Introduction

Chemical analyses and laboratory studies of glasses from the ancient and medieval worlds have provided a great deal of useful information for archaeologists, curators, and historians,¹ but until now, few such studies have been carried out on glasses from more recent periods. One reason is that in more recent times, from about the seventeenth century onwards, sources of raw materials have become more numerous and commercial trade in these materials has grown more competitive and farther-reaching. Therefore, it has been expected that there might have been as much variation in chemical composition within the production of a given factory as there was variation among different factories. If that were true, the chances of characterizing the products of a given factory on the basis of chemical composition would be poorer than when dealing with glass manufactured in earlier times.

To some extent then, the analytical study of Amelung glass reported here is opening a new field, and as it is with most new fields of research, one must proceed with caution when interpreting the results. Because so few analyses have been made of glasses of this period, either American or European, there is as yet little information available in the way of comparative data.²

This study was initiated as a part of the excavation of the New Bremen factory site, but it quickly broadened with the development of analytical techniques which made possible the nondestructive analysis of complete glass objects. Through this development, it became possible to compare the compositions of excavated fragments with known-and suspected Amelung objects. The result was a cooperative analytical program between The Corning Museum of Glass and The Henry Francis du Pont Winterthur Museum, a program which has been most fruitful. It will become apparent to the reader that the analysis of American glass is indeed a very promising field of research, a finding which seems especially appropriate in the nation's bicentennial year.

The investigation consists of five parts, each of which was directed towards specific goals. The five parts are described briefly below, and results from the two major parts are presented in the sections which follow. Several important aspects of this research are still in progress, and it can be expected that our knowledge of Amelung's glass and its overall place in the history of glass will be expanded even further in the future.

1) Chemical Analyses of Amelung Objects

Because of the great historical importance of Amelung objects, they can only be analyzed by completely nondestructive, non-sampling methods, and the only method currently available for analyzing glass within this restriction is the Energy-Dispersive X-ray Fluorescence Analyzer. During the past three years, a large group of Amelung glasses and related glasses has been analyzed by Mr. Victor F. Hanson and his colleagues on this instrument in the Analytical Laboratory of the Winterthur Museum. This marks the first time that such an extensive study of museum glass objects has been conducted. To date, more than a hundred objects have been analyzed.

The objective of this part of the study was to determine If any specific chemical compositions or peculiarities could be associated with the products of Amelung's factories. Therefore, the most important part of the study was the analysis of pieces with unquestioned attributions. An ancillary objective was the strengthening or weakening of attributions for objects whose origins are less certain. A critical point in the logic behind this study was to determine not only whether the known Amelung glasses resemble one another, but also whether they differ as a group from glasses from other factories.

2) Chemical Analyses of Fragments Excavated at the Amelung Factory Site

During the course of digging at the site of the New Bremen factory, thousands of fragments of glass were uncovered. From among these, thirty pieces were selected for chemical analysis. Some of the pieces came from recognizable glass forms, while others were pieces of cullet or waste glass. Because the pieces are fragments, it was permissible to sacrifice samples large enough for full quantitative chemical analysis. These analyses were carried out as part of the scientific research program of The Corning Museum of Glass. The objective of this part of the study was to characterize the chemical compositions of the glass made at the factory and to use the findings to learn as much as possible about how the factory operated.

3) Flouresence of Amelung Glasses

When examined under ultraviolet light, most glasses will fluoresce, that is, they emit visible light. The color of the light and its intensity vary depending primarily upon the chemical composition of the glasses examined. Thus, fluorescence is often a useful clue to chemical composition and, more importantly, is useful for identifying glasses which might have been made in the same factory.

Most of the Amelung objects mentioned throughout Mr. Lanman's and Ms. Palmer's contribution and all of the fragments used for chemical analysis have been examined under both long-wave and short-wave ultraviolet lamps. Several different types of fluorescence were observed, which serve as an additional basis for classifying the objects. These observations were of a subjective and qualitative nature only, because it was experimentally impossible to determine actual fluorescence spectra on large pieces of glass. For this reason comparisons between groups of objects were especially important. The human eye is a remarkably sensitive instrument for detecting small differences when comparing colors, but it is relatively poor at remembering colors.

Excavated fragments representative of the various types of fluorescence found among the glasses have been analyzed chemically so as to determine what compositional features are associated with fluorescence colors observed in the objects. This research is still in progress, and the results will be reported in the future.

4) Other Studies

Viscosity-temperature curves were estimated for Amelung glasses. These curves were extrapolated from data obtained from simple pull tests carried out on small fragments of the glasses. The resulting curves allow one to estimate the minimum temperatures required for forming various types of objects and thus tell something about the furnace capabilities. A gradient-furnace experiment was carried out on a sample of mudstone removed from the furnace. This also yielded some information on the temperatures achieved in the furnaces. Several experiments have been planned for various other materials excavated at the site, including crucible fragments, bricks, and possible raw materials. However, these experiments have not yet been completed.

