Introduction
The Eurasian Steppe communities are now widely recognised as key drivers of Bronze Age transformations across the broader continental landscape (c. 3500–1000 BC), shaping patterns of population movement and broader cultural change (Chernykh Reference Chernykh1992; Anthony Reference Anthony2007; Librado et al. Reference Librado2021). Among these, the intensification of copper and bronze production played a pivotal role in shaping social and economic systems from the Atlantic to the Pacific (Vandkilde 2016). Steppe communities exploited abundant ore resources and responded to inter-regional demand for copper-based goods (Stöllner & Samašev Reference Stöllner and Samašev2013; Chernykh Reference Chernykh2017).
Prevailing interpretations often portrayed steppe societies as culturally and technologically homogeneous, a view now challenged by extensive research in the area (e.g. Frachetti Reference Frachetti2012; Narasimhan et al. Reference Narasimhan2019). Metallurgical debris, such as slag, technical ceramics and production waste, provides a robust and high-resolution record of technological practice, preserving evidence of operational sequences and technical choices that is often inaccessible through the study of finished metal artefacts alone. Despite significant advances in the study of artefacts and exchange networks (Stöllner & Gontscharov Reference Stöllner and Gontscharov2020; Artemyev et al. Reference Artemyev2024), the dynamics of production, its variability and innovation processes are still insufficiently understood. Building on frameworks such as Frachetti’s (Reference Frachetti2012) concept of ‘non-uniform complexity’, our study addresses this gap by placing production contexts, and the technological decisions embedded within them, at the centre of analysis.
This article introduces a substantial new dataset from the Bronze Age settlement of Taldysai (1900–1600 cal BC) in central Kazakhstan, including slags, minerals and metal debris recovered from the metallurgical workshops at Taldysai, and geological minerals from the nearby mine field of Dzhezkazghan, for a total of 50 samples. Through reconstruction of the chaîne opératoire, we demonstrate the sustained, parallel production of pure copper, arsenical copper and Cu-As-Sn (copper-arsenic-tin) polymetallic alloys over nearly three centuries by Petrovka and Alakul/Fedorovo communities. Building on earlier work (Calgaro et al. Reference Calgaro2023), our analyses of ore inputs and smelting recipes reveal a specialised, adaptable and scalable metallurgical system. These results underscore the role of steppe metallurgists as key innovators in inter-regional metal exchange and, potentially, in the broader transmission of metallurgical technologies.
Taldysai in its geological and archaeological context
Taldysai lies in the Ulytau district of central Kazakhstan (48°20′41.38″N, 67°0′6.76″E), near the Zhezdy River and 70km from the Dzhezkazgan copper-mining district (Figure 1). The Dzhezkazgan ore field spans approximately 100km2, with copper sandstone mineralisation, primary sulfides (pyrite, chalcopyrite) and secondary sulfides (bornite, chalcocite) overlain by copper carbonates and oxides (Box et al. Reference Box2012). Archaeological investigations at Taldysai began in 1994, revealing three metal production areas (excavations 1–3) (Kurmankulov et al. 2012; Yermolayeva et al. Reference Yermolayeva2020a).
A) The location of Taldysai and Dzhezkazgan (map by authors); B) aerial view of the modern village and the eponymous Bronze Age site of Taldysai, marked by a blue dot (adapted from Kurmankulov et al. 2012: 8).
Chronology
These three metallurgical areas formed the core of metal production activity at Taldysai between 1900 and 1600 cal BC (Figure 2A) (Hermes et al. Reference Hermes2020). Stratigraphic evidence, ceramic typology and associated metallurgical materials corroborate this chronological framework, allowing the site’s operational span to be defined with more confidence.
A) Aerial view of the Bronze Age site of Taldysai. Excavation 1 includes eastern (a), western (b) and northern (c) complexes; B) vaulted pit furnace with horizontal canal (northern complex); C) above-ground bowl-shaped furnace (eastern complex). The large scale bars in B and C are 100mm each segment (adapted from Yermolayeva et al. Reference Yermolayeva and Kurmankulov2020b: photographs 3, 5 & 21).
