Sources of geomaterials in the Sicani Mountains during the Early Middle Ages: A case study of Contrada Castro, central western Sicily

From 2017, an unknown rural settlement in Contrada Castro at Corleone (Palermo Province, western Sicily) was investigated as part of the ‘Harvesting Memories Project’. The stratigraphic sequence, supported by radiocarbon dating, has demonstrated a reoccupation of a pre‐Roman site during the transition between the Byzantine and Islamic periods. In particular, the main occupation occurred in the late 8th–9th century when pottery kilns and a probable warehouse were constructed. During the 10th–11th century, a new structure with different orientations replaced the previous buildings that had already collapsed. Specifically focusing on a perspective of the household production and its relationship to the surrounding landscapes of this site, this paper presents the results of a geological survey and petrographic analysis of ceramic finds and lithic samples to identify the source area of the geomaterials used in the studied settlement. The ceramic finds were divided into different Paste Groups based on the characteristics of a polarizing microscope study. The use of some locally available raw materials was recorded both for ceramic and lithic samples. Such an approach enables us to better understand ceramic craft technology, clay and lithic procurement strategies and, more broadly, the consumption of household pottery in the Early Medieval site in Contrada Castro. Furthermore, this study verifies the close relationship that this rural settlement had with the surrounding resources in the area and reveals a connection with the city of Palermo during the Early Middle Ages.


| INTRODUCTION
In recent decades, interest in the archaeology of early medieval Sicily, which includes the Byzantine and Islamic periods (6th-11th century, AD), has grown considerably (Ardizzone & Nef, 2014;Molinari, 2016;Nef & Prigent, 2010). Specifically, owing to an increase in the knowledge of the chronological indicators in early medieval pottery in Sicily (Arcifa, 2010;Sacco, 2020), a better comprehension of the evidence for this period is possible. Archaeological surveys and new excavations have lent a new impetus to the debate. However, the period between the end of the 7th and 9th centuries remains the least known from an archaeological perspective for rural and urban areas. Specifically, in the rural world, little evidence of Byzantine and/ or Islamic settlements exists. Examples of these include the Byzantine castrum of Monte Kassar (Carver et al., 2019;Vassallo et al., 2015), the Byzantine village of Rocchicella di Mineo (Arcifa, 2016), a rural site with Byzantine and Islamic phases of Colmitella in the province of Agrigento (Rizzo, 2014) and the Islamic fortified granary of Pizzo Monaco (Rotolo & Martín Civantos, 2013). Byzantine and Islamic structures, which showed long-term occupation (Castrorao Barba, 2016), have been found more frequently within Roman sites, such as the village of the Islamic and Norman period built on the site of the late antique Villa del Casale at Piazza Armerina in the territory of Enna (Pensabene & Bonanno, 2008), in the large villages along the roadside including Casale San Pietro in Castronovo di Sicilia (Carver et al., 2019; and in Sofiana/Philosophiana (Vaccaro, 2017).
The discovery of new late Byzantine and rural Islamic sites in Contrada Castro (Corleone, Palermo) ( Figure 1) is an important opportunity to gain knowledge about this 'dark age' in the history of the Sicilian countryside (Castrorao Barba et al., 2018. As a part of the 'Harvesting Memories: Ecology and Archaeology of Monti Sicani landscapes (central-western Sicily)' project, the investigations into the Contrada Castro site have identified an occupation sequence during the period between the 7th and 11th centuries, which were well supported by radiocarbon dating (Castrorao . The essential elements for reconstructing the 'material culture' of ancient rural communities include assessing technological expertise, trading circuits and cultural influences (Tite, 2008). This can be achieved by laboratory-based compositional characterization of the artefacts in the form of geomaterials in their natural state or after a transformation by pyrotechnology (i.e., ceramics), as well as by identifying raw material-supplying areas. By identifying pottery kilns related to the period of the late 8th-9th century, the relationship between Contrada Castro and the surrounding geomaterials, as well as the connectivity with other far areas, can be established. In this case study, provenance determination of archaeological samples was supported by a database concerning all the clayey materials traditionally destined for ceramic production in western and central Sicily (Montana, Cau, et al., 2011). Mountains. This plateau is occupied by the remains of interconnected enclosures with 1-m-high dry walls, which were used for transhumant herding activities and were constructed under dry conditions using square blocks of various sizes.
The first excavation area was opened in the southeastern part of the plateau (Castrorao Barba et al., 2018Barba et al., ) during 2017Barba et al., , 2018 and partially in 2019. According to the current research (summer 2019), the more ancient evidence of the site's occupation dates back to the late archaic/classical period (6th-5th c. B.C.). This is supported by a residual material and a collapsed squared/rectangular building, which have not yet been fully investigated. According to the current survey and excavation data, the site was reoccupied in the Byzantine period after a long hiatus. The only evidence above the classical period layers comprises two perinatal burials. The death of these individuals occurred at a gestational age of 40 ± 2 weeks, estimated according to the regression formulas of Scheuer et al. (1980). Based on a radiocarbon analysis (Castrorao Barba et al., 2018) of bone samples, the deaths of these individuals date back to the late 7th to early 8th c. AD (at confidence levels of 1-sigma [65%] AD 662 to AD 778, or 2-sigma [95%] AD 620 to AD 906). This chronology is confirmed by the discovery of a roof tile decorated with stamped striations that are typical of the Byzantine period in Sicily (e.g., Arcifa, 2010, pp. 108-111) on the surface of the layer where the burial area is cut.
The main occupation of this site can be articulated into two different periods: period A (late 8th-9th c. AD); period B (10th-11th c. AD).