5) Activation Analyses of the Glasses

Very small samples of pontil glass have been removed from several Amelung objects. Although these samples were too small to be analyzed by the x-ray fluorescence method or wet-chemical methods, they were large enough for analysis by activation methods. The samples were removed only from those objects where a convenient protrusion of pontil glass was found. In no case were samples removed from objects where there was any danger of breakage. The sacrifice of a tiny piece of glass from a pontil mark is justified on the basis that the information to be gained by its chemical analysis is potentially more valuable than a few milligrams of glass itself. Samples have been submitted to Dr. Philip Lafleur at the National Bureau of Standards for analysis by his group.

The question arises whether pontil glass is necessarily the same as glass from which the body of an object itself was made. In most cases, they obviously should be the same, but one cannot be certain that this is always true. The ultraviolet fluorescence colors of the pontil glass samples were compared to the bodies of the parent glasses, and in all but one instance the fluorescence was identical to that of the parent glass. It is, therefore, reasonable to assume that for the glasses analyzed here (other than the one with the non-matching fluorescence), the pontil samples do truly represent the compositions of the objects. These analyses have not yet been completed.

Chemical Analyses of Glasses Attributed to John Frederick Amelung

V.F. Hanson

In 1969 the Henry Francis DuPont Winterthur Museum established a research laboratory to develop methods for characterizing the objects in its collection on the basis of chemical compositions. The analytical programs at the laboratory center around a new instrument, the Energy-Dispersive X-ray Fluorescence Analyzer.³ The most remarkable feature of this instrument—a unique feature for the chemical analysis of certain types of materials—is that it employs a method which is completely nondestructive and non-sampling. The instrument enables the analyst to identify all the chemical elements above element 19 (potassium) in the periodic table without removing a sample or defacing an object in any way. The X-ray Analyzer has been used for numerous analyses of metallic and ceramic objects and plays a key role in the glass-analysis program at Winterthur.

In 1973 the Winterthur Museum and The Corning Museum of Glass undertook a cooperative venture in which the X-ray Analyzer was used to analyze Amelung glasses in connection with Corning's bicentennial exhibition and the publication of the archaeological excavations of the Amelung site. Since the beginning of this project, just over a hundred objects have been analyzed. The key pieces in the study are 10 signed or documented Amelung objects and about 15 others with strong Amelung attributions. The balance consists of some objects with looser Amelung associations along with an assortment of glasses from other American factories or of European origin.

The study has so far included about 80 glasses from the Winterthur and Corning collections and more than 20 glasses lent for chemical analysis by other institutions and private collectors. (A list of the objects analyzed appears below.) After the analyses were completed, the objects were returned to the shelves of the owners perfectly intact and in exactly the same condition as when they had left.

A detailed description of the analytical procedures and a complete compilation of the data will appear elsewhere in the near future, because the full exposition of these subjects on a rigorous scientific basis goes beyond the scope of this publication. The objective here is to outline a few analytical aspects of the work and to summarize some of the more important findings.

The Energy-Dispersive X-ray Fluorescence Analyzer brings together radioactive x-ray-emitting isotopes, solid state x-ray detectors, minicomputers, and compact data storage devices. In simplest terms, the object to be analyzed is positioned above a radioactive source. X-rays from the source excite atoms within the material being analyzed, causing them to emit a pattern of secondary x-rays. The pattern emitted is uniquely characteristic of the types and numbers of atoms within the material. The secondary x-rays are collected and counted and ultimately lead to the identification of the chemical elements within the object and to the measurement of their concentrations. About twenty minutes is required to irradiate the sample and obtain a teletype readout of weight percentages of thirty elements.

In order to adapt this method to the analysis of glasses, a procedure was worked out which makes use of separate sequential irradiations by three radioactive isotopes, Fe⁵⁵, Cd¹⁰⁹, and Am²⁴¹. This procedure provides the potential for quantitative determinations of 25 chemical elements in the glasses analyzed. The method makes use of a series of several reference glasses of known chemical compositions, some of which were especially prepared so as to simulate the compositions anticipated for the Amelung glasses.⁴

One limitation of the method in its present stage of development at Winterthur is that the analyses do not provide information on low-atomic-number elements, and it is known that such data are useful for classifying glasses. This was recognized at the outset. But in view of the fact that the X-ray Analyzer is the only method of analysis which is truly non-sampling and nondestructive for glasses, it is also therefore the only method by which the Amelung objects may now be analyzed. Instrumental improvements planned for the future should allow for analyses of some of the light elements as well. It should also be noted that this method, unlike wave-length dispersive x-ray fluorescence, does not pose any hazard of discoloration of the glass.

Several special problems had to be solved first involving the geometry of the objects; that is, the effects of thickness of the glass, distance from the source, and shape of the object. These factors were all taken into account and compensated for before the objects were analyzed.

It often happens that problems of handling and interpreting large bodies of analytical data can be as difficult and time-consuming as the processes which go into acquiring the data. That would certainly have been true in this instance were it not for the computerized techniques of data handling built into the instrument. Even so, this does not entirely relieve the investigators of the responsibility for deciding what parts of the data are most significant for the ultimate interpretation relative to the objects themselves.

