Content uploaded by Eddy Faber
Author content
All content in this area was uploaded by Eddy Faber on Feb 12, 2018
Content may be subject to copyright.
THE PRODUCTION AND CIRCULATION OF CARTHAGINIAN
GLASS UNDER THE RULE OF THE ROMANS AND THE
VANDALS (FOURTH TO SIXTH CENTURY AD): A CHEMICAL
INVESTIGATION*
l. SIU†, J. HENDERSON and E. FABER
Department of Archaeology, University of Nottingham, University Park, Nottingham, UK;
Fifty-seven glass samples from Carthage dating to the fourth to sixth century AD were analysed
using the electron microprobe. The results show that these samples are all soda–lime–silica
glass. Their MgO and K
2
O values, which are below 1.5%, suggest that they were made from
natron, a flux that was widely used during the Roman period. The major and minor elements
show that these samples can be divided into four groups, three of which correspond to the late
Roman period glass groups that were found throughout the Roman Empire: Levantine I, and
‘weak’and ‘strong’HIMT. Of particular interest is our Group 2, which is technologically and
compositionally similar to HIMT glass and the CaO and Al
2
O
3
values of which are similar to
those of Levantine I. Glass of similar composition has been reported by several authors and is
predominantly found dating from the late fifth to seventh century. This could represent a ‘new’
glass group; therefore further study is needed to determine its origin. Also, this study suggests
that the Vandal invasion in North Africa did not disrupt the glass trade between Carthage and
the Levantine coast.
KEYWORDS: LATE ROMAN GLASS, NATRON, HIMT, LEVANTINE I, ELECTRON
MICROPROBE, CARTHAGE
INTRODUCTION
The extensive evidence of glass-working and use dating from the Punic period (eighth to second
century BC) to the eighth century AD from various excavations at Carthage has been key to our
understanding of the glass industry in the Mediterranean and Near East during these periods
(Tatton-Brown 1984, 1994; Funfschilling 1999, 2009, 2010; Foy 2003; Sterrett-Krause 2006,
2009). Sterrett-Krause’s (2006, 2009 studies are among the most important for understanding
the Carthaginian glass industry because they provide evidence of secondary glass-working in
Carthage, including raw glass, malformed vessels, vessels with tool marks, and production waste
such as tooled blobs and blowpipe moils. In particular, 16 free-blown bases that have been dated
to the third or fourth century AD, and are believed to have come from a single workshop, suggest
that a late Roman glass workshop existed in Carthage (Sterrett-Krause 2009, 241–3). These pub-
lications have undoubtedly enriched our knowledge of the glass vessel forms and the industry
from Carthage spanning from its foundation to the late antique period. However, it is still the case
that very little is known about the glass industry of the late Roman and Vandal period, especially
in terms of its organization, the location of the glass workshop(s) (primary and/or secondary) and
the trading of glass between Carthage and the Mediterranean. This paper aims to address this
*Received 7 April 2015; accepted 24 March 2016
†Corresponding author: email acxis2@nottingham.ac.uk
Archaeometry ••,•• (2016) ••–•• doi: 10.1111/arcm.12252
© 2016 University of Oxford
bs_bs_banner
deficiency by using the results of electron microprobe analyses (EPMA–WDS) of glass from
Carthage to address a range of questions:
(1) What are the compositional groups of glass in late Roman Carthage?
(2) Where was the glass made?
(3) Is there evidence of glass mixing?
(4) Can we construct a trade/exchange system for late Roman glass found in Carthage?
(5) Did the Vandal invasion have any detectable effect on the glass compositions found at Carthage?
(6) Can we identify any links between compositional groups of glass and glass vessel forms?
The understanding of the Carthaginian glass industry has undoubtedly benefitted from the
Save Carthage Project, launched in 1972 by the Tunisian government and UNESCO, which in-
vited teams of archaeologists from the United Kingdom, Bulgaria, Canada, Denmark, France,
Germany, the Netherlands, Italy, Poland, Sweden, Tunisia and the United States to excavate, sur-
vey and conserve the archaeological remains in particular areas of Carthage (Hurst and Roskams
1984). As a result, a series of excavation reports have been published on the sites of Carthage.
Tonnes of archaeological materials including glass have been excavated, studied and catalogued,
and have provided valuable insights into the economy and trade of Carthage and the Mediterra-
nean, especially through ceramic evidence, which was the most abundant find and the most du-
rable material to be found in the archaeological excavations.
Although much archaeometric work has been carried out on Iron Age and Roman glass from
Carthage (see, e.g., Eremin et al. 2012 on Iron Age glass beads from Carthage; also, Tom Fenn
has given several conference papers on Roman, Vandal and Byzantine Carthaginian glass, which
are currently unpublished), there have been relatively few publications of analytical work on the
late Roman glass. Freestone (1994, 290) has analysed a few raw glass chunks that were found at
the Circular Harbour and Foy (2003, 86) has published results of the analyses on Carthage glass
from the fifth to sixth century AD. This paper presents the results of EPMA–WDS analysis of 57
samples of vessel glass and raw glass from Carthage dating to the fourth to sixth century AD. The
analysis will provide valuable information on the chemical composition of the glass samples.
This allows for interpretations concerning the raw materials used, their provenance, evidence
for glass trading and the possibility of identifying glass workshop(s) with unique compositions.
ROMAN GLASS–MAKING
Throughout the Roman period, it is known that glass is of a soda–lime–silica type. There is a gen-
eral consensus that glassmakers in this period used a two-ingredient recipe to produce late Ro-
man soda–lime–silica glass; namely, sand, which can contain mineral impurities such as
feldspars, and natron. Although the natron probably derived from the Wadi al-Natrun in Egypt,
other possible sources include Lake Van in Turkey, the salt lakes at al-Jaboul in northern Syria
and a lake near the city of Chalastra, northern Greece (Shortland 2004, 501; Dotsika et al.
2009, 133; Henderson 2013, 52; Ignatiadou et al. 2005, 64–6). The sand sources may also have
contained calcium-rich shell fragments. The concentration of calcium carbonate in sand in the
Bay of Haifa is around 15%, and this level is enough to produce a soda–lime–silica glass with
about 8.4% CaO (Brill 1988, 267; Freestone 2008, 86–7). However, the two-ingredient recipe
claim is disputed in a study by Wedepohl and Baumann (2000, 129) and Freestone (2006, 12;
2008, 91), who have suggested that shells could have been added to glass to provide CaO.
Other sand sources in the Western Mediterranean have also been studied recently, particularly
using isotopic analyses (strontium and neodymium), by Degryse and Schneider (2008, 1998–9)
and Brems et al. (2013a, 222–32; 2013b, 457–61). Degryse and Schneider (2008) found that not
2l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
all locations might have been suitable for glass production after analysing sand deposits from the
Italian peninsula to the French and Spanish coast and those from north-western Europe. Brems
et al. (2012, 2013a, 2013b have suggested that beach sand from between the mouths of the
Basento and the Bradano Rivers in south-eastern Italy, the north-eastern side of the Salentina
peninsula and the western part of the Follonica Gulf, Tuscany are the possible locations for glass
production in the West. However, without any archaeological evidence to support these analyses,
all these suggestions remain conjectural at this stage.
