Primary Glass from Israel and the Production of Glass in Late Antiquity and the Early Islamic Period - article ; n°1 ; vol.33, pg 65-83

-

English
21 Pages
Read an excerpt
Gain access to the library to view online
Learn more

Description

Travaux de la Maison de l'Orient méditerranéen - Année 2000 - Volume 33 - Numéro 1 - Pages 65-83
On présente les résultats des analyses des éléments majeurs de blocs de verre et de quelques fragments de vaisselle en verre provenant d'une série de sites d'Israël, datant de l'époque byzantine et de l'époque islamique (Ve -IXe s.). Tous les verres sont de composition « soda-lime-silica » et les données sont compatibles avec l'idée d'un changement de la source sodique au IXe siècle qui passe du natron aux cendres de plantes. Les verres au natron utilisent un sable calcaire comme source de silice et de chaux tandis que les verres aux cendres de plantes emploient une source de silice relativement pure, la chaux provenant des cendres de plantes. La dalle de verre de Bet She'arim semble être le résultat d'un échec dû à l'utilisation de deux composants porteurs de chaux, des cendres de plantes et du sable calcaire. Les verriers de Bet Eli'ezer produisaient un verre au natron et semblent avoir chargé leur four en différentes fournées de matières premières. Les analyses des verres provenant de quatre fours à bassin de Bet El'ezer indiquent que les verriers produisaient un verre relativement stable qui avait une composition de 73-76% de S1O2 et de 11-13% de Na2O. Les verres byzantins des VIe et VIIe siècles des sites de Dor et d'Apollonia, ont des pourcentages plus élevés de soude et moins élevés de silice, mais les verres de Bet El'ezer trouvent un bon parallèle dans des verres du VIIIe siècle provenant de Ramla. Les verres d'Israël peuvent être distingués des verres égyptiens de la seconde partie du premier millénaire en se fondant sur leurs éléments majeurs ; ils diffèrent aussi de verres romains d'Europe occidentale, jetant quelques doutes sur l'idée selon laquelle la majorité des verres romains était fait avec le sable de la rivière Bélus. Des analyses préliminaires des éléments traces confirment les distinctions entre les groupes de verre, qui peuvent être expliqués en termes de géologie régionale et de matières premières employées.
Major element analysis has been conducted on unformed chunk glass, with some vessel glass, from a number of sites in Israel, of Byzantine to Islamic date. All glasses are soda-lime-silica compositions, and the data are consistent with a change from a natron to a plant ash soda source in the ninth century. The natron glasses utilised a calcareous sand as a source of silica and lime, whereas the plant ash glasses used a relatively pure silica source, the lime deriving from the plant ash. The glass slab at Bet She'arim appears to have failed due to the use of two lime-bearing components, plant ash and calcareous sand. The glassmakers at Bet Eli'ezer produced a natron glass and appear to have charged their furnaces in batches. Analysed glass from four of the tank furnaces at Bet Eli'ezer indicates that the glassmakers produced a relatively consistent product, which had a composition of 73-76% SiO2 and 11-13% Na2O. Sixth to seventh century Byzantine glass, from sites at Dor and Apollonia, has higher soda and lower silica but the Bet Eli'ezer glass is matched by eighth century vessel glass, from Ramla. The glasses from Israel may be distinguished from later first millennium Egyptian glasses on the basis of their major elements, and also differ from much of the Roman glass of western Europe, casting some doubt on the idea that most Roman glass was made from the sand of the River Belus. Preliminary trace element data confirm the distinctions between glass groups, and may be explained in terms of regional geology and raw materials used.
19 pages
Source : Persée ; Ministère de la jeunesse, de l’éducation nationale et de la recherche, Direction de l’enseignement supérieur, Sous-direction des bibliothèques et de la documentation.

