Earthen Architecture and Craft Practices of Early Iron Age Ramparts: Geoarchaeological Analysis of Villares de la Encarnación, South-Eastern Iberia

1 Introduction

Beginning in the second half of the seventh century BC, Early Iron settlements in the southeast of the Iberian Peninsula were progressively protected by the construction of defensive walls. Although some walled settlements existed in the eighth century BC (García Menárguez & Prados Martínez, 2014), a new period of instability began at this time, which turned the safeguarding of Phoenician and autochthonous settlements into a priority. The probable increase in territorial tensions and social conflict resulted in the construction of new defensive systems and enceinte walls at the Late Bronze Age long-term settlements, such as Peña Negra (Lorrio Alvarado et al., 2020), at newly created towns such as Los Almadenes (Sala Sellés et al., 2020) and Castellar de Villena (Esquembre Bebia & Ortega Perez, 2017), and the construction of fortresses that controlled access roads or areas complementary to the main settlements as was the case with Cabezo de la Fuente del Murtal (Lomba Maurandi & Cano Gomáriz, 2002), Les Barricaes and Cantal de la Campana (Trelis Martí & Molina Mas, 2017). Among the Phoenician settlements, the most representative example of this instability was the construction of a massive defensive wall at Fonteta around 600 BC, which was successively reinforced throughout the century (González Prats, 2011, pp. 21 and 42; Lorrio Alvarado, López Rosendo, & Torres Ortiz, 2021; Rouillard, Gailledrat, & Sala Sellés, 2007, p. 126).

Most of the cited fortifications (Figure 1) were constructed mainly with stone, with few or poorly preserved examples employing earthen building materials (EBMs). Mudbricks, square or rectangular in shape, sun-dried and mass produced with a mould (Pastor Quiles, 2017, p. 51), had been in use in the domestic architecture of the Iberian southeast already since the eighth century BC. Autochthonous and Phoenician settlements, such as Castellar de Librilla (Ros Sala, 1989, pp. 100–101), Saladares (Arteaga Matute & Serna González, 1980, pp. 84–85), Peña Negra (González Prats, 1983, pp. 140–141), or Fonteta (Rouillard et al., 2007, pp. 99–100), provide us with numerous concrete case studies. However, the employment of mudbricks as a key construction element in defensive walls or other public structures (e.g., terrace walls) is not so common in the autochthonous settlements, which directly contrasts with the situation of Fonteta where mudbricks were an essential element in the construction of defensive structures. The excavations carried out at the site have revealed the use of mudbricks in various parts of the fortification, especially the wall elevations (Lorrio Alvarado et al., 2021, pp. 344, 346), one of the towers linked to the defensive rampart (González Prats, 2011, p. 56, Figure 26), buttresses (Rouillard et al., 2007, p. 128, Figure 114; pp. 132–133, Figure 124), and a low outer wall to prevent direct access to the defensive walls defined as antemuralla (Lorrio Alvarado et al., 2021, pp. 360–361, Figures 33 and 34).

Figure 1

Location of the fortified settlement of Villares de la Encarnación, with other Early Iron Age sites of the Iberian southeast (DTM: Spanish National Geographic Institute).

Therefore, the integration of mudbricks into defensive architecture was not fully adopted by local communities, perhaps because it was perceived as a foreign architectural tradition. However, the situation had to change in the Late Iron Age, when mudbricks readily appeared in the fifth century BC as a commonly used material in the elevations and the reinforcements of ramparts and watchtowers (see García Cano, 2008; Uroz Rodríguez, Lorrio Alvarado, & Uroz Sáez, 2022).

One of the earliest examples of integrating mudbricks into autochthonous walls of the Early Iron Age is identified at the site of Villares de la Encarnación (Caravaca de la Cruz, Región de Murcia). The initial systematic survey on the site resulted in recognition of its chronological sequence and its role as the main settlement in the highlands of northwest Murcia, at least between the seventh and sixth centuries BC (Brotons Yagüe, 2022; Ros Sala, Brotons Yagüe, & Ramallo Asensio, 2016). The feature demonstrating its remarkable geopolitical development is the defensive programme implemented, with two lines of solid walls reinforced by towers enclosing the areas not protected by steep escarpments. Despite the predominant building material being stone, the systematic survey identified a significant mudbrick concentration around one of the outer wall towers and its linked structures. The conservation degree of these materials is exceptional due to their exposure to a conflagration event, probably a low-temperature one, which has not left any burning marks on the detected mudbricks. This has clearly hardened the material, allowing their preservation in still-standing architectural remains.

