Objects were observed for traceological markers, photographed, and sampled with a stereoscopic microscope (SM), an Olympus SZ61-TR, equipped with a Bresser MikroCamII 12 megapixel digital camera and operated by MikroCamLabII software. The images were recorded at multiple focal planes and reconstructed using the focus stacking software Helicon Focus.

The detached chips (dimensions on average 500–700 x 500–700 µm) were embedded in epoxy resin, ground with silicon carbide paper (1200–4000 mesh), and polished with diamond suspension (5 and 1 µm) in the laboratories of the Dipartimento di Geoscienze, Università degli Studi di Padova (Department of Geosciences, University of Padua, Italy), to obtain a flat and smooth cross-sectional surface. Samples were analyzed with an FEI Quanta 200 environmental scanning electron microscope (ESEM) equipped with tungsten filament and an Xplore 30 (Oxford Instruments) energy dispersive (EDS) X-ray detector at the Centro Analisi e Servizi per la Certificazione (CEASC; Center of Analyses and Services for Certification) at the University of Padua. Imaging in secondary (SE) and backscattered electrons (BSE) was used to study the texture and the weathering processes, and to map and quantify mineral inclusions related to coloring/opacifying agents, relics from the batch, and newly formed phases. Point and areal analyses were performed at standard operating conditions of 20 kV voltage, chamber pressure of 0.53 Torr, spot size of 1–5 µm, and a duration of 30 seconds per point/areal acquisition. All the semiquantitative EDS analyses are normalized at 100% by default. The detection limit is variable but in the order of 0.1 wt %, except for elements with a low atomic number or with overlapping peaks.

After SEM-EDS analyses, the vitreous phase of 20 samples (plus 1 zoning) and a selection of inclusions were investigated by electron probe microanalysis (EPMA) to obtain chemical data with very high precision and accuracy. The instrument used was the JEOL 8200 Super Probe with five wavelength-dispersive spectrometers (WDS) at the Dipartimento per le Scienze della Terra, Università degli Studi di Milano Statale (Earth Sciences Department of the University of Milan). Eighteen elements were analyzed: Na, Mg, Al, Si, P, S, Cl, K, Ca, Ti, Mn, Fe, Co, Cu, As, Sn, Sb, and Pb. Corning glass standards A and B were analyzed before each session (results in Table 2) to evaluate the accuracy (values) and the precision in terms of reproducibility of the data (Standard Deviations), which turned out to be consistent with those recorded in previous experimental rounds with the same instrument.7

Working conditions were set at 15 kV accelerating voltage and 5 µA sample current; spot size was varied according to the textural characteristics of the samples: for the transparent glasses and the interparticle glass in sample AM-BV-29, we opted for 5 measures with beam defocused at 10 µm to reduce alkali loss. Opaque samples with homogeneously dispersed crystalline inclusions were investigated with 4–6 measures at 5 µm spot size; since opaque glasses in this set were all heavily weathered, we didn’t need any additional point at 10 µm spot size to evaluate the sodium volatilization rates. Mineral inclusions were analyzed with a 1 µm probe diameter to reduce the contribution of the glassy matrix to the acquired data. The minimum detection limit of this technique is approximately 0.03% for most of the analyzed oxides.

The archaeological sites in which the analyzed materials were discovered are reported in Figure 1. In this section, a brief synthesis on the archaeological contexts and the consistency of the finds in each site will be shortly provided. Table 1 summarizes the main archaeological and physical data related to the analyzed samples, including provenance, chronology, typology, state of conservation, dimensions, color, weight, archaeological data on the label, and the number and location of the samples detached. For ease of reference, it also includes the analytical code of each sample, and a set of relevant notes. Plate 1 presents photographs of all the analyzed samples.

The major and minor elements in the amorphous phase of the analyzed samples are reported in Table 2 as weight percent (wt %) of the oxides. The chemical analyses show a wide range of compositions and a non-negligible number of weathered glasses (Na₂O + K₂O < 4%), which include all opaque whites (CR-AAF-003Bi, CD-BoD-04Bi, PN-0878-Bi) and three out of seven transparent green/blue glasses from Cabezo Redondo (CR-GA-498, CR-BS-555, CR-GV-2255) (Figs. 5, 6). To avoid any confusion, compositional data related to altered samples will not be discussed further or plotted in the next diagrams; they will only be considered for textural observations and for the identification of the mineral opacifiers.

