Cova Cartanyà is a cave with a small corridor to the right of its entrance, followed by a main chamber and a chasm. Pottery and animal bones were found at the corridor. The main chamber yielded pottery (including Bell Beaker pots), bone objects, faunal remains, stone tools, and sporadic human bones. The limited contextual information available prevents further specifics, but its excavator assigned a chronology between the Late Neolithic-Chalcolithic and the Late Bronze Age based on artefact typology (Vilaseca Anguera, 1926, 1973), and more recent research concurs (Miró, 1995; Martín Cólliga et al., 1999; Rovira and Ambert, 2002). The materials recovered suggest that the cave may have been sporadically inhabited and that, despite the presence of human remains, it was not primarily used for burial.

The crucible analysed, held by the Museu d’Arqueologia Salvador Vilaseca (Reus), was not initially recognised during excavations, but later found during a revision of materials (Martín Cólliga et al., 1999). Although it could not be accessed for photographic documentation, Martín Cólliga et al. (1999) describe it as a hemispherical vessel, with a diameter of 20 cm. Previous lead isotopes analyses (LIA henceforth) on the crucible residue ruled out the Montsant and the MBF mining districts as metal sources (Fig. 1) (Montero, 2017; Rafel Fontanals et al., 2018a).

Cova del Buldó is a cave with two entrances communicated by a corridor. Below the second entrance, a lower level can be accessed. The corridor and one of the lower-level chambers host communal inhumations. Finds from the corridor include at least five adults and three children. Only one burial (burial A) had associated perforated shells and some pottery. In contrast, the lower-level chambers yielded scattered human remains, including adult victims of the Carlist Wars and two infant mandibles attributed to prehistoric times. Pottery sherds were also identified together with a crucible fragment (Fig. 2). No Bell Beaker pots were found. Despite the lack of habitational evidence, it is possible that the cave was used for this purpose too (Vilaseca and Iglésies, 1929; Vilaseca Anguera, 1973; Martín Cólliga et al., 1999; Soriano Llopis, 2010).

The crucible sampled is held at the Museu d’Arqueologia Salvador Vilaseca (Reus). It has a ~18-20 cm diameter. Previous LIA indicated that the copper processed in this crucible might come from Turquesa mine (Fig. 1). However, there was some disagreement regarding specific ratios (Soriano Llopis, 2010; Rafel et al., 2016; Montero, 2017).

Excavation of Balma del Duc was undertaken following a partial looting. The site is a rock shelter 3 m deep and 23 m wide. It had temporary occupations between the Late Neolithic-Chalcolithic and the Middle Bronze Age, based on pottery from different periods (including Bell Beaker), and bone and flint tools. Finds dated to the Iberian Iron Age, Roman, and Medieval times were also recorded (Solé, 1982; Estrada Martín and Nadal Lorenzo, 1993). The excavators (Solé, 1982) do not mention the three crucible fragments belonging to two vessels later reported by Soriano Llopis (2010). Thus, they lack contextual information. One vessel is described as very similar to Cova del Buldó specimen, and the possibility that they belong to the same specimen has been suggested (Martín Cólliga et al., 1999; Soriano Llopis, 2010). It was not possible to access these crucibles at the Museu-Arxiu de Montblanc in order to corroborate this.

Rovira and Ambert (2002) dated Balma del Duc crucibles to Chalcolithic times, but their poorly defined chrono-stratigraphic context(s) (Rafel et al., 2016) has led other authors to link them to the EBA instead (Soriano Llopis, 2010). They have 10 cm and 9 cm diameters, respectively (Soriano Llopis, 2010). Previous LIA on one of them matched the copper minerals from Turquesa mine (Fig. 1) (Rafel et al., 2014; Montero-Ruiz, 2018; Rafel Fontanals et al., 2018b).

Cova Josefina, excavated in the 1920s, is a cave with no stratigraphic sequence. It has two chambers communicated by a corridor. The materials that have been recovered are dated from the Late Neolithic-Chalcolithic up to Visigothic times: flint tools, stone beads, perforated shells, and pots of different types, including Bell Beaker material. Three crucible sherds, a flat copper axe, two arrow heads (dated to the Bronze Age), and an awl mould fragment were recovered. Three stone hammers and one flat stone (a polisher?) might be also related to metallurgical work (Serra-Vilaró, 1925; Martín Cólliga et al., 1999; Soriano Llopis, 2010; Rafel Fontanals and Soriano, 2017a). Together with the habitational and metallurgical evidence, remains of at least four individuals were identified at the bottom of the cave with no reported connections with the metallurgical finds (Serra-Vilaró, 1925).

