1 Introduction

The Sicilian Middle Bronze Age (MBA), identified with the Thapsos culture (fifteenth–thirteenth century BC), represents the most important phase of the Bronze Age with respect to the emergence of a common technological koine for pottery production, characterized by new ways of clay procurement strategies, pottery-making practices, and firing technology (Cuomo di Caprio 2007; Bietti Sestieri 2015; Borgna and Levi 2015). Such a technological koine will be fully established only later in the course of the subsequent Late Bronze Age (LBA) culture of North Pantalica (thirteenth–eleventh century BC) (Tanasi 2008). The circulation of Mycenaean pottery and the contact with foreign and more skilled potters triggered an intense period of experimentalism by local artisans, which resulted in various districts’ production of Siracusan and Etnean Thapsos (Alberti 2013, 2017; Bietti Sestieri 2015; Jones et al. 2014; Russell 2017; Tusa 2000).

After a series of traditional studies which tackled the typological classification of the Thapsos pottery (Alberti 2004, 2008; Tanasi 2008, 2010a, 2015; Veca 2014a, b, 2015), in the last few years new research focused on the technical aspects of the ceramic production using as a case study the evidence from the territory of Catania, in particular from the sites of Monte San Paolillo di Catania and Grotte di Marieneo di Licodia Eubea.

At the core of these technological changes, two main shifts within the production strategies adopted by local communities can be placed. On the one hand, during the second half of the MBA, there is an increasing interest towards high-calcium-oxide-based clays, which will become more common than alluvial clays in use during the earliest part of the MBA (Barone et al. 2011a, b, 2012; Raneri et al. 2015a, b; Rodrıguez et al. 2015, Tanasi et al. 2013; Veca 2014a, b, 2015). On the other hand, such a change in the clay procurement is also accompanied by a change in firing techniques, which started to rely on kilns with underneath chambers instead of those with a side chamber, thus reaching very high temperatures (> 900 °C) during the latest phases of the MBA (Borgna and Levi 2015; Cultrone et al. 2001; Jones et al. 2014; Raneri et al. 2015b).

The opportunity to expand the range of data at our disposal and to attempt a more in-depth analysis of the Thapsos pottery production in the territory of Catania comes from the reassessment of a large assemblage of ceramics collected during excavations undertaken by the Musei Civici “Luigi Sturzo” of Caltagirone in the late 1990s on the hill of St. Ippolito at Caltagirone. The combination of both chemical and petrographic analyses of the mineral composition of ceramics from the hill of Sant’Ippolito has the potential of exploring and assessing some aspects of local ceramic production during the Middle Bronze Age of eastern Sicily.

After more than 10 years since its recovery, the ceramic assemblage from Caltagirone has finally become the focus of an interdisciplinary study aimed at detecting, examining, and defining the occupation of the hill, spanning throughout the MBA and the early LBA. The goal is to tackle the discussion concerning the main technological features of Thapsos ceramics in the Caltagirone area, to understand the practices that were commonly accepted by such a large community within the broader geographical context of southeastern Sicily, and ultimately to define the local ceramic production and distribution.

Although this portion of the island is very well known to archeologists for the widespread presence of several social groups sharing similar cultural and technological practices during that period, almost no information has been published yet for the westernmost part of the Hyblean Platform at the time of the Thapsos culture. Against this picture, the case study of Sant’Ippolito may provide additional new data.

2 The case study of St. Ippolito Hill

The Hill of Sant’Ippolito at Caltagirone is not new to the archeological literature, being strictly connected with the first discovery of an important eponymous Copper Age pottery style (Crispino and Ippolito 2014). During excavations carried out in the 1990s, the important evidence of a Middle Bronze Age (1550–1250 BC) settlement emerged (Amoroso 1987). The site also showed a later phase of use, represented by a low stone precinct surrounding a floor lever with a hearth by which a fragment of a terracotta stove and a cooking jar were found, dated to the Early Iron Age (1050–950 BC) (Tanasi et al. 2017). In the frame of an overall reappraisal of the archeological evidence from this excavation, still unpublished at this point, a research project with a strong archeometric component has recently been carried out (Caso et al. 2017), which also led to the reappraisal of the stratigraphic deposit and the related ceramic materials.

