Archaeometry 47, 4 (2005) 763– 780. Printed in Singapore Archaeometry ARCH 0003-813X © Oxford, 4November 47University UK 2005 of Oxford, 2005 ORIGINAL Making C. M. Jackson colourless ARTICLE glass in the Roman period Blackwell Publishing, Ltd. MAKING COLOURLESS GLASS IN THE ROMAN PERIOD * C. M. JACKSON Department of Archaeology, University of Sheffield, Northgate House, West Street, Sheffield S1 4ET, UK This paper discusses the compositional analysis of Roman colourless glass from three sites in Britain. The findings suggest that two broad compositional groups can be identified on the basis of the choice of the initial raw materials selected for glass production, in particular the sand. The largest of these groups is inherently different from the naturally coloured, blue–green glasses of the same period, while the other group is compositionally similar. Further subgroups are apparent on the basis of the decolorizers used. These glass groups are explored in the light of the current theories concerning the organization of glass production in the Roman world. KEYWORDS: BRITAIN, ROMAN, COLOURLESS GLASS, ANTIMONY, MANGANESE, SAND *Received 18 August 2004; accepted 14 December 2004. INTRODUCTION Roman glass is recognized as a masterful feat of technology in terms of both manufacture and design. The most striking visible sign is the control over colour, where colorizers and decolorizers are used skilfully, in conjunction with the control of furnace parameters, to provide a wide variety of hues. This study argues that colourless glasses are a good medium through which to show the skill of the Roman glassmaker and illustrate the influence of the choice of raw materials upon the final glass composition. These choices are explored in the light of current theories relating to diachronic changes in the nature of colourless glass manufacture and the organization of glass production in the Roman world. Colourless glass is known from the early periods in glass-making history (Bimson and Freestone 1988), but its popularity increased during the Roman period. From the late first century ad and into the second century it was produced in volume for high-quality tablewares, its popularity only declining in the late third century (Price and Cottam 1998, 16). In fact, Pliny specifically mentions it, suggesting that ‘the most highly valued glass is colourless and transparent, as closely as possible resembling rock-crystal . . .’ (NH XXXVI, 200). Such glass forms, which are clear and ‘sparkle’, can be found in high-quality vessels such as facet-cut beakers and dietrata. However, the statement also implies that colourless glasses could be found in a number of different qualities, from the truly colourless to those that have a slight blue or green tint. This range of hues within colourless glass has been noted by other authors (e.g., Sayre 1963) and is explored in this paper. The greatest proportion of Roman glass is blue–green (Price and Cottam 1998, 15) owing to iron impurities introduced to the glass from the raw materials (Bamford 1977, 79). A *Received 18 August 2004; accepted 14 December 2004. © University of Oxford, 2005 764 C. M. Jackson colourless glass can be produced either by selecting raw materials that are low in iron or by the addition of a decolorizer to the glass. The selection of high-purity raw materials Roman glasses are produced from silica sand and natron. Natron, or trona, is a crystalline mineral composed predominantly of sodium compounds with few impurities. The iron impurity found in blue–green Roman glasses is therefore assumed to derive from the sand. Although it has been suggested that decolorizing could have been achieved by calcining the iron-containing raw materials before melting, thus oxidizing the iron to its more colourless state (Stern 1990, 37), it is more likely that low-iron, high-purity sands were selected for the manufacture of colourless glasses. In fact, Pliny specifically mentions sand from the River Volturno for colourless glass production (NH XXXVI, 192–5). The use of manganese and antimony as decolorizers in Roman glasses An alternative way of making colourless glass is to add a ‘decolorizer’. Antimony and manganese decolorize glass by oxidizing the iron, although the relationship between the iron, manganese and other compounds in the glass is a complex one (Sellner et al. 1979; Freestone et al. 1990). Antimony, a stronger decolorizer than manganese, also acts as a fining agent in glass by removing dissolved gases (Bamford 1977, 80; Weyl 1981, 118), so producing a more brilliant glass. The relative amount of either oxidizing agent will depend upon the concentration of iron in the glass-making raw materials and equilibrium between the two (or more) elements (Sanderson and Hutchings 1987, 103). Manganese and antimony are derived from minerals. The purest commonly found manganesecontaining mineral is pyrolusite (MnO2) (Green and Hart 1987, 276). Others include psilomilane (∼ 75% MnO, 15% BaO, 5% K2O and 1% CaO), diallogite (between 70% and 90% manganese, 2–3% CaO, 2–7% MgCO3 and up to 11% FeO), wad (which can contain up to 50% iron oxide) (Greg and Lettsom 1977 [1858], 286; Davy 1815, 117) and rhodochrosite (up to 50% MnCO3) (Mondadori 1983, 92). Many of these would have required processing before use. Of the antimony-containing minerals, stibnite (Sb2S3) is the best known. Bindheimite (Pb2(Sb,Bi)2O6(O.OH)) could also have been used, introducing, at the same time, small amounts of lead (Mass et al. 1998; Biek and Bayley 1979, 9). It is difficult to suggest which minerals were used in glass manufacture from the interpretation of chemical analyses of the glasses because of the complex nature of the glass, the uncertainty as to whether the minerals were processed before use and the inclusion of cullet. Some authors suggest that both decolorizers could have been added accidentally as contaminants of the raw materials (Newton 1980, 175; Newton 1985, 98; Gebhard 1989, 167). This is unlikely in the case of antimony, which is not usually present above a few parts per million in commonly occurring geological materials. However, it may be possible in the case of manganese, which has been recorded in soils and sands from Egypt (Turner 1956, 48T) and at levels approaching 0.5% in sands from the River Volturno, a source noted by Pliny for the production of colourless glass (Brill 1999, 475). Other authors note the presence of manganese in glasses, but suggest that its inclusion was a result of recycling or the use of glassmaker’s soap to clarify the glass (Velde and Gendron 1980, 185; Mirti et al. 1993, 234; Velde and Hochuli-Gysel 1997). In 1961, Sayre and Smith proposed a model for the use of decolorizers in Roman colourless glasses. They found that colourless