The crucibles are stored at the Museu Diocesà de Tarragona (Fig. 2). CJ3236 has a ~12 cm diameter; CJ3238 and CJ3237 diameters could not be calculated. Although the pXRF analyses on CJ3237 showed enrichments in copper, arsenic, and lead comparable to CJ3238 and CJ3236 (Tab. SF.13 in Supplementary File 1), it was not sampled because it lacks substantial residues. Thus, a sample was taken only from the other two.

Cova de l’Heura is a cave that consists of a corridor followed by a sub-rectangular area. In its deeper area, over 60 bodies were uncovered. Moreover, numerous flint tools and production residues point to the existence of a flint knapping station. Stone beads and ceramics were also reported. No Bell Beaker pots were recovered, but other items like the stone wrist-guard found are typical of these communities. A crucible sherd (~34 cm diameter) and other (s)melting residues/ore fragments support metallurgical activity too (Fig. 2). A bronze bead, a lead bead, an arsenical copper awl, and an elongated bronze item were also described.

The first habitational occupation, the flint knapping station, the lead bead, and the awl were dated to the Late Neolithic-Chalcolithic. During the early 3^rd millennium BC, the collective burial took place, and the metallurgical activities and the bronze bead were attributed to the beginning of the Bronze Age (Vilaseca Anguera, 1952; Rovira and Ambert, 2002; Soriano Llopis, 2010; Rafel et al., 2016).

LIA on the copper awl were consistent with Turquesa mine. The crucible charge was linked to either Barranc Fondo (Cornudella del Montsant, Tarragona) or Solana del Bepo mines (Fig. 1) although minor elements do not perfectly match. LIA on two mineral fragments (CH4403(1) and CH4403(10)) fit Barranc Fondo mine, although minor elements do not perfectly fit either. Previous pXRF analyses on these minerals indicated iron-rich copper with zinc, silver, tin, antimony and lead, and occasional bismuth impurities, which are not consistent with the known geochemistry of the deposit. The lead bead was linked to the MBF mining district (Fig. 1) (Montero-Ruiz et al., 2012; Rafel et al., 2014, 2016; Montero, 2017; Montero-Ruiz, 2018). These materials are kept at the Museu d’Arqueologia Salvador Vilaseca (Reus). A sample was taken from the crucible, as well as from four production residues/ore fragments, including the two previously sampled for LIA.

Although the materials already presented are the bulk of the available late 3^rd millennium BC production remains in the Northeast, there are a few other objects not considered in this study that can also be generally ascribed to the same context. Tab. 1 (and Fig. 1 for location) summarises them for reference.

CC1008 crucible slag is rich in Fe (59.9 wt%FeO in bulk). Its glassy matrix is a Ca-Fe aluminosilicate (15.6 wt%CaO; 26 wt%FeO). Ca enrichment is probably related to charcoal addition, as this element was not part of the ore fragments characterised (see below), but minor gangue contributions cannot be ruled out. Minor S, Cu, Zn, As and Ba detected in the glassy matrix (Tab. 2) were linked to the ore.

While Fe-Si and Mg-Fe-Si crystals occur in the glassy matrix, this is dominated by Fe oxide crystals (magnetite). These usually trap Cu-based prills in an arrangement consistent with Cu ore relics. These prills are metallic Cu (or Cu-Cl derived from corrosion), or Cu sulphides (up to 23.3 wt%S) with As ranging between BDL-9.1 wt%As (Fig. 3A). Some of the sulphides also contain minor Ag, and between 1.2-9.9 wt%Fe. Both Cu and Cu-S prills are roughly equally frequent. This characterisation is consistent with a sulphidic, Fe-rich Cu ore such as chalcopyrite (CuFeS₂), which may have contained some oxidic patches, minor As, Ag, Ba, and Zn impurities, and some Mg.

CB-PA22253 crucible slag is richer in Si, S, Cl, Ca, Mg, Fe, Cu, and Zn than the associated ceramic (Tab. 2). The small Mg enrichment might be derived from charcoal ash contributions as opposed to gangue materials (Fig. 4A). Cl most likely results from post-depositional corrosion. Some Si might have entered the charge as gangue.