| Period A (late 8th-9th c. AD)
A squared 5 × 5 m building was constructed in this area not much later than the indicated period (probably in the second half of the 8th c. AD), thereby representing Phase A1. This building is partially sunken, cutting into the bedrock in the south and in an earlier layer to the west (i.e., the earlier layer was cut by the construction of this building during a later phase), while the entrance to the building is located on the east side through a sloping ramp. The building was likely not entirely covered in the first phase, containing only a partial roof made of tiles. However, inside of the house, craft activities related to pottery and tile production took place. In fact, on the south side of the building, the remains of two circular kilns are connected to burnished layers and a large number of productive wasters (Figure 3).
In particular, both kiln and ceramic wasters have been found (Figure 3 (1)) such as amorphous fragments featuring a purplish/greyish colouration, a bulgy surface and a bloated texture belong most likely belonging to the clay's structural features of the oven (either the floors/walls of the chamber or the kiln's dome); a whole stack of deformed pottery (most probably combed tiles), also, in this case, the purplish surface, burnt core and bloated clay body, point towards a failure in the firing process that caused the artefacts to collapse and fuse.
The first kiln was abandoned early on, while the second one was well preserved and was built nearby and had a circular stone structure. Directly in front of the kiln mouth, the soil was strongly reddened by thick ash layers filled with charcoal derived from combustion processes. One of the charcoals, a fragment from a small shrub of Pistacia, was dated using a radiocarbon analysis (Castrorao , which confirmed that the building was used between the second half of the 8th and 9th centuries (reported 1-sigma [65%] accuracy age of AD 774 to AD 878; 2-sigma [95%] AD 762 to AD 900). The inner parts of the firing chamber and dome were probably made from yellow clay that was also used to create a curblike structure around a stone-made pilaster that supports a possible cooking floor. These structures look much simpler than the other examples found in Sofiana/Philosophiana in Central-Eastern Sicily (Vaccaro & La Torre, 2015), and they were dated to the late 8th and 9th c. AD. Evidence for a possible cooking floor has not been found because the building was transformed during the second phase (Phase A2): the western entrance was closed, restorations were identified along the eastern walls, a new earth floor covered the disused south-east kiln, the other kiln was converted into a cooking oven (possibly for bread) and a hearth base was added in the northern part. This building completely collapsed, as confirmed by a thick stone layer (Phase A3). A radiocarbon analysis  performed on an animal bone found in this layer indicates a 9th-century date for the collapse (age at 1-sigma, 65% accuracy, of AD 800 to AD 878; age at 2-sigma, 95% accuracy, of AD 766 to AD 897).  (Troia et al., 2012). In fact, despite long-term exploitation, the area still maintains a high level of plant diversity and a well-preserved rural landscape Bazan, Speciale, et al., 2020;Gianguzzi & Bazan, 2019.