In this study it was decided that the analyses of the 10 signed or documented Amelung pieces was the place to start the interpretation. The first important observation is that all are potash-lime-silica glasses, and although several contain traces of lead, none are high-lead compositions. Beyond that it appears that these glasses have in common the compositional features listed below. It was also discovered that several additional glasses having strong Amelung attributions, and a few of less certain attribution, share these same compositional features, which are:

1. The mean potassium:calcium ratio is 16:10.
2. The average ratios Rb:Sr:Y:Zr are 7:10:2:6.
3. The anitomy content is 0.1-0.2 percent.
4. The barium content is 0.02-0.08 percent.

A most important feature of Amelung glasses is the presence of significant amounts of antimony. Based on present studies, it is evident that objects having compositional features matching those of Amelung glass, except for antimony, were made in another factory.

The analytical results for the ten signed or documented objects are shown in Table 1 (above). The chemical similarities and variability can be judged from the table. For comparison, analyses of seven Lauenstein objects are presented in Table 2 (above). There are differences in the concentration of two major ingredients, potassium oxide and calcium oxide, and in six trace elements: titanium, manganese, rubidium, lead, antimony, and barium.⁵ The ratios of rubidium, strontium, yttrium, and zirconium are distinctly different from those of Amelung glass. There is, however, overlapping among some of these elements. In Table 3 (below) are given analyses of some fragments excavated at the New Bremen factory. These were reported from among a total of nineteen fragments analyzed, because they were also analyzed by The Corning Museum.

The data in Table 1 also contain an intriguing hint of possible systematic differences between the first five objects dated up through 1789 and those dating from 1791 and 1792. The distinction shows up especially in the lead contents. The difference could mark a distinction between the glasses made before the fire of 1790 and those made afterward. About half of the signed objects contain no titanium, whereas half have titanium in concentrations from 0.003 to 0.03 percent. Titanium occurs widely in the earth's crust as ilmenite, the black magnetic particles frequently found in sand. It appears that either the source of sand was changed during the fabrication of this group of signed pieces or that a method was found to remove the ilmenite from the sand. The iron in ilmenite produces an objectionable greenish color in the glass which was often offset by adding manganese. While the arsenic contents vary considerably and arsenic does not yet appear to qualify as a characterizing feature, it may prove significant in future studies.⁶

After the chemical features of the documented Amelung objects had been established, the analyses of the remaining glasses were considered. These were classified solely on the basis of chemical similarities, grouping together those which have the most chemical features in common. Three other major chemical groups were defined, and the large majority of all the glasses analyzed (79) fit into the four categories. The categories are built up around 13 chemical elements. The first group, the A group, contains 38 objects, all but five of which have some Amelung association. The group also includes a patty of glass made by melting together crushed cullet from the excavation. Three objects attributed to Stiegel and two with European attributions also fell in this group. Further studies of the data will be required to explain the chemical differences among the groups in terms of glass technology, but at least part of the variation is related to the color chemistry.

Beyond the four major groups, eleven other chemical types appeared, but for the moment some of these are only "splinter groups," and none contains more than three glasses. This is significant and encouraging, however, in that these groups imply a diversity of compositions among glasses of this period. It is expected, therefore, that as this project continues, compositional categories matching particular factories, regions, or periods will eventually emerge.

Acknowledgements The success of this project was due in large part to the cooperation of those institutions and individuals who lent objects for chemical analysis. The author thanks them and also Arlene Palmer and Dwight P. Lanmon who delivered the objects to the Winterthur Laboratory for analysis. Special thanks are expressed to the author's associates at the Winterthur Museum, especially those listed below: Karen Anderson, Janice Carlson, Charles Hummel, Justine M. Mataleno, Karen M. Papouchado, George J. Reilly, and Helen F. Szczecinski. Special thanks are also expressed to P.H. Gaither of Winterhtur's Scientific Advisory Committee for his valuable contribution to the computer programming used in this study.

Chemical Analyses and Other Laboratory Studies of Glass Fragments Excavated at the Site of the New Bremen Glassmanufactory

R.H. Brill

The object of this study was to determine what chemical compositions or peculiarities might characterize the glass excavated at the site of John Frederick Amelung's New Bremen Glassmanufactory.⁷ From among the thousands of fragments excavated, thirty were selected for chemical analysis. The sampling was not intended to be representative, in a proportional sense, of the entire body of glass uncovered. Instead, there was a strong bias in the sampling towards colorless glasses, which evidently were used for fine quality production. About half of the specimens analyzed (fourteen fragments) are of this colorless glass. These were chosen in the hope of establishing compositional connections with museum objects attributed to Amelung. The remainder consists of a small assortment of colored glasses (purples and blues) and several specimens thought to represent production of more ordinary wares or window glass. A list of sample descriptions is appended at the end of this text.