CB-PA22253 glassy matrix is a Ca-Fe aluminosilicate (9.8 wt%CaO; 29.1 wt%FeO) with clusters of Fe oxide crystals (magnetite) trapping Cu prills, which are interpreted as Cu ore relics (Fig. 3B). The prills contain sporadic Ag, Sn, As, and/or Bi, and between 1.6-6.1 wt%Fe. No Cu-S phases were detected. A semi-dissolved Ca-Al-Si mineral was identified in the glassy matrix too, with no parallels in the altered ceramic analysed, so it might be residual gangue contributing to the Ca enrichment in the slag (0.6 CaO/Al₂O₃ and 0.04 CaO/Al₂O₃ for the ceramic and slag respectively, Fig. 4A). Some Fe-Al-Si crystals were also identified, likely resulting from the reaction of Fe gangue with the siliceous matrix. This evidence supports a Cu smelting operation of a Fe-rich Cu-ore bearing some Ca, and with minor Ag, Sn, As, Zn, S and Bi.

The ceramic fabric is non-calcareous (3.2 wt%CaO). No voids consistent with organic (plant) inclusions could be identified in the altered ceramics preserved. The mineral inclusions analysed are summarised in Tab. 3. Their small size (below 400 µm), and their inhomogeneous distribution suggest that they are natural clay inclusions.

BD-PA22251 crucible slag, when compared to the ceramic proper, is richer in Mg, Si, S, Ca, Fe, Zn and Cu (Tab. 2). The small Mg increase (0.05 MgO/Al₂O₃ and 0.21 MgO/Al₂O₃ for the ceramic and slag respectively, Fig. 4A) might be consistent with fuel ash as opposed to gangue. However, the larger Ca increase (0.05 CaO/Al₂O₃ and 0.7 CaO/Al₂O₃ for the ceramic and slag respectively) is comparable to that of Cova del Buldó crucible, which was related to Ca-bearing gangue. However, as Ca was not found in the ore relics analysed (see below), it might alternatively represent charcoal ash. The ceramic preserved is non-calcareous (0.8 wt%CaO), so Ca-related ceramic contributions were discarded. Finally, the considerable Si enrichment observed (2.2 SiO₂/Al₂O₃ and 3.3 SiO₂/Al₂O₃ for the ceramic and slag respectively) might derive from siliceous gangue.

BD-PA22251 glassy matrix is a Ca-Fe aluminosilicate (14.3 wt%CaO; 21.4 wt%FeO). Fe-Si-Mg crystals populate the ‘devitrified’ glassy matrix together with big clusters of Fe oxide crystals (magnetite), Cu prills, Cu sulphides (with up to 27.6 wt%S), and Cu-Fe sulphates that were interpreted as Cu ore relics (Fig. 3C). The Cu-bearing phases sometimes have up to 1.2 wt%As. Similar prills also appear outside the magnetite clusters.

As in Cova Cartanyà, the ore smelted was possibly sulphidic, perhaps combined with Cu carbonates. The charge would also contain Zn and As impurities, and potentially Ca. Moreover, previous pXRF analyses (Soriano Llopis, 2010; Montero, 2017) detected 0.16 wt%Sn, so minor Sn could be present.

The ceramic preserved is considerably altered, preventing the observation of voids consistent with organic (plant) inclusions. Tab. 3 summarises the mineral inclusions found. Their varied shapes, inhomogeneous distribution, and maximum size (200 µm) support their natural presence in the clay.

Surface pXRF analyses on CJ3236 and CJ3238 show enrichments in S, Cu, As, Sn, and Pb in the inner altered surface of both specimens. CJ3236 inner surface was also richer in P and Zn, and CJ3238 had more Ba (Tab. SF.13 in Supplementary File 1). This is consistent with SEM-EDS results.

CJ3236 has a thin red inner slag layer (~100-500 µm thick, Fig. 5A) with 1.6 wt%CuO. This layer is richer in Ca, Mg, S, Cu, As, and Ba than the ceramic (Tab. 2). Given its thinness, the Ca and Mg enrichments might be attributed to charcoal ash contributions instead of gangue components, from which more substantial slag might be expected. However, it must be noted that the Ca increase in the slag (0.06 CaO/Al₂O₃ and 0.8 CaO/Al₂O₃ for the ceramic and slag respectively, Fig. 4A) is consistent with that of the samples from Cova del Buldó and Balma del Duc related to Cu smelting operations instead. The Ca-Al-Si crystals analysed (anorthite) might represent the reaction of charcoal with the decomposing ceramic. The Fe-Mg-Al-Si crystals observed are most likely derived from ceramic inclusions (Tab. 3).