| MATERIALS AND METHODS
The sample objects of this study (29 samples) are mostly representative of period 3 and, to a lesser extent, period 4 of the Contrada Castro settlement (Table 1). They were selected for the purpose of representing the different categories of local natural and transformed geomaterials: ceramic kiln waste (4 samples), tiles (10 samples), daily use pottery (8 samples), raw clay (2 samples), bread baking stone plate (2 samples) and stone ashlars (3 samples). All the collected samples were subjected to petrographic analysis by optical microscopy using transmitted polarized light (section thickness of 0.03 mm) to define their textural and compositional characteristics and identify the corresponding raw material.
The assessment of textural and mineralogical similarities or differences between specific geomaterials (i.e., raw clays) and artefacts may allow compositional groups and productive contexts to be established and allow archaeologists to make historical and socioeconomic inferences (Montana, 2020 and references therein). Specifically, in the case of ceramic artefacts, thin-section microscopy can significantly contribute to establishing a production chain, distribution and use, as well as to infer the technical skills and even the social levels of the involved cultures (Quinn, 2013). For these purposes, the quantification of monomineralic grains and lithic fragments constituting the aplastic components of the ceramic paste highlights several diagnostic parameters that allow the correct provenance identification.
Thin-section petrography was performed using a Leica DC 200 polarizing microscope equipped with a digital camera. The relative abundances (modal mineralogy expressed in area percentage, %) of aplastic inclusions were determined using comparative tables (Matthew, 1991;Van der Plas & Tobi, 1965).

| RESULTS
All the ceramic samples, regardless of their intended use, were characterized by petrographic observations based on compositional and textural characteristics. Based on the obtained petrographic data, they were divided into four different paste groups according to the nature, abundance and size of minerals and lithic fragments composing the aplastic inclusions. With the aim of recognizing local productions and distinguishing them from F I G U R E 5 Geological map of the Contrada Castro area with details on the lithologies outcropping in the study area and in the immediate surroundings [Color figure can be viewed at wileyonlinelibrary.com] imports, a cross-comparison was made with the lithological characteristics of the local clayey raw materials, which are considered suitable for ceramic use. This provenance study conducted using thin-section petrography was basically supported by an available database on the microfabric acquired after experimental firings of all the clayey materials traditionally destined for ceramic production in western and central Sicily (Montana, Cau, et al., 2011).
With regard to the lithoid samples constituting bread baking plates (samples CST-22 and CST-23) or structural elements (samples CST-27, CST-28, CST-29), which were discovered after the excavation, a similar procedure was adopted: petrographic characterization and comparison with the lithologies emerging in the surroundings of the Contrada Castro.

| Local clayey materials
Two of the samples analysed can be considered as representative of the ceramic raw materials that are available in the close vicinity of the

| Ceramic materials
The minero-petrographic paste group MPP-1 includes samples CST-1, CST-2, CST-3 and CST-4 (Table 1) The latter is an indication of quick and coarse processing of the ceramic paste. The compositional and textural characteristics found for MPP-1 are well correlated with those known for the clays of the Castellana Formation (outcrops 2-3 km away from the site of Contrada Castro) as described by Montana, Cau, et al. (2011) and .
The minero-petrographic paste group MPP-2 includes the following samples: CST-5, CTS-6, CST-8, CST-9, CST-13 and CTS-14 (Table 1) Based on the microscopic observations, we conclude that the raw clay material was likely sourced from the Marne di Cardellia Formation (outcrops~1 km from the site of Contrada Castro).
The minero-petrographic paste group MPP-3 includes samples CST-7, CTS-10, CTS-11 and CST-12 (Table 1) (Montana, Cau, et al., 2011;. Therefore, it is believed that the ceramic products described above were produced in Palermo and imported to the study site. The minero-petrographic paste group MPP-5 includes samples CST-20, CST-21 and CST-24 (Table 1). This ceramic paste is re- Due to their peculiar textural and compositional characteristics, two ceramic samples, CST-16 (amphora) and CST-19 (cooking pot/ olla), could not be assigned to any of the ceramic paste groups previously described and were, therefore, classified as ungrouped objects, which were likely not locally produced.
Object CST-16 is characterized by a paste that was derived from the firing of a marly clay, with sporadic and very fine-grained quartz show an evident preferential orientation, likely due to working by the pottery wheel. Based on these observations, we conclude that the potential raw clay material used for the object CST-19 was sourced from the Flysch Numidico Formation that outcrops~70 km away (Palermo/Monreale area) from the site of Contrada Castro (Montana, Cau, et al., 2011;. The CST-23 sample can be described as planktonic foraminiferal (Globigerina) limestone, with particle size ranging from 0.06 to 0.2 mm. Benthic foraminifera were much more subordinate. The texture was grain-supported, with an intergranular micritic matrix that is particularly rich in iron oxides. No sparry carbonate cement was observed; therefore, the lithotype was petrographically classified as packstone (Dunham, 1962). Very fine-grained quartz The CST-29 sample was slightly different from the above description due to a relatively greater abundance of monocrystalline quartz grains (angular to subangular). Glauconite particles were also common ( Figure 9d). This lithotype matches the raw material used to make the stone bread baking plate CST-22. The textural and biostratigraphic features allow this sample to be associated to the Marne di Cardellia Formation that crops out~0.5 km away from the studied site (Basilone, 2011;Di Stefano et al., 2013).