Without entering into the details of the results of the excavation, which will be the subject of another publication currently under preparation, it is possible to see in the aerial view of St. Ippolito (Fig. 1) how the hill comprises an upper hill and a lower hill, and how its original profile and shape were drastically transformed by the activities connected to a gypsum quarry. The hypothetical location of the Copper Age site identified by Paolo Orsi when he first investigated the site (Crispino and Ippolito 2014) should be in the valley between the San Giorgio River and the St. Ippolito Hill itself. The excavation carried out in the 1990s focused on the lower terrace, where two large trenches were opened (Saggio A and Saggio B). While Saggio A provided just data about an occupation dated to the Greek Archaic period, Saggio B offered evidence of a consistent occupation mostly during the Middle Bronze Age and poorer traces of use in the Late Bronze Age and Early Iron Age, with remnants of huts. The floor level of similar huts was and still is visible both in the upper and the lower terraces (purple dots in Fig. 1). The Saggio B comprises 17 quadrants of irregular sizes ranging from 2.50 to 1.50 m, added progressively one after the other. As emphasized in Fig. 2, better preserved structures associated with diagnostic materials dated to the Early Iron Age were identified in quadrants 10/8/5/17 (Tanasi et al. 2017) and remnants of two superimposed Bronze Age structures were found in quadrant 4, which produced the most complete stratigraphic sequence (Fig. 3). The vast majority of the ceramics recovered in Saggio B were of Thapsos type and represent the three main sub-phases of this production: I, II, III (Alberti 2017) (Fig. 4).

Fig. 1
figure 1

Aerial view of St. Ippolito Hill with indication of the two main excavation areas

Full size image
Fig. 2
figure 2

Schematic plan of the main chronological phases of the structured uncovered in Saggio B (Caso et al. 2017)

Full size image
Fig. 3
figure 3

Saggio B, quadrant 4, section view North–South (elaborated by authors)

Full size image
Fig. 4
figure 4

Saggio B, assemblage of Thapsos pottery of sub-phases I, II, and III (drawings C. Veca and G. Caso), scale 1:2

Full size image

3 Materials

The traditional assessment by archeologists of the Thapsos pottery from Sant’Ippolito via direct examination led to the identification of six main fabrics on the basis of parameters such as texture, manufacture, and function (Table 1): very fine hard fabric with smoothed and burnished surface (fabric I); fine hard fabric with smoothed and painted surface (fabric II); fine hard fabric with smoothed and undecorated surface (fabric III); coarse soft fabric with a rough and rarely slipped surface (fabric IV); medium soft and undecorated fabric (fabric V); and coarse medium fabric with unsmoothed (fabric VI) which occurs only on large storage jars.

Table 1 Main fabrics of Thapsos pottery found at St. Ippolito Hill, identified via direct examination
Full size table

To test the validity of the initial fabric classification and to better understand the relationship between fabrics and typologies, a group of 23 Thapsos pottery samples representing the most common shape types and fabrics was sampled for archeometric analyses. To this group, four samples of cooking plates and adobe from the same stratigraphic context where the other samples came from were added together with two samples taken from later North Pantalica vessels to define possible technological changes at the transition between the two periods, for a total of 29 samples (Table 2; Fig. 5).

Table 2 List of samples for this study, including sample IDs, type, and class
Full size table
Fig. 5
figure 5

Cup typologies belonging to fabrics I, II, III

Full size image

4 Methods

Samples were subject to an array of optical and chemical comparative analyses to enhance the quality of the data gathered to either confirm or reject the interpretation given by archeologists concerning pottery fabrics (Table 3).

