Scholarly article on topic 'Changes in glass consumption in Pergamon (Turkey) from Hellenistic to late Byzantine and Islamic times'

Changes in glass consumption in Pergamon (Turkey) from Hellenistic to late Byzantine and Islamic times Academic research paper on "History and archaeology"

CC BY
0
0
Share paper
Academic journal
Journal of Archaeological Science
OECD Field of science
Keywords
{"Mineral natron glass" / "High-boron glass" / "Glass working" / Pergamon}

Abstract of research paper on History and archaeology, author of scientific article — Th. Rehren, P. Connolly, N. Schibille, H. Schwarzer

Abstract We present compositional data for nearly 100 glass samples from Pergamon, western Turkey, spanning 1500 years from the Hellenistic to Late Byzantine and Islamic periods. The data shows the use of already-known Roman glass groups during the first half of the time frame, for imported vessels as well as locally worked glass. No compositional change is seen related to the introduction of glass blowing for either of the glass groups in use during this time. During the first half of the 1st millennium AD, two previously little-known boron- and alumina-rich compositional groups emerge. These glass groups, thought to be regionally produced, dominate glass compositions in Pergamon during the mid-to late Byzantine and Islamic periods, indicating a major shift in glass supply and a fragmentation of the economy into more regional units. Plant-ash glass, from the 9th century AD replacing mineral natron glass in the Levant, plays only a minor role in Byzantine and Islamic Pergamon.

Academic research paper on topic "Changes in glass consumption in Pergamon (Turkey) from Hellenistic to late Byzantine and Islamic times"

Accepted Manuscript

Changes in glass consumption in Pergamon (Turkey) from Hellenistic to late Byzantine and Islamic times

Th. Rehren, P. Connolly, N. Schibille, H. Schwarzer

PII: S0305-4403(15)00004-7

DOI: 10.1016/j.jas.2014.12.025

Reference: YJASC 4306

To appear in: Journal of Archaeological Science

Received Date: 23 August 2014 Revised Date: 26 December 2014 Accepted Date: 27 December 2014

Please cite this article as: Rehren, T., Connolly, P., Schibille, N., Schwarzer, H., Changes in glass consumption in Pergamon (Turkey) from Hellenistic to late Byzantine and Islamic times, Journal of Archaeological Science (2015), doi: 10.1016/j.jas.2014.12.025.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

400 --

♦ Eastern Med ■ HBAI AHLiBAI

—300

200 --

100 --

▲ ▲

▲ ▲

800 1200 B (ppm)

Changes in glass consumption in Pergamon (Turkey) from Hellenistic to late Byzantine and Islamic times

Th. Rehrena\ P. Connolly3, N. Schibilleb and H. Schwarzerc

a UCL Qatar, Georgetown Building, PO Box 25256, Doha, Qatar. * Corresponding author; Phone: 00974 7777 0275, email: th. rehren@ucl .ac.uk

b Research Laboratory for Archaeology and the History of Art, University of Oxford, Oxford, UK. Current address: Art History, University of Sussex, Falmer/Brighton, UK

c Institut für Klassische Archäologie und Christliche Archäologie / Archäologisches Museum Westfälische Wilhelms-Universität Münster, Domplatz 20-22, D-48143 Münster, Germany

Abstract

We present compositional data for nearly 100 glass samples from Pergamon, western Turkey, spanning 1500 years from the Hellenistic to Late Byzantine and Islamic periods. The data shows the use of already-known Roman glass groups during the first half of the time frame, for imported vessels as well as locally worked glass. No compositional change is seen related to the introduction of glass blowing for either of the glass groups in use during this time. During the first half of the 1st millennium AD, two previously little-known boron- and alumina-rich compositional groups emerge. These glass groups, thought to be regionally produced, dominate glass compositions in Pergamon during the mid- to late Byzantine and Islamic periods, indicating a major shift in glass supply and a fragmentation of the economy into more regional units. Plant-ash glass, from the 9th century AD replacing mineral natron glass in the Levant, plays only a minor role in Byzantine and Islamic Pergamon.

Key words: mineral natron glass; high-boron glass; glass working; Pergamon.

1. Introduction

For nearly two millennia, from 1000 BC to the late 1st millennium AD, glass making in the

Eastern Mediterranean was based on mineral natron from the Wadi Natrun in Egypt. Little is known about Hellenistic glassmaking, with production evidence so far only known from Rhodes (Rehren et al. 2005, and references therein). At least from the Roman period it seems

to have been concentrated in a relatively small area stretching from lower Egypt (Nenna 2000; Nenna et al. 2005) to the northern Levant (e.g. in Bet Eli'ezer, Freestone et al. 2002a and Beirut, Kouwatli et al. 2008), where it was fused with local sand (Fig. 1). The glass composition directly reflects impurities in the sand used by each producer, resulting in chemically distinct glass groups (Freestone 2005; 2006; Degryse et al. 2009). From these primary production centres the finished glass was then exported to the consumption centres for working into artefacts. Significantly, the various compositional groups have limited life spans, as documented from archaeological finds, suggesting that individual large-scale producers operated only for a few centuries before giving way to others.

Fig. 1: Map of the Eastern Mediterranean with some primary glass production sites and the position of Pergamon and Sardis. The region of major borate deposits is shaded in grey, east of Pergamon. Drawing: Robert Dylka.

Much of the literature concerning relatively early glass compositions (pre-4th century AD) is based on glasses from the northern and western provinces (e.g. Foster and Jackson 2005, 2009, 2010; Paynter 2006), and Italy (e.g. Mirti et al. 1993; Silvestri et al. 2005, 2008; Silvestri 2008; Gallo et al. 2013). Here, dominating compositional groups include Roman blue/green glass (Rb/g), antimony-decoloured glass and manganese-decoloured glass, and HIMT glass. In contrast, much of the later analysed glass has been found in the Eastern Mediterranean, with dominating groups including Levantine I and II, HIMT, and more regionally restricted, Egypt I and II (e.g. Freestone et al. 2002b, 2008; Foy et al. 2003; Freestone 2005, 2006; Nenna et al. 2005; Kato et al. 2009, 2010; Abd-Allah 2010; Rehren et al. 2010; Rosenow and Rehren 2014).

In contrast, and despite its economic and political importance and its closeness to the primary production centres, relatively little is known about the composition of glass used in Asia Minor. The analyses published up to now are predominantly from southwest Turkey; Brill (1999) lists some 35 analyses of glasses from Sardis and seven from Aphrodisias; Uhlir (2004; Uhlir et al. 2010) reports glass compositions for 106 glass samples from Hanghaus 1 in Ephesos, ranging from the 2nd century BC to the 6th and 7th century AD; and Degryse et al. (2006) report 11 analyses of mid-1st millennium glass from Sagalassos in southern Turkey. This situation is corroborated by a similarly inadequate situation concerning typological studies of ancient and mostly Byzantine glass from Asia Minor, and stands in contrast to the cultural and economic importance and prosperity of the region. Only recently research concentrates on these aspects (e.g. Lafli [ed.] 2009). A comprehensive typological study of glass from Pergamon (Schwarzer 2009; Schwarzer and Rehren 2015; Schwarzer in preparation) revealed a complex picture of imported and locally produced luxury glasses as

well as every-day mass-produced vessels, and changing preferences for the use of glass as a medium to produce functional or decorative items of a wide spectrum. It also provided an opportunity to investigate the change in composition of the glass used in this important city, spanning more than 1500 years from the Classical era to the Islamic period.

1. 1 Research aims

Long-term trends in the production, consumption and trade of glass in a particular site or region have so far been largely ignored by analytical studies. Fischer and McCray (1999) traced glass compositions at Sepphoris in modern-day Israel over more than a millennium, identifying a marked change in glass composition around the BC/AD turn which they link to the introduction of glass blowing and an associated adjustment of the glass recipe. A further major change occurred during the 8th to 9th century AD when glassmaking in the Levant reverted to plant-ash based recipes (e.g. Kato et al. 2009, 2010), possibly due to an interruption in the production of mineral natron (Whitehouse 2002; Shortland et al. 2006).

The assemblage from Pergamon is of particular significance not only due to the city's importance, but also because it encompasses both these major developments which may have had an influence on the nature of the glass worked and consumed in Pergamon. The earliest samples pre-date the invention of glass blowing, while the latest samples post-date the introduction of plant-ash based glass making in the Levant and Egypt. We want to see on a qualitative level how these events may have affected glass use in Pergamon, and what the Pergamenian assemblage tells us about the wider validity of the observations made in the earlier studies. Other major political changes, such as the schism of the Roman Empire, are not thought to have influenced glassmaking and glass use, while different levels of prosperity enjoyed by the city's inhabitants clearly influenced the quality and quantity of glass consumption (Schwarzer 2009).

1.2 Pergamon

Pergamon, one of the most important cities in antiquity, is situated near the western coast of modern Turkey (Radt 2011). The earliest settlement on the acropolis hill goes back at least to the Late Bronze Age, and prospered during the Archaic and Classical periods. The city obtained supra-regional significance with the Hellenistic dynasty of the Attalids who made it the capital of their kingdom. In the 2nd century BC during the reign of king Eumenes II this realm comprised a major part of Asia Minor. In 133 BC the kingdom was bequeathed by the last ruler Attalos III to the Roman people, and became part of the new province of Asia.

Pergamon remained a powerful metropolis and prospered in the Roman Imperial period, especially during the 2nd century AD. With the division of the Roman Empire in the late 4th century AD the city became part of the Byzantine realm. Since then Pergamon became less important; however, the seat of a bishopric was established here and several churches were erected. In 716 AD the city was sacked by the Umayyads who enslaved the inhabitants. This dramatic event led to an interruption of the settlement up to the 10th century AD. Widespread building activities took place again in the middle/late Byzantine period (12th/13th century AD), mostly culminating in a spacious fortification. In the early 14th century AD the city was conquered by the Seljuks and then absorbed into the Ottoman Empire. Parts of the archaeological site of Pergamon, especially the lower city, are now covered by the modern city of Bergama, home to more than 60,000 people.

2. Materials and Methods

The long-term excavations in Pergamon conducted by the German Archaeological Institute have yielded many thousands of glass fragments, from almost all periods of the city's history. During cataloguing these finds, 100 small samples were taken from a cross section of the material found in the so-called Stadtgrabung on the southern slope of the acropolis hill and with the permission of the Turkish authorities exported for chemical analysis; of these, 96 were artificial glass, one obsidian, one quartz, one a faience bead, and one fused ceramic. Sampling intended to cover all visually and typologically defined main types of glass (Schwarzer 2009; Schwarzer and Rehren 2015; Schwarzer in preparation), as well as some extraordinary pieces, covering the entire chronological sequence present. The material chosen includes glass vessels, worked in different techniques (core-formed, mould-formed, free-blown and mould-blown) and different colours, window panes, jewellery and unformed chunks. As a result, the samples represent the range and diversity of glass used at Pergamon from the mid-4th century BC up to the beginning of the Islamic period in the early 14th century AD. More than forty percent of the samples date to the period up to the 2nd century AD; about a third date to the Late Roman and Early Byzantine periods at Pergamon, roughly speaking from the 3rd to 7th centuries AD, while the remaining circa twenty percent date to the 12th to 14th centuries AD. Only two samples are dated to the 8th/9th century AD when mineral natron glass production is thought to have come to an end (Whitehouse 2002; Shortland et al. 2006), and none to the 10th and 11th centuries. It has to be stressed that the numbers of samples analysed are not representative of the relative proportions of different glass types excavated at Pergamon. Due to the difficult stratigraphic situation in the excavation areas of Pergamon resulting from the continuous settlement the presumed dates of the samples were established not only in terms of their context and the associated material but

also through typological comparison with finds from other sites, sometimes leading to rather broad date ranges (Table SOM 1).

