JULIAN HENDERSON
THE RAW MATERIALS OF EARLY GLASS PRODUCTION
Summary. This paper discusses raw materials used in ancient glasses. Following a consideration of some archaeological reasons f o r studying glass, the discussion concentrates on the evidence provided by chemical analysis of the glass, and focuses on glass f r o m later prehistoric Western Europe. Consideration of the major, minor and trace components of the glass leads to a conclusion that prehistoric glass artisans were able closely to control the addition of small quantities of colorants, opacifiers and clarifiers to the glass melt. Some possible ways of introducing such small quantities of these substances are suggested. A related implication is that glass production in prehistoric Europe was a mainly specialist industry, being part of a ‘highstatus’ socio-economic sphere. When interpreting technical analyses a full consideration of the socio-economic niche occupied by the glass industry is urged. The technical study of ancient glass can lead to a range of specific archaeological implications, possibly generating informed speculation about the identity of raw materials used, and their sources. However, frequently such studies omit important archaeological analyses of the glass which lead to implications about the socio-economic significance of the glass to the site as a whole, and also to larger systems. Indeed it is true to say that both archaeological and technical studies of glass have reached a stage where a consideration of the socioeconomic niche occupied by the glass industry and its products in ancient communities needs to be closely considered. The glass industry can be regarded as a form of specialisation in later European prehistory, and before considering how glass
raw materials form part of this specialisation it is worth briefly examining an example of the way in which the glass industry fits into this broader perspective. A more detailed paper on the subject will appear elsewhere (Henderson, in prep.). As a concept, industrial specialisation such as glass production should be seen in the context of social evolution. Industrial specialisation is, for example, particularly relevant to the rise of urban centres in Europe and the emergence of early states in the Middle East (Childe 1934 and 1950). Long-distance trade in materials is generally considered necessary for the formation of the state (Sherratt 1972) and glass, or glass raw materials, within prehistoric European society are sufficiently exotic to form part of such a trade-system. In the hierarchical
0OXFORD JOURNAL OF ARCHAEOLOGY 4(3) 1985
267
RAW MATERIALS OF EARLY GLASS
society of later Iron Age Europe, glass is likely to have formed part of a distinct system of circulation which also probably included Mediterranean coral (Webster 1970 and Champion 1976,36) and decorated bronzes. With glass and bronzeworking, intensification of labour and skill are required, although the hierarchy amongst artisans can depend on the scale of the operation within a given settlement system. Using ethnographic data Rowlands (1971) has discussed some of the possible socioeconomic conditions in which European bronze age metal-working might have taken place, and has suggested (Rowlands 1980) that weapons, ornaments and livestock can be seen as forming part of a relatively highstatus system of exchange. In bridging the gap between technical studies and archaeology, ethnographic data could also provide invaluable models for the interpretation of the modes of prehistoric glass production. In attempting to suggest the social and economic contexts in which glass production as a specialist industry operated, a study of the raw materials is essential. By researching the possible sources and uses of glass raw materials it becomes possible to infer some culturally-related phenomena such as the details of technical skills involved, the emergence of ‘fashions’ or demands for particular glass types, and to posit models for interaction and cooperation between artisans which would have been essential for the efficient functioning of a workshop or guild. By qualifying or possibly quantifying the degree of industrial specialisation at a particular stage of social evolution through technical analysis it becomes possible to infer a hierarchy of importance for the glass industry within a settlement system. In this paper some technical features of ancient glasses will be drawn together allowing general implications to be made about 268
the nature of prehistoric glass industries. Past contributions to the study of ancient glass raw materials include an important article by Turner (1956a) in which he described the chronology of glassmaking constituents, including the identification and use of separate element oxides in glass. Matson (1951) in another paper described some of the properties of ancient glassmaking materials as did Sayre and Smith (1967). Biek and Bayley (1979) discuss many of the materials used in glass, particularly from Roman and later contexts. Here, the discussion will focus on raw materials used in prehistoric glass with an emphasis on Later Prehistoric western Europe. The functions of some of the raw materials used in the glass will also be described. 1. TYPES
OF EVIDENCE
There are two types of evidence which must be considered. The first, which can be subdivided, is the remains of the raw materials from glassmaking sites together with the published chemical analyses of glass products. The second type of evidence is the historical accounts of glassmaking processes. In considering raw materials and the industrial evidence from glassmaking sites, it is worth making a distinction between sites where only shaping of glass objects (from cullet or ready-made scrap glass) took place and sites where fusion of primary raw materials of glass occurred as well. In the absence of crucibles, furnace remains or well-defined working areas, the chemical relationships between raw materials, byproducts and products (where found) need to be established. If this can be achieved, it helps to define the nature of the industry and in some cases its efficiency for a given set of socio-economic circumstances. It is OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
often easier to establish that glass-shaping rather than glass-making has taken place because once the working area has been deserted burial conditions can weather away unfused alkalies and semi-formed glasses. At the fourth-century A.D. glass factory at Jalame in western Galilee for example it has been possible to infer the recipe used for manufacturing the glass from the chemical analysis of a range of glass types from the site (Dr. Robert H. Brill, pers. comm.). From the chemical analysis of semi-formed glass beads and finished products from the c. third-century B.C. levels of the site of Meare Lake Village West, Somerset it has also been possible to infer a glass recipe for the lead glasses which were definitely being worked if not actually made, using primary raw materials on the site (Henderson 1980 and 1981; Henderson and Warren 1981 and 1983). It is, however, still possible that deposits of sand, lumps of lead and the byproducts of glass-manufacture in the form of vitreous or sintered lumps remain to be excavated and may indicate the existence of a glass industry. The second important area of evidence for the nature of the glass industry in antiquity is written descriptions of glass manufacture. They can provide invaluable information about the use of particular raw materials of glass, the furnace atmosphere, the time of firing and the fuels used. Descriptions of glass-manufacture can also include ritual practices which would otherwise be impossible to infer from the archaeological record and would be lost to us. By using the available descriptions of glass manufacture we can conduct experiments in which the raw materials of glass are fused according to textual sources. This experimental archaeology serves two purposes. One is that it can help to confirm the accuracy of translations, and secondly it can OXFORD JOURNAL OF ARCHAEOLOGY
generate ,. greater knowledge of the parameters affecting melting conditions. The reconstitution of ancient Near Eastern glasses by Brill (1970) has been a notable success in these respects. Information on raw materials and the melting conditions used were gleaned from translations of recipes of Middle Babylonian texts dating to c. 1400-1200 B.C. and from texts from the Library of Assurbanipal at Nineveh (dated 668-626 B.c.). Still, there are problems related to environmental conditions. A significant factor is that the environments, from which the plants which provided the alkalies were selected, were well defined (marshland and desert) and had only a limited range of plant species when compared to temperate environments. In addition to such experimental work, in this paper reference will also be made to medieval descriptions of glass manufacture and related processes of glaze manufacture. Since some basic features of glass technology are long-lived the evidence from the medieval period cannot be ignored since it substantiates the evidence from prehistory. Overall, however, the emphasis here is placed on the practical aspects of raw materials used in prehistoric glass production with specific reference to the European Iron Age. While at present there is no direct evidence for the manufacture of glass from its raw materials during the Iron Age in Europe, chemical analysis and certain archaeological characteristics of the glass indicate that it was a possibility. 2.
PRODUCTION OF GLASS
Glass is produced in nature as a volcanic material (obsidian) and can be manufactured accidentally whenever fusion of a siliceous material (such as silica-rich plants and sand) and an alkali (plant ashes or an 269
RAW MATERIALS OF EARLY GLASS
evaporite) takes place. Instances of this are seen in high-temperature industries such as Iron Age coin-flan production (Tournaire et al. 1982), other metal-working (Biek and Bayley 1979), cremations (Henderson, Janaway and Richards, forthcoming), or during the accidental firing of high-silica plants such as wheat (Folk and Hoops 1982), or the firing of prehistoric forts (Youngblood et al. 1978). When glass is manufactured deliberately the alkali, sand (and/or lead oxide) and a calcium-containing compound, which can be sand, are generally fritted to bring about some fluxing of the silica, and this process reduces the melting temperature of the glass. Turner (1%6b, 293T-295T) discusses the fritting process in some detail (for a soda-lime-silica glass). He notes that a slow chemical reaction between sodium carbonate and sand occurs below 600°C long before melting occurs. The solid compounds remain in a powdered state below 700°C and only assume a sintered or fritted condition above 750”C, up to which point most impurities would be retained. There is evidence that at Tel el-Amarna, Egypt in the 14th century B . C . a specific shallow rectangular crucible form was used for the fritting process (Saleh et al. 1972), and it is interesting to speculate that such vessels might have been used in prehistoric Europe. The addition of separate fractions of decolorisers, opacifiers and colorants to the glass batch can either take place before the final melt, or after the basic glass is fully formed. 2.1 Preparation of the glass batch Before discussing the various kinds of raw materials which are involved in glass manufacture, it is necessary to describe some of the factors which might affect the success of 270
the glass melt. Preparation of batch raw materials prior to the melting process is one of these. In order to increase the surface area to volume ratio for accretion, fluxing and melting, the raw materials of glass production should be ground to a small size. Since the silica (sand or quartz) determines the minimum melting temperature of the batch, grinding to a small size is essential. Stoch et al. (1978) found that shape., ‘texture’, and defects in sand-grain shape affected the speed of melting of the glass batch. Therefore, in economic terms (consumption of fuel and time expended) the grinding of raw materials is an important part of batch preparation. The thorough mixing of ingredients will help to produce homogeneous glass, and if colorants are present it will help to disperse them and prevent streaking. A thorough dispersal of fining agents such as antimony oxide will reduce the possibility of producing bubbled glass. The glass batch may also include scrap glass (‘cullet’) which will reduce the overall melting temperature. 3.