The sampling contains specimens recognizable as vessel fragments, some cullet, and waste glass, such as drippings and knock-offs. One can rarely be absolutely certain that particular glass vessel fragments or even pieces of cullet must necessarily have been made at a factory site because they were excavated there. Some scrap glass and cullet could always have been brought into a factory. However, we feel quite confident that the colorless specimens and cullet analyzed here really were made at this factory. Moreover, the indications are that the specimens of more ordinary glasses also were, because, as will be demonstrated below, their compositions closely match those of knock-offs and drippings. Only two specimens are really suspect of having outside origins. The dates of the specimens analyzed are assumed to fall somewhere within the factory's operating period, that is, 1785 through 1795, although they are associated with the building which is believed to have burned in 1790 and not been rebuilt afterwards.

Quantitative analyses by atomic absorption were carried out for the following elements, expressed throughout as weight percentages of the oxides:⁸ potassium (K₂O), calcium (CaO), sodium (Na₂O), magnesium (MgO), aluminum (Al₂O₃), iron (Fe₂O₃), and manganese (MnO). Phosphorus (P₂O₅) was determined colorimetrically. All of the other elements at minor and trace levels were analyzed by semi-quantitative emission spectroscopy. Duplicate determinations starting with new samples were run for any questionable values. Silica (SiO₂) was estimated by difference from 100 percent.

Several special reference glasses were prepared for these analyses and for the object analyses carried out at Winterthur. These included "synthetic Amelung glasses" which contained major, minor, and trace elements at levels corresponding to those obtained in preliminary studies. Analyses of the synthetic Amelung glasses yielded excellent agreement with the theoretical compositions calculated from the known batch compositions.⁹ Therefore, the analytical procedures used are closely calibrated for the analyses of both the Amelung fragments and the objects.¹⁰

The data are summarized in Table 4 and compiled in Tables 5–8. The specimens are grouped according to two main categories which emerged from the results and are arranged within those categories according to color.

The first important observation made was that all but one of the specimens are potash:lime:silica glasses (K₂O:CaO:SiO₂). Whereas lead (PbO) was frequently found at trace or low minor levels, a fact of some significance, it was a major component in only one specimen, no. 4283. This fragment, a stem base with faceted cutting, is a true lead glass, containing 34.5 percent PbO. The question arises as to whether it was actually manufactured at the factory or if it is an intrusion or glass brought in for cullet. It appears that the majority of glass made at this factory, if not all of it, was of non-lead formulations.

The second important observation is that although all the glasses are potash:lime formulations (excluding the one lead glass), two distinctly different categories emerge from the data. One is a low-lime formulation, and the other a high-lime formulation. The low-lime group has an average CaO content of 9.05 percent. It contains eighteen glasses ranging from 7.08 to 10.8 percent CaO. The high-lime group has an average CaO content of 19.6 percent. It contains five glasses ranging from 18.6 to 20.6 percent CaO. The separation between the groups is well-defined, and there is no overlapping.

The data for the other chemical elements reinforce the separation. The same specimens which cluster together in the low-lime group also cluster together in their percentages of other chemical elements. Similarly, the specimens in the high-lime group remain clustered in the other chemical elements. In all, ten different chemical elements serve to discriminate between the two chemical types. The discriminating oxides are: CaO, K₂O, Na₂O, MgO, Al₂O₃, Fe₂O₃, SrO, BaO, PbO, and Sb₂O₅. There are two partially-discriminating elements, P₂O₅, and TiO₂, while MnO and B₂O₃, are non-discriminating. Repeat analyses will be required to establish whether or not Li₂O and Rb₂O are discriminating.

The sharp separation into two groups proves to have a rational basis in that all the glasses in the low-lime group are either colorless, purple, or blue. Clearly, this composition reflects the basic formulation used for manufacturing the "fine glass" production of the factory. All of the specimens in the high-lime group are green, aqua, or amber—the colors characteristic of more ordinary production. The chemical groups are tight and the separation is complete, there being no crossovers between the two groups.

There are six glasses which have been designated intermediate or uncertain as shown in Table 8. There are various ways of interpreting their analyses. Two of them (nos. 346 and 573) could possibly be glasses made by melting together a mixture of glasses of the low-lime and high-lime groups. The others could be badly off-composition examples of the low-lime and high-lime groups. Alternatively, some of these six might be intrusions on the site—especially the bottle fragments—or could be scrap glass brought in for remelting as batch cullet. Among these, only no. 350, an irregularly-shaped piece of waste glass, was certainly melted at the site. Its unusual color, probably the result of devitrification, is consistent with its identification as an off-composition waste glass. The others appear to be fragments of vessel or window glass. It is probably wiser not to press too far in interpreting these analyses now, but to await the results of follow-up analyses of additional specimens.

It is noteworthy that the purple and blue glasses fall in the same category as the colorless glasses. The vessels made from the colored glasses must have been regarded as being of the same level of quality as the colorless vessels, because for practical purposes, as far as the colors would have been concerned, the addition of the same colorants to the high-lime formulation would have yielded an identical appearance to the eye.