For analysis, small (around 3 mm long) fragments from all samples were mounted in transparent resin blocks, where possible as cross sections, and ground and polished to expose uncorroded glass for electron probe micro analysis (EPMA). Two thirds of the samples were analysed at the Wolfson Archaeological Science Laboratories at the UCL Institute of Archaeology, while the remaining samples were analysed in the Research Laboratory for Archaeology and the History of Art, University of Oxford (Schibille 2011). Both instruments were calibrated using elemental or simple stoichiometric compounds, and the calibration tested by analysing Corning glasses A and B alongside the unknown samples (Table 1 for the UCL EPMA). The analyses are in close agreement within a few percent relative of the published values (Brill 1999; Vicenzi et al. 2002); the Oxford data are as reported in Schibille (2011).

About three quarter of the assemblage was further analysed for their trace element content by Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICPMS), partly in collaboration with Dr Bernard Gratuze at the IRAMAT laboratory in Orleans (UCL samples, labelled Lo/Or in Tables 2 and 3), and partly by Dr Laure Dussubieux at the Field Museum in Chicago (Oxford samples, labelled Ox/Ch in Tables 2 and 3; Schibille 2011). A comparison of trace element concentrations for Corning A (Brill 2012) and measured values from the laboratory in Orleans is given in Table 1. While no direct comparison of the performance of the two LA-ICPMS laboratories was done for these samples, there is no systematic difference visible for data from similar glasses analysed in the different labs; the data is assumed to be fully compatible.

We first used a visual assessment of the concentration of diagnostic minor oxides (Al2O3, CaO, TiO2, MnO and Sb2O5) to allocate the analysed samples to specific major glass compositional groups. The initial allocation was then checked by comparing the remaining minor oxides in the newly-analysed samples with values typically found in the published glass groups to obtain a subjective best fit. With very few exceptions, these allocations are unambiguous. However, it is important to remember that none of these groups have formally defined compositional ranges, and different authors not only use different names for similar, even identical groups, but there are also subtle differences in composition within groups. We also stress that even the diagnostic oxides are not always suitable for a strict definition of compositional 'space' to which a sample must adhere in order to be recognized as a member of that group. This is probably best illustrated by the case of the antimony-decoloured group including three samples with no antimony oxide detectable by electron microprobe analysis, and a more yellowish tint (Per 019, 055 and 090; Table 2). However, since all other diagnostic oxides match with this group, but not the others, we assigned them to this group

177 regardless. They were probably produced from the same sand as the antimony-decoloured

178 glass, but for some reason had no antimony added to them.

179 For this paper, we refer predominantly to published data from Jackson and co-workers

180 (Jackson 2005; Foster and Jackson 2009, 2010), Paynter (2006) and Gallo et al. (2013) for

181 Roman blue/green and colourless glass; Silvestri and co-workers (Silvestri et al. 2008) for

182 antimony- and manganese-decoloured glass; and Mirti et al. (1993), Freestone (2005, 2006;

183 Freestone et al. 2002a, b), Foy and co-workers (Foy et al. 2003), Foster and Jackson (2009,

184 2010) and Rehren et al. (2010) for Levantine and HIMT glass.

186 3. Results

187 The results of our analyses are reported in Tables 2 (oxides) and 3 (trace elements); a

188 catalogue of samples is provided in the supplementary online material. The archaeological

189 context indicated prosperity during the Hellenistic and Roman era, followed by a less active

190 period in the mid-1st millennium AD, a hiatus between AD 716 and the 10th century, and

191 resurgence in the early second millennium. The presentation of the data is following these

192 broad periods.

194 3.1 Early glasses: the Roman tradition

195 More than half of the samples (52) can be linked to glass compositions known as antimony-

196 decoloured, Roman blue/green, and manganese-decoloured, including all samples from the

197 2nd century BC to the 2nd century AD. Of these groups, the ten antimony-decoloured glasses

198 show the most consistent compositional pattern (Table 2), in line with data from the Iulia

199 Felix ship wreck (Silvestri et al. 2008) for such glass, and similar to Romano-British

200 antimony-decoloured glass (Foster and Jackson 2010), even though three of them (Per 019,

201 055 and 090) have no antimony above the detection limit of the EPMA. Four of the ten glass

202 sherds stem from the 1st century AD, including both Roman mould-formed and blown

203 vessels, while a few other examples belong probably to the later Roman Imperial period.

204 Among the antimony-decoloured objects of the early Roman Imperial period are fragments of

205 a colourless mould-formed bowl (Per 004), a mould-blown lotus-beaker (Per 019, Fig. 2a)

206 and a colourless free-blown beaker with facet-cut decoration (Per 058). The beaded stem of a

207 goblet going back to the 4th to 6th centuries AD belongs to the latest pieces (Per 063). The

208 earliest piece (Per 055, a Hellenistic mould-formed grooved bowl) dates to the mid-2nd to

209 early 1st century BC and has good compositional similarity with this group, but has slightly

higher lime and potash, somewhat lower soda, and no antimony; its relationship to the antimony-decoloured glass is therefore somewhat tentative.

Fig. 2a-h

The largest compositional group, totalling 23 samples, matches the Roman blue/green and Rb/g manganese-decoloured group. The typical Roman pale blue to blue-green ('aqua') glass has been described inter alia by Silvestri (2008) and Gallo et al. (2013), and the manganese-decoloured glass from the same ship wreck by Silvestri et al. (2008), where it forms a very tight compositional group. In contrast, the manganese-decoloured glass among the Pergamon samples is much more variable in its manganese content and other minor oxides. Significantly, there appears to be a seamless transition into the Roman blue/green glass, which is differentiated from the manganese-decoloured glass primarily by a lower, or no, manganese content. The manganese-decoloured glass reported in Silvestri et al. (2008) has consistently more than 1 wt% manganese oxide; among the Pergamon samples, the manganese content varies from as high as 1.8 wt% MnO, decreasing almost continuously down to the detection limit (assumed for our analyses by EPMA as 0.01 wt%). We interpret this as a sign of recycling and mixing of normal Rb/g glass with manganese-decoloured glass. Apart from the manganese content there is no significant compositional difference within this group.

These glasses also comprise both mould-formed and free-blown vessels. The mould-formed vessels include ribbed bowls from the late Hellenistic and early Imperial Roman periods (Per 045 [Fig. 2b], Per 002, 007, 077) and a network glass (Per 035) presumably imported from Italy around the turn of the ages (von Saldern 2004, 181f). The time range of the blown glass goes from the 1st to the 4th century AD. The fragment of an early Roman inscribed beaker of probably Syrian or Cypriote provenance (von Saldern 2004, 252f) is of special interest (Per 070). It formerly displayed the dictum „ЛАВЕ THN NIKHN - Gain the victory".

Another 13 closely related samples is coloured blue by cobalt and copper oxide (see Table 3). The minor oxide concentrations of nine of these are indistinguishable from the uncoloured Rb/g and manganese-decoloured glasses. However, their levels of transition metal oxides differ substantially. All of them have significantly increased levels of iron oxide, on average more than 0.9 wt% compared to the 0.3 wt% on average found in the Rb/g and manganese-decoloured glasses, as well as consistently around half of one percent manganese oxide. They include Hellenistic core-formed (alabastra Per 024, 086 [Fig. 2c]) and mould-formed vessels (mosaic glass bowl Per 014) as well as mould-formed vessels (mosaic glass bowl Per 023),

ribbed cups (so-called Zarte Rippenschalen) (Per 012, 050) and free-blown beakers (Per 068, 073) of the early Roman Imperial period.

The combination of copper and cobalt is reminiscent of the Egyptian Late Bronze Age cobalt-blue glass coloured using a preparation derived from cobaltiferous alums (Kaszmarzcyk 1986; Rehren 2001; Tite and Shortland 2003; Smirniou and Rehren 2013), even though it does not have the high alumina, nickel and zinc contents typical of those earlier cobalt-blue glasses. A compositionally very similar glass sample, also of an early date (50 BC to AD 130) was recently published from Bubastis in northern Egypt (Rosenow and Rehren 2014: Mn 06, a mould-cast ribbed bowl), suggesting a wider use of this colourant across the Eastern Mediterranean.

The remaining four cobalt-blue samples (Per 033, 064, 066 and 080) have more than one percent of antimony oxide, and the first three of these have also similarly elevated levels of lead oxide. Three also have higher copper (Per 064, 066 and 080), overall suggesting a different colourant source for these. Compared to the other cobalt-blue glasses, these have higher lime and lower soda concentrations, making them more similar to Levantine I glasses; their alumina and barium levels, however, are still more in line with the Rb/g glass set. They also are later than the other group (Tables 2, 3), further underlining their difference from the earlier eight cobalt-coloured glasses.

Seven samples contain antimony oxide at between 0.1 and 0.6 wt% as well as between 0.2 and 1 wt% manganese oxide, labelled Sb-Mn decol in Table 3 and 4 (Per 010, 020, 061, 067, 069, 083 and 089). Their minor oxide content falls between the antimony-decoloured and the Rb/g manganese decoloured glass groups, suggesting that they represent glass obtained from mixing cullet during recycling of decoloured glass. In the majority the samples are from Roman vessels, for example a mould-blown bottle with a base moulding in shape of a rosette (Per 010, Fig. 2d). An exception is a mould-formed grooved bowl from the 2nd century BC (Per 083).

All three glass groups discussed so far, the antimony-decoloured, the Roman blue/green (with or without manganese) and the mixed glass, have the same general chronological setting in the last few centuries BC and up to about the fourth century AD, and were used for the same range of glass objects. It appears that they co-existed side-by-side, rather than one following the other. The presence of several pieces that were most likely imported as finished objects (such as the network glass Per 035 from Italy, the inscribed bowl Per 070 from Syria or Cyprus, and most likely also the delicate ribbed bowls Per 012 and 050), but do not stand out compositionally, is particularly noteworthy.

3.2 Mid-1st millennium AD: Levantine and HIMT glass

The glass compositions which during this period dominate elsewhere in the Eastern Mediterranean world are only represented here by eight samples. Three samples were identified as Levantine I (Per 003, 029 and 100), based on their higher alumina and lime content compared to the previous samples. A single sample each was identified as Levantine II (Per 065), based on its even higher alumina and slightly lower lime content compared to the Levantine I samples, and as HIMT glass (Per 099), based on the high titania and iron oxide content. Both are late antique/early Byzantine window panes. The other analysed window panes are Levantine I glass (Per 100), Co-blue glass (Per 064) and HLiBAl glass (see below, Per 087). Three samples (Per 056 [Fig. 2e], an early Roman Imperial mould-blown ribbed bowl, Per 085, a goblet, and Per 094, a polycandelon-lamp, the latter both from the early Byzantine period) have slightly elevated levels of titania and iron oxide, in line with the HIMT 2 group defined by Foster and Jackson (2009) for Britain, or the weak HIMT from northern Egypt (Rosenow and Rehren 2014). Six of these eight Levantine and HIMT samples date to the mid-1st millennium AD, the remaining two date as early as the 1st and 2nd century AD, based on their archaeological context and on typological comparisons (Per 003 Lev I and Per 056 weak HIMT).

3.3 High boron high alumina glass of the mid-1st to early 2nd millennium

Almost all archaeological glass analysed so far from Europe and the Near East has less than 250 ppm B and less than 3-4 wt% Al2O3. The occurrence at Pergamon of glass with much higher boron and alumina is therefore noteworthy, as first reported by Schibille (2011). We define high-boron glasses as having in excess of 500 ppm B. Nearly one third of all analysed samples (28 out of 97) from Pergamon, and the large majority of glasses dating later than the 5th century AD, belong to this new glass type. There are two sub-types of this group, one with about 1000 ppm B and 9 wt% Al2O3 on average, and the other with nearly 1500 ppm B, around 300 ppm Li and around 5 wt% Al2O3 (Fig. 3, 4). We suggest labelling the first of the two sub-groups as HBAl, for High Boron and Alumina, in parallel to the HIMT label, and the second sub-group as HLiBAl, for High Lithium Boron and Alumina.

Fig. 3: Li vs B in all Pergamon glasses. 'Eastern Med' captures all glasses in this assemblage that have been assigned to one of the established 1st millennium AD eastern Mediterranean compositional groups, including the various decoloured and cobalt-coloured glasses, Levantine I/II and the HIMT / weak HIMT glasses.

