampled from the deposit of Misdağı 1. In point of fact, the ore deposits of Nakhchivan and southern Armenia have all been identified as CuMo‑deposits by geologists; they are part of the copper and molybdenum ore deposits of the Zangezur ranges (Aliev et al. 1974, p. 155). During our survey in 2016, we sampled the deposit of Ağdere in eastern Nakhchivan, where one of the former Soviet molybdenum and copper-ore mines was located.

49Lead isotope analyses (LIA) were also carried out to determine the comparative lead isotope ratio content between metal objects and natural ores. Thirty-seven samples from two types were analysed all together by MC‑ICP‑MS (tab. 4). The results are shown in the ²⁰⁸Pb/²⁰⁶Pb versus ²⁰⁷Pb/²⁰⁶Pb diagrams (fig. 16).

Tab. 4 – Lead isotope data of the ores from Misdağı 1 and 2, of the metal objects from Ovçular Tepesi and Kültepe I, as well as of the ores from Ovçular Tepesi and Zirinçlik, by MC‑ICP‑MS with 2σ absolute standard deviation.

Fig. 16 – Lead isotopic results of the ores and metal objects plotted into the ²⁰⁸Pb/²⁰⁶Pb versus ²⁰⁷Pb/²⁰⁶Pb reference diagram; the diagrams are restricted to the range of isotope values of the samples.

50First, we may note a point cluster at the centre of the graphic, which testifies to interesting similarities between the ores and the objects from Ovçular Tepesi, the ores from Misdağı 1 and Zirinçlik and one artefact from Kültepe I (spearhead 7942). This suggests a local origin for all these artefacts, which were probably crafted with some ore from Misdağı I or from the same mountain range: the Zangezur.

51Secondly, at the periphery of the first group, one may note a number of artefacts showing that Ovçular Tepesi was supplied by a diversity of copper ores. The ore from Locus 23063, Locus 6468 and Locus 6530 have very specific lead isotope signatures. As far as the ore from Locus 23063 is concerned, it is close to the copper-ore sample from Misdağı 2, while the ore from Locus 6530 is close to the composition of an awl found at Ovçular Tepesi from Locus 6320 in the same excavation square.

52Lastly, a high ratio of ²⁰⁸Pb/²⁰⁶Pb versus ²⁰⁷Pb/²⁰⁶Pb characterize some of the objects from Kültepe I (ring 7701 and a needle 8062), as well as a flat axe (12079) and a bracelet (Locus 17000) from Ovçular Tepesi. The significance of these ratios is not clear yet; it may reflect different practices like co-smelting or the re-melting of several copper objects together.

53All these data show that the ore was obtained locally, but not from a single source. Late Chalcolithic communities seem to have extracted their copper ore from different deposits. As far as the Early Bronze Age is concerned, the high ratio of ²⁰⁸Pb/²⁰⁶Pb versus ²⁰⁷Pb/²⁰⁶Pb in the Kura-Araxes objects from Kültepe I suggests the use of other copper deposits, together with tin, in order to obtain bronze. At Ovçular Tepesi, the Kura-Araxes earrings from Tomb 11285 also have a very high ratio of ²⁰⁸Pb/²⁰⁶Pb versus ²⁰⁷Pb/²⁰⁶Pb, totally different from that of the Late Chalcolithic copper objects of the same site. This raises the question of the origins of the copper supplies: does the copper used by Kura-Araxes copper and bronze smiths come from another region, different from the one visited by Chalcolithic groups?

Discussion

54The results presented above demonstrate the presence of sustained metallurgical activities during the Chalcolithic and the Early Bronze Age in Nakhchivan, from which it appears that extractive metallurgy was not restricted to large sedentary sites in the valleys. Quite on the contrary, the assemblage under study is suggestive of a keen knowledge and of regular metallurgical practices on a variety of sites located both in valley and mountainous environments: it offers a good insight into the variety and technological organisation of early extractive metallurgy in the South Caucasus. The presence of ore, slag, and crucibles demonstrate the practice of sophisticated metallurgy at Ovçular Tepesi from the middle of the 5th millennium BCE onwards: it is clear that South Caucasian extractive metallurgy was much more developed during the Late Chalcolithic period than once thought, while the discovery of copper ores, a fragmented nozzle and mould pieces at Zirinçlik suggests that similar activities were carried out more or less at the same time in mountainous areas as well.

55The elemental analyses indicate a particular preference of the Ovçular metalworkers for malachite ores, which were most certainly collected from the local deposits of the Zangezur mountains. Similar ores, which are quite distinctive because of their molybdenum content and their lead isotope signature, were also found at Zirinçlik, which suggests that smelting activities during the Late Chalcolithic took place in pastoral camps, as well as large valley settlements. During the Early Bronze Age, on valley sites such as Kültepe I, metallurgical activities were also performed as shown by the presence of moulds, but the bronze smiths seem to have used ores from other copper sources. Unfortunately, not much more is known altogether about metallurgical practices at Kültepe I, since most of the Kura-Araxes occupation levels were dug away by the Soviet excavations – the Kura-Araxes levels that could be investigated by the French-Azerbaijani team were very limited in size.