glasses from the Syrian coast are characterized by the Making colourless glass in the Roman period 765 increasing use of manganese oxide (MnO) rather than antimony oxide (Sb2O5), in concentrations in the order of 1% MnO towards the end of the Roman period. This model has since been used widely to characterize the use of decolorizers in colourless glasses throughout the Roman Empire (Velde and Gendron 1980, 185; Freestone et al. 1990, Mirti et al. 1993; Fleming 1999, 141). Later work by Sayre (1963), Sayre and Smith (1967) and Smith (1971, 616) showed that in Italy and northern Europe, glasses were generally decolorized with either antimony or antimony/manganese until the end of the third century, when an increase in manganese is observed. These observations assume the deliberate addition of decolorizers, above 0.2% in each case (Sayre 1963, fig. 1, 265). Models of production The use of high-purity sands and/or the addition of decolorizers in the production of colourless glass have implications for the organization of production. Current theories, relating to late Roman glass production, are centred around two models (Jackson et al. 2003a). The first assumes large primary manufacturing centres near raw material sources, with a host of secondary working centres throughout the Empire (Nenna et al. 2000; Freestone et al. 2002). The second model proposes local glass-making and -working centres (Wedepohl et al. 2003), to which raw materials were locally available and/or imported. Both models have been projected to earlier periods, although the archaeological and scientific evidence for either is difficult to interpret (Baxter et al. 2005). Assuming that glass production sites were located near to raw material sources, the use of high-quality sands from a small number of sources may indicate that a few (large) glass-making centres specialized in colourless glass production. This may be indicative of centralized production. The use of impure raw materials, with the addition of a decolorizer, implies that colourless glass could be produced using the same raw materials as coloured glasses and therefore could be produced at the same centres. This may be more indicative of dispersed production. Of course, the assumption in both of these cases is that ‘pure’ sands are only found at a small number of locations, and that impure sands, used to manufacture coloured glasses, were found widely and used in glass manufacture. Therefore, colourless glass could have been produced in a number of ways relating to the choice of primary raw materials, decolorizers and the production technology. Using compositional analysis, this paper will investigate these technological choices, to determine the nature of the raw materials used and the selection of decolorizers, in a group of first- to fourthcentury ad glasses from Britain. The implications arising from this, relating to the organization of production and possible distribution of glasses to Britain, will be explored. THE GLASS The glass derives from three Romano-British sites spanning the first to fourth centuries ad (see the Appendix) (Jackson et al. 1990; Jackson 1992, 1994). The largest group, excavated from 16–22 Coppergate, York, between 1976 and 1981, consists of vessel and window glass, predominantly second- to third-century, but spanning the first to fourth centuries (Jackson 1992). The glass from Blue Boar Lane, Leicester (third-century), excavated in the 1970s by J. Wacher, and that from Hartshill near Mancetter (second-century), excavated by K. Hartley, again in the 1970s (both unpublished; noted in Price and Cool 1991) are the remains of glassblowing waste. These two assemblages consist of drips and trails and, in the case of Mancetter, some heat-affected vessel fragments that appear to be waste glass (cullet). C. M. Jackson 766 EXPERIMENTAL The glass was analysed by inductively coupled plasma spectroscopy (ICPS) at Royal Holloway, University of London, Egham. The instrumentation and sample preparation protocol are described in Jackson et al. (2003b), and are adapted from the method for the analysis of silicates detailed in Thompson and Walsh (1983). In addition to the samples, four European Science Foundation glass standards (which would best mimic the archaeological glasses to be analysed: 76-C-144, 76-C-150, 76-C-151 and 77-C-33), two blanks (acids and diluent only to check the purity of the reagents, contamination from any of the equipment used and to assess background levels), two internal rock standards (KC10 and KC11, used for calibration and drift monitoring) and two duplicate ‘unknowns’ (to check for reproducibility within the run) were analysed. Independent measurements on the glass standards provided data for accuracy and precision, as well as affording a check on inter-batch variability. Synthetic sodium nitrate was prepared from 0–30% Na2O to reflect the sample matrix, as this element was outside the concentration of the rock standard. Antimony and lead ICP stock solutions were independently prepared at levels similar to those expected in the unknown samples, to analyse for these elements, which were outside the concentrations of the synthetic rock standards and/or absent in the glass standards. Separate calibration lines were computed for each run and the data corrected: the line for soda generally formed a quadratic, and this is reflected in the levels of accuracy and precision. Silica cannot be measured using this method, because heating with hydrofluoric acid removes it by evaporation as silicon tetrafluoride (SiF4). Evaluation of the results The analysis of blanks showed no contamination above minimum detectable levels. The precision and accuracy, defined by the uncertainty in the estimate of concentration as derived from that value (Boumans 1987, 162) and the comparison between the measured concentration and the published value respectively, can be seen in Table 1. Accuracy and precision could not be calculated for lead, as it was not present in the standard. Those elements that could be determined with precision and accuracy, and that occur in the archaeological samples in concentrations above their minimum detectable levels (Floyd et al. 1980), are given in Table 1 and are used in the discussion below. THE COMPOSITIONAL CHARACTERISTICS OF COLOURLESS GLASS The base glass composition The compositional groups described below are shown in Figure 1: primarily, the colourless glasses fall into two broad groups, on the basis of those oxides and elements associated with the primary glass-making raw materials. The first and larger group (Group 1) is compositionally distinct from the more common blue–green glasses, which tend to characterize Roman glass assemblages (Fig. 2). It has lower concentrations of alumina (< 2%, cf., 2.5%), iron oxide (0.3%, cf., 0.5%) and phosphorus pentoxide (0.04%, cf., 0.10%), and is termed low iron-alumina glass. A positive correlation between iron and phosphorus, alumina and titanium suggests that these oxides originate in the same glass-forming raw materials. These glasses are also slightly lower in calcium oxide and potash, and are depleted in barium, zinc, vanadium and copper. Making colourless glass in the Roman period Table 1 Al2O3 Fe2O3 MgO CaO Na2O K 2O TiO2 P 2O 5 MnO Pb Sb Ba Cu Li Ni Sr V Y Zn 767 Minimum detectable levels, accuracy and precision for ICPS analysis Spectral line (nm) Minimum detectable level (ppm) Accuracy (%), inter-day (one day) Accuracy (%), inter-day ( five days) Precision (%), intra-day (n = 25) Precision (%), inter-day (n = 131) 308.21 259.94 383.83 315.89 588.99 766.49 335.05 178.22 257.61 220.35 206.83 455.4 324.75 670.78 231.6 407.77 290.88 371.03 213.86 21.67 69.96 26.12 530.37* 296.51* 72.22 67.89 16.27 6.97 4.95 84.7 0.049 0.62 0 5.29 0.077 2.04 0.31 0.86 1.4 6.8 2.3 0.6 −0.9 −0.9 3.6 6.3 −7.1 −1.7 2.4 −1.3 −2.9 −7.3 −7.9 0.9 6.3 −7.1 3 2.5 2.9 2.6 4.8 5.3 3.5 2.9 6.7 4.4 4.4 4.6 3.8 7.8 7.6 4.5 2.9 6.7 −0.8 −4.1 −3.7 −3.7 4.6 2.1 4.8 14.3 −2.5 8.4 −0.4 −5.6 −3.7 0.4 −1.4 1.4 7.1 −2.5 2.3 4 5 5 2 3 3 3 3 4.4 6 5 7 4 6 4 3 5 *Quadratic equation used for calculation. Accuracy and precision could not be calculated for Pb because it was not present in the standard. Co and Sc were measured but found to be below the MDL in the glasses. Cr, Mo, Nb and Zr were found to be either imprecise, inaccurate or both and were discounted from further analysis. Figure 1 Compositional groups of Roman colourless glass. The second group (Fig. 1, Group 2) falls within the compositional range of the more common blue–green glasses, but has slightly lower mean levels of iron oxide and alumina, and marginally lower calcium oxide than the blue–green glasses (Fig. 2 and Table 2). These are described here as high-iron alumina glasses (in comparison with Group 1 above). 768 C. M. Jackson Figure 2 An illustration of the iron and alumina content of colourless and nearly colourless glasses. The ellipse shows the range of blue–green glass compositions from the same sites, taken from Jackson (1992). Decolorizers Four subgroups can be identified on the basis of the decolorizers used (Figs 1 and 3). The first two of these fall into the low iron-alumina group (Group 1). One is predominantly decolorized using antimony (Group 1a), while the other, consisting here of only one example (York 14108), contains very little of either decolorizer (Group 1b). This latter sample is decolorized by using low-impurity raw materials and careful control of the redox conditions within the furnace (Stern 1990). The third group (Group 2a) contains approximately equal quantities of both antimony and manganese oxides (∼ 0.4%). The majority of these glasses fall into the high iron-alumina group (Group 2), although some are transitional between Groups 1 and 2 (e.g., cylindrical cups 10781 and 13995 from York; see the Appendix). The fourth group of glasses (Group 2b) has high levels of manganese oxide (> 0.8%) and all fall within the high iron-alumina composition (Group 2). In summary, the glasses that are low in iron and alumina (Group 1) appear to be decolorized predominantly using antimony, while those of the high iron-alumina composition (Group 2) contain antimony and manganese, or are predominantly decolorized using manganese. The differences between the groups are also noticeable visually. The glasses in Group 1 are more clearly colourless, they do not have a green tinge when viewed on the cut edge and they ‘sparkle’ in the light. The glasses in Group 2 are weakly coloured, with a green tinge on the cut surface. These glasses are therefore ‘nearly colourless’. DISCUSSION A comparison of the colourless glass analysed here with published groups The trends observed in the data presented can be seen in other groups of Roman glasses, although few publications specifically note the glass colour. The most prevalent examples are Table 2 All glasses % Al2O3 Fe2O3 MgO CaO Na2O K 2O TiO2 P 2O 5 MnO PbO Sb2O5 ppm Ba Cu Li Ni Sr V Y Zn 1.93 ± 0.13 0.35 ± 0.07 0.45 ± 0.09 5.69 ± 0.53 19.34 ± 0.57 0.52 ± 0.08 0.07 ± 0.01 0.04 ± 0.01 0.07 ± 0.09 0.04 ± 0.07 0.53 ± 0.15 152 18 11 12 394 10 7 22 ± ± ± ± ± ± ± ± 19 18 5 3 59 3 1 5 Leicester, glassworking waste (third century) Group 2, nearly colourless (n = 19) York, domestic assemblage ( first to fourth centuries) Group 1, colourless (n = 5) Group 2, nearly colourless (n = 3) Blue–green* (n = 66) Group 1, colourless (n = 53) Group 2, nearly colourless (n = 8) Mancetter, glassworking waste (second century) Blue– green* (n = 70) 2.30 0.55 0.60 6.29 18.85 0.70 0.10 0.10 0.58 0.05 0.32 ± ± ± ± ± ± ± ± ± ± ± 0.12 0.13 0.10 0.72 1.39 0.11 0.02 0.03 0.36 0.07 0.18 1.83 0.34 0.39 5.13 19.3 0.48 0.07 0.04 0.02 0.06 0.61 ± ± ± ± ± ± ± ± ± ± ± 0.05 0.05 0.07 0.83 0.6 0 0.01 0 0.01 0.11 0.08 2.34 0.49 0.56 6.11 19.6 0.74 0.10 0.11 0.35 0.05 0.46 ± ± ± ± ± ± ± ± ± ± ± 0.04 0.12 0.03 0.12 0.12 0.06 0.01 0.01 0.02 0.01 0.01 2.37 0.69 0.55 6.56 18.3 0.71 0.10 0.12 0.27 0.03 0.35 ± ± ± ± ± ± ± ± ± ± ± 0.13 0.17 0.03 0.77 1.2 0.08 0.01 0.01 0.08 0.01 0.16 1.95 0.36 0.47 5.79 19.4 0.52 0.07 0.04 0.09 0.04 0.52 ± ± ± ± ± ± ± ± ± ± ± 0.13 0.09 0.10 0.53 0.6 0.09 0.02 0.01 0.15 0.07 0.16 2.35 0.55 0.67 6.52 18.6 0.73 0.10 0.07 0.70 0.06 0.19 ± ± ± ± ± ± ± ± ± ± ± 0.29 0.15 0.10 1.22 1.47 0.17 0.02 0.02 0.41 0.11 0.18 2.55 0.54 0.60 7.07 18.1 0.78 0.09 0.13 0.50 0.03 0.19 ± ± ± ± ± ± ± ± ± ± ± 0.28 0.19 0.15 0.74 1.07 0.21 0.02 0.05 0.26 0.03 0.15 241 88 18 18 432 20 8 33 ± ± ± ± ± ± ± ± 92 79 4 4 66 5 1 11 144 7 14 10 347 9 6 21 ± ± ± ± ± ± ± ± 6 1 1 2 73 1 1 1 208 113 17 17 404 18 8 34 ± ± ± ± ± ± ± ± 6 28 1 1 7 1 0 4 206 82 20 16 398 16 8 35 ± ± ± ± ± ± ± ± 15 74 11 2 28 2 1 11 155 19 11 12 404 10 7 22 ± ± ± ± ± ± ± ± 22 18 5 3 67 4 1 5 301 99 17 16 427 19 8 33 ± ± ± ± ± ± ± ± 150 118 8 4 24 6 1 16 247 88 13 17 432 18 8 29 ± ± ± ± ± ± ± ± 4 91 8 3 52 4 1 12 Group 1, colourless (n = 1) 1.78 0.34 0.38 5.78 19.4 0.53 0.07 0.05 0.02 0.01 0.49 138 7 15 11 454 8 6 17 Group 2, nearly colourless (n = 8) Blue–green* (n = 87) 2.31 0.52 0.54 6.26 18.2 0.68 0.10 0.12 0.56 0.03 0.47 ± ± ± ± ± ± ± ± ± ± ± 0.06 0.19 0.04 0.57 1.8 0.04 0.02 0.02 0.34 0.02 0.36 2.45 0.48 0.53 7.08 17.5 0.70 0.08 0.14 0.43 0.03 0.16 ± ± ± ± ± ± ± ± ± ± ± 0.11 0.09 0.05 0.66 1.05 0.15 0.02 0.02 0.17 0.02 0.15 223 69 17 20 419 20 8 34 ± ± ± ± ± ± ± ± 27 45 2 5 35 4 <1 8 230 59 16 17 424 17 8 26 ± ± ± ± ± ± ± ± 21 51 3 3 27 4 1 5 Making colourless glass in the Roman period Group 1, colourless (n = 59) Means and standard deviations for glasses from Leicester, York and Mancetter *Blue–green glass compositions were taken from Jackson (1992). 