There are other differences between the two sub-groups, with lime, sulphate, rubidium and strontium all being much higher in HLiBAl glass, and soda, iron oxide, titania, phosphate and arsenic higher in HBAl glass (Tables 2, 3; Fig. 5, 6). Remarkable is the very low concentration of chlorine in the HLiBAl glasses, reaching only a fraction of the usual levels of around 1 wt% in other ancient glass, and pointing to an unusually chlorine-poor natron source. Schibille (2011) has developed the argument why this boron-rich glass is likely based on an evaporate deposit related to the major borate deposits in western Asia Minor (grey shaded area in Fig. 1), a few hundred kilometres northeast from Pergamon, but utilising two different sand sources and possibly also two different evaporate deposits. Our data here further corroborates this distinction.

Fig. 4: CaO vs Al2O3 in all Pergamon glasses - note the clear separation of the early glasses from the two new glass groups.

Fig. 5. Titania vs iron oxide in Pergamon glass. The HBAl group has much higher values in both oxides than the other glasses. Note also the elevated iron oxide values in the cobalt-coloured early glasses.

Fig. 6: Sr vs B (in ppm). Note the extremely high Sr content of the Li-rich glass.

The two types are broadly equally represented among the total Pergamon data set, with 16 HBAl and 12 HLiBAl samples, respectively (Table 3). HBAl glass often appears almost opaque due to the very dark colour of the glass and relatively thick working. The HBAl group includes vessels from the early Byzantine (Per 046, a lamp with loop-like handles, and Per 072, a mould-blown spiral-ribbed beaker) to the late Byzantine times (Per 047, a bowl, and Per 071, a lamp). There are also several bracelets with and without decoration (Per 008, 011, 031, 038, 040, 091) that predominantly occur in the HBAl-group. Islamic vessel imports of the 8th/9th (Per 043, 053 [?], 096) and 12th/13th century (Per 062 [Fig. 2f]) are noteworthy.

HLiBAl glass is typically transparent and faintly coloured to dark green glass, similar in appearance to HIMT glass. It contains in particular locally produced beakers of the 12th/13th century decorated with applied threads and prunts (Per 037, 048, 051, 078, 081), but also a vessel sherd with enamel-painted decoration, a product most likely from a Mamluk glass factory (Per 032). Furthermore one bracelet (Per 041), a spindle whorl (Per 080) and a crown glass window pane (Per 087) belong to this group. A beaker of the 2nd half of the 1st century

349 AD (Per 034) and a snake-trailed lamp of the mid-3rd century AD (Per 036, Fig. 2g) have to

350 be considered as extraordinary early in date, however their find contexts are well dated.

352 3.4 Other glasses

353 Four glasses from the 12th to 13th century are typical plant ash glasses (Per 016, 017, 052,

354 074). They include two enamel-painted beakers imported from the Mamluk realm and dating

355 to the 2nd half of the 13th century (Per 016 [Fig. 2h], Per 074). A fifth plant ash glass is an

356 Ottoman tulip vase of the 18th century (Per 025). Another sample from the later Ottoman

357 period (Per 088) is very unusual and apparently a European import, containing only two

358 percent by weight soda and potash each, but nearly 7.5 wt% alumina and nearly 23 wt% lime.

359 This composition resembles European window glass of the 17th century, which has very

360 similar levels of soda, potash, lime and magnesia, but only about half the alumina and iron

361 oxide levels of this glass (Dungworth 2012; Scott et al. 2012).

362 Five suspected chunks of raw glass were analysed; of these, two can be allocated to plant ash

363 glass (Per 052) and HBAl-group (Per 009), respectively. One sample (Per 092) was identified

364 as pure quartz, probably rock crystal, another as obsidian, a black natural glass (Per 093).

365 Both materials were worked in antiquity into artefacts also produced in glass, such as small

366 vessels, beads and other jewellery. Indeed, glass is often seen as a substitute for the rarer and

367 more difficult to work natural precious stones, and the inclusion of such natural materials in

368 an archaeological glass assemblage is not surprising. The last of these (Per 030) is a piece of

369 fully vitrified ceramic and not related to glass working.

371 4. Discussion

372 The chemical analysis of the glass samples from Pergamon revealed a similarly complex

373 pattern of different types and groups as was already apparent from the typological study

374 (Schwarzer 2009; Schwarzer and Rehren 2015; Schwarzer in preparation), consistent with the

375 changing fortunes of the city over more than one and a half millennium. They also throw

376 light on several issues of much wider significance, such as the relationship between glass

377 composition and glass working, the primary production of natron glass outside Egypt and the

378 Levant, and the resurgence of plant ash glass making in the early Islamic period.

380 4.1. Cast vs blown glass

The early glasses in the Pergamon assemblage span the transition from cast glass to blown glass around the first centuries BC and AD. In a previous paper Fischer and McCray (1999) suggested that this change in working technology led to a change in base composition of the glass, from 19 wt% soda to 14 to 15 wt% soda, thought to adjust the viscosity of the glass to suit the new working technology. We therefore compared the average compositions of cast glass with blown glass from the first few centuries AD. There is no noticeable difference between the cast and blown glass compositions respectively, suggesting that at least in the workshops which supplied glass artefacts to Pergamon the change in technology did not trigger a change in glass composition. Fig. 7 illustrates this for the ratio of lime vs soda; the cast and blown glasses overlap almost perfectly, regardless of whether they are decoloured by antimony or manganese, or not decoloured.

Fig. 7: Lime vs soda values for all early glasses from Pergamon, spanning the transition from cast to blown glass around the first centuries BC and AD. The cast and mould-formed glasses (diamonds) display no different composition than the free- and mould-blown glasses (open squares).

4.2. High boron high alumina glass

Dussubieux et al. (2010) recently revisited the complex pattern of ancient high-alumina glasses and identified several distinct groups. One of them is of particular interest for us, as it is closely related to our own analyses, established by several samples from Sardis (Brill 1968; 1999)1. Schibille (2011) has built on this, using a sub-set of the current Pergamon samples, and linked this high-alumina glass group to a most likely western Asia Minor production origin related to the borate deposits in western Turkey (see Fig. 1).

Since then, Swan (2012: 193) has reported analyses of 16 medieval bracelets from Hisn al-Tinat in southern Turkey, which include eleven samples with high boron levels similar to the Pergamon glasses. Five of these are similar to our HLiBAl group, while six samples are intensely coloured and characterised by very high alumina and low lime contents, similar to HBAl glass. However, the match between the two pairs of chemical groups is not perfect, with systematic differences in Rare Earth Element concentrations and some minor oxides.

An even higher boron level has been found in seven mid-1st millennium AD glasses from Aphrodisias, southeast of Izmir in western Turkey (Brill 1968; 1999). These, however, do not have elevated alumina levels. Lauwers et al. (2010) report a bracelet of similar composition from Sagalassos, also in Asia Minor, and Borisov (1989: 292, Table 24) mentions two out of

1 Brill (1999) reports six high-alumina glasses in his tables; four of these are rich in boron (0.1 to 0.25 wt% B2O3). Two are flat transparent glass, two are black bracelets. The other two are labelled 'slag' in the catalogue, and have low boron (0.01wt%).

four bracelet analyses with high boron content (0.13wt%), but no elevated alumina; these date to the 11th and 12th century AD from Djadovo in Bulgaria. This scarcity of comparative data is most likely due to the paucity of boron analyses in the literature; the main analytical methods used in glass analyses, such as SEM-EDS, EPMA and XRF, cannot easily detect boron at the low levels typically present.

Taken together, the western Asian high-boron glasses form a complex family of compositional sub-groups, within the overarching group of mineral natron glasses. At present, it is not clear whether the elevated boron levels are introduced with the sand or the natron. However, based on the geographical distribution of these high-boron glass finds, it is highly likely that at least one of the two raw materials would have come from the vicinity of the borate deposits in western Asia Minor. This then suggests that the primary production of these glasses took place somewhere in this region, and not in the known traditional glassmaking regions of Syro-Palestine and Egypt. Significantly, the emergence of this regional glassmaking tradition is not linked to the end of mineral-natron based glass making seen in the south-eastern Mediterranean in the 8th or 9th century AD. The earliest examples of these locally produced high-boron glasses date to the 1st century AD (Per 034) and the 3rd century AD (Per 036), with several others dated to the middle of the 1st millennium AD (Per 046 'early Byzantine', Per 072 '4th to 5th century AD'), and they dominate the Pergamon assemblage from the 6th century AD onwards. Clearly, primary glass production in western Asia Minor coexisted for about half a millennium with the Levantine and Egyptian glassmaking centres, and persisted well into the 2nd millennium AD. The continuing, even increasing dominance of this glass group during the early 2nd millennium, when the supply of mineral natron from the Wadi Natrun had supposedly long come to an end (Whitehouse 2002), is further strong indication for a local or regional natron source for these glasses. We therefore argue that both the sand and the natron used for these glasses were from the region.

The compositional variability within this broad glass group covers elements which are clearly linked to specific sand sources, such as iron, titanium, and zircon, as well as elements which are most likely entering the glass with the natron source, such as lithium, chlorine and sulphur. The existence of discrete compositional groups suggests that there were a number of different glassmaking sites, using their specific unique sand and individual natron sources rather than relying on a single natron source that was shared more widely.

4.3. The end of mineral natron glass in Pergamon

Relatively few samples date to the late mid-1st millennium AD; about half of them are either Levantine or HIMT glass, while the other half is of the new regional composition rich in

boron and alumina. Hardly any glass is known from the late first millennium, after the sack of the city by the Umayyads in 716 AD and the resulting hiatus in settlement. The period which in the Levant saw the switch from mineral natron to plant ash glass is therefore not represented among the Pergamon assemblage.

Only from the 10th and especially in the 12th century AD do we see a resurgence of building activity, and accordingly new deposition of glass in the archaeological record. Interestingly, all of this post-Umayyad glass is of the regional high-alumina composition, with just a few (and often imported) plant-ash based pieces among the analysed samples. Thus, the picture here differs considerably from the re-emergence of plant ash glass as the dominant glass type after the 8th or 9th century AD in the Levant. Significantly, the transition to a new glass recipe does not seem to be linked to the events in the 8th or 9th century in the Nile Delta which have been implicated in the disappearance of mineral natron glass making (Whitehouse 2002; Shortland et al. 2006). In Pergamon, the new glass composition emerges already several hundred years earlier, at a time when HIMT, Levantine I and II glass was still being produced in large quantities and available even in remote areas such as northern Bulgaria (Rehren and Cholakova 2010, 2014) or northern England (Freestone and Hughes 2006). This change in glass consumption is therefore not driven by a lack of production of glass in the Levant, but more likely by changes in the regional connectivity across the Eastern Mediterranean and the regionalisation of the Byzantine economy more generally.

4.4. Regional economy and glass supply

Chunks of raw glass, vitreous slag, manufacturing waste and deformed glass objects suggest local glass working in Pergamon and can be linked to more common vessel types, while rare vessel types within the Pergamenian assemblages are likely imported glass. Glass vessels were extremely rare in Pergamon prior to the 1st century BC (Schwarzer and Rehren 2015). The results of the sondages in the foundation of the Great Altar, erected shortly before the middle of the 2nd century BC, found thousands of pottery sherds but not a single piece of glass (de Luca and Radt 1999). Among the earliest glass vessels in Pergamon are a fragment of a mould-formed bowl with leaf decoration and a few pieces of core-formed amphoriskoi and alabastra from the 4th century BC. All of these are imports. The demand for glass rose with the integration of Pergamon into the Roman Empire, as seen in the increase of mould-formed vessels (grooved bowls, ribbed bowls) during the late 2nd and the entire 1st century BC. They were probably made locally, as indicated by the relatively high number of sherds excavated. The production of ribbed bowls in Pergamon continued until the end of the 1st century AD.

Glassblowing was introduced in Pergamon most likely not before the mid-1st century AD. Fragments of mould-blown vessels of presumably Syrian or Cypriot provenance (von Saldern 2004, 252f) (Per 070) and a small number of luxurious vessels of the early Imperial period, including mould-formed mosaic glass (Per 023), network glass (Per 035) (von Saldern 2004, 181f) and vessels with cut decoration (Per 058) presumably imported from Italy around the turn of the ages, suggest extensive trade connections. Their composition does not differ from the locally worked glass, suggesting that glass workshops in Italy and Pergamon used glass made at the same primary factories. Glass working continued throughout the Roman Imperial period but was restricted to utilitarian glass in a broad repertoire of forms. All analysed glasses from these early periods match known compositional groups used extensively elsewhere, confirming the model of centralised glass production, long-distant trade of raw glass, and local glass working, with some import of luxury objects produced elsewhere, but from glass of the same composition.