Thus, the early chronology of the site (Table 1) and the availability of earthen materials from several locations around the tower allowed us to raise a pilot study of the mudbricks based on a geoarchaeological approach. Our main aim is to use the chemical and petrographic characterization of these earthen materials from Villares de la Encarnación as a means to investigate the process of incorporating this new building pattern, including essential issues such as the exploitation of the landscape or the complexity of architectural craftsmanship. The selected framework of cultural encounters between autochthonous and Phoenician communities also reveals the complexity hidden in these undervalued archaeological materials of the Iberian Peninsula, and how public/defensive and domestic architecture did not always follow the same dynamics of change and continuity.

2 Archaeological Background: Villares de la Encarnación

The fortified settlement of Villares is part of the Estrecho de la Encarnación archaeological area, a privileged location in the valley of the Quípar river made up of hills of different elevations occupied since prehistory (Figure 2a). The communities that settled in this inland region have benefited from its strategic position within the natural corridor that connects the Levante of the Iberian Peninsula with the area of the Alto Guadalquivir (Chapa et al., 2019) and prospered in an area rich in water, rain-fed crops, and abundant pasture for livestock.

Villares de la Encarnación is an Early Iron Age site located on a bio-calcarenite hill approximately 800 m in length and 300 m in width, emerging from the landscape, thanks to its steep slope of more than 40 m, affecting its immediate surroundings (Brotons Yagüe, 2022, p. 186). This location provides the settlement with broad control of the territory of the basin, which includes a large area of arable and pasture land between the Argos and Quípar rivers (Ros Sala et al., 2016, pp. 222–223). Although there are written references to the site from modern times (Cuadrado Díaz, 1945), the only archaeological works carried out to date consist of systematic archaeological surveying (Ros Sala et al., 2016).

The first results, which will need to be confirmed by archaeological excavation, suggest an initial occupation around the end of the eighth or the beginning of the seventh century BC, with an abandonment around the middle or the third quarter of the sixth century BC according to the surface ceramic materials. The survey also points to a certain specialization in some areas dedicated to productive activities, such as pottery manufacture or metallurgy work, which provide us with evidence of its role as the main settlement of this region and its commercial links with other inland and coastal areas through the Segura riverbed (Ros Sala et al., 2016). The necropolis extends outside the walls, specifically in the southwest area where several cremations were located under the foundations of the Late Iron Age sanctuary of La Encarnación (Ramallo Asensio & Brotons Yagüe, 2014, p. 30). The date of the abandonment will need to be verified by archaeological excavations (Brotons Yagüe, 2022, p. 191). Still, the present data would fit into the crisis and restructuring dynamics that spread throughout the Iberian Peninsula during the sixth century BC (Cutillas-Victoria, 2020).

Despite the site’s excellent natural defence by escarpments and ravines, the 18-hectare summit was heavily fortified by two sets of ramparts endowed (Figure 2b) with a significant number of towers or bastions (Brotons Yagüe, 2022; Ros Sala et al., 2016). The first defensive wall is arranged with an NE-SW orientation, and its total length is approximately 310 m. This rampart has been defined as a transversal barrier because of the division of the summit it creates, enclosing the north-western sector of the site, which was likely the location of the earliest settlement (Brotons Yagüe, 2022, pp. 187–188). This first fortification consists of a drywall of 5 m thickness made of local stone blocks, reinforced by ten rectangular or slightly trapezoidal towers built at regular intervals.

Figure 2

Villares de la Encarnación site. (a) Aerial view of the Estrecho de la Encarnación with the different sites located (PNOA: Spanish National Geographic Institute). (b) Shaded Relief Map (SRM) of Villares (elevation 45°, azimuth 50°) with the different walls and the study area marked (modified from Brotons Yagüe, 2022, p. 188, Figure 11). (c) Image of the study area during the sampling season (Author: F. Brotons).

The second “perimeter” wall construction probably corresponds to a second phase when the settlement expanded its urban area in a framework of emerging territorial tensions. This new rampart enclosed almost the entire plateau, except for the areas of the Barranco de la Virgen ravine and the side of the Quípar river, where the 40 m escarpments already provided an excellent defence (Brotons Yagüe, 2022, p. 188). The southern and southeastern areas are characterized by numerous towers or bastions to protect the more accessible plateau area. The remains of at least one tower built with mudbricks (Figure 2c) have been identified in this latter area (Brotons Yagüe, 2022, p. 190). These EBMs have been identified as part of the tower superstructure, laid on a stone socle, and the ones that collapsed from the structure as part of the outside landslide. However, not only were mudbricks used in the construction of this tower/bastion, but some nearby structures located inside the walled area were also built with mudbricks, although the functions of the latter remain unclear (Figure 3).