THE MAIN COMPONENTS OF EARLY GLASS MANUFACTURE
3.1 Silica Sources of silica would have provided few problems for prospective glassmakers during prehistory. However to have access to relatively pure sand may have proved to be more difficult. It is therefore worth considering some possible sources of silica and their impurities. Haevernick (1960, 20) suggests that quartz from the Hunsriick mountains in Germany would have provided a usable source for the putative manufacture of Iron Age German glass at sites such as Manching in Bavaria. Suitable sand with variable quantities of impurities is OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
widespread in island situations and the British Isles is no exception to this rule. Possible inclusions in sand are feldspar and clay, which account for much of the alumina and iron in the final glass (Matson 1951,84). Other possible impurities in British sand which can contribute minor and trace components in glass are titanite or sphene (Ca0.TiOz.Si02) providing titanium; chromite (FeO.Cr203) providing chromium and iron; and epidote (Caz(A1,Fe)3Si04) 30H) providing aluminium and iron; and feldspars, either plagioclase, providing aluminium, calcium and sodium, or alkali feldspars, providing aluminium, sodium and potassium. Lucas (1948) among others has reported a high lime (i.e. shell) content in sand which would provide at least some of the calcium oxide in soda-lime-silica glasses of the European Iron Age. Both titanite and epidote are also calcium-bearing, providing additional small quantities of lime to the glass (Goerk 1977,48 and Highley 1977,9). Analyses have been carried out on sand deposits with the intention of relating them to the chemical analyses of ancient Egyptian and Syrian glass. Sand from the River Belus on the Syrian coast (Turner 1954), where Pliny and Strabo describe the ‘discovery’ of glass taking place by accidental fusion of sand and natron, contained 81.3% Si02, 4.26% A1203, 0.13% Fe203, 8.81% CaO with 6.23% loss of weight on ignition (these figures are the mean of three analyses). This sand can certainly be considered relatively pure when compared with the analysis of sand from Tell el-Amarna and Luxor (Turner, ibid. and Table 1) which contained traces of TiOz, MgO, KzO, P2O3, SO3 and BaO. However there are insufficient data available to determine the variability in composition of sand from Tell el-Amarna or the Theban shore opposite Luxor, since only one sample was taken from each site. OXFORD JOURNAL OF ARCHAEOLOGY
The mechanical sorting and mixing of sands by various natural agencies must surely have an effect on the concentrations of minerals in both of these cases. Ball (1907) describes specific concentrations of magnetite and hornblende at different locations at the first Aswan cataract on the Nile, so in some cases it might be possible to increase the probability of identifying sources of sand. At the glass factory at Jalame in western Galilee a history of the coastal landscape was considered in assessing the possible sources of sand used (Dr. Robert H. Brill, pers. comm.). The mineralogical composition of sand samples from Tell el-Amarna, the Theban shore opposite Luxor and the Belus river by Haifa in Syria are given in Table 1 (p. 272). There is also an additional possible source for minor elements which derive from sand: those associated with the sand contained in the wall of the crucible in which the glass was prepared (Saleh et al., 1972). It is worth noting that analyses of various sand deposits in Scotland have indicated that they are not sufficiently homogeneous to allow their chemical characterisation and sourcing (Stanley Warren, pers. comm.). Refining, mixing and fritting procedures will probably affect the extent to which impurities in sand are carried through from the raw materials via the glass batch to the finished glass.
3.2 Alkalies Alkalies are obviously essential to the formation of alkaline glasses, which.inc1ude the majority of transparent glasses in prehistoric Europe. Alkalies dissolve or flux sand or quartz grains, lowering the melting temperature of pure quartz from 1710°C to a minimum liquidus temperature (the absolute melting point of the glass above which nuclei and crystals cannot form) at
271
RAW MATERIALS OF EARLY GLASS
TABLE. I MINERALOGICAL ANALYSIS OF ECYFTIAN SANDS (PERCENT WEIGHT)
(Derived from Turner 1956(b)) Tell el-Amarna
Thebes
Quartz Calcite Feldspars
50-55 30-33 5
Pyroxenes
5
Ilmenite Mud and mica
-
55 4 (+Barite: BaSO,) 16-18 orthoclase and plagioclase feldspars 16-20 amphiboles and pyroxenes (mainly hornblende) 2-3 2-3
1
the ternary eutectic of 725°C for a composition of 21.9% Na20, 5% CaO and 73.1% S O 2 (Morey 1964,-Fig. 20, Table 13, 33). Most analyses of western European Later Prehistoric glasses including those from Britain indicate that. sodium (present at c. 20% Na20), probably introduced as a sodium salt, was the main alkali used with much lower contents of potash (K20) of up to 2.5% (Caley 1962, Hahn-Weinheimer 1955 and 1960, Besborodov 1975 and Henderson 1982). The liquidus temperature for a glass containing 18.8% Na20, 7% CaO and 74.2% Si02 is 867°C which rises to 1060°C for a glass having a composition of 17.6% Na20, 15% CaO and 67.4% SiOz (Morey, ibid.). The former composition is closest to that for prehistoric European glass although the S i 0 2 level is higher. Dr. Robert Brill (pers. comm.) has carried out some detailed experimental work on the base soda-lime-silica glass composition from the Jalame factory site. The basic chemical composition is c. 68%-73% Si02, c. 14%17.4% Na20 and c. 7.4%-10% CaO. He concluded that the softening point was c. 70O0C, the softening temperature was c. 1000°C and the marvering and gathering temperatures were c. 100O"C-1100"C. It is 272
River Belus, Haifa Bay 70-75
16-18 4 plagioclase
4
1 -
therefore valid to assume that the softening and working temperatures for many sodalime-silica European prehistoric glasses, which frequently have a composition of c. 18%-20% Na20, 6-9% CaO and 68%72% SOz (Henderson 1982), is between c. 1000°C and 1100°C. A range of 'representative' prehistoric glass compositions is given in Table 2, including two soda-lime-silica glasses: Analyses 2 and 4. As is evident from Analysis 1 in Table 2, prehistoric glasses sometimes contain more potash than we have considered so far for the soda-limesilica system (Guido et al. 1984, Raftery and Henderson, in press), although they do not appear to have contained the much higher levels of potash found in medieval western European glasses (Cox and Pollard 1981). It is worth noting that the durability of prehistoric soda-lime-silica glasses is greatly superior to medieval potash glasses, and this constitutes an example of technical regression (in the long term). It has always been assumed that the alkali source for prehistoric soda-lime-silica glass was plants of the genus Sulicorniu or the mineral evaporite natron (Forbes 1957, 142). Natron has a variable composition; OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
TABLE 2 REPRESENTAITVE* EUROPEAN PREHISTORIC GLASS ANALYSES
Site: Colour:
Henderson (1982) analysis no. Date B.C. NazO MgO A1203 SiOz KZO CaO SnO, Sb,OS TiO, MnO FeZ03
coo CUO PbO
Rathgall turquoise
640
9th-8th C. 6.3 3.6 1.4 72 10.4 2.2 1.1 nd .O1 0.6 nd 2.49 0.01
MLVW none
MLVW opaque yellow
Aulnat blue
41
16
612
3rd C. 17.3 1.1 .4 70.9 .9 7.0 nd .6 .09 nd .92 nd .01 .37
3rd C. 11.5 1.2 2 59.1 .4 3.4 nd .9 nd 1.6 nd .O1 19.65
2nd C. 14.8 .6 4.7 71.3 .6 6.3 nd nd nd .I7 .62 .22 .17 nd
Hayling Is. opaque yellow
247 50 B.c.-A.D. 50 17.2 1.6 6.1 57.8 1.6 9.5 .7 .5 .63 .93 nd nd 3.58
*These ‘representative’ glass compositions are given here to provide an idea of the salient chemical features found in a range of prehistoric European glasses. The quoted analyses are in no way meant to provide a compositional mean or average. The glasses are transparent unless otherwise specified. nd = not detected - = not analysed for C = century The sites. Rathgall is in Co. Wicklow, Ireland, excavated by Raftery (Raftery and Henderson, in press); MLV = Meare Lake Village (west), Somerset excavated by Bulleid and Gray (1948) and Coles and Orme (Orme et al. 1981); Aulnat is in central France near Clermond-Ferrand excavated by Collis (1981); Hayling Island, Hampshire was excavated by Downey, King and Soffe (1979).