The transition into the late antique and early Byzantine period followed on seamlessly although the scope of forms was reduced significantly (Schwarzer 2009). Imports are now rare, probably as a result of the regionalisation of the early Byzantine economy visible elsewhere (Keller 2006; Hodges 2012). In Pergamon, this is reflected in the emergence of the regional glass groups rich in boron and alumina (HBAl and HLiBAl), and a paucity of glass groups that are much more dominant elsewhere in the Levant, such as Levantine I and HIMT. Interestingly, glass windows appear to be made predominantly from these imported compositions even though some of the regional glass is as transparent as HIMT; whether they were imported as ready panes, or manufactured locally, remains open.

The interruption of the settlement in Pergamon caused by the conquest of the Umayyads is reflected in a low number of glass finds from the early 8th to 10th centuries AD. With the resettlement of the citadel hill in the middle and late Byzantine period glass workshops were established again in Pergamon, now working almost exclusively the regionally produced glass. During the 12th and 13th centuries locally made beakers with nubbed decoration dominate. These vessels are so far without parallels in Asia Minor but known from the Mamluk realm and from Frankish sites in Europe. Thus, the inhabitants of late Byzantine Pergamon seem to have had contacts to the Mamluk realm, otherwise one cannot explain the remarkable number of Mamluk glass imports uncovered in the excavations on the citadel hill. The large amount of glass bracelets (ca. 1000) is also remarkable, representing almost five percent of all glass finds of the Stadtgrabung (ca. 20.000). This phenomenon is already known on other sites in Asia Minor (Lauwers et al. 2010). The Byzantine glass production ceased with the Seljuk conquest of Pergamon at the beginning of the 14th century.

5. Conclusion

The nature of the glass assemblage from Pergamon, and the changes it underwent over time initially reflect the broad chronological trends known from previous analytical studies of 1st millennium AD glass, from Roman Britain through Italy, Egypt and the Levant. The early phase is dominated by Roman blue/green glass with various levels of manganese decolouration, and by glass decolourised by antimony. The assemblage includes high-quality imported objects as well as locally-produced ones, and no distinction can be made between the main glass groups with regard to their use for particular objects. The import of finished objects as well as raw glass chunks reflect Pergamon's strong position as a major cultural centre, while the similarity of the glass compositions found in Pergamon to those used elsewhere in the Roman world is in line with the prevailing model of a centralised primary glass production supplying raw glass to secondary workshops elsewhere. Glassworkers in Italy and Syria worked glass from the same primary production centres as their colleagues in Pergamon. The introduction of glass blowing, visible in Pergamon from the first century AD, does not affect the composition of the glass used. The recipes for both antimony-decoloured and for Roman blue/green to manganese-decoloured glass remain constant from the Hellenistic period to the end of the Roman period.

The mid-1st millennium AD then sees a decline in the city's fortunes and only limited use of Levantine I and HIMT glass, but also the emergence of new glass groups, rich in boron and alumina. The identification of regionally made glass from the early to mid-1st millennium AD onward is of major importance, as it is the first evidence for regular glass making outside Egypt and the Levant in this period. On geological grounds it is reasonable to assume that this glass was made in the wider region north of Pergamon or Sardis, near the borate deposits in western Asia Minor (see Fig. 1). By the time that Pergamon is re-settled in the mid- to late Byzantine period this glass dominates the assemblage, with only a handful of plant ash glasses among the analysed fragments, many of which were imported as finished objects. The regionally produced high-boron glass falls into several chemically distinct sub-groups, indicating the existence of several discrete production sites. Its use is not restricted to Pergamon, since it seems to also have been found as far as Bulgaria and in Hisn al-Tinat in southeast Turkey near the Syrian border. It is probably only due to the limited number of analysed glass assemblages from Asia Minor that this group is not more widely visible. Its introduction had earlier been tentatively linked to the assumed collapse of natron supply during the 8th century AD. However, our wider data set now shows that it already appears in some quantity well before this collapse. Instead, it seems more likely that the emergence of the boron-rich glass groups is due to the availability of boron-rich mineral natron in western Asia Minor and the broader political and economic pattern within the Byzantine Empire at the time, resulting in a more regionalised economy and a reduction in international trade.

Acknowledgements

We are most grateful to the General Directorate of Cultural Assets and Museums (Kultur Varliklari ve Muzeler Genel Mudurlugu) in Ankara for permission to export these samples for analysis. The access to the material by the German excavation at Pergamon (German Archaeological Institute, Istanbul Department) and funding from the Deutsche Forschungsgemeinschaft for a three-year project to catalogue and investigate all glass finds from Pergamon are gratefully acknowledged by Holger Schwarzer. The LA-ICPMS analyses were done in the laboratories of Bernard Gratuze (IRAMAT, Orleans) and Laure Dussubieux (Field Museum, Chicago), partly with the help of James Lankton, and their support and advice are gratefully received. The part of the research done by Nadine Schibille was carried out at the Research Laboratory for Archaeology & the History of Art at the University of Oxford, supported by a Marie Curie Intra European Fellowship within the 7th European Community Framework Programme, Project No. 236809. Some of the research was done by Biljana Milevska in partial fulfilment of the requirements of the MSc in Technology and Analysis of Archaeological Materials at the UCL Institute of Archaeology; her contribution is gratefully acknowledged. Comments from three reviewers helped us to improve the text and are gratefully acknowledged; any remaining shortcomings are ours.

References

Abd-Allah, R., 2010. Chemical characterisation and manufacturing technology of late Roman to early Byzantine glass from Beit Ras/Capitolias, Northern Jordan. Journal of Archaeological Science 37, 1866-1874.

Borisov, B., 1989. Djadovo. Vol 1 Mediaeval Settlement and Necropolis (11th - 12th Century). Tokai University Press, Tokyo.

Brill, R., 1968. The scientific investigation of ancient glasses. Proceedings of the VIIIth International Congress on Glass, London, Sheffield, The Society of Glass Technology. pp 47-68.

Brill, R., 1999. Chemical Analyses of early Glasses. Corning Museum of Glass, New York.

Brill, R., 2012. Chemical Analyses of Early Glasses. Volume 3: The Years 2000-2011, Reports, and Essays. Corning Museum of Glass, New York.

Degryse, P., Schneider, J., Haack, U., Lauwers, V., Poblome, J., Waelkens, M., Muchez, Ph., 2006. Evidence for glass 'recycling' using Pb and Sr isotopic ratios and Sr-mixing lines: the case of early Byzantine Sagalassos. Journal of Archaeological Science 33, 494-501.

Degryse, P., Schneider, J., Lauwers, V., Waelkens, M., Muchez, Ph., 2009. Radiogenic isotopes in the provenance determination of raw materials. A case of lead and glass recycling at Sagalassos (SW Turkey). Journal of Nordic Archaeological Science 16, 1523.

Dungworth, D., 2012. Historic window glass - the use of chemical analysis to date manufacture. Journal of Architectural Conservation 18, 7-25.

Dussubieux, L., Gratuze, B., Blet-Lemarquand, M., 2010. Mineral Soda Alumina Glass -Occurrence and Meaning, Journal of Archaeological Science 37, 1646-1655.

Fischer, A., McCray, P., 1999. Glass production activities as practised at Sepphoris, Israel (37 BC-AD 1516). Journal of Archaeological Science 26, 893-905.

Foster, H., Jackson, C., 2005. 'A whiter shade of pale'? Chemical and experimental

investigation of opaque white Roman glass gaming counters. Glass Technology 46, 327333.

Foster, H., Jackson, C., 2009. The composition of 'naturally coloured' late Roman vessel glass from Britain and the implications for models of glass production and supply, Journal of Archaeological Science 36, 189-204.

Foster, H., Jackson, C., 2010. The composition of late Romano-British colourless vessel glass: glass production and consumption. Journal of Archaeological Science 37, 30683080.

Foy, D., Picon, M., Vichy, M., Thirion-Merle, V., 2003. Caractérisation des verres de la fin de l'Antiquité en Méditerranée occidentale: l'émergence de nouveaux courants commerciaux, in Foy, D., and Nenna, M.-D. (Eds.), Échanges et commerce du verre dans le monde antique: actes du colloque de l'Association française pour l'archéologie du verre, Aix-en-Provence et Marseille, 7-9 juin 2001, Éditions Monique Mergoil, Montagnac, 4186.

Freestone, I., 2005. The provenance of ancient glass through compositional analysis. Mater. Res. Soc. Symp. Proc. 852, OO8.1.1-14.

Freestone, I., 2006. Glass production in Late Antiquity and the Early Islamic period: a geochemical perspective. In: M. Maggetti and B. Messiga (eds), Geomaterials in Cultural Heritage, Geological Society London Spec Publ 257, 201-216.

Freestone, I., Hughes, M., 2006. The origins of the Jarrow glass. In: R. Cramp (ed) Wearmouth and Jarrow Monastic Sites Vol. 2, 147-155. English Heritage.

Freestone I. C., Greenwood, R., Gorin-Rosen, Y., 2002a. Byzantine and early Islamic glassmaking in the Eastern Mediterranean: production and distribution of primary glass, in: Kordas, G. (Ed.), Hyalos - Vitrum - Glass. History Technology and Conservation of glass and vitreous materials in the Hellenic World. 1st International conference Rhodes -Greece 1-4 April 2001, Athens, 167-174.

Freestone, I., Ponting, M., Hughes, M., 2002b. The origins of Byzantine glass from Maroni Petrera, Cyprus. Archaeometry 44, 257-272.

Freestone, I., Jackson-Tal, R., Tal, O., 2008. Raw glass and the production of glass vessels at Late Byzantine Apollonia-Arsuf, Israel. Journal of Glass Studies 50, 67-80.

Gallo, F., Silvestri, A., Molin, G., 2013. Glass from the Archaeological Museum of Adria (North-East Italy): new insights into Early Roman production technologies. Journal of Archaeological Science 40, 2589-2605.

Hodges, R., 2012. Adriatic Sea trade in an European perspective. In: S. Gelichi and R. Hodges, eds., From One Sea to Another: Trading Places in the European and Mediterranean Early Middle Ages. Turnhout, Brepolis, 207-34.

Jackson, C., 2005. Making colourless glass in the Roman period. Archaeometry 47, 763-780.

Kaczmarczyk, A., 1986. The source of cobalt in ancient Egyptian pigments. In: Olin, J., Blackman, M.J. (Eds.), Proceedings of the 24thInternational Archaeometry Symposium. Smithsonian Institution Press, Washington, DC, pp. 369-376.

Kato, N., Nakai, I., Shindo, Y., 2009. Change in chemical composition of early Islamic glass excavated in Raya, Sinai Peninsula, Egypt: On-site analysis using a portable X-ray fluorescence spectrometer, Journal of Archaeological Science 36, 1698-1707.

Kato, N., Nakai, I., Shindo, Y., 2010. Transitions in Islamic plant-ash glass vessels: on-site chemical analyses conducted at the Raya/al-Tur area on the Sinai Peninsula in Egypt. Journal of Archaeological Science 37, 1381-1395.

Keller, D., 2006. Die Gläser aus Petra, in: Petra ez Zantur III. Ergebnisse der SchweizerischLiechtensteinischen Ausgrabungen. Mainz. pp 1-256.

Kouwatli, I., Curvers, H., Stuart, B., Sablerolles, Y., Henderson, J., Reynolds, P., 2008. A pottery and glass production site in Beirut (BEY 015). BAAL 10, 103-130.

Lafli, E. (ed), 2009. Late Antique/Early Byzantine Glass in the Eastern Mediterranean. Izmir.