Researchers who have studied the production of late antique currently glass consider that glass
was made in primary production centres (including in the Near East), and that it was exported in
the form of raw chunks or cullet to secondary glass workshops located in the Roman Empire (Free-
stone et al. 2002, 259; Freestone 2003, 111; Fontaine and Foy 2007, 236; Nenna 2007, 125; Free-
stone et al. 2008a, 30–1). Recent discoveries of primary glass production centres in the Near East,
including Bet Eli’ezer (Hadera) (eighthcentury AD), Apollonia (sixth and seventh centuries AD)and
Bet She’arim (early ninth century AD), which are located in close proximity to the Levantine sand
sources suitable for glass-making (Gorin-Rosen 2000, 62), have helped to validate this model. Also,
archaeological evidence for shipwrecks that contain large amounts of raw glass further suggest that
raw glass was produced in a limited number of production centres in the Near East, was broken up
into blocks and was transported along with glass vessels and cullet to secondary glass workshops in
the West (Foy et al. 2000, 51). Examples of such shipwrecks include: the late second or early third
century AD shipwreck at the coast of Provenance, near the archipelago of Embiez in France; the
shipwreck at the bay of Ajaccio in Corsica, dated to the late second or early third century AD;and
the first century AD shipwreck of Mjlet in Croatia (Foy et al. 2000, 52; Fontaine and Foy 2007,
235–42). However, these instances are relatively rare and while most of the shipwrecks do contain
glass vessels, they only represent part of the cargo. Moreover, currently there is no evidence for late
antique shipwrecks transporting raw glass.
There are six main groups of glass that have been chemically identified in the literature from
the first millennium AD: Levantine I (fifth to seventh century AD) and II (seventh to eighth cen-
tury AD), Egypt Groups I (eighth century AD) and II (eighth to ninth century AD), and HIMT 1
and 2 (late fourth to fifth century AD). What they all have in common is that they were first iden-
tified in the Near East, Egypt and North Africa by Bimson and Freestone (1988), Foy (2003, 84–
6), Nenna et al. (2000, 105–11), Picon and Vichy (2003, 23–4), Foy et al. (2003, 45–6) and Free-
stone et al. (2000, 70–2; 2002; 2005; see also Freestone 2003, 19–32). There are numerous pub-
lications concerning the identification of these glass groups in Turkey, Cyprus, Jordan, the UK,
Italy, Germany, Slovenia, Albania and Bulgaria (e.g., Mirti et al. 1993; Degryse et al. 2006; Fos-
ter and Jackson 2009, 197–204; Arletti et al. 2010; Rehren and Cholakova 2010; Rehren et al.
2010; Schibille 2011; Silvestri et al. 2011; Šmit et al. 2014; Gallo et al. 2014).
These glass groups are in agreement with the division of labour model noted above. Lumps of
raw glass belong to Levantine I and II and Egypt I and II were found in the primary glass
production workshops, and glass vessels belonging to these five groups have also been identified
in secondary glass workshops across a wide geographical region.
ANALYTICAL METHODOLOGY
For the purpose of this project, 57 fragments of glass wereselected for electron-probe micro-analysis
(EPMA–WDS). They represent a range of vessel forms and raw glass, including indented bodies, a
fire-rounded everted-rim bowl, flasks, bottles, beakers, jugs, goblets and knock-off rim vessels. They
also include decorations such as linear cut, mould blown, trail and pinched (Table 1). Where
Carthaginian glass under the rule of the Romans 3
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
Table 1 Descriptions of the glass samples from Carthage, together with contextual information
Samples Vessel forms Context Date Colours
Car01 Indented body 1H7 Fourth to early fifth
century
Colourless
(light green tint)
Car02 Indented body 1 M40 Fourth to early fifth
century
Colourless
(yellow–green tint)
Car03 Indented body 1 M41 Fourth to early fifth
century
Colourless
(light green tint)
Car04 Fire-rounded everted-rim
bowl
Unknown 1 a Late fourth to fifth
century
Colourless
(light green tint)
Car05 Fire-rounded everted-rim
bowl
Unknown 1 b Late fourth to fifth
century
Colourless
(light green tint)
Car06 Indented body Unknown 2 a Fourth to early fifth
century
Colourless
(pale green tint)
Car07 Indented body Unknown 2 b Fourth to early fifth
century
Colourless
(pale green tint)
Car08 Body fragment, linear
cut decoration
1R26 Fourth to fifth century Pale green
Car09 Flask base 1 M37 Fourth to fifth century Pale green
Car10 Flask neck 1 M37 Fourth century? Yellow–green
Car11 Strap handle, bottle 2D66 Second to third century Yellow green
Car12 Ribbed mould-blown
body fragment
2F6 Fifth century Colourless
(with green tint)
Car13 Trail-decorated body
fragment
1 M42 Fourth to fifth century Yellow–green
Car14 Raw glass chunk 1 T34 Fifth to sixth century? Green
Car15 Beaker base with
pinched feet
1H13 Late fourth to early fifth
century
Colourless
Car16 Fuel slag 1R12 Unknown Black
Car17 Handle terminal 1H10 Sixth century Green
Car18 Jug handle 1 U6 Sixth century Green
Car19 Tubular rim bowl 1 M4 Late fourth to early fifth
century
Aqua
Car20 Kicked-up base 1 U6 Late fifth to sixth
century
Olive green
Car21 Tubular rim-bowl 1 U6 Late fourth to early fifth
century
Olive green
Car22 Body fragment 2D66 Fifth century? Yellow–green
Car23 Jug? handle 1H8 Fourth century? Green
Car24 Mould-blown ribbed 1H8 Fifth century? Light green
Car25 Pontil knock-off 8D34 Fourth to early fifth
century
Aqua
Car26 Pontil Knock-off 1 N17 Fourth to early fifth
century
Pale green
Car27 Knocked-off rim 1 N17 Fourth to early fifth
century
Green
Car28 Knocked-off rim 1 N17 Fourth to early fifth
century
Yellow–green
Car29 Knocked-off rim 1H7 Fourth to early fifth
century
Pale green
Car30 Knocked-off rim 1H17 Light aqua
(Continues)
4l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
Table 1 (continued)
Samples Vessel forms Context Date Colours
Fourth to early fifth
century
Car31 Knocked-off rim 1 N17 Fourth to early fifth
century
Yellow–green
Car32 Knocked-off rim 1 N17 Fourth to early fifth
century
Yellow–green
Car33 Thick body fragment 1S11 Fourth to early fifth
century
Olive green
Car34 Knocked-off rim 1 N17 Fourth to early fifth
century
Yellow–green
Car35 Knocked-off rim bowl 1 M36 Fourth to early fifth
century
Olive green
Car36 Knocked-off rim bowl 2G45 Fourth to early fifth
century
Colourless
(with green tint)
Car37 Knocked-off rim bowl,
linear cut below rim
1 M4 Fourth to early fifth
century
Pale green
Car38 Knocked-off rim bowl 1 N17 Fourth to early fifth
century
Colourless
(with green tint)
Car39 Goblet base 1 T53 (Cistern
2)
Late fifth century Yellow–green
Car40 Rim of? bowl 1 T53 (Cistern
2)
Late fifth century Colourless
(with purple tint)
Car41 Tubular rim bowl 1 T52 (Cistern
2)
Late fifth century Yellow–green
Car42 Mould-blown ribbed base
fragment of? beaker
1 T51 Late fifth century Yellow–green (with
brown tint)
Car43 In-turned fire-rounded
rim bowl
1 T52 on 1 T11
(C2)
Late fifth century Light green
Car44 Base fragment of? bowl 1 T51 on 1 T11
(C2)
Late fifth century Yellow–green
Car45 Tubular base fragment 1 T51 on 1 T11
(C2)
Late fifth century Yellow–green
Car46 Fire-rounded rim of bowl 1 T51 Late fifth century Yellow–green
Car47 Mould-blown ribbed
bowl base
1 T52 on 1 T11
(C2)
Late fifth century Yellow–green
Car48 Goblet base 1 T50 (Cistern
2)
Late fifth century Yellow–green
Car49 Goblet stem and bowl 1 T53 (Cistern
2)
Late fifth century Yellow–green
Car50 Goblet stem, bowl and
part of base
1 T51 on 1 T11
(C2)
Late fifth century Yellow–green
Car51 Goblet base 1 T50 (Cistern
2)
Late fifth century Yellow–green
Car52 Tessera 2D28 Third to fourth century? Black
Car53 Fire-rounded rim of bowl 8E11 Fifth century Blue
Car54 Raw glass chunk I52 Unknown Yellow–green
Car55 Raw glass chunk I52 Unknown Yellow–green
Car56 Raw glass chunk IN24 Unknown Green
Car57 Raw glass chunk IHN Unknown Green
Carthaginian glass under the rule of the Romans 5
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
possible, the samples were selected according to diagnostic vessel forms, with 11 samples deriving
from the late fifth century AD Cistern 2 (1T50–53) in Area 2CC. The glass samples include represen-
tative examples of the different colours present in the assemblage: the majority are pale green,
yellow–green and olive green, characteristic of the late Roman period, while rarer colours include
aqua, blue and ‘black’(very dark green) (Table 1). The majority of the samples exhibited evidence
of weathering due to their prolonged contact with soil and groundwater; thus most of the samples
were covered with an iridescent, flaky coating (Salviulo et al. 2004, 296).