Figure 3

Group of displaced mudbricks from the internal structures, sampling point SE (Author: B. Cutillas-Victoria).

Focusing on the defensive tower, its appearance in the Early Iron Age had to be prominent due to its size and position in defensive and urban planning. Considering the importance these structures acquired in this specific socio-political context, especially if its construction is framed within the episodes of instability experienced by the Iberian southeast between the end of the seventh century and the sixth century BC, could explain the introduction of mudbrick to enhance and monumentalize these defences (Figure 4). Moreover, the possible relationship of the tower with the main entrance gate to the settlement, a hypothesis that we plan to test by excavating this sector in the future, could reinforce this approach. Thus, the construction of at least one massive tower using EBM would reinforce the most accessible flank of the settlement, as well as transmit a renewed image of the power and strength of the elites who inhabited and controlled the settlement and its territory.

Figure 4

Mudbricks forming part of the external landslide corresponding to the tower (Author: F. Brotons).

3 Materials and Methods

The sampling of the earthen materials from Villares de la Encarnación was conducted in 2023 and authorized by the Heritage Service of the Region of Murcia (EXC294/2022). The 31 samples were obtained from different areas of the SE sector of the site, including mudbricks from the tower and its collapse, in addition to samples collected from the internal structures inside the perimeter wall (Table 2). The information on the layout and architectural plan of these structures is limited because of the absence of archaeological excavations, but their location and the georeferencing of the samples (Figure 5) allow us to attribute them to different architectural elements. Another noteworthy aspect was identifying several marks on five of the mudbricks analysed – VIL 15, 28, 29, 30, and 31 – in the shape of circles, blades, and impressions, which are currently under study.

Figure 5

Aerial view of Villares de la Encarnación from the southeast and locations of samples (Source: Paisajes Españoles).

The samples have been analysed by X-ray fluorescence (XRF) and CHN elemental analyses in order to characterize their chemical composition and explore possible cluster formations that can help us identify raw source materials (Georgakopoulou, Hein, Müller, & Kiriatzi, 2017; Nodarou, Frederick, & Hein, 2008). Sample preparation consisted of homogenising the mudbricks with a Mixer Mill (MM 400, Retsch) to prepare the pressed beads using 8 g of sample and 2 g of wax that were analysed by a commercial spectrometer wavelength dispersive X-ray fluorescence (WDXRF) (Bruker S4 Pioneer). The analysis was performed in vacuum mode, allowing the detection of low-presence elements. The elemental compounds determined were Na₂O, MgO, Al₂O₃, SiO₂, P₂O₅, SO₃, Cl, K₂O, CaO, TiO₂, V₂O₅, Cr₂O₃, MnO, Fe₂O₃, CoO, NiO, CuO, ZnO, As₂O₃, Ga₂O₃, Rb₂O, SrO, Y₂O₃, ZrO₂, Nb₂O₅, PbO₂, BaO, and La₂O₃. The elemental analysis of carbon (C) and hydrogen (H) was performed using a 628 Series Elemental Determinator (LECO) to record the volatile elements contained in the samples in the form of carbon dioxide (CO₂) and water (H₂O). The sum of major, minor, and trace element concentrations and CH is located within a range of 99.9–100%, and all the elements are expressed as oxide concentrations (Table S1). In our analysis, we included elements that were not affected by postdepositional processes.

The chemical analyses were integrated with the study of the assemblage through thin-section petrography. Using XRF in association with petrographic analysis is an effective technique that allows the study of geochemical fingerprints and fabric compositions, thus bringing the study of earthen material to the forefront of archaeological research and integrating two diverse types of data. The petrographic analyses were carried out using a polarizing microscope, working with a magnification between 5× and 40×. The samples were described and grouped based on the description system of ceramics (Quinn, 2013; Whitbread, 1995) and soils (Bullock, 1985; Nicosia & Stoops, 2017; Stoops, Marcelino, & Mees, 2010).

