Introduction

In recent years it has become apparent that laboratory examinations of ancient glass might provide archaeologists and historians with valuable information about both ancient glass and the history of technology. At present several scientists working in this field are developing those techniques which show most promise of being useful, and are accumulating the necessary background of data against which new research results can be evaluated. Perhaps the greatest need on the part of these workers is the continued cooperation of archaeologists and museum curators. This cooperation should reward us in the future with additional new techniques and approaches for studying archaeological finds.

In this paper are reported the findings of a laboratory investigation of two fragments of the Agora plate described in Dr. Gladys Davidson Weinberg's article, "An Inlaid Glass Plate in Athens" (Journal of Glass Studies, v. 4, pp. 28–36). It will be evident that even as extensive an investigation as this one does not at present tell us everything we would like to know about the object studied. However, some conclusions of definite interest have been drawn about the Agora plate and its method of manufacture. Although several experimental methods were employed, the two which proved most useful for this particular object were chemical analyses and microscopic examination. In addition, a batch of glass was synthesized in the laboratory, duplicating the chemical composition of the Agora plate. Experiments conducted with this glass have allowed us to learn something about the conditions under which the plate was manufactured.

Description of Samples

The fragments are covered with weathering crusts which appear to be almost one millimeter thick at the points of greatest thickness. The weathering crusts are buff-colored but in certain regions they are covered with patches of a very thin outer layer of dark brown color. The crusts have been partially worn away from most of the surfaces leaving a faintly iridescent appearance in some areas. A small sample was cut away from the end of one fragment, revealing that the glass itself is light blue in color. Although it is somewhat turbid, it does not appear to be a truly opaque glass.

Microscopic Examination

When glass objects become weathered the decomposition often follows inhomogeneities in the original glass and produces clearly defined patterns in the weathering products. Thus the structure of weathering crusts records very vividly the structural inhomogeneities in the original glass. In the case of the Agora plate the weathering process seems to have created or emphasized swirled patterns on the weathered surface which give the impression that the object was formed by fusing together small mosaics of glass, as was done in making millefiori objects. Discontinuities were found in some areas which apparently mark the boundaries between what once were separate mosaics. A careful microscopic investigation showed that the "grain" on the edges of each fragment is generally perpendicular to the flat surfaces of the plate. No evidence was found that the surface of the plate had been ground or polished, but such evidence may have been obscured by the heavy weathering.

A small sample of one fragment was mounted in plastic and polished on a metallographic polishing wheel. In cross-section the weathering crust can be seen to average about 0.5 millimeter in thickness. Although it shows a layered structure, the weathering crust clearly is not completely intact, and therefore the object cannot be elated by the counting technique.¹ By transmitted light the glass is seen to be very heterogeneous. As is shown in Fig. 1, the portion of the glass examined appears to consist of an almost colorless base glass with turbid streamers of blue color passing through. The glass was quite obviously made by mixing (not very thoroughly) a blue glass, which was probably opaque, with a relatively clear and colorless glass.

The glass contains a great many tiny inclusions, some of which have a crystalline appearance. They are of irregular shape and non-uniform sizes, the smaller inclusions may be particles of opacifier, although there certainly seems to be an abundance of extraneous foreign matter in addition.²

A great many air bubbles are trapped in the glass. Without exception these are spherical. The photomicrograph in Fig. 2 shows some of these bubbles.

In the areas examined, the turbid streamers of blue glass run roughly perpendicular to the flat surfaces of the plate. This provides an important clue to the possible method of manufacture.

A few tiny inclusions too small to be seen by the unaided eye were investigated with the electron microbeam probe.³ The results on this particular sample were somewhat ambiguous but seem to indicate that the one inclusion analyzed was a fragment of unfused glass of a composition different from the rest of the glass. It appeared to have a higher SiO₂ content (75–80 percent), a higher CaO content (10 percent), and lower alkali content (Na₂O + K₂O = 9 percent). Such a composition would melt at a higher temperature than the base glass and thus it might represent the imperfectly fused remains of a piece of scrap glass used in the manufacture of the base glass.