that from Wadi-el-Natrun has the composition of 22.4%-75% sodium carbonate, 5-32.4% sodium bicarbonate with impurities of sodium chloride (2.2%-26.8%) and sodium sulphate (2.3%-29.9%) (Turner 1956(b), Table IV). Analysis of 12 samples of ‘natron’ from Wadi-el-Natrun at the Corning Glass Center, New York, revealed that its identity is in reality trona, the sesquicarbonate Na2C03.NaHC0,.2H,0 (Dr. Robert H. Brill, pers. comm.). It will, for convenience, however be referred to as OXFORD JOURNAL OF ARCHAEOLOGY
‘natron’ throughout this paper. Natron is found in Egypt, especially at Wadi-elNatrun, and in the Beheira province of Lower Egypt, a source which was worked in antiquity. Natron was easily mined, having accumulated adjacent to the lake of Wadiel-Natrun as well as along its bed. However, during the early history of Egypt it seems that ready-made glass was imported from the Near East (Oppenheim 1973), and that in any case natron was not used exclusively for glass production (Turner 1956(a), 40 and 273
RAW MAERIALS OF EARLY GLASS
Shackley 1977, 131). Although scholars have always assumed that Egyptian natron might well have been a source of alkali in ancient soda-lime-silica glass, the distances from Egypt to glass-making sites in Iron Age western Europe are such that one might suggest that a European evaporite or alternative source was used, especially in the Later Iron Age when the volume of glass used increased (see below). With reference to the sources of alkali for glassmaking in India, Bhardwaj (1979, 68) mentions the probable source of a crude natron which results from the decomposition of mineral particles such as plagioclase. There are igneous rocks which are present in alluvium and which contain additional minor quantities of calcium and magnesium salts. Bhardwaj (ibid., 69) mentions that igneous minerals such as nepheline (sodium, potassium, aluminium silicates) and nepheline-bearing rocks, such as nepheline syenite, are possible sources of alkali. After the possible sources of alkali in Egypt and India, the relationships between analyses of alkali sources and ancient glass will now be considered. Besborodov (1957, 183) suggested that variability of glass chemical composition could be related to the variation in chemical composition in the raw materials used and in particular the alkalies. However, as the discussion unfolds it will become apparent that this suggestion is largely untenable. Brill (1970), in his discussion of the chemical interpretation of ancient Assyrian and Mesopotamian texts, has noted that ashes of what is referred to as the Nugu plant, liable to have been a source of alkali, may well have been a species within the genera Salicornia or Salsola among others (and see Turner 1956(a), 42T). These plants are found growing at the margins of marshes 274
in Iraq, Iran and Syria and also within the desert. Rye (1976, 181) points out that plant-ash chemical compositions may be categorised according to their silica and phosphorous contents (in addition to sodium and potassium). However of course once these ashes are used to form glass only the phosphorous content may still be of discriminative value. Brill points out that the chemical composition of these plants is quite variable ‘depending on the particular plants which are chosen, the environments within which they grew, and the ways in which they were burned’ (ibid., 110). Besborodov (1975, Table 5) tabulates the differences in chemical composition between the different parts of plants which could potentially provide sources of alkali for glassmaking. A distinction is made, for example, between the chemical composition of the leaf and the stem of the beech tree with a difference of over 300% in the potash contents and over 600% in the silica contents. Geilmann et al. (1955) indicate that the Mn304 contents of beech ashes from different sources vary from 0.2% to 13%. Besborodov (ibid.) also tabulates the chemical analysis of two species of marine plants with respect to the part of the plant, providing similar though less marked variation in ash composition (the highest is nearly 300% difference between the potash contents for the data for Caladium caspicum). The same point is illustrated by other plant-ash analyses such as wheat, barley, oak, fern, reed, heather and mulberry, and in addition it also underlines the wide range of possible sources of alkalies for glass manufacture in temperate Europe. Brill (1970, 124) published marine and desert plant-ash analyses which again illustrate the variability in composition between species. He went further and prepared some glasses from these ashes with quartzite OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
pebbles using the ‘recipes’ for Mesopotamian glasses (Oppenheim 1970, 35-38). He found that analysis of the glasses he manufactured matched those of the archaeological specimens (Brill 1970, l l l ) , implying that the raw materials used in prehistory were similar to those picked for analysis. Geilmann and Briickbauer (1954, 458 and Table 5 ) also attempted to relate analysed beech ashes to glasses he made out of them. Of most archaeological significance these analyses and experiments illustrate the difficulty of sourcing alkali used to manufacture glass, and also of identifying the species of plant involved. Even when analysis of a reconstituted glass based on historical descriptions matches analyses of ancient specimens, it is never possible to be absolutely certain of a correct selection of identical raw materials. Brill (1970,110-11) mentions some of the problems in attempting to infer the use of a specific plant in glass production by analysing its ash and subsequently combining it with silica according to ancient glass recipes. Non-active ingredients which formed part of the original net weight rank high amongst the potential hazards. Low magnesia correlated with low potash in soda-lime-silica glass implies that natron has possibly been used (Calvi et al. 1963, 312). As Brill points out, however, there may well have been plant ashes which were used as sodium sources which produce similar compositional ratios to those produced by natron in the final glass (1970, 111). Whatever was the true situation in terms of the plant species used, it has been established by Sayre and Smith (1967, 285, 287) that during prehistoric periods in Mycenean Greece, Anatolia, Egypt, Mesopotamia and south-west Persia, high magnesium (HMG) glasses persisted from c. 1500 B.C. until c. 800 B . c . , when a change to low magnesium glasses (LMG) OXFORD JOURNAL OF ARCHAEOLOGY
occurred. ihis change to low magnesiumcontaining glasses must indicate a change in the raw-material source used. The use of a predominantly low magnesia, low potash composition glass during the Iron Age in western Europe (Henderson 1982) suggests the use of a natron-like alkali source. Seaweed, however, has been suggested as a possible alternative (HahnWeinheimer 1955, 6 and Haevernick 1960, 19). This would imply that glass was made from European raw materials in the first place, but at the moment there is no substantive archaeological evidence for this taking place. Pliny ( N . H . XXXI, 107) however refers to the use of oak as an alkali ash for glassmaking having died out in his day, so it is conceivable that he was referring back to the high magnesium-high potash glasses of the 8th century B.C. and earlier, discussed above. A search for suitable evaporite sources for glassmaking in Britain has revealed that the deposits (other than sodium chloride) which have not been dissolved away either lie too deeply for their exploitation during antiquity, or are of a potassium-rich composition and are also possibly not mineable. The ancient bed of the Zechstein Sea, which outcrops in north-eastern England is an example of the latter where it is ‘concealed below a considerable thickness of Mesozoic strata’ (Dunham et al. 1978, 304). If we assume for a moment, even without archaeological evidence, that glass was manufactured in western Europe during the Iron Age, then the evaporation of sea water might be considered a possible alkali source. However if sodium chloride has been used we would expect to detect levels of chlorine in the glass. Geilmann et al. (1955) found levels of between 1% and 2% in the soda-lime-silica glasses they analysed. Fritting sodium chloride as one of the raw 275
RAW MATERIALS OF EARLY GLASS
materials for glass manufacture at 750°C is unlikely to have boiled off the chlorine. Even solubility of sodium chloride in molten glass of composition 75.5% S O 2 , 9.2% CaO and 15.8% NazO at 1400°C was found to be 2.3% (Bateson and Turner 1939). In support of these findings, examination with Dr. Robert Brill of some experimental melts carried out at the Corning Museum of glass revealed that negligible fusion of NaCl and Si02 had taken place at 1400°C. The use of NaHC03 instead of NaCl was, as expected, highly successful at forming a glass. In the medieval period Neri (1612) mentions the import of poverine and rochetta as sources of alkali, but it cannot be assumed that similar trzide relations necessarily existed for the import of alkalies for glassmaking during the Iron Age. The problem of identifying the prehistoric European sources of alkali remains unsolved and may indeed be insoluble.
3.3 Lead Although there is no definitive archaeological evidence for the use of lead as a primary raw material in glassmaking in prehistoric Europe, metallic lead and galena have been found at the glass-working site of Meare Lake Village West in Somerset (Bulleid and Gray 1948, Gray and Bulleid 1953, 250). It is quite possible that on this site, where lead glass was worked during the Iron Age (Henderson 1981), high-lead glasses were being created by adding metallic lead or more probably lead oxide to imported low-lead or lead-free glass. The lead oxide content of up to 44% is definite evidence for the intentional addition of lead to the glass batch (here it acts as a glassformer or flux), and analysis certainly suggests that lead was added to a low-lead glass at the site (Henderson and Warren 276
1981 and 1983). As well as a reduction in the glass’s melting point, lead increases the solubility of antimony, tin and copper oxides at high temperatures, leading to glass opacification if crystals precipitate out at lower temperatures (Sayre and Smith 1967, 303). For some representative European Iron Age lead glasses see Table 2, Analyses 3 and 5 . The regular appearance of deliberately added lead oxide in glass first occurs in Egypt in the 18th Dynasty (1550-1307 B.c.) in opaque yellow glass (Turner and Rooksby 1959, Table 1B: glass from Thebes dated to 1450-1425 B.C. and Sayre 1964), in northeast Iraq in probable 15th century contexts (Vandiver 1983) and in faience (Kaczmarczyk and Hedges 1983). The earliest dated opaque red glasses which contain deliberately added lead are from ninthcentury B.C. contexts of Hasanlu (Brill 1970, 120) and Nimrud of the 8th-7th centuries B.C. (Turner 1954, 455T). The earliest historical reference to the use of lead in glass is found in the Mesopotamian texts (Brill 1970, 121). Lead has been used in both transparent and opaque glasses up to and including the Middle Ages (Henderson and Warren, in press). A n extremely useful partly medieval reference to the use of lead in glass is made by Heraclius in De Coloribus et Artibus Romanorum of 12th to 13thcentury date: ‘Take good and shining lead and put it in a new jar and burn it in the fire until it is reduced to powder. . . Afterwards take sand and mix it well with that powder, so that two parts may be of lead and the third of sand, and put it in an earthenware vase. Then do as before directed for making glass, and put that earthen vase into the furnace and keep stirring it until it is converted into glass’ (Merrifield 1849, 216). Neri (1612, Book 6), however. mentions that calcined lead should OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
be used because lead tends to sink through the crucible base. Heraclius makes the creation of a lead glass sound an easy process, which to some extent supports the view that where lead glass-working is discovered in association with metallic lead the two are related. At the Hellenistic glass-factory on Rhodes which dates to the third century B.c., sheets of lead as well as many half-made leadcontaining opaque yellow glass beads were discovered (Weinberg 1969 and 1983, pers. comm. Dr. R.H. Brill). Again it is possible that lead oxide was added to a low or nonlead glass, or to a glass batch, and the resulting lead glass used for the production of beads or for the decoration of beads and vessels. The presence of opaque yellow lumps of cullet on the Rhodian site (Weinberg 1969, 148) however, may imply that part, at least, of the yellow glass came from elsewhere (or the cullet may have been produced from the Rhodian factory itself). For the medieval period one of the most useful pieces of evidence for the use of lead comes from a glassmaking workshop of 11th to 13th-century A.D. date in Galitch, western Russia. Remains of glassworking crucibles with layers of glass sticking to them were found, as well as fragments of glass and some metallic lead (Dovjenock 1955). This site provides us with the types of archaeological evidence which we would hope to find in European Iron Age contexts. The evidence from the late 11th-century site in Kiev (Kievo-Percherskaya-havra region), which produced mosaic glass for the decoration of Uspenski cathedral, consisted of the remains of two glass-melting crucibles and fragments of metallic and red lead (Bogusevitch 1954). Support of a two-part (lead oxide-silica) melt is afforded by an 8thcentury recipe producing a 70% lead glass for Japanese bead production (Harada et al. OXFORD JOURNAL OF ARCHAEOLOGY
1965), which is corroborated by the analysis of lead-glass beads of that date. Recent analyses of medieval glass from York, England have also provided evidence for a two-part lead oxide-silica recipe (Henderson and Warren, in press). 3.4 Calcium Practically all analyses of prehistoric glass reveal the presence of calcium. Low-lead soda-lime-silica compositions contain up to c. 8% CaO, and lead glasses usually c. 2-5% or less. Calcium acts as a network stabiliser and reduces weathering in glasses, but if present in excessive quantities tends to increase the melting temperature of the batch. Calcium oxide may well not have been recognised as a separate and necessary major raw material in glass (Turner 1956(a), 45T). Brill, in his discussion of the chemical composition of zukO, a substance mentioned in the ancient Mesopotamian texts, and considered almost certainly to have been glass (1970, 109), points out that a calciumcontaining compound is not alluded to as a major additive. Pliny makes reference to ‘shells’ as a glass-maker’s raw material ( N . H . , XXXVI, 66). The inclusion of shell fragments in the sand used to make prehistoric glasses must be highly ranked as a possible source for the calcium detected in the glasses. On rare occasions where the molar ratio of Ca0:MgO is close to unity, dolomite or a dolomitic sandstone cqn be suggested as a source for the calcium in the glasses (Matson 1951). Pernicka and Malissa (1980, 103) infer the use of dolomite in glazes. However the frequent positive correlation between MgO and K 2 0 in ancient glasses indicates that this is generally not the case, and that the magnesium was introduced with the K 2 0 . 277
RAW MATERIALS OF EARLY GLASS
4.
