• 1 This article is part of the PhD of Francesca Colangeli, who conducted the majority of the research (...)

1The increase in archaeometric and typological studies on glass in the last decades has highlighted the complexities of the manufacture and trade of ancient glass, and how little we still know about the transition from the centralised production and circulation system of the Roman age to the fragmented system of the Medieval period. In this context, the need to start collecting data on the regions of southern Italy has become increasingly evident. The research carried out in collaboration with the European projects Sicily in Transition and GlassRoutes are a step in this direction.1

• 2 Schibille 2022.
• 3 Colangeli – Sacco in this volume.

2The regime changes that affected Sicily between the 6^th and 13^th centuries make the island a privileged place to study the impact that these changes may have had on society, economy and networks. Moreover, understanding the effects of the Islamic conquest of Sicily on the artisanal manufacture of glass is particularly important because, as mentioned in the introduction to the book, between the 9^th and 10^th centuries several regions of the Islamic world adopted new ways of producing glass independently, especially through the replacement of natron with soda plant ash based fluxes.2 In contrast, natron glass continues to be recycled in many regions of the Italian peninsula.3

• 4 Excavation data have been recently published in open access. See Molinari – Meo 2021.

4The Via Romano site is located in the northern area of Mazara del Vallo city centre, on the Sicilian western coast (fig. 1). An emergency excavation was conducted in 1997 by the AR.CO cooperative that brought to light several structures and pits dated from the late 7^th century to modern times.4 Medieval glass comes from late 7^th-century frequentation layers and from 10^th- to 11^th-century latrines and wells.

• 5 Schulze-Dörrlamm 2009, tipo E17, p. 52
• 6 Muresu 2018, p. 120, fig.72; Ripoll – Velázquez 2013, p. 142 with references.

5The Castronovo area is the focus of the Sicily in Transition ERC project. The town is in the Sicilian hinterland between Palermo and Agrigento (fig. 1). From 2014 to 2019, surveys and excavations have been carried out in different parts of the territory. The glass finds analysed in this article come from two excavation areas: the Byzantine fortress of Mount Kassar and the Casale San Pietro, located in the Platani valley near the Palermo-Agrigento highway. Several excavations were carried out in different parts of the fortress. In this context, we will consider just the top site (intervention 6) excavated between 2014 and 2016. The operations revealed a rectangular structure built against the defensive wall of the fortress. In this building, a layer dated to the first half of the 8^th century was discovered under the collapse of the walls and roof caused by the abandonment of the site. The chronology of the layer was determined by the typological study of pottery and metal finds and by the compositional analyses of the glass finds. Interestingly, the metal finds include a Hyppo Regius buckle (fig. 2), a type that spread in the central Mediterranean mainly during the 6^th and the 7^th centuries,5 but that has been found also in some Sardinian and Spanish 8^th-century contexts.6 In this specific case the redecoration marks suggest a long use of the object.

• 7 See Sacco – Meo – Capelli in this volume.

6Regarding Casale San Pietro, the extensive excavation area brought to light a stratification that ranges from the 2^nd to the 13^th centuries. In this context we will treat just the finds from the Islamic period (10^th -11^th century) and few specimens dated to the 8^th century. The excavation brought to light tons of 10^th– to 11^th-century pottery (tableware, cooking pots, amphorae). Imports come mainly from Palermo. As capital of Islamic Sicily, the city was an important manufacture and commercial hub.7

• 8 Foy 2003, type C p.74 and p. 77, fig. 64-72.
• 9 For a summary of the geographical spread and dating of this type see Golofast 2009, p. 318-319.

717 individual glass objects have been identified from late Byzantine layers (10 from Via Romano, 4 from Kassar fortress and 3 from Casale San Pietro). In spite of the limited number, these finds have a high informative potential, due to their provenance from well-dated archaeological contexts. From a typological point of view, almost all of the Byzantine glass fragments are identifiable as goblets with an applied solid stem that is tapered in the middle8 (fig. 3a and 3l). Generally, this type of stem (indicating a two-stage production) is dated between the 7^th and 8^th centuries, although it seems to appear as early as the 6^th century.9 Disk feet usually have a tubular edge (fig. 3b, c, d, e, h). In addition to goblets, one beaker was also identified (fig. 3g). The base has some purple streaks that could be intentional, for a decoration purpose, or a possible production mistake in the colouring or decolouring practice. All the specimens have different shades of blue-green, ranging from light blue-green to darker aqua shades.