Lauwers, V., Degryse, P., Waelkens, M., 2010. Middle Byzantine (10th-13th century A.D.) glass bracelets at Sagalassos (SW Turkey), in: J. Drauschke and D. Keller, eds., Glass in Byzantium - production, usage, analyses. RGZM, Mainz, 145-152

Luca, G. de, Radt, W., 1999. Sondagen im Fundament des Grossen Altars, Pergamenische Forschungen 12, Berlin - New York.

Mirti, P., Casoli, A., Appolonia, L., 1993. Scientific analysis of Roman glass from Augusta Praetoria, Archaeometry 35, 225-240.

Nenna, M.-D., 2000. Ateliers de production et sites de consommation en Égypte, Ve s. av. J-C. VIIe s. apr. J.-C., in: Annales du 14e congrès de l'AIHV: Venise-Milan 1998, 20-24.

Nenna, M.-D., Picon, M., Thirion-Merle, V., Vichy, M., 2005. Ateliers primaires du Wadi Natrun: nouvelles decouvertes, Annales du 16e Congrès de l'Association Internationale pour l'Histoire du Verre: London 2003, 59-63.

Paynter, S., 2006. Analyses of colourless Roman glass from Binchester, County Durham, Journal of Archaeological Science 33, 1037-1057.

Radt, W., 2011. Pergamon. Geschichte und Bauten einer antiken Metropole. Darmstadt.

Rehren, Th., 2001. Aspects of the production of cobalt-blue glass in Egypt. Archaeometry 43, 483-489.

Rehren, Th., Cholakova, A., 2010. The early Byzantine HIMT glass from Dichin, Northern Bulgaria. Interdisciplinary Studies 22/23, 81-96.

Rehren, Th., Cholakova, A., 2014. Glass supply and consumption in the late Roman and early Byzantine site Dichin, northern Bulgaria. In: D. Keller, J. Price and C. Jackson (eds), Neighbours and successors of Rome - Traditions of glass production and use in Europe and the Middle East in the later 1st millennium AD. Oxbow Books, 83-94.

Rehren, Th., Spencer, L., Triantafyllidis, P., 2005. The primary production of glass at Hellenistic Rhodes. In: H. Cool, ed, Annales du 16e Congres de l'Association Internationale pour l'Histoire du Verre, Nottingham, 39-43.

Rehren, Th., Marii, F., Schibille, N., Swann, C., Stanford, L., 2010. Glass supply and

circulation in early Byzantine southern Jordan. In: J. Drauschke and D. Keller, eds., Glass in Byzantium - production, usage, analyses. RGZM, Mainz, 65-81.

Rosenow, D., Rehren, Th., 2014. Herding cats - Roman to Late Antique glass groups from Bubastis, northern Egypt. Journal of Archaeological Science 49, 170-184.

Schibille, N., 2011. Late Byzantine mineral soda high alumina glasses from Asia Minor: a new primary glass production group, PLoS ONE 6(4): e18970. doi: 10.1371/j ournal. pone. 0018970

Schwarzer, H., 2009. Spätantike, byzantinische und islamische Glasfunde aus Pergamon, in: E. Lafli (ed), Late Antique/Early Byzantine Glass in the Eastern Mediterranean. Izmir, 85109.

Schwarzer, H. in preparation, Antikes, byzantinisches und islamisches Glas aus Pergamon, Pergamenische Forschungen, Darmstadt.

Schwarzer, H., Rehren, Th., 2015. Antikes Glas aus Pergamon. Erste Ergebnisse

archäologischer und naturwissenschaftlicher Untersuchungen, in: U. Kästner - A. Scholl (eds), Pergamon als Zentrum der hellenistischen Kunst. Bedeutung, Eigenheiten und Ausstrahlung, Petersberg, in press.

Shortland, A. J., Schachner, L., Freestone, I. C., Tite, M. S., 2006. Natron as a flux in the early vitreous materials industry: sources, beginnings and reasons for decline, Journal of Archaeological Science 33, 521-530.

Scott, R., Shortland, A., Power, M., 2012. The interpretation of compositional groupings in 17th century window glass from Christ Church Cathedral, Oxford. In: Annales du 18e congres de l'AIHV (D. Ignatiadou and A. Antonaras, eds), 425-429.

Silvestri, A., 2008. The coloured glass of Iulia Felix. Journal of Archaeological Science 35, 1489-1501.

Silvestri, A., Molin, G., Salviulo, G., 2005. Roman and Medieval glass from the Italian area: bulk characterisation and relationships with production technology. Archaeometry 47, 797-816.

Silvestri, A., Molin, G., Salviulo, G., 2008. The colourless glass of Iulia Felix, Journal of Archaeological Science 35, 331-341.

Smirniou, M., Rehren, Th., 2013. Shades of blue - cobalt-copper coloured blue glass from New Kingdom Egypt and the Mycenaean World: a matter of production or colorant source? Journal of Archaeological Science 40, 4785-4792.

Swan, C., 2012. Scientific Investigation of Middle Byzantine Glass Bracelets from Hisn al-Tinat, Southern Turkey. New Evidence for High Alumina and High Boron Glasses. Poster presentation at 19e Congres d'Association Internationale pour l'Histoire du Verre (AIHV), Piran/Slowenia.

Tite, M.S., Shortland, A.J., 2003. Production technology for copper- and cobalt-blue vitreous materials from the New Kingdom site of Amarnada reappraisal. Archaeometry 45, 285312.

Uhlir, K., 2004. Naturwissenschaftliche Untersuchungen an antiken Gläsern aus Ephesos mittels m-RFA und REM/EDS. PhD dissertation, University Vienna.

Uhlir, K., Melcher, M., Schreiner, M., Czurda-Ruth, B., Krinzinger, F., 2010. SEM/EDX and [i-XRF investigations on ancient glass from Hanghaus 1 in Ephesos/Turkey. In J. Drauschke and D. Keller (eds) Glass in Byzantium - Production, Usage, Analyses. RGZM - Tagungen 8, Regensburg, 47-64.

725 Vicenzi, E., Eggins, S., Logan, A., Wysoczanski, R., 2002. Microbeam characterization of

726 Corning archeological reference glasses: new additions to the Smithonian Microbeam

727 Standard Colelction. Journal of Research of the NIST 107, 719-727.

728 von Saldern, A., 2004. Antikes Glas, Handbuch der Archäologie, München.

729 Whitehouse, D., 2002. The transition from natron to plant ash in the Levant, Journal of Glass

730 Studies 44, 193-196.

733 Table captions:

735 Table 1: Comparison of EPMA and LA-ICPMS analyses of Corning reference glasses.

736 Corning A and B were measured at the UCL Institute of Archaeology together with the

737 Pergamon samples, to their published values (Brill 1999; Vicenzi et al. 2002). Each of the

738 four individual measurements reports the average of five area analyses done on the glass.

739 The LA-ICPMS analyses were done at Orleans, and are reported against trace element

740 values from Brill (2012).

742 Table 2: Major and minor oxide composition of 96 Pergamon glasses, obtained by EPMA,

743 sorted by glass groups in broadly chronological order, and reported in wt%. In bold are

744 those element concentrations that are diagnostic for specific groups. The line within the

745 antimony-decoloured group (Sb-decol) separates those three samples which match the

746 base glass signature of this group, but have no added antimony. The line within the cobalt-

747 blue glass group separates the early from the later samples, which have different base glass

748 compositions. The column 'Lab' refers to the laboratory where the analyses were done;

749 see text for details. A fuller description of the analysed samples, including drawings,

750 photographs and details on their dating, is provided as supporting online material, SOM

751 Table 1. Sample Per 026 was analysed using SEM-EDS at UCL Qatar.

753 Table 3: Trace element compositions of 96 Pergamon glasses, obtained by LA-ICPMS, sorted

754 by glass groups in broadly chronological order, and reported in mg/g. In bold are those

755 element concentrations that are diagnostic for specific groups. For lines see caption to

756 Table 2. Data is reported to single ppm; for higher concentrations in particular we have to

757 assume an error margin in the order of tens of ppm, or more. 0 ppm indicates values below

758 0.5 ppm.

SiO2 Na2O CaO K2O MgO AI2O3 FeO TiO2 Sb2O5 MnO CuO CoO SnO2 PbO NiO ZnO BaO SrO P2O5 CI SO3 Total

Corning A 67.6 14.4 4.96 2.79 2.61 0.90 0.92 0.81 1.69 1.03 1.15 0.17 0.17 0.05 0.03 0.03 0.53 0.13 0.11 0.09 0.15 100.3

EPMA UCL 67.4 14.2 4.91 2.78 2.59 0.87 0.88 0.81 1.69 1.00 1.16 0.16 0.18 0.07 0.01 0.03 0.50 0.17 0.11 0.09 0.13 99.8

67.8 14.3 4.95 2.77 2.64 0.89 0.95 0.83 1.69 1.01 1.20 0.15 0.17 0.05 0.01 0.02 0.55 0.16 0.11 0.09 0.15 100.5

67.7 14.5 4.97 2.76 2.59 0.89 0.93 0.81 1.72 1.03 1.10 0.17 0.16 0.09 0.01 0.03 0.47 0.14 0.10 0.10 0.13 100.3

Average 67.6 14.4 4.94 2.77 2.61 0.89 0.92 0.82 1.70 1.02 1.16 0.16 0.17 0.06 0.01 0.03 0.51 0.15 0.11 0.09 0.14 100.2

Published 66.6 14.3 5.03 2.87 2.66 1.00 0.98 0.79 1.75 1.00 1.20 0.17 0.23 0.10 0.02 0.05 0.47 0.14 0.13 0.09 0.13 99.7

Corning B 62.7 16.8 8.64 1.01 1.06 4.13 0.29 0.11 0.44 0.25 2.66 0.04 0.03 0.46 0.09 0.16 0.11 0.02 0.83 0.17 0.50 100.5

EPMA UCL 62.7 17.2 8.50 1.02 1.05 4.08 0.29 0.10 0.42 0.25 2.64 0.06 0.00 0.52 0.09 0.17 0.09 0.05 0.70 0.17 0.52 100.6

62.6 17.1 8.57 1.03 1.03 4.07 0.32 0.11 0.38 0.24 2.55 0.05 0.01 0.50 0.11 0.12 0.12 0.07 0.74 0.17 0.52 100.4

62.9 17.1 8.41 1.02 1.04 4.03 0.29 0.10 0.41 0.24 2.58 0.04 0.02 0.47 0.08 0.13 0.10 0.06 0.71 0.18 0.50 100.4

Average 62.7 17.0 8.53 1.02 1.05 4.08 0.30 0.10 0.41 0.25 2.61 0.05 0.01 0.49 0.09 0.15 0.10 0.05 0.74 0.17 0.51 100.5

Published 61.6 17.0 8.56 1.00 1.03 4.36 0.31 0.09 0.46 0.25 2.66 0.04 0.03 0.50 0.10 0.19 0.09 0.02 0.82 0.16 0.45 99.7

Li2O B2O3 TiO2 V2O5 Cr2O3 MnO CoO NiO CuO ZnO AS2O3 SnO2 Sb2Oa PbO Rb2O SrO Y2O3 ZrO2 BaO CeO2 ThO2 UO2

Corning A 117 1857 7520 63 32 9916 1794 232 11122 504 34 1570 16299 710 97 1065 0.95 57 4448 0.35 0.38 0.24

LA-ICPMS Orleans 130 2017 7521 67 31 9787 1768 243 12192 560 35 1616 16329 745 102 997 0.72 51 4254 0.33 0.33 0.24

August 2012 118 1923 7744 64 31 10584 1827 238 11025 541 33 1658 17132 700 93 1114 0.84 57 4686 0.32 0.36 0.23

123 1994 7626 68 32 9817 1724 238 11065 521 35 1561 17857 690 96 1050 0.86 55 4315 0.37 0.37 0.24

124 1912 7670 65 31 10130 1820 249 11921 513 32 1699 17406 717 97 1057 0.84 54 4327 0.33 0.34 0.24

127 1965 7582 66 35 10000 1778 241 11817 505 33 1656 16586 674 99 1029 0.89 55 4282 0.34 0.32 0.23

Average 123 1945 7611 65 32 10039 1785 240 11524 524 34 1627 16935 706 97 1052 0.85 55 4386 0.34 0.35 0.23

Brill 2012 110 2200 7900 67 29 10000 1700 200 11700 440 33 1900 17000 725 93 1100 0.46 55 4600 0.29 0.37 0.23

Per 004

Per 027 Per 028 Per 054 Per 058 Per 060 Per 063

Lo/Or Ox/Ch Lo/Or Lo/Or Lo/Or Ox/Ch

Sb decol

Sb decol Sb decol Sb decol Sb decol Sb decol Sb decol

Colour

colourless

colourless

light yellow green, nearly colourless white

colourless

colourless

light yellow green, nearly colourless

working

mould-

formed

Dating (H.S.)