Table 3 List of the samples with indication of the analytical techniques applied (petrographic examination of thin sections, portable X-ray fluorescence spectrometry, electron microprobe analysis)
Full size table

4.1 Petrographic analysis of thin sections

Among the entire group of samples, 17 most representative specimens were selected for the petrographic study (USF29350, 29351, 29354, 29355, 29357, 29358, 29359, 29360, 29361, 29362, 29363, 29365, 29368, 29369, 29370, 29376, 29377), to either confirm or reject the interpretation provided by the direct examination of the fabrics. Thin section production was carried out in the Department of Geosciences at the University of South Florida. Thin sections were described using an Olympus B120C polarizing microscope, following the standards proposed by Whitbread (1989), Quinn (2013), and Santacreu (2014).

By doing so, a new classification for the samples can be proposed, according to specific features of the ceramics including the structure of the clay paste, grain size and distribution, and overall texture. The resulting groups are, therefore, established according to the parameters that have been used in the literature to explain technological features related to manufacture and firing techniques of pottery production throughout the broader geographical context of the eastern side of Sicily (Barone et al. 2011a, b, 2012; Raneri et al. 2015a, b; Rodriguez et al. 2015; Tanasi et al. 2013).

4.2 Chemical analysis with a portable X-ray fluorescence spectrometer

The chemical analysis of the selected samples was carried out at the Laboratory for Archaeological Science & Technology (LAST) in the Department of Anthropology at the University of South Florida using a Bruker Tracer III-SD X-ray fluorescence spectrometer.

The analysis was performed with settings of 40 kV and 10 μA, using a 12-μm Al, 1-μm Ti, and 6-μm Cu filter to minimize background and enhance measurements of peak areas (Emmitt et al. 2018; Forster et al. 2011; Newlander et al. 2015; Stovel et al. 2016), and according to the standards proposed by Hunt and Speakman (2015) and Tykot (2016). The instrument was placed in an upright position on a plastic stand with samples balanced on top. Three readings of 120 s each were taken for each sherd over the edge, inner, and outer surfaces with a beam area of 5 × 7 mm. Flat areas on the specimens’ surface were chosen to reach the instrument’s maximum potential, after following a two-cycle cleaning procedure involving washing and then scrubbing the samples’ surface with a very fine abrasive paper. Raw data for each of the trace elements were obtained using Excel software that incorporates values for 40 standards tested with the instrument. To evaluate the consistency of each trace element and to interpret the geochemical data in accordance with established analytical conventions efficiently (Emmitt et al. 2018; Forster et al. 2011; Hunt and Speakman 2015; Karacic and Osborne 2016; Newlander et al. 2015), the three measurements taken on each sherd were averaged, and then transformed using base log 10.

The resulting statistical analysis combined principal components analysis (PCA) and Euclidean distance hierarchical clustering. pXRF has been increasingly utilized in non‐destructive archeological research for a variety of projects (Karacic and Osborne 2016; McKendry 2015; Mommsen 2004; Neff 2002; Stovel et al. 2016; Tanasi et al. 2017). It was chosen for this study for several reasons: since it requires minimal sample preparation, it is relatively inexpensive, and the data it generates can potentially be used to differentiate between ceramics with different elemental compositions in a relatively short amount of time. Overall, most of the elements which are of interest for compositional analysis of archeological ceramics can be obtained accurately and precisely by pXRF (Emmitt et al. 2018; Forster et al. 2011; Hunt and Speakman 2015; Newlander et al. 2015; Tykot 2016). To potentially distinguish among different groups of ceramics, Rb, Sr, Y, Zr, and Nb were selected for the analysis.