Stratigraphic and material culture analyses (Yermolayeva et al. Reference Yermolayeva2020a) outline three major occupation phases. The earliest corresponds to the Petrovka (Nurtai) tradition, marking the establishment of the settlement and the first metal production activities. This is most clearly represented in excavation 1, where ceramic typology indicates that the northern metallurgical complex dates to 1900–1700 cal BC (Yermolayeva et al. Reference Yermolayeva2018; Hermes et al. Reference Hermes2020).
A second phase reflects continuity and cultural transformation through the Alakul (Atasu) and Fedorovo (Nura) traditions. The western and eastern complexes in excavation 1 contain mixed Petrovka and Alakul/Fedorovo materials, indicating sustained occupation and metal production between 1800 and 1600 cal BC.
After 1600 cal BC, the archaeological record shows marked socioeconomic restructuring. Metallurgical installations were sealed and abandoned, and the site was reorganised with round enclosures associated with the Sargary-Alekseevka cultures (1600–1400 cal BC) (Hermes et al. Reference Hermes2020). No furnaces from this phase have been identified, and the presence of only a few metal objects suggests limited, possibly secondary, metallurgical activities such as casting (Yermolayeva et al. Reference Yermolayeva2019).
Two burials associated with this final Bronze Age phase were recovered. One burial, unearthed from a cultural context dated to 1600–1400 cal BC, contained two adult males who had been interred near a furnace and channel system in the eastern complex, suggesting a possible symbolic link between metallurgy and mortuary practice. A second burial contained a child; the remains were radiocarbon dated to 1379–1196 cal BC and were found in a disturbed trench alongside Alakul ceramics, though stratigraphic mixing complicates cultural attribution (Narasimhan et al. Reference Narasimhan2019; Hermes et al. Reference Hermes2020).
Archaeological, cultural and chronological evidence from Taldysai strongly supports the interpretation of sustained metallurgical activity over approximately 300 years. Material culture from the site spans three successive Bronze Age horizons: Petrovka (Nurtai), Alakul/Fedorovo and Sargary-Alekseevka (Yermolayeva et al. Reference Yermolayeva2020a). The presence of these sequential occupations, which correspond with well-established regional developments across northern Eurasia, indicates continuous or repeatedly renewed occupation and metallurgical activity rather than a single, intensive, short-term phase.
Characterising metallurgical debris at Taldysai
The site has yielded a remarkably diverse assemblage of metallurgical remains, reflecting multiple operational steps, recipes and furnace types. Two distinct furnace types were identified in excavations 1, 2 and 3 (Figure 2B & C), spanning Petrovka and Alakul/Fedorovo contexts.
The first, and technologically most original, type is the vaulted pit furnace (‘shatnovo tipa’). Examples of this type are identified at the site as deep installations, reaching down to 2m, with radial ducts carved into their inner walls. These pits were originally enclosed by domed superstructures built from stone slabs and plaster, many of which collapsed inward over time. Each is accompanied by a horizontal channel (extending up to 8m); experimental reconstructions support a functional interpretation of this channel as being engineered to enhance airflow and redirect toxic fumes (Yermolayeva & Rusanov Reference Yermolayeva and Rusanov2022). While similar in design to smaller Sintashta-type furnaces, the Taldysai vaulted pit furnaces are notable for their scale and complexity. No parallel examples are known outside the Eurasian steppes and the Bronze Age (Kuznetsova & Teplovodskaya Reference Kuznetsova and Teplovodskaya1994), marking them as a unique and advanced technological innovation of the steppe zone.
The second type comprises above-ground, bowl-shaped furnaces (‘nadzemnovo tipa’), more common and widely distributed across the Sintashta, Petrovka and Alakul traditions. Often located near the vaulted pit furnaces, they are associated with ash, soot, burnt bone and metallurgical debris. No evidence for preferential distribution or spatial organisation of the two smelting furnace types was identified across the site.
