In 1973, a sponge diver reported the discovery of an underwater shipwreck at Serçe Limani on the south coast of Turkey, opposite Rhodes. The wreck was investigated by Professor George Bass, of the Institute of Nautical Archaeology, between 1977 and 1979. His investigation revealed that part of the ship’s hold contained (among other commodities) a cargo of three tonnes of cullet (glass destined for remelting) in the form of chunks of raw glass, which were by-products from a glass factory, and broken glass vessels. In addition to about 150 amphorae and other pots, the rest of the hold must have contained more cargo (to balance the ship), but no trace of it survived.
The glass from Serçe Limani offers a unique insight into medieval glassmaking because the sample is very large, it comes from a closely datable context, the excavation was documented with extraordinary care, and because the finds were studied in minute detail. Professor Bass’s final report on this remarkable discovery is in active preparation and probably will appear in the next 12 to 24 months.
Thanks to the meticulous work of Bass and his colleagues, we know that one or more of the ship’s last ports of call were in the Levant, that the ship sank in or shortly after AD 1025, and that the glass in the hold was already broken when it was brought on board and stowed in baskets. In other words, the broken glass was an article of commerce. My purpose here is to place the cargo of the Serçe Limani ship in the context of medieval glassmaking in the Levant. To convey an idea of the %%scale%% of production, I propose to translate quantities of glass into numbers of old-fashioned glass Coca Cola bottles (which weigh about 250 grams). The cullet on the Serçe Limani ship was sufficient to make 12,000 Coke bottles.
The sources of information about medieval glassmaking in the Levant are scattered and, although a great deal of work has been done on individual sites and documents, they have never been studied comprehensively. No medieval description of Levantine glassmaking is known but anecdotal, eyewitness information about this industry exists in Latin, Hebrew, and Arabic documents. Archaeological discoveries tell us about the range of products and the distribution of particular forms, and about factories where glass was made. Chemical analyses inform us about raw materials, and ethnographic studies suggest how these materials may have been melted. Together, our sources furnish enough information for us to understand, in general terms, how the Levantine glass industry functioned in the Middle Ages. They show clearly, for example, that the industry had three components: primary, secondary, and tertiary production.
The essential raw materials for making glass are soda, lime, and silica (collectively known as batch ). Sand is the source of the silica that makes up about 70%. Soda (about 15%) was added to reduce the temperature at which the sand melted. Lime (about 8%) makes the glass chemically stable. The remaining 7% consist of impurities and sometimes of additives to change the colour of the glass. It sounds improbable, but seems to be true, that most early glassmakers were unaware that lime was a necessary ingredient; they introduced it by accident, mixed with the sand or the soda.
In Roman times, the preferred source of the sand (according to Pliny) was the beach at the mouth of the Belus (Na’aman) River near Akko, Israel. There, the sand contains quantities of broken shells, and these provided the lime necessary to stabilise the glass. Roman and early medieval glassmakers in the Levant used a mineral form of soda (natron), which they imported from Egypt. During the 9th century, however, they began to use soda derived from the ashes of halophytic plants, which grow locally. The plant ash also contained lime and, since too much lime ruins the glass, this must have compelled glassmakers to use sand with a reduced amount of shell, or crushed quartz pebbles, or to purify the plant ash. Nevertheless, in the decades around AD 1200, two eyewitnesses, William of Tyre and Jacques de Vitry, noted that glassmakers used sand from the beaches of Akko and Tyre. How the medieval glassmakers reduced the amount of lime in their plant ash to compensate for the lime in the sand, is not yet known.