4.3 Chemical analysis using an electron microprobe (EPMA)

To enhance the interpretation of more problematic samples, 12 specimens (USF29349, 29352, 29353, 29364, 29366, 29367, 29371, 29372, 29373, 29374, 29375, 29378) were analyzed with an EPMA JEOL 8900R Superprobe, equipped with a wavelength-dispersive spectrometer, at the Florida Center for Analytical Electron Microscopy at the Earth and Environment Department of the Florida International University of Miami, and according to standardized procedures proposed by Malainey (2011). Electron probe microanalysis (EPMA) was performed on the clayey matrix. This method of analysis can help to either confirm or exclude suspected rocks as sources of raw materials by detecting phases < 15–20 μm, including minerals such as feldspars, micas and quartz, which are relevant for the identification of the technological condition of firing and the clay’s properties (Cultrone et al. 2001; Cuomo di Caprio 2007). The possibility to combine backscattered electron (BSE) images, and secondary electron (SE) images, without destroying the samples, offers a solid basis for integrated interpretations of the ancient ceramics. The precise compositional and fabric features resulting from EPMA are overall indicators of raw materials and their provenance, and even of occurring technological constraints, such as firing temperature and atmosphere.

5 Results

5.1 Petrographic data

The analysis of the 17 thin sections was carried out following Whitbread’s classification (1989). Such an approach allowed the characterization of the entire group of samples into petro-groups, or fabrics, according to the occurrence of different features within the clay paste. Samples’ mineral and textural features, such as grains’ distribution, optical activity, angularity, porosity, and clay matrix, have been described to detect the use of different clay sources, or a specific production recipe, thus highlighting the practice in use for the pottery from the hill of Sant’Ippolito during the Middle Bronze Age. Samples 29355 (cup), 29360 (cup), and 29370 (cooking pan) were not included in the petrographic analysis due to the poor preservation of their texture after the treatment. According to the petrographic analysis, two different fabrics can be distinguished: SIP1 and SIP2. On the basis of groundmass texture and temper, SIP1 can be further divided into two sub-groups, SIP1a and SIP1b.

Fabric SIP 1a (Fig. 6, Table 4) includes samples 29354 (cup), 29361 (situla), 29376 (hot-plate), and it is characterized by a semi-fine light brown groundmass. The coarse fraction is mainly composed by poorly sorted grog, and frequent microfossils, while the fine fraction is characterized by very fine quartz particles, and rare vitrified grains.

Fig. 6
figure 6

Fabric SP1a: thin section of sample USF29354 (B4/7/6) XP (×40)

Full size image
Table 4 Petrographic description of fabric SIP1a
Full size table

Fabric SIP1b (Fig. 7, Table 5) includes samples 29350 (foot), 29363 (cooking pan), 29368 (cooking pan), 29369 (cooking pan), and it is characterized by a semi-fine light brown groundmass. The coarse fraction is mainly composed by moderately sorted grog and no microfossils, while the finer fraction is mainly quartz.

Fig. 7
figure 7

Fabric SP1b: thin section of sample USF29369 (SN2) XP (×40)

Full size image
Table 5 Petrographic description of fabric SIP1b
Full size table

Overall, the SIP1 group, including both SIP1a and SIP1b, is characterized by very porous and micritic clay containing both sedimentary and intrusive rocks. The temper is overall composed by clasts of sedimentary rocks, grog, and microfossils, which are very common in the geological unit around Mt. Etna and in eastern Sicily (Barone et al. 2012; Basilone 2018; Rodriguez et al. 2015). Group SIP1a differs on average from group SIP1b mainly because of the lack of abundant fossils among samples of the latter, most likely coming from micrite carbonates.

Fabric SIP2 (Fig. 8, Table 6) includes samples 29351, 29357, 29358, 29359, 29362, 29365, 29377, and it is characterized by a coarse and semi-fine dark and very dark brown texture, composed mainly by grog and fossiliferous groundmass.

Fig. 8
figure 8

Fabric SIP2: thin section of sample USF29359 (B4/14/4) XP (×40)

Full size image
Table 6 Petrographic description of fabric SIP2
Full size table

This petrographic group is characterized by the presence of a coarse fossiliferous clay, whose temper can be described as mainly composed of homogeneous microfossils, high amounts of grog and secondary grog, and fine silicate grains. Its overall texture is more porous than SIP1, and inclusions are randomly distributed across the thin sections, with a much higher presence of microfossils and small silica grains.