Slags and archaeological copper minerals are among the most abundant finds, followed by technical ceramics such as crucibles and their fragments. This article focuses on the analysis of slag and mineral samples, which provide the most direct evidence for metal production at the site. Technical ceramics, which have been examined in detail as part of the first author’s doctoral research (Calgaro Reference Calgaro2025), will be addressed in a separate publication. Here, we focus specifically on reconstructing the smelting recipes and metallurgical practices associated with metal extraction during the Petrovka (1900–1700 cal BC) and Petrovka-Alakul/Fedorovo (1800–1600 cal BC) phases.
Materials
Two main slag types were identified, each reflecting specific steps in the chaîne opératoire or different smelting inputs. Amorphous slags are dense, irregular, brown fragments, sometimes oxidised to red or green, with visible copper prills (metallic copper droplets entrapped within the slag matrix; Figure 3A), resulting from the smelting of mixed secondary copper sulfides and carbonates. Flat slags are thin, vesicular and dark brown, with copper oxide patches (Figure 3B & C), linked to smelting copper carbonate ores or co-smelting copper ores with arsenic-rich minerals like arsenopyrite. Vitrified and slagged clay fragments (Figure 3D & E) likely derive from the furnace linings of above-ground installations.
Typical metallurgical slags from Taldysai: A) amorphous (BAE 37); B & C) flat and thin, with bubble imprints preferentially formed on one side (B: BAE 45; C: BAE 40); D & E) amorphous lump of clay showing a slagged (D) and a reddened side (E) (BAE 138a) (figure by authors).
The Taldysai copper mineral assemblage includes malachite and azurite (Figure 4A & B), likely residual or partially processed ores. We also analysed geological samples, including blue minerals and chalcopyrite (Figure 4C & D), from the Dzhezkazgan ore field at Satpayev (47°54′01.3″N 67°30′53.6″E) held in the archive collection at the Margulan Institute of Archaeology to help contextualise the archaeological mineral remains identified at Taldysai.
Archaeological minerals from the Taldysai workshops (A & B) and geological mineral samples from the Dzhezkazgan ore field (C & D). A) Green malachite (BAE 60); B) blue azurite (BAE 157); C) fragment of blue-grey mineral composed of copper carbonates and sulfides (BAE 198a); D) copper-iron sulfide of the chalcopyrite type (BAE 197) (figure by authors).
Geological samples have previously been used in experimental smelting in vaulted pit furnaces modelled on the examples found at Taldysai (Yermolayeva & Rusanov Reference Yermolayeva and Rusanov2022). Our study thus combines analysis of geological minerals collected at the Dzhezkazgan mine (n = 2), archaeological minerals found at Taldysai (n = 8), slags (n = 23) and associated metal debris (n = 17), and compares the results with previous research to provide refined insights into metallurgical choices and operational sequences across 300 years.
Methods
Samples (minerals, slags, metal debris) were assigned sequential Bronze Age Eurasia (BAE) ciphers (Table S1). Initial portable x-ray fluorescence screening, magnetic tests, epoxy embedding and polishing preceded optical microscopy (Leica DM4500). Scanning electron microscopy with energy-dispersive x-ray spectroscopy (SEM-EDS) was conducted using a Zeiss EVO with Oxford Instruments X-Max80. All data presented in the main text have been corrected with factors based on certified reference materials. Full protocols are available in the online supplementary material (OSM).
Results
Compositional and microstructural analyses of the samples reveal two distinct operational systems: the primary extraction of pure copper (production line A) and the production of arsenical copper and Cu-As-Sn alloys (production line B), both dating to the suggested 300-year-long span (1900–1600 cal BC).
Production line A employed two copper smelting recipes. The dominant method used chalcocite-type sulfide ores (Figure 4C), potentially mixed with copper carbonates (Figure 4A & B), producing amorphous slags (Figure 3A), most likely in vaulted pit furnaces (Figure 2B). The less common method involved smelting malachite and azurite without sulfides, yielding flat slags (Figure 3B & C), likely within above-ground furnaces (Figure 2C). Iron-rich ores such as chalcopyrite (Figure 4D) were not primary inputs. Production line B corresponds to the smelting of arsenic-rich sulfureous ores, combined with either copper carbonates or copper sulfides to produce Cu-As-Sn alloys. The smelting occurred in two steps: desulfurisation (step 1, reflected in flat slags and slagged clay fragments; Figure 3B–E) and iron removal (step 2, as evidenced by one flat slag). Tin bronze metal was added to the charge during step 1 but no tin ore was smelted.
