Introduction

Traditionally, ancient glass is studied using typological and art-historical approaches using decorative styles. These approaches can help us to investigate the evolution of the glass vessel forms across space and time, to some extent, and the provenance of the glass vessels. More recently, chemical analysis of glass has been proven beneficial to investigate the technology of glassmaking including raw material use, glass provenance, glass distribution patterns and trade. A combination of archaeological, typological and chemical studies of glass can help us to investigate the technological, social and economic aspects of ancient glass. This study attempts to apply these approaches to the study of glass from middle Byzantine (11th–12th centuries AD) Turkey.

A limited number of investigations of middle Byzantine glass has been carried out using combined archaeological, typological and chemical approaches. Previously, many scholarly discussions and debates on middle Byzantine glass centred on luxury glasses, such as the painted and enamelled glasses from the Treasury of San Marco and vessels in the British Museum (Henderson and Mundell Mango 1995; Ristovska 2009; Tait 1995). More recently, researchers have begun to use a combined archaeological and scientific approach to study Byzantine glass such as that from Hosios Loukas, Dalphni, Amorium, Pergamon, Nafaru, Zeyrek Camii and Kariye Camii, and glass from the eleventh century AD shipwreck Serçe Limani, in order to understand the technological and economic aspects of middle Byzantine glass (Arletti et al. 2010; Brill 2005, 2009; Brill and Stapleton 2012; Bugoi et al. 2013; Schibille 2011; Wypyski 2005). However, such researches are relatively few and they sometimes only focus on the technological aspects of Byzantine glass.

Therefore, we considered it appropriate to study window, vessels and raw furnace glass from Zeyrek Camii in Istanbul (historically known as Constantinople) using a combination of different approaches. Although based on a relatively small number of samples, this paper aims to address a range of research questions using the results of electron microprobe analysis (EPMA-WDS):

  1. (1)

    What is the chemical composition of middle Byzantine glass from Zeyrek Camii and what is its provenance?

  2. (2)

    What raw materials were used to produce these glasses?

  3. (3)

    Are there any relationship between chemical compositions and colours?

  4. (4)

    Can we use a combination of historical, archaeological and chemical evidence to look at the glass trading network between the Byzantine Empire and the Islamic East in this period?

The site of Zeyrek Camii and the glass finds

Zeyrek Camii (formerly known as the Pantokrator Church) was first built in AD 1118–1136 by the Byzantine Emperor John II Komnenos and the Empress Eirene (Ousterhout 2001). The complex consisted of three large interconnected churches, an almshouse, a hospital, a pharmacy and it was also an imperial mausoleum (Canav-Özgümüş and Kanyak 2015; Ousterhout 2001; Ousterhout et al. 2000).

The church had undergone a series of restorations in 1953, 1966–1967 and more recently in 1997–1998, 2001–2005 and 2009. Excavation was also carried out in and around the structure under the supervision of the Directorate of Istanbul Archaeology Museums in 2009 and 2013 (Canav-Özgümüş and Kanyak 2015). Their progress was summarised in Megaw (1963), Ahunbay and Ahunbay (2001), Ousterhout et al. (2000) and Canav-Özgümüş and Kanyak (2015).

During the 2009 restoration (carried out by the Istanbul Municipality and the restorers were Baris Han and his team) and excavation campaign, glass finds such as glass cakes and chunks, glass vessels including small bowls, small bottles and jars were found during excavation work around the structure (including the western part of the building). The glass vessels were used to prepare drugs or to perform medical investigations. Glass oil lamps and pieces of goblets were an indication of daily life in the church and were made of glass (Canav-Özgümüş and Kanyak 2015). In the 2013 excavation under the supervision of Istanbul Archaeological Museums, a substructure was discovered and cast, painted window glass (formed in a mould and inserted into plaster frames with small openings) and crown glass were found (Canav-Özgümüş and Kanyak 2015). With the permission of Istanbul Archaeological Museum, the glass finds were recorded and some of the glass materials from the 2009 restoration and excavation campaign and the 2013 excavation were analysed in this study.