5.2 Chemical analyses

5.2.1 pXRF

The results of the chemical analysis carried out over the entire ceramic assemblage (= 29) showed the occurrence of a significant range for Sr, Zr and Rb, while the range is small for Y and Nb. However, an overall homogeneity among Rb, Sr, Y, Zr, and Nb values is still noticeable. Elements such as Sr, Zr and Rb are among the most meaningful trace elements for discriminating between groups of compositionally different samples, and it seems that variations occurring among those elements also proportionally affect the content of both Y and Nb. However, the same elements do not seem to consistently affect one another (Table 7).

Table 7 Descriptive statistics of the total (= 29) ceramic assemblage
Full size table

Some noticeable variations are present for sample 29353, which shows the highest concentration of Sr (5497 ppm), and very high concentrations of Y (28), Zr (178), and Nb (19). Sr concentrations for samples 29372 (4086) and 29375 (2019) are among the highest recorded. However, the latter represents a very interesting specimen, since it also bears the lowest amount for Rb (60), Y (16), Zr (88), and Nb (10). This sample is a fragment of the adobe found on the ground floor of the hut. Minor compositional variations affect sample 29349, which bears one among the lowest concentrations of Rb (83) and Nb (17), and very high values for Sr (1740) and Zr (203). The lowest Sr concentration (706) has been recorded for sample 29374, which is also characterized, on the other hand, by the highest Zr concentration (228) among the entire group of samples. Other meaningful variations have been recorded for Rb concentrations among samples 29364 (73), 29366 (77), and 29373 (79). Sample 29364 also bears a very low concentration of Nb (16). Rb values are low among the samples 29372 (83) and 29373 (79), which also show very low concentrations of Y (21). Table 8 includes the ppm values recorded for all the samples included in this study.

Table 8 pXRF results for trace elements, with values reported as ppm values
Full size table

What is important, however, is the interpretation of the further statistical analysis, which provides clues about elements’ load for each sample. For this reason, principal components analysis was performed using a variance–covariance relation, due to samples’ overall high homogeneity (Fig. 9).

Fig. 9
figure 9

Biplot of the two principal components (PCA) showing all the samples analyzed for this study

Full size image

The scatter plot for principal components 1 and 2, which accounts for more than 95% of the cumulative variation among variables, clearly shows one main group of 24 specimens clustered altogether, including all the samples regardless of their shape, function, and fabric detected. Three major outliers (29353, 29372, 29375) are, instead, plotted separately far from the main group. These samples are strongly characterized by an overall anomalous content of Sr, Zr, and Y in relation to all the other specimens, and to one another, and they consistently differ from the main cluster. Other minor outliers, as they are plotted closer to the main cluster than the major outliers, are samples 29349 and 29374. PCA loadings for PCA1 suggest Sr is the main trace element driving the resulting distinction between different compositional groups of samples, while Rb and Zr play the major role for PCA2. Variations occurring among Zr concentration, however, are not able to discriminate among groups by itself, and its loading can be understood only in regard to Sr concentration. The hierarchical cluster, computed according to Euclidean distance (Fig. 10), further confirms what PCA analysis previously suggested, as it displays the samples 29372 and 29353 in the top right of the graph lacking any immediate links with their compositionally neighboring counterparts, including a minor outlier (29374). The adobe fragment is, instead, displayed closer to the other minor outlier (sample 29349) in the top left, thus showing how these two groups differ compositionally.

Fig. 10
figure 10

Euclidean distance of the entire group of samples for trace elements (Rb, Sr, Y, Zr, Nb)

Full size image

5.2.2 EPMA

The analysis of 12 samples by microprobe aimed at investigating their ambiguous chemical signature detected via pXRF, in relation to the entire ceramic assemblage. Although this method concerns the petrographic assessment of the samples, it also provides important clues in regard to their chemical composition. Such an approach, therefore, allowed the further characterization of the selected samples into different groups according to differences among the peaks detected for Si, Fe, and Ca.