4 Results

4.1 Geochemical Analyses: WDXRF and CHN Elemental Analyses

The chemical composition of Villares samples detected by WDXRF was evaluated through statistical analysis performed with R (R Core Team, 2023). The elements retained for the analysis are as follows: Na₂O, MgO, Al₂O₃, SiO₂, P₂O₅, SO₃, K₂O, CaO, TiO₂, V₂O₅, Cr₂O₃, MnO, Fe₂O₃, NiO, CuO, ZnO, Rb₂O, SrO, Y₂O₃, ZrO₂, and BaO (Table 3). Furthermore, the PCA incorporates the CO₂ and H₂O from CHN. The rest of the major and minor elements were discarded by low analytical precision (Table S1).

Table 3

Mean (m) and standard deviation (sd) of the chemical composition groups and loner samples cited in the text

	Cluster VIL-A (n = 15)		Cluster VIL-B (n = 12)		Cluster VIL-C (n = 2)		Loner VIL 5	Loner VIL 15
Elements	m	sd	m	sd	m	sd
Na2O	0.0888	0.0120	0.0928	0.0197	0.0585	0.0120	0.0450	0.1310
MgO	0.8881	0.1540	1.0835	0.1328	1.1250	0.0354	0.7800	0.9540
Al2O3	4.9203	0.4343	6.4389	0.3841	6.1810	0.0283	4.5700	4.2100
SiO2	15.9673	1.9594	19.5633	1.3934	21.9800	1.0748	14.5300	14.4500
P2O5	0.0716	0.0110	0.0848	0.0143	0.0922	0.0139	0.1170	0.0520
SO3	0.1727	0.0545	0.1031	0.0278	0.0793	0.0248	0.1840	0.1990
K2O	0.9533	0.1449	1.3918	0.1408	0.7110	0.1061	0.5490	0.8690
CaO	39.3100	1.8679	35.5767	1.7495	34.3900	0.7495	41.6200	40.0600
TiO2	0.2976	0.0289	0.3750	0.0436	0.3630	0.0141	0.2880	0.2480
V2O5	0.0082	0.0014	0.0088	0.0017	0.0084	0.0014	0.0072	0.0021
Cr2O3	0.0066	0.0010	0.0078	0.0012	0.0057	0.0014	0.0098	0.0057
MnO	0.0344	0.0123	0.0387	0.0105	0.0496	0.0007	0.0400	0.0271
Fe2O3	2.2583	0.2486	2.7555	0.2591	2.6615	0.1704	2.2180	1.5930
NiO	0.0032	0.0009	0.0037	0.0005	0.0035	0.0010	0.0033	0.0015
CuO	0.0043	0.0004	0.0050	0.0007	0.0044	0.0008	0.0045	0.0033
ZnO	0.0053	0.0008	0.0066	0.0006	0.0065	0.0001	0.0053	0.0029
Rb2O	0.0031	0.0005	0.0054	0.0009	0.0019	0.0004	0.0023	0.0024
SrO	0.0689	0.0131	0.0530	0.0060	0.0609	0.0156	0.0740	0.0727
Y2O3	0.0017	0.0003	0.0019	0.0003	0.0024	0.0006	0.0016	0.0010
ZrO2	0.0140	0.0025	0.0149	0.0023	0.0177	0.0040	0.0138	0.0118
BaO	0.0232	0.0062	0.0243	0.0041	0.0240	0.0014	0.0180	0.0180

The PCA performed with the normalised values explores the chemical clustering of the assemblage showing two main preliminary clusters as well as a small separate group (Figure 6). Although the PCA only comprises 61.2% of the observed variance in the data, explaining PC1 47% and PC2 14.2% of the total variability, respectively, these groupings are confirmed by the hierarchical clustering analysis (HCA) based on the squared Euclidean distance and the average algorithm. The resulting dendrogram (Figure 7) presents a main branch subdivided into two clear clusters, VIL-A and VIL-B, a small grouping composed of two samples, VIL-C, and one individual outlier. Although the chemical composition is generally similar, the differences between certain elements drive us to make these divisions.

Figure 6

Results of PCA considering 23 variables (Na₂O, MgO, Al₂O₃, SiO₂, P₂O₅, SO₃, K₂O, CaO, TiO₂, V₂O₅, Cr₂O₃, MnO, Fe₂O₃, NiO, CuO, ZnO, Rb₂O, SrO, Y₂O₃, ZrO₂, BaO, CO₂, and H₂O) for the 31 mudbrick samples collected at Villares de la Encarnación.

Figure 7

Dendrogram of 31 mudbricks from Villares de la Encarnación after the hierarchical cluster analysis, CLR transformed, including archaeological location and petrographic fabric.