Chemical Analyses

Three totally independent analyses of the glass were made at three different laboratories.⁴ Two of these were spectrographic analyses. The third was done by methods of wet chemistry. The results shown in the table are a composite of the three separate analyses, all of which agreed very well with one another.⁵ In addition, separate spectrographic analyses were made of the weathering crust, the dark outer surface of the weathering crust, and a tiny speck of inlay which had become dislodged from the larger fragment. The results of these analyses are shown in Table 1.

Table 1: Results of Chemical Analyses

		Unweathered Glass	Light Inner Weathering Crust	Dark Outer Weathering Crust	Inlay Fragment
Major Constituents
Silica	SiO2	64-88	Major	—	Major
Soda	Na2O	18.1	0.2	0.2	—
Lime	CaO	6.8	2	2	—
Minor Constituents
Potash	K2O	0.49	n.d.	n.d.	0.X
Magnesia	MgO	0.80	0.5	0.5	1-5
Alumina	Al2O3	2.0	2	≤1	Major
Iron oxide	Fe2O3	≃0.5	≃0.4	≃0.3	0.X
Manganese dioxide	MnO2	0.X	< glass	>3x light crust	0.0X
Titanium oxide	TiO2	≃0.01	≃0.01	≃0.02	0.0X
Coloring Agents
Copper oxide	CuO	≃0.5	≃0.5	≃0.5	1-5
Cobalt oxide	CoO	≃0.002	n.d.	≤0.002	n.d.
Antimony oxide	Sb2O5	0.87	0.5	0.3	0.01-.05
Other Oxides
Lead oxide	PbO	≃0.5	0.2	0.1	≤0.5
Tin oxide	SnO2	≃0.1	0.1	0.3	0.5
Strontium oxide	SrO	0.0X	—	—	0.0X
Barium oxide	BaO	0.0X	—	—	0.0X
Zirconium oxide	ZrO2	0.00X	—	—	—
Silver oxide	Ag2O	≃0.005	—	—	—
Arsenic oxide	As2O5	0.00X	—	—	0.00X
Boric oxide	B2O3	≃0.01	—	—	—
Chromic oxide	Cr2O3	≃0.002	—	—	—
Nickel oxide	NiO	≃0.004	—	—	—
Zinc oxide	ZnO	0.01(?)	—	—	—
Phosphorus pentoxide	P2O5	n.d.	—	—	—
Vanadium oxide	V2O5	0.005(?)	—	—	—
Density=2.544 gm./c.c.			Coefficient of Expansion (0°-300°)=103.4x10-7/°C.
n.d.=not detected			<=less than
≃=approximately			>=greater than
≤ equal to or less than			≥ equal to or greater than

The composition of the glass is about what one would predict for a glass of the Roman period. However, except for the elements antimony and manganese, which are discussed below, the composition is not uniquely Roman. Glasses of considerably earlier dates could have very much the same composition. The potash (K₂O) is low, indicating with certainty that the source of alkali was not plant ash from plants grown in an ordinary inland environment and probably was not ash from plants grown in a coastal environment. The source of alkali most likely to produce such a preponderance of soda over potash is natron, although ashes from at least some desert plants show a similar soda/potash ratio.⁶ The whole problem of alkali sources is one which deserves much further experimental attention.

The principal coloring agent is copper oxide, although the level of cobalt oxide in the glass is sufficient also to contribute somewhat to the blue color. The level of cobalt oxide (0.002%) is low enough to make one question its presence as an intentional additive, but since it is believed that the color was produced by adding a blue glass to a relatively colorless glass, with its consequent dilution, it is reasonable to assume that cobalt oxide was an intentional additive to the original blue glass component.