In general terms, silica contents range from 59.6% to 67.7% in glasses and reach the peak value of 70.9% in the glassy faience sample from El Amarejo. Soda goes from 6.1% to 19.9%, potash from approximately 0.2% to 11.6%, lime from 1.2% to 5.7%, magnesia from 0.4% to 7.1%, and alumina from 1% to 6.5%.

The classic plot K₂O vs. MgO (Fig. 6) emphasizes the presence of at least three compositional macro-groups, which are essentially related to the type of fluxing agent employed, plus three outliers all originating from the site of Peña Negra (the dark-blue glasses PN-O878-B and PN-OB-13, and the dark-green glass PN-AN-80), which will be presented separately. The first group includes only the light-blue glassy faience from El Amarejo (AM-BV-29). The sample is characterized by a low-magnesium (MgO 0.37%), high-potassium (K₂O 11.6%) (LMHK) composition,34 also known as mixed alkali, due to the presence of inversely proportional amounts of soda and potash (Na₂O 6.15%). The second set includes all the non-weathered green, brown, and blue glasses from Cabezo Redondo and Cueva de las Delicias, for a total of seven samples (CR-GA-544, CR-AAF-03M, CR-AV-25, CR-GV-85, CD-GV-02, CD-GV-03, CD-BoD-04V). It comprises high-magnesium (MgO 5.4–7.1%) glasses (HMG) that contain relevant levels of potash (2.9–3.5%) and lime (4.9–5.7%) typical of a plant-ash flux coming from eastern Mediterranean regions.35 The third cluster incorporates two very dark-green samples from Peña Negra (PN-GN-978, PN-GV-87), whose composition can be framed in the class of low-magnesium glasses (LMG), showing K₂O in the order of approximately 1%, low MgO (on average 0.6–0.7%) and CaO (always < 2%), but also very high levels of FeO (about 10.7%). LMG glasses, including the FeO-rich variant (Figs. 6, 7), were all produced using natron as fluxing agent.36 This peculiar LMG subgroup was already recognized in several LBA–EIA sites in Italy, France, and Portugal and is often named “natron-black.”37

Finally, three glasses have an intermediate HMLK composition with MgO from 2% to 4% and potash in the range of classic natron glasses (K₂O < 1.5%). This class can be further subdivided in two distinct groups:

Considering the moderate to advanced surface weathering of all the beads, the color of the glass was determined during sampling from the detached micro-chips (see Figure 4). In general terms, the set comprises mostly transparent green and blue beads (with either a light or a very dark tone), one transparent brown/amber sample (CR-AAF-03M), and three opaque white glasses (CD-BoD-04Bi, PN-O878-Bi, and CR-AAF-03Bi, which was later recognized as a transparent glass of unknown color during SEM investigations, as noted above).

The ratio CoO/CuO (Fig. 8) shows that all seven transparent blue and green glasses from Cabezo Redondo and Cueva de las Delicias, except CR-GV-85 and CD-BoD-04V, are colored with copper (CuO in the range from 1% to 1.2%). SnO₂ is aways under detection limit in all the glasses (< 0.03 wt %); however, it reaches 0.47 wt % in the glassy faience from El Amarejo, with a correspondingly higher CuO (4.1%). The Cu/Sn ratio (8.9) agrees with the use of a Cu alloy with 10% of tin, which is perfectly compatible with the composition of the ancient bronze objects produced in the timespan considered. On the contrary, cobalt exceeds the detection limits only in the two dark-blue samples from Peña Negra (PN-O878-B, PN-OB-13).