Samples were obtained following the same procedures previously adopted for the optical analysis, and later on treated to make them suitable for the microprobe analysis by applying a very thin layer of gold on the sections’ surface. According to the results, the samples showed, on average, very similar chemical signatures. Slight differences have been detected among samples’ main clay components (Al, Si, Fe), although this result, per se, is not really indicative for the definition of specific production features. However, the quantitative variation between these elements among the group of samples has highlighted interesting compositional patterns, which might be useful for the resulting distinction between groups of fabrics. Two main groups can be distinguished: Group A and Group B.

Group A (Fig. 11), consisting of five samples (29349, 29352, 29353, 29364, 29378), was detected according to the association between high amounts of Al, Si, and Ca and low Fe content. The digital image (× 40) shows a slightly porous clay matrix with meso- and macro-vughs, which likely suggests the use of degradable compounds added as temper. The upper foot fragment (29353) that has been considered an outlier shares such features.

Fig. 11
figure 11

Group A: backscattered electron image (×40) 100 μm and spectrum indicating peaks for Al, Si, Ca and Fe (sample USF 29378—B5/4/SIP4). The picture shows heterogenous voids that follow a specific pattern, thus suggesting the use of a higher concentration of organic matter as temper, which disappeared after the firing process and, then, left voids behind within the matrix

Full size image

Group B (Figs. 12, 13) counts five samples with low amounts of Al, Si, and Ca, always associated with higher Fe content (29367, 28371, 29372, 29373, 9374). All the samples in this group showed very similar composition, thus suggesting they were more likely to have been made with closely related raw materials. Interestingly, all the pithoi are compositionally related to two cooking pans (29367, 29371). Besides, the presence of sub-rounded meso-vughs, the matrix of these groups show, on average, the highest degree of compactness, which further confirms the importance of making these types more resistant to eventual shocks.

Fig. 12
figure 12

Group B: backscattered electron image (×40) 100 μm and spectrum indicating peaks for Al, Si, Ca and Fe (sample USF 29372—B4/7)

Full size image
Fig. 13
figure 13

Group B: backscattered electron image (×40) 100 μm and spectrum indicating peaks for Al, Si, Ca and Fe (sample USF 29373—B13/14/7)

Full size image

Besides, these two main groups minor compositional variations occur among two samples (29366, 29375), although they do not seem to share any compositional pattern with one another to the extent that they be considered as a group apart. It is important to remember that sample 29375 is adobe and it was also identified as an outlier by the pXRF analysis.

6 Discussion

The findings from the application of the three analytical techniques have provided significant new data to expand the discussion on the significance of the evidence from the site of St. Ippolito (Table 9).

Table 9 List of the samples with the indications of the results from the three analytical approaches
Full size table

With respect to the outcomes of the petrographic analysis, the Thapsos pottery samples from Area B at the hill of Sant’Ippolito show a relative homogeneity of production, characterized by the use of fossiliferous clay tempered mainly with grog, additional carbonate and siliceous rocks, probably related to the geological setting of the hill itself (Barone et al. 2012; Basilone 2018; Rodriguez et al. 2015). Intrusive rocks are far less abundant.

The most interesting aspect of the petrographic study is the recognition of specific “pottery recipes” for both tableware (SIP1a and SIP2) and cooking ware (SIP1b). The use of finer and less porous clays for SIP1b, characterized by low carbonate concentrations, absence of microfossils, and high concentration of silica particles, may be related to the choice of the potter to enhance the resistance of this class to thermal shocks occurring during the firing stage. This interpretation is further supported by the optical inactivity of the samples from this fabric across the sections, which highlights even more interesting clues in regard to the choices made by ancient potters. The light brown color, which is distinctive for this fabric, may suggest the occurrence of a setting that supposedly could reach very high temperatures (> 900 °C), causing the vitrification of finer minerals across the sections. The combination of these features also suggests that the firing more likely occurred within a kiln, rather than in a firing pit, since reaching higher temperatures commonly requires skills for controlling the fire and the atmosphere.