1st c. AD

1st c. AD 69.7

uncertain (without known parallels) 71.7

2nd/3rd c. AD 72.2

2nd half 1st c. AD 71.5

probably Roman Imperial period 70.4

late antique/early Byzantine (or 71.2

earlier?)_

Na2O K2O MgO

18.0 0.38 0.39

18.9 0.34 0.32

17.2 0.49 0.38

17.8 0.34 0.43

18.1 0.40 0.38

18.9 0.41 0.40 18.0 0.44 0.50

Al2Oa CaO FeO TO Sb2O5 MnO CuO CoO

1.68 5.8 0.28 0.05 0.79

1.70 6.3 0.36 0.05 0.63

1.78 6.2 0.29 0.08 0.89

1.53 5.7 0.32 0.07 0.51

1.63 5.3 0.32 0.06 0.71

1.83 5.5 0.34 0.05 0.87

2.00 6.1 0.34 0.08 0.58

bdl bdl bdl

bdl bdl bdl

bdl bdl bdl

bdl bdl bdl

bdl bdl bdl

bdl bdl bdl

bdl bdl bdl

0.02 0.03 0.03 0.02 0.02 0.02

1.09 0.92

1.10 1.00 1.26

0.32 0.28 0.28 0.26 0.31 0.25

Per 090 Per 055 Per 019

Lo/Or Lo/Or

Sb decol Sb decol Sb decol

Per 061 Lo Sb-Mn decol

Per 067 Lo/Or Sb-Mn decol

Per 010 Lo/Or Sb-Mn decol

Per 089 Lo Sb-Mn decol

Per 083 Lo/Or Sb-Mn decol

Per 020 Lo Sb-Mn decol

Per 069 Lo/Or Sb-Mn decol

Per 042 Lo/Or n/Mn decol

Per 005 Lo

n/Mn decol

Per 079 Lo/Or n/Mn decol

Per 001 Lo n/Mn decol

Per 007 Lo/Or n/Mn decol

Per 077 Lo/Or n/Mn decol

Per 002 Lo/Or n/Mn decol

yellowish olive yellowish brown yellowish green

greenish yellow yellowish green greenish blue aqua

yellowish green aqua

greenish blue purple

yellowish green

light yellow green, nearly colourless colourless

mould-

formed

mould-

mould-

mould-

formed

mould-

formed

mould-

formed

mould-

formed

probably 3rd/4th c. AD (perhaps 73.1 17.4 0.36 0.33 earlier)

mid-2nd - early 1st c. BC 72.2 16.4 0.62 0.58

1st c. AD 72.0 17.2 0.39 0.30

1.72 5.6 0.25 0.04 1.87 6.6 0.33 0.06 2.07 5.9 0.29 0.04

uncertain

presumably late Roman Imperial period

mid-1st - beginning 2nd c. AD 1st/2nd c. AD 2nd c. BC

mid-1st - 2nd c. AD

presumably early Roman Imperial period

1st half 1st c. AD probably 3rd/4th c. AD 3rd - 4th c. AD

2nd c. AD or perhaps little later mid-1st c. BC - mid-1st c. AD 1st half 1st c. AD

3rd third 1st c. BC - mid-1st c. AD

68.0 18.0 0.51 0.68 2.01 5.5 0.74 0.10

70.3 69.7

17.4 0.63 0.48 16.9 0.84 0.89

69.7 16.5 0.74 0.57

66.9 18.2 0.62 0.66

70.9 15.9 1.11 0.58

71.2 16.1 0.72 0.52

2.14 6.5 0.36 0.05

2.06 6.8 0.58 0.07

2.54 7.1 0.51 0.09

2.11 8.6 0.32 0.04

2.28 6.9 0.47 0.06

2.32 7.0 0.40 0.07

17.6 0.54 0.53

69.0 15.4 0.59 0.61

71.3 15.0 0.50 0.49

71.2 14.8 0.51 0.48 67.8 18.2 0.44 0.49 70.8 15.3 0.73 0.52

70.3 16.5 0.60 0.49

2.34 2.38

2.35 2.32

8.0 0.38 0.05 8.8 0.40 0.07

7.4 0.36 0.05

7.5 0.39 0.06 7.4 0.41 0.04

2.35 7.6 0.36 0.05

2.39 7.2 0.35 0.04

bdl bdl bdl bdl 0.08 1.07 0.17

bdl bdl bdl bdl 0.11 0.95 0.12

bdl 0.29 bdl bdl 0.13 1.12 0.14

0.61 0.96 bdl bdl 0.09 1.09 0.32

0.43 0.51 bdl bdl 0.08 1.01 0.21

0.25 0.24 bdl bdl 0.23 0.98 0.22

0.22 0.71 bdl bdl 0.09 0.94 0.21

0.15 0.82 bdl bdl 0.08 0.89 0.35

0.14 0.36 bdl bdl 0.15 0.93 0.16

0.11 0.30 bdl bdl 0.13 1.06 0.12

bdl 1.77 bdl bdl 0.13 0.92 0.25

bdl 1.70 bdl bdl 0.14 0.93 0.20

bdl 1.45 bdl bdl 0.11 1.05 0.12

0.04 1.43 bdl bdl 0.13 0.95 0.12

bdl 1.08 bdl bdl 0.09 1.15 0.27

bdl 0.74 bdl bdl 0.18 0.98 0.13

bdl 0.58 bdl bdl 0.12 1.00 0.15

Per 035 Per 075 Per 049 Per 057 Per 059 Per 082 Per 097 Per 018 Per 084 Per 044 Per 013 Per 021 Per 045 Per 098 Per 022 Per 070

Per 012 Per 014 Per 023 Per 024 Per 026 Per 050 Per 068 Per 073 Per 086

Lo/Or Lo

Lo/Or Lo/Or Lo/Or Qa

Lo/Or Lo/Or Lo/Or Lo/Or

n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol n/Mn decol

Co bl Co bl Co bl Co bl Co bl Co bl Co bl Co bl Co bl

colourless greenish blue olive yellow aqua

greenish blue aqua

yellowish green yellowish green yellowish olive greenish blue yellowish green yellow green yellowish brown olive yellow yellowish green yellowish green

ultramarine cobalt blue blue

pale blue

ultramarine

ultramarine

ultramarine

ultramarine

cobalt blue

mould-

formed

mould-

formed

mould-

mould-

formed

mould-

formed

formed

formed

end 1st c. BC - early 1st c. AD 2nd half 1st - 1st half 2nd c. AD 3rd third 1st - 1st third 2 c. AD 1st c. AD

mid-1st - mid-2nd c. AD 1st c. AD

2nd half 3rd - 1st half 4th c. AD

3rd - 4th c. AD

probably 3rd - 4th c. AD

mid-1st c. AD

mid-1st - 2nd c. AD

2nd - beginning 3rd c. AD

3rd - 2nd quarter 1st c. BC

1st half 1st c. AD

1st - early 2nd c. AD

1st c. AD

1st - 3rd quarter 1st c. AD

end 2nd - early 1st c. BC

end 1st c. BC - early 1st c. AD

2nd - 1st half 1st c. BC

late Hellenistic/early Roman Imperial

1st half 1st c. AD 1st c. AD 1st half 1st c. AD 3rd - 2nd c. BC

72.3 70.1

69.7 71.9

71.1 72.3 71.6

71.9 72.8

15.1 16.8

15.6 15.4

15.8 16.1

15.5 15.2 16.4

0.51 0.43 0.49 0.39 0.47 0.74 0.52 0.58 0.50 0.45 0.37 0.49 0.66 0.44 0.48 0.55

69.2 69.7

0.50 0.44 0.46 0.40 0.42 0.47 0.45 0.53 0.41 0.40 0.40 0.39 0.76 0.50 0.42 0.45

68.7 17.3 0.56 0.50

17.3 0.65 0.57

16.4 0.53 0.56

67.4 17.0 0.61 0.57

68.5 16.3 0.61 0.60

68.2 17.5 0.56 0.56

2.03 2.59 2.34 2.26 2.11

2.38 2.34 2.44 2.57 2.32

2.37 2.37 2.29 2.48

8.6 0.27

7.0 0.28

7.7 0.33 6.9 0.22

7.2 0.27

7.8 0.37

7.3 0.31

7.9 0.34

7.1 0.29 7.8 0.33 7.6 0.27 7.5 0.28

7.8 0.48

7.4 0.30

6.9 0.30 7.1 0.30

2.41 2.22 2.46 2.26

0.04 0.04 0.04 0.04 0.05 0.04 0.06 0.06 0.02 0.04 0.06 0.04 0.05 0.04 0.05 0.05

2.28 7.8 0.81 0.04 1.78 6.6 1.28 0.07

7.6 1.19 0.05

8.1 1.33 0.05 8.0 1.39 0.07

8.2 0.75 0.04

69.2 16.9 0.76 0.52 2.39 7.3 0.83 0.03

0.05 bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl

0.51 0.37 0.35 0.26 0.25 0.24 0.19 0.10 0.08 bdl bdl bdl bdl bdl bdl bdl

bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl

bdl 0.08

bdl 0.08

bdl 0.15

bdl 0.10

bdl 0.13

bdl 0.16

bdl 0.15

bdl 0.13

bdl 0.11

bdl 0.09

bdl 0.10

bdl 0.10

bdl 0.09

bdl 0.12

bdl 0.12

bdl 0.16

0.98 1.13 1.01 1.07 1.03 0.95 0.86 0.82 1.03 1.10 1.03 1.07 0.89 1.15 0.99 1.06

0.20 0.15 0.14 0.16 0.13 0.21 0.13 0.16 0.11 0.14 0.11 0.11 0.26 0.13 0.08 0.14

68.4 17.7 0.53 0.50 2.36 65.0 15.9 0.75 0.62 2.17

7.6 0.79 0.05 8.3 0.76 0.05

bdl 0.42 0.05 0.04 0.08 0.85 0.39

0.12 0.45 0.39 0.29 0.10 0.82 0.22

bdl 0.17 0.10 0.09 0.10 0.78 0.19

bdl 1.07 0.12 0.04 0.11 0.85 0.29

0.18 0.36 0.11 0.10 0.09 0.82 0.23

bdl 0.26 0.06 0.06 0.13 0.99 0.27

bdl 0.55 0.09 0.07 0.14 0.75 0.28

bdl 0.33 0.07 0.06 0.11 1.08 0.25

0.04 0.51 0.45 0.07 0.12 0.82 0.31

Per 033 Ox/Ch Co bl

ultramarine

early Roman Imperial period

69.6 15.1 0.50 0.47

.2 0.77 0.05 2.05 0.42 0.10 0.07 0.16 0.96 0.24

Per 064 Ox/Ch Co bl greenish blue cast early or late Byzantine 67.2

Per 066 Ox/Ch Co bl greenish blue free- late antique/early Byzantine 66.3

Per 080 Lo/Or Co bl cobalt blue late Roman or Byzantine 66.8

Per 003 Lo Lev I colourless free- mid-1st - 2nd c. AD 71.0

Per 029 Lo/Or Lev I yellowish green free- mid-3rd c. AD 70.0

Per 100 Ox/Ch Lev I light olive, nearly cast probably late antique 68.7

colourless

Per 065 Ox/Ch Lev II greenish blue cast early or late Byzantine 73.5

Per 056 Lo/Or w HIMT yellowish green mould- 2nd half 1st c. AD 66.3

Per 094 Lo w HIMT greenish blue free- 5th - 7th c. AD 69.3

Per 085 Ox/Ch w HIMT aqua free- 5th/6th c. AD 66.5

Per 099 Ox/Ch HIMT greenish olive cast probably late antique 65.8

Per 008 Lo/Or HBAl black olive bracelet 12th/13th c. AD 60.9

Per 009 Ox/Ch HBAl yellowish olive chunk uncertain (find context 13th c. AD) 53.0