It has recently been shown by E.V. Sayre and R.W. Smith⁷ that ancient glasses can be divided into five different compositional categories which seem to reflect different traditions in glassmaking. These traditions, which are separated chronologically, are based on the concentrations of MgO, MnO, K₂O, Sb₂O₅ and PbO in the glasses.⁸

The two groups pertinent to this discussion are those described by Sayre and Smith as the "Antimony-Rich Group" which dates roughly from the 6th century B.C. to the 4th century A.D., and the "Roman Group," dating from the 4th century B.C. to the 9th century A.D.⁹ The glasses in these two groups are distinguishable from one another by relatively high antimony or relatively high manganese concentrations. (Sb₂0₅ ≃ 1–3 percent or MnO ≃ 1 percent). It is tentatively believed that these two ingredients were used as decolorizers¹⁰ to remove the greenish color imparted to most glasses by iron impurities present in raw materials. There also appear to be glasses which contain relatively large amounts of both antimony and manganese, which may represent a transitional period,¹¹ during which a change-over from antimony to manganese was being made.

The composition of the Agora plate would place it in the transitional group containing both high antimony and high manganese, except that the turbid nature of the blue streamers suggests strongly that the blue colorant in the glass was originally an opaque glass. This seems to demand that the presence of at least part of, and possibly all of, the Sb₂O₅ be attributed to its intentional addition as an opacifying agent instead of a decolorizer. It has been well established that Sb₂O₅ in combination with CaO was used as an opacifying agent during ancient times.¹²

The most satisfactory interpretation seems to be that the source of high antimony in the final composition of the Agora plate was the opacifying agent used in the blue colorant, and that probably the high manganese content resulted from the use of manganese dioxide as a decolorizer for the colorless glass to which the blue colorant was added. The objection might be raised that no decolorizer would he used for a blue plate, but the original clear glass might well have been a glass prepared earlier for other uses. The re-use of scrap glass or cullet is a factor which must constantly be kept under consideration. It could and possibly already has caused erroneous conclusions to be drawn from analytical data.

All this leads to the conclusion that the composition of the Agora plate places it most properly in the "Roman Group," although the earlier group cannot be completely ruled out. At present no clear-cut time limits have been established for this "Roman Group." We can only hope that with the further accumulation of analytical data from well-authenticated samples it will become possible to arrive at more or less precise time limits for this and the other groups.

The fragment of inlay which had become dislodged from the plate had been very thoroughly weathered and the analysis could not be expected to tell much about the glass itself. Indirectly, however, by comparing the composition of this weathered inlay fragment with the composition of the weathering crusts on the blue base glass it is possible to draw some conclusions, tentative though they may be, about the original color of the inset. The most apparent differences between the weathering crusts are that the inset apparently is richer in Al₂O₃, MgO, CuO, and SnO₂, while it is quite a bit lower in Sb₂O₅. If the inlay had originally been a yellow opaque or a white opaque glass it would be expected to have a higher Sb₂O₅ content in its weathered remains than the weathering crust of the plate itself, since Sb₂O₅ was a constituent of both the yellow and white opacifiers in most common use during the Roman period. The higher concentration of copper in the inset may imply that it had originally been a blue or perhaps an opaque red glass, since copper was used for producing both of these colors. Generally copper is rather thoroughly removed from glass by the weathering process so that a concentration of one to five percent of CuO in the residue implies that there had originally been a very high concentration of copper. From this one might conclude that the glass had been an opaque red since these glasses often contained in excess of 10 percent copper. A color rendering of the plate shows the inlay parts as being buff-colored. This is the present color of the weathering crust, but the original glasses undoubtedly were other colors.

The presence of tin was unexpected and might mean that a tin oxide opacifier was used in the original glass, or it might just be that the tin is associated with the copper, as would be the case if a piece of bronze had been used as a source of copper. In a recent paper¹³ H.P. Rooksby has reported a yellow opacifier with a composition approximating Pb₂SnO₄. Mr. Rooksby believes that the use of antimony oxide as an opacifier was abandoned in favor of tin oxide and that tin oxide was used in the period from the fourth century A.D. to the seventeenth century.

The analysis of the weathering crust indicates that virtually all of the Na₂O and K₂O was removed by the weathering process. This process consists of the leaching out, or dissolving, of the "more soluble components" of the glass by soil water.¹⁴ In addition, considerable portions of the CaO, MgO, and CuO were removed, along with part of the Fe₂O₃, and MnO. The difference in color between the weathering crust and the remains of the dark outermost surface can be attributed to the redeposition of leached cobalt and manganese, both of which form dark colored oxides.