The nature of the medieval glass industry in the Levant is best understood in the context of the Roman industry from which it evolved. The investigation, in the 1960s, of two sites in Israel, Bet She’arim and Jalame, followed by investigations at other sites, revealed that in Late- and post-Roman times the manufacture of glass from raw materials (primary production) and the manufacture of objects from raw glass (secondary production) sometimes—perhaps habitually—occurred at different places: raw glass was produced at Bet She’arim, for example, and glass vessels at Jalame. This separation of primary and secondary production is implied in a 3rd-century AD section of the Babylonian Talmud, which draws a distinction between transporting raw glass and glass vessels, and thereby tells us that raw glass might be moved from one place to another. Moreover, chemical analyses of vessels from sites in many parts of the Roman Empire, including secondary workshops in France and Italy, some of which are as early as the 1st century AD, reveal that they have a small number of strikingly similar compositions. When these compositions are compared with analyses of sand from different regions, they indicate that most of the glass had been made with sand from the Levant or Egypt, and with natron. In other words, even in the 1st century AD, either sand and natron, or raw glass, were traded over very long distances. Indeed, archaeologists now believe that manufacturers in the eastern Mediterranean supplied raw glass to secondary workshops all over the Roman Empire.
The export of raw glass to secondary workshops in distant parts of the Roman Empire implies that primary production was on a very large %%scale%%, and this is confirmed by four discoveries in Israel and Lebanon. In the early 1950s, excavations at coastal Arsuf (ancient Apollonia), Israel, revealed the remains of a rectangular furnace with a melting tank 3.8m long and 2.5m wide. The tank was filled with glass to a depth of 0.25m. A few years later, excavators at inland Bet She’arim, also in Israel, uncovered a glass slab measuring 3.8 x 1.95 x 0.45m and weighing very nearly nine tonnes. Chemical analyses of the slab indicated that the batch contained far too much lime, rendering the glass unusable; consequently, as at Arsuf, it was abandoned. Perhaps, without knowing it, the glassmakers were adding lime with both the sand and the plant ash. This possibility has led to the suggestion that the Bet She’arim slab was the result of an early attempt to substitute plant ash for natron.
At a third site in Israel, Bet Eli’ezer, excavations in 1992 revealed the remains of no fewer than 17 tank furnaces similar in size to the furnaces at Arsuf and Bet She’arim. Finally, the remains of several tank furnaces exist at Tyre, Lebanon, the largest of which is 6.0m long and 3.0m wide. The last time it was used, the tank (now somewhat smaller because it had been relined at least twice) was filled to a depth of about 80cm: in other words, it contained more than 30 tonnes of glass - enough to make at least 120,000 Coca Cola bottles. All of these furnaces—Arsuf, Bet She’arim, Bet Eli’ezer, and Tyre—were built in Late Antiquity or the early Middle Ages; the chronology is not yet certain.
It is interesting to note that a document from the Cairo Geniza, written in 1011, mentions 37 ‘bales’ of glass exported from Tyre. Were they, one wonders, similar to the baskets that went down with the Serçe Limani ship, a decade or two later? According to S. D. Goitein, who spent a lifetime studying the Geniza documents, the total weight of 37 bales would have been almost eight and a half tonnes - the equivalent of 34,000 Coke bottles. Incidentally, the apparent weight of the shipment in 1011 is sufficiently close to the weight of the Bet She’arim slab, and the presumed capacities of the tanks at Bet Eli’ezer, to hint at the possibility that the shipment may have been the usable glass from a single melt.
The size of the furnaces raises questions about the duration of the melting process and the methods used in the Middle Ages to maintain the required temperature of 1000°C or thereabouts for a significant length of time. We have no direct means of answering these questions. The ability to melt batch on the scale accomplished at Bet Eli’ezer or Tyre no longer exists outside industrial plants in either the Mediterranean region or the Near East; indeed, traditional methods of making glass on this scale probably survive in very few places worldwide. One such place, however, is Jalesar, 150km from Delhi, India, where a traditional factory provides clues about how the medieval Levantine establishments may have operated. Two Danish scholars, Jan Kock and Torben Sode, have reported that the tank furnaces at Jalesar are reverbatory furnaces with capacities of about three tonnes. The melting process requires a team of seven, working in shifts, and it continues uninterrupted for 30 days, during which about 300kg of raw materials are poured into the tank every third day, together with chunks of raw glass from an earlier melt. The long duration of the process, the fact that only a small amount of batch is melted at any one time, and the presence of cullet (which melts at a lower temperature than the batch) allow the glassmakers to operate the furnace at a relatively low temperature. At the end of the melting process, the furnaces at Jalesar are allowed to cool for two weeks, after which the huge ‘cakes’ of glass are broken up and removed. Only the top half of the cake is of a quality good enough to sell; the bottom half is crushed and recycled as cullet.