Subjects

Informations

Published by
Published 01 January 2000
Reads 30
Language English
Document size 2 MB
Report a problem

Ian C. Freestone
Madame Yael Gorin-Rosen
Michael J. Hughes
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. Colloque
organisé en 1989 par l'Association française pour l'Archéologie du Verre (AFAV). Lyon : Maison de l'Orient et de la
Méditerranée Jean Pouilloux, 2000. pp. 65-83. (Travaux de la Maison de l'Orient méditerranéen)
Citer ce document / Cite this document :
Freestone Ian C., Gorin-Rosen Yael, Hughes Michael J. 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. Colloque organisé en 1989 par l'Association française pour l'Archéologie du Verre (AFAV). Lyon : Maison de l'Orient et de
la Méditerranée Jean Pouilloux, 2000. pp. 65-83. (Travaux de la Maison de l'Orient méditerranéen)
http://www.persee.fr/web/ouvrages/home/prescript/article/mom_1274-6525_2000_act_33_1_1874Résumé
On présente les résultats des analyses des éléments majeurs de blocs de verre et de quelques
fragments de vaisselle en verre provenant d'une série de sites d'Israël, datant de l'époque byzantine et
de l'époque islamique (Ve -IXe s.). Tous les verres sont de composition « soda-lime-silica » et les
données sont compatibles avec l'idée d'un changement de la source sodique au IXe siècle qui passe du
natron aux cendres de plantes. Les verres au natron utilisent un sable calcaire comme source de silice
et de chaux tandis que les verres aux cendres de plantes emploient une source de silice relativement
pure, la chaux provenant des cendres de plantes. La dalle de verre de Bet She'arim semble être le
résultat d'un échec dû à l'utilisation de deux composants porteurs de chaux, des cendres de plantes et
du sable calcaire. Les verriers de Bet Eli'ezer produisaient un verre au natron et semblent avoir chargé
leur four en différentes fournées de matières premières. Les analyses des verres provenant de quatre
fours à bassin de Bet El'ezer indiquent que les verriers produisaient un verre relativement stable qui
avait une composition de 73-76% de S1O2 et de 11-13% de Na2O. Les verres byzantins des VIe et
VIIe siècles des sites de Dor et d'Apollonia, ont des pourcentages plus élevés de soude et moins élevés
de silice, mais les verres de Bet El'ezer trouvent un bon parallèle dans des verres du VIIIe siècle
provenant de Ramla. Les verres d'Israël peuvent être distingués des verres égyptiens de la seconde
partie du premier millénaire en se fondant sur leurs éléments majeurs ; ils diffèrent aussi de verres
romains d'Europe occidentale, jetant quelques doutes sur l'idée selon laquelle la majorité des était fait avec le sable de la rivière Bélus. Des analyses préliminaires des éléments traces
confirment les distinctions entre les groupes de verre, qui peuvent être expliqués en termes de géologie
régionale et de matières premières employées.
Abstract
Major element analysis has been conducted on unformed chunk glass, with some vessel glass, from a
number of sites in Israel, of Byzantine to Islamic date. All glasses are soda-lime-silica compositions, and
the data are consistent with a change from a natron to a plant ash soda source in the ninth century. The
natron glasses utilised a calcareous sand as a source of silica and lime, whereas the plant ash glasses
used a relatively pure silica source, the lime deriving from the plant ash. The glass slab at Bet She'arim
appears to have failed due to the use of two lime-bearing components, plant ash and calcareous sand.
The glassmakers at Bet Eli'ezer produced a natron glass and appear to have charged their furnaces in
batches. Analysed glass from four of the tank furnaces at Bet Eli'ezer indicates that the glassmakers
produced a relatively consistent product, which had a composition of 73-76% SiO2 and 11-13% Na2O.