The glass samples were mounted in cold-setting epoxy resin, and then were grounded and
polished using standard sample preparation procedures down to a 0.02 μmfinal polishing solu-
tion. The samples were coated with a thin film of carbon prior to analysis, to allow the conduction
of the electron beam. The polished samples were analysed by EPMA–WDS, using a JEOL JXA-
8200 electron microprobe in the Department of Archaeology, University of Nottingham.
Quantitative compositional analyses were carried out using the following analytical set-up:
20 kV accelerating voltage, 50 nA beam current and a 50 μm defocused beam. The counting
times were 30 s on the peak and 15 s on the background to either side of the peak. A defocused
beam is used to reduce the effect of the migration of alkalis within the samples (Henderson 1988,
78–9). The EPMA–WDS was calibrated against a combination of certified standard reference
materials, including minerals (orthoclase, jadeite, pyrite, wollastonite and MgO), pure metals
(Mn, Ti, Cu, Ag, V, Sb, Zn, Sn, Ni, Co., Cr and Zr) and synthetic standards (PbTe, GaP, InAs,
KCl, BaF and SrF). The compositions of 26 elements were sought and were expressed as weight
percentages. Three areas of interest (at ×1000) were analysed in each sample and the mean and
standard deviation were calculated (Appendix A). Some elemental compounds (Ag
2
O, SnO
2
,
NiO, As
2
O
5
, BaO, CoO, V
2
O
5
,Cr
2
O
3
and SrO) were below the minimum detection limit of
the microprobe under the analysed conditions and their values are therefore not reported.
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 during the analytical run (at the start and end of each sample set, and between each
sample set; for the results, see Table 2).
RESULTS
The EPMA results are given in Appendix A. They show that the glasses are all of the soda–lime–
silica type except for Car16 and Car52, which have a mixed alkali composition and will not be
discussed further. The majority of the glass samples have MgO and K
2
O compositions of less
than 1.5%. This suggests that natron was the primary alkali flux for these glasses (Henderson
2013, 51). Four samples, Car01, Car26, Car29 and Car30, have slightly elevated levels of
K
2
O, at above 1%. This is due to the prolonged heating of glasses in furnaces over a long period
of time, which led to the alkali-rich waste gases modifying the composition of the glass slightly,
resulting in increased concentrations of alkali, particularly potash (Paynter 2008, 280–1; Rehren
et al. 2010, 75). All of the glasses in this study were produced using sand as their silica source. It
may contain mineral impurities, including iron-bearing minerals and potassium oxide, feldspars
(introducing alumina) and titinite or sphene (introducing titanium oxide) (Henderson 2000, 27).
Four compositional groups have been identified in this study, using the major and minor ele-
ments. Their average compositions and compositions are presented in Table 3 and Appendix A
respectively:
•Group 1: Group 1 (n=10) is generally dated to the fourth to fifth century AD. It is characterized
by low Al
2
O
3
(av. 2.12%) and CaO (av. 6.49%) levels. It is also noted that this glass group has
6l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
Table 2 The results of the analysis of the Corning Glass Standard B
Na
2
O CuO K
2
O ZnO SiO
2
CaO Al
2
O
3
FeO TiO
2
Cl MgO PbO MnO SO
3
P
2
O
5
Known 17.26 2.7 1.1 0.2 61.55 8.71 4.22 0.35 0.1 0.2 1.19 0.4 0.23 0.54 0.84
Measured (n= 8) 17.24 2.74 1.07 0.2 60.32 8.71 4.26 0.29 0.07 0.17 1.12 0.4 0.23 0.59 0.71
SD 0.1 0.03 0.01 0.01 0.57 0.1 0.1 0.01 0.01 0.004 0.01 0.02 0.01 0.02 0.05
Error (%) 0.12 1.48 2.73 0 2 0 0.95 6.45 30 15 5.88 0 0 9.26 15.48
Detection limits 0.03 0.02 0.01 0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.01 0.03 0.01 0.01 0.01
Carthaginian glass under the rule of the Romans 7
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
elevated levels of FeO (av. 0.62%), TiO
2
(av. 0.09%) and MnO (1.30%). The MnO in the glass
does not appear to have been present as an impurity. It is likely that manganese was introduced
intentionally in the glass to counteract the iron content (Henderson et al. 2004, 461). This
group is also characterized by a high MgO level of 0.89% on average, and a relatively high
soda level of 19.08% on average.
•Group 2: In terms of Group 2 (n= 11), the majority of the glasses are from the contexts of
Cistern 2 (Table 1). They are mainly dated to the late fifth century AD, but Car23 is dated to
the fourth century AD. A piece of raw glass (Car56) belongs to this group. It is characterized
by higher levels of CaO (av. 8.12%) and Al
2
O
3
(av. 2.64%) than Group 1. Like Group 1, it also
contains elevated levels of FeO (av. 0.95%), TiO
2
(av. 0.16%) and MnO (av. 1.57%): they are
generally higher than those observed in Group 1 (Table 3 and Appendix A). Like Group 1, MnO
was intentionally added to the glass to counteract the undesired colours imparted by the high
iron content. Also, its K
2
O (av. 0.71%) is significantly higher than for Groups 1 and 4.
•Group 3: For Group 3 (n=15), the glasses from this group are dated to the fourth to fifth century
AD.Itissignificantly different from Groups 1 and 2. It is characterized by high CaO (av. 8.55%)
and Al
2
O
3
(av. 3.0%). It has no significant elevated levels of FeO (av. 0.35%) and TiO
2
(av.
0.03%), but its high MnO level (av. 0.82%) may suggest that MnO was intentionally added to
counteract the iron content in the glass. Moreover, its soda (av. 16.07%) and magnesium (av.
0.60%) levels are significantly lower than for other groups in this study (Table 3).