THE
CHEMICAL
AND
ARCHAEOLOGICAL
NATURE OF PREHISTORIC GLASS COLORANTS
By adding a controlled amount of colouring substance to a colourless or weaklycoloured glass, a final predictable colour may be achieved in a glass. However not only would there be a choice of several possible forms of chemical colorant, but also the way in which the colorant material is mixed into the batch, and the furnace conditions employed in the melt, will be important determining factors in achieving the final glass colour. Some of the possible compounds which might be used as a glass colorant are: (1) An unrefined mineral ore containing a proportion of the colorant element but associated with others. (2) A mineral ore which, as in the case of a metal ore, after preparation and purification, may have been crushed, washed and roasted before use. ( 3 ) A colorant frit, containing the colorant oxide in a diluted form, since it is combined with other raw materials, enabling more control over the desired colour than if a raw mineral was used. The term ‘frit’ in this instance is used in its normal technical sense of a calcined mixture of sand and fluxes (i.e. primary raw materials) which are used in glassmaking (see page 270). (4) Either glass cullet (including broken or even whole glass beads), or cakes of highly-coloured ready-prepared glass in the form of canes or cakes which like (3) would enable a degree of control over the resulting colour. Another important group of factors which determine the final colour of the glass melt are the furnace conditions under which the glass is made. These factors include the gaseous atmosphere, the firing time and the 278
type of fuel used, and these have only recently been addressed in the context of medieval glass technology (Sellner et al. 1979, Newton 1978 and Biek and Bayley 1979). On an atomic scale the actual physical properties of the glass which are affected by changing the furnace atmosphere relate to the coordination of the colorant ions in the silicate network. In order to understand the type of structural changes which take place it is worth describing the situation for the role of cobalt as a colorant in glass, partly because it is one of the most-frequently used glass colorants in Iron Age Europe. In a soda-lime-silica glass, cobalt normally occurs in its divalent form. It may either be in a fourfold or a sixfold co-ordination with oxygen atoms in the silicate network. The sixfold co-ordination is rare (Bamford 1977, 42) and it imparts a pink colour to the glass. The optical absorption of cobalt in soda-lime-silica glass, expressed by the linear absorption coefficient at its maximum value, is at least a factor of five greater than that of any other transition metal ions (ibid.). This means in effect that less cobalt is required to colour the glass a certain colour intensity than other ions, and that cobalt will ‘overpower’ other colorant ion impurities. 4.1 Cobalt Farnsworth and Ritchie (1938, 159) attributed the dark blue colour of the Egyptian 18th-Dynasty glasses they analysed to the presence of copper and cobalt modified by manganese. Having pointed out that manganese was a possible modifier of the glass colour they did not expand on this to any extent to include the possible effects that a strong oxidising agent such as Mn02 might have had on the redox OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
equilibrium in the glass. They explicitly state that they regarded the presence of COO in Egyptian 18th-Dynasty glass as wholly deliberate and in an ‘optimum’ amount (Farnsworth and Ritchie 1938,160). They suggest (ibid., 160) that Egyptian alum and nickel ores which contain traces of cobalt might have provided a source. Garner (1956a and 1956b), who also worked on Egyptian glass, suggested that the cobalt was introduced in the form of a cobalt mineral ore. As a result of chemical analysis, Garner suggested that a Persian cobalt ore was used to colour the glass (1956(a), 148), an idea originally put forward by Farnsworth and Ritchie (ibid., and see Lucas 1934, 218). Historical evidence presented by Oppenheim (1973) has indicated that Egypt might have been -partly dependant on Asia (probably ‘Upper Syria’) as a source for at least part of its basic glass raw materials. However Kaczmarczyk (in press) has analysed a probable source for the Egyptian cobalt pigment used in glass (see p. 280 below). In western Europe Jope and Wilson (1957, 83) discuss the association of cobalt with copper in a later Iron Age armlet from ‘Loughey’, near Donaghadee, Co. Down. So far we have not fully established the nature of the cobalt colorant used in the glass: it could easily have been added to a basic glass in the form of a frit or as a fragment of glass cullet. Garner (1956(b)) usefully discusses the main types of cobalt ores, and he attributes the use of either cobaltite or erythrite arsenical cobalt ores to the coloration of a piece of glass from Eridu of c. 2000 B.C. which is one of the earliest pieces of glass to have been analysed. No arsenic was detected. Garner considered that the arsenic level was either too low for detection, or that the arsenic had been significantly reduced by boiling it off at OXFORD JOURNAL OF ARCHAEOLOGY
temperatures above 1250°C. This might have taken place during a purification process, in a high partial pressure of silica or in the actual glass production process. Garner concludes that a relatively low manganese content in a cobalt blue glass indicates the use of an arsenical cobalt ore. The addition of a cobalt-rich ore directly to a glass is unlikely, and the preparation of a colorant material would probably retain the element oxide ratios of the ore. A problem arises in determining the mineral ore type used: if any of the glass ingredients added to the colorant contain elements in common with the colorant, they would be indistinguishable from each other in the analysis of the glass. Copper and cobalt are unlikely to have been introduced with the primary raw materials used to prepare the glass. However in a prehistoric situation alteration of cobalt colorants prior to being added to the melt might easily have taken place. Such an alteration which certainly occurred was the medieval practice of roasting zaffre (or ‘Damascus pigment’) before using it in order to remove any sulphur or arsenic. One positive result which derives from a study by Young (1956) of a later cobaltcontaining vitreous substance, Chinese blue-and-white glaze, was the discovery of a distinctive and datable change in Mn:Co ratio in the glass, which probably corresponded with the use of particular cobalt-rich ore types. The significance of this is that it illustrates that, given a sufficient number of well-dated samples, the chemical analysis of cobalt blue glass has archaeological potential, and in this case the presence or absence of arsenic takes on less significance in the discussion. In considering the chemical analysis of the final glass, it is worth remembering the potential complexities involved in the study of colorant ores, including the fact that cobalt may occur in com279
RAW MATERIALS OF EARLY GLASS
bination with copper. These oxides include, for example, black trianite (2Co2O. Cu0.6H20), iron and manganese ores (absolites), arsenic and sulphur (as cobaltite CoAsS), or nickel and arsenic (as Skutterudite (Co,Ni,Fe)As3). Trace elements often associated with cobalt minerals which might occur in the finished glass melt are Pb-Sb; Ni-Mn-Zn and Bi-Fe. Arsenical cobaltites often contain zinc. There are a number of problems in attempting to relate glass chemical analyses to the actual location of the minerals used for colouring the glass. However Geilmann (1962, 191) suggests that the geological correlations between elements in cobalt-rich ores should be relatable to the ratios of colorant elements in the glasses. This is probably only true in general terms for distinguishing between different families of ores. The Black Forest is one of the major locations for ‘Erdkobalt’ minerals within Celtic Europe. Even here, however, Geilmann (ibid.) advises caution in attempting to establish its use in glass artefacts. Andrews (1962) lists a wide range of cobalt-rich mineral sources in the area of Celtic southern Europe (modern Germany, Austria, Switzerland and Czechoslovakia). The ore types he discusses include smaltite, erythrite, cobaltite and safflorite, the outcrops of which vary considerably in size. Given the wide geographical and compositional range of these cobalt minerals, seriously to suggest that the cobalt which contributed to blue Mycenean glass resulted from smelting a cobalt-rich silver ore in the area of Schneeburg (Dayton 1981) should be regarded as a questionable proposition. Sayre (1964, 7-8) has distinguished between cobalt blue western Roman glasses and those from Mesopotamia and south-western Iran, particularly on the basis of the manganese oxide contents and also accord280
ing to the variation of iron, nickel, copper, tin and lead oxides. He ascribes these differences to the use of manganese-containing ores as opposed to the arsenical lowmanganese cobalt ores of the Near East described by Young (ibid.). Kaczmarczyk and Hedges (1983, 46) have also discovered the association of Mn, Zn, Ni and A1 in cobalt blue faience and cobalt-containing objects of the New Kingdom having abnormally high aluminium concentrations. Farnsworth and Ritchie (1938, 160) suggested that an Egyptian alum could have provided the cobalt source. Kaczmarczyk has analysed the alum, and there is now every reason to believe that because the elements correlated with aluminium and cobalt are found in both alum and blue glass he has confirmed the identity of the cobalt source. So while a change in the use of a general chemical type of ore can be suggested from the analysis of some glasses, the chemical characterisation of specific ore deposits is a greater problem (apart from in exceptional circumstances) not least because of the heterogeneity of the ore deposits. The identity of such ores is therefore likely to be established by the presence of characteristic elements rather than by the ratios of absolute quantities of each. As a caveat to this discussion it is worth noting that cobalt was detected as an impurity in the glaze of pre-New Kingdom faience analysed by Kaczmarczyk and Hedges (1983, 43). An illustration of the potential complexity of glass coloration is provided by the analysis of three early Iron Age archaeological groupings of blue glasses from Wetwang Slack, North Humberside which can be chemically distinguished using their iron, manganese, copper and cobalt contents (Henderson, in press). These glass compositions indicate that three separate cobaltOXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
