ig 2 An early Roman ostrakon in Greek from Berenike (Red Sea Coast of Egypt) bearing a note written by Herennius to a military colleague. © Photo by John Peter Wild. By courtesy of the Berenike Project. Archaeology฀ and฀the฀dating฀of฀textiles John฀Peter฀Wild To an archaeologist confronted with an archaeological artefact two questions spring to mind immediately: Where was it found? What date is it? The questions are obviously linked; but it is the second, fundamental, question that needs to be addressed here. Until the scientiic dating revolution, which began in 1949 and was led by radiocarbon dating,1 archaeologists had their own well-tried methods for constructing chronological frameworks. For historical archaeologists, coins which carry a legend and/or date underpin their thinking (ig 1a–b). I want to concentrate primarily on these traditional methods, insofar as they can be applied to textiles from the Nile Valley. The term ‘Coptic’, applied to the best known group, may not be politically correct any more, but it is a convenient label because there is broad agreement about what it means: textiles, mostly from burials, dating from the early Roman occupation until at least the tenth century ad. But I propose to look sideways, too, at textile inds which are broadly speaking Roman, but are not from burials. When Coptic textiles irst became an object of interest to antiquaries, art-historians, collectors and dealers, little was known about their background. Now, however, we are much better informed about the archaeological, historical and cultural matrix to which they belong, thanks to recent archaeological excavations and the parallel world projected by papyrology. The ability to date accurately and closely, I need hardly add, is essential to pin these disparate ields together, and encourage them to shed light on one another. Archaeologists generally draw a distinction between inds from settlement sites, where most material uncovered has been thrown away deliberately or lost, and formal burials, which (until Christian ideas prevailed) were oten furnished with high-value objects demonstrating the status of the deceased and the wealth of his or her family. The same distinction can be translated into textile terms: a stag in gold-thread embroidery from a Viking grave at Birka,2 for example, contrasts with the old rags from the Viking port of Hedeby which served for caulking and tarring ships.3 In Egypt the contrast lies between the textiles from inhumation burials – conventionally 19 On฀methods ig 1a–b Some Roman coins from Egypt showing their obverse (a) and the reverse (b). Top let: Constantius ii (346–361 ad); top right: Diocletian (284–305 ad)(mint of Alexandria); below let: Constantius ii; below right: Hadrian (117–138 ad) (mint of Alexandria). © Photos by D. Trillo at G-ten. 1 2 3 Aitken 1990. Geijer 1938, 118, pl. 36, 1–2. Hägg 1984, 15–19; Hägg 1991, 268–273. ig 3 The pottery processing area at Berenike, showing an amphora surrounded by bags and baskets of potsherds. © Photo by John Peter Wild. By courtesy of the Berenike Project. 4 Wild/Cork/Cooke et al. 1998; Wild/Cooke/Cork et al. 1997. called ‘Coptic’, and including the best clothes of the deceased – and the myriad scraps of cloth excavated from rubbish-dumps and occupation layers in or near towns, forts and industrial and religious centres. They represent opposite ends of the same spectrum; but, taken together, they open for us a window upon what was available to the Egyptian wearer and consumer in the Roman period. When we ask what the date of a textile is, what do we mean? There are ive horizons in a textile’s life: 1. its manufacture; 2. its primary use as a garment or furnishing; 3. its secondary use, that is, recycled in some form; 4. its circulation as rubbish; and 5. inal deposition in the ground. Archaeology฀and฀the฀dating฀of฀textiles Radiocarbon dating focuses on the date of manufacture, the death of the organic raw material from which a fabric was made. Graves contain textiles in primary use, supplemented by material from horizons 3 or even 4. Middens on settlement sites, the by-product of everyday life, are the last resting place of rubbish (horizons 4 and 5). Ideally, we should date each of these horizons