56Judging by the analysis of metal objects, however, there are notable differences between the Late Chalcolithic and the Kura-Araxes metallurgical procedures, as shown by the presence of a percentage of tin in some of the Early Bronze Age objects from Kültepe I, and a specific lead isotopic signature for the EBKA earrings of Ovçular Tepesi – which differs from the signature of the Late Chalcolithic artefacts. Few objects of the Kura-Araxes period are available for analysis at either Ovçular or Kültepe I, but there seems to be a change in the provenance of copper ores between the two periods, while the use of alloys is more and more common in the Bronze Age. Otherwise, metallographic studies confirm the use of well-spread technics such as the high-temperature treatment of copper minerals, annealing, and casting in open and bivalve moulds for all periods.

57All the above data are particularly interesting for analysing the role of mobile groups in the procurement of copper ores during the Chalcolithic and Early Bronze Age. Some form of mobility, probably seasonal transhumance, is certainly attested at Ovçular Tepesi, as shown by the results of the first Sr, Ca and O isotopic analyses carried out on the teeth of caprines and cattle (Berthon et al., this volume). The exact routes followed by the Late Chalcolithic herds from Ovçular, as well as the different pasturelands they actually reached, are not known yet – this information will hopefully be available through the establishment of a reference map of the ⁸⁷Sr/⁸⁶Sr geological values, which is in progress. Iconographic data, however, indicate close ties between the valley site of Ovçular, and the pasturelands of eastern Nakhchivan, as illustrated by the caprines depicted on the jars from Ovçular Tepesi and the petroglyphs of Gemikaya (fig. 17). This is all the more interesting as the rich pasturelands of eastern Nakhchivan, which are still used by transhumant shepherds today, are located in the vicinity of major copper-ore deposits, as we noted during the survey we carried out in 2016.

Fig. 17 – Late Chalcolithic jar from Ovçular Tepesi, Phase I, Locus 12066, and petroglyph from Gemikaya (MBA; left: N. Gailhard; right: V. Bakhshaliyev).

58All these data put together suggest that the Nakhchivani copper ores were extracted by mobile groups and brought over to valley sites in the wake of seasonal transhumance. The available data is not sufficient for us to say whether the procurement of copper ore was operated by shepherds from valley sites that would take their flocks (or part of their flocks, see Berthon et al., this volume) in spring to spend the summer in the highlands. Or whether this activity was carried out by independent mobile groups, who would bring copper ore to valley communities on coming back from pasturelands in the autumn. Both systems may well have coexisted: the available data from Ovçular suggest that part of the valley community may have spent their summers in the mountains, while the evidence from Zirinçlik, a site that is partly contemporary with Ovçular, but has a very distinctive pottery assemblage, indicates that this mobile group was certainly independent from the Ovçular community.

Conclusion

59As seen from the above, issues related to the development of metallurgy, the circulation of technological knowledge, and the production of sophisticated metal objects can only be examined through the combination of archaeometallurgical, archaeometric and socioeconomic data.

60This paper was focussed on the Chalcolithic period and Early Bronze Age in Nakhchivan, when the mobility of people and the circulation of raw materials seem to have been an inherent part of the metal production. The diversity of the sites – permanent settlements, semi-permanent settlements, seasonal camps in the foothills – combined with different categories of metallurgical artefacts (objects, ores, crucible, moulds…) give us the opportunity to catch a glimpse of the organisation of the early metal production.

61As a tentative hypothesis, we argue that the development of extractive metallurgy in the Lesser Caucasus was linked to mobility and the spread of vertical pastoralism. On reflecting on the diffusion of metallurgical practices4 (Lis, Rückl, Choleva 2015, p. 63), it is usual to study the geographical extent of their diffusion through the trajectory of artefacts over long distances; but the mobility of metallurgists is also a criterion that should be taken into account. Some of the tools produced in Nakhchivan during the Late Chalcolithic display strong links with artefacts from distant regions such as south-eastern Europe, but the Nakhchivani artefacts were made with local ores, while the whole chaine opératoire seems to have been mastered by the domestic community: it is possible that early metallurgical techniques were imported to the South Caucasus by South European itinerant coppersmiths, a hypothesis that has been put forward, among others, by Chernykh (1992). At the same time, the work carried out in Nakhchivan demonstrates that the highlands had started to be exploited by local communities for stock-breeding and probably for obtaining raw materials, among which copper ores, at least from the Chalcolithic onwards: the proximity of certain copper deposits to water-rich pasturelands, as well as the practice of metallurgy on pastoral campsites, as evidenced at Uçan Ağıl and Zirinçlik, suggests the existence of strong ties between pastoralists and early metallurgists, who in certain cases might even have been the same persons.

62In this period of nascent interactions between human groups and their mountain environment, and considering the available data, it is still difficult, however, to decide whether Caucasian extractive metallurgy was developed by pastoral nomads or whether several forms of mobility, including nomadic pastoralism, were in fact favoured by the rise of metallurgy.