VIL-A (n = 15) shows higher values of CaO and SrO compared to VIL-B (n = 12), which presents samples with lower concentrations of CaO and SrO but high concentrations of Al₂O₃, SiO₂, K₂O, TiO₂, Fe₂O₃, and Rb₂O. Both groups are also easily recognized in the PCA. Regarding VIL 5, this sample presents exceptionally low levels of Na₂O and K₂O, separating it from VIL-A, although it has similar values of Al₂O₃, SiO₂, CaO, TiO₂, and Fe₂O₃. Similarly, the two individuals of VIL-C, samples 22 and 24, present low values of Na₂O, K₂O, and Rb₂O, while the rest of the chemical values show strong similarities with VIL-B. Finally, the chemical composition of VIL 15 is particularly interesting to us because it is an outlier in both HCA and PCA. Most chemical values are comparable with those defined in the main clusters, but the low presence of Fe₂O₃ and ZnO separates it from the main mudbrick assemblage of Villares de la Encarnación.

In this regard, the existence of two main clusters, VIL-A (including VIL 5) and VIL-B/C, is also supported by the complementary statistical analyses that we have carried out. The ternary phase diagram SiO₂–Al₂O₃–CaO + MgO confirms the existence of these two clusters, which are differentiated by the presence of the calcareous or silicate materials, highlighting the distinctly calcareous clay in the mudbricks (Figure 8). CHN results also show a pattern in which VIL-A exhibits higher CO₂ percentages than samples belonging to VIL-B and VIL-C (Table 3), strengthening our previous interpretation of the geochemical results.

Figure 8

Triangular scattergram SiO₂−Al₂O₃−CaO + MgO, including the archaeological location of the mudbricks.

Yet, there is an exception with sample VIL 23. This mudbrick clusters with group VIL-B/C in the PCA and the triangular scattergram, but in the hierarchical cluster analysis, it appears in the VIL-A branch. The analysis of its chemical composition reveals high values of MgO, Al₂O₃, and SiO₂, and low values of CaO, an image close to the VIL-B/C averages, but a high value of SO₃ has also been detected in this sample. This latter characteristic could be linked to a chemical anomaly or contamination, and it could explain the group swap, although the chemical – and petrographic – evidence points to its inclusion in the VIL-B/C cluster.

4.2 OM Petrographic Results

Thin-section analysis allowed the differentiation of three petrographic fabrics (PF) containing the mudbrick samples from Villares de la Encarnación (Figure 9). The best-represented fabric group is VIL-1 (n = 22), where internal differences have also been identified, leading to the differentiation into subgroups. With fewer individuals, we have recognized the groups VIL-2 (n = 7) and VIL-3 (n = 2). The petrographic study follows a similar trend to that shown in the chemical results, indicating that combining different analyses in archaeometry is necessary for data accuracy and feedback.

Figure 9

Representative photomicrographs of petrographic groups identified among the analysed mudbricks from Villares de la Encarnación, crossed polars (XP).

Fabric 1: limestones and biocalcarenites (n = 22). The main characteristic of this fabric is the widespread presence of limestones of marine origin, resulting in a wide variety of this type of inclusion, which includes biocalcarenites and biocalcirudites. The coarse fraction is dominated by coarse-grained calcite, biocalcirudites mainly represented by bryozoan fragments, bioclast particles, and, in a lesser proportion, rounded bioclast wackestone and benthic foraminifers, echinoderms, and isolated shells. The size of these inclusions varies over a wide range from very fine sand (<0.05 mm) to pebbles (8–10 mm). Other inclusions also present in the matrix at a low frequency are monocrystalline quartz, subangular chert, and rare mudstone particles. The structure of the coarse fraction is poorly sorted, and its distribution is generally single to double-spaced, although this varies according to the subgroups that make up this fabric. The groundmass has a slightly orange-brown colour, with a fine fraction composed of quartz, limestone, and calcite inclusions, and the presence of embedded microfossils is very rare. Concerning the voids, they mainly present an elongated and slightly curved shape, as well as vughs, that should be linked to using vegetal temper while preparing the soils (Figures 10a–10c). The voids exhibit secondary calcite in some exceptional cases (Figure 10c).

Figure 10

Detailed images of some of the identified voids linked to vegetal temper (a, b, c, d, e; plane polarized light [PPL]), presence of secondary calcite in the voids (c), traces of degraded vegetal remains (e) and orange clay coating creating a rolling pedofeature in some clasts (b and f, indicated with an arrow) from Villares EBM assemblage. Crossed polar light (XP) in 10.f.