We cannot assume that medieval Levantine glassmakers worked in the same manner as their modern counterparts at Jalesar, but at the very least these observations from India suggest that the production of glass at factories like Tyre was a lengthy business. If it takes six weeks to melt and cool three tonnes of glass at Jalesar, how long did it take to melt and cool nine tonnes at Bet She’arim or 30 tonnes at Tyre? Consequently, we have to imagine that the glassmakers of Tyre probably spent months (not weeks) preparing for, and completing, their largest glass melting operations. The observations by Kock and Sode also suggest that it would have been sensible to introduce raw materials and cullet gradually throughout the melting process.
The glass produced at a few factories like Tyre supplied many secondary workshops that might be in the same region or, as the Serçe Limani shipwreck demonstrates, at a significant distance. Although several medieval workshops have come to light in the Levant, only one has been published: at es-Samariya (Crusader Somelaria), 5km north of Akko. The workshop, which apparently operated in the 13th century, made bottles and beakers. It had a single furnace with a tank for remelting glass that measured only 40 x 50cm and had a capacity of about 100kg, enough to make just 400 Coca Cola bottles.
The remains of a somewhat better preserved secondary workshop were uncovered at Corinth, Greece, in 1937. Excavators uncovered evidence of two medieval workshops on the site of the ancient agora. No structures survived at the Agora North-east workshop, but a building in the Agora South Centre area contained a small furnace, surrounded by working debris and fragments of cups, beakers, and bottles. The excavators believed that the workshop dates from the 11th and 12th centuries; in which case, it would have been operating at the time when the Serçe Limani ship was wrecked. However, a reassessment of the finds suggests that it did not function until the decades around 1300, and so it postdates the Serçe Limani shipwreck by nearly 200 years.
Just as primary and secondary production took place in different workshops, so did the third stage in the manufacturing process, which consisted of decorating a small percentage of the objects after they had been formed and annealed. The only tertiary technique that is known to have been practiced in the Levant around 1025, when the Serçe Limani ship went down, was cutting. This involves incising the surface of the object with a rotating wheel fed with an abrasive such as emery. The Arabic writer al-Muqqadasi reported that cut glass was one of the exports of Tyre in 985 and, 40 years later, the Serçe Limani ship was carrying fragments of cut glass in the cargo and complete vessels in the quarters occupied by the crew and passengers.
I began this survey by describing the Serçe Limani shipwreck and what it tells us about medieval glassmaking in the Levant. I propose to end by noting two things it does not tell us. First, we have no idea of the ship’s destination. Was it heading north towards Byzantium, whose glass industry in the 11th century is almost completely unknown to us? Was it heading west towards Greece, about which we are equally ignorant, now the workshop at Corinth has been redated to the years around 1300? Or was it travelling even farther afield, perhaps to Italy?
Secondly, we have no idea what was in the rest of the hold, except (since nothing remained) that it was perishable. One possibility is that it was carrying Levantine plant ash which, 200 years later, was the soda most sought-after by Venetian glassmakers. In any case, wherever it was going and whatever else it contained, the ship was carrying enough glass to make 12,000 Coke bottles. Thus, together with the documents that refer to Tyre, the Serçe Limani cargo casts a unique light on primary, secondary, and tertiary glass production in the Levant in the early 11th century.
Dr. David Whitehouse
Senior Researcher, The Corning Museum of Glass
Trained as an archaeologist, David Whitehouse has directed excavations in Europe, North Africa, and the Middle East. In 2001, he organised, with Dr. Stefano Carboni, the exhibition Glass of the Sultans (Minerva, Nov/Dec 2001, pp. 8–16).
For further details about underwater excavations conducted by the Institute of Nautical Archaeology, Texas A & M University, see: http://ina.tamu.edu/.