769 770 Figure 3 Britain. C. M. Jackson The concentration of the decolorizers measured in colourless and nearly colourless Roman glass from glasses that are similar in composition to Group 1. The analysis of around 170 samples of Romano-British colourless glass from Colchester and Lincoln, 40 from Binchester and 240 glasses from various sites around Britain also showed a predominance of low iron-alumina glasses decolorized with antimony (Heyworth et al. 1990; Mortimer and Baxter 1996; Paynter 2004; Baxter et al. 2005). This same pattern was observed in second- to third-century reliefcut colourless vessels from Britain (Boon 1985, 15), in two second- to fourth-century vessels from Caerleon and four third- to fourth-century vessels from Caerwent (Brill 1999, 118 –9). Mirti et al. (1993, 233) analysed eight samples of Italian colourless glass and many were low iron-alumina glasses, although antimony was not analysed for. This pattern is not geographically discrete; it is also seen in contemporary colourless glass from Sedeinga, Sudan (Brill 1991, 11). Brill suggests that it is likely the latter glass was imported from Alexandria, or from another glass-making centre. It is not only Roman colourless glass that appears have been produced using low ironcontaining raw materials (Group 1). Low iron-alumina concentrations are seen in Egyptian glasses dating from around 1500–1300 bc (although these glasses do not appear to contain either antimony or manganese and so would fall into Group 1b; Bimson and Freestone 1988), in first- to third-century Iron Age beads from Meare and Glastonbury (Henderson and Warren 1981) and in Saxon cone beakers (Sanderson and Hutchings 1987). Colourless glasses exhibiting higher levels of iron and manganese (Group 2) are more difficult to trace in the literature. Sayre (1963, 277) cites examples of ‘nearly colourless’ glass from Italy which are high in manganese and some similar Rhenish examples from the third to fourth centuries, although a full composition is not presented. Similarly, Stawiarska (2005) also cites some second- to third-century high-manganese colourless glasses from Poland. There are also a number of examples from the late third and fourth centuries, given in Foster (2004). It is increasingly apparent that glasses that appear to be made of low-impurity sands in conjunction with higher levels of antimony are often the more elaborate and ‘high-status’ vessels, such as the facet-cut beakers and cage cups discussed by Sayre (1963, 279). It has been suggested that these types were fabricated in the Rhineland (Harden 1987, 107), although this interpretation is based on distributions of vessel styles rather than manufacturing evidence. An exploration of the use of decolorizers in specific vessel types is given in Jackson et al. (2003a) and Baxter et al. (2005). Making colourless glass in the Roman period 771 Glass raw materials—different sands? The compositional trend of low iron-alumina seen in the largest group of colourless glasses (Group 1) may be explained in a number of ways. A strict refining procedure of the sands or subsequent frits could have been practised (Cole 1966, 47), although washing would probably have been inadequate to remove the undesirable iron-containing fraction or the alumina and phosphorus-bearing heavy clay minerals. Potash and calcium oxide, associated with feldspar minerals such as anorthite, anorthoclase and orthoclase, would also have been difficult to remove by this method. Chemical washing may have had the desired effect, although it is difficult to suggest which strong acids may have been used. The model that best explains this compositional patterning is the deliberate selection of specific low-impurity sand sources, as Pliny documents and as seen in modern glass-making (Turner 1940, 199; Davies and Rees 1945a; Davies and Rees 1945b, 277). This suggestion is reinforced by the low concentrations of trace elements seen in these glasses, which are indicative of pure silica sands. The roles of antimony and manganese in decolouration The analysis of the colourless glasses suggests that in some cases both decolorizers are present. Either of these in sufficient quantity would have the desired effect and, therefore, the presence of both is puzzling. The level of the decolorizer used is relative to the amount of iron and, as antimony is a stronger decolorizer than manganese, lower quantities will render the glass colourless. Although Sayre (1963) suggests that levels above 0.2% of either would be the result of intentional addition, Brill (1988) puts this value for manganese at ∼ 0.4%. This level is upheld here, as concentrations of manganese are often found in the blue–green glasses at 0.5% or higher (Table 2), where it is assumed that manganese was not being deliberately added as a decolorizer. The use of antimony as the decolorizer in the colourless glasses (Group 1) is supported by a weak to moderate correlation with iron (r = 0.62); a similar relationship between iron and manganese in these glasses is not evident. It is suggested, therefore, that where both decolorizers are present in low quantities, manganese is probably entering the glass either unintentionally or for purposes other than as a decolorizer. Unintentional addition could be explained by manganese-containing raw materials. Brill (1999, 457) notes manganese concentrations of 0.43% in sand from the mouth of the River Volturno, which would give a base level of around 0.3% in a natron glass. Alternatively, Freestone et al. (2005) suggest that the presence of manganese in some high iron-manganese fourth-century green glasses (HIMT) is to oxidize sulphur and prevent the glass from going black. As some of the glasses analysed here are also late Roman, they may be part of the same tradition. They are also high in iron and titanium (e.g., Appendix, York 12628). Of course, the effects of recycling cannot be discounted. In summary, it is suggested that for the majority of glasses analysed here, antimony is the primary decolorizer. This is because the colourless glass often has lower (or similar) values of manganese to that seen in the blue–green glass, but has relatively higher concentrations of antimony. The concentration of manganese required to decolorize these glasses (which have iron concentrations generally below 0.5%) would appear to be > 0.5%, rather than the 0.2% suggested by Sayre (1963). This may suggest that the manganese was not deliberately added as a decolorizer in most cases. It is, however, clear that some greenish-tinged glasses are decolorized using manganese in relatively high proportions (≥ 1%). In these cases, the addition of manganese is deliberate. 