Based on the frequency and grain size of the inclusions, we identified three subgroups within Fabric 1. Sub-fabric 1.1 (n = 6; c:f:v: ca. 40:50:10; samples VIL 2, 6, 9, 13, 18, and 19) includes the samples which, following the characteristics of the group, present a dominant coarse fraction and their matrix is characterized by a high percentage of inclusions. The higher presence of sub-rounded to angular monocrystalline quartz suggests that sand plays a major role in this subgroup. Sub-fabric 1.2 (n = 9; c:f:v: ca. 20:70:10; samples 1, 3, 4, 11, 15, 20, 21, 30, and 31) shows the same type of aplastic inclusions, but of smaller size and in lower frequency. In two of the samples, VIL 1 and 11, we have identified a continuous orange clay coating creating a rolling pedofeature surrounding some clasts (Figure 10f), a characteristic that can be linked to the kinesthetic process that presses the mud mixture into itself in a rolling pattern (Lorenzon & Iacovou, 2019). Finally, Sub-fabric 1.3 (n = 7; ca. 10:80:10; samples VIL 22, 23, 24, 26, 27, 28, and 29) is characterized by the dominance of a fine fraction composed of large amounts of limestone, calcite, and quartz particles, although some limestone and biocalcarenite grains and small pebble are also eventually found in the coarse fraction.

Fabric 2: bioclasts and limestone inclusions in a microfossil-rich soil (n = 7; c:f:v: ca. 35:55:10; samples VIL 5, 7, 10, 12, 14, 16 and 17). The groundmass exhibits a homogeneous deep dark brown and densely packed structure where the aplastic inclusion represents very fine sand (0.1–0.05 mm) to coarse pebbles (ca. 8–12 mm) that follow a poorly sorted distribution. The space distribution of the particles is measured within the single to double-spaced frame. The coarse fraction is mainly represented by subrounded limestones, including coarse-grained calcite with polygonal grain boundaries, biocalcarenites, and oolitic limestones. A small part of these particles also present evidence of dolomitization. The other main characteristic of this fabric is the common presence of microfossils embedded in both the matrix and the limestone grains. The former includes fragments of bryozoans, benthic and planktonic foraminifera, and echinoids, as well as bivalve fragments in very low frequencies. Other inclusions of the coarse fraction in smaller proportion are subrounded to rounded monocrystalline quartz, round mudstone, and subangular chert. The fine fraction is mainly composed of calcite and reddish-brown mud particles. The voids are not so frequent in the samples of this fabric, but they reach a large size with a structure of planar voids and vughs clearly linked to vegetal temper, sometimes exceeding 2 µm in thickness (Figure 10d).

Fabric 3: fine matrix with predominant mudstone (n = 2; c:f:v: ca. 5:85:10; samples VIL 8 and 25). The groundmass presents a deep orange-brown background and consistent structure with few aplastic inclusions except for the predominant presence of a very homogeneous greyish-green mudstone. The coarse fraction is barely represented by isolated and small inclusions of angular monocrystalline quartz, together with subangular to rounded limestone, oolites, and biocalcarenite particles. Very rare microfossils have also been identified, such as benthic foraminifera and algae neomorphized in calcite. The fine fraction is mainly composed of quartz and calcite sand. The space distribution of the inclusions, except the mudstone, is measured within the single to double-spaced frame. The presence of voids is irregular, and they are not oriented to present a random basic distribution pattern. The samples present channels caused by internal microfractures and planar voids, vughs and vesicles linked to vegetal temper (Figure 10e). The use of human-induced tempering can be directly verified since, in sample VIL 8, abundant traces of degraded vegetal remains have been preserved. Only in exceptional cases do the voids show some secondary calcite.

5 Discussion and Archaeological Implications

The analyses carried out on the mudbricks from the tower and internal structures of Villares identify different clusters and allow us to classify them into two main groups and one outlier sample. Group A corresponding to chemical cluster A and PF 1.2 and 2 (samples VIL 1, 3, 4, 5, 7, 10, 11, 12, 14, 16, 17, 20, 21, 30, and 31); Group B corresponding to chemical clusters B and C, and fabrics 1.1, 1.3, and 3 (samples VIL 2, 6, 8, 9, 13, 18, 19, 22, 23, 24, 25, 26, 27, 28, and 29). The isolated sample VIL 15 corresponds to fabric 1.2, but its chemical composition suggests a differentiation from the two previous clusters. The combination of different analytical techniques has been effective, as indicated by the match between the chemical and petrographic results, allowing us to identify these specific groups even though they share common compositional and geological features.

