Glass as subject of study

Printer-friendly versionSend to friend

Glass is a find category encountered very often, if not always, on Roman and Early Byzantine sites. This illustrates the widespread use of glass vessels (and to some extent window panes and bracelets) in every day life of this period. Therefore it is no surprise that in Sagalassos glass sherds also turn up in almost all deposits. Studying the archaeology of glass is difficult, however. Not only is one obliged to know the archaeologicae, such as vessel types and their distribution and function, workshops for glass production and the related structures, but also the body of archaeometrical studies is becoming so large and intriguing, that it can no longer be omitted by the archaeological discipline. The necessary exercise of composing vessel typologies of sites or regions is no longer the aim, but a means. It is a shackle in the chain out of which contemporary glass studies exist. No longer is it a discipline made up by a connoisseur-ship in the true meaning of the word. Scientific analysis of glass compositions, performed by chemists or geochemists, has forced archaeologists to upgrade themselves. Recent discoveries of production centres, application of new techniques of analysis and more intense collaboration of archaeologists and other scientists have given the discipline a much needed new stimulus. The present day scientific focus is the reconstruction of the craft as a whole, from the batch preparation (by selection of the raw materials) to the final use of the vessel.

The story of the glass of Sagalassos is typical of material found in a Roman or early Byzantine city in the East. These sites with a long and intense history of occupation provide us with a glass record that spans several centuries (even millennia). Most of the material is very fragmented due to its delicate state, the intense recycling programme and its burial history. Almost all the glass finds are originating from refuse or levelling layers and are often found in secondary position. Additionally it needs to be stressed that a lot of the rim and base fragments are so small or know such a broad use, they can be attributed to a wide range of types. As such, making a typology and quantifying a glass assemblage is a difficult and time consuming exercise. The last 3 decennia however, the scientific world has taken a keen interest in the fascinating case of Roman and early Byzantine raw glass production. Questions as to where the raw materials came from, how and where glass was made and how the geological origin of the components can reflect the actual place of production of the raw glass are ranked high on the present day research programme.

Glass is a material artificially made by the fusing of a vitrifying agent (silica), a flux (natron or plant ashes) and a stabilizing agent (calcium or manganese). It lacks the crystalline structure characterising other materials, like metals, meaning that the molecules do not have a regular arrangement. Glass is also a supercooled liquid. This means that when cooled down from temperatures above 1100°C (for a soda-silica-lime glass) it is of the outermost importance to avoid crystallisation, leading to breakage. If the melt is allowed to cool to slowly a silicate can be formed, containing crystals. This can be avoided by cooling the glass at a moderate rate in the actual furnace or in a specially made annealing kiln.

Natron glass was the predominant type of ancient glass in the Mediterranean and Europe from the middle of the first millennium BC onwards until the ninth century AD. Work by Foy et al. suggests that there are likely to have been around 10 major glass groups in the Mediterranean and Western European region between the 1st and 9th centuries A.D.

Although substantial databases of major element analyses of glass exist, meaningful groupings with respect to the geographical origin of the mineral resources have seldom been possible. For example, all Roman glass was found to be relatively homogeneous natron glass with little variation in major element composition. Though significant advances have been made, an understanding of the exploitation of raw materials, technology and trade through main and trace element analysis still remains limited. However, Rare Earth Element (REE) patterns have proved promising to distinguishing between sand raw materials, as these may be typical for the geological environment of the sand.

The constituents of late antique glass production

Sand or quartz pebbles provide silica (SiO2), the principal component for glass production (accounting for 45 to 70% of the final weight). The iron impurities in the sand are responsible for the blue greenish colour raw glass possesses. We can presume that sands were selected for the impurities which they did or did not contain and their influence on the glass melt, e.g. calcium as a stabilizing agent. It is by means of these impurities, we can try to ‘trace’ the sand back to its origin. Quartz pebbles contain much less impurities, but were only rarely used in Roman, natron based, soda-silica-lime glasses.