The two geological mineral samples from the Margulan Institute of Archaeology archive (BAE 197 & BAE 198a) align well with the mineralisation of the Dzhezkazgan, characterised by copper carbonate and sulfides and iron sulfides in sandstone and shale. BAE 197 (Figure 4D) consists of chalcopyrite with galena, rutile and chalcocite inclusions, typical of copper sandstone deposits (also, Rovira Reference Rovira1999). BAE 198a (Figure 4C), catalogued in the archives as chalcocite, was confirmed via SEM-EDS to be a low-sulfur copper carbonate.
The eight archaeological ores collected from Taldysai contain malachite, azurite, copper sulfate, quartz, iron oxides and chalcocite, also aligning with the mineralisation found in Dzhezkazgan (BAE 41, 46, 60a–k; Figure 4A & B; see also OSM).
Slag analysis confirms two main copper recipes corresponding with production line A: the most common used chalcocite-type sulfides with copper carbonates, producing amorphous slags; the second used pure carbonates (malachite and azurite), yielding flat slags. Our data suggest chalcopyrite was not a major ore source. Dark blue ores such as BAE 198a appear to have been smelted in vaulted pit furnaces, offering more efficient returns than lower-grade carbonates.
Production line A: recipe 1
SEM-EDS microstructural and chemical analyses show that both amorphous and flat slags from Taldysai relate to copper production (production line A) and derive from two distinct smelting recipes using different copper ore types. These compositional differences likely reflect the use of different furnace types, as seen in slag morphology. Both slag types appear in all excavation contexts, including Petrovka and Petrovka-Alakul/Fedorovo phases (Table S1). Of the 20 newly analysed copper slags, 19 are amorphous, the dominant type. Only one flat slag (BAE 174) is identified. Published data from two more copper flat slags also found at Taldysai are included for comparison (Ankushev et al. Reference Ankushev and Kurmankulov2020).
The amorphous slags are chemically distinct and result from smelting ores similar to geological sample BAE 198a (Figure 4C), as verified through experimental reconstructions (Yermolayeva & Rusanov Reference Yermolayeva and Rusanov2022). These slags contain high silica (SiO2, ∼60wt%), moderate alumina (Al2O3, ∼14.5wt%) and low iron oxide (FeO, 4–10wt%) (Table S15). Their fully liquified matrix includes metallic copper prills, oxides, matte and minor sulfide inclusions (Figure 5A). Newly formed cuprite, delafossite and occasional iron-manganese spinels suggest smelting in a mildly oxidising atmosphere (Table S16). All prills are high-purity copper with low iron (<3.3wt%) and trace cobalt, arsenic, silver and lead (<0.9wt%) (see Table S17), excluding chalcopyrite as a source. Experimental smelting of BAE 198a in vaulted pit furnaces produced slags matching the archaeological material (Calgaro & Radivojević Reference Calgaro, Radivojević, Yermolayeva and Rusanov2022). Although such ores were not found onsite, their low sulfur trioxide (SO3) content (21.1wt%) supports the hypothesis of co-smelting with secondary sulfides to generate matte and facilitate copper oxide reduction (see OSM).
Photomicrographs (plane polarised light/crossed polarised light, 100–200×) and SEM-EDS backscattered electron imaging of amorphous and flat slags from production line A (pure Cu metal) from Taldysai. A1–3) Amorphous slags composed of glassy matrix (gl), partially reduced chalcocite (ch) and bornite (bor), copper prills (Cu) and cuprite (Cu2O). B1–3) flat slag composed of glass (gl), delafossite (del), cuprite (Cu2O), magnetite (mg) and copper prills (Cu). Samples: BAE 50a, 61a, 53, 174. See Tables S15 and S16 for details (figure by authors).