772 C. M. Jackson Recycling The effects of indiscriminate recycling of cullet may account for the presence of manganese in some samples. Indeed, elevated levels of decolorizers, colorants and opacifiers, such as manganese, lead, copper and antimony, are commonly associated with recycling practices (Jackson 1997; Freestone et al. 2002). The glasses produced using low iron-alumina sand, decolorized predominantly with antimony (Group 1) have low levels of copper and manganese, which suggests that they were not produced using recycled material (Jackson 1997). In contrast, the nearly colourless samples produced using higher iron-alumina sands (Group 2) have comparable levels of copper, manganese and antimony to the blue–green examples, which suggests recycling in both cases. Recycling may also explain the presence of both decolorizing agents in the blue–green and nearly colourless glasses. Chronological patterning The conclusion that antimony is being used as a decolorizer in the majority of these Roman glasses, in preference to manganese, can be tested against the earlier work of Sayre (1963), Sayre and Smith (1967) and Velde and Gendron (1980), who found that trends of antimony and manganese oxide ratios varied over time in the Roman period. Work by Sayre (1963) showed that northern European colourless glass from the late first and early second centuries appears to have been decolorized using manganese, while by the late second century it was predominantly decolorized using antimony, or both antimony and manganese oxides. The trend is towards increased relative manganese concentrations from the end of the third through to the fourth centuries. Thus, in northern Europe the most dominant groups throughout the second to late third centuries ad appear to be the antimony and manganese glasses, although the ratios of the two oxides are not consistent between samples from the same century. Using the criterion outlined by Sayre (1963, 265), that intentional decolorizers are present above 0.2%, most of the colourless glass waste from the mid-second-century site at Mancetter is antimony/manganese decolorized, while that from the third-century site at Leicester appears to be mainly antimony decolorized (see the Appendix). The glass from York, spanning the first and fourth centuries, is predominantly antimony decolorized but includes all types, including those with high manganese. These may fit well into Sayre’s chronological development of the use of the two oxides in northern Europe, where in the second and third centuries both antimony and manganese were used (Sayre 1963, 277). However, of the dated fragments, there are no manganese-decolorized examples from the late first or early second centuries. In contrast, if we assume that manganese was only used intentionally to decolorize glasses at levels above 0.5%, as suggested earlier, then only seven samples, from Mancetter and York, appear to be intentionally decolorized using manganese. These glasses all contain manganese, at around 1%, with only traces of antimony. The two samples from Mancetter can be dated only broadly to the second century, but dated examples from York exist from the fourth century. This evidence cannot wholly support the idea of a move to manganese glasses in the fourth century; further analyses of later glasses are needed. However, it is interesting to note that in the fourth century manganese features heavily in HIMT glasses, thought to be produced in Egypt (Freestone et al. 2005). Making colourless glass in the Roman period 773 CONCLUSIONS These results show that the production of colourless glass was not just a highly developed skill that required knowledge of the nature of decolorizing agents, but that also the raw materials for the base glass were chosen specifically to enable the decolorizing properties of the minerals to have a greater effect. The results presented here suggest that antimony was the preferred decolorizer, especially for earlier glasses, and that a specific high-grade sand was chosen for the production of the majority of truly colourless glasses found in the north-west provinces throughout the Roman period. In these glasses, there is no evidence of recycling. A smaller subsample of less successfully decolorized glasses shows different compositional characteristics. These ‘nearly colourless’ glasses are either decolorized with manganese (around 1% and above), or with manganese and antimony together (∼ 0.5%), where antimony acts as the stronger decolorizer. Manganese only appears acting as a decolorizer at concentrations above 0.5%; below this level, it was probably not deliberately added for this purpose. In these cases, the base glass composition, and hence the raw materials used to produce the glass, are similar in composition to those used for coloured glasses. There is no evidence that refined or selected sands were used, and many glasses show evidence of recycling. These glasses are predominantly later types, which tantalisingly upholds Sayre’s suggestion that there was a move to manganese in the fourth century. However, this change in the technology of production may be linked to the compositional and stylistic changes seen in all Roman glasses at this time. Both types of colourless glass are seen at the three sites studied. These findings would indicate that there are at least two broad models of colourless glass production, which may reflect some diachronic variation. The first is that the majority of glasses in the second and third centuries were produced using high-purity sands. Sands such as these, with high levels of strontium, are suggested by Freestone et al. (2002) to derive from beach sands, typically Mediterranean coastal sands. There is no evidence for recycling in these compositions. The second is that other colourless glasses, perhaps those that were less elaborate or of lower status, were produced using raw materials similar to those used to produce coloured glasses. For these latter glasses, mixtures of decolorizers were used, either intentionally or otherwise, and cullet was added to the melt. It is probable that these two (or more) glass compositions were produced at different centres. It is not possible to deduce from these analyses whether this fits with models of centralized primary glass production suggested by Freestone et al. (2002) and Nenna et al. (2000) for late Roman and early post-Roman glasses or with a more dispersed pattern indicated by Baxter et al. (2005), or both—the picture is complex. It is clear, however, that high concentrations of low-impurity colourless glasses in the second and third centuries indicates centralized control over raw materials and processes. In the later, higher-impurity glasses, decolorized using a mixture of decolorizers and recycled material, the