containing colorants have been used. A fourth chemical grouping of cobalt blue glass from Wetwang, although made up of the analysis of beads of the same type, overlaps two of the other groupings of glass compositions. When examined more closely this fourth grouping was found to be made up of three sub-groupings. The compositions within this fourth group include both low- and high-manganese glasses and illustrate a far more complex situation in the use of cobalt-containing colorants than has been recognised to date. It is of significance that this fourth chemical group was made up of the analysis of the commonest type of bead from the site which in contrast were undecorated. This implies that fewer constraints were imposed on the selection of a colorant than for the chemically more closely-defined decorated beads (assuming a single source). Alternatively the beads might have been collected from a variety of sources, which explains the variation in chemical composition. Whatever the correct explanation, the point still remains that the decorated beads form coherent chemical groupings, implying a tight control over the colorant raw materials used in the glass. The recent discovery of cobalt blue cakes of glass cullet on a Mycenean shipwreck off the coast of Turkey near KaS (The Washington Post, 5 December 1984) indicates that at an early stage in the Bronze Age in Europe cobalt-blue coloured glass was being shipped along the coast of Asia Minor. This is the first definitive proof for the use of a stock of such glass. Research into the use of cobalt as a colorant in ancient glasses thus has not only developed through the use of increasingly more sophisticated and quicker analytical techniques, but has also left certain technical ad archaeological questions only half answered. OXFORD JOURNAL OF ARCHAEOLOGY
4.2 Copper The use of copper as a glass colorant to produce an opaque red colour in glass appears to date from the 15th century B.C. with examples from Nuzi in north-eastern Iraq (Vandiver 1983, 239, 245) and 18thDynasty Egyptian contexts (Brill 1970,120). A recipe involving the use of copper derives from Classical times, when copper sulphide called ferrotto was roasted to produce metallic copper. The production of this glass carried on into the Iron Age and into the Roman occupation of Europe, and it is worth considering the chemically sophisticated technique to which this raw material was subjected in order to produce opaque red glass. A red colour is caused by the presence of cuprous oxide (Cu20) or a mixture of Cu20 and/or metallic copper in suspension. In order to prepare the glass it is necessary to hold the glass under reducing conditions, precipitating out the opacifiers as a suspension and preventing them from going into solution. The opacifiers dissolve under oxidising conditions and even if the hot glass is exposed to air momentarily. The use of a charcoal blanket produces the required reducing conditions. The presence of lead oxide in these glasses, at the 1% level or more, greatly facilitates the precipitation of Cu20. The size of Cu20 crystals in the glass is determined by the temperature and the period during which heat treatment is carried out, and also determine the colour of the cupric glass. Ahmed and Ashour (1981, 3 2 ) , who discuss experiments using a 19.06% PbO glass, indicate that the temperature of the heat treatment determines whether opaque red, yellow or orange glass is produced, which in turn is directly related to crystal size. The Babylonian texts almost certainly describe the preparation of these so-called lead opaque red ‘haematinon’ 281
RAW MATERIALS OF EARLY GLASS
glasses (Brill 1970, 120). The process of ‘striking’ copper out of solution refers strictly to the technique of re-heating the glass at temperatures below its melting point to produce ruby glasses in Europe in the medieval and later periods. The use of copper in glass to produce an opaque red colour dates back to at least as far as the mid second millennium B . C . Interestingly enough the opaque red glasses undergo a change in their lead oxide contents from about 1% before the 9th century B.C. and including the Egyptian 18th Dynasty to deliberately-added amounts of c . 3% between the 9th and the 6th century B . C . The lead oxide contents then increased to between 15% and 30% between the 6th and 3rd centuries B.C. (Brill 1970, 120; Henderson and Warren 1983). Hughes (1972) analysed British opaque red glasses of Iron Age and Romano-British date and found that the high PbO levels continued in use for the rest of the first millennium B.C. and into the 1st century A . D . One important result of Hughes’s work is that by breaking down the analyses into archaeological groupings (hoards) it is possible to assess the compositional variation in the glass used in a given workshop (Henderson 1982) or the production of a particular phase of production. Spratling (1980, 114-15) has commented that there is probably a close connection between the stylistic affinity of the decoration on the bronze objects and the composition of the opaque red glass used to decorate them. Other prehistoric European opaque red glasses which have been analysed include a recent detailed microscopic and chemical examination of a second-millennium B.C. glass bead from Wilsford in Wiltshire (Guido et a]. 1984). These analyses have revealed that copper was deliberately added 282
to a glass which contained negligible lead oxide. One explanation for overcoming the difficulty of dissolving copper in the lowlead glass (which was borne out by replication of the glass analyses) was that between 1.37% and 2.29% of iron oxide in the glass provided a suitable chemical environment for the introduction of copper (ibid., 251) and for its subsequent precipitation. No comparable opaque red glass artefact has been found in prehistoric Europe to date. When copper is used as a glass colorant in its oxidised form (CuO) it imparts a distinctive turquoise colour to the glass. Weyl (1953, 164-5) notes that the presence of lead oxide produces a ‘green’ instead of a ‘blue’ colour. However several transparent turquoise glasses from prehistoric Europe have been analysed which contain both minimal amounts of lead oxide (see analysis 1 in Table 2 for example) and greater quantities (Henderson 1981, Table 16, analysis of 2815). It is therefore evident that further research needs to be carried out into the effect that a variation of lead oxide content has on the colour of a cupric oxide giass. It is also of significance that translucent cobalt blue glasses often contain copper as an ‘impurity’, and this may indicate the use of cobalt-rich copper ore. 4.3 Bronze The presence of copper, lead and tin in ancient glasses in some cases occurring in fixed ratios suggests that leaded bronze has been added to produce a turquoise green colour (Sayre and Smith 1967, 307-309). Where this has been observed it ties in with changes in the lead contents of the bronzes from the areas concerned.
OXFORD JOURNAL OF ARCHAEOLOGY
J . HENDERSON
4.4. Iron The green colouring-effect of iron in glass, often imparted by an impurity in the sand used, normally derives from a mixture of reduced and oxidised ions, with Fez+ions forming the greater proportion (Weyl 1953, 91). Bamford (1982, 6) has recently pointed out that the colour resolves into one major absorption band for Fe2+ and three minor Fe3+bands in spectral transmission. Sellner et al. (1979) have shown that variation in furnace atmosphere produces a range of glass colours in medieval potash glass when iron and manganese are present. These vary in colour from green through purple-brown to blue. However, for the European Iron Age glasses analysed (Henderson 1982). in which iron probably made a major contribution to the glass colour bottle green, brown and ‘black’ (often very dark brown) were the main colours found. In the ‘black’ glass deliberate addition of iron was definitely involved (Henderson, in press), the metal of course being a common part of the technology of the period. In the same way that lead and antimony would facilitate the solution of a metal like copper in a glass, it is also quite possible that over 4-5% iron would have the same effect (Sayre and Smith 1967, 306). Schreurs and Brill (1984) have recently produced a detailed discussion of the role of the iron-sulphur chromophore in the production of yellow glasses in antiquity. The procurement of iron ores for deliberately colouring glasses does not involve complex exchange networks. It is however evident from the glasses studied that deep cobalt blue glasses were preferred to the rather weaker iron blue glases during the Iron Age. 4.5. Manganese
Manganese is rarely deliberately used as a OXFORD JOURNAL OF ARCHAEOLOGY
purple colorant in British Iron Age glasses but. where this has possibly occurred, the colour is dependent on the equilibrium between Mn2+ and Mn3+ ions in the glass. The trivalent ion is thought to be largely responsible for the deep purple colour and is quite stable in glass as compared to its behaviour in aqueous solution (Weyl 1937, 118), although of course in glass it is modified by the presence of other ions. Haevernick (1960) indicates the existence of a high proportion of manganese purple glasses from Mainland European Iron Age sites and specifically from Manching, Bavaria where lumps of purple glass were found (Kramer 1960), similar in composition to lumps of purple glass from Hengistbury Head, Hampshire (Henderson 1982). Where manganese oxide is found in glasses at the 1% level or above, it was probably added deliberately as a manganese-rich compound. As mentioned in the discussion on cobalt ores, manganese ores sometimes occur in association with cobalt, forming a major cobalt-rich ore type. However where manganese is the principal colorant in purple glasses cobalt has not been detected in prehistoric European glasses (HahnWeinheimer 1955 and 1960, Henderson 1982). This implies that manganese-rich compounds were available to the artisans who manufactured the purple glass in a relatively pure form compared with those used for cobalt coloration in prehistoric Europe. A commonly-occurring manganese-rich mineral is pryolusite (MnOZ). While this deliberate addition of a manganese-rich compound as a colorant in prehistoric soda-lime-silica glasses certainly took place, manganese and associated phosphorous contents introduced into the medieval glass batch in the form of beechwood ash (Geilmann and Briickbauer 1954) is likely to have been an accidental use of a 283
RAW MATERIALS OF EARLY GLASS
potential colorant. Newton (1978, 60) suggests that the medieval glass artisans took ‘pot luck’ in the final colour that was achieved. His comments were developed from the experimental work carried out by Sellner et al. (1979) on the effect of furnace conditions on the glass colour. Newton (1985) has however extended this discussion of medieval glass colour and the role of manganese oxide to include a consideration of ‘northern’ and ‘southern’ furnace designs with inferences about the furnace atmosphere. 4.6.