On the other hand, the abundance of fossils and different types of carbonate rock-based tempers and grog suggests a different approach to the firing strategy for this class. Here, the combination between the dark color in the matrix and the presence of microfossils is linked to the reduced atmosphere setting. This triggers the occurrence of choices made by potters to eliminate the risk of fractures, since the formation of calcium oxide creates a significant stress in the pottery (between 650 and 898 °C) and, thus, it may generate cracks in the ceramic body (Barone et al. 2012; Borgna and Levi 2015; Cuomo di Caprio 2007; Quinn 2013; Veca 2015). Although such features cannot be clearly confirmed at this stage due to the lack of any assessment about the crystallographic properties of the minerals detected among these samples, the results may provide important clues in regard to technology. Therefore, this class of pottery may have been fired differently, and at different temperatures, than the former one, perhaps according to constrains derived from the nature of the clays in use and the skills kept by potters. The lack of organic matter among all the fabrics can be also addressed as further confirmation for the idea that both types were fired at temperatures above 700 °C, either in a kiln or in a firing pit. This information is crucial because it highlights the importance of petrography as a necessary tool that cannot be excluded from the study and the interpretation of ceramic artifacts, especially to understand the evolution of ancient technology (Levi and Muntoni 2014; Mommsen 2004; Quinn 2013; Santacreu 2014; Veca 2014b; Whitbread 1989).

The compositional results of the Thapsos ceramic assemblage also revealed an overall homogeneity. The pXRF provided precious clues for the understanding of the compositional proximity among the specimens. Some of the micro-variations detected among the elements might be the result of the overall high presence of grog within each sample, which might potentially affect the overall analysis by pXRF due to the width of its beam (Hunt and Speakman 2015; Tykot 2016). Nonetheless, the samples were made according to very close manufacturing strategies, as the detected chemical homogeneity demonstrates.

The PCA suggests the occurrence of three outliers, specifically an upper foot cup (29353), a storage jar (29372), and the adobe fragment (29375). The cup represents an interesting case, considering its typological relation with the tableware types we encountered during this study. It is noticeable that its values (highest Sr and Zr concentrations, and lowest Y content) are slightly off from the overall picture of the dataset. Unfortunately, no clay source had been investigated at the time of the recovery, and the artifacts cannot be properly correlated to any of the nearby clay sources at this stage, which we aim to accomplish in the future. Further investigations might provide the proper data to understand additional clues in regard to procurement strategies, given also the occurrence of an adobe fragment (29375) among the detected outliers. The other outlier consisted of a storage jar (29372). The likelihood for the pithos to be considered a common specimen is very low. According to the nature of the archeological evidence, the location of the site, and the context of the ceramic assemblage, it is more likely that the pithos was manufactured differently, perhaps from a different workshop or with a different recipe than those that overall produced the repertoire here investigated. The sample 29374, clustered in a different reference group than the major one, further confirms this hypothesis. The complex debate concerning the movement of big storage jars across the landscape might provide precious insights, whether they were the material product of a larger koine of potters or a privileged specialization for specific socio-political agents within the broader prehistoric political network (Alberti 2013; Doonan 2001; Raneri et al. 2015a; Tanasi 2010b; Veca 2014a, b, 2015). The computation of the Euclidean distance served as a further confirmation that the petrographic fabrics and the chemical composition could match. Among all the samples, groups according to the classification resulting from the petrographic assessment are also chemically distinguishable. The overall occurrence of more or less fossiliferous clays, along with the use of siliceous- and carbonate rock-based tempers, may be related to specific choices made among potters and to the diversification of their recipes among workshops (Buxeda i Garrigos et al. 2001; Quinn and Day 2007; Santacreu 2014; Stovel et al. 2016; Whitbread 1986). However, at this stage, it is too early to address how extensive was the network that made the potters choose specific recipes, given how little we know about clay sources, workshops, and settlement networks throughout this area during the Middle Bronze Age. PXRF was very useful in combination with the petrographic methods, which helped at defining groups of artifacts that can be distinguished within the broader and chemically homogeneous ceramic assemblage.