Sixth to seventh century Byzantine glass, from sites at Dor and Apollonia, has higher soda and lower
silica but the Bet Eli'ezer glass is matched by eighth century vessel glass, from Ramla. The glasses
from Israel may be distinguished from later first millennium Egyptian glasses on the basis of their major
elements, and also differ from much of the Roman glass of western Europe, casting some doubt on the
idea that most Roman glass was made from the sand of the River Belus. Preliminary trace element data
confirm the distinctions between glass groups, and may be explained in terms of regional geology and
raw materials used.La route du verre
TMO 33, Maison de l'Orient, Lyon, 2000
PRIMARY GLASS FROM ISRAEL AND THE PRODUCTION OF GLASS
IN LATE ANTIQUITY AND THE EARLY ISLAMIC PERIOD
Ian C. Freestone *, Yael Gorin-Rosen ** and Michael J. Hughes
abstract
Major element analysis has been conducted on unformed chunk glass, with some vessel glass, from a
number of sites in Israel, of Byzantine to Islamic date. All glasses are soda-lime-silica compositions, and
the data are consistent with a change from a natron to a plant ash soda source in the ninth century. The
natron glasses utilised a calcareous sand as a source of silica and lime, whereas the plant ash glasses used
a relatively pure silica source, the lime deriving from the plant ash. The glass slab at Bet She'arim appears
to have failed due to the use of two lime-bearing components, plant ash and calcareous sand. The
glassmakers at Bet Eli'ezer produced a natron glass and appear to have charged their furnaces in batches.
Analysed glass from four of the tank furnaces at Bet Eli'ezer indicates that the glassmakers produced a
relatively consistent product, which had a composition of 73-76% SiO2 and 11-13% Na2O. Sixth to seventh
century Byzantine glass, from sites at Dor and Apollonia, has higher soda and lower silica but the Bet
Eli'ezer glass is matched by eighth century vessel glass, from Ramla. The glasses from Israel may be
distinguished from later first millennium Egyptian glasses on the basis of their major elements, and also
differ from much of the Roman glass of western Europe, casting some doubt on the idea that most Roman
glass was made from the sand of the River Belus. Preliminary trace element data confirm the distinctions
between glass groups, and may be explained in terms of regional geology and raw materials used.
RESUME
On présente les résultats des analyses des éléments majeurs de blocs de verre et de quelques fragments
de vaisselle en verre provenant d'une série de sites d'Israël, datant de l'époque byzantine et de l'époque
islamique (Ve -IXe s.). Tous les verres sont de composition "soda-lime-silica " et les données sont compatibles
avec l'idée d'un changement de la source sodique au IXe siècle qui passe du natron aux cendres de plantes.
Les verres au natron utilisent un sable calcaire comme source de silice et de chaux tandis que les verres
aux cendres de plantes emploient une source de silice relativement pure, la chaux provenant des cendres de
plantes. La dalle de verre de Bet She'arim semble être le résultat d'un échec dû à l'utilisation de deux
composants porteurs de chaux, des cendres de plantes et du sable calcaire. Les verriers de Bet Eli'ezer
produisaient un verre au natron et semblent avoir chargé leur four en différentes fournées de matières
premières. Les analyses des verres provenant de quatre fours à bassin de Bet Elïezer indiquent que les
verriers produisaient un verre relativement stable qui avait une composition de 73-76% de S1O2 et
de 11-13% de Na2O. Les verres byzantins des VIe et vif siècles des sites de Dor et d'Apollonia, ont des
pourcentages plus élevés de soude et moins élevés de silice, mais les verres de Bet ElVezer trouvent un bon
parallèle dans des verres du vme siècle provenant de Ramla. Les verres d'Israël peuvent être distingués des
verres égyptiens de la seconde partie du premier millénaire en se fondant sur leurs éléments majeurs ; ils
diffèrent aussi de verres romains d'Europe occidentale, jetant quelques doutes sur l'idée selon laquelle la
majorité des verres romains était fait avec le sable de la rivière Bélus. Des analyses préliminaires des
éléments traces confirment les distinctions entre les groupes de verre, qui peuvent être expliqués en termes
de géologie régionale et de matières premières employées.