In this regard, the PFs exhibit certain characteristics that connect directly with the geological deposits near the settlement, revealing the local origin of the selected sediments used in mudbrick manufacture. The geology of the Estrecho de la Encarnación area is mainly characterized by Neogene deposits rich in biogenic limestones and silts, which are complemented by older geological deposits such as Jurassic and Cretacic layers characterized by limestones, dolomites, marlstones, conglomerates, and gypsum. As mentioned above, the settlement of Villares is located on a hill made up of Tortonian marine biocalcarenites, a deposit that expands towards the southern slopes of the site. There is also an important Pliocene deposit southwest of the site, characterized by conglomerates, silts, and clays of marine origin. Finally, marls are also present along the Quípar river, where the watercourse has created a Quaternary layer of alluvial gravel and sand (Figure 11).

Figure 11

Geological map of Estrecho de la Encarnación area, with the Early Iron Age site of Villares (Source: Cutillas-Victoria, based on the Geological and Mining Institute of Spain, sheet number 931 [Baena Pérez, 1972] and the PNOA-Spanish National Geographic Institute).

With this geological landscape, it is a complex issue to propose possible raw material source procurement areas for Group A since two different PF are located within this chemical cluster. Although there are common characteristics, such as the coarse fraction, the main difference between fabric 1.2 (n = 9) and fabric 2 (n = 7) is the presence of a microfossiliferous matrix in the latter. Nevertheless, the grouping of the samples from these fabrics suggests that the two procurement areas were close to each other and in the vicinity of the site. At 650 m SW of the settlement, we have the so-called Barranco de los Canteros (Figure 11), which is characterized by a Pliocene deposit with clays of marine origin. This latter deposit is located near the biogenic limestone on which the site is, and it could explain the presence of the microfossils’ inclusions in the mudbrick sediment. Thus, the source of raw material procurement of fabric 2 might be located in this ravine, perhaps close to the Quípar river, which presents characteristics close to the fabric 1.2 fine fraction with a sandy matrix and the presence of biocalcarenites and limestones as aplastic inclusions. Still, the three PF classified as Group A – fabrics 1.1, 1.3, and 3 – are quite similar to each other and point to a very localized procurement area. Thus, we consider that this group of manufacturers had a good knowledge of the suitable sediments close to the site.

Group B comprises two different but compatible PF, which could be related to the Neogene deposits of limestone and biocalcarenite material that constitute the lithological deposit on which the settlement is located. On the one hand, the difference between fabric 1.1 (n = 6) and fabric 1.3 (n = 7) is based on the frequency and size of the aplastic inclusions in the coarse fraction since the characteristics of the fine fraction are similar between the two fabrics. However, this separation is not random, but we consider that it portrays the existence of two extraction sources within the same catchment area, a hypothesis also supported by the chemical data as the composition of fabric 1.1 is homogeneous. The characteristics of the fine fraction are very similar to those seen in fabric 3 (n = 2), although in this case, the predominant presence of mudstones separates them, as indicated by other techniques such as the CH elemental analysis. Thus, it is likely that the sediment procurement areas for the Group B mudbricks were located close to each other in areas where the sediments were rich in sand and mudstone inclusions, such as a riverbed. On the basis of the detailed geological map (Figure 11), the presence of limestone and biocalcarenite materials, and our own analysis, we propose the Quípar river, probably on its path to the west of Villares de la Encarnación, as the raw material procurement area for these mudbricks.

Although these proposed procurement areas are indicative and need further investigation, including sediment sampling, what is evident is the existence of two different raw source procurement areas that served as the basis for the chaîne opératoire of different groups working in mudbrick manufacturing. This hypothesis is not only supported by the geochemical fingerprint but also by the PF. The introduction of temper and stabilizers indicates articulated craftsman strategies in manufacturing these EBM. The use of vegetal temper as a binder or stabilizer in mudbricks is a well-evidenced process that reinforces the structure of these materials (Lorenzon, 2021; Love, 2012), and its quantification in thin sections makes it possible to extract data about recipes and sediment treatments as in the present case study. For instance, Group B comprises numerous voids associated with organic temper and significant SiO₂, which we correlate with sand content (Oliver, 2008), providing us with a recipe that showcases relevant human-induced interventions to transform the local sediment. Other examples are the samples VIL 8 and 25, belonging to fabric 3, which present high CO₂ values and numerous planar voids in thin sections linked to the quantity of vegetal temper added as a degreaser to balance a fabric rich in mudstone and the lack of other aplastic inclusions.