Other colorants
If nickel and chromium occur in glass they are normally introduced as impurities at trace levels (Goerk 1977 and see pp. 270 f. above). However occasionally nickel is found in greater quantities when introduced as a nickel-rich cobalt ore. Such glasses are unusual in prehistoric Europe though some have been discovered in cobalt blue glasses of early Iron Age date from north-western Yugoslavia (Henderson 1982). Nickel alone in a soda-lime-silica glass produces a grey colour (Weyl 1953), but this colour has not been found in prehistoric European glass.
5.
GLASS DECOLORANTS
Manganese can act not only as a colorant in ancient glasses but also a decolorant. In its role as a decolorant the manganese oxidises green ferrous ions in the glass. A representative glass which probably started life as a colourless glass is Analysis 5 in Table 2 which contains a deliberately added 0.63% MnO. The other principal decolorant used in prehistory was antimony, which normally produced a more brilliant colourless glass than manganese. Sayre and Smith (1967, 284
300-1) note that colourless glasses in the Near East of the second half of the first millennium B.C. (first appearing in the 7th century B.c.) are mainly decolorised by antimony pentoxide and continue in use until the late first century B.c., when manganese oxide appears to replace antimony pentoxide as a clarifier. Also Sayre (1963, 279, figs. 10 and 11) has found that in Italy and the Rhineland between the 2nd and 4th centuries A . D . both manganese and antimony oxides were found in colourless glasses. This might either imply that the antimony, which is a more effective decolorant than manganese, was deliberately added to the glass to bring about a more brilliantly clarified glass (Sayre, ibid.), or that antimony-containing glass cullet was added to the manganese-containing glass. Sanderson et al. (1984) argue that for British, Dutch and Scandinavian glass of the first millennium A . D . a mixture of the two potential decolouring oxides noted by Sayre (antimony and manganese) indicates their accidental use. The use of cullet in the Roman centuries was a common practice (Price 1978, 70-1) and probably brought about a mixture of antimony and manganese glasses. However Sayre (ibid., 281, figs. 9-12) notes distinct trends in the antimony and manganese contents of colourless glasses from the 1st century A . D . to the later parts of the first millennium A . D . which, for Italian and Rhenish glass at least, reflect definite and presumably deliberate changes in the use of glasses of particular compositions. Perhaps what we are observing here are compositional differences brought about by the existence of differing regionalised technical traditions. If basic colourless glass was imported from the Near East into Europe during the Iron Age, or if a Near Eastern recipe was used to make European glass, it should be OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
possible to monitor any changes in the basic glass composition through the analysis of Iron Age glass, although the change from antimony- to manganese-based glasses appears to take place slightly earlier in the 2nd-1st centuries B . C . (Henderson and Warren 1983). Indeed, this is the case; and we can therefore infer some form of interaction between the two areas, whether it takes the form of movement of raw materials or of artisans, or transmission of technical ideas. The socio-economic factors which lead to a change in glass decolorants and therefore chemical composition are difficult to ascertain, although the expanding political influence of Rome in northwestern Europe, particularly during the 2nd and 1st centuries B . C . is extremely significant. The political climate in which supplies of raw materials and/or glass cullet changed indicates that the influence of Rome had economic consequences. On an economic level there was certainly also a relationship between the Roman empire and Belgic Britain during the 1st century B . C . and the first half of the 1st century A . D . , so that one cannot expect to be able to talk in terms of distinct ‘Later Iron Age’ and ‘Roman’ glass chemical compositions in all cases. 6.
GLASS OPACIFIERS
The study of opacified ancient glasses has greatly benefited from the work of Turner and Rooksby (Rooksby 1959, 1962 and 1964; Turner and Rooksby 1959, 1961 and 1963). The two main opacifying elements used in ancient glasses are antimony and tin, which sometimes form a compound with lead. To introduce antimony intentionally into the glass batch a mineral like stibnite (antimony sulphide), as inferred from the chemical analysis of Roman glasses (HahnWeinheimer 1954 and Turner 1954, 280), or OXFORD JOURNAL OF ARCHAEOLOGY
Bindheimite (Pb2(Sb,Bi)206(0, OH), a lead antimonate (Rooksby 1962, 23), was probably used. During the process of heating lead-containing batches, a reaction between lead and antimony would produce opaque yellow lead pyroantimonate (Pb2Sb207) which would remain incompletely dissolved in the glass under oxidising conditions. Under the microscope the opaque yellow crystals are in fact seen to be embedded in a matrix of colourless glass. Alternatively if antimony is added to sodalime-silica glass it can react with the calcium in the glass producing opaque white crystals of calcium antimonate (Ca2Sb207 or Ca2Sb206)(Turner and Rooksby 1961, 3). Tin oxide (SnO,) dissolves relatively easily in soda-lime-silica glasses up to a concentration of 10% to 15% weight by melting at temperatures above 1050°C (Turner and Rooksby 1961, 1). As an opacifying agent tin either produces a white colour when present as tin oxide (SnO,) or a yellow colour in the glass when present as cubic or orthorhombic lead-tin oxides (Pb2Sn207 or Pb2Sn04 respectively). The possibility of free tin oxide remaining uncombined in the glass is discussed by Rooksby (1964, 21) when it is added in quantities in excess of the stoiciometric requirement for the formulation of the orthorombic Pb2Sn04. This certainly appears to be the case for some tin-opacified prehistoric European glasses (Henderson 1982 and Henderson and Warren 1983). Tin has also been detected in transparent turquoise later bronze-age glasses from Rathgall, Ireland (Raftery and Henderson, in press). In the cases of both tin and antimony-opacified glasses Rooksby (1964, 25) states that the previously-prepared opacifying agent should be doped with silica, and specifically in the case of tin two molar parts of silica to three of Pb2Sn04 should be
285
RAW MATERIALS OF EARLY GLASS
heated in air to the temperature range 800°C to 900°C. In both lead-tin and lead-antimony opacified glasses the colour and opacification can be destroyed if the glass is heated above c . 1000°C-1100°C. Turner and Rooksby’s contention is that tin oxide began to be used as an opacifying agent in glasses somewhere between the 2nd and 5th centuries A . D . and that ‘simultaneously antimony oxide . . . was ceasing to be employed’ (Turner and Rooksby 1961, 1-2). However Henderson (1982) and Henderson and Warren (1983) have discovered several tin-opacified glasses from Celtic Europe which date to the 2nd-1st centuries B . C . and are the earliest yet found. Henderson (1982) has also discovered that there are several opaque white antimonycontaining glasses of 9th-century date from Viking Ribe in Jutland. This therefore alters our picture of the chronology of the use of opacifiers in ancient glasses. In the 2nd century B . C . the appearance of tin is probably a reflection of the emergence of organised, relatively large-scale industries in later Iron Age oppida. Other opacification of glass can be caused by crystalline soda-lime-silicates which may be due to incomplete vitrification of the glass, or formed by subsequent devitrification. Also masses of air or other gas bubbles or impurities in the glass can produce opacity. Air bubbles can make a glass appear opaque white, especially if they were present in what was originally a colourless glass (Stevenson 1976, 51). Further opacifying agents which have not yet been discovered in European Iron Age glasses are phosphates, arsenic oxides (As203 or As205) in conjunction with lead, and fluorides (Turner and Rooksby 1961, 2, Turner and Rooksby 1963, 306, Werner and Bimson 1963, 303, and Besborodov 1975, 73). 286
7.
SUMMARY AND CONCLUSIONS
The deliberate use of small quantities of colouring, clarifying and opacifying elements clearly indicates that prehistoric glass artisans understood their materials and were able closely to control the amount of the effective colouring element added. The use of the term ‘glass colorant’ however does not indicate that Iron Age artisans measured specific quantities of a single element. Instead, they probably used small quantites of colorant-containing frit or cullet, and added it to a colourless or weakly-tinted glass so that the colouring effect could be effectively controlled by diluting the additive. The chemical and physical study of the possible sources of alkalies used in ancient glass has an inherent range of problems. The compositional variations for different parts of the same species of plant used as an alkali source, which itself will be determined by the specific geochemical conditions in which the plant grew, prevents a successful chemical characterisation. Similarly trace and minor-element analysis of sand deposits, while potentially a worthwhile research topic, may also suffer from excessive heterogeneity due to a range of possible formation and depositional processes of the sand deposit. While reference to the texts in which glass-melting is described provides useful pointers to the ways in which the materials can be recognised or identified, it is difficult to be certain about the precise r61e that an individual artisan within the organisation of a European Iron Age glassmaking centre might have played. There is however little doubt that the techniques involved in making and working glass were highly refined, involving complex procedures and con-
OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
ditions. In a broader economic sense, the ways in which the use of raw materials of glass relate to those consumed in other hightemperature prehistoric industries still needs to be elucidated. The study of glass raw materials in their proper archaeological setting will ultimately add substantially to our knowledge of the forms of social and economic interactions within prehistoric communities which lead to the formation of glass and its use. This is the tip of the iceberg however, since the analysis of fully-formed glass and the distribution of chemically-characterised glass from production centres will reflect on the role of glass in the societies concerned and
can be related to the study of other socioeconomic features. Acknowledgements
I am very grateful to the following people who have commented on parts or all of this paper and made constructive suggestions: Stanley Warren, Professor Martyn Jope, Dr. Robert Brill, Leo Biek, Dr. Bill Melson, Dr. Edward Sayre, Dr. Lambertus van Zelst. Jacqueline O h , Dr. Rita Wright, Dr. Pat McGovern and Dr. Alex Kaczmarczyk.