* British Museum, Department of Scientific Research, London WC IB 3DG, UK.
** Israel Antiquities Authority, P.O.B. 586, Jerusalem 91004, Israel. I.C. Freestone, Y. Gorin-Rosen & M.J. Hughes 66
Recent investigations of the remains of large-scale glass making in Israel ! and Egypt 2 are leading
to a re-evaluation of our concepts of early glass production. Instead of a model whereby fresh glass was
produced from its raw materials in the same workshops that produced glass vessels, a division of
production appears to have existed whereby large quantities of glass were made from alkali and sand in a
relatively limited number of locations. These centres of primary glass production were located close to the
favoured sources of sand, such as that at the mouth of the Belus River, near modern Akko 3 or close to
sources of alkali, such as the natron of the Wadi Natrun, Egypt. The freshly made primary glass was then
broken up into lumps or chunks and distributed to glass workshops for fabrication into vessels.
The foregoing model is similar to that outlined some years ago by Brill 4 on the basis of his
investigations of the nine tonne slab of glass at Bet She'arim, Israel (Fig. 1). It contrasts with the
conventional idea, based on medieval sources such as Theophilus, 5 as well as on early modern practice,
that glass workers were also glass makers and made their own glass from sand and ash. A number of
arguments support the view that a division of production was widespread in antiquity. Firstly, Roman glass
composition is very homogeneous in terms of its major element chemical composition, a characteristic
which has been widely commented upon. 6 This homogeneity is most readily explained by the use of a
small number of raw material sources, mixed to a very simple recipe. Secondly, unambiguous evidence for
glass making (as opposed to glass working) has not been forthcoming from detailed archaeological
investigations of peripheral areas such as Britain, 7 suggesting that the raw glass material was imported
Fig. 1 - The glass slab at Bet She 'arim (photo: composite of two colour transparencies
by Ian Freestone, processed by Tony Milton).
1. Gorin-Rosen 1995.
2. Nennaeia/. 1997.
3. See Trowbridge 1930 and Turner 1956 for references to the Belus sand by classical authors.
4. Brill 1967.
5. Hawthorne and Smith 1963.
6. e.g. Sanderson and Hunter 1981; Baxter et al. 1995; Wedepohl 1998 and Nenna et al. 1997.
7. Price and Cool 1991. Primary Glass from Israel 67
from elsewhere. Thirdly, such a division of production appears to have existed until about a decade ago in
India, where different furnaces were used for glass making and glass working. 8 Finally, it is noted that a
similar situation was typical in the ancient metallurgical industries, where the mining and smelting of
copper, for example, was separated from the workshops which fabricated objects from the early Bronze
Age.9
In the absence of archaeological evidence, the extent to which the foregoing model is applicable to
the ancient world beyond the Near East, where the factories have been discovered, can only be determined
by the investigation of glass composition. If it can be shown that the outputs of different primary glass-
making centres may be differentiated on the basis of their composition, then it may be possible to relate
workshop products - vessels, trinkets, windows etc. - to the original sources of glass. The present paper is
concerned with the characterisation of unformed or 'primary' glass from Israel, and its relationship to glass
from elsewhere.
Materials Analysed
Samples analysed are listed in Table 2, below. This work focuses particularly upon the site of Bet
Eli'ezer, Hadera (see map, Fig. 2 for locations of sites mentioned). Here the remains of seventeen
rectangular furnaces, each of which appears to have produced at least one slab of glass similar in
dimensions to that at Bet She'arim, have been excavated. 10 Even if each furnace had been fired only once,
the amount of glass produced would have been sufficient for the order of one million glass vessels of
typical weight around 150 g. At present there appears to be no straightforward approach to an estimation
of the number of firings but, while direct comparisons are risky, it may be pertinent that the Indian furnace,
reported by Kock and Sode, n appears to have been intended for repeated use. The Bet Eli'ezer furnaces
are believed, on the basis of associated small finds, to have been operational in the late Byzantine period
(sixth to seventh centuries AD). However, material culture in the region does not change abruptly with the
arrival of Islam and, as discussed below, the analytical evidence suggests that the furnaces were still in use
in the early eighth century. Between four and eleven glass fragments, predominantly blue-green in colour,
from each of four of the Bet Eli'ezer furnaces, have been analysed.