The amber bead CR-AAF-03M (see Figure 4b) as well as the green samples CR-GV-85 and CD-BoD-04V do not contain any ionic chromophore in significant concentration (see Figures 7, 8). This is a typical feature of transparent brown glasses, whose color has been widely attributed to the Fe-S complex (FeO 0.28% and SO₃ 0.32%).39 Likewise, the green sample CR-GV-85 probably derives its color from FeO (0.3%), but under different redox conditions. Finally, iron is surely the chromophore responsible for the very dark greenish-black color of three samples from Peña Negra (PN-AN-80, PN-GN-978, PN-GV-87), which contain FeO in the range 9.3–10.8% (see Figure 7).

In addition to revealing the different patterns of weathering (which will not be discussed here for the sake of brevity), BSE images showed the presence of various mineral inclusions, comprising coloring and opacifying agents, but also relics from the batch and newly formed phases. The (limited) remaining portions of non-altered glass in the beads from Cabezo Redondo and Cueva de las Delicias are always homogeneous. The only exception is CD-BoD-04, whose dark-green glass shows an even dispersion of colloidal Cu sulfide particles. Due to the sub-micrometric dimension of the inclusions (Fig. 9a), it was not possible to precisely determine their shape or their stoichiometry, which would allow us to identify their exact mineral nature. However, the Cu/S wt % ratio derived from EDS data (3.75) suggests them to be most probably chalcocite (Cu₂S). The white glass from the same bead (CD-BoD-04Bi) is equally interesting: as expected, it contains small and finely dispersed aggregates of Ca antimonates (EDS data), which are mostly anhedral and partly dissolved into the (severely weathered) glassy matrix. However, a big lump of euhedral crystallites is also present (Fig. 9b) and EPMA determined them to be Na antimonates (mineralogically, brizziite). We interpret this as an evidence of combined ex situ addition (Ca antimonates) and in situ precipitation (brizziite), where this last occurrence can most probably be associated with the very low CaO (only 0.8%) in the vitreous phase.

The glasses from Peña Negra show only minor signs of weathering on the external surface. Air bubbles are always present. In terms of inclusions, PN-GN-978 contains one residual silica grain (Figs. 4a, 9c); and PN-O878-B a few newly formed wollastonite (CaSiO₃) crystals close to the outer surface (Fig. 10a), which are considered indicators of high temperatures or found to be the result of weathering.40 All other transparent glasses are very uniform. The white decoration in PN-O878-Bi is opacified by means of euhedral Ca antimonates with rectangular morphology and limited substitutions of PbO (up to approx. 2%) (Fig. 10b). BSE imaging suggests that they belong to the orthorhombic variant with formula Ca₂Sb₂O₇.41

The biconical bead from El Amarejo AM-BV-29 is composed of a comparable quantity of mineral inclusions and an amorphous phase, confirming its preliminary interpretation as glassy faience (Figs. 4c, 10c). There is no preserved glaze, and the body of the bead is very uniform, with the interaction layer (IL) not actually distinguishable from the core. The porosity is rather scarce, but several bubbles have significant dimension, up to approximately 60–70 µm. Silica grains constitute the majority of the mineral inclusions, with a dimension from a few microns up to 100–120 µm. Chemical mapping (Fig. 11) and EDS and EPMA data also highlight the presence of numerous alkali feldspars, a few augites, and frequent cassiterite (SnO₂) inclusions.

BSE observations of the Egyptian blue bead from Peña Negra PN-DEB-15 demonstrate the total lack of any residual vitreous component. This is most probably a result of weathering,42 but it could also be—less probably—derived from a complete synthesis of the raw materials into cuprorivaite crystals, the calcium-copper tetrasilicate with formula CaCuSi₄O₁₀ which confers the typical blue color to the Egyptian blue frit. From a microstructural perspective, only big, angular silica grains (darker gray in Fig. 12) and proportionally fewer cuprorivaite crystals (lighter gray in Fig. 12) survive in the sample, within a diffuse porosity (black in Fig. 12).