separately in turn, and one day that may be possible. New research, for instance, already enables us to estimate for how long a garment was worn, efectively the span of horizons 2 and 3.4 Let us now look at some actual examples of how archaeological textiles from settlement sites can be dated through their stratigraphic contexts and what can be learnt thereby. On฀methods 20 ig 4 Fragment of wool furnishing fabric in wet-faced compound tabby weave (taqueté) from Berenike (be97 118), dumped together with datable pottery and dated ostraka in the early Flavian period. © Photo by John Peter and Felicity C. Wild. By courtesy of the Berenike Project. 0 5 The inhabitants of the Roman port of Berenike on the Red Sea let large dumps of rubbish in and around the town which with time have spread out and consolidated. Two such middens or midden complexes have been excavated. The north-western one is of early Roman date. Scraps of cloth accumulated in its layers as it built up around the middle of the irst century ad,5 associated (and that is a technical term) with pottery that can be dated from sites elsewhere to no later than about ad 70.6 Documents written on potsherds, ostraka, were also being dumped there (ig 2, p. 18) and the latest of them carries a written date of about ad 75, in Vespasian’s reign.7 That must also be the date by which the textiles in the midden had been dumped there. The second midden (probably in fact a group of contemporary middens) built up during the last century of Berenike’s existence when the settlement focus had moved further to the south east. The pottery in this case belongs to the late fourth and ith centuries ad, much of it imported wine and oil amphorae (ig 3). There are few helpful coins and no papyri or ostraka, but an historical source tells us that the port had almost ceased to function by ad 525,8 and that must be the terminal date of the textiles, too. The textile assemblages from these two deposits, some 300 years apart in date, are quite diferent from one another in character. The irst century textiles are mainly ine wool and linen fabrics, some with simple tapestry decoration. Some layers have a high proportion of pieces of sailcloth and webbing and other utilitarian items.9 The ith-century deposit by contrast contained all sorts of highly coloured discarded fabrics in a range of qualities: pile hangings, resist-dyed curtains, scraps of igured tapestry and fragments of well-worn wet-faced compound tabby (taqueté) (ig 4).10 Berenike is geographically isolated, and there are probably only two signiicant factors which could account for the contrast between the early and the late Roman textile assemblages. The irst is the population itself. Early Berenike was strongly inluenced by the military units stationed nearby, while late Roman Berenike saw an inlux of desert dwellers from the south. Their respective needs and tastes must have conditioned local textile use. The second factor is the changing repertoire of Roman weavers in Egypt as a whole who supplied Berenike. Take the wet-faced compound tabbies. There are ten examples from late Roman Berenike11 – and they are of 21 Archaeology฀and฀the฀dating฀of฀textiles On฀methods 5 6 7 8 9 10 11 12 Wild/Wild 2000a. Tomber 1999, 132; Tomber, personal communication. Bagnall/Helms/Verhoogt 2000, 7. Wild/Wild 1998, 235. Wild/Wild 2001a; Wild 2002. Wild/Wild 1996; Wild/Wild 1998. Wild/Wild 2000a, 260–261. Vogelsang-Eastwood 1988. ig 5 Sections through the midden which built up during the irst and second centuries north of the Roman fort of Maximianon/Al-Zarqâ in the Wadi Hammamat, Eastern Desert of Egypt. ig 6 © Drawings by J.-P. Brun. By courtesy of J.-P. Brun. Histogram demonstrating the changing proportions through time of textile types found on Elephantine island in the Upper Nile. © Drawing Roberta Cortopassi. By courtesy of Roberta Cortopassi.         13 14 15 16 17 18 19 20 Wild/Wild 2000a, 256, igs 11–12, pls. 11–13; Pliny, nh viii, 196. Cardon 2003a. Brun 2003, 69–71. Adam/Brun/Reddé 2003, 122–123; cf. Bingen 1996; Wilson 2002. Cortopassi 2004b. Kortüm/Lauber 2004. Taylor-Van Rosevelt, personal communication; Batcheller 2002, 47–48; Van Minnen 1994, 229; Van Minnen 1995. Rassart-Debergh 1997, 35–51; Cauderlier