Also analysed are chunks of unformed glass from sites where the evidence for glass making, as
opposed to glass working, is less clear. Thus they may represent primary glass which was brought onto the
site for the purpose of glass working, and which was made elsewhere. Two glass making furnaces were
found in excavations in the early nineteen-fifties at the site of Apollonia (Arsuf) by I. Ben-Dor and
P. Kahane 12 but unfortunately the site was never written up. The remains of a large glass slab from those
excavations are now in the Ha'aretz Israel museum, Tel Aviv. 13 The blue-green glass chunks from
Apollonia analysed in the present study are from another area of the site, excavated more recently by
Roll; 14 it is not known if these were produced by the furnaces excavated earlier. Several glass vessels of
6-7th centuries from Apollonia, also blue-green, were also analysed. Blue-green chunk glass of Byzantine
date has also been excavated at Dor 15 (Fig. 2) and a group of this material was included in our sample.
8. Kock and Sode 1994.
9. e.g. Shalev 1994.
10. Gorin-Rosen 1995 and this volume, p. 49-63.
11. Kock and Sode 1994.
12. Perkins 1951.
13. Weinberg 1988, p. 25 footnote 3, refers to this slab.
14. Roll 1995; Roll and Aviram 1993.
15. Dauphin 1997. 68 I.C. Freestone, Y. Gorin-Rosen & M.J. Hughes
Banias
River Bolus
•Jalame
*BetShe'arim
Bet She'an
Jerusalem
Fig. 2 - Map showing locations of sites mentioned in the text. Glass from Israel 69 Primary
For comparative purposes, we have analysed a group of chunk glass which is clearly Islamic, from
the excavations of V. Tsaferis at Banias 16 (Fig. 2). This material dates from the 11th - 13th centuries and
differs from the earlier glass analysed in its colour, which is colourless-green to colourless-purple, as
opposed to the pale blue-greens of the Byzantine glasses. Also Islamic in date, from the 8th- 1 lth centuries,
is a group of vessel glass from Ramla. Ramla was founded as an administrative centre in the early eighth
century, so that it is certain that the vessels analysed from the site were in use during the Islamic period.
The Ramla data are reported in detail elsewhere 17, but are important for some of the dating arguments
employed here, and are referred to in some detail.
Analytical Methods
For the determination of major elements, small fragments were removed from the selected glasses,
set in epoxy resin blocks, ground flat and polished using diamond pastes down to Ιμΐη grade. The polished
sections were vacuum-coated with a thin conductive layer of carbon and analysed for major elements in a
JEOL JSM 840 scanning electron microscope, using an Oxford Instruments ISIS energy dispersive X-ray
spectrometer, with a GEM germanium detector (SEM-EDXA). Conditions were 35° take-off angle, beam
current of 1.7 nA, accelerating potential 15 kV and counting livetime 200 s. Count rate on metallic cobalt
was around 3000 cps.
The X-ray spectrometer was calibrated using pure elements, oxides and simple compounds of known
composition. Results were checked against Corning glass standards A and Β and a number of commercially
available glass standards. Commercial soda-lime-silica standard RM01 was analysed throughout the two-
month period over which the bulk of these analyses were obtained, and the mean analysis of the standard
is compared with the given composition in Table 1. Reproducibility and agreement with the standards were
improved by normalising results to 100% and this procedure has been adopted throughout. On the basis of
the standard analysis, it is expected that a single analysis is within 2% relative of the SiO2 content, 4% of
the Na2O and CaO, 6% of the A12O3 and 20% of the MgO and K2O. Results are presented in Table 2.