Materials and methods

Samples

Fifteen glass samples from Zeyrek Camii dating to the 11th–12th centuries AD and were selected for the electron microprobe analysis (EPMA-WDS). A range of glass artefacts were selected (Table 1). Four painted window fragments of a range of colours (ZC-09, ZC-10, ZC-11, ZC-12) were sampled. A range of colours is included (Table 1).

Table 1 Description of the glass samples from Zeyrek Camii
Full size table

Sample preparation

Samples of approximately 0.5 cm were removed from each of the glass artefacts so that unweathered glass samples could be used for analysis. Each sample was then mounted in epoxy resin. Each epoxy resin block was ground and polished using standard sample preparation procedures down to a 0.02-μm final polishing solution. This removed any weathering and exposed produced a flat surface. Each block was then carbon coated to prevent surface charging and distortion of the electron beam during analysis (Henderson 1988; Henderson et al. 2004).

Analytical procedure

The polished samples were analysed by EPMA-WDS with a Cameca SX100 electron microprobe in the School of Environment, Earth and Ecosystem Sciences in the Open University. The system is equipped with five wavelength-dispersive X-ray spectrometers with LTAP, PET, TAP, LPET and LLIF crystals and a single energy-dispersive X-ray spectrometer and both secondary and backscattered detectors. The analytical set-up for quantitative compositional analyses was as follows: 20 kV accelerating voltage, a 20 nA beam current and a 25 μm defocused beam. The counting times were between 10 and 40 s on the peak and 15 s on the background either side of the peak. A defocused beam was used to minimise the effect of the migration of alkalis (e.g. Na2O) from the samples (Henderson 1988).

A combination of standard reference materials, including minerals (baryte, bustamite, cobaltite, crocoite, haematite, stibnite, willemite), pure metal (Cu, Ag, V, Ni) and synthetic standards (Cr2O3, KCl, LiF), was used to calibrate the EPMA-WDS. Twenty-six elements were sought and the results expressed as weight percentage oxides. Five areas of interest (at × 1000) were analysed in each sample and the mean was calculated. So as to check the accuracy and precision of the EPMA-WDS system and to monitor any drift in the instrument (Meek et al. 2012, 790), repeat analyses of a secondary standard, Corning B, were included at the start and finish of each analytical run. The gross analytical errors resulting from the repeat analysis of the glass standards are presented in Table 2.

Table 2 The results of the analysis of the Corning Glass Standard B (expressed in wt.%). The known composition of the Corning Glass Standard B is from Brill (1999b)
Full size table

Results

The electron microprobe results are given in Appendix A. The result shows that all of the glasses from Zeyrek Camii are soda-lime-silica glasses. Using the elements K2O and MgO, which derive from the alkali source, it is evident that seven samples were made with plant ashes and sand with their high contents of K2O (2.20–2.64%) and MgO (2.82–3.87%). On the other hand, the other eight samples belong to a glass type referred to here as mixed natron-plant ash glass. They have lower levels of K2O (1.28–2.34%) and MgO (1.44–2.87%) than the plant ash glasses. The glass sample ZC-06 is an outlier and has a higher concentration of K2O (4.41%) than the other two glass groups.

The two glass types can be identified by their varying concentrations of major and minor element. The mean and standard deviations of the glass groups are presented in Table 3.

Table 3 The means and standard deviations of the glass groups from Zeyrek Camii
Full size table

Group 1

Seven glasses (ZC-08, 09, 10, 11, 12, 13 and 15) belong to group 1. It is characterised by a relatively low concentration of Al2O3 (0.90–1.42%), Fe2O3 (0.31 and 0.74%) and TiO2 (0.07–0.09%). This suggests that a relatively pure source of sand was used. These glasses have relatively high Na2O (11.38–14.18%) and CaO (8.04 and 10.15%) concentration. The levels of MgO (2.82–3.87%) and K2O (2.20–2.64%) in these glasses are also relatively high, suggesting that the source of alkali was ashed plants.