| DISCUSSION
The analysis carried out on the studied geomaterials allows us to make a series of interpretations regarding the relationship between the sources of materials and the different objects. In fact, the main aim of this study is to attempt to understand the supply dynamics linked to the production context found for period A (late 8th-9th c. AD). which is largely distributed in the region and used in several other production sites (Montana, Cau, et al., 2011; and crop out very close from the site of Contrada Castro (approximately 3 km). The presence of this group of tiles could be interpreted either as the short-distance import of products made in a kiln located outside the site or, more likely, as a result of an internal production, but for which it was decided to use a different raw material more distant than the closest Marne di Cardellia Formation. Of the samples analysed for period A, the cooking pots confirmed a local production linked to the Marne di Cardellia Formation (MPP-5). In fact, the two types of ollae (CST-20 and CST-21) refer to this group of pastes: CST-20 ( Figure 10 (1a,b)), a fragment of the olla with an extroverted and enlarged rim with a small external recess between the attachment of the rim and the shoulder, and CST-21, a fragment of the olla with an everted rim and blackened on the external surface.
The ceramic handmade bread-baking plate (CST-24) ( Figure 11 (1a,b)), also known as 'testo/testello' (Giannichedda & Zanini, 2011), whose paste has been grouped in MPP-5, is recorded (in terms of typology) in many rural contexts of peninsular Italy from the 6th/7th century throughout the Middle Ages (Augenti, 2016). Furthermore, it has been argued that similar forms of flatbread baking plates were not only recorded in Roman and Byzantine contexts in Egypt and Chios (Reynolds, 2016) but also in southern Spain such as in Visigothic and Byzantine sites and, more often, related to the so-called ṭābaq (Gutiérrez Lloret, 1991) Rizzo et al., 2014). In contrast, the most commonly used objects were the so-called 'testi in pietra' (Arcifa & Bagnera, 2014) stone trays/plates. These are well known and closely connected to the Islamization of the island, especially of the western part, from the end of the 9th into the early 10th c. AD (Ardizzone et al., 2018), and were used at least until the 13th century.
The two samples CST-22 and CST-23 can be related to the Islamic period (10th c. AD). In particular, sample CST-22 was made with raw material exploited from local outcrops of the Marne di Cardellia Formation (Figure 11 (2)), while sample CST-23 was made with the planktonic foraminiferal limestone belonging to the Trubi Formation, which outcrop approximately 7 km from the site (Figure 11 (3a,b)). The CST-15 sample ( Figure 10 (2a,b)) relates to a vertical, slightly enlarged amphora with a rib in relief on the outside immediately below the rim, and a series of creases on the neck. On the neck, it has a brown decoration with a motif formed by two parallel and horizontal sinusoids. In addition, there were two compensation holes in the body of the amphora. On the body, it has a painted decoration with vertical bands alternating with sinusoidal bands (typical decoration of late 9th c. AD, but probably also earlier; see Sacco, 2018).
The question remains open about the chronological origin of these amphorae and whether they were already in circulation before the Aghlabid period. The CST-17 sample (Figure 10 (3a,b)) is an amphora with a thickened rim that contains traces of red paint and a slightly pendulous triangular section. This typology currently finds compar- isons with examples found in the late 9th-century contexts during the excavations of Palermo, such as Castello San Pietro, Palazzo Bonagia and the church of the Gancia (Sacco, 2018). The CST-18 ( Figure 10 (4)) is a small flask with a probable hint of spout, a handle with an oval section and a horizontal ear and with a slightly rounded bottom. The decoration covers the entire surface with a continuous horizontal sinusoidal red band, and a sinusoidal pattern alternating with bands was also observed on the bottom. Currently, there is no comparison of this form that has a painted decoration equal to the amphorae with a sinusoidal motif.
It is interesting to note that, to date, the exploitation of Argille di Ficarazzi clays has been related to specific types of amphorae from the Aghlabid period, between the late 9th and beginning of the 10th c. AD (Alaimo & Giarrusso, 2004;. However, the Argille di Ficarazzi clays (Montana, Cau, et al., 2011) were already widely known and exploited for the production of ceramics, at least from the 7th and/or 6th century B.C., well before the arrival of the Muslims in Palermo, and with a continuity up to the present day . Regarding the Contrada Castro site, the stratigraphic sequence is confirmed by robust absolute radiocarbon dating and allows us to date these materials to the 9th century. The site collapsed afterwards and was no longer used at the end of the 9th century. Therefore, we can still hypothesize an early connection between Aghlabid Palermo and the territory of Corleone in the decades following the Islamic conquest of this town in AD 840.
It is not possible to exclude a long-term city-countryside relationship, starting in the Byzantine period and running through to the first half of the 9th century. In fact, in the stratigraphic context of the late 8th to early 9th century, sherds of an olla with an extroverted rim (CST-19) made of clays of the Flysch Numidico Formation, likely coming from an area near the city of Palermo, are an element that can be better explored with further studies and analyses, to verify this city-countryside connectivity in the complex decades that preceded the Islamic conquest of western Sicily. Interestingly, the network of short-distance exchanges is illustrated by the import in the early 10th century of the small amphora (CST-16) (Figure 10 (5)), which was most likely sourced from the territory of Monte Iato, intensely occupied in the Islamic period, especially between the 10th and 11th centuries (Alfano & Sacco, 2014).