The color chemistry of the purple and blue glasses is familiar. The three blue glasses are colored by cobalt oxide (CoO). The two pale blue glasses (nos. 363 and 1821) have about 0.03 percent CoO and no. 349, a much darker blue color, contains about 0.4 percent CoO. Certain other trace elements seem to be higher in the blue glasses than in their colorless companions within the low-lime group. These elements could have been introduced, probably unintentionally, with the ingredient which contained the cobalt. They are bismuth (Bi₂O₅), copper (CuO), lead (PbO), probably arsenic (As₂O₅), probably iron (Fe₂O₃), and possibly nickel (NiO). On several other occasions arsenic has been found to accompany cobalt in early glasses.¹¹ This is because arsenic often occurs in association with cobalt in natural minerals, such as cobaltite, CoAsS. The blue glasses made at the Amelung factory were probably colored by the addition of cobalt imported in the form of zaffre or smalt.

The purple glasses (nos. 347, 348, 362, and 568) are all colored, as is to be expected, with manganese (MnO) at a level of approximately 2–3 percent. The ingredient used to introduce the manganese also brought in barium (BaO), lead (PbO), nickel (NiO), copper (CuO), vanadium (V₂O₅), some aluminum (Al₂O₂), and possibly bismuth (Bi₂O₃), zinc (ZnO), and tin (SnO₂).

The presence of manganese in all of the fragments from the site, the low-lime, the high-lime, and even the uncertain group, ties the glasses together somewhat. At the concentrations found (a mean value of 0.40 percent), it is reasonable to assume that the manganese was an intentional additive meant to act as a decolorizer. Its function was to offset the greenish color introduced by iron impurities. Because manganese was so commonly used for this purpose, it is not likely to prove useful for distinguishing Amelung glass from other glasses of the period.

A rather vexing question grows out of the fact that some of the discriminating elements in this study occur at levels which are difficult to interpret. The same is true of some of the "Amelung features” in Mr. Hanson's object analyses. For example, lead concentrations in the range of a few tenths of a percent do not seem sufficient to confer any advantageous properties to the glasses. Similarly, the level of the antimony in some of the glasses is somewhat low for performing its usual functions as a decolorizer or as a fining agent. Because the glasses are believed to contain manganese as a decolorizer, presumably the antimony was intended as a fining agent, that is, an ingredient used to remove "seed" or small bubbles from glass. Antimony is probably not very effective as a fining agent at concentrations less than about 0.05 percent. In this connection, the Amelung glasses do not seem to be particularly well-fined, even by eighteenth-century standards. The presence of arsenic (As₂O₅) may also be explained by its use as a fining agent, particularly in the colorless glasses, where it would not have been associated with any colorant.

The trace levels of the copper and zinc and the occasional traces of vanadium might be explained in either of two ways. All three could be associated with the manganese, or the vanadium might have come in through the corrosion of a crucible, since it is a common impurity in clays.¹² On the other hand, the plant ashes used for preparing the potash ingredient must have been purified by leaching, and the alkali could have picked up copper and zinc from brass pots or other utensils used in that process. If the alkali had been prepared in lead-sheathed vats, that might also explain the presence of lead in the low-lime glasses. The alkali would have had to contain about one percent of lead to account for the 0.23 percent average PbO content of the glasses. It is possible that some lead was introduced accidentally through the use of high-lead cullet. (Recall, for example, the piece of high lead glass, no. 4283, which was analyzed.) One would expect, however, that if scrap cullet was the source of the lead, that the lead contents would be more variable.

In any event, the most important point is that the lead and antimony are present, regardless of how they got there, and may someday serve to differentiate the glass made at this factory from glasses made elsewhere. Whether or not that hope materializes depends upon the outcome of analyses of glasses from other sources.

There are two bodies of analytical data for American glasses at The Corning Museum, excluding miscellaneous analyses of single objects. One relates mainly to pressed glasses, but these objects are mostly lead glasses and differ too much in date for direct comparison with the Amelung fragments. (It might be noted in passing, however, that even those which are not lead glasses are quite different from the Amelung compositions.) Of more immediate interest here is a group of twelve glasses from the Wistar factory. Among this suite of samples, containing bottle fragments, cullet, and trailings, there are two somewhat different compositions. The Wistar glasses can be distinguished from both Amelung compositions (the low-lime and high-lime groups) by four elements: calcium, sodium, lead, and antimony. There are other elements, too, which may be discriminating.

The Wistar glasses contain lead, but the concentration is only of the order of 0.00X percent, markedly lower than in the fine Amelung glasses. Antimony was not detected. Actually, the Wistar analyses of bottle fragments are quite close to the composition of the stray sample no. 346 from the Amelung site. Although the agreement is not perfect, the composition of that vessel fragment seems as close to the Wistar glass as it does to its Amelung companions, lending support to the supposition that this piece is an intrusion.