It is also notable that this glass group is characterised by a high concentration of MnO (0.96–2.21%). It is likely that MnO was added deliberately to the glass but for different purposes: MnO was likely used to decolourise glass samples ZC-08, 11, 12, 13 and 15 and to colour sample ZC-09 a lilac and ZC-10 purple.

Group 2

Group 2 consists of eight samples (ZC-01, 02, 03, 04, 05, 06, 07, 14). Its composition is significantly different from group 1. It is characterised by a relatively higher level of Al2O3 (1.81–3.13%), Fe2O3 (0.98–2.08%), TiO2 (0.11–0.29%) than in group 1. This suggests that a less purer sand source was used than the sand source used to make group 1. This group can also contain relatively low MgO (1.44–2.87%), K2O (1.28–2.34%) and CaO (6.91–8.73%).

Many of these glasses are highly coloured. The yellowish-green and green colours of ZC-01, 02, 04 and 05 are likely to be coloured by ferric oxide (0.69–1.34%) introduced in the sand. The colourless glass sample (ZC-14) contains 0.99% MnO used to decolourise them. The blue glasses (ZC-03 and 07) are likely coloured by cobalt-rich compounds with CoO (0.11–0.15%), CuO (0.28–0.32%), ZnO (0.11–0.30%) and PbO (0.18%). Zinc and lead impurities are frequently found in cobalt blue glass in the Middle East (Henderson 2013).

Outlier–ZC-06

Glass sample ZC-06, a lamp, is compositionally different from groups 1 and 2. It is characterised by a relatively high concentrations of Al2O3 (2.73%), Fe2O3 (0.94%) and TiO2 (0.14%), showing that an impure sand was used. High levels of Fe2O3 and TiO2 may have derived from impurities such as chromite and titinite found in the sand source (Henderson 2000). The glass contains high levels of MgO (2.52%) and K2O (4.41%), which suggests that the alkali source was ashed plants. What is significant about this glass is its exceptionally high K2O level, which is not found in other glasses from groups 1 and 2. This could potentially suggest that a different species of plant was used to produce this glass or that the plant used was growing on a different geological environment from the plants used to make the other glasses. Such high levels of K2O in plant ash glasses are also characteristic of a central Asian origin (Brill 1999a, b; Rehren et al. 2010). It is also noted that ZC-06 is a colourless glass and has a relatively high concentration of MnO (0.72%), which was used to decolourise it.

Discussion

The use of raw materials and glass provenance

A comparison of Zeyrek Camii glass groups with contemporary Zeyrek Camii and Kariye Camii glasses analysed by Brill (2005) and Henderson and Mundell Mango (1995), and with 10th–14th centuries AD glasses from the eastern Mediterranean shows that they belong to different glass production groups and probably production zones. Figures 1, 2 and 3 show that group 1 corresponds to Brill (2005)’s low boron plant ash glasses. Both groups are characterised by a relatively low Al2O3 and Fe2O3 (Fig. 1). They have relatively high CaO levels of between 8.04 and 9.57% (Fig. 2) and a relatively high concentration of Na2O and K2O (Fig. 3).

Fig. 1
figure 1

A bi-plot of Fe2O3 versus Al2O3 for Zeyrek Camii groups 1 and 2 from 11th–12th centuries AD Zeyrek Camii with relevant 10th–14th centuries AD glasses made in the eastern Mediterranean. It is displayed according to composition groups and sites. Data for Tyre glasses: Freestone (2002); Banias glasses: Freestone et al. (2000); Damascus glasses: Henderson et al. (2016); Kariye Camii glasses: Brill (2005), Henderson and Mundell Mango (1995); Zeyrek Camii glasses: Brill (2005)