Per 011 Lo/Or HBAl brownish red bracelet Roman or early Byzantine 60.1

Per 015 Ox/Ch HBAl dark yellowish free- presumably 12th - beginning 13th 56.5

green blown c. AD

Per 031 Lo/Or HBAl black olive bracelet 13th c. AD 59.1

Per 038 Lo/Or HBAl black olive bracelet 12th/13th c. AD 58.6

Per 040 Lo/Or HBAl yellowish green bracelet Byzantine 72.9

Per 043 Ox/Ch HBAl dark red marbled free- 8th/9th c. AD (early Islamic) 56.7

Per 046 Lo/Or HBAl yellowish green free- early Byzantine 65.7

Per 047 Ox/Ch HBAl brownish green free- late Byzantine 63.8

Per 053 Ox/Ch HBAl yellowish brown free- 8th - early 9th c. AD (early Islamic) 55.3

Per 062 Ox/Ch HBAl reddish brown free- 12th/13th c. AD (Islamic, probably 57.4

blown Mamluk)

Per 071 Ox/Ch HBAl yellowish olive free- 12th/13th c. AD 57.8

Per 072 Lo/Or HBAl dark olive green free- 4th/5th c. AD 57.1

Per 091 Lo/Or HBAl black olive bracelet mid- or (most probably) late 52.9

Byzantine

15.0 0.70 0.63 2.55 8.5 0.68 0.01 0.98 0.65 0.37 0.02 0.18 0.89 0.20

14.5 0.57 0.62 2.43 8.4 0.99 0.09 1.29 0.53 0.52 0.03 0.16 0.67 0.21

13.0 1.07 0.63 2.20 9.2 0.92 0.05 3.97 0.30 0.30 0.24 0.18 0.29 0.64

13.2 0.37 0.55 3.04 8.7 0.39 0.06 bdl 1.46 bdl bdl 0.07 1.00 0.07

14.1 0.53 0.62 3.02 8.1 0.57 0.06 bdl 1.85 bdl bdl 0.10 0.94 0.09

16.2 0.73 0.57 2.81 9.1 0.35 0.07 bdl 1.15 bdl bdl 0.09 0.94 0.21

14.5 0.53 0.44 3.13 7.3 0.49 0.12 bdl bdl bdl bdl 0.05 0.79 0.15

19.2 0.46 0.96 2.24 6.7 1.08 0.16 bdl 1.09 bdl bdl 0.07 1.17 0.31

17.7 0.59 0.78 2.80 5.9 0.82 0.16 bdl 0.46 bdl bdl 0.11 0.73 0.26

18.0 0.78 0.95 2.50 8.6 0.85 0.16 bdl 0.79 bdl bdl 0.15 0.84 0.21

18.6 0.49 0.89 2.53 6.3 1.20 0.47 bdl 2.64 bdl bdl 0.05 1.00 0.28

15.2 1.81 1.43 10.4 4.8 1.69 0.59 bdl bdl bdl bdl 0.19 0.95 0.10

19.5 1.71 2.21 11.4 7.4 2.85 0.64 bdl 0.27 bdl bdl 0.27 1.01 0.13

13.1 1.63 1.42 9.5 5.1 3.99 0.59 bdl 0.66 0.74 bdl 0.23 0.74 0.10

17.3 1.77 1.42 9.7 5.0 1.60 0.65 bdl 2.95 bdl bdl 0.36 0.98 0.13

17.0 1.56 1.45 10.1 5.0 1.71 0.63 bdl bdl bdl bdl 0.22 1.14 0.20

18.9 1.58 1.27 9.8 4.3 1.50 0.57 bdl bdl bdl bdl 0.23 1.23 0.26

12.0 1.20 0.80 3.6 5.9 1.10 0.27 bdl 1.09 bdl bdl 0.09 0.36 0.06

14.5 2.14 2.24 11.0 7.4 2.74 0.72 bdl bdl 0.67 bdl 0.28 0.99 0.16

14.5 1.73 0.83 8.0 4.5 1.10 0.35 bdl 0.61 bdl bdl 0.15 0.53 0.22

18.3 1.21 1.18 6.7 4.4 1.41 0.50 bdl 0.63 bdl bdl 0.27 1.15 0.15

17.8 1.78 1.58 9.9 5.3 1.95 0.63 bdl 3.66 bdl bdl 0.36 0.95 0.15

22.3 1.08 1.34 8.1 4.8 1.85 0.60 bdl 1.38 bdl bdl 0.22 1.14 0.31

19.2 1.27 1.57 9.8 5.3 2.16 0.82 bdl 0.59 bdl bdl 0.24 1.14 0.06

19.1 1.50 1.36 9.5 4.4 1.60 0.60 bdl 1.40 bdl bdl 0.26 1.10 0.12

24.2 1.23 1.30 9.7 4.6 1.79 0.58 bdl bdl bdl bdl 0.39 1.45 0.18

Per 096 Ox/Ch HBAl

dark red marbled

free-blown

8th/9th c. AD (early Islamic)

57.6 18.5 1.82 1.47

4.7 2.12 0.68

bdl 0.16 1.50

0.32 1.19 0.16

Per 032

Per 034

Per 036

Per 037

Per 039

Per 041 Per 048

Per 051

Per 076

Per 078

Per 081

Per 087

Lo/Or Ox/Ch

HLiBAl

HLiBAl

HLiBAl

HLiBAl

HLiBAl

HLiBAl HLiBAl

HLiBAl

HLiBAl

HLiBAl

HLiBAl

HLiBAl

Per 016 Ox/Ch PA

Per 017 Ox/Ch PA

Per 025 Ox/Ch PA

Per 052 Ox/Ch PA

Per 074 Ox/Ch PA

olive green

colourless

yellowish green

light olive green, nearly colourless light reddish brown bluish green

olive green green olive colourless yellowish green reddish brown reddish brown

colourless olive

yellowish brown blue

olive yellow

bracelet

free-blown

end 12th - beginning 13th c. AD

2nd half 1st c. AD (find context)

mid-3rd c. AD (find context)

12th/13th c. AD

late Byzantine

Byzantine 12th/13th c. AD

13th c. AD

early or mid-Byzantine

12th/13th c. AD

12th/13th c. AD

mid- or (most probably) late Byzantine

2nd half 13th c. AD (Mamluk)

12th - 13th c. AD

18th c. (Ottoman)

uncertain (Byzantine find context) 2nd half 13th c. AD (Mamluk)

60.3 15.5 1.62 1.26

72.0 13.4 0.69 0.87

64.1 15.0 1.63 0.96

67.9 14.5 1.25 0.89

65.1 14.0 2.06 0.84

68.7 13.2 1.14 1.21

62.0 15.2 2.22 1.12

59.7 18.4 1.51 1.23

65.4 13.3 2.23 0.76 64.0 15.8 1.58 1.02

67.4 14.8 1.00 0.97

61.5 16.6 1.75 0.99

70.0 10.7 2.54 3.59

70.1 12.0 2.16 3.13

59.6 14.0 3.27 2.31

66.3 14.6 3.47 4.52

69.0 12.0 1.69 2.55

.3 1.09 0.30

0.9 9.4 0.40 0.04

8.2 0.68 0.19

3.2 9.4 0.55 0.10

5.9 9.1 0.64 0.12

2.6 11.1 0.67 0.09

5.9 10.0 0.90 0.15

5.4 10.8 0.94 0.21

6.5 7.9 0.61 0.09

5.3 9.6 0.66 0.11

2.4 10.1 0.71 0.12 5.2 10.1 0.51 0.12

0.9 1.2

8.4 0.37 0.22 8.4 0.57 0.23

7.5 6.8 1.69 0.28

0.7 8.5 0.38 0.03 1.4 10.3 0.97 0.25

bdl 1.43

bdl 0.73

bdl 0.56

bdl 1.46

bdl bdl

bdl bdl

bdl bdl

bdl bdl

bdl 1.28 bdl bdl

bdl 0.12

bdl 0.72

bdl 0.92

bdl 0.87

bdl 0.56

bdl 1.37

bdl bdl

bdl bdl

bdl bdl

bdl bdl

bdl bdl

bdl bdl

bdl 1.56 bdl bdl

0.13 0.14

0.34 0.15

0.19 0.33

bdl 0.90 bdl bdl 0.25 0.72 0.28

bdl 1.00 bdl bdl 0.32 0.81 0.21

bdl 2.68 bdl bdl 0.45 0.74 0.07

bdl bdl bdl bdl 0.29 0.86 0.23

bdl 1.60 bdl bdl 0.35 0.90 0.16

Per 088 Ox/Ch European

brownish olive free- Ottoman (apparently European

blown import)