Strabon in his Geographica (XVI, 2:25) which was most likely finished during the reign of Tiberius, Flavius Josephus in De Bello Judaico (II, 10: 12-13) , written in between 75 and 79 A.D. and Pliny the Elder in the Naturae Historia (XXXVI, 190) (76-c.78) mention several sand beds used for glass production in the Roman world. All these authors stress the exceptional qualities of the sand found on the banks and in the estuary of the river Belus (present day Na'aman River, near Akko).
Although it is true that these are all early Roman authors, analysis done on glass of the secondary workshop (were they worked glass, hence reheated raw glass and blown it into vessels) of Jalame (in Israel) showed that this type of sand was used in the 4th to 8th century to make glass with.

Other well defined geo-chemical clusters, hence representing different kinds of glass compositions/receipts, the so-called Levantine 1 and Levantine 2 clusters, can through isotopic analysis, also be traced back to the Levantine shores.

For the first time analysed in the glass assemblages (dating to the 4th century AD) excavated in the circular harbour of Carthage, Ian Freestone described a glass consisting out of a high iron, manganese and titanium oxide content, hence called HIMT.
Combining the data obtained by isotopic and major elements compositional analysis it is now clear that HIMT, indeed was made in the eastern Mediterranean but differs significantly with the Levantine glasses. It is argued that most probably HIMT was made of sands originating from a region close to the mouth of the river Nile, possibly somewhere between Alexandria and Gaza. As a possible primary production site, the city of Ostrakine (near the Bardawil Lagoon) is put forward as large quantities of glass were attested in the industrial area. Although it was shown that HIMT glass was present, further research is needed to support this hypothesis.

Aditionally, alkalis were used to lower the melting temperature of pure quartz from 1710°C to 725°C for natron glass. Natron in its pure form, does not frequently occur naturally, one of the best known sodium salt mixtures (commonly known as natron) deposits in the world is the Wadi Natrun in Egypt. Situated 100 km North West of Cairo, this depression incorporates 8 saline lakes. Whether the natron originating from Egypt was exclusively used in late antique glass production is an issue still under debate.

Production and dispersal of glass

Strikingly, for the early Roman period, we don’t have evidence yet, which is related to the production infrastructure of raw glass. And also for the late antique period, the evidence is scarce and limited to the Levant and Egypt.

Several hypotheses can be brought forward concerning the production of raw glass. According to a first model, glass could be made locally in workshops that also shaped the glass either into complete vessels or architectural elements (window panes, tesserae, …). As a second posibility glass would be recycled over and over again. A third option is that the production of raw glass was limited to the eastern part of the empire (the so-called primary workshops) and spread from there throughout the Empire. Ian Freestone argues that the limited number of well-defined glass compositions found throughout the Roman Empire from the 4th to the 9th century AD reflects the existence of such a small number of workshops that were responsible for the production of all the glass. However, Freestone lists some pros and contras for these three models. If, like claimed in the first model, raw glass was to be produced on a pure local scale with local sands then the result would be a plethora of different chemical compositions (if the sand was not manipulated). Also the second model of constant recycling of glass is not feasible either, because this would result in big blurry and fuzzy clusters of chemical compositions. Moreover, if a consequent programme of recycling would be followed, this would also result in a new distinct composition, as shown by the early Byzantine pale green glass of Sagalassos. The third model starts from the presumption that all secondary workshops made their products with imported raw glass coming from the East. This would explain the well-defined cluster of a limited amount of basic glass compositions. This hypothesis is backed up by the discovery of primary workshops in Egypt and Israel, the analysis of glass vessels, chunks and window panes coming from different sites all over the Roman Empire and the excavations of several shipwrecks in the Mediterranean where parts of the cargo existed out of big lumps of raw glass with again the same composition. Additionally, this practice existed until a decade ago in India, where different furnaces and kilns were used for glass making and working. The existing difference in chemical compositions would then reflect the chronological situation of the workshops and the raw materials available.