CJ3238 only has glassy patches made of altered ceramic. Its inner surface is richer in Ca, Fe, and Cu (1.0 wt%CuO, Tab. 2). Ca enrichment is minor, and likely related to fuel ash, and to Ca-bearing minerals identified in the ceramic paste (Tab. 3; Fig. 4A). A Cu-Fe-Al-Si prill (8.3 wt%Cu), Fe-Si prills with up to 1.5 wt%Cu, a Ba-S prill, and a Ba-S oxidised phase attached to the inner surface of the crucible were also identified (Fig. 3D). Ca-Al-Si, Ca-Fe-Al-Si, and Fe-Al-Si (with minor Cu) crystals were also found. They are consistent with the decomposition of ceramic inclusions that absorbed some of the available Cu (Tab. 3).

The thin slag layers of both crucibles and the scarcity of mineral phases within them might indicate that they were involved in quick operations, perhaps Cu melting. There is no conclusive evidence to support smelting. SEM-EDS and pXRF analyses support the use of Cu containing S, As, Ba, Sn, Zn, and Pb impurities.

SEM-EDS analyses of the clay matrices (Tab. 2) show a broadly similar composition involving non-calcareous clays (below 3 wt%CaO). Similar mineral inclusions of up to 3000 µm (CJ3238) and 2000 µm (CJ3236) were documented. This potentially indicates the use of the same clay for both ceramics. The variability of mineral inclusions, inhomogeneous distribution, and varied sizes in both samples might support natural presence. CJ3236 preserved some organic (plant) inclusions (Fig. 5B) of about 5000 µm. No organic inclusions were observed in CJ3238.

Starting with the ore fragments analysed, all of them turned out to be dominated by chalcopyrite at different stages of thermal decomposition. CH4403(1) is a Cu-S mass with variable Fe (4.9-15.5 wt%Fe) (Tab. 4). It consists of Cu sulphide crystals (matte) with up to 26.9 wt%S, surrounded by a Cu oxide or a Cu-Fe-O (23.0 wt%Fe) phase (Fig. 6A). A Bi-Ca-O phase (61.1 wt%Bi, 8.1 wt%Ca) was sporadically observed too. CH4403(1) likely represents the initial thermochemical alteration of a sulphidic Cu mineral such as chalcopyrite. In the absence of SiO₂, the Fe reacted with some of the Cu available. It is possible that some Cu carbonate patches were also dispersed among the dominant hypothesised chalcopyrite ore, facilitating the Fe absorption. This ore also contained minor Bi and Ca.

CH4403(2) has high Fe (25.4 wt%FeO), Si (56.9 wt%SiO₂) and Cu (8.6 wt%CuO), and some S (3.5 wt%SO₃), and Ba (0.3 wt%BaO) in its bulk composition (Tab. 4). Zn (0.2 wt%ZnO) was detected in the glassy matrix (Tab. SF.24 in Supplementary File 1), made of a Cu-Fe silicate with up to 7.2 wt%SO₃. All these elements were linked to the Cu ore used. Within this glassy matrix, big patches of Cu-Fe sulphides (matte) with Ag and Zn impurities were identified. These were sometimes surrounded by Cu-Fe sulphates resulting from their corrosion, and Fe oxide crystals. Ba-S-O prills with minor Sr were also analysed. Moreover, well-delimited areas of a Cu-Fe aluminosilicate phase with 4.7 wt%SO₃ were identified close to these phases, probably resulting from the reaction of the Fe and Cu with the silicate gangue (Fig. 6B). This would be in agreement with the quartz relics seen in other areas. In the inner cracks of the quartz grains, Cu oxide corrosion phases accumulated, together with Fe-Al Si phases and Cu-Fe phases with up to 2.2 wt%SO₃ and occasional Pb (1.0 wt%PbO).

In sum, CH4403(2) is likely the resulting product of a Fe-bearing sulphidic Cu ore thermochemically altered, potentially chalcopyrite. It had silicate gangue, as well as minor Ag, Zn, Ba, Sr, and Pb impurities.