Suggested Method of Manufacture

As Dr. Weinberg has suggested, the Agora plate appears to have been fashioned by methods similar to those used in making millefiori objects. There is considerable evidence which is consistent with this suggested means of manufacture, some of which has already been mentioned. Consider, for example, the observation that the turbid streamers of blue glass found in the microscopic examination of the sample run roughly perpendicular to the flat surfaces of the plate, at least in those areas which were examined. There seems to be no other way to account for this unusual perpendicular "grain" of the glass except to assume that the plate was formed by fusing together separate mosaics cut or broken from an elongated ingot or canes of stock glass. To visualize this grain, imagine a circular branch of wood sawed into flat discs. If the discs were laid flat and arranged to form a plate, the grain of the wood would be perpendicular to the flat surfaces. The appearance of a cross-section of such a plate would be similar to that of the plate from the Agora. The linear blue streamers which would have been formed initially by elongation of a gather of poorly-mixed molten glass into an ingot are analogous to the grain of the wood. There is some question as to just how an ingot, which in this case must have been about two centimeters in diameter, would have been cut or broken into individual mosaics.

Another interesting observation is that the air bubbles trapped in the glass are perfectly spherical despite the fact that the glass shows evidence of having been elongated into an ingot. When air bubbles are trapped or gas bubbles are formed in molten glass, the bubbles invariably take on a spherical shape. This is because the surface tension of a liquid always tends to reduce to a minimum the exposed surface area of the interface between a gas and a liquid. For a given volume of gas the geometrical configuration which has the minimum surface area is a sphere. (For example, gas bubbles in carbonated beverages always seem to have perfectly spherical shapes, and droplets of mercury assume a spherical shape except for the slight deformation caused by gravity.) If a gather of molten glass containing bubbles is elongated into a rod or cane, the shape of the bubbles is distorted, and they are elongated to ellipsoidal shapes, or, if the extension is great enough, into long hollow channels running through the length of the glass. These distorted shapes are characteristic of the bubbles seen in many manufactured glass objects, since after shaping the glass, which in a sense is always a distortion operation, the glass is usually cooled at a rate which is too quick to permit surface tension to restore the bubbles to a spherical shape. In general, smaller bubbles are less subject to distortion than larger bubbles, since their smaller radii and greater surface to volume ratios accentuate the relative importance of the surface tension forces which oppose distortion.

At first glance, then, it is surprising to see such perfectly shaped spherical bubbles in the glass of the Agora plate. The formation of such a very bold grain in the glass would certainly have been accompanied by distortion of the air bubbles. On further thought, however, the sphericity of the bubbles is consistent with the suggested method of manufacture. In order to fuse the mosaics together, to fashion the plate itself, and to soften the plate so that the inlay design could be pressed into it, the glass must have been heated at least two or three times. As will be shown below, these reheating processes would soften the glass sufficiently to restore the bubbles to spherical shapes. The microscopic examination of Roman millefiori fragments usually shows the same effect; that is, perfectly spherical bubbles despite the obvious linear distortion of the fabricated glass.

It is not necessarily the case that the plate was formed into its final shape at the same time that the mosaics were fused together. It is quite possible that the individual mosaics were fused into a flat plate of glass and that this plate was later shaped into the final form by sagging the glass into a ceramic mold. By investigating the distortion of individual mosaics on millefiori objects, it should be possible to determine whether or not the shaping and fusion steps were simultaneous.

Assuming that the inlay design was affixed by pressing a preformed mosaic plaque into the softened plate, one must postulate a reheating process to soften the plate for each separate inlay operation. These reheating operations must have brought the glass to temperatures of 765° C. or above. This becomes a rather tricky operation if all of the insets are not placed into the glass at the same time, since a reheating operation might seriously distort any inlay pattern that had been placed in the plate previously.;

Experiments with a "Synthetic" Glass

In order to learn something about the conditions under which the Agora plate was manufactured, a batch of glass having the same composition was synthesized in the laboratory. Two series of experiments were conducted with this glass to determine the conditions under which mosaics can be fused together to form glass objects and to define what temperature and time relationships will restore bubbles to sphericity. Both of these series are concerned with the softening of glass, and it is therefore necessary to make use of the concept of viscosity and viscosity-temperature relationships in glasses. A brief description of these topics is given in this Journal in "A Note on the Scientist's Definition of Glass." The physical properties measured for this synthetic glass are reported in the same note. These include the viscosity-temperature curve for the glass, the coefficient of expansion, and the density.