Furthermore, PCA clearly showed how the samples 29349 and 29374 form a compositionally distinguished group from the main group of Thapsos ceramics. 29349 represents an early North Pantalica specimen, proving that the well-known cultural discontinuity occurring between the two pottery traditions was not just with respect to style and shape repertoire, but it perhaps affected clay procurement and manufacturing strategies (Tanasi 2008). Also remarkable is the proximity between this sample and one of the three pithoi (29374) analyzed, which further confirms the assumption that some specific types here represented could have been produced somewhere else in relation to the entire assemblage. The resulting data are precious clues that will provide additional pieces to the broader understanding of the Middle Bronze Age of Sicily.

These assumptions become clearer with the inclusion of the results coming from the compositional study via EPMA of a selected group of samples. Although the samples on average confirm a high compositional homogeneity, they were classified into two distinctive groups of samples according to differences in the clayey matrixes that are either high in Ca (A) or Fe (B). The results highlight the possibility that some recipes required a mixture of clays according to potters practices and functionality. Yet, understanding the provenance and the relation among each production type and clay source cannot be fully discerned. Within Group A, higher Ca concentrations among samples of the tableware group, (29349, 29352, 29353), one of the coarse group (29364), and a coarse cooking plate (29378), might suggest the use of highly fossiliferous clays, and the use of carbonate rocks added as temper. Such an interpretation strengthens the previous petrographic assessment. Samples within Group B, which are functionally related to cooking pans (29367, 29371) and storage jars (29372, 29373, 29374), show very high concentrations of Fe that are always associated with very low levels of Ca, thus triggering the use of clays lower in fossiliferous content, along with the use of less carbonate rocks as temper, than Group A. The distinctive association between Ca and Fe concentrations within the ceramic groups is here interpreted as a marker for the identification of two distinct clay pastes that are characterized either by clays higher in Fe or in Ca content. However, whether this difference is the result of specific choices made by potters or imposed by the availability of local raw sources still represents a deficiency within the broader understanding of local ceramic production.

Finally, the presence of the two North Pantalica samples 29349 and 29364 in the same Group, already grouped together by pXRF in a cluster apart, further reinforces the hypothesis of a discontinuity in the manufacturing processes between the Thapsos (MBA) and North Pantalica (LBA) types.

7 Conclusions

The results of the archeometric analyses testify to a petrographic and chemical homogeneity for Thapsos pottery production at the site of St. Ippolito. Comparing the petrographic data with those previously presented on the materials from Monte San Paolillo di Catania and Grotte di Marineo di Licodia Eubea, it is now possible to draw some preliminary conclusions on the manufacturing strategies and traditions of the Etnean territory spanning from the Southeastern slope of the Etna volcano and the eastern side of the Hyblaean plateau. The discriminating petrographic characteristic between the three productions is the occurrence of volcanic inclusions (rocks, glass, plagioclase, pyroxenes), largely attested at Monte San Paolillo di Catania and Grotte di Marineo di Licodia Eubea and basically absent at St. Ippolito.

On the other hand, the only fabric without volcanic materials observed at Grotte di Marineo (fabric A: Tanasi 2015, p. 56) shares same petrographic features of SIP 1A and SIP 2, such as limestone inclusions and fossiliferous groundmass. The use of grog and microfossils and limestone inclusions also seems to be another feature in common between the production of Grotte di Marineo and St. Ippolito.

In this perspective, the use of volcanic tempers becomes the principal indicator of the Etnean production of Thapsos pottery. The different evidence offered by St. Ippolito shows how such a site should be considered as part of another technological koine, characterized by different manufacturing strategies. Considering that the territory of Caltagirone is closer to the geographical district of Siracusa than to that of Catania, it is not inappropriate to suggest that Thapsos potters at St. Ippolito shared the same technological knowledge of those from Siracusa. However, the lack of archeometric data for the Thapsos pottery from the Megarian and Siracusan district leaves the problem open still, but a gap that we plan to bridge by extending the archeometric survey to Thapsos ceramic assemblages from that area.