• 10 For an in-depth analysis of the Via Romano Islamic specimens see Colangeli 2021, p. 473-474.

89^th-century glass finds are virtually absent, meaning there is no direct material evidence from the time of the Islamic conquest. Moreover, the investigations carried out in the fortress of Mount Kassar did not return any evidence of a frequentation of the area in the Kalbid period. In contrast, 10^th- to 11^th-century contexts present larger quantities of glass: 36 objects from Via Romano and 31 from Casale San Pietro.10 The glass artefacts are mostly related to vessels and they are mainly cosmetic objects or tableware (fig. 4). While the two categories seem to be more balanced in the Casale San Pietro site, in Via Romano there seems to be a clear predominance of vessels for cosmetic use (fig. 4). They are essentially small free-blown flasks that show signs of a fast and often inaccurate manufacture. When preserved, the rim is folded inwards and it is often irregular, while the bottom and the profile are often decentred and crooked. In most cases these objects are in the natural colours of glass.

• 11 Specimens have been identified in the 1995 under the Regional Archaeological Museum “Antonio Salin (...)
• 12 Foy 2020

9Most of the balsamaries found in the Via Romano present a truncated cone body and a tubular neck (fig. 5m-n), but there are also specimens with a square section and two recessed sides (fig.5o), as well as sporadic samples with a globular body. A possible microvase was also reported (fig.5p). Some balsamaries with truncated-cone bodies had already been found in Palermo as well as at Villa del Casale.11 However, given the contemporaneity and extent and richness of typological variety, the main reference material for all the identified types are the finds from the Tunisian site of Ṣabra al-Manṣūriyya.12

• 13 Scanlon – Pinder-Wilson 2001, p. 43, fig. 18a-b
• 14 See respectively Puche Acién 2000, p. 157, fig. 1, no. 4 and Gomez 2012, Q282, p. 441 and p. 444, (...)

10The comparison between Via Romano and Casale San Pietro shows a lower amount of cosmetic and medicinal vessels (fig. 4) and also a lower typological variability for the latter. Specifically, only 9 balsamaries have been identified in Casale San Pietro and the majority of them belong to the type with a square section and two recessed sides (fig. 5c-d). These balsamaries are associated with a microvase and an everted rim (fig. 5f-e). This last object has a narrow neck with a central bulge (fig. 5e). The fragmentary nature of the piece impedes the identification of the body profile. However, the wall is slightly introflected suggesting a tubular body. The object is free-blown in a translucent glass with a yellow/brown colour. The typology has no published comparisons in Sicily. From an extra regional point of view, the specimen resembles two balsamaries found in Fustat dated to the 9^th century.13 The type has also been found in al-Andalus in Alicante and at the castle of Silves, where it appears in brown glass with marbled decoration in 11^th century layers.14 Although these comparisons are not precise and minor differences can be noted, the object would seem to resemble the same prototype.

• 15 Colangeli 2021.

11Concerning tableware, the 6 objects found at Via Romano are essentially beakers and bottles. The beakers have a truncated cone profile and a flat or slightly umbonate base and they are free-blown or mould-blown. These objects can either be not decorated or with mould-blown or abraded decorations (fig. 5q). This is the case of the MZ247 specimen that, from a macroscopic point of view, differs also in terms of greater thickness, less surface alteration and colour (colourless/pale yellow). Among the Via Romano bottles, the most representative type is characterised by a pyriform body displaying a ribbed decoration obtained by mould-blowing (fig. 5r). For the moment, no comparisons has yet been found for this specimen. However, the archaeological context suggests a mid-11^th-century date.15 7 tableware specimens have been identified from Casale San Pietro and they consist of beakers with truncated cone profile and flat bases and goblets with short stems (fig. 5a-b).

• 16 Giannetti – Giuliani – Turchiano 2017, p. 206-207; Uboldi 1995, type I.4.
• 17 Lamps with a truncated cone or globular profile, a tubular rim and handles set on the rim have bee (...)
• 18 For example, see Carboni 2001, p. 166-167; Rontomé Notario, 2000, p. 109, fig. 2, n. 2; Foy 1999, (...)