Turning to CH4403(4), this is a Cu-Fe-S silicate (16.8 wt%SO₃; 23.9 wt%FeO; 39.3 wt%CuO) (Tab. 4) made of Cu sulphide patches (matte) with up to 23.4 wt%S and 1.2 wt%Fe. These are surrounded by clusters of Fe oxide crystals related to a Fe-Si phase and a Fe-Cu-Si phase with up to 2.3 wt%SO₃ (Fig. 6C). CH4403(4) was thus also characterised as a sulphidic Cu ore fragment, probably chalcopyrite, subjected to thermochemical alteration. This made the Fe to oxidise and react with the scarce available gangue and surrounding Cu.

To conclude with the altered minerals, CH4403(10) has a Cu sulphide matrix (matte, 24.5 wt%S) with up to 9.0 wt%Fe and minor Zn. This is populated by tiny Cu-Zn-S and Cu-Zn-Pb-S prills (the latter containing 1.2 wt%In). Clusters of Fe-Ca crystals associated to Cu oxide prills were observed within the matte matrix (Fig. 6D), as well as clusters of Fe and Fe-Ca-Mg-Si crystals and a Fe-Ca-Si phase with 0.6 wt%BaO (Fig. 6E). Within these clusters, Cu prills with minor Ag, Sn, and Pb were identified. Many were biphasic, with the secondary phase being a Cu-S-Pb alloy (7.8 wt%S; 8.4 wt%Pb). In other areas, Cu-Si phases within the matte matrix were close to Fe-Mg crystals.

CH4403(10) was interpreted as another sulphidic Fe-rich Cu ore fragment, perhaps chalcopyrite, subjected to initial thermochemical transformations. This allowed some of the available Ca, Mg and Si gangue to react with the Fe and Cu. It is important to note the initial formation of metallic Cu with very low S (in bulk) already in the shape of small prills. The mineral processed also had minor Ag, Pb, Zn, Sn, In, and Ba impurities.

While SEM-EDS analyses helped to establish the use of chalcopyrite, pXRF –with lower detection limits– indicates that the impurities of the ore fragments are broadly similar, which might be indicative of a similar source (Tab. SF.21 in Supplementary File 1). According to pXRF, all samples have Si, Fe, Ca, Zn, As, Sn, Ba, and Pb. CH4403(1) also has Ag, Zr and Sr, which should be added to the Bi found in sporadic Bi-Ca-O phases in this sample. The consistent presence of substantial amounts of S in all mineral samples indicates that these were not roasted –the latter usually producing a S-poor, red surface (Rostoker, 1975). Thus, these are likely remnants of a smelting operation in a relatively reducing atmosphere, as seems to be confirmed by their attested processing within crucible CH4769.

CH4769 is the crucible analysed. Its slag layer is enriched in Mg, Si, S, Ca, Fe, Cu, Zn, and Ba when compared to the preserved ceramic. As Mg and Ca enrichment is considerably small (0.04 MgO/Al₂O₃ and 0.12 MgO/Al₂O₃; 0.4 CaO/Al₂O₃ and 0.7 CaO/Al₂O₃ for the ceramic and slag respectively), it is more consistent with charcoal ash rather than gangue (Fig. 4A). Although the Si enrichment is not very high (2.1 SiO₂/Al₂O₃ and 2.5 SiO₂/Al₂O₃ for the ceramic and slag respectively), some siliceous materials could have entered as gangue. The rest of listed elements were interpreted as linked to the Cu ore used.

The glassy matrix is a Fe-Ca aluminosilicate (24.4 w%CaO; 20.5 wt%FeO) with Cu ore relic microstructures. Firstly, clusters of Fe oxide crystals (magnetite) trap numerous Cu-based prills (Fig. 3E). These are mostly pure Cu with up to 0.2 wt%S, and in some occasions, they are Cu sulphides with up to 19.9 wt%S. Importantly, solid-state smelting microstructures have been identified within these clusters, which supports their interpretation as Cu ore relics (Montes-Landa et al., 2024). Based on the low S of the trapped prills, these Fe-bearing Cu ore relics are mostly of oxidic nature. S was only a minor impurity.