•Group 4: Group 4 (n= 17) contains glasses that are dated to the fourth and fifth centuries AD.
Only Car17 and Car20 are dated to the late fifth to sixth century AD. This group is characterized
by a high Al
2
O
3
(av. 2.99%) concentration, but has the lowest CaO (av. 5.87%) content among
the Carthage groups. It also has a high concentration of soda (av. 19.27%) and the highest con-
centration of MgO, at 1.23%, out of all glass groups. Like Groups 1 and 2, it has elevated
levels of FeO (av. 1.66%), TiO
2
(av. 0.49%) and MnO (av. 1.85%), all of which are signifi-
cantly higher than for all other groups in this study.
Table 3 The mean compositions of each group that has been identified in this study, as weight percentages
Group 1,n=10 Group 2,n=11 Group 3,n=15 Group 4,n=17
Na
2
O 19.11% 0.53 18.93% 0.82 16.07% 1.19 19.27% 1.44
CuO 0.05% 0.05 0.08% 0.08 0.08% 0.06 0.07% 0.06
K
2
O 0.50% 0.08 0.71% 0.09 0.86% 0.25 0.50% 0.13
ZnO 0.02% 0.01 0.02% 0.01 0.02% 0.01 0.02% 0.01
SiO
2
65.72% 1.51 64.24% 1.30 66.93% 0.98 64.32% 1.47
CaO 6.49% 0.85 8.12% 0.59 8.55% 0.57 5.87% 0.61
Al
2
O
3
2.12% 0.11 2.64% 0.16 3.0% 0.13 2.99% 0.20
FeO 0.62% 0.10 0.95% 0.23 0.35% 0.05 1.66% 0.35
TiO
2
0.09% 0.02 0.16% 0.10 0.03% 0.01 0.49% 0.10
Cl 1.03% 0.21 0.86% 0.06 0.80% 0.15 0.99% 0.14
MgO 0.89% 0.17 1.24% 0.14 0.60% 0.06 1.27% 0.20
Sb
2
O
5
0.03% 0.01 0.06% 0.04 0.05% 0.02 0.02% 0.01
MnO 1.30% 0.53 1.57% 0.26 0.82% 0.30 1.89% 0.23
SO
3
0.33% 0.12 0.35% 0.06 0.24% 0.07 0.27% 0.05
P
2
O
5
0.04% 0.02 0.10% 0.03 0.09% 0.02 0.05% 0.02
ZrO
2
0.01% 0.01 0.01% 0.01 bdl* 0.03% 0.01
*bdl, Below detection limit.
8l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
A number of glass samples have been omitted from the following discussion due to their sig-
nificantly different chemical compositions. Samples Car15 and Car53 have chemical composi-
tions similar to those of early Roman glass. Car15 has a higher level of antimony pentoxide
than manganese oxide, which suggests that antimony, a decolourizer that was used in the early
Roman period (Jackson 2005, 769, Table 2), was added to make a colourless glass; Car53 has
a composition similar to that of the third century Roman glass from the Iulia Felix, and has high
iron and copper oxide contents due to their addition in a blue colourant (Jackson 2005, 769,
Table 2; Silvestri 2008, 1494).
Car16 and Car52 have unusually low Na
2
O (6.77% and 4.48%) and K
2
O (4.15% and 4.63%)
contents, which could point to mixed-alkali glasses. But their high contents of Al
2
O
3
(7.78% and
12.69%) and FeO (7.33% and 2.27%) and low levels of CaO (0.25% and 0.76%) suggest that
they are fuel slag and recycled tesserae (Appendix A).
DISCUSSION
The comparison of the major and minor elements of the Carthage glass assemblage with glasses
dated to the late Roman/early medieval period suggests that they belong to different glass groups
that were produced in the Mediterranean during this period.
It shows that Groups 1 and 4 are consistent with HIMT glasses (Fig. 1), which have a wide
range of Al
2
O
3
concentrations, between 1.88% and 3.17%, and CaO between 5.54% and
Figure 1 A biplot of CaO versus Al
2
O
3
(wt%), showing the Carthage assemblage and its relation to HIMT and Levan-
tine I glasses. Data for HIMT glasses: Maroni Petrera Group 2, Cyprus, Freestone et al. (2002); Butrint HIMT 1 and 2,
Albania, Conte et al. (2014); Galeata, Italy, Arletti et al. (2010); Augusta Praetoria HIMT, Italy, Mirti et al. (1993); Brit-
ish HIMT 1 and 2, Foster and Jackson (2009); Carthage, Tunisia, Freestone(1994). Data for Levantine I glasses: Maroni
Petrera Group 1, Cyprus, Freestone et al. (2002); Butrint Levantine I, Albania, Conte et al. (2014); Apollonia, Israel,
Freestone et al. (2008b) and Tal et al. (2004); British Levantine Ia and Ib, Foster and Jackson (2009); Petra, Jordan,
Schibille et al. (2008).
Carthaginian glass under the rule of the Romans 9
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
7.83%. It is observed that both groups have a high concentration of Na
2
O (av. 19.0%), which is
characteristic of HIMT glasses (Fig. 2) (Foster and Jackson 2009, 192). Moreover, HIMT glasses
have a high concentration of MgO (>0.98%), which was observed by Foster and Jackson (2009,
192) and is confirmed in this study (Table 3). Using the compositional data for FeO, TiO
2
, MnO
and Al
2
O
3
, they show that the Carthage HIMT has high concentrations of these elements
(Table 3). Moreover, there is a strong positive correlation between Al
2
O
3
, FeO and TiO
2
in
Groups 1 and 4 (Figs 3 and 4). All of these are the key characteristics of HIMT glass and are re-
ported by Conte et al. (2014), Foy et al. (2003), Freestone (1994, 2003, 2005, Freestone et al.
(2002, 2005, Foster and Jackson (2009), Gallo et al. (2014), Rehren and Cholakova (2010)
and Silvestri et al. (2011) in their studies of HIMT glasses.
On the basis of FeO, TiO
2
and MnO, it is also possible to divide the Carthage HIMT group
into HIMT 1 and HIMT 2. The Group 1 glasses have lower FeO, TiO
2
and MnO concentrations
than glasses from Group 4 (Table 3). Figures 3 and 4 show that Group 1 falls into the ‘weak’
HIMT (or HIMT 2) group, which includes the British HIMT 2, HIMT from Galeata in Italy,
two HIMT glasses from Maroni Petrera in Cyprus, and Butrint HIMT 2 from Albania.
On the other hand, the Group 4 glasses have higher average concentrations of FeO, TiO
2
and
MnO than Group 1 (Table 3). It displays a wider compositional spread than Group 1 and there is
a very strong positive correlation between Al
2
O
3
, FeO and TiO
2
(Figs 3 and 4). It falls into the
‘strong’HIMT (or HIMT 1) group, which includes HIMT 1 glasses from Britain, Butrint HIMT 1
and the majority of the Carthage glass analysed by Freestone (1994). It is also associated with the
higher-iron members of the HIMT group, such as those from Dichin (Rehren and Cholakova
2010).
So far, there is no archaeological evidence of a primary production centre(s) that produced
HIMT glasses. But Nenna (2014, 187) has suggested that HIMT glass was probably produced
using the Nilotic sands in Egypt. The levels of titanium, iron and magnesium in Groupes 1
Figure 2 A biplot of Na
2
O versus CaO (wt%) for the Carthage assemblage and relevant HIMT and Levantine I data.