pattern suggests more dispersed production. ACKNOWLEDGEMENTS The Science and Engineering Research Council are thanked for financial assistance (Award 88803864), and Drs J. N. Walsh and S. James of the NERC ICP–AES facility at Royal Holloway, Egham, are thanked for help with chemical analysis. Dr H. E. M. Cool and Professor M. J. Baxter are acknowledged and thanked for discussions concerning colourless glass. The helpful comments from the three referees are acknowledged, and Professor I. Freestone is thanked for 774 C. M. Jackson reading an earlier manuscript. Without the support from Dr S. E. Warren and Professor J. R. Hunter, this project would not have been instigated or completed. REFERENCES Bamford, C. R., 1977, Colour generation and control in glass, Elsevier, Amsterdam. Baxter, M. J., Cool, H. E. M., and Jackson, C. M., 2005, Further studies in the compositional variability of colourless Romano-British vessel glass, Archaeometry, 47, 47–68. Biek, L., and Bayley, J., 1979, Glass and other vitreous materials, World Archaeology, 11(1), 1–25. Bimson, M., and Freestone, I. C., 1988, Some Egyptian glasses dated by royal inscriptions, Journal of Glass Studies, 30, 11–15. Boon, G. C., 1985, A new Roman relief-cut vessel from Caerwent, Journal of Glass Studies, 27, 11–17. Boumans, P. W., 1987, Inductively coupled plasmas (ICP): state of the art in research and routine analysis, in XX Colloquium Spectroscopicum Internationale, Prague, 7–26. Brill, R. J., 1988, Scientific investigations of the Jalame glass and related finds, in Excavations at Jalame, site of a glass factory in late Roman Palestine (ed. G. D. Weinberg), 257–94, University of Missouri Press, Columbia, MO. Brill, R. J., 1991, Scientific investigations of some glasses from Sedeinga, Journal of Glass Studies, 33, 11–29. Brill, R. J., 1999, Chemical analyses of early glasses, vol. 2, The tables, The Corning Museum of Glass, New York. Cole, H., 1966, Analyses and discussion of the Caerleon window glass, Journal of Glass Studies, 8, 46 –7. Davies, W., and Rees, W. J., 1945a, A note on the heavy minerals present in some British glass-making sands, Journal of Glass Technology, 29, 266–7. Davies, W., and Rees, W. J., 1945b, The millstone grit of the Yorkshire dales as a source of glass-making sand, Journal of Glass Technology, 29, 273–8. Davy, H., 1815, Some experiments and observations on the colours used by the Ancients, Philosophical Transactions of the Royal Society of London, Part I, 97–124. Fleming, S., 1999, Roman glass: reflections on cultural change, University of Pennsylvania Museum of Archaeology and Anthropology, Philadelphia, PA. Floyd, M. A., Fassel, V. A., and D’Silva, A. P., 1980, Computer-controlled scanning monochromator for the determination of 50 elements in geochemical and environmental samples by inductively coupled plasma–atomic emission spectrometry, Analytical Chemistry, 52, 2168 –73. Foster, H. E., 2004, Late Romano-British vessel glass: investigating compositional and typological changes to the assemblage in the historical context, Unpublished Ph.D. thesis, University of Sheffield. Freestone, I. C., Bimson, M., and Buckton, D., 1990, Compositional categories of Byzantine glass tesserae, in Annales du 11e Congrès de l’Association Internationale pour l’Histoire du Verre, Bâle, 29 août—3 septembre 1988, 271– 81, AIHV, Amsterdam. Freestone, I. C., Ponting, M., and Hughes, M. J., 2002, The origins of Byzantine glass from Maroni Petrera, Cyprus, Archaeometry, 44, 257–72. Freestone, I. C., Wolf, S., and Thirlwall, M., 2005, The production of HIMT glass: elemental and isotopic evidence, in Annales du 16e Congrès de l’Association Internationale pour l’Histoire du Verre, 153–7, AIHV, London. Gebhard, R., 1989, Der Glasschmuck aus dem Oppidum von Manching, Die Ausgrabungen in Manching, Band 11. Green, L. R., and Hart, F. A., 1987, Colour and chemical composition in ancient glass: an examination of some Roman and Wealden glass by means of ultraviolet-visible-infra-red spectrometry and electron microprobe analysis, Journal of Archaeological Science, 14, 271–82. Greg, R. P., and Lettsom, W. G., 1977 [1858], Manual of the mineralogy of Great Britain and Ireland, with supplementary list by L. J. Spenser and P. G. Embrey, Lapidary Publications, Broadstairs, Kent. Harden, D. B., 1987, Glass of the Caesars, Olivetti, Milan. Henderson, J., and Warren, S. E., 1981, X-ray fluorescence analysis of Iron Age glass: beads from Meare and Glastonbury lake villages, Archaeometry, 23, 83 –94. Heyworth, M., Baxter, M. J., and Cool, H. E. M., 1990, Compositional analysis of Roman glass from Colchester, Essex, Ancient Monuments Laboratory Report 53/90. Jackson, C. M., 1992, A compositional analysis of Roman and early post-Roman glass and glassworking waste from selected British sites, Unpublished Ph.D. thesis, University of Bradford. Jackson, C. M., 1994, Appendix 1. Compositional data from the Romano-British glassmelting debris from Mancetter and Leicester, in Exploratory multivariate analysis in archaeology (M. J. Baxter), 228 –9, Edinburgh University Press, Edinburgh. Making colourless glass in the Roman period 775 Jackson, C. M., 1997, From Roman to early medieval glasses: many happy returns or a new birth, in Annales du 13e Congres de l’Association Internationale pour l’Histoire du Verre, Pays Bas, 28 août—1 septembre 1995, 289 –302, AIHV, Lochem, The Netherlands. Jackson, C. M., Cool, H. E. M., and Baxter, M. J., 2003a, Identifying group and meaning, in Échanges et commerce du verre dans le monde antique (eds. D. Foy and M.-D. Nenna), 33–9, Monographies Instrumentum, vol. 24, Editons Monique Mergoil, Montagnac. Jackson, C. M., Hunter, J. R., Warren, S. E., and Cool, H. E. M., 1990, The analysis of blue–green glass and glassy waste from two Romano-British glass working sites, in Archaeometry ’90. International Symposium on Archaeometry, Heidelberg, Germany 2–6 April 1990 (eds. E. Pernicka and G. A. Wagner), 295–305, Birkhäuser Verlag, Basel. Jackson, C. M., Wager, E. C. W., Joyner, L., Day, P. M., Booth, C. A., and Kilikoglou, V., 2003b, Small-scale glass making at Coppergate, York: analytical evidence for the nature of production, Archaeometry, 45, 435–56. Mass, J. L., Stone, R. E., and Wypyski, M. T., 1998, The mineralogical and metallurgical origins of Roman opaque colored glasses, in The prehistory and history of glassmaking technology (eds. P. McCray and W. D. Kingery), 121–44, Ceramics and Civilization, vol. VIII, American Ceramic Society, Westerville, OH. Mirti, P., Casoli, A., and Appolonia, L., 1993, Scientific analysis of Roman glass from Augusta Praetoria, Archaeometry, 35, 225– 40. Mondadori, A., 1983, The Macdonald encyclopaedia of rocks and minerals, trans. C. Atthill, H. Young and S. Pleasance, Macdonald, London. Mortimer, C., and Baxter, M. J., 1996, Analysis of samples of colourless Roman vessel glass from Lincoln, Ancient Monuments Laboratory Report 44/96. Nenna, M.