Secondly, Cu-Fe sulphide masses (matte) with up to 1.8 wt%Zn were identified too. They present internal Cu oxide masses and cracks also filled up by Cu oxide phases. The Cu oxide masses probably represent the post-depositional corrosion of metallic Cu (comparison in Hauptmann, 2011, fig.19.13). These Cu-Fe sulphide masses are usually surrounded by (Cu)-Fe-Ca-Al-Si crystals with sporadic Ba and Pb impurities probably related to the ore (Fig. 3F). These masses were interpreted as sulphidic Cu ore relics, likely chalcopyrite. It is possible that it was mixed with some Cu carbonate patches that facilitated the development of metallic Cu inside the matte matrix.

The use of sulphidic ores is also confirmed by independent Cu sulphide prills observed under OM and analysed by SEM-EDS. This evidence links CH4769 to the chalcopyrite ore fragments already analysed. Moreover, pXRF on the inner surface of this crucible also detected over 0.02 wt% of Sn and Ag which fits the impurities in the altered chalcopyrite samples above. Although As was not detected in the crucible slag, its volatility might explain its absence in this sample.

Finally, semi-dissolved quartz and pyroxenes were detected in the slag, but these are consistent with other ceramic mineral inclusions (Tab. 3). Their maximum size (250 µm) and their inhomogeneous distribution point to their natural presence in the clay. Elongated voids compatible with burnt organic (plant) inclusions of up to 1500 µm were observed. Interestingly, this crucible presents a calcareous fabric (8.4 wt%CaO in ceramic matrix, and 10.3 wt%CaO in the bulk ceramic composition).

The identification of both oxidic and sulphidic copper minerals potentially links this crucible to a co-smelting operation to produce copper, whereby copper is reduced by direct oxide-sulphide interaction without an intermediate matte stage. This can happen incidentally (through the selection of naturally occurring rocks that exhibit distinct colours and contain both minerals) or by mixing separate ores (also distinguishable by colour) in the same crucible. Whatever the case, the extraction of copper out of this co-smelting operation likely happened in a single step, as proposed by Rostoker et al. (1989).

It is important to understand how the characteristics of the Northeast Iberian technological tradition affect copper-based production in subsequent periods.

Going back to the EMBA (2100-1650 BC), a previous study of Minferri and Cantorella (Maldà, Lleida) assemblages had revealed the arrival of a range of metallurgical innovations to the Northeast at this time, presumably transmitted from Polada groups in Italy, through Southern France. These are (1) the know-how necessary for alloying bronze, (2) a specific type of purpose-made crucibles with a thicker base and pouring spout, and (3) stone moulds, of which early examples in the Northeast date to 1900-1800 BC (Soriano and Amorós Gurrera, 2014; Soriano and Escanilla, 2016; Peres et al., 2024). Remarkably, the connections between the Northeast and Southern France during this time are also supported by the documented Baltic amber trade that, although started back in Neolithic times, remained active up until this period in the Northeast but absent in other Iberian areas (Murillo-Barroso et al., 2023, 2025b).

Although both purpose-made crucibles and repurposed common pots were used for metallurgical activities at Minferri (2100-1650 BC), purpose-made crucibles disappear from the archaeological record in the Northeast after this time. Arguably, this type of specialised technological ceramics did not provide behaviourally significant improvements in this context. This is supported by the analyses carried out on one purpose-made crucible from Minferri that, when compared to other repurposed fabrics, show that both types of vessels have non-refractory calcareous pastes of comparable qualities (Montes-Landa, 2023; Montes-Landa et al., 2025).

The rejection of purpose-made crucibles and the adoption of bronze technology and stone moulds in the Northeast would be consistent with the already discussed selective adoption of features from Southern Iberia and Southern France technological traditions during Chalcolithic-EBA times. Moreover, it illustrates a key technological characteristic of the regional tradition, which will influence technological choices later on: a commitment to incurring only in low investments during metallurgical practice.

The evolution of copper smelting and bronze alloying technologies documented in Northeast Iberia through the analyses of Minferri, Vilars, Mas Castellar and Ullastret production residues suggests that the technological tradition of this area remained broadly unchanged between the 3^rd millennium and at least the 3^rd century BC (Montes-Landa, 2023; Montes-Landa et al., 2025). With the exception of bronze technology introduction and stone moulds (Peres et al., 2024), no further remarkable innovations were adopted, including proper furnaces. Metallurgy continued to take place as small-scale operations in crucibles heated from the inside within very simple open structures. Reducing conditions were unstable, resulting in what we could consider inefficient operations that were, however, good enough for the standards of the time. This is demonstrated by the apparent continuation of this way of doing even within the workshop context studied at Ullastret dated to the 450-350 BC (Montes-Landa, 2023; Montes-Landa et al., 2025). This characterisation is consistent with prehistoric Iberian metallurgy in general, which connects this area with the rest of the Peninsula despite the peculiarities already highlighted.