For trace elements, small fragments of a subset of the Bet Eli'ezer and Apollonia samples were
abraded using a diamond burr to remove weathered material, then powdered in an alumina ball mill.
Samples of c.20 mg were accurately weighed into 10 ml PTFE beakers with covers and digested with
hydrofluoric and nitric acids. After evaporation to dryness, including a repeated addition of nitric acid, the
residue was taken up in 20 ml dilute nitric acid with a small amount of hydrochloric acid added. These
solutions were submitted for analysis using inductively coupled plasma mass spectrometry (ICP-MS) to
CARE, Imperial College, London, under the supervision of Dr J. G. Williams. Selected trace element data
are presented in Table 3.
Results and Discussion
Raw materials
All glasses analysed are soda-lime-silica glasses, but fall into two main groups, those with potash and
magnesia in excess of 1.5% and those with very low potash and magnesia (Fig. 3). The high-potash, high- group is represented by glasses from Banias, which also have slightly elevated phosphate, P2O5.
This is the generally accepted sub-division between glasses made using natron, or mineral soda (low
potash) and those made using plant ash (high potash). Analysis of dated coin weights has shown that, in
16. Tsaferis and Israeli 1995.
17. Freestone in press. I.C. Freestone, Y. Gorin-Rosen & M.J. Hughes 70
Egypt, natron-based glasses were replaced by plant ash-based glasses in the ninth century, 18 while analysis
of glass from Raqqa, Syria by Henderson 19 shows that both natron- and plant ash-based glasses were in
use at the end of the eighth century. Our own data from Ramla 20 indicate that plant ash replaced natron as
the flux in glass between the eighth and the tenth centuries. A single fragment of low-potash, low-magnesia
window glass from Banias (no. 54, Table 2) is considered likely to be earlier, probably from the Byzantine
period.
A further difference between the Byzantine and post-ninth century Islamic glasses is in their alumina
contents, which are typically around 3% in the glasses from Bet Eli'ezer, Apollonia and Dor, but 1% in the
glasses from Banias {Table 2). This is likely to result from the use of different sources of sand for the two
glass types. Brill 21 has demonstrated that the compositions of fourth century glasses from the workshop
at Jalame (Fig. 2) may be explained as simple mixtures of Egyptian natron and sand collected from the
mouth of the Belus River. The coastal sand has a moderate content of calcium carbonate, which is present
as shell fragments. Thus the lime in the glass is supplied by the sand. Analyses of Levantine plant ash, 22
however, show that it typically has a relatively high lime content. Therefore a lime-rich sand was not
needed for the plant ash-based glasses. Taking "Islamic glass" as post-ninth century, then the basic
formulations of the glasses are thus expressed as:
Roman/Byzantine glass = Natron (high Na2O, low CaO) + Belus-type sand (high CaO, high SiO2)
and
Islamic glass = Plant ash (high Na2O, high CaO) + Sand/crushed quartz (low CaO, high SiO2)
Different silica sources were therefore utilised for the Byzantine and Islamic glasses. If the Belus sand
had been utilised in conjunction with plant ash for the manufacture of glass, a high lime composition would
have resulted, as both raw materials would have contributed lime to the batch. Freestone and Gorin-
Rosen 23 have shown that such a "plant ash plus Belus-type sand" formulation was probably used in the
production of the Bet She'arim slab, causing the resulting glass to contain about twice as much lime as
typical vessel glass from the region. Hence the melting process failed, as the slab either devitrified very
easily, or did not fully melt under normal furnace conditions. The tripolar diagram (Fig. 4) illustrates the
relationship between the various raw materials and the compositions of Levantine glasses. The mixtures of
Belus sand and natron on the one hand, or lime-free sand (quartz) and plant ash on the other, were selected
to produce glasses which had very similar ratios of silica: lime: soda, and plot in the same part of the
diagram. The diagram does not show magnesia and potash which distinguish plant ash- from natron-based
glasses. The Bet She'arim slab can be seen to be exceptional, as it lies on the tie-line joining plant ash and
Belus sand. It is therefore the result of an attempt to combine the sand of the Roman tradition with the ash
of the later, Islamic, tradition. This would appear to date the slab to the ninth century, when the changeover
between the traditions was occurring.