10 l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
and 2 analysed by Foy et al. (2003) are similar to the levels found in the Nilotic sands in Egypt.
Therefore it is possible that this sand was used to manufacture Foy et al.’sGroupes 1 and 2. On
the other hand, the elevated levels of manganese oxide found in the glass were thought to be in-
troduced to the glass deliberately, as the Nilotic sands in Egypt have lower levels of manganese
Figure 3 A biplot of FeO versus Al
2
O
3
(wt%) for the Carthage assemblage and HIMT and Levantine data from the
literature.
Figure 4 A biplot of TiO
2
versus FeO (wt%) for the Carthage glasses and HIMT and Levantine I data.
Carthaginian glass under the rule of the Romans 11
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
than the glass samples. Nenna (2014, 188–9) further speculates that the primary production cen-
tres for HIMT glass were located in northern Sinai, especially at the sites of Ostrakine
(Ostrakine) and Pelusium (Pelusio) and its surrounding region.
Group 3 resembles the glasses from the Levantine I group, which show a homogeneous com-
position and have a relatively higher concentration of CaO than HIMT glasses (Fig. 1). It is also
clear from Figure 2 that the low Na
2
O concentration of Group 3 separates it from the HIMT
glasses. This group has low FeO, TiO
2
and MnO concentrations and, unlike the HIMT glasses,
shows no correlation between FeO, Al
2
O
3
and TiO
2
(Figs 3 and 4). The titanium level in this
glass group is consistent with the levels found in Roman and late Roman glasses that were
manufactured in the Syro-Palestinian region, which are below 0.10% (Nenna 2014, 180). Exca-
vations of primary glass production centres in Apollonia and Dor have yielded large amounts of
Levantine I raw glasses, which confirm that this glass group originated from and was produced in
the Syro-Palestinian region.
The results also indicate that manganese oxide in the Carthage Levantine I group (except for
two samples, Car19 and Car25) was being added intentionally, since the typical maximum impu-
rity level for manganese oxide is about 0.4% (Brill 1988, 258–9). Examples of manganese being
added to Levantine I glass have been reported by Brill (1988, 258–62) and Foster and Jackson
(2009, 193).
On the other hand, Group 2 belongs to neither the HIMT nor the Levantine I glass groups.
Figure 2 shows that it has a CaO concentration very similar to that of the Levantine I group
and a Na
2
O concentration very similar to that of the HIMT group, which marks it out as a distinc-
tive glass group. It is shown in Figures 5 and 6 that this group is not associated with Groups 1, 3
and 4 based on the values of iron, aluminium and titanium oxides. Also, there is no significant
correlation between these elements in Group 2.
Figure 5 A biplot of FeO versus Al
2
O
3
(wt%) for the Carthage groups and the post-Roman glasses (Anglo-Saxon Period
I, Freestone et al. 2008a; Vicq, Velde 1990; WD2 from Butrint, Schibille 2011; Crypta Balbi, Mirti et al. 2000).
12 l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
Such a phenomenon is reported by Velde (1990, 110) in his study of Merovingian (sixth to
seventh century AD) glass from Vicq in France, Freestone et al. (2008a, 32–7) in Anglo-Saxon
Period I glass from Britain, Schibille (2011, 2945–6) in WD2 glass from Butrint and Mirti
et al. (2000, 364–5) in seventh century AD glasses from Crypta Balbi in Rome. Figures 5 and
6 show that Group 2 is associated closely with the Vicq, Anglo-Saxon Period I, Butrint and
Crypta Balbi samples in terms of its Al
2
O
3
, FeO and TiO
2
levels.
Could Group 2 result from the mixing of HIMT and Levantine I glasses? Other studies have
suggested that recycling did occur for this group, though not all samples have been found to
be recycled, so this cannot be the sole explanation (Freestone et al. 2008a, 36–7; Schibille
2011, 2945). Unfortunately the lead, copper and tin oxides that could help to suggest that
recycling had occurred are not high enough. Trace element analysis would help to resolve this
issue. Therefore it is difficult to tell whether they have been recycled or not. But from their major
and minor element concentrations, we can conjecture that Group 2 could not have been the result
of the mixing of HIMT and Levantine I glasses, since the group would overlap with the Levan-
tine I and HIMT groups more: as we can see from the above, there are minimal overlaps.
Colouring agents
The majority of the glass vessels are green, yellow–green or olive green, with a few aqua, blue
and ‘colourless’samples; three are green and two others are yellow–green. In terms of the green,
yellow–green, olive green and aqua colours, iron, an impurity in the sand source, is the colourant
responsible. Iron can exist in two oxidation (ionic) states in glass: ferrous (Fe
2+
) and ferric (Fe
3+
)
ions. The aqua colour in Car19, 25 and 30 is due to the glasses being melted under reducing con-
ditions, and the Fe
2+
ion is therefore dominant. As Brill (1988, 273) has pointed out, most of the
late Roman green-coloured glasses could be the result of a mixture of ferrous blue with traces of
the ferrous sulphide amber chromophore helping to dilute the aqua colour with yellow, producing
Figure 6 A biplot of TiO
2
versus FeO (wt%) for the Carthage groups and the post-Roman glasses. Group 2 is clearly
associated with post-Roman glasses that have limited correlation between these two elemental compounds.
Carthaginian glass under the rule of the Romans 13
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
the green colour that we see in the majority of the Carthage glasses. On the other hand, the olive
green colour is due to a combination of reduced Fe
2+
and a higher concentration of ferri-sulphide
(Brill 1988, 269). For the yellow–green-coloured glasses and the ‘colourless’glass, high levels of
manganese are always found, suggesting that it was added intentionally to counteract the iron.
However, it is known that in order for manganese to be really effective as a decolourant in glass,
the MnO/Fe ratio needs to be higher than 2.2:1 (Silvestri et al. 2011, 3412). It is noticeable that
green tints can be found in the ‘colourless’Carthage glass because the ratio was not as high as
2.2:1. The manganese helps to oxidize the iron, lowering the concentration of the strongly
colouring ferrous ion and replacing it with the weaker ferric colourant, and modifying the aqua
to a paler yellowish green colour (Brill 1988, 276–7).
The blue-coloured Car53 has a high concentration of iron (1.32%), at a similar level to HIMT
glass, but also a high concentration of CuO (0.35%) and a very low concentration of MnO
(0.37%). This suggests that MnO was not intentionally added to counteract the effect of the iron
but that it is an impurity; its base composition corresponds to Levantine I glass. Copper was prob-
ably added deliberately to produce the blue colour.
Compositional groups and typology
Five different diagnostic vessel types have been sampled in this study: the indented vessels
(n= 5), dated to the fourth to early fifth century AD; the fire-rounded rims of bowls (n= 4), dated
to the late fourth to late fifth century AD, goblets (n= 5) dated to the late fifth century AD; ribbed
mould-blown vessels (n= 4), dated to the fifth century AD; and vessels with knocked-off rims
(n= 13), dated to the fourth to early fifth century AD. It is interesting to investigate whether glass-
workers in late antiquity used the same compositions of glass to produce the same vessel type.