Production line A: recipe 2
The flat slags (BAE 174 and two from Ankushev et al. Reference Ankushev and Kurmankulov2020) reflect copper-carbonate smelting. BAE 174 shows similar ratios of SiO2 and copper oxide (CuO), elevated lead oxide (PbO), Al2O3, manganese oxide (MnO), low FeO, and newly formed mineral phases pointing to oxidising conditions (Figure 5B; Tables S15 & S16), suggesting production in above-ground furnaces. Significant readings of PbO in the composition of copper flat slags reflect the occurrence of galena (lead ore) in the mineralisation of Dzhezkazgan.
Production line B: arsenical copper and Cu-As-Sn
Metallurgical debris associated with the production of arsenical copper and Cu-As-Sn alloys (production line B) includes flat slags (BAE 29, 39b, 40, 45) and slagged clay fragments (BAE 137, 138a) (Figure 3B–E). These materials are found across all excavated contexts at Taldysai, including Petrovka and Petrovka-Alakul/Fedorovo workshops, reflecting the distribution of copper slags from production line A. Two smelting steps are identified. Step 1 (desulfurisation) is reflected in BAE 39b, 40, 45, 137 and 138a, with lower SiO2 (average 39.5wt%), higher FeO (average 33.2wt%), increased lime (CaO, average 10.3wt%) and reduced Al2O3 (≤9.4wt%) (Table S15). BAE 138a is an exception, with high SiO2 (65.3wt%) and low FeO (4.7wt%), likely representing a slagged furnace fragment. Mineralogically, the flat slags produced within step 1 contain a fully vitrified matrix with quartz, fayalite, pyroxene and augite (Figure 6A; Table S16). Matte and copper-rich speiss (iron arsenide) are common and metallic prills show arsenic up to 6.4wt% and tin between 0.2 and 20.2wt%, suggesting intentional Cu-As-Sn alloying. Although the separate production of speiss at Taldysai cannot be entirely excluded, given the absence of in situ archaeological speiss and the possibility that it was later added to copper metal (as proposed by Rehren et al. Reference Rehren2012), our evidence does not support this scenario. In line with previous analyses from Taldysai, including arsenical copper slags (Ankushev et al. Reference Ankushev and Kurmankulov2020; Calgaro et al. Reference Calgaro2023), all analysed slags contain copper. This contrasts with the distinct ‘speiss slags’ documented at Chalcolithic and Early Bronze Age sites such as Tepe Hissar (late fourth millennium BC) and Arisman (3100–2900 BC), respectively (Thornton et al. Reference Thornton2009; Rehren et al. Reference Rehren2012). Chromite mineral inclusions were documented in sample BAE 137 (see OSM). Chromite represents a relict phase, most likely related to the use of ores hosted in ultramafic rocks, and has been documented in slags from Mid to Late Bronze Age settlement and mining contexts in the Urals (c. 2000–1700 BC) (Zaykov et al. Reference Zaykov2012; Asmus Reference Asmus, Koryakova and Krause2021), hinting at a network of raw materials supply and metalmaking knowledge extending through the Ural steppes into central Kazakhstan, to be explored elsewhere.
Photomicrographs (plane polarised light, 50×) and SEM-EDS backscattered electron imaging of flat slags and slagged clay lumps from production line B (Cu-As and Cu-As-Sn alloys). A1–3) Step 1: fayalite (fa), magnetite (mg), augite (aug), clinopyroxene (clpx) and Cu-As prills with chalcocite (ch) and arsenic-rich phases (Cu3As) (samples: BAE 39b, 40, 137). B1–3) Step 2: notable is the presence of wüstite (wü) and absence of sulfur phases associated with Cu-As prills (sample BAE 29). See Tables S15–S16 for details (figure by authors).
Step 2 (iron removal), represented by BAE 29, produced slag with extremely high FeO (66.2wt%) and no sulfides or speiss, suggesting targeted iron removal from arsenopyrite or chalcopyrite ores (Figure 6B and OSM).