The explanation of why there were two formulations in use at the factory can be seen from either of two viewpoints, both involving aesthetic, technological and economic factors. Seen from the viewpoint of what is known about glass compositions today, the low-lime compositions would be taken as the norm, and the high-lime compositions seen as unusual. The high-lime content made the glass appreciably harder, so that it would have stood up better under heavy usage and might have been more resistant to breakage, both of which are clearly advantageous properties for utilitarian wares. This was confirmed by the results of simple scratch tests on some of the analyzed glasses. These tests showed that the five glasses with the high-lime composition have a hardness somewhere between 6.5 and 7 on the Mohs hardness scale, while the glasses in the low-lime group have a hardness of about 5.5 to 6.¹³ A further advantage to the glassmaker was an economic one, for the high-lime composition would have resulted from the use of a low-grade, impure potash as alkali. More expensive, purified potash would have been needed for the colorless and softer low-lime composition of the fine glass.

Seen from a contemporaneous viewpoint, however, the situation takes a different twist. The period of this factory's operation began just at the time that Lavoisier's revolutionary discoveries which set the foundation for our modern concepts of chemistry were gaining acceptance. In fact, in the very month when Amelung was advertising his glass in the Maryland Journal and Baltimore Advertiser (February 11, 1785), Lavoisier was demonstrating crucial experiments of The Chemical Revolution before a group of eminent scientists assembled in Paris for the occasion.

It is clear that the glassmakers of the time would not have had the same understanding as we do of the compositions of their glasses. Instead, they would have taken a purely empirical approach, thinking in terms of the properties of the glasses which would result from their choices of starting materials. What we call the high-lime group of glasses were really the norm of the day, and the low-lime glasses, which required the expensive, purified materials, were the specialty glasses. But the formulation which led to the colorless glass had another equally important technological advantage. The fact that the glass was appreciably softer meant that it also could be cut and engraved more easily not as easily as high-lead glasses, but certainly it was an improvement over the hard high-lime glasses. This practice should not be thought of as being an innovation by Amelung or as being unique to the Amelung factory. It had probably prevailed for some two centuries previously in the glasshouses of Europe and England.

Some approximate calculations have been made starting with the composite compositions given in Table 9 for the Amelung glasses of the low-lime and high-lime groups. By making assumptions as to probable compositions of probable raw materials, it was estimated that the basic recipes used for preparing the batches could have been as follows:

	Best quality colorless glass for fine wares
Sand	100 lb.
Pearl ash	40 lb.
Lime	20 lb.
Decolorizing cullet	1 lb.
Fining agent (?)	2 oz.
	Green glass for common wares and window glass
Sand	100 lb.
Low-grade Potash	40 lb.
Lime	20 lb.
Decolorizing cullet	1 lb.

The significance of these estimates is that both grades of glass could have been made by following the same whole-number weight-proportion recipes with different grades of raw materials. In each case, some quantity of crushed cullet would also have been added to facilitate melting. The numbers calculated are admittedly conjectural and are presented here mainly as an illustration of the way in which archaeological chemists like to explore their data.

Although the physical-properties studies have not been completed, there are some preliminary findings which aid in visualizing the working qualities of the two types of Amelung glass. Viscosity determinations have shown that both the low-lime and the high-lime glasses have similar, and very steep, viscosity temperature curves.¹⁴ In order to have gathered either glass it would have had to have been brought to a temperature of 1150-1200°C (log viscosity ~3). The glasses would have gathered like a modern soda-lime glass, but would have begun to set up very quickly as they cooled, giving the glassblower a relatively short working time in which to finish his blowing, shearing, and other shaping operations before having to reheat the glass. Present-day glassblowers would not like working with these glasses at all. The coefficients of expansion are not very different from modern soda-lime glasses, with the ordinary Amelung formulation being somewhat better on this count than the formulation used for the fine wares. Some of the estimated properties, based upon averages of four glasses of each type, are:

	Softening Point	Annealing Point	Strain Point	Coefficient of Expansion (x107/0C)
High-lime glasses	800°C	610°	565°	91
Low-lime glasses	820°	640°	600°	77

Eight of the excavated fragments reported here were also included among the nineteen fragments analyzed by Dr. Hanson.¹⁵ All eight of these glasses are of the low-lime type. They are nos. 337, 339, 349, 562, 563, 565, 568, and 572. In all, twenty-one elements were sought in common by the two laboratories. These include two major components (calcium and potassium) and one minor component (manganese). The remainder, even if they are intentional additives such as antimony and cobalt, are at trace levels.

A comparison of the analyzed values for K₂O and CaO shows that the agreement on four glasses (nos. 562, 563, 568, and 572) is satisfactory, but that the x-ray fluorescence data for the other four glasses (nos. 337, 339, 349, and 565) are systematically lower than the atomic absorption-analyses.¹⁶ The x-ray values are consistently about 60 percent of the atomic absorption values. For the most part, the agreement on the trace elements is acceptable. In other instances, however, the agreement is poor, even taking into account that the two methods are only semi-quantitative at trace levels. Thus, the agreement between the two sets of data is not as good as one would like to have, but that can be improved in the future, and the two procedures should be brought into better cross-calibration as further comparative studies are made. Nonetheless, the analyses are generally in agreement in a qualitative sense, and each is self-consistent. For example, the agreement on lead, one of the important elements, is good quantitatively, and that on antimony is good qualitatively. This matter has an important bearing on any attempts to compare the atomic-absorption/emission spectroscopic data for the excavated fragments (which have the advantage of containing information on the light elements) with the x-ray fluorescence data on the Amelung objects. With these reservations in mind, such a comparison has been attempted.