All seven non-weathered samples from Cabezo Redondo and Cueva de las Delicias were classified as plant-ash glasses based on the soda, magnesia, potash, and lime content. The HMG recipe is typical of the glasses produced using halophytic plant ashes in the Near East and Egypt starting from the second millennium BCE.43 This composition is extremely widespread in central Europe and in the whole Mediterranean region during the LBA, with occurrences unearthed even in the far north of Europe and throughout Romania, Germany, and Scandinavia.44

The samples from Cabezo Redondo and Cueva de las Delicias cover a wide timespan from the end of the seventeenth to approximately the tenth century BCE. In general, they show very low impurity content, suggesting that they are freshly batched glasses or, alternatively, that in recycling, the glassmakers took care to remelt only similar colors together. The alumina content, associated with the very low amounts of FeO and TiO₂, is compatible with very pure or purified sands, but it could also, less probably, be due to a contamination from the quartz-grinding process and from the ashes, or even to an enrichment from the crucible. Most of the beads are colored with copper (with cobalt under detection limits), which is the preferred solution for the majority of the light-blue and green beads from this period. However, one sample is made of HMG amber glass. This color is rather unusual in LMG glasses before the Hellenistic period,45 but it is especially rare in Europe during the LBA. There is, indeed, a subgroup of plant-ash glasses known as high magnesium brown glasses (HMBG) documented in central and northern Italy during the local RBA and FBA.46 However, the texture of HMBG glasses is different from that of CR-AAF-03M: while the Spanish sample is perfectly transparent and very homogeneous, HMBG glasses show abundant euhedral crystals of diopside and/or augite and droplets of copper sulfides; the chemistry is also not compatible, due to the higher content of CaO (6-7%) and especially Al₂O₃ (2-3.5%) and FeO (4-5.5%) in the HMBG class. Transparent HMG brown glasses are, however, attested in the well-known glassmaking centers of Amarna and Malkata (Egypt) during the fourteenth century BCE. In the western Mediterranean, the only strict compositional comparison for CR-AAF-03M is an annular bead from the Nuragic Giants’ Tomb of Gonnosfanadiga (Cagliari, Sardinia) dated to about 1450–1250 BCE.47 The chronology is in perfect accordance with the one proposed for the sample from Cabezo Redondo (Beta-332581, 3110 ± 30 BP, 1442–1286 cal BCE, 95.4% 2σ), confirming that Sardinia and southeastern Spain were served by similar maritime circuits probably departing from the far east shores of the Mediterranean during the early second half of the second millennium BCE.

The barrel bead with helicoidal decoration CD-BoD-04 has an HMG composition. This is somewhat unexpected considering that many FBA beads with the same typology recovered from across Europe are mixed-alkali glasses. There are, however, two known exceptions from France, one found on the site of the Kermorvan peninsula (Le Conquet, Brittany) and the other in Eckwersheim (Bas-Rhin), both dated to the eleventh to tenth century BCE.48 Trace element analyses on those two beads suggest a Near Eastern provenance, which is most probably the case also for our sample based on the similarity of the major element composition. The opaque white decoration is predominantly colored by ex situ addition of Ca antimonates, but in situ precipitated brizziite was also identified. Brizzite appears sporadically in ancient glasses—most of the time unintentionally, being a consequence of insufficient Ca amounts in the glassy matrix. It was, however, recognized in other white decorations of the FBA and EIA which frequently have low CaO levels, and is also sporadically present both in the Near East and Europe up to late antiquity.49

Despite the wide chemical variations, the five beads from Peña Negra dated to the tenth to ninth century BCE are all produced using mineral natron as flux. In Egypt and the Near East, this composition systematically substitutes for HMG glasses starting from the tenth to ninth century BCE and remains the most common recipe used throughout the IA and beyond. The discovery of LMG samples in central and western Europe as early as the tenth to ninth century BCE is uncommon but confirmed by several specimens. Among these early natron glasses, the samples from Peña Negra belong to three different variants—(1) Fe-rich LMG; (2) Fe-rich HMLK; (3) Al-Co blue HMLK—which have already been recognized in Italy as well as Portugal, France, Greece, Hungary, former Yugoslavia, and Czechia.50 The very dark appearance of Fe-rich LMG and HMLK beads is due to the voluntary addition of iron as a raw material, probably in the form of highly impure sands, rich in heavy minerals. However, the discovery of all the beads from Peña Negra belonging to the Fe-rich compositional variants in the dump of a metallurgical workshop might tentatively suggest a connection with metal production, for example, a coloring process triggered by adding a certain amount of slag chunks or powder to the molten batch, as already suggested for HMBG glasses.51 The bead from Gatas, which was already considered anomalous when first analyzed by Julian Henderson in 1999, remains an open problem.52 While it can be framed generally in the heterogeneous class of Fe-rich greenish-black glasses, its composition with high K₂O and MgO (both > 2%) and its very early chronology (1700–1500 cal BCE) might tentatively identify it as an ancestor of the later Fe-rich LMG productions, sharing a similar coloring technology.