Research Laboratory for Archaeology and the History of Art, Oxford University, 6 Keble Road, Oxford, OX1 3QJ
REFERENCES
A.A. and ASHOUR, G.M. 1981: Effect of heat treatment on the crystallisation of cuprous oxide in glass. Glass Technology 22(1), 24-33. AHMED,
ANDREWS, R.W.1962: Cobalt, H.M.S.O., London. BALL,J . 1907: A description of the 1st Aswan Cataract of the Nile, Egypt Survey Department. C.R. 1977: Colour generation and control in glass, Elsevier Scientific Publishing Co., Amsterdam. BAMFORD,
BAMFORD, C.R. 1982: Optical properties of flat glass. Journal of Non-Crystalline Solids 47( I ) , 1-20.
BATESON, H.M.and TURNER, W.E.S.1939: Journal of the Sociely of Glass Technology 23 (1939), 265-67T. BESBORODOV, M.A. 1957: A chemical and technical study of ancient Russian glasses and refractories. Journal of the Society of Glass Technology, 41, 16884T. BESBORODOV, M.A. 1975: Chemie und Technologie der antiken und mittelalterlichen Glasser. Mainz.
BHARDWAJ, H.C. 1979: Aspects of Ancient Indian Technology, Delhi.
Academy of Sciences U.S.S.R . , No. 41, 47-8, Moscow.
BRILL,R.H. 1970: The chemical interpretation of the texts. In A.L. Oppenheim et al., 105-28. BULLEID, A. and GRAY, H.ST.GEORGE 1948: The Meare Lake Village, Volume i . Taunton Castle. CALEY, E.R. 1962: Analysis of Ancient Glasses 17901957. The Corning Museum of Glass, Corning Glass Center, New York. CALVI, c.M., TORNATI, M. and SCANDELLARI, M.L. 1963: Recherches historiques et technologiques sur le verre ancient d'Aquilte 6th International Congress on Glass: Advances in Glass Technology (Eds. F.R. Matson and G.E. Rindone) Part 2, New York: Plenum Press, 308-28. CHAMPION, s. 1976: Coral in Europe: commerce and Celtic ornament. In P.-M. Duval and C.F.C. Hawkes (eds.), Celric Art in Ancient Europe: Five protohistoric centuries. Seminar Press, London, 29-37. C:HILDE, V.G.1934: New light on the most ancient east: the orienral prelude of European prehistory. London, Kegan Paul, 1st ed.
BIEK.
L. and BAYLEY, J. 1979: Glass and other vitreous materials. World Archaeology 11(1), 1-25.
CHILDE,
V.A. 1954: Archaeological excavations in 1950 in the Podol on Kiev. Short contributions to the Institute of the History of Material Culture of the
COLLIS, J.R. 1980: Aulnat and urbanisation in France: a second interim report. Archaeological Journal 137, 40-9.
BOGUSEVITCH,
OXFORD JOURNAL OF ARCHAEOLOGY
V.G.1950: The urban revolution. The Town Planning Review 21, 3-17.
287
RAW MATERIALS O F EARLY GLASS
cox, G . A . and POLLARD, A . M . 1981: The multivariate analysis of data relating to the durability of medieval window glass. Proceedings of the Symposium for Archaeometry , Paris 26-29/3/80. Revue- d’Archkometrie no. 5, vol. iii, 119-28. DAYTON, I . E . 1981: Geological evidence for the discovery of cobalt blue glassin Mycenean times as a byproduct of silver smelting in the Schneeburg area of the Bohemian Erzgebirge. Symposium on Archaeometry, Pans 26-29/3/80, Revue d’Archaeometrie Supplement no. 5 , volume iii, 57-61.
DOVIENOCK, V . I . 19.55: Villages and Sites of Towns in the Vicinity of Ancient Galitch. Short Contributions of the Institute of Archaeology of the Academy of Sciences, Ukrainian S.S.R. No. 4, Kiev, 12-13. DOWNEY, R . , KING, A . and SOFFE, G . 1979: Hayling Island Temple, 3rd interim report on the excavations of the Iron Age and Roman temple 1976-78. DUNHAM,
K.,
BEER,
K.E.,
ELLIS,
R.A.,
GALLAGHER,
w. and BRUCKBAUER, T. 1954: Beitrage zur Kenntnis alter Glaser 11. Der Mangangehalt alter Glaser. Glastechhnische Berichte 27. 456-59.
GEILMANN,
GOERK, I . H .
1977: Glass raw materials and batch
preparation. Proceedinns . - of. the 9th International ConRress on Glass, Prague, - 4-8 July, Vol. 2, 38-111. GRAY, H.ST.GEORGE and BULLEID, A . 1953: The Meare Luke Village, Volume ii, Taunton Castle.
c., BAYLEY, J . and 1984: A Bronze Age glass bead from Wilsford, Wiltshire: Barrow G42 in the Lake Group. Proceedings of the Prehistoric Society 50, 245-54. GUIDO, c.M., HENDERSON, J . , CABLE, BIEK, L.
HAEVERNICK, T.E. 1960: Die Glasarmringe und Ringperlen der Mittel und Spatlarenezeit auf dem Europiiischen Festland, Rudolf Habelt, Bonn.
1954: Uber specktrochemische Untersuchungen an romischen Fensterglasern. Glastechnische Berichte 27, 459-64. HAHN-WEINHEIMER, P.
and WEBB, B . C . 1978: United Kingdom, in Mineral Deposits of Europe; Volume I: Northwest Europe, Eds. S.H.U. Bowie, A. Kvalheim and H.W. Haslam, Institute of Mining and Metallurgy, 263-3 17.
HAHN-WEINHEIMER,
FARNSWORTH, M . and RITCHIE, P . D . 1938: Spectrographic studies on ancient glass. Egyptian glass mainly of the 18th Dynasty, with special reference to its cobalt content. Technical Studies in the Field of the Fine Arts 6(3), 155-73.
P. 1960: Die spektrochemische Untersuchung von Glasarmringen und Ringperlen der Mittel und Spatlatknezeit. In T.E. Haevernick 1960, 266-72.
and HOOPS, G . K . 1982: An Early Iron Age layer of glass made from plants at Tel Yin’am, Israel. Journal of Field Archaeology 9(4), 455-66.
objects in Shdsoin, Tokyo.
M . J . , NU=,
M.J.C.
FOLK, R . L .
1957: Glass. In Studies in Ancient Technology Vol. 5, 110-231, Brill, Leiden. FORBES, R . J .
GARNER, H . 1956a: The use of imported and native cobalt in Chinese blue and white. Oriental Art, 2(3),
48-50.
1956b: An early piece of glass from Eridu. Iraq 18, 147-49. GARNER, H .
1962: Beitrage zur kenntnis alter Glaser. VII Kobalt alsfarbungsmittel. Glastechnische Berichte 35(4), 186-92. GEILMANN, w .
and 1955: Die chemische Zusammensetzung einiger alter Glaser. Glustechnische Berichte 28, 14656.
GEILMANN, w . , BEYERMANN, K . , BRUCKBAUER, TH.
JENEMAN, H .
288
P . 1955: Spektrochemische und Physikalische Untersuchungen an latknezeitlichen Glasfunden aus dem Oppidum von Manching. Beilage zum Sammelblatt des Historischen Vereins lngolstadt 65, Frankfurt-am-Main. HAHN-WEINHEIMER,
HARADA, Y . , YAMASAKI, K .
and
OKADA, J .
1965: Glass
HENDERSON, J. 1980: Some new evidence for Iron Age glassworking in Britain. Antiquity liv, 60-1. HENDERSON, I . 1981: A report on the glass excavated from Meare Lake Village West in 1979. Somerset Levels Papers 7, 55-60, Ed. J.M. Coles. HENDERSON, I. 1982: X-ray fluorescence analysis of Iron Age glass, Unpublished Ph.D. thesis, University of Bradford.
3 . (in press). Archaeological, technical and analytical discussion of the glass beads from Wetwang Slack. North Humberside. In J. Dent Wetwang Slack excavations.
HENDERSON,
HENDERSON, J . (in prep.). The archaeological significance of prehistoric glass analyses.
,,
HENDERSON, and WARREN, s,E, 1981: X-ray fluorescence analyses of Iron Age glass: beads from Meare
OXFORD JOURNAL O F ARCHAEOLOGY
J. HENDERSON
and Glastonbury Lake Villages. Archaeometry 23(1), 83-94. HENDERSON, J . and WARREN, S . E . 1983: Analysis of Prehistoric lead glass. Proceedings of the 22nd International Symposium on Archaeometry , Bradford, 168-80. HENDERSON, J . and WARREN, S . E . (in press). The glass and glassy waste from 34 The Shambles, York. In D . Tweddle: Artefacts from recent small site excavations in the city of York, C.B.A. Research Report, Ed. P.V. Addyman. HENDERSON,
J.,
JANAWAY, R.C.
and
RICHARDS,
I.,
forthcoming: A curious clinker. 1977: Silica, Mineral Dossier No. 18,
HIGHLEY, D.E.