| CONCLUDING REMARKS
The discovery of an early medieval production context with kilns for and a rational exploitation of the resources available locally for the production of tiles, cooking ware, daily-use ware, lithic trays, breadbaking plates and building stone materials.
This close link with the sustainability of the resources of the surrounding area was also confirmed by the intersection of information between the phytosociological characteristics of the area, the palaeoenvironmental data on wood species and crops associated with the charcoals and seeds found in the sites and the composition of the zooarchaeological record connected to the strategies of animal husbandry Castrorao Barba et al., 2021).
Despite this, the picture that emerges is not that of an entirely self-sufficient and isolated settlement. In fact, some data with regard to amphorae and, to a lesser extent, cooking pots indicate the existence of a short-scale network of exchanges, at least limited to the area of the Sicani Mountains, and a connection with the city of Palermo. In general, with regard to the local ceramic production at the settlement of Contrada Castro, it should be emphasized that several good raw clays can be sourced within a radius of 2-3 km, which are suitable for both brick (relatively richer of aplastic inclusions, up to 25%) and daily use pottery production. In particular, the use of clay materials sourced from the Terravecchia Formation has been frequently and diachronically recorded in western Sicily, likely due to its excellent quality that is particularly suitable for the production of the finest pottery (Montana, Cau, et al., 2011).
This connectivity of the rural site of Contrada Castro with the city of Palermo reveals new interesting questions about the relationship between the city and its territory in the different and complex phases between the decades immediately preceding the arrival of the Aghlabid army and the consolidation of Palermo as the capital of Islamic Sicily.
Future research and integration with a comparison of contemporary contexts and well-dated stratigraphic contexts from Palermo (that have emerged in recent rescue excavations, Vassallo, 2020) will clarify the nature of this city-countryside relationship and its relationship with the historical macro-changes from the Byzantine to the Islamic periods, while always taking into account the longue durée factors in rural communities and its different and variable rhythms of resilience and response to political regime changes.
In summary, these new data show us how an integrated, interdisciplinary approach including archaeometric analysis, stratigraphic excavation and geological survey of the territory is fundamental for an enhanced appreciation of the economic and settlement dynamics associated with the poorly understood Sicilian countryside during a historical period such as the Early Middle Ages.