Full size image
Fig. 2
figure 2

A bi-plot of CaO versus MgO for Zeyrek Camii groups 1 and 2 from 11th–12th centuries AD Zeyrek Camii with relevant 10th–14th centuries AD glasses made in the eastern Mediterranean. It is displayed according to composition groups and sites. Data for Freestone (2002); Banias glasses: Freestone et al. (2000); Damascus glasses: Henderson et al. (2016); Kariye Camii glasses: Brill (2005), Henderson and Mundell Mango (1995); Zeyrek Camii glasses: Brill (2005)

Full size image
Fig. 3
figure 3

A bi-plot of Na2O versus K2O for Zeyrek Camii groups 1 and 2 from 11th–12th centuries AD Zeyrek Camii with relevant 10th–14th centuries AD glasses made in the eastern Mediterranean. It is displayed according to composition groups and sites. Data for Tyre glasses: Freestone (2002); Banias glasses: Freestone et al. (2000); Damascus glasses: Henderson et al. (2016); Kariye Camii glasses: Brill (2005), Henderson and Mundell Mango (1995); Zeyrek Camii glasses: Brill (2005)

Full size image

Recent studies of plant ashes show that the level of CaO in the plants is determined by the geology of the environment where the plants were grown and that the alkali levels in the plants are determined by the physiognomy of different plant species and genera (Barkoudah and Henderson 2006; Henderson 2013). The similarities in the alkali and lime contents between group 1 and Brill’s low boron glasses could suggest that they were probably produced in the same glass workshop using the same plant species and/or genera from the same geochemical environment but trace element analysis is needed to be more confident of this interpretation.

A comparison of group 1 and Brill’s 2005 low boron glass with 10th–14th centuries AD glasses from the Middle East (Figs. 1, 2 and 3) shows that they have similar Al2O3, Fe2O3, CaO, MgO, Na2O and K2O levels to some of the 12th–14th centuries AD plant ash glasses from Damascus, 10th–11th centuries AD glasses from Tyre and 11th–13th centuries AD Banias (Freestone et al. 2000; Henderson et al. 2016).

This may suggest that a common source of silica and plant ashes (similar plant species from similar geological environments) was used to produce these glasses during the 11th–14th centuries AD. It is possible that glasses from group 1 and Brill (2005)’s low boron plant ash glasses were made in Israel, Lebanon or Syria. According to the Jewish traveller Benjamin of Tudela, there were about ten glassmakers in Tyre or Antioch in the twelfth century AD where plant ash was used to manufacture glass (Boas 1999, 151).

Group 2 is rather different from group 1. One way of interpreting the relative oxide levels in Figs. 4, 5, 6, 7 and 8 is that group 2 is a mixed natron-plant ash glass group rather than the Syro-Palestinian plant ash glass of group 1. The main characteristics of this type of glass are low K2O and MgO levels (Fig. 6). The typical value of K2O in mixed natron-plant ash glasses is usually below 2.5% (Andresscu-Treadgold et al. 2006; Bugoi et al. 2013, 2016). In addition, mixed-natron plant ash glasses have high concentrations of Al2O3 (1.50% or more) and Fe2O3 (up to 4.54%) (Fig. 4), which can also be associated with mixed silica sources were used (Andresscu-Treadgold et al. 2006; Bugoi et al. 2013, 2016; Lauwers et al. 2010).

Fig. 4
figure 4

A bi-plot of Fe2O3 versus Al2O3 for Zeyrek Camii groups 1 and 2 from 11th–12th centuries AD Zeyrek Camii with relevant 10th–13th centuries AD glasses found in the Middle East and Byzantine Empire. Data for Tyre glasses: Freestone (2002); Banias glasses: Freestone et al. (2000); Zeyrek Camii glasses: Brill (2005); Nufaru glasses: Bugoi et al. (2013); Torcello Basilica glasses: Andresscu-Treadgold et al. (2006)