These data indicate that arsenical copper production at Taldysai likely relied on arsenic-rich ores such as tennantite (a copper sulfarsenide) and arsenopyrite (an iron sulfarsenide), possibly accompanied by scorodite, a weathering by-product of arsenopyrite. Although such ores have not been recovered at the site, they are documented in the deeper primary deposits of the Dzhezkazgan region (Satpaeva Reference Satpaeva2007), and their use aligns with known regional mineralisation patterns. These arsenic-bearing ores were likely co-smelted with chalcopyrite or copper-iron secondary sulfides. Occasional additions of metallic tin bronze into the charge to produce Cu-As-Sn alloys are also consistent with previous analysis (Ankushev et al. Reference Ankushev and Kurmankulov2020). The flat morphology of most arsenical slags suggests their formation in shallow, controlled smelting environments, under reducing conditions (e.g. in above-ground furnaces or crucibles), as indicated by mineral phases like fayalite and wüstite.
Additional support comes from Park’s (Reference Park and Kurmankulov2020) analysis of finished artefacts at Taldysai, which include pure copper, arsenical copper, Cu-As-Sn ternary alloys and tin bronzes. If these artefacts were produced onsite, it was either by adding tin to the smelting charge or by co-smelting arsenical copper with tin bronze. Our analysis of Cu-As, Cu-Sn and Cu-As-Sn metal debris shows as-cast dendritic structures, with seven out of eight samples exhibiting significant porosity, indicative of casting in oxidising conditions. On the other hand, 113 out of 122 finished metal artefacts from Taldysai, including copper, Cu-As and Cu-As-Sn items, show evidence of post-casting working (Park Reference Park and Kurmankulov2020). Conversely, the pure copper droplets analysed in our assemblage display α-eutectic structures and partial recrystallisation, suggesting slow cooling, and elevated copper oxide concentrations at their margins reflect oxygen saturation during solidification, which enhanced α-grain growth (Figure 7). Optically blue matte phases observed in copper droplets (e.g. BAE 57a, BAE 167) confirm primary smelting activity, likely within vaulted pit furnaces (see OSM).
Biplot comparing metal debris from this study (A: tin bronze; B: pure copper; C: arsenical/Cu-As-Sn alloys) with finished metal artefacts from Taldysai analysed by Park (Reference Park and Kurmankulov2020). Images A–C are photomicrographs of samples BAE 62, 44 and 56, respectively, etched with FeCl3 (plane polarised light, 50×, image width: 3mm) (figure by authors).
Discussion
During the first half of the second millennium BC, copper, arsenical copper and occasional Cu-As-Sn alloys were smelted at Taldysai using well-defined and standardised metallurgical practices that were replicated across three centuries. Our dataset, along with evidence from Ankushev et al. (Reference Ankushev and Kurmankulov2020) and Park (Reference Park and Kurmankulov2020), confirms two distinct recipes for copper production and a separate protocol for arsenical copper. While pure and arsenical copper dominate the assemblage, the presence of Cu-As-Sn alloys appears limited and likely experimental.
At Taldysai, pure copper was primarily produced by smelting a mixture of secondary sulfides and carbonates in vaulted pit furnaces with horizontal channels. Such ores, compositionally akin to BAE 198a from nearby Dzhezkazgan, have not been found at Taldysai but experimental smelting of BAE 198a-type minerals supports this interpretation (Yermolayeva & Rusanov Reference Yermolayeva and Rusanov2022). Resulting amorphous slags show minimal compositional variation (Figures 8 & 9), indicating a highly standardised process over approximately 300 years. A less common but equally enduring recipe involved smelting malachite and azurite in above-ground furnaces. Both furnace types reflect locally adapted Sintashta-Petrovka traditions. Copper was then crucible-refined, cast and worked onsite (Figure 10A).
Principal component analysis of slags from Taldysai (data reported in Table S15). Circles represent samples by Ankushev et al. (Reference Ankushev and Kurmankulov2020), coloured according to the corresponding production line in this study (figure by authors).