Fusion Experiments. A number of solid canes were draw from the experimental glass and sawed into small mosaics. Each mosaic disc had a roughly circular cross-section about one centimeter in diameter, and was about seven millimeters thick. The upper and lower flat surfaces had been roughened by the sawing but the cylindrical surfaces of the mosaics were the original surfaces of the drawn canes. The mosaics were placed in small circular ceramic molds and heated for one hour at varying temperatures ranging from 750° C. to 850° C. Each mold contained about twenty mosaics packed tightly in a single layer on the bottom of the mold. Figure 3 shows the results of this experiment.

The mosaics heated at 750° C. were slightly deformed by the heating and became fused to one another at the points of contact between the individual mosaics. At 765° the mosaics were more deformed but large gaps still remained between them. Those heated at 775° C. had deformed sufficiently so that there were no gaps remaining, but the over-all surface still preserved the separate contours and identities of the individual mosaics.

The mosaics heated at 800° C. flowed together to form one larger disc of glass. The roughened sawed surfaces of some of the mosaics were still visible as cloudy areas on the surface of the disc. The mosaics fused at 850° flowed into one smooth disc and still showed faint traces of the roughened surfaces. By examining these discs with transmitted light it is still possible to discern visible lines of demarcation between what had originally been separate mosaics. It is worth noting here that when one examines millefiori objects the individual mosaics sometimes appear to have hexagonal or pentagonal shapes. This does not mean that the original mosaics had hexagonal or pentagonal cross-sections. If canes of circular cross-section are packed into an area and then fused, the glasses flow together and form a series of polygons which are generated by the planes of intersection formed by the deforming canes.

The results of this series of experiments indicate that in order to fuse separate mosaics of a glass having the composition of the Agora plate it is necessary to heat the mosaics to a temperature of 785° C. or greater. Since the viscosity-temperature relationship for this glass is known it is possible to draw the more general conclusion that in order to fuse individual mosaics of glass into a new piece having a smooth surface it is necessary to heat the glass to at least the temperature at which the viscosity becomes 7 x 10⁵ poises.¹⁵ This is for a heating time of one hour; it is not likely that reasonably prolonged heating at a much lower temperature would accomplish the same purpose.

"Sphericalization" Experiments. The "sphericalization" process has also been verified experimentally. Two glass pitcher handles of Islamic origin were chosen for experimentation. Their chemical compositions are probably quite similar to that of the Agora plate. Each fragment was a long cane of glass approximately one centimeter in diameter and about twelve centimeters long, bent into a hook shaped handle. In each were observed hundreds of elongated air bubbles. In one fragment, as shown in Fig. 4, the bubbles were only slightly elongated into a shape roughly resembling that of a football (North American variety). In the other fragment, as seen in Fig. 6, the bubbles had been extended to such a degree that they were ellipsoids which were sometimes 10 to 20 times longer than their thicknesses. Samples cut from these fragments were heat-treated at various temperatures for various times and the general conditions required for the restoration of sphericity were established. The results are summarized below and illustrated in Fig. 4 through 8.

1. Heating for one hour at a viscosity of 10⁷ poises is not sufficient to restore the larger bubbles to sphericity.

2. Heating for four hours at a viscosity of 10⁷ poises does restore sphericity.

3. Heating for one hour at a viscosity of 10⁶ poises does restore sphericity.

Knowing the viscosity-temperature relationship for the glass of the Agora plate it can be concluded that, for the bubbles to have been restored to spherical shapes, the plate must have been heated to a temperature of about 770° C. or greater for one hour, or 710° C, for four hours or longer. One hour at 710° C. would not be sufficient to restore the bubbles to a spherical shape. The heating step required to fuse the mosaics together (a minimum temperature of 785° C.) would certainly have been sufficient to restore the bubbles to sphericity.