The results suggest that the indented vessels belong to Groups 1, 3 and 4. Data for the knock-
off rim vessels indicate that all apart from Car36 belong to Groups 3 and 4 exclusively. On the
other hand, all fire-rounded rim bowls belong to Groups 2 and 3, all goblets belong to Group
2 and all ribbed mould-blown vessels belong to Groups 1, 2 and 4. Therefore, this shows that
the fourth to early fifth century AD vessel forms are made with Levantine I and ‘weak’and
‘strong’HIMT compositions. After the early fifth century AD, the quantity of ‘weak’HIMT
and Levantine I glasses decreased. On the other hand, Groups 2 and 4 dominate the majority
of the assemblage from the late fifth to sixth century AD. This suggests that during the fourth
to early fifth century AD, both glassmakers and consumers in Carthage favoured Levantine I,
‘weak and strong’HIMT glasses. However, after the early fifth century AD, consumers in Car-
thage favoured Group 2 and ‘strong’HIMT products. This can be seen in Britain, where con-
sumers favoured the stronger British HIMT 1 over the weaker British HIMT 2 and Levantine I
glasses from the fifth century AD onwards (Foster and Jackson 2009, 194–5). However, it is cur-
rently unknown why ‘strong’HIMT and Group 2 fared better than other glass compositional
groups in the fifth century AD. It has been suggested that this could due to a change in fashion,
or to the fact that it was cheaper to make HIMT and Group 2 in large quantities using lower-grade
raw materials than for Levantine I glass (Foster and Jackson 2009, 195). With further scientific
research of the selected vessel forms, a more rigorous statistical assessment could be used.
CONCLUSION
Based on the results, four compositional groups of glass have been identified: Levantine I (Group
3), ‘weak’HIMT (Group 1), ‘strong’HIMT (Group 4) and Group 2. Their compositional
14 l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
characteristics suggest that different sand sources have been used to make these glasses. The
Levantine I glass was made with Belus River (or similar) sand. Group 3 has lower titanium oxide
levels when compared to other known Levantine I glasses, suggesting that a different sand source
within the Palestinian region had been used. The HIMT groups were probably made using
Egyptian sand, as discussed by Freestone et al. (2005, 155–6) and Nenna (2014, 187). For Group
2, the glassmakers may have used a sand source located between the Palestinian and Egyptian
regions. It is technologically or geochemically related to HIMT (or both), and to some extent re-
sembles Levantine I, especially in its lime content. The Carthage Group 2 is similar to groups de-
fined by Freestone et al. (2008a, 32–7), Mirti et al. (2000, 364–5), Schibille (2011, 2945–6) and
Velde (1990, 110). Yet it is still difficult to determine its origin and the extent of its circulation in
late antiquity. It appears to have been produced between the fifth and ninth centuries AD.
In terms of its distribution, it has been mainly identified in the western half of the Roman Em-
pire, and it has now been identified in Carthage. It is yet to be discovered in many parts of the
Eastern Mediterranean, but such a wide geographical distribution suggests that this group could
have been widely traded throughout Western Europe (Freestone et al. 2008a, 36).
In terms of its origin, no primary or secondary glass production sites have been found for this
glass group. Freestone et al. (2008a) performed trace element analysis on the Anglo-Saxon
Period I samples and found that they show trace elemental compositions similar to those of Le-
vantine I, Levantine II and HIMT glasses. Therefore it is possible that this group might have
come from somewhere between the Nile in Egypt and Haifa in Israel, since it displays mineral
assemblages similar to those of the sands that come from the Nile, which are moved up the east-
ern coast of the Mediterranean by longshore drift and tidal currents (Freestone et al. 2008a, 36–7,
fig. 9).
Using our data set, it appears that minimal glass recycling had occurred in the late antique pe-
riod. These four groups did not overlap with each other: there was a clear distinction between the
four in terms of their major elements. Although other studies have shown that Group 2 could be
the result of mixing HIMT glass with Levantine I glass, the present study shows that there was
minimal mixing. If it had happened, Group 2 would have overlapped with HIMT and Levantine
I to a greater extent. Further trace element analysis is needed to determine whether recycling oc-
curred extensively in Carthaginian glass.
Therefore, the chemical analysis of the Carthage glass may suggest that Carthage was
importing raw glass from the Palestinian region and using glass made in Egypt. The glass may
have been made into vessels in secondary glass workshops located in Carthage. Archaeological
evidence, such as working debris from Carthage, Area 2CC and the study by Sterrett-Krause
(2009), who studied tooled blobs and blowpipe moils from Carthage (Sterrett-Krause 2009,
242), suggests that secondary glass production occurred there, including in the Vandal period.
There is still a question about where the workshop(s) were located. The identification of four
glass groups suggests that under the rule of the Vandals, Carthage was still receiving imports
from different primary glass production centres in the East, probably along a similar route to wine
imports. Glass from the East evidently remained popular under Vandal rule. Therefore, the tradi-
tional view that the Vandals had brought chaos to Carthage and North Africa is not entirely true
(Leone 2007, 283; Christie 2012, 1).
Glass production centres in the Eastern Mediterranean may have been competing with each
other in late antiquity, as shown by the presence of different glass groups at Carthage. This prop-
osition has been discussed by Freestone et al. (2002, 173), Foster and Jackson (2009, 194–6) and
Nenna (2014, 186). The possible reasons for this competition are: (a) aesthetics in terms of col-
ours—HIMT is yellow–green while Levantine I has an aqua shade; and (b)—technologically
Carthaginian glass under the rule of the Romans 15
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
related—HIMT used low-quality sands and would only require relatively low temperatures to
melt it due to its higher soda content. Therefore it was more easily produced in volume using un-
specialized raw materials and less sophisticated technology. Levantine I glass required higher
temperatures, which made it less economic and more costly to produce. However, the extent
of the competition is not clear. The present study suggests that HIMT was more popular than
the Levantine I glass, as shown by its use in a number of archaeological assemblages. However,
because this study has only analysed 57 samples of Carthage glass, a larger number of samples is
required to prove that the population of Carthage in late antiquity favoured HIMT glass. But
existing research also suggests that HIMT glass proved to be more popular. For instance 66%
of the glass from Britain dated to AD 400–550 analysed by Foster and Jackson (2009) belongs
to the HIMT group (Nenna 2014, 186).
A limitation of this study is that it is still uncertain whether Carthage had exported any of its
glass products to other regions in the Mediterranean or Europe. The major and minor elemental
analysis used here has suggested that some of these glasses came from the Eastern Mediterra-
nean, and there is no archaeological or chemical evidence that suggests that Carthage produced
its own glass type under the Roman and Vandal rules. In the future, trace element and isotope
analyses could be used in order to investigate the extent of recycling and provenance.
REFERENCES
Arletti, R., Vezzalini, G., Benati, S., Mazzeo Saracino, L., and Gamberini, A., 2010, Roman window glass: a comparison
of findings from three different Italian sites, Archaeometry,52, 252–71.
Bimson, M., and Freestone, I. C., 1988, The discovery of an Islamic glass-making site in middle Egypt, in Annales du 10
e
Congrès de l’Association Internationale pour l’Historie du Verre, Madrid–Segovie, 1985, 237–43, AIHV,
Amsterdam.
Brems, D., Degryse, P., Hasendoncks, F., Gimeno, D., Silvestri, A., Vassilieva, E., Luypaers, S., and Honings, J., 2012,
Western Mediterranean sand deposits as raw material for Roman glass production, Journal of Archaeological Science,
39, 2897–907.
Brems, D., Ganio, M., Latruwe, K., Balcaen, L., Carremans, M., Gimeno, D., Silvestri, A., Vanhaecke, F., Muchez, P.,
and Degryse, P., 2013a, Isotopes on the beach, part 1: strontium isotope ratios as a provenance indicator for lime
raw materials used in Roman glass-making, Archaeometry,55, 214–34.