2.1 2.14 2.59

7.4 22.8 2.04 0.33

bdl bdl bdl

0.15 0.22 0.28

Per 093 Ox/Ch Obsidian

find context late Byzantine

4.2 5.02 0.12 13.5 0.8 0.92 0.14

bdl bld bdl

bld 0.05 bdl

I.D. Lab type Li B Ti V Cr Mn Co Ni

Per 027 Lo/Or Sb decol 0 142 397 6 8 169 1 2

Per 028 Ox/Ch Sb decol 3 127 296 6 6 155 1 3

Per 054 Lo/Or Sb decol 3 157 414 8 6 91 0 1

Per 058 Lo/Or Sb decol 0 123 379 7 8 150 1 2

Per 060 Lo/Or Sb decol 4 201 410 6 7 134 0 2

Per 063 Ox/Ch Sb decol 3 167 315 8 6 150 1 3

Per 055 Lo/Or Sb decol 1 85 330 8 9 285 1 4

Per 019 Lo/Or Sb decol 1 40 316 10 11 2954 5 7

Per 067 Lo/Or Sb-Mn decol 4 184 453 18 9 3625 4 4

Per 010 Lo/Or Sb-Mn decol 0 175 508 14 11 2871 4 6

Per 083 Lo/Or Sb-Mn decol 2 96 273 14 40 7627 6 10

Per 069 Lo/Or Sb-Mn decol 4 168 455 12 10 2524 5 6

Per 042 Lo/Or n/Mn decol 0 117 334 32 10 16482 44 21

Per 079 Lo/Or n/Mn decol 3 110 410 55 12 9968 9 9

Per 007 Lo/Or n/Mn decol 0 220 309 21 9 9964 12 12

Per 077 Lo/Or n/Mn decol 0 128 334 18 10 7048 14 15

Per 002 Lo/Or n/Mn decol 0 123 279 13 9 5430 19 12

Per 035 Lo/Or n/Mn decol 0 119 269 10 8 4706 5 7

Per 075 Lo/Or n/Mn decol 3 75 337 13 8 4186 3 7

Per 049 Lo/Or n/Mn decol 3 214 330 11 8 2688 2 6

Per 057 Lo/Or n/Mn decol 0 74 294 9 9 2601 4 5

Per 059 Lo/Or n/Mn decol 0 91 300 14 12 2571 4 4

Per 082 Lo/Or n/Mn decol 0 132 331 8 10 2484 4 5

Per 044 Lo/Or n/Mn decol 0 98 317 5 10 743 2 3

Per 045 Lo/Or n/Mn decol 1 85 399 11 10 459 2 4

Per 070 Lo/Or n/Mn decol 1 132 330 7 10 190 1 3

Per 014 Lo/Or Cobl 4 192 499 14 10 4814 1949 65

Per 023 Lo/Or Cobl 4 280 342 13 11 2567 577 16

Per 024 Lo/Or Cobl 0 203 401 21 12 10767 178 21

Per 050 Lo/Or Cobl 0 138 331 11 11 2903 444 15

Cu Zn As Sn Sb Pb Rb Sr Y Zr Ba La Ce Nd Th u

8 12 112 10 4951 33 4 469 4 40 123 5 10 5 1

9 13 0 0 3994 16 6 397 5 34 126 5 8 4 0

7 10 13 12 3486 19 4 463 6 46 127 6 11 6

17 11 13 13 5237 111 5 413 5 43 136 5 10 6

3 15 13 11 5995 11 5 411 5 47 142 6 10 5

8 15 0 0 2567 23 5 344 5 37 124 5 9 5

12 12 1 7 9 37 7 326 5 30 156 5 9 5

9 14 2 11 33 4 6 337 5 34 185 5 10 5

10 17 13 13 2955 20 6 462 7 48 215 7 11 6

22 22 9 17 2291 91 6 461 5 44 210 6 11 6

6 16 5 7 2019 78 9 601 7 29 229 6 11 6

41 14 4 26 968 199 7 462 7 47 208 7 12 7

112 26 3 18 323 96 7 641 6 30 336 6 11 6

12 13 4 15 429 17 4 498 8 45 368 7 12 7

9 15 2 8 17 7 2 507 5 29 346 5 10 6

23 26 2 9 147 238 8 463 5 30 243 6 11 6

40 20 2 7 36 21 2 418 4 22 208 5 10 5

8 13 3 8 589 426 8 516 6 28 199 6 13 6

4 9 3 10 0 0 5 461 7 37 239 7 11 7

5 10 2 13 50 56 6 463 7 35 215 6 11 6

4 10 2 7 0 0 6 394 5 29 204 5 15 6

3 10 2 6 0 1 6 427 6 37 232 6 11 6

28 13 2 10 138 17 9 429 6 31 294 6 12 6

11 8 2 8 46 5 6 402 6 32 229 6 12 6

6 10 2 8 0 0 11 441 6 31 191 6 11 6

5 10 2 7 1 0 7 360 5 30 198 5 11 5

2763 57 36 177 1143 177 7 440 6 37 173 6 10 6

626 44 4 15 112 201 7 451 6 35 229 7 11 6

556 96 4 11 122 422 8 614 7 39 262 7 12 7

369 23 3 24 109 24 7 480 6 31 220 6 11 6 2

Per 068 Lo/Or Co bl 0 99 354 15 10 5809 502 17

Per 073 Lo/Or Cobl 0 124 310 10 10 3656 393 12

Per 086 Lo/Or Co bl 1 147 350 16 10 5111 478 51

Per 033 Ox/Ch Cobl 3 117 206 15 12 2849 458 24

Per 064 Ox/Ch Cobl 4 137 307 19 9 4674 101 21

Per 066 Ox/Ch Cobl 4 127 331 17 9 3666 169 16

Per 080 Lo/Or Cobl 0 187 389 12 9 3371 1622 67

Per 029 Lo/Or Levi 3 79 414 33 11 13671 10 24

Per 100 Ox/Ch Levi 3 108 199 17 10 8847 5 6

Per 065 Ox/Ch Lev II 5 50 400 10 9 178 2 5

Per 056 Lo/Or wHIMT 0 154 1044 35 19 11353 12 15

Per 085 Ox/Ch wHIMT 7 149 679 27 15 7116 12 14

Per 099 Ox/Ch HIMT 5 188 1232 50 50 18016 11 16

Per 008 Lo/Or HBAI 21 665 3983 52 75 283 4 29

Per 009 Ox/Ch HBAI 24 1424 3156 78 92 902 9 50

PerOll Lo/Or HBAI 58 832 3936 165 74 8411 27 59

Per 015 Ox/Ch HBAI 18 941 3123 310 83 20431 27 40

Per 031 Lo/Or HBAI 21 953 4031 55 78 298 5 32

Per 038 Lo/Or HBAI 20 1120 3544 51 66 463 5 28

Per 040 Lo/Or HBAI 126 1123 1766 23 60 9651 8 22

Per 043 Ox/Ch HBAI 26 580 1965 60 97 516 10 61

Per 046 Lo/Or HBAI 103 940 2216 91 40 6325 13 15

Per 047 Ox/Ch HBAI 16 1066 1264 67 70 4051 9 26

Per 053 Ox/Ch HBAI 23 954 2621 57 85 26383 6 43

Per 062 Ox/Ch HBAI 17 657 1909 43 66 8115 5 27

Per 071 Ox/Ch HBAI 24 712 3757 82 103 4000 16 34

Per 072 Lo/Or HBAI 17 1124 3810 288 58 11703 14 29

Per 091 Lo/Or HBAI 26 1750 3889 69 78 294 4 32

Per 096 Ox/Ch HBAI 19 694 1952 66 81 1108 8 39

Per 032 Ox/Ch HLiBAl 258 1337 1641 153 44 10601 21 22

549 26 4 38 113 41 11 517 7 37 260 7 12 7 1 2

558 28 3 29 131 76 6 440 6 29 229 6 11 6 1 1

3584 102 154 5935 447 22727 9 562 8 35 209 8 11 7 1 1

1073 32 10 9 10581 6465 7 441 9 40 230 7 12 7 1 1

3385 100 8 334 6491 24095 8 429 6 35 240 6 11 6 1 1

4277 93 6 551 5922 34970 7 438 7 42 239 7 11 6 1 1

2055 52 33 68 33983 141 13 499 7 35 242 7 12 7 1 1

16 22 4 7 5 5 6 610 8 34 497 8 12 8 1 1

28 13 3 1 1 12 10 446 7 33 432 6 12 6 1 1

27 8 0 3 45 176 9 329 7 40 210 6 13 6 1 2

110 23 10 20 331 78 5 702 8 88 614 9 16 9 1 1

89 31 0 18 50 277 9 546 8 65 260 7 13 7 2 2

146 27 5 12 5 101 5 424 9 173 1236 8 15 8 2 1

14 19 323 10 1 9 46 196 26 284 431 30 60 28 9 4

37 38 259 3 1 18 34 188 29 282 537 29 58 26 10 9

7258 26 498 61 229 724 36 478 29 390 2651 59 101 41 24 4

21 41 227 2 3 14 35 207 29 279 5259 29 59 26 8 7

7 32 225 9 1 6 36 195 27 279 424 32 66 30 9 6

72 20 447 24 1 16 36 178 23 226 375 27 58 24 8 5

32 73 70 14 30 64 23 192 12 134 239 12 21 11 4 1

5698 40 169 554 41 356 55 230 33 311 369 29 52 25 7 2

5 25 299 8 7 44 61 850 18 180 1769 23 46 19 11 4

9 22 1451 1 1 12 29 161 22 232 484 20 41 18 5 5

31 672 524 7 196 44 36 408 27 266 733 27 54 25 9 5

15 44 341 13 578 29 20 193 20 215 249 20 42 19 6 14

10 31 140 2 2 19 25 221 35 400 1334 36 74 33 12 6

333 26 604 45 4 28 33 180 29 295 4212 31 64 31 11 4

10 16 803 11 0 9 24 182 29 311 428 33 59 31 10 6

11218 31 214 115 19 492 47 156 25 237 514 26 54 24 7 3

11 34 6 2 15 24 71 1986 22 169 2766 19 41 17 11 8

Per 034 Lo/Or HLiBAl 347 1195 285 141 7 6912 9 5 9 15 22 8 10 2 53 3596 2 15 1826 3 8 3 1 1

Per 036 Lo/Or HLiBAl 247 1223 1100 68 21 5043 8 7 23 17 67 10 7 12 69 2656 12 89 1360 15 33 13 9 3

Per 037 Ox/Ch HLiBAl 303 1433 308 226 9 9996 20 7 15 29 27 1 42 36 75 2895 5 31 2977 8 15 6 5 2

Per 039 Ox/Ch HLiBAl 277 1367 366 123 9 9448 14 6 10 31 34 1 10 35 112 2517 11 43 2714 13 57 10 9 4

Per 041 Lo/Or HLiBAl 376 1181 552 22 9 1399 2 5 4 14 12 9 11 6 84 5563 3 24 517 7 14 5 3 1

Per 048 Ox/Ch HLiBAl 339 1451 363 66 10 5190 13 9 14 26 28 1 8 18 117 2663 9 49 1139 12 27 9 10 4

Per 051 Ox/Ch HLiBAl 300 1362 791 82 30 6483 20 12 9 27 242 1 8 18 87 3079 17 129 1973 15 30 13 9 3

Per 076 Lo/Or HLiBAl 256 1093 520 112 5 6946 12 4 7 19 23 9 8 16 102 2652 9 33 1749 11 21 8 10 4

Per 078 Ox/Ch HLiBAl 352 1784 592 66 6 3836 6 7 8 26 0 1 9 27 97 2644 7 26 1139 10 26 7 12 5

Per 081 Ox/Ch HLiBAl 388 1810 587 169 8 9865 16 7 15 24 2 0 14 8 66 3120 5 28 2754 8 15 6 3 1

Per 087 Ox/Ch HLiBAl 438 1722 447 172 6 10766 14 7 8 35 0 7 9 24 112 2979 9 27 2789 8 24 7 9 4

Per 016 Ox/Ch PA 6 70 1047 20 16 7091 7 9 41 69 0 1 0 39 9 541 7 125 359 6 13 6 1 1

Per 017 Ox/Ch PA 6 57 1072 18 19 8224 8 9 33 94 0 48 0 149 9 554 7 119 208 6 12 6 1 1

Per 025 Ox/Ch PA 22 2 1373 30 19 21094 12 22 55 114 0 15 1 170 39 327 17 129 1095 27 54 20 7 3

Per 052 Ox/Ch PA 11 93 139 6 3 191 59 5 113 98 0 11 2 232 10 544 2 8 44 2 3 1 0 0

Per 074 Ox/Ch PA 8 55 809 23 23 11213 7 14 62 131 2 1 1 22 8 505 10 118 234 7 14 7 1 2

Per 088 Ox/Ch European 32 51 1705 42 61 519 6 23 14 53 0 3 1 27 64 382 21 182 286 25 48 23 9 3

Per 093 Ox/Ch Obsidian 63 32 451 3 3 584 0 2 3 39 5 2 1 52 173 90 9 96 390 39 74 18 18 6

Fig, 2a; Mould-blown lotus-beaker, nominally antimony-decoloured glass but with a yellowish-green tint and no added antimony; sec text for discussion. Sample Per 019 (diam. body 4.5-5 em).

Fig. 2b; Mould-formed late Hellenistic ribbed bowl, Roman b/g glass. Sample Per 045 (diam. rim 18 cm).

Fig. 2c: Hellenistic alabastron, Co-blue glass. Sample Per 086 (diam. body 1.5-2.3 cm).

Fig, 2d: Mould-blown bottle base with a rosette. Mixed Sb and Mn decol. Sample Per 010 (diam. base 7 cm).

Fig. 2e: Early Imperial mould-blown ribbed bowl, weak H1MT glass. Sample Per 056 (diam. rim 7.6 em).

Fig. 2f: 12th century Islamic import, plant ash glass. Sample Per 062 (diam. rim 10.8 cm; diam. base 4.4 cm).

•■W . 1 r »U

Fig. 2g: Snake-trailed lamp, 3rd c AD, I ILiBAl glass. Sample Per 036 (max, length 5.2 cm).

Fig. 2h: Mamluk enamel-painted beaker, import, plant ash glass. Per 016 (4.4 x 2,7 cm [left]; 1,8 x 2.2 cm [right])

400 --

—300

200 --

100 --

♦ Eastern Med ■ HBAI A HLiBAl

▲ ▲

A A ▲

800 1200 B (ppm)

▲ A ♦ Eastern Med

X ▲ A ♦ A ♦ A A A A ■ HBAI A HLiBAI x Plant ash

x X X V % A A A

c* X ■ ■

■ ■ ■

--1-1— -1-1- -1-1

AI2O3 (wt%)

1.0 -r

0.8 --

♦ Eastern Med ■ HBAI HLiBAI

SO 0.6 "

¡- 0.4 +

0.2 --

#4 ♦

% k ♦ ♦ ♦

-1—*

1.5 2.0 2.5

FeO (wt%)

6000 -T

5000 --

♦ Eastern Med ■ H6AI ▲ HLiBAI x Plant ash

4000 +

£ 3000 + t/5

2000 --

A A aA

1000 --

800 1200 B (ppm)

♦ Mould-formed □ Free-blown

Dd № ° ^

Q «□ □ ♦

□ ♦

N320 (wt%)

□ □