-D., Picon, M., and Vichy, M., 2000, Ateliers primaries et secondaires en Égypte à l’époque gréco-romaine, in La route du verre. Ateliers primaries et secondaires du second millénaire av. J.-C. au Moyen Âge (ed. M.-D. Nenna), 97–112, Travaux de la Maison de l’Orient Mediterranéen no. 33, Lyon. Newton, R. G., 1980, Recent views on ancient glasses, Glass Technology, 21, 173– 83. Newton, R. G., 1985, W. E. S. Turner—recollections and developments, Glass Technology, 26(2), 93 –103. Paynter, S., 2004, Analyses of colourless Roman glass from Binchester, County Durham, Ancient Monuments Laboratory Report 21/04. Pliny, 1962, Natural history, trans. D. E. Eichholz, Heinemann, London. Price, A. J., and Cool, H. E. M., 1991, The evidence for the production of glass in Roman Britain, in Ateliers de verriers de l’antiquité a la periode pre-industrielle. Actes du 4ème Recontres de l’Association Francaise pour l’Archéologie du Verre, Rouen, 24–25 novembre 1989 (eds. D. Foy and G. Sennequier), 23 –9, l’Association, Rouen. Price, J., and Cottam, S., 1998, Romano-British glass vessels: a handbook, Practical Handbook in Archaeology 14, Council for British Archaeology, York. Sanderson, D. C. W., and Hutchings, J. B., 1987, The origins and measurement of colour in archaeological glasses, Glass Technology, 28(2), 99 –105. Sayre, E. V., 1963, The intentional use of antimony and manganese in ancient glasses, in (eds). Advances in glass technology, part 2 (eds. F. R. Matson and G. Rindone), 263–82, Plenum Press, New York. Sayre, E. V., and Smith, R. W., 1961, Compositional categories of ancient glass, Science, 133, 1824 – 6. Sayre, E. V., and Smith, R. W., 1967, Some materials of glass manufacturing in antiquity, in Archaeological chemistry: a symposium (ed. M. Levey), 279–311, University of Pennsylvania Press, Philadelphia, PA. Sellner, C., Oel, H., and Camera, B., 1979, Untersuchung alter Gläser (Waldglas) auf Zusammenhang von Zusammensetzung, Farbe und Schmelzatmosphäre mit der Elektronenspektroskopie und der Elektronenspinresonanz (ESR), Glastechnische Berichte, 52, 255–64. Smith, R. W., 1971, The analytical study of glass in archaeology, in Science in archaeology: a survey of progress and research (eds. D. R. Brothwell and E. Higgs), 614 –24, Thames and Hudson, London. Stawiarska, T., 2005, Roman period glass beakers with thread decoration (E188-192) from Poland, in Annales du 16e Congrès de l’Association Internationale pour l’Histoire du Verre, 75–9, AIHV, London. Stern, W. B., 1990, The composition of Roman glass: problems of non-destructive analysis, in Annales du 11e Congrès de l’Association Internationale pour l’Histoire du Verre, 37–41, AIHV, Amsterdam. Thompson, M., and Walsh, J. N., 1983, A handbook of inductively coupled plasma spectrometry, Blackie, Glasgow. Turner, W. E. S., 1940, Notes on British sources of sands for making colourless glass, Journal of Glass Technology, 24, 197–239. Turner, W. E. S., 1956, Studies in ancient glasses and glassmaking processes, part III; the chronology of the glassmaking constituents, Journal of Glass Technology, 40, 39T–52T. 776 C. M. Jackson Velde, B., and Gendron, C., 1980, Chemical composition of some Gallo-Roman glass fragments from central western France, Archaeometry, 22, 183–7. Velde, B., and Hochuli-Gysel, A., 1997, Correlations between antimony, manganese and iron content in Gallo-Roman glass, in Annales du 13e Congres de l’Association Internationale pour l’Histoire du Verre, Pays Bas, 28 août— 1 septembre 1995, 185–92, AIHV, Lochem, The Netherlands. Wedepohl, K. H., Gaitzsch, W., and Follmann-Schulz, A.-B., 2003, Glassmaking and glassworking in six Roman factories in the Hamback Forest, Germany, in Annales du 15e Congrès de l’Association Internationale pour l’Histoire du Verre, Corning, New York, 2001, 56 – 61, AIHV, Nottingham. Weyl, W. A., 1981, Coloured glasses, Society of Glass Technology, Sheffield. appendix The composition of ‘colourless’ glass from York, Leicester and Mancetter, UK Ba Cu Li Ni 1.82 1.86 1.75 1.88 1.83 2.37 2.34 2.30 2.31 2.22 2.43 2.31 2.24 1.78 2.33 2.32 2.30 1.97 2.00 2.04 0.37 0.41 0.31 0.30 0.30 0.57 0.35 0.55 0.56 0.36 0.44 0.57 0.63 0.34 0.51 0.55 0.56 0.28 0.46 0.42 0.40 0.50 0.35 0.35 0.33 0.55 0.59 0.53 0.56 0.44 0.56 0.52 0.57 0.38 0.56 0.55 0.55 0.37 0.51 0.43 5.59 6.25 5.15 4.45 4.20 6.12 6.22 5.99 6.17 6.30 7.47 6.01 5.47 5.78 6.49 6.10 6.09 5.52 5.60 5.58 0.08 0.08 0.06 0.07 0.07 0.10 0.10 0.11 0.11 0.06 0.07 0.10 0.12 0.07 0.10 0.11 0.11 0.05 0.07 0.07 0.04 0.05 0.04 0.04 0.04 0.11 0.11 0.10 0.12 0.15 0.15 0.11 0.11 0.05 0.12 0.11 0.12 0.03 0.04 0.04 0.02 0.03 0.02 0.02 0.02 0.37 0.35 0.33 0.41 0.96 1.22 0.30 0.40 0.02 0.44 0.36 0.35 0.01 0.06 0.06 0.25 0.01 0.01 0.01 0.01 0.06 0.06 0.04 0.05 0.01 0.01 0.03 0.05 0.01 0.04 0.04 0.04 0.02 0.06 0.02 0.63 0.74 0.57 0.61 0.52 0.47 0.47 0.45 0.47 0.15 0.03 0.41 0.53 0.49 0.37 0.45 0.47 0.64 0.50 0.69 141 140 140 155 146 214 207 203 214 244 280 202 202 138 224 209 207 145 151 163 9 5 6 7 6 132 125 81 124 14 17 54 73 7 68 63 142 13 58 15 14 16 14 13 12 18 17 17 20 19 14 18 19 15 16 17 17 9 14 14 2.06 2.04 0.35 0.25 0.38 0.30 5.70 19.86 0.59 0.05 5.60 19.66 0.63 0.04 0.03 0.02 0.02 0.01 0.27 0.39 0.74 0.88 158 165 1.84 0.26 0.34 5.06 19.15 0.54 0.04 0.03 0.01 0.02 0.65 145 1.83 0.35 0.53 5.80 19.62 0.45 0.06 0.03 0.01 0.02 0.61 133 20.02 19.76 18.91 18.82 18.97 19.74 19.50 19.59 19.49 17.53 14.20 17.76 19.86 19.44 18.92 19.19 18.70 19.61 19.30 20.24 0.51 0.50 0.45 0.50 0.44 0.71 0.70 0.80 0.70 0.63 0.63 0.65 0.68 0.53 0.69 0.74 0.71 0.64 0.53 0.64 10 13 10 11 8 17 17 16 16 26 28 15 19 11 17 16 19 12 14 14 Sr V Y Zn 379 446 360 286 266 409 406 396 412 426 497 384 392 454 427 410 400 386 382 358 10 10 8 8 9 17 18 18 18 20 28 15 19 8 20 19 18 7 11 11 7 7 6 6 6 8 8 8 8 8 9 8 8 6 8 8 8 7 7 7 Assigned group Context/ Colour Description SF no. 20 20 22 21 21 32 39 31 32 31 35 29 33 17 27 29 52 19 25 28 1 1 1 1 1 2a 2a 2a 2a 2b 2b 2a 2a 1a 2a 2a 2a 1a 1a 1a L19 L19 L19 L19 L19 L19 L19 L8 M64 M63 M64 M24 M24 M24 M63 M24 M63 14331 12266 13493 C C C C C NC NC NC NC NC NC NC NC C NC NC NC C C C 16 12 15 416 10 12 15 533 8 7 17 7 7 16 1a 1a 12461 11172 C C 8 12 13 359 8 7 15 1a 13955 C 13 12 14 415 11 7 20 1a 12582 C Water-rounded lump Pinched fragment Water-rounded lump Pinched fragment Water-rounded lump Cylindrical moile Cylindrical moile Pinched fragment Trail Cylindrical moile Bubbly fragment Trail Chip Chip Chip Chip Fragment Chunk Body fragment Cylindrical cup, Isings 85b, late second/third century Body fragment Facet-cut beaker, Isings 21 Cylindrical cup, Isings 85b, late second/third