12Finally, only 3 fragments can be attributed to lighting devices, and they have been identified exclusively in Casale San Pietro (fig. 4). They include a lamp with a tubular rim and a horizontal handle applied directly to the rim (fig. 5g). This typology is known in the Mediterranean, and particularly in Southern Italy, from the 5^th century16 but is still attested in the 10^th -11^th century in the Islamic Regions of the Western Mediterranean.17 Moreover, a piece of a possible Islamic/Mosque lamp has also been identified (fig. 5h). Finally, a tube probably represents a wick holder18 (fig. 5i).

• 19 On Sicilian glass weights see Balog 1975; D’Angelo 2005; Weiss 2011; D’Ottone Rambach 2017, p. 175 (...)

13Glass was not used just for vessels, but also for objects of personal adornment (beads and bracelets) as well as for architectural use and jetons (category ‘other’ in fig. 4). The majority of these finds comes from the Islamic layers at Casale San Pietro (fig. 4). In this respect, it is worth mentioning that the site revealed two glass jetons (fig. 5k-l) which based on the stratigraphy can be dated to the second half of the 10^th century.19

• 20 Method is given in Schibille – Colangeli 2021.
• 21 Henderson 2013, p. 91-92 and 98; Sayre – Smith 1961.

14In order to investigate the composition and thus obtain information on the production technique, provenance and, in some cases, even the chronology, most of the finds just described have been analysed by LA-ICP-MS.20 A total number of 32 and 15 samples dated between the end of the 7^th and the 11^th century have been analysed from Mazara del Vallo and Castronovo. Based on their level of sodium, potassium and magnesium, all analysed specimens are soda-lime-silica glass that can be divided into natron and soda plant-ash glass. Both natron and soda plant-ash groups show high soda levels, but they differ in terms of magnesium and potassium content (fig. 6). Compared with plant ashes, natron is a purer source of alkali, therefore natron glass is characterized by low levels of both these two elements, whereas plant ashes have high levels of MgO and K₂O.21

• 22 Schibille et al. 2019; Phelps et al. 2016; Freestone – Gorin-Rosen – Hughes 2000.
• 23 Foy et al. 2003.
• 24 Freestone 2020; Brems et al. 2018; Phelps et al. 2016.
• 25 Phelps et al. 2016, p. 60.
• 26 Ivi p. 65-66.
• 27 For an in-depth analysis of Mazara finds see Schibille – Colangeli 2021.
• 28 Schibille et al. 2019.

15In order to detect the origin of natron glass finds, the samples have been compared with 7^th- to 9^th-century compositional groups produced in Egypt and in the Levant.22 The ratios of Al₂O₃/SiO₂ and TiO₂/Al₂O₃ separate the groups according to the geographical origin of the silica source, because it is generally assumed that Egyptian silica sources have higher levels of titanium and zirconium compared to Levantine sands.23 The plot clearly shows the Levantine origin of the natron-type glass samples from Mazara, as well as of two Casale San Pietro samples (fig.7a). Specifically, they belong to the glass associated with the primary production installations at Apollonia (6^th-7^th century) and Bet Eli’ezer (8^th century).24 From a chemical point of view, these two groups are very similar. Both are characterized by mature high silica sands (Bet Eli’ezer shows highest silica levels) with relatively high alumina contents low in heavy elements such as titanium, iron, and zirconium.25 One of the distinguishing features of Bet Eli’ezer glass is a significant decrease in the soda contents that has been linked to problems in the natron supply.26 This is reflected in a comparison of the Na₂O/SiO₂ and CaO/Al₂O₃ ratios, which illustrates the changes in the composition of the silica source and the decline of soda content over time (fig. 7b). While Mazara samples clearly belong to the Apollonia glass group,27 CSP288 and CSP180 show a higher silica (SiO₂ ≈ 73-74%), and the lower soda content (Na₂O ≈ 13%), which corresponds to the typical characteristics of Bet Eli’ezer glass. They also share the lower calcium content (CaO ≈ 7%). Just one specimen coming from the Kassar fortress has an Egyptian signature, more specifically, it belongs to Egypt 1b dated between 720 and 78028 (fig. 7a).

• 29 Freestone 2006; Lilyquist – Brill 1993; Sayre – Smith 1961.
• 30 Schibille – Colangeli 2021.
• 31 Henderson 2013 p. 23 and 48-49; Mirti et al. 2009.
• 32 For a more detailed analysis of the composition of the Via Romano and Casale San Pietro medieval g (...)
• 33 Schibille – Colangeli 2021.