The Chalcolithic-EBA data on metal provenance should also be discussed in relation to later finds. The LIA results available so far are consistent with the use of Les Ferreres (La Bauma), Linda Mariquita (Ullastret), Solana del Bepo (La Bauma, Ullastret), Barranc Fondo (Cova de l’Heura), Turquesa (Cova del Buldó, Balma del Duc, La Bauma?, Ullastret?) and Linares (Minferri) mines (Rafel et al., 2016; Montero, 2017; Montero-Ruiz, 2018; Montes-Landa, 2023; Montes-Landa et al., 2025). For Solana del Bepo case, whose minerals have been found both at La Bauma and Ullastret smelting slag, this would imply transport of minerals from El Priorat to Girona from Chalcolithic (La Bauma) up until Iberian Iron Age times (Ullastret), perhaps using very similar routes. This is particularly noteworthy when one considers that metal trade had been established by the Iron Age. Moreover, an unknown source of copper identified in the production residues from Mas Castellar (dated to the 550-500 BC) is consistent with EMBA metallic items from Cova de Porta Lloret, Rocallaura, and Covas del Cingle Blanc (Arbolí, Tarragona) (Montes-Landa et al., 2025). This also supports a continuity in the use of the same metal sources in the region.

Furthermore, the versatile procurement pattern observed at La Bauma, which combined five sources located between 25 and 200 km away is important, as well as the possible exploitation of two mineral sources at Cova de l’Heura. As already mentioned, these cases exemplify a highly adaptive and versatile strategy to secure metal procurement. This permeated in the later technological tradition in two ways. Firstly, a multi-source exploitation strategy was maintained at sites such as Vilars and Mas Castellar during the Iron Age, Ullastret during Iberian Iron Age times, and possibly Minferri during the EMBA. As in previous times, such strategy allowed sustaining bronze demand. Secondly, the adoption of a flexible pattern of choice towards making technological choices (as seen in the combination of several metal sources within a site, but also in the combination of direct copper smelting and co-smelting of oxidic and sulphidic ores) influenced bronze-making strategies too. Thus, at Minferri, Mas Castellar and Ullastret, different bronze alloying techniques coexisted. Such variability and flexibility in the selection of techniques has been interpreted as a resilience mechanism to sustain demand (Montes-Landa et al., 2025). The adaptive character of the Chalcolithic Northeast Iberian metallurgy remains at the core of such strategy two millennia later.

Also, we should note the matte nodules found at Vilars and Ullastret sites, dated to 450-375 BC and 250-200 BC respectively (Montes-Landa, 2023; Montes-Landa et al., 2025). While these are rare specimens in the context of Iron Age/Iberian times bronze metallurgy, which is dominated by oxidic minerals, they denote the sporadic handling of sulphidic ores that could have relied on Chalcolithic know-how.

Finally, the contextualisation of the metallurgical operations allows a reassessment of the location of the chaîne opératoire in the landscape over time. So far, mainly based on Minferri assemblage and the presence of purpose-made vessels there, it had been contended that since the EMBA the chaîne opératoire was broken down into the landscape, with copper smelting being conducted by the mines, and only metal melting occurring on habitation sites (Soriano Llopis, 2010). However, recent results prove that copper smelting and metal alloying were conducted within the settlement (Montes-Landa, 2023; Montes-Landa et al., 2025), thus continuing the Chalcolithic tradition of smelting in domestic spaces, even if the nature of the settlement type had changed. A separation of copper smelting from melting or alloying in the landscape is not observed until the Iron Age at Mas Castellar, but this statement can be misleading: although metallic copper for bronze cementation operations was produced elsewhere, copper ores were still co-smelted with cassiterite to produce bronze on site too. Moreover, at the contemporaneous site of Vilars, copper smelting was still conducted on site within domestic contexts, and this keeps happening later on at Ullastret too. All in all, this data reinforces the weight of the Chalcolithic technological tradition two millennia after its development.