Ternary plots of (A) SiO2-FeOtot-Al2O3 and (B) SiO2-FeOtot-CaO for the bulk of copper, arsenical and Cu-As-Sn slags from Middle to Final Bronze Age contexts of the Southern Urals and central steppes. Data averaged and normalised to 100. The values from Taldysai appear in colour. Note how sites spanning the Middle to the Final Bronze Age produced copper, arsenical copper, Cu-As-Sn alloys and tin bronze by combining different ore types according to broadly similar metallurgical recipes. These practices resulted in comparable compositional patterns across sites. The dashed line separates those sites that predominantly used copper carbonates and copper sulfides for smelting copper (upper region of the plot) from those that employed iron-rich sulfidic ores to produce copper and arsenical copper. The production of Cu-As-Sn at Taldysai appears well aligned with that observed in the early Sintashta-Petrovka sites of the Urals (Kamennyi-Ambar, Levoberezhnoe, Ustye). Full data and comparative references reported in the OSM. Amorphous slag samples associated with Cu-As-Sn production from Taldysai (reported by Ankushev et al. Reference Ankushev and Kurmankulov2020) are treated as slagged lumps of clay with composition transitioning between more silica or iron rich, comparable to the sample we report (BAE 138a) (figure by authors).
Chaîne opératoire of A) production line A, copper metal, and B) production line B, arsenical copper and Cu-As-Sn alloys, at Taldysai, 1900–1600 cal BC (figure by authors).
Smelting debris reveals that both Petrovka and Alakul/Fedorovo groups also produced arsenical copper over three centuries at Taldysai, occasionally experimenting with tin to create Cu-As-Sn alloys. Such activity aligns with contemporaneous metalmaking contexts of the Eurasian Steppe and at the early Sintashta-Petrovka sites of the Urals (Figure 9; see also OSM). While no arsenic ores have been found onsite or in the collection from Taldysai, slag composition points to the use of sulfarsenides (e.g. tennantite, arsenopyrite) co-smelted with primary ores such as chalcopyrite or bornite and copper carbonates, possibly from Dzhezkazgan. This production line was characterised by a two-step process of desulfurisation and iron removal (Figure 10B).
Bogdanov (Reference Bogdanov and Makarov2023) recently conducted experimental co-smelting of secondary copper sulfides and carbonates with arsenopyrite, the results of which help to interpret the production of Cu-As alloys at Taldysai. In the experiment, 1570g of Kargaly copper minerals (mineralogically akin to Dzhezkazgan) were smelted in a bowl-shaped, above-ground furnace. The charge included secondary sulfides, carbonates, silica gangue and 150g of crushed arsenopyrite, with 4.5kg of charcoal. This yielded an arsenical copper ingot (∼1.5wt% As) beneath a 4mm flat slag layer. Microstructural analysis revealed bornite-like relict sulfides, copper arsenates, Cu-As droplets and olivine-magnetite lathes, resembling production line B slags at Taldysai. These findings support the view that co-smelting arsenopyrite with secondary sulfides and/or carbonates was the dominant Cu-As alloying method at Taldysai.
Flat slags of production line B (Figure 3B & C) likely formed atop smelting charges. Their high calcium content suggests either interaction with a ceramic lining that differed in composition from the vaulted pit furnaces, or the use of a different fuel mix for smelting (e.g. including bone ash), as seen in the lime-rich slag BAE 138a (Figure 9).
The production of arsenical copper would have entailed considerable health risks due to arsenic fumes. At Taldysai, no distinct spatial clustering of production line B can be identified, but as the site appears to have operated entirely as a workshop without domestic structures, spatial zoning may have been of limited relevance. Studies from Bronze Age Iberia indicate that arsenical copper production often occurred alongside general copper working activities, suggesting that communities managed these hazards through practical workshop organisation rather than dedicated specialist areas (Murillo-Barroso et al. Reference Murillo-Barroso2025). Analyses of crucibles and technical ceramics from the Taldysai workshops suggest that the production of arsenical copper could have been conducted either in above-ground furnaces, as discussed here, or in crucibles (Calgaro Reference Calgaro2025; Dong Reference Dong2025). The use of crucible vessels from Taldysai in metalmaking will be addressed elsewhere. Metallic tin phases recorded across slags of production line B suggest experimental alloying, but not as a regular practice. Instead, arsenical copper and tin bronze may have been co-cast to produce the 10 Cu-As-Sn artefacts identified at Taldysai (Park Reference Park and Kurmankulov2020).