16Magnesium and potassium oxide contents of the plant-ash glass mostly range between 1.5% and 3% (fig. 4), indicating the use of halophytic plant ash as fluxing agent.29 Some Mazara specimens have higher potassium to magnesium levels that may have been caused by contamination by fuel ash.30 For this reason and considering the variability of plant ash,31 the definition of the groups has mainly been based on the characteristics of the silica source. Compared to other Mediterranean groups, the 10^th- to 11^th-century samples from Mazara and Casale San Pietro clearly show differences in terms of the silica source related elements such as titanium and zirconium (fig. 8). The zirconium content in the majority of the samples range between 100 and 150 ppm which is higher than in Levantine glass, while the Ti to Zr ratios differ from both Egyptian and Iberian glass groups. Moreover, titanium ranges between 600 and 1000 ppm which is higher than in the Samarra finds but at the same time significantly lower than in the Spanish samples.32 Except for the Levantine engraved beaker (fig. 5q – MZ247), and the Egyptian small flask (fig. 5e – CSP182) most of the glass falls into this new group, which can be identified with great probability as Sicilian production.33

• 34 Foy et al. 2003; Gratuze 2000.
• 35 See for instance Rome and Comacchio, as well as recent research of the Iberian contexts of Gózquez (...)
• 36 Ceramic studies conducted in various areas of the region have revealed substantial continuity (alb (...)

17The typological and compositional study of the two archaeological contexts makes it clear that not only the coastal towns of Sicily were probably still supplied with new glass as late as the end of the 7^th century, but possibly also the inland centres until the 8^th century. The identification of goblets made of Levantine glass from Bet Eli’ ezer in Casale San Pietro and a sample of Egypt 1b glass in the Byzantine fortress of Kassar suggests a circulation of these goods within the island. Although studies conducted in different areas of the Mediterranean demonstrate the persistence of a Levantine (Apollonia) fresh glass trade even in the late 7^th century, it seems to be more common in Tunisia and eastern territories.34 In contrast, the varying frequency and increasing recycling indicates that this type of glass does not arrive in large quantities in north-western Mediterranean cities.35 As for Sicily, the close connection with North Africa and the arrival of goods from the eastern Mediterranean suggests that it would be easier to obtain this material for the island as well. Nevertheless, these data are consistent with the pattern of Sicily ‘long late antiquity’ already evidenced by ceramic studies.36

• 37 Meo 2021a.

18Regarding the 8^th century, however, it should be considered that the Castronovo area was closely linked to the Byzantine Empire, thanks to a fortress and a village that may have had a fiscal function. Consequently, the interests of the state may have facilitated the arrival of imports in these two centres, while other sites may have been excluded from redistribution circuits. It is important to note that Egypt and the Levant were part of the territories of the Umayyad Caliphate at the end of the 7^th to the middle of the 8^th century and that, even if the study of the Via Romano pottery had already shown the presence of imports from both Byzantine and Umayyad Caliphate territories (Egyptian amphorae),37 the Levantine coast remained invisible.

• 38 On the different uses of the term ‘Islamisation’ see Peacock 2017.

19The typological and compositional change that emerges from the 10^th century onwards clearly shows the impact of the Islamisation38 of the island. The extensive use of glass for cosmetics, as well as the morphological repertoire similar to other Islamic contexts (first and foremost Sabra al-Manṣūriyya) and the possible regional production of soda plant ash glass, illustrate Sicily’s inclusion in the Islamic technical environment and networks.

• 39 Sacco forthcoming.
• 40 Meo 2021b.
• 41 A hypothesis has been put forward on the structures found at the Steri in Palermo, but the publish (...)
• 42 Goldberg 2012.

20Future research will seek to better understand Palermo’s possible role in glass production. As ceramic studies show,39 the city was not only the capital of Islamic Sicily, but also the main centre of production and trade until at least the middle of the 11^th century. This is also evident in the contexts of Mazara and Casale San Pietro, where amphorae, cooking pots and glazed pottery from Palermo are strongly represented.40 This could indicate that glass production, at least until the mid-11^th century, was also predominantly based in Palermo, from where glass was exported either as raw glass or as finished objects. For the time being, there is no archaeological evidence of glass production on the island during the Islamic period.41 Nonetheless the connection between Palermo and other centres may also have favoured the circulation of extra-regional imports. This is not so relevant for Mazara, which as the second most important port after Palermo certainly had a direct connection with Ifrīqiya and Egypt,42 but it was probably crucial for inland centres like Castronovo.