century Chunk Making colourless glass in the Roman period Al2O3 Fe2O3 MgO CaO Na2O K2O TiO2 P2O5 MnO PbO Sb2O5 777 778 appendix Ba 2.07 0.38 0.44 5.95 18.89 0.61 0.07 0.06 0.45 0.02 0.42 212 14 13 18 384 18 7 24 1/2a 10781 C 1.79 0.33 0.39 5.32 19.87 0.47 0.05 0.03 0.01 0.02 0.57 137 12 12 12 364 1a 13634 C 2.09 0.39 0.44 6.04 19.08 0.65 0.07 0.06 0.43 0.02 0.44 215 15 15 19 389 19 8 24 1/2a 10781 C 1.91 2.10 1.94 2.00 1.67 2.03 2.04 2.02 2.08 1.86 0.39 0.44 0.44 0.50 0.36 0.41 0.44 0.42 0.49 0.35 0.51 0.63 0.48 0.66 0.46 0.55 0.58 0.55 0.50 0.39 6.67 5.93 5.88 6.38 6.08 5.97 5.71 5.97 6.72 5.31 0.07 0.07 0.07 0.10 0.07 0.08 0.09 0.08 0.08 0.07 0.03 0.03 0.05 0.04 0.03 0.03 0.03 0.03 0.04 0.03 0.01 0.01 0.14 0.01 0.01 0.01 0.01 0.01 0.04 0.08 0.02 0.02 0.06 0.02 0.02 0.02 0.02 0.02 0.13 0.02 0.61 0.45 0.53 0.45 0.53 0.46 0.50 0.41 0.88 0.61 131 16 14 19 486 11 8 25 157 14 13 12 429 11 7 22 169 105 22 14 392 12 7 32 138 13 12 15 567 13 7 19 127 15 11 13 436 11 7 22 145 11 11 15 409 12 7 18 158 15 11 12 419 11 8 22 144 11 13 13 406 12 7 19 144 13 15 15 579 11 7 19 156 17 14 15 352 11 7 23 1a 1a 1a 1a 1a 1a 1a 1a 1a 1a 12287 14130B 14291A 11993 14321A 11984 14264B 11894 11420 8625 C C C C C C C C C C 1.84 0.37 0.43 5.47 19.47 0.51 0.07 0.04 0.13 0.02 0.58 162 19 16 16 356 12 7 25 1a 8207 C 2.05 1.95 0.44 0.34 0.65 0.41 5.91 19.21 0.40 0.08 5.50 20.51 0.48 0.07 0.03 0.03 0.01 0.01 0.02 0.01 0.45 0.61 156 158 12 13 15 424 11 7 16 10 12 13 396 9 7 15 1a 1a 12201 13657 C C 1.67 2.05 1.98 1.87 0.34 0.43 0.44 0.28 0.46 0.57 0.55 0.32 5.67 6.01 5.95 4.49 0.03 0.04 0.05 0.03 0.01 0.10 0.09 0.01 0.02 0.06 0.08 0.12 0.89 0.48 0.46 0.72 132 164 156 143 11 13 12 394 9 7 16 61 25 13 416 11 7 25 58 18 11 398 10 7 26 15 7 11 286 5 6 16 1a 1a 1a 1a 13443 14311A 14311B 14346B C C C C 19.61 18.81 18.44 20.03 18.62 19.26 20.38 19.49 19.62 18.72 19.68 19.75 19.84 18.60 0.54 0.44 0.63 0.45 0.47 0.45 0.39 0.42 0.56 0.49 0.52 0.55 0.52 0.44 0.06 0.07 0.08 0.06 Cu Li Ni Sr V Y Zn 9 7 22 Assigned group Context/ Colour Description SF no. Cylindrical cup, Isings 85b, late second/third century Cylindrical cup, Isings 85b, late second/third century Cylindrical cup, Isings 85b, late second/third century Body fragment Chunk Chunk Chunk Body fragment Chunk Chunk Chip Body fragment Body fragment (fourth century?) Beaker/flask, late second century Body fragment Globular flask, late second/early third century Chunk Chunk Chunk Body fragment C. M. Jackson Al2O3 Fe2O3 MgO CaO Na2O K2O TiO2 P2O5 MnO PbO Sb2O5 Ba Cu 2.13 0.35 0.52 6.54 19.72 0.62 0.07 0.04 0.03 0.00 0.49 140 18 8 8 7 30 1a 12103 C 1.84 1.62 2.07 1.85 1.99 0.22 0.25 0.33 0.29 0.31 0.33 0.43 0.46 0.41 0.43 5.13 5.13 5.43 5.51 5.70 0.05 0.06 0.07 0.06 0.06 0.03 0.03 0.06 0.04 0.05 0.02 0.02 0.16 0.11 0.11 0.00 0.00 0.01 0.01 0.01 0.40 0.74 0.42 0.41 0.39 142 129 172 152 162 8 5 23 17 17 5 9 354 6 7 16 8 10 358 8 6 15 7 11 355 11 7 28 8 9 355 9 7 23 9 10 377 9 7 23 1a 1a 1a 1a 1a 13835 10864 12392 13288 12037 C C C C C 1.97 1.87 0.29 0.30 0.45 0.48 6.73 19.62 0.52 0.07 6.90 20.62 0.52 0.07 0.04 0.05 0.02 0.09 0.00 0.01 0.05 0.41 138 151 13 19 6 9 9 469 9 8 21 8 406 10 7 28 1b 1a 14108 13871 C C 1.90 2.01 0.26 0.30 0.41 0.43 5.32 19.83 0.51 0.05 5.59 18.86 0.62 0.06 0.03 0.06 0.02 0.21 0.00 0.01 0.29 0.35 137 174 6 6 24 11 9 359 6 7 19 9 358 10 7 23 1a 1a 13351 13995 C C 1.76 0.23 0.37 5.60 18.69 0.42 0.06 0.03 0.02 0.00 0.45 137 12 8 384 7 7 18 1a 12457 C 2.10 0.36 0.53 6.14 19.25 0.66 0.08 0.06 0.20 0.03 0.45 176 56 13 10 403 13 7 32 1a 12252 C 1.93 2.01 2.12 1.78 1.79 1.95 0.29 0.34 0.34 0.27 0.24 0.42 0.42 0.45 0.50 0.39 0.36 0.46 5.54 5.43 5.53 5.19 5.16 5.84 19.59 19.94 18.60 19.62 19.06 19.64 0.55 0.55 0.37 0.47 0.36 0.62 0.07 0.06 0.08 0.05 0.06 0.08 0.05 0.04 0.03 0.04 0.03 0.06 0.11 0.06 0.02 0.10 0.02 0.24 0.00 0.04 0.00 0.01 0.00 0.02 0.42 0.52 0.31 0.49 0.35 0.53 156 158 146 149 135 185 14 10 7 362 9 14 10 9 370 8 7 7 11 358 9 16 8 8 334 7 8 6 8 326 6 21 9 12 388 13 1.87 1.83 1.76 2.18 1.92 0.37 0.31 0.32 0.55 0.27 0.49 0.37 0.40 0.64 0.32 5.77 5.51 5.45 6.43 5.59 18.48 19.21 19.37 19.61 18.44 0.59 0.47 0.44 0.79 0.46 0.08 0.06 0.07 0.09 0.05 0.04 0.04 0.03 0.07 0.03 0.03 0.02 0.02 0.26 0.02 0.01 0.01 0.01 0.03 0.24 0.64 0.40 0.53 0.40 0.57 149 141 136 214 146 5 8 10 14 6 10 11 9 11 62 28 13 8 6 10 18.55 19.37 19.48 19.42 19.01 0.48 0.42 0.59 0.49 0.55 Li Ni 6 Sr 8 504 V Y Zn Assigned group Context/ Colour Description SF no. 7 7 7 6 7 7 24 21 19 26 18 27 1a 1a 1a 1a 1a 1/2a 14309A 14310A 14320 15522 14308A 9707 C C C C C C 400 9 6 345 7 6 370 8 6 430 15 8 401 6 6 18 19 24 40 16 1a 1a 1a 2a 1a 13992 9501 13778 11325 13864 C C C NC C Cylindrical bottle, late second/third century Body fragment Chip Body fragment Body fragment Wheel-cut beaker, late first/second century Body fragment Cylindrical cup, Isings 85b, late second/third century Body fragment Cylindrical cup, Isings 85b, late second/third century Body fragment (fourth century?) Body fragment (fourth century?) Body fragment Body fragment Body fragment Body fragment Body fragment Vertical-ribbed jug, late second/third century Wheel-cut bowl Circular facets Body fragment Body fragment Body fragment Making colourless glass in the Roman period Al2O3 Fe2O3 MgO CaO Na2O K2O TiO2 P2O5 MnO PbO Sb2O5 779 780 appendix Ba Cu 1.86 0.37 0.45 5.81 18.04 0.46 0.07 0.04 0.02 0.01 0.65 143 22 7 12 403 10 7 36 1a 8442 C 1.99 2.00 0.47 0.41 0.53 0.47 6.03 19.70 0.63 0.09 6.01 19.22 0.51 0.08 0.06 0.04 0.07 0.03 0.03 0.01 0.72 0.61 154 146 14 14 7 13 409 10 7 25 6 8 438 9 7 24 1a 1a 8516 13716 C C 1.97 2.27 0.33 0.73 0.47 0.77 4.99 17.99 0.49 0.07 5.53 19.99 0.85 0.13 0.03 0.07 0.02 0.92 0.01 0.04 0.25 0.16 150 231 6 6 8 324 8 6 15 84 22 18 394 24 8 38 1a 2b 8674 12628 C NC 2.23 2.65 2.12 2.16 0.71 0.44 0.51 0.69 0.74 0.59 0.52 0.77 5.95 8.10 6.05 5.64 0.12 0.07 0.08 0.12 0.07 0.08 0.07 0.07 1.00 1.06 0.22 0.87 0.03 0.02 0.30 0.03 0.08 0.08 0.54 0.16 318 85 24 581 25 11 183 350 27 239 124 18 44 18 60 35 2b 2b 2a 2b 10315 12788 9718 13770 NC NC NC NC 2.19 2.25 0.35 0.75 0.66 0.81 5.91 18.92 0.43 0.08 7.66 20.08 0.59 0.13 0.03 0.05 0.02 0.93 0.00 0.01 0.31 0.11 152 229 11 8 10 433 9 7 16 11 13 19 675 28 8 19 2? 2b 14309B 12096 NC NC 19.76 17.78 19.37 18.85 0.98 0.74 0.67 0.75 C = colourless; NC = nearly colourless. Context/SF no.: L = Leicester; M = Mancetter. All other samples are from York. Li Ni 21 16 16 20 Sr 430 461 401 418 V 22 21 15 24 Y Zn 8 9 7 7 Assigned group Context/ Colour Description SF no. Cylindrical cup, Isings 85b, late second/third century Chip Cylindrical bottle neck Body fragment Isings 96/106, fourth century Body fragment Body fragment Body fragment Body fragment, fourth century Body fragment Body fragment C. M. Jackson Al2O3 Fe2O3 MgO CaO Na2O K2O TiO2 P2O5 MnO PbO Sb2O5