Conclusions
This study offers the first integrated reconstruction of metal production at the site of Taldysai, revealing three centuries of continuity in metallurgical practice during the first half of the second millennium BC. Through detailed analysis of slags, ores, furnaces and metal artefacts, we show how local Petrovka and Alakul/Fedorovo communities sustained and transmitted copper-based alloying technologies across generations.
For the first time, we trace this tradition through the lens of technology itself: the scale and design of the vaulted pit furnaces and the standardisation of slag composition point to operations rooted in deep knowledge across approximately three centuries. We identify two distinct copper smelting practices within production line A, alongside a two-step process for arsenical copper and Cu-As-Sn alloys in production line B. Far from being peripheral suppliers, the metallurgists of Taldysai emerge as expert metalsmiths who combined traditional knowledge with local experimentation. Vaulted pit furnaces, resembling a developmental update on the Sintashta models found in the Urals, were uniquely adapted to local ore compositions and, possibly, to specific fuel availability in the semi-desert conditions of the central steppe (Yermolayeva et al. Reference Yermolayeva2020a). The broader distribution of similar furnaces across the Eurasian Steppe, from the Urals to the Zarafshan Valley (Avanesova Reference Avanesova and Berdimuradov2015), highlights a widespread technological network that fostered innovation in optimising production capacity.
Central Kazakhstan formed a major Bronze Age copper production core, closely linked to rich ore deposits such as Dzhezkazgan. Middle and Late Bronze Age mining and settlement sites, including Kresto, Mylikuduk, Ainakol, Sorkuduk and Zlatoust, cluster around these resources, reflecting an integrated landscape of extraction and production (Kuznetsova & Teplovodskaya Reference Kuznetsova and Teplovodskaya1994). Within this system, Taldysai likely functioned as an intermediary production centre, connecting mining zones with wider distribution networks. The scale of exploitation was considerable, with Dzhezkazgan alone estimated at around one million tonnes of ore (60 000–70 000 tonnes of copper) (Satpayev Reference Satpayev1941). Together with other major mining systems across the central Kazakh steppes, this points to a highly organised network of extraction, production and circulation, within which Taldysai occupied a strategic role.
Metalmaking sites were not isolated from other communities of metal consumers, rather, they exhibited a distinct specialisation in metal production, establishing the Eurasian Steppe communities not as passive recipients but as full participants in the Bronze Age world of metallurgical innovation.
Acknowledgements
The authors sincerely thank the A.Kh. Margulan Institute of Archaeology in Almaty for over a decade of support and access to the Taldysai collection, which made this international collaboration and research possible.
Funding statement
This research was supported through funding from the Institute for Archaeo-Metallurgical Studies (IAMS) and the London Arts and Humanities Partnership awarded to Dr Ilaria Calgaro for the completion of her PhD at the University College London Institute of Archaeology, and by the European Research Council-awarded and UK Research and Innovation-funded project DREAM, ‘Discovering the (R)evolution of EurAsian steppe Metallurgy’, granted to Dr Miljana Radivojević (EP/Z00022X/1) at the University College London Institute of Archaeology.
Author contribution: CRediT categories
Ilaria Calgaro: Conceptualization-Equal, Investigation, Methodology-Equal, Formal Analysis-Lead, Data Curation-Lead, Visualization, Funding acquisition-Supporting, Writing-Original Draft, Writing – Review & Editing-Equal. Miljana Radivojević: Conceptualization-Equal, Methodology-Equal, Formal Analysis-Supporting, Data Curation-Supporting, Funding Acquisition-Lead, Project Administration, Resources-Equal, Supervision, Writing – Review and editing-Equal. Antonina Sergeevna Yermolayeva: Field Investigation, Resources-Equal.
All authors contributed to the interpretation of the results and approved the final version for publication.
Data availability statement
All data available in OSM.
Online Supplementary material (OSM)
To view supplementary material for this article, please visit https://doi.org/10.15184/aqy.2026.10390 and select the supplementary materials tab.
Open access