Melting Temperature. Accepting Turner's estimate¹⁶ that the maximum temperature available to ancient glassmakers was something less than 1200° C., then it is very unlikely that the Agora plate could have been made by any sort of casting process. At 1200° C. the viscosity of the molten glass would have been about 400 poises, which means that the glass would not have been fluid enough to have been poured into a mold at that temperature. Having such a viscosity the glass would have poured so slowly that it would have cooled sufficiently during the pouring to have become a taffy-like mass before the pouring could have been completed. During experiments made when synthesizing the experimental glass used in these experiments, Dr. August Erickson, who conducted the melting, attempted to pour the glass at a temperature of 1250° C. into a flat open mold. He remarked that the glass was so viscous that it could not be poured from the crucible and would have just slid out slowly as one stiff blob.

Professor Turner has done some work on the determination of the temperatures required to manufacture Egyptian XVIIIth Dynasty glasses.¹⁷ It should be worthwhile to extend this line of research to other ancient compositions by studying various mixtures of raw materials, fritting operations, and melting conditions. Preliminary experiments of this sort have been conducted at this laboratory using a gradient furnace. Various mixtures of raw materials can be placed in a platinum boat which is then placed in an electric furnace. The furnace is constructed in such a way that one end of the boat is at a higher temperature than the other, and the temperature at any given point between the two extremes can be accurately measured. A series of experiments was conducted using a mixture of sand, “synthetic natron," and lime. The over-all temperature gradient ranged from 952° to 1405° C. The proportions of the raw materials were chosen so that the resulting glass would have a composition very closely resembling that of the Agora plate, and the typical Roman glass mentioned in "A Note on the Scientist's Definition of Glass." The experiment showed that a minimum temperature of 1316° C. was necessary to melt these raw materials completely to a glass, allowing four hours of melting time. However, in the temperature range of 1210°–1315° C. only a very few grains of unreacted sand remained in the resulting glass after four hours. With an increase in melting time to the order of twenty hours, an acceptable glass could probably be made without exceeding 1210° C. On the other hand, it appeared that below 1000° C. it would be virtually impossible to melt a glass of this composition from these raw materials. At intermediate temperatures of the order of 1100° C., one or two successive melting operations, allowing longer melting times and possibly including some cullet, might have allowed ancient glassmakers to manufacture a creditable glass from these raw materials.

A similar experiment was conducted using a batch of identical raw materials which had been "fritted" or sintered beforehand by heating in an open tray for four hours at 700° C. This treatment caused the batch materials to clump together somewhat into a form having a more granular texture. The fritted material melted completely at 1270° C., but aside from that difference, showed final results identical to the unfritted mix at all other temperatures. The fritted material did show one advantage, as was anticipated, in that during the melting of the glass the reaction proceeded more smoothly, since the ingredients had undergone the major part of their gas evolution during the fritting operation. Although one would hardly expect that a fritting operation would permit a given batch of raw materials to be melted to a glass at lower temperatures, nevertheless the fritting does appear to promote greater "meltability" and leads to convenience in handling.

These experiments are being continued to define more precisely the probable temperature requirements for making ancient glasses.

The laboratory examination of this object has perhaps raised more questions about ancient glassmaking than it has answered. With further examination of fragments of this. plate it would be possible to tell considerably more about its method of manufacture and the nature of the glasses used. In the coming months inlay objects and millefiori glasses will be investigated extensively at this laboratory in the hope that the examination of such objects may lead to laboratory experiments which will provide further insight into the techniques used in making these remarkable objects.

---

This article was published in the Journal of Glass Studies, Vol. 4 (1962), 37–47.

---

Acknowledgment: The author thanks Dr. August A. Erickson, Supervisor of Glass Melting Services, Corning Glass Works, and his co-workers, for preparing the experimental glass used in this study, and for his advice and assistance in conducting the fusion and sphericalization experiments. The author is also indebted to the following people for the parts they played in carrying out the chemical analyses and physical measurements: Dr. Robert H. Bell of Lucius Pitkin, Inc.; Mr. Robert H. Close, Miss Doris Evans, Mr. Edward Friebis, and Mr. Arthur Malenfant, all of Corning Glass Works. Mr. Christos Roulidis of Corning Glass Works conducted the gradient furnace experiments.