Another difference between the Islamic and the Byzantine glasses is the presence of manganese oxide
at around 1% in the Banias material, whereas it has not been found above 0.1% in any of the Byzantine
chunk glass that we have analysed so far. Manganese, the "glass makers' soap" of the early modern period,
was commonly added to glass to counteract the colourant effect of iron oxides, and this practice was
18. Sayre and Smith 1974; Gratuze 1988; Gratuze and Barrandon 1990.
19. Henderson 1995; 1997.
20. Freestone in press.
21. Brill 1988.
22. e.g. Brill 1970; Ashtor and Cevidalli 1983; Verità 1985.
23. Freestone and Gorin-Rosen 1999. Primary Glass from Israel 7 1
common in the ancient world. 24 Although it was utilised at Jalame in the fourth century, 25 its use in
northern Israel in the Byzantine period seems to be unusual, and it does not appear to be re-introduced until
the ninth-tenth centuries. Furthermore, the evidence of the Bet She'arim slab, which itself contains around
1% MnO 26 appears to be that manganese was added by the glass makers to the glass batch, rather than by
the glass worker at a later stage.
Bet ElVezer
Glass has been analysed from four furnaces at Bet Eli'ezer, denoted by different context numbers
(Table 2), giving a total of 27 samples. Compositions are similar throughout and most glasses have between
73 - 76% SiO2 and 10-13% Na2O. A few samples, however, have markedly higher Na2O. Fig. 5 shows that
virtually the full spread of composition can be found within glass sampled from a single furnace (e.g. L22)
and that significant and consistent compositional differences between furnaces in terms of soda and silica
do not occur. Rather, the relationship between these components is inversely correlated (R2 = 0.88), and is
likely to reflect variations in the proportions of sand and natron used to make the glass. This is a reflection
of the way that the raw materials were mixed and the way each furnace was charged. If furnaces were
charged with single, nine tonne batches of raw materials before firing, then the analyses suggest that such
batches were not completely homogeneous. Alternatively, the furnaces may have been charged in a number
of batches throughout the firing process, and these may have varied in composition.
Evidence for the repeated charging of the furnace comes from the Bet She'arim slab, which may be
seen to show some evidence of horizontal layering in the form of a tendency to fracture horizontally
(Fig. 1) and also shows some compositional layering, with more soda-rich glass towards the top of the
slab. 27 Although such layering could conceivably reflect the cooling of a horizontal slab, it may be that
the batches of glass added in the case of Bet She'arim failed to mix with those below due to the failure of
the glass to melt fully, so that an interface between the layers was maintained. Repeated charging of the
furnace would have had several of advantages. Firstly, it would have allowed organisation of the workforce
so that mixing of batch material could be carried out while the furnace was being fired. Secondly, the
design of the furnaces appears to have required that the heat diffused through the glass at the top of the
slab, through to the glass at the base. It would have been more efficient in fuel to heat the raw materials in
layers, as the time required to heat each layer of glass to the operating temperature would have been
reduced. If it is assumed that the occasional compositions with higher soda (Fig. 5) were derived from the
upper layers of slabs, as found by Brill and Wosinski 28 at Bet She'arim, then an explanation may be that
the glass makers added batch with excess soda towards the end of a firing, to maintain the fluidity of the
glass as the temperature of the furnace began to fall.