Table 9 contains data which allow a comparison between the documented Amelung objects analyzed by x-ray fluorescence at the Winterthur Museum and the excavated fragments analyzed by The Corning Museum using a combination of atomic absorption and emission spectroscopy. The data are composite compositions consisting of mean values for each element. The means were computed from ten objects in the case of the documented Amelung pieces,¹⁷ and nineteen glasses in the case of the excavated fragments.

Adjustments were made in the fragment analyses to offset the effects of colorants or colorant-associated elements. Thus, the glasses of these groups are comparable on the basis of colorless glasses containing impurity trace elements. The lead and antimony values have been split so as to reflect the fact that both bodies of data contained these elements at two different levels. To have reported a single mean value would have been misleading.

Of the twelve elements reported for x-ray fluorescence, ten agree either well or acceptably with the compositions of the low-lime fragments. Only titanium and arsenic do not agree. The author is inclined to see the match as being quite persuasive towards the inference that the documented Amelung objects were made in the same factory which yielded the fragments—or at least in a closely-related factory. But the evidence is not conclusive. The discrepancies in the arsenic and titanium analyses might be attributable to experimental errors, or in the case of the arsenic, an intentional additive, it may have been used for only certain short periods.

The close match between the split means of the lead in the objects with those for the fragments is intriguing, recalling that the split may separate the objects made before 1790 from those made afterwards.¹⁸ If it had turned out that the New Bremen factory fragments were all of the higher-lead content (the "early level"), then a straightforward interpretation would have been possible. It could have been concluded that the higher-lead glasses were made there, that the factory did not reopen after the 1790 fire, and that the lower-lead glasses came from another (later) Amelung factory. But that possibility is not borne out by the evidence, because both levels of lead are found among the New Bremen fragments. This problem becomes quite complicated if one attempts to sort out all the possibilities, but the author has come to a tentative conclusion that two of the hypotheses involved are mutually exclusive (unless one assumes that the change in composition occurred at about the time of the fire by coincidence). These hypotheses are (1) that the split in lead separates the pre-1790 glasses from the post-1790 glasses and (2) that the excavated factory did not reopen after the 1790 fire. However, the data are really too few at present to accept either one of these important hypotheses at the expense of rejecting the other. Therefore, the question should be held in reserve.

A minor flaw in the logic of comparison lies in the fact that the x-ray fluorescence data are lacking in light-element information. On the basis of the restricted number of elements included in Table 9, one might well find compositions matching the fragments—marginally so, at least—among the objects known not to be Amelung glass. Without detracting from the prowess of the x-ray fluorescence method in any way, it is now recognized that data on such elements as sodium, magnesium, aluminum, phosphorus, and lithium are very useful for characterizing glasses. The situation is similar to that arising in the study of compositions of medieval stained glasses, where major- and minor-element contents are more useful than trace elements in characterizing glasses from different factories or periods. In addition, neither method of analysis included determinations of sulfate or chloride, each of which is believed to be helpful for characterizing glasses in the potash:lime:silica system.

Perhaps the best way to describe the author's present feelings is to say that we now look forward to obtaining activation analyses of the pontil glass samples which may confirm the inference that the documented objects came from the excavated factory. In the meantime we shall continue the analyses of excavated materials (including additional glass specimens) and complete our research on the physical properties and fluorescence behavior of Amelung glasses. Having gained encouragement from the results of this study, we shall also pursue more vigorously the analysis and laboratory examination of other American glasses.

Acknowledgements
The author of this section gratefully acknowledges the assistance and encouragement of the following persons in various stages of this work: Ivor Noël Hume, Kenneth M. Wilson, Paul N. Perrot, John F. Wosinski, Robert H. Bell, Dwight P. Lanmon, Victor F. Hanson, Charles F. Hummel, and George J. Reilly. The Mss. Judy Seal and Linda Randall assisted in the handling of the data.

Sample Descriptions

The primary entries are Analytical Sample Numbers of the Scientific Research Department of The Corning Museum of Glass. "AS" numbers refer to excavation field numbers, and Figure references are those in Mr. Noël Hume's section of the Journal.

The term colorless refers here to glasses which appear to have been decolorized. In some cases the colors vary somewhat from being "water-white" and show a smoky or faint purplish tinge. In the author's judgment, however, they were intended to be colorless. The purple color is synonymous with amethyst.

---

Robert H. Brill and Victor F. Hanson
This article was published in the Journal of Glass Studies, Vol. 18 (1976), 216–237.