H.M.S.O. HUGHES, M . J . 1972: A technical study of opaque red glass of the Iron Age in Britain. Proceedings of the Prehistoric Society 38, 98-107.
and WILSON, B . C . S . 1957: A burial Of the 1st century A.D. near Donaghadee, Co. Down. Ulster Journal of Archaeology 20, 73-95. JOPE, E.M.
(in press). The.source of the cobalt in ancient Egyptian pigments. Proceedings of the 24th International Symposium on Archaeometry, Washington, D.C. KACZMARCZYK, A.
and HEDGES, R.E.M. 1983: Ancient Egyptian faience, Aris and Phillips Ltd., Warminster. KACZMARCZYK, A .
K R ~ M E Rw. , 1960: The oppidum of Manching. Antiquity 34, 191-200. LUCAS, A . 1934 (1921): Ancient Egyptian materials and industries, London. LUCAS, A . 1948: Ancient Egyptian materials and industries, London.
NEWTON, R . G . 1985: Eighth W.E.S. Turner memorial lecture: W.E.S. TurneF recollections and developments. Glass Technology 26(2), 93-103. OPPENHEIM, A . L . 1973: Towards a history of glass in the ancient Near East. Journal of the American Oriental Society 93, 259-66. BRILL, R . H . , BARAG, D. and VON 1970: Glass and Glassmaking in Ancient Mesopotamia, Corning, New York. OPPENHEIM, A . L . , SALDERN, A .
and BAILEY, 1981: Meare Lake Village West 1979. Somerset Levels Papers 7, 12-69. ORME, B . J . , COLES, J . M . , CASELDINE, A . E .
G.N.
PERNICKA, E . and MALISSA, H . 1980: Examination of Islamic glazes with the electron microprobe. In Proceedings of the 16th International Symposium on Archaeometry and Archaeological Prospection (Eds. E.A. Slater and J.O. Tate), Edinburgh, 1976, 96-112. PRICE, J . 1978: Trade in glass. In Roman shipping and trade: Britain and the Rhine provinces, C.B.A. Research Report no. 24, eds. J. du plat Taylor and H. Cleere, Hertford, 70-8. RAFTERY, B . and HENDERSON. J . (in press). Some later Bronze Age glass beads in Ireland. Marburg Studienfur Vor- und Friigeschichte.
1959: An investigation of ancient opal glasses with special reference to the Portland vase. Journal of the Society of Glass Technology 43, 285T88T. ROOKSBY, H . P .
ROOKSBY, H.P. 1962: Opacifiers in opal glasses. G.E.C. Journal of Science and Technology 29(1), 20-6.
1964: A yellow cubic lead tin oxide opacifier in ancient glasses. Physics and Chemistry of Glasses 5, 20-5.
ROOKSBY, H . P .
MATSON, F . R .
ROWLANDS, M . J . 1971: The archaeological interpretation of prehistoric metalworking. World Archaeology 3, 210-24.
1849: Original treatises on the arts of painting, London.
ROWLANDS, M . J . 1980: Kinship, alliance and exchange in the European Bronze Age. In The British Later Bronze Age, eds. J . Barrett and R. Bradley, British Archaeological Reports, British Series No. 83, 15-56.
1951: The composition and working properties of ancient glass. Journal of Chemical Education 28, 82-7. MERRIFIELD, M .
MOREY, G.W. 1964: Phase-equilibrium relations of the common rock-forming oxides except water. U.S. Geological Survey Professional Paper No. 440-L, Chapter L. NERI, ANTONIO
1612: L’Arte Vetraria, Venice.
1978: Colouring agents used by medieval glassmakers. Glass Technology 21(4), 173-83.
RYE, O.S. 1976: Khir: sintered plant ash in Pakistan. Appendix 3 in O.S. Rye and C. Evans: Traditional pottery techniques of Pakistan, Smithsonian contributions to Anthropology No. 21, Smithsonian Institution Press 1976, 180-5.
NEWTON, R . G .
OXFORD JOURNAL OF ARCHAEOLOGY
SALEH,
s.A.,
GEORGE, A . W .
and
HELMI, F.M.
1972:
289
RAW MATERIALS O F EARLY GLASS
Study of glass and glass-making at Wadi-el-Natrun. Studies in Conservation 17, 143-72. HUNTER, J . and WARREN, S . E . 1984: Energy dispersive x-ray fluorescence analyses of 1st millennium A . D . glass from Britain. Journal of Archaeological Science 11(1), 53-69.
and COLLIS, J . 1982: Iron Age coin moulds from France. Proceedings of the Prehistoric Society 48, 417-36.
SANDERSON, D . c . w . ,
TURNER, W.E.S.
1963: The intentional use of antimony and manganese in ancient glasses. VI International Congress on Glass: Advances in Glass Technology, Part 2, Eds. F.R. Matson and G.E. Rindone, Plenum Press, New York, 263-82.
TURNER, W . E . S .
SAYRE, E.V. 1964: Some ancient glass specimens with compositions of particular archaeological significance. Brookhaven National Laboratory, New York.
TURNER, W . E . S .
and SMITH, R . W . 1967: Some materials of glass manufacturing in antiquity. In Archaeological Chemistry: A Symposium, 3rd Symposium on Archaeological Chemistry, Atlantic City, New Jersey (ed. M. Levey), University of Pennsylvania Press, 279-312.
TURNER, W.E.S.
SAYRE, E.V.
SAYRE, E.V.
SCHREURS, J . W . H . and BRILL, R . H . 1984: Iron and Sulfur related colours in ancient glasses. Archaeometry 26, 2, 199-209.
c., OEL, H . J . and CAMERA, B. 1979: Untersuchung alter Glaser (Waidglas) auf Zusammenhang von Zusammensetzung, Farbe und Schmelzatmosphare mit der Elektronenspektroskopie und der Elektronenspinresonam (ESR). Glastechnische Berichte 52, 255-64. SELLNER,
A.G. 1972: Socio-economic and demographic models for the neolithic and bronze ages of Europe. In D.L. Clarke, ed., Models in Archaeology, London, Methuen. SHERRATT,
SHACKLEY, M .
1977: Rocks and Man, London.
SPRATLING, M . G . L . 1980: Observations on the analysis of the compositions of the opaque red glass (‘red enamel’) of the British Pre-Roman Iron Age. Proceedings of the 16th International Symposium on Archaeomerry and Archaeological Prospection, Edinburgh 1976 (Eds. E.A. Slater and J.O. Tate), 113-25.
1976: Romano-British glass bangles. Glasgow Archaeological Journal 4, 45-54.
STEVENSON, R . B . K .
and STOCH, L. 1978: The influence of the structure and shape of grains of mineral raw materials upon the melting speed of soda-lime batches. Szklo Ceramic 29(9), 238-41. STOCH, Z . , PAKULSKA,
M.
TOURNAIRE, I., BUCHSENSCHUTZ,
290
O.,
HENDERSON,
I.
1954: Studies of ancient glass and glassmaking processes. Part 11: the composition, weathering characteristics and historical significance of some Assyrian glasses of the eighth to sixth centuries B.C. from Nimrud. Journal of the Society of Glass Technology 38,455T-56T.
1956(a). Studies of ancient glass and glass-making processes. Part 111. The chronology of glass-making constituents. Journal of the Society of Glass Technology 40, 39T-52T. 1956(b): Studies of ancient glass and glass-making processes. Part V. Raw materials and melting processes. Journal of the Society of Glass Technology 40, 277T-300T. 1959: Studies of ancient glass and glassmaking processes. Part VI. The composition and physical characteristics of the glasses of the Portland Vase. Journal of the Society of Glass Technology 43, 262T-88T. and ROOKSBY, H . P . 1959: A study of opalising agents in ancient opal glasses throughout three thousand four hundred years. Glastechnische Berichte 32K, Heft VIII, 17-28.
TURNER, W.E.S.
W . E . S . and ROOKSBY, H . P . 1961: Further historical studies based on x-ray diffraction methods of the reagents employed in making opal and opaque glasses. Jahrbuch des Romisch-Germanischen Zentralmuseum 8, 1-16.
TURNER,
and ROOKSBY, H . P . 1963: A study of the opalizing agents in ancient glasses throughout 3400 years, part 11. Proceedings of the 6th International Congress on glass: Advances in Glass Technology (eds. F.R. Matson and G.E. Rindone), Plenum Press, New York, 306-7 (abstract).
TURNER, W.E.S.
1983: Glass technology at the midsecond-millennium B . C . Hurrian site of Nuzi. Journal of Glass Studies 25, 239-47. VANDIVER, P.
WEBSTER, R . 1970: Gems: their sources, descriptions and identification. WEINBERG, G. 1969: Glass manufacture in Hellenistic Rhodes. Archaiologikon Deltion 24, 143-51.
1983: A Hellenistic glass factory on Rhodes: progress report. Journal of Glass Studies 25, 37. WEINBERG, G.
OXFORD JOURNAL OF ARCHAEOLOGY
J. HENDERSON
and BIMSON, M . 1963: Some opacifying agents in Oriental glass. Proceedings of the 6th International Congress on Glass: Advances in Glass Technology (eds. F.R. Matson and G.E. Rindone), Plenum Press, New York, 303-305.
s. 1956: An analysis of Chinese Blue-andWhite. Oriental Art 11, 43.
WERNER, A . E .
YOUNG,
1937: The chemistry of colored glass: 11. The Glass Industry April 1937, 117-20.
FREDRIKSSON, B . J . , KRAUT, F . and 1978: Celtic vitrified forts: implications of a chemical-petrological study of glasses and source rocks. Journal of Archaeological Science 5, 99-121.
WEYL, W . A .
WEYL, W . A .
YOUNGBLOOK, E., FREDRIKSSON, K .
1953: Coloured glasses, Corning, New
York.
OXFORD JOURNAL OF ARCHAEOLOGY
29 1