Therefore, we admit that it is difficult to assess the degree of specialization of the two groups who made the mudbricks from Villares, being more feasible to speak of two types of know-how reflecting different areas of raw material procurement and treatment of the mixture based on available environmental and climate resources. The print marks currently under study also contribute to this picture (Figure 12), probably revealing specific gestures of both groups that show a willingness to identify the products of each workshop. Even VIL 15, the geochemically isolated sample, presents a different mark with three fingerprints and a smaller module than the rest of the mudbricks, which could also point to a third manufacturing group or a specific variation for a particular reason. This need for singularity could also indicate the difference between the production and construction domains. One of the main results of the geoarchaeological analysis is demonstrating the use of mudbricks of different provenances for the building of each of the constructions analysed. The results from the structures with the largest sampling indicate that the tower has mudbrick of Group A (fabrics 1.2 and 2), Group B (fabric 1.1 and 3), and the lone individual VIL 15, while the internal NE structure is composed of mudbricks from Group A (fabric 1.2) and Group B (fabrics 1.3 and 3).

Figure 12

Print marks observed on mudbricks VIL 28 (left) and VIL 30 (right) during the sampling conducted at the site in 2023 (Author: F. Brotons).

In this regard, the results of this pilot study reveal the technological and socio-economic complexity behind the production of mudbricks at Villares de la Encarnación, as well as at the construction level, the use of materials from different production workshops. The indiscriminate employment of diverse mudbricks in the same structures is attested, thanks to the results of the mudbricks VIL-1 to VIL-5, which are found in situ and are part of the same bricklaying in the main tower.

The incorporation of adobe as a building material for various parts of the perimeter fortification and some annexed structures may have been due to the new possibilities and the speed with which these materials could be worked in a probable phase of expansion of the settlement. However, there are still important research questions to be answered, particularly about the integration of this material of foreign origin: was it already used by this autochthonous community in older domestic structures or even in the transversal wall? In addition to the articulation of the structures analysed here, the above question about the adaptation of mudbrick as one of the EBM in the region is one of the main aims of the project that we hope to be able to develop at the site through extensive research and archaeological excavation. The investigation of other sectors of the walls and intra-site will allow us to understand whether or not there are other mudbrick remains and their functionality, providing new data to understand when and how the integration of this new construction technique took place, and whether the structures analysed in this work are the result of dynamics of change or continuity with respect to the construction patterns of the inland Early Iron communities of the Iberian southeast.

6 Final Remarks and New Perspectives

The preliminary study of the mudbricks from Villares de la Encarnación reveals the complexity of mudbrick construction on the Iberian Peninsula during the Early Iron Age. Chemical and petrographic techniques reveal the existence of various clusters that, despite indicating the material is of local origin, suggest different procurement areas and treatment of the sediments from which the mudbricks were made. On the one hand, this complexity is surprising because mudbricks were hardly used in the ramparts of autochthonous settlements, in contradistinction to the walls of Villares. On the other hand, mudbricks were a well-known construction material for these communities, as revealed by their previous use of mudbricks in domestic architecture throughout the Iberian southeast.

Despite the absence of extensive archaeological excavations, the analytical characterization of the mudbricks suggests the existence of several social groups working in manufacturing these materials, which also reveals the importance of this material among the socio-economic activities of the Early Iron Age communities. However, it should be noted that there is no relationship pattern between the types of mudbricks produced and their use in specific structures. In other words, mudbricks of different raw material provenances were used for the building of the tower as well as in the interior structures. This situation could indicate two distinct dimensions between the mudbrick manufacturers and the builders, i.e., the manufacturing of the mudbricks was separated from the follow-up construction. To explore these hypotheses further, we plan to start archaeological excavations at the site in the near future. The fieldwork will be essential to understanding this settlement development and its dynamics of interaction with other autochthonous and Phoenician communities, including specific aspects such as construction practices and the integration of elements of Levantine tradition, such as the introduction of mudbrick architecture.

Finally, this study is also a wake-up call on the potential of these earthen materials for research on the Iberian Peninsula. This perspective has been highlighted in recent years (Pastor Quiles, 2017), but it still offers new possibilities if analytical techniques are applied as they are being done in Near Eastern and eastern Mediterranean scenarios (e.g., Cammas, 2018; Homsher, 2012; Lorenzon & Iacovou, 2019; Love, 2012). The rigorous application of analysis and intervention protocols in the excavation and the laboratory will depend on our understanding and exploitation of these materials as archaeological sources of enormous potential.