Neglecting the outliers in Fig. 5, it appears that the glass makers of Bet Eli'ezer were working to a
single recipe, with a target composition of between 73-76% SiO2 and 11-13% Na2O. Comparison of the
Bet Eli'ezer glass with the of the Byzantine glasses from Dor and Apollonia, analysed here,
and also the Roman period glass from Jalame, analysed by Brill, 29 reveals that the composition of Bet
Eli'ezer glass is extreme, in that its soda contents are very low while silica contents are high (Fig. 6). This
is not just a reflection of the relative proportions of natron to sand, however, because the Bet Eli'ezer glass
has lower lime but higher alumina than the other glasses (Fig. 7), suggesting that different sand sources
24. Sayre 1963.
25. Brill 1988.
26. Brill and Wosinski 1965.
27. Brill and 1965.
28. Brill and Wosinski 1965.
29. Brill 1988. 72 I.C. Freestone, Y. Gorin-Rosen & M.J. Hughes
were being utilised. Brill 30 demonstrated that the Jalame glass composition could be produced from Belus
sand and, in view of the compositional similarities, it may be inferred that the glass from Dor and Apollonia
was made from the same material (Fig. 6-7). Thus the Bet Eli'ezer glass makers are likely to have utilised
a different sand source which is likely to have been relatively close to the furnaces and more accessible
than the Belus sand.
The extreme composition of the Bet Eli'ezer material challenges the view that the glass produced was
used routinely for vessels in 6th-7th century glass making. Indeed, we were initially concerned that this
unusually silica-rich glass was some intermediate product, to be mixed with additional soda, or with
another, more soda-rich glass, before use. However, the problem is resolved by the analysis of the early
Islamic glass from Ramla. Eighth century glass vessels from Ramla match the Bet Eli'ezer glass very
closely (Fig. 6), 31 and it appears that Bet Eli'ezer production extended into this period when the glass was
used directly in vessel manufacture. The high silica composition is likely to have made a suffer glass with
a shorter working range than the more typical Roman/Byzantine type and to have suited the development
of a distinctive approach to glass working. Similar high-silica natron-based glass, dating to the eighth
century, has been reported from Raqqa, Syria 32 but it is not possible to determine precisely how similar it
is to that produced at Bet Eli'ezer as the data are not yet fully published. It may be an indication of the area
supplied by the Bet Eli'ezer production, or an indication of a shared tradition of glass making and glass
working. The initial reasons for the shift to a high-silica, low-soda composition may relate to the cost or
ease of availability of natron which, as the raw material with the most remote source, is likely to have
controlled the amount of glass that could be produced. The mean contents of soda in the Dor and Apollonia
glasses are 16.5% and 15.2% respectively. The reduction of the sodium content to 12.1% in the Bet Eli'ezer suggests a reduction in natron content of the batch of 20-25% relative, allowing the production of
20% more glass for the same quantity of alkali.
Regional Context
It has been possible to identify at least two compositional groupings of natron-based glass produced
in Israel in the mid-late first millennium AD. The group defined here as "Levantine I", is Roman-
Byzantine, and comprises sixth to seventh century glass from Dor and Apollonia, as well as fourth century
glass from Jalame. 33 The sand upon which this group is based was probably derived from the Bay of
Haifa, close to the mouth of the River Belus of antiquity. The second "Levantine II" group was produced
at Bet Eli'ezer and is dated Late Byzantine - Umayyad. From Egypt, Gratuze 34 and Gratuze and
Barrandon 35 recognised two main groups of glass in their study of Islamic glass coin weights. Their Group
I, here termed "Egyptian I", is characterised by high alumina (3-4.5%) and low lime (3-4%); a similar low-
lime composition was identified by Sayre and Smith 36 as a product of the Wadi Natrun factories and this
is confirmed by the work of Nenna et al. 37 We tentatively attribute all of these glasses to the Wadi Natrun
production. Egyptian I compositions appear to have been used up to the 8th century 38 but the date of
30. Brill 1988.
31. See Freestone, in press.
32. Henderson 1995; 1997.
33. Brill 1988.
34. Gratuze 1988.
35.and Barrandon 1990.
36. Sayre and Smith 1974.
37. Nenna et al. 1997.
38. Gratuze and Barrandon 1990.