Brems, D., Ganio, M., Latruwe, K., Balcaen, L., Carremans, M., Gimeno, D., Silvestri, A., Vanhaecke, F., Muchez, P.,
and Degryse, P., 2013b, Isotopes on the beach, part 2: neodymium isotopic analysis for the provenancing of Roman
glass-making, Archaeometry,55, 449–64.
Brill, R. H., 1988, Scientific investigations of the Jalame glass and related finds, in Excavations at Jalame: site of a glass
factory in late Roman Palestine (ed. G. D. Weinberg), 257–94, University of Missouri Press, Columbia, MO.
Christie, N., 2012, The fall of the western Roman Empire: an archaeological and historical perspective, Bloomsbury
Academic, London.
Conte, S., Chinni, T., Arletti, R., and Vandini, M., 2014, Butrint (Albania) between Eastern and Western Mediterranean
glass production: EMPA and LA–ICP–MS of late antique and early medieval finds, Journal of Archaeological
Science,49,6–20.
Degryse, P., and Schneider, J., 2008, Pliny the Elder and Sr.–Nd isotopes: tracing the provenance of raw materials for
Roman glass production, Journal of Archaeological Science,35, 1993–2000.
Degryse, P., Schneider, J., Haack, U., Lauwers, V., Poblome, J., Waelkens, M., and Muchez, P., 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.
Dotsika, E., Poutoukis, D., Tzavidopoulos, I., Maniatis, Y., Ignatiadou, D., and Raco, B., 2009, A natron source at
Pikrolimni Lake in Greece? Geochemical evidence, Journal of Geochemical Exploration,103, 133–43.
Eremin, K., Degryse, P., and Erb-Satullo, N., 2012, Iron Age glass beads from Carthage, in Historical technology,
materials and conservation: SEM and microanalysis (eds. N. Meeks, C. Cartwright, A. Meek, and A. Mongiatti),
30–5, Archetype, London.
16 l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
Fontaine, S. D., and Foy, D., 2007, L’épave Ouest-Embiez 1, Var: le commerce maritime du verre brut et manufacturé en
Méditerranée occidentale dans l’Antiquité, Revue archéologique de Narbonnaise,40, 235–65.
Foster, H. E., and Jackson, C. M., 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.
Foy, D., 2003, Le verre en Tunisie: l’apport des fouilles recentes tuniso-françaises, Journal of Glass Studies,45,59–89.
Foy, D., Vichy, M., and Picon, M., 2000, Lingots de verre en Mediterranee occidentale (III
e
siècle av. J.-C. –VII
e
siècle
ap. J.-C.), in Annales du 14
e
Congrès de l’Association Internationale pour l’Historie du Verre, Italia/Venezia–Milano,
27 October –1 November 1998,51–7, AIHV, Lochem.
Foy, D., Picon, M., Vichy, M., and Thirion-Merle, V., 2003, Caracterisation des verre de la findel’Antique en
Mediterranee occidentale: l’emergence de nouveaux courants commerciaux, in Échanges et commerce du verre dans
le monde antique (eds. D. Foy and M.-D. Nenna), 41–85, Editions Monique Mergoil, Montagnac.
Freestone, I. C., 1994, Chemical analysis of ‘raw’glass fragments, in Excavations atCarthage, volume II, part 1. The Circular
Harbour, north side: the site and finds other than pottery (ed. H. R. Hurst), 290, The British Academy, Oxford.
Freestone, I. C., 2003, Primary glass sources in the mid first millennium AD,inAnnales du 15
e
Congrès de l’Association
Internationale pour l’Histoire du Verre, New York –Corning, 15 –20 October 2001, 111–5, AIHV, Nottingham.
Freestone, I. C., 2005, The provenance of ancient glass through compositional analysis, in Materials issues in art and
archaeology VII (eds. P. B. Vandiver, J. L. Mass, and A. Murray), 195–208, Materials Research Society Symposium
Proceedings Volume 852, Materials Research Society, Warrendale, PA.
Freestone, I. C., 2006, Glass production in late antiquity and the early Islamic period: a geochemical perspective, in Geomaterials
in cultural heritage (eds. M. Maggetti and B. Messiga), 201–16, Special Publication 257, Geological Society, London.
Freestone, I. C., 2008, Pliny on Roman glassmaking, in Archaeology, history and science: integrating approaches to
ancient materials (eds. M. Martinon-Torres and T. Rehren), 77–100, Left Coast Press, Walnut Creek, CA.
Freestone, I. C., Gorin-Rosen, Y., and Hughes, M. J., 2000, Primary glass from Israel and the production of glass in late
antiquity and the early Islamic period, in La route du verre: ateliers primaires et secondaires du second millénaire av.
J.-C. au Moyen Âge (ed. M.-D. Nenna), 65–84, Maison de l’Orient Mediterraneen –Jean Pouilloux, Lyon.
Freestone, I. C., Hughes, M. J., and Stapleton, C. P., 2008a, The composition and production of Anglo-Saxon glass, in
Catalogue of Anglo-Saxon glass in the British Museum (eds. V. I. Evison and S. Marzinzik), 29–46, Research
Publication 167, British Museum, London.
Freestone, I. C., Jackson-Tal, R. E., and Tal, O., 2008b, Raw glass and the production of glass vessels at late Byzantine
Apollonia–Arsuf, Israel, Journal of Glass Studies,50,67–95.
Freestone, I. C., Ponting, M., and Hughes, M. J., 2002, The origins of Byzantine glass from Maroni Petrera, Cyprus,
Archaeometry,44, 257–72.
Freestone, I. C., Wolf, S., and Thirlwall, M., 2005, The production of HIMT glass: elemental and isotopic evidence, in
Annales du 16
e
Congrès de l’Association Internationale pour l’Histoire du Verre, London, 7–13 September 2003,
153–7, AIHV, Nottingham.
Funfschilling, S., 1999, Gläser aus den Grabungen des Deutschen Archäologischen Instituts in Karthago. Die Grabungen
‘Quartier Magon’und Rue Ibn Chabaat sowie kleinere Sondagen, in Die deutschen Ausgrabungen in Karthago
(Karthago III) (ed. F. Rakob), 435–529, Verlag Phillip von Zabern, Mainz.
Funfschilling, S., 2009, Special relationship between the glass finds from Carthage and Rome and comparisons with finds
north of the Alps, in Annales du 17
e
Congrès de l’Association Internationale pour l’Histoire du Verre, Anvers, 2006
(eds. K. Janssens, P. Degryse, P. Cosyns, J. Caen, and L. Van’t dack), 143–9, University Press Antwerp, Antwerp.
Funfschilling, S., 2010, Byzantine glass finds from Carthage: from the German excavations, the English excavations at
Bir Messaouda and from the Canadian excavations, in Glass in Byzantium—production, usage, analyses (eds. J.
Drauschke and D. Keller), 215–24, RGZM Tagungen 8, Mainz.
Gallo, F., Marcante, A., Silvestri, A., and Molin, G., 2014, The glass of the ‘Cassa delle Bestie Ferrite’:afirst systematic
archaeometric study on late Roman vessels from Aquileia, Journal of Archaeological Science,41,7–20.
Gorin-Rosen, Y., 2000, The ancient glass industry in Israel: summary of the finds and new discoveries, in La route du
verre: ateliers primaires et secondaires du second millénaire av. J.-C. au Moyen Âge (ed. M.-D. Nenna), 49–64,
Maison de l’Orient Mediterraneen –Jean Pouilloux, Lyon.