1.R.H. Brill, "The Record of Time in Weathered Glass," Archaeology, 14, No, 1, Spring 1961, pp. 18–22; and R.H. Brill and H.P. Hood "A New Method for Dating Ancient Class," Nature, 189, January 7, 1961, pp. 12–14.

2. X-ray diffraction studies of a very minute sample of the powdered glass confirmed that some of the material is indeed crystalline. Intensity maxima were found which indicate lattice spacings of 1.76, 2.70, 2.80 and 3.17 A°, but it was not possible to identify the crystals specifically. It is certain that the spacings found do not correspond to the opacifier Ca₂Sb₂O₇ which is mentioned later in this paper. This opacifier, if present, is in such a dilute concentration, however, that X-ray experiments probably would not have disclosed its presence.

3. R.H. Brill and S. Moll, "The Electron Beam Probe Microanalysis of Ancient Glass," in: The International Institute for Conservation of Historic and Artistic works, [Conference Papers] Rome Conference 1961, mimeographed, pp. 260–276.

4. Lucius Pitkin, Inc., N.Y.C.; Spectrographic Laboratory, Corning Glass Works; Analytical Laboratory, Corning Glass Works.

5. After completion of these analyses, it was learned that Dr. E.V. Sayre of Brookhaven National Laboratories had done a spectrographic analysis of a sample believed to have come from the same object. The results of his analysis check these results very well.

6. W.E.S. Turner, "Studies in Ancient Glasses and Glassmaking Processes. Part V. Raw Materials and Melting Processes," Journal of the Society of Glass Technology, XL, 1956, pp. 277F–3OOT.

7. E.V. Sayre and R.W. Smith, "Compositional Categories of Ancient Glass," Science, 133, June 9, 1961, pp. 1824–1826.

8. These are the oxides of magnesium, manganese, potassium, antimony and lead.

9. Notice that these overlapping dates correspond to two traditions. The name "Roman" is intended to mean a tradition which apparently culminated in Roman glass but which extends in time beyond the time limits implied by the phrase "Roman period."

10. The decolorizing properties of manganese are well known, and MnO₂ is still in use today as a decolorizer. The case for Sb₂O₅ is somewhat less certain, since the oxidizing and reducing properties of antimony oxides in glasses seem to be more complicated. Sb₂O₅ is also useful as a "fining" agent. At high temperatures it evolves oxygen gas which sweeps out small undesirable glass bubbles from molten glass.

11. E.V. Sayre, private communication to the author, November 22, 1961; and unpublished results of the author.

12. R.H. Brill and S. Moll, "The Electron Beam Probe Microanalysis of Ancient Class," in: The International Institute for Conservation of Historic and Artistic Works, [Conference Papers] Rome Conference 1961, mimeographed, pp. 260–276; and W.E.S. Turner and H.P. Rooksby, "A Study of the Opalising Agents in Ancient Opal Classes Throughout Three Thousand Four Hundred Years," Glastechnische Berichte, Vol. 32K, No. VIII, 1959, pp. VIII/17–28.

13. H.P. Rooksby, "Opacifiers in Opal Glasses through the Ages," G.E.C. Journal of Science and Technology, 29, No. 1, 1962, pp. 20–26.

14. R.H. Brill, Archaeology, 14, No. 1, Spring 1961, pp. 18–22; and R.H. Brill and H.P. Hood, Nature, 189, January 7, 1961, pp. 12–14.

15. 7x10⁵ = 700,000 poises.

16. W.E.S. Turner, Journal of the Society of Glass Technology, XL, 1956, pp. 277T–300T.

17. W.E.S. Turner, “Studies of Ancient Glass and Glassmaking Processes. Part I. Crucibles and Melting Temperatures Employed in Ancient Egypt at about 1370 B.C.," Journal of the Society of Glass Technology, XXXVIII, 1954, pp. 436T–444T,

Published on July 23, 2013