Full size image
Fig. 5
figure 5

A bi-plot of CaO versus MgO for Zeyrek Camii groups 1 and 2 from 11th–12th centuries AD Zeyrek Camii with relevant 10th–13th centuries AD glasses found in the Middle East and Byzantine Empire and 7th–13th centuries AD glasses made in Central Asia. Data for Tyre glasses: Freestone (2002); Banias glasses: Freestone et al. (2000); Zeyrek Camii glasses: Brill (2005); Nufaru glasses: Bugoi et al. (2013); Torcello Basilica glasses: Andresscu-Treadgold et al. (2006)

Full size image
Fig. 6
figure 6

A bi-plot of K2O versus MgO for Zeyrek Camii groups 1 and 2 from 11th–12th centuries AD Zeyrek Camii with relevant 10th–13th centuries AD glasses found in the Middle East and Byzantine Empire and 7th–13th centuries AD glasses made in Central Asia. Data for Tyre glasses: Freestone (2002); Banias glasses: Freestone et al. (2000); Zeyrek Camii glasses: Brill (2005); Nufaru glasses: Bugoi et al. (2013); Torcello Basilica glasses: Andresscu-Treadgold et al. (2006)

Full size image
Fig. 7
figure 7

A bi-plot of P2O5 versus MgO for Zeyrek Camii groups 1 and 2 from 11th–12th centuries AD Zeyrek Camii with relevant 10th–13th centuries AD glasses found in the Middle East and Byzantine Empire. Data for Tyre glasses: Freestone (2002); Banias glasses: Freestone et al. (2000); Zeyrek Camii glasses: Brill (2005); Torcello Basilica glasses: Andresscu-Treadgold et al. (2006)

Full size image
Fig. 8
figure 8

A bi-plot of P2O5 versus K2O for Zeyrek Camii groups 1 and 2 from 11th–12th centuries AD Zeyrek Camii with relevant 10th–13th centuries AD glasses found in the Middle East and Byzantine Empire. Data for Tyre glasses: Freestone (2002); Banias glasses: Freestone et al. (2000); Zeyrek Camii glasses: Brill (2005); Torcello Basilica glasses: Andresscu-Treadgold et al. (2006)

Full size image

All glasses in group 2 have K2O between 1.28 and 2.34%. Six out of ten glass vessels analysed by Brill (2005) also have K2O contents below 2.50% and display similar characteristics as the suggested mixed-natron plant ash glasses (Fig. 6). High concentrations of Al2O3 (1.81–3.13%) and Fe2O3 (0.94–2.08%) are also found in group 2 and Brill’s 2005 mixed-natron plant ash glass vessels (Al2O3: 1.02–2.72%; Fe2O3: 0.60–2.25%) (Fig. 4).

This study also shows that natron and plant ash glasses were mixed in different ratios to produce positive correlated mixing lines in Figs. 5, 6, 7 and 8. In these instances, at the left-hand end of the line is the natron glass from Torcello Basilica, Banias and Tyre and at the right-hand end is the plant ash glasses from Zeyrek Camii, Banias and Tyre. Group 2 falls in the middle between the natron and plant ash glass and are clearly separated from the other two types of glass (Figs. 5, 6, 7 and 8). If plant ash glasses with different K2O, MgO, CaO and P2O5 levels had been used to mix, many of these mixed-natron plant ash glasses would not plot on the positively correlated mixing lines (Andresscu-Treadgold et al. 2006; Henderson 2013).

Mixed-natron plant ash glass is the result of mixing old natron glass, which ceased to be produced after the ninth century AD, and plant ash glass which was increasingly produced and used after the late 8th and 9th centuries AD (Henderson et al. 2004). They are found in the west such as Sagalassos in Turkey, Torcello’s basilica in Italy, Nufaru and Isaccea in Romania, Daphni and Hosios Loukas in Greece during the 10th–13th centuries AD (Andresscu-Treadgold et al. 2006; Arletti et al. 2010; Bugoi et al. 2013, 2016; Lauwers et al. 2010).