Henderson, J., 1988, Electron probe microanalysis of mixed-alkali glasses, Archaeometry,30,77–91.
Henderson, J., 2000, The science and archaeology of materials: an investigation of inorganic materials, Routledge,
London.
Henderson, J., 2013, Ancient glass: an interdisciplinary exploration, Cambridge University Press, Cambridge.
Henderson, J., McLoughlin, S. D., and McPhail, D. S., 2004, Radical changes in Islamic glass technology: evidence for
conservatism and experimentation with new glass recipes from early and middle Islamic Raqqa, Syria, Archaeometry,
46, 430–68.
Carthaginian glass under the rule of the Romans 17
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
Hurst, H. R., and Roskams, S. P., 1984, Excavations at Carthage: the British mission, volume I, part I, in The Avenue du
President Habib Bourguiba, Salammbo: the site and finds other than pottery, H, Huddersfield, Charlesworth.
Ignatiadou, D., Dotsika, E., Kouras, A., and Maniatis, Y., 2005, Nitrum chalestricum: the natron of Macedonia, in
Annales du 16
e
Congrès de l’Association Internationale pour l’Histoire du Verre, London, 7–13 September 2003,
64–7, AIHV, Nottingham.
Jackson, C., 2005, Making colourless glass in the Roman period, Archaeometry,47, 763–80.
Leone, A., 2007, Changing townscapes in north Africa from late antiquity to the Arab conquest, Edipuglia, Bari.
Meek, A., Henderson, J., and Evans, J., 2012, Isotope analysis of English forest glass from the Weald and Staffordshire,
Journal of Analytical Atomic Spectrometry,27, 786–95.
Mirti, P., Casoli, A., and Appolonia, L., 1993, Scientific analysis of Roman glass from Augusta Praetoria, Archaeometry,
35, 225–40.
Mirti, P., Lepora, A., and Sagui, L., 2000, Scientific analysis of seventh-century glass fragments from the Crypta Balbi in
Rome, Archaeometry,42, 359–74.
Nenna, M.-D., 2007, Production et commerce du verre à l’époque impériale: nouvelles découvertes et problématiques,
Facta: A Journal of Roman Material Culture Studies,1, 125–47.
Nenna, M.-D., 2014, Egyptian glass abroad: HIMT glass and its markets, in Neighbours and successors of Rome: tradi-
tions of glass production and use in Europe and the Middle East in the later first millennium AD (eds. D. Keller, J.
Price, and C. Jackson), 177–93, Oxbow Books, Oxford.
Nenna, M.-D., Picon, M., and Vichy, M., 2000, Ateliers primaires et secondaires en Egypte a l’epoque Greco-Romaine, in
La route du verre: ateliers primaires et secondaires du second millénaire av. J.-C. au Moyen Âge (ed. M.-D. Nenna),
97–112, Maison de l’Orient Mediterraneen –Jean Pouilloux, Lyon.
Paynter, S., 2008, Experiments in the reconstruction of Roman wood-fired glassworking furnaces: waste products and
their formation processes, Journal of Glass Studies,50, 271–90.
Picon, M., and Vichy, M., 2003, D’Orient en occident: l’origine du verre a l’époque romaine et Durant le haut Moyen
Àge, in Échanges et commerce du verre dans le monde antique (eds. D. Foy and M.-D. Nenna), 17–31, Editions
Monique Mergoil, Montagnac.
Rehren, T., and Cholakova, A., 2010, The early Byzantine HIMT glass from Dichin, northern Bulgaria,
Интердисциплинарни Изследвания (Interdisciplinary Studies), XXII–XXIII,81–96.
Rehren, T., Marii, F., Schibille, N., Stanford, L., and Swan, C., 2010, Glass supply and circulation in early Byzantine
southern Jordan, in Glass in Byzantium—production, usage, analyses (eds. J. Drauschke and D. Keller), 65–81,
Romisch-Germanisches Zentralmuseum, Mainz.
Salviulo, G., Silvestri, A., Molin, G., and Bertoncello, R., 2004, An archaeometric study of the bulk and surface
weathering characteristics of early medieval (5th–7th century) glass from the Po valley, northern Italy, Journal of
Archaeological Science,31(3), 295–306.
Schibille, N., 2011, Supply routes and the consumption of glass in first millennium CE Butrint (Albania), Journal of
Archaeological Science,38, 2939–48.
Schibille, N., Marii, F., and Rehren, T., 2008, Characterization and provenance of late antique window glass from the
Petra church in Jordan, Archaeometry,50, 627–42.
Shortland, A. J., 2004, Evaporites of the Wadi Natrun: seasonal and annual variation and its implication for ancient
exploitation, Archaeometry,46, 497–516.
Silvestri, A., 2008, The coloured glass of Iulia Felix, Journal of Archaeological Science,35, 1489–501.
Silvestri, A., Tonietto, S., and Molin, G., 2011, The palaeo-Christian glass mosaic of St. Prosdocimus (Padova, Italy):
archaeometric characterisation of ‘gold’tesserae, Journal of Archaeological Science,38, 3402–14.
Šmit, Ž., Milavec, T., Fajfar, H., Rehren, T., Lankton, J. W., and Gratuze, B., 2014, Analysis of glass from the post-
Roman settlement Tonovcov grad (Slovenia) by PIXE–PIGE and LA–ICP–MS, Nuclear Instruments and Methods
in Physics Research B,311,53–9.
Sterrett-Krause, A. E., 2006, Evidence for glass production in the Yasmina Necropolis at Carthage, Unpublished M.A.
dissertation, University of Cincinnati, Cincinnati, OH.
Sterrett-Krause, A. E., 2009, Evidence of glass working from Yasmina Necropolis at Carthage, in Annales du 17
e
Congrès
de l’Association Internationale pour l’Histoire du Verre, Anvers, 2006 (eds. K. Janssens, P. Degryse, P. Cosyns, J.
Caen, and L. Van’tdack), 240–6, University Press Antwerp, Antwerp.
Tal, O., Jackson-Tal, R. E., and Freestone, I. C., 2004, New evidence of the production of raw glass at Late Byzantine
Apollonia–Arsuf, Israel, Journal of Glass Studies,46,51–66.
Tatton-Brown, V. A., 1984, The glass, in Excavations at Carthage: the British mission, Volume I, Part I. The Avenue du
President Habib Bourguiba, Salammbo: the site and finds other than pottery (eds. H. R. Hurst and S. P. Roskams),
194–212, H. Charlesworth, Huddersfield.
18 l. Siu, J. Henderson and E. Faber
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
Tatton-Brown, V. A., 1994, The glass, in Excavations at Carthage: the British mission, volume II, part 1. The circular
harbour, north side: the site and finds other than pottery (ed. H. R. Hurst), 282–90, The British Academy, Oxford.
Velde, B., 1990, Alumina and calcium oxide content of glass found in western and northern Europe, first to ninth
centuries, Oxford Journal of Archaeology,9(1), 105–17.
Wedepohl, K. H., and Baumann, A., 2000, The use of marine molluskan shells for Roman glass and local raw glass
production in the Eifel area (Western Germany), Naturwissenschaften,87, 129–32.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher’s
web–site:
Appendix S1 The EPMA results and group attributions of the Late Roman Carthaginian glass. Results are
presented as weight%
Carthaginian glass under the rule of the Romans 19
© 2016 University of Oxford, Archaeometry ••,•• (2016) ••–••