---

1. For general information on the scientific investigation of early glasses see: R.H. Brill, "The Scientific Investigation of Ancient Glass," Proceedings of the VIIIth International Congress on Glass, London, Sheffield, England: The Society of Glass Technology 1968, pp. 47–68.

2. A few analyses are scattered among the literature, and at The Corning Museum of Glass there are as yet unpublished analyses of some American and European glasses of the eighteenth and nineteenth centuries. Some analyses of surface finds at the New Bremen site were published earlier. See: Donald B. Hubbard, Lillie B. Jenkins, and Elizabeth M. Krumrine, "Amelung Glasses Compared with Some Modern Commercial Glasses," The Scientific Monthly, LXXV, No. 6, Dec. 1952, pp. 327–338.

3. Victor F. Hanson, "Quantitative Elemental Analysis of Art Objects by Energy-Dispersive X-Ray Fluorescence Spectrometry," Applied Spectroscopy, 27, No. 5, 1973, pp. 309–334.

4. The following lists the reference glasses. All were prepared by Dr. A.A. Erickson of Corning Glass Works and his co-workers.
a) Nos. 611, 612, 614, and 616 distributed by the National Bureau of Standards.
b) Reference glasses A, B, and D of a series prepared for The Corning Museum of Glass for inter-laboratory comparison analyses.
c) Three glasses prepared for The Corning Museum of Glass, which contain only the major ingredients in the reference glasses A, B, and D and not the trace elements.
d) Glasses SPE, TVX, and TVY prepared for The Corning Museum of Glass simulating the compositions anticipated for the Amelung glasses.

5. In keeping with glassmaking convention, all compositions were recorded in weight percentages of the oxides. For readers unfamiliar with chemical symbols, the formulas below represent the oxides of the corresponding chemical elements.

K2O	potassium
CaO	calcium
TiO2	titanium
MnO	manganese
Fe2O3	Iron
Rb2O	rubidium
SrO	strontium
Y2O3	yttrium
ZrO2	Zirconium
PbO	lead
As2O5	arsenic
Sb2O5	antimony
BaO	barium

 

6. Some remarks on the functions of the various ingredients and their levels of concentration appear in Section III. Also included there are comparisons between the analyses of the documented objects and the excavated fragments.

7. The factory will be referred to here simply as "the Amelung factory" without intending to imply either that it was the only "Amelung factory" or that it was always under the control of John Frederick Amelung.

8. For readers unfamiliar with chemical symbols, the formulas below represent the oxides of the corresponding chemical elements.

SiO2	silicon	PbO	lead
K2O	potassium	BaO	barium
CaO	calcium	SrO	strontium
Na2O	sodium	As2O5	arsenic
MgO	magnesium	Li2O	lithium
Al2O3	aluminum	Rb2O	rubidium
Fe2O3	iron	B2O3	boron
TiO2	titanium	V2O3	vanadium
Sb2O5	antimony	NiO	nickel
MnO	manganese	ZrO2	zirconium
CuO	copper	Bi2O3	bismuth
CoO	cobalt	P2O5	phosphorus
SnO2	tin

 

9. The reference glasses used for all analyses reported here are listed below. All the glasses were prepared by Dr. August A. Erickson of Corning Glass Works and his co-workers. a) Nos. 611, 612, 614, and 616 distributed by the National Bureau of Standards. b) Reference glasses A, B, and D of a series prepared for The Corning Museum of Glass for interlaboratory comparison analyses. c) Three glasses prepared for The Corning Museum of Glass, which contain only the major ingredients in the reference glasses A, B, and D and not the trace elements. d) Glasses SPE, TVX, and TVY prepared for The Corning Museum of Glass simulating the compositions anticipated for the Amelung glasses.

10. The only discrepancy was in one glass, where the Y₂O₃ by conventional x-ray fluorescence yielded a value of 0.06 percent instead of the theoretical 0.10 percent.

11. Unpublished analyses of The Corning Museum of Glass.

12. In this connection there is an interesting quote in W. Rosenhain, Glass Manufacture, New York: D. Van Nostrand Company, 1912, p. 189."... vanadium occurs in small proportions in a number of fireclays, including some of those of the Stourbridge district, and glass melted in pots containing this element is liable to have its colour spoilt by taking up the vanadium from the clay."

13. The absolute values of the hardness values estimated may be in error, because of the nature of the tests, but the difference of approximately one unit (on the Mohs scale) between the two groups of glasses is about right.

14. The measurements reported here were made by Mr. Loren Morse and Mr. Eugene Fontana, both of Corning Glass Works. For a discussion of viscosity temperature relationships see: R.H. Brill, "A Note on the Scientist's Definition of Glass," Journal of Glass Studies, IV, 1962, pp. 127–138.

15. See Table 3.

16. This might be a result of geometry, because the samples in some cases are small and have curved or irregular surfaces.

17. Analyses of parts of some objects were omitted, because they seemed suspiciously low in the reported K₂O and CaO values, possibly because of problems of geometry.

18. See Table 1.

Published on July 24, 2013