Although it goes without saying that these three models may have occurred at any one time and any one place, the centralized model is becoming the traditional view on glass making and working in the late antique period (4th to 9th century AD). It is shown in the literature that supra-regional models of trade cannot be reconstructed. The boost of intrinsic wealthy regions can only be secured through linking them to larger interactive networks, like the grain provision of Rome and Constantinople or the army’s guarding the limes. In this respect, the monopoly of Levantine glass in the late Roman period cannot be looked at as an isolated fact, but needs to be set in the context of overall regional economic growth, as illustrated by the large amount of locally made amphorae excavated all over the Empire, which reflects the impact of Palestinian wine trade.

The Sagalassos glass assemblage

It was shown that the sand sources used for the raw glass used in Sagalassos, originated either from Egypt, the Belus estuary, the Syro-Palestinian coast or more northern Levantine source(s). The latter seems to have known a restricted use in the early Imperial period, whereas the Egyptian source was used for the production of HIMT glass from the 4th century onwards. In the primary production centres these sands were fused with a flux, being natron or certain plant ashes for the middle Byzantine period. Natron for years was treated in a rather axiomatic way. In future research, questions as to where it was extracted, whether Egypt (the salt lakes in the Delta) was the only supplier and how its trade and market in general were organised, should be addressed. Future isotopic research hopefully will enlighten us further on the origin of the raw materials used in plant ash glasses and shed some light on the question as to where this raw glass was produced.
In future research the possibilities for the underlying causes for the occurrence of certain glass compositions should be explored. Is the changing occurrence merely an illustration of the different (clusters of) workshops that flood the market at a certain moment with a specific kind of raw glass, reflecting the geological situation of the region they are situated? And if there is a difference between Roman and early Byzantine raw (and vessel) glass trade, then what brought on this changing pattern? Are the reasons of geo-political nature (does not seem likely since all the glass compositions are situated in the Levant)? And why does HIMT suddenly appear on the international glass market in the 4th century AD, while Egypt was producing its own raw glass already for centuries (even millennia)?

Only few examples of these raw glass production installations have been excavated so far. And since we only can trace the glass clusters up to the level of their geological origin (and it is thus not a one to one relation with the primary production centres), it is difficult to reconstruct the trade patterns of this raw glass. Most likely Sagalassos only produced its own vessels from the early Byzantine period (mid 5th century AD) onwards when chunks or raw glass coming from the Levant (Levantine 1 group) or Egypt (the HIMT glass) were imported. These were then locally reheated and blown into shape. Additionally it was also proven that Levantine 1 and HIMT glass were locally remelted into a pale green glass. Making use of local pyrolysite, glass was decolourised on the spot, most likely in the Sagalassos glass workshop. Where this workshop was situated within the urban fabric and how the artisans interacted with their costumers and society in general is a question that could not be resolved. Neither was there an answer found on the question where the early to late Imperial glass vessels were produced. Are they indeed the products of glass workshops on the Levantine coast, as was claimed in the traditional literature, or were they made in centra in Asia Minor that used imported raw glass coming from the northern Levant and other places in Syro-Palestine? Also the production of window panes remains largely unknown to us. It was claimed that it does not seem likely that this type of glass object was produced on the site itself. This interpretation needs to be treated with caution since no evidence for window pane producing installations in the Roman or early Byzantine world are discovered so far. It can be suggested however, that the crown glasses and discoïdal panes were made in the local glass workshop, as they required no techniques other then the ones used in the blowing process of vessels.

To conclude we would like to emphasise the fact that we have tried to find a place for the glass of Sagalassos in the present day knowledge on Roman and early Byzantine glass. It was shown that even a town, high up in the mountains of Anatolia, had already from the early Imperial (possibly even earlier) period, access to a network of trade in glass vessels and later in raw glass. It can only be hoped that future archaeological research will focus more on glass studies, that large datasets become quantified and analysed, that more primary and secondary workshops (preferably of early to late Imperial date) get excavated and most of all that we keep this spirit of scholarly enthusiasm in this rather new and fascinating field of research going.