There is evidence that a significant amount of mixed-natron plant ash glass was produced in the Islamic east and was shipped to Byzantine territories. About three metric tons of glass cullet and raw glass were found on the eleventh century AD Serçe Limani shipwreck (Bass 1984; Brill 2009; van Doorninck 1990, 2009). Chemical analysis of the glasses shows that a number of the Serçe Limani glasses were made with mixed-natron plant ash glasses which compositionally overlap with some of the mixed-natron plant ash glasses from Zeyrek Camii (Arletti et al. 2010; Brill 2005, 2009).

The source of some of the glass found on the Serçe Limani shipwreck is probably from glass workshops along the Fatimid Syrian coast. It has been suggested that the plant ash portion of the glasses may have come from Tyre as it was a famous glass production centre during this period (Bass 2009). The presence of raw glass that belongs to the mixed-natron plant ash glass type in Zeyrek Camii raised the possibility that a glass workshop was set up near the church during this period (Canav-Özgümüş and Kanyak 2015): perhaps, natron and plant ash glasses of group 2 were mixed there. These authors have also noted that some of the raw glasses were the colours of gold-foiled mosaics. It has been suggested that the workshop must have been set up in order to supply glass mosaics and window glass used in the church (Canav-Özgümüş and Kanyak 2015). If the Serçe Limani was sailing towards Constantinople, it would seem that not all of it was mixed there.

Figure 4 shows that ZC-06 has high concentrations of Al2O3 and Fe2O3 that is similar to group 2. The high concentrations of MgO (2.52%) and K2O (4.41%) show that a plant ash was used to produce this piece of glass and that it does not belong to either group 1 or 2 (Fig. 6). Such a high level of potash is usually found in Central Asian glasses, which contains potash levels as high as 7% (Brill 1999b; Rehren et al. 2010). Therefore, based on its high concentrations of Al2O3, Fe2O3 and K2O, ZC-06 is likely to have been produced in Central Asia.

Compositional groups, colours and typology

Typological and art-historical analyses have been useful in identifying the craftsmanship of the Zeyrek Camii window glass and its manufacturing techniques. It is also beneficial to examine any relationships between the chemical compositions of and attempt to explain any relationships in terms of the social, economic and political contexts of the glass. Glasses of group 1 are both cast and painted and crown window glasses. Only one sample (ZC-15) is a vessel rim. Two of the window glasses are lilac (ZC-09) and purple (ZC-10, four window glasses (ZC-08, 11, 12 and 13) and the glass vessel are colourless. The low B2O3 window glasses from Zeyrek Camii published by Brill (2005) are dark blue, purple, aqua and amber (Brill 2005).

The majority of the glasses from group 2 are in the forms of raw glass and vessels. The glass vessels are a cup (ZC-04) and a beaker (ZC-14). Window glasses are a crown glass (ZC-05) and cast window glass (ZC-07). Moreover, the colours of group 2 glasses are rather different from those in group 1: only one sample is colourless (ZC-14), the rest are translucent yellow, blue-green, cobalt blue and green. Some of the vessel glasses analysed by Brill (2005) were made with mixed-natron plant ash glass. However, the glasses he analysed are different: mainly colourless and aqua. The majority of other mixed-natron plant ash glasses found in the Byzantine Empire that have been analysed are glass vessels, tesserae and bracelets (Andresscu-Treadgold et al. 2006; Arletti et al. 2010; Brill and Stapleton 2012; Bugoi et al. 2013, 2016). ZC-06 is a colourless glass lamp that was produced with a Central Asian glass recipe.

Although a relatively small number of samples are involved, this study shows that there is a clear relationship between chemical composition, artefacts type and colours. Mixed natron-plant ash glass was mainly used to produce everyday objects (glass vessels and bracelets) and tesserae; (2) plant ash glass (group 1 and Brill’s low boron plant ash glass) and high boron glass were mainly used to produce the painted window glass in Zeyrek Camii (this study and by Brill (2005)).