Al-Co blue HMLK glasses derive their peculiar chemical composition from the use of Egyptian colbaltiferous alums as coloring agent. Beads with a similar composition from the ninth and eighth centuries BCE were discovered in France and Italy and constitute some of the earliest natron glasses ever recorded in western Europe.53 Their presence in Spain as early as in the tenth to ninth century BCE shows once again how the Iberian Peninsula was directly and precociously integrated in the maritime traffic which reached the western side of the Mediterranean Sea, paving the way to the Phoenician-Punic colonization of southern Spain.

The bead from El Amarejo was classified as a glassy faience LMHK based on a combination of textural and chemical features. Mixed-alkali vitreous materials are attested in central Europe, Italy, Britain, Ireland, and France starting from the EBA (about 2200–1900 BCE). At the beginning, this composition was reserved for faience beads, but later, as of about 1700/1650 BCE, production of glassy faience ornaments began. While the peak number of occurrences of glassy faience LMHK dates to the period 1700/1650–1450 BCE, they continue to be sporadically found up to the FBA north of the Alps. LMHK faience, glassy faience, and glass objects are considered a local European production, in contrast with the coeval plant-ash vitreous materials which were imported as semi-finished or finished products from Egypt and the Near East.54

Judging by the quantity of alumina (1.5%), and especially by the very abundant presence of alkali feldspars and other relics from the batch, sands were certainly used as silica source. The color is due to copper, likely introduced in the form of bronze. The composition is generally consistent with the conical buttons and beads from northern and central Italy which are common during the local MBA1–2,55 but also with the later Swiss occurrences from the FBA.56 Unfortunately, the composition of LMHK glassy faience objects is rather variable even in pieces from the same assemblage, possibly due to the production technology, or to the existence of multiple workshops (not yet identified), or even the use of non-standardized processes and raw materials, depending on availability. Even more probable is that we are confronted with a combination of the three. This complicates the identification of chemical markers with chronological significance in these materials and, ultimately, prevents us from assigning with certainty the El Amarejo bead to one or the other phase. However, its presence testifies to relevant contacts with Italy and central Europe, marking an important additional, previously unknown locale in the distribution of mixed-alkali vitreous materials.

At present, sample PN-DEB-15 is the only published occurrence of an Egyptian blue bead recovered in the western Mediterranean / western Europe with a BA dating. Egyptian blue has a long history as a pigment, but in ancient times it was also employed to produce small ornamental objects. Although this technology was introduced in Egypt and the Near East around the early third millennium BCE, it appears around the Mediterranean only about a thousand years later, when it was essentially confined to Minoan and Mycenean sites. Its discovery in Spain at the turn of the second to the first millennium BCE is a clear indicator of long-distance trade networks.

Due to severe weathering, the microstructural examination evidenced only the presence of abundant silica grains and of small cuprorivaite crystals. The lack of interparticle glass is unfortunately a well-known issue that affects the vast majority of the BA Egyptian blue samples recovered in Greece and the Aegean that have been analyzed to date. This being the case, it is not possible to gain information regarding the raw materials or to suggest a tentative provenance for the bead on a compositional basis. The smaller quantity and dimensions of the cuprorivaite inclusions (max. 25–30 µm) with respect to those of silica account for the very light blue color of the bead. In the literature this has been considered as an indicator that the beads in Egyptian blue are secondary products derived from grinding, shaping, and refiring primary lumps.57