The trading of glass between the Byzantine Empire and the Islamic East in the 11th–12th centuries AD

Textual sources such as the Jewish manuscripts from the Cairo Geniza suggest that a strong trading relation existed between the Byzantine Empire, the Fatimid Caliphate and later the Crusader States (after AD 1099) in the Middle East in the 11th–12th centuries AD. For instance, trading networks between Fatimid Egypt and Byzantine ports located in southern Turkey, namely Tarsus, Seleucia and Antalya, were well established (Goitein 1996). Products such as pepper were demanded by the Byzantines, and goods such as timber (which was scarce in the Middle East) cheese, corals, silks and pharmaceutical products) from the Byzantine Empire were exported to the Fatimid Caliphate (Goitein 1996; Jacoby 2000; Laiou 2002).

This study also shows a well-established and thriving glass trading network between the Byzantine Empire and the Middle East in the 11th–12th centuries AD. The similarity between group 1 and plant ash glass from Tyre in Lebanon, Banias in Israel and Damascus in Syria would suggest the Byzantines possibly acquired plant ash glass from these regions to produce window glass for a major religious building commissioned by the Komnenoi family. Other studies also show that plant ash glass was imported to other regions in the Byzantine Empire for architectural decorations and personal adornment throughout the 10th–13th centuries AD, such as the mosaic glass from the monasteries of Hosios Loukas and Daphni, and bracelets and plant ash glass chunks from Hisn al-Tinat and Sagalassos (Arletti et al. 2010; Lauwers et al. 2010; Swan et al. 2017).

The finding of mixed-natron plant ash glass chunks and vessels from group 2 shows that plant ash glass was imported from the Middle East (possibly Fatimid Syria) and mixed with old natron glass in Byzantine glass workshops (e.g., in Constantinople and Sagalassos), possibly to keep the cost down (Canav-Özgümüş and Kanyak 2015; Keller 2004; Lauwers et al. 2010). But it is equally possible that the mixed-natron plant ash glass could have come from the Fatimid Middle East. The Serçe Limani shipwreck contained a number of glass cullts with this composition that could have come from Fatimid Syria (Bass 2009; Brill 2009). Nonetheless, the archaeological and scientific evidence suggest a matured glass trading network existed between the Byzantine Empire and the Fatimid Middle East in this period.

Conclusion

By the tenth century AD, plant ash glasses were exported from the Syro-Palestinian coast to regions controlled by the Byzantine Empire. Using EPMA-WDS analysis to examine glass from Zeyrek Camii in Istanbul, we have identified two groups of glass from 11th–12th centuries AD Constantinople: group 1 is a type of plant ash glass characterised by high Na2O and CaO, very similar to glasses from Damascus, Tyre and Banias, suggesting that this glass type might have originated from the Syro-Palestinian region; on the other hand, group 2 is a type of mixed-natron plant ash glass and is characterised by lower MgO, K2O and CaO levels. This type of glass is widespread across the Byzantine Empire and Byzantine influenced territories. Similar type of glass is found in Greece, Romania and Italy.

Using a combination of historical, archaeological and chemical evidence, although based on a relatively small number of samples, this study suggests that there was a well-established and thriving trading network of plant ash glass between the Byzantine Empire and the Islamic throughout the 11th–12th centuries AD and continued uninterrupted into the thirteenth century AD.

In future, trace element analysis (e.g. LA-ICPMS) will be used to investigate the technological and economic aspects of Byzantine glass. A recent study by Henderson et al. (2016) has shown that trace levels of elements such as chromium (Cr), caesium (Cs) and lanthanum (La) are very useful for defining provenance, for discriminating between glass production zones and for providing evidence for glass trade across the Middle East. Therefore, this technique should therefore cast further light on the economic and technological links between glass production in the Byzantine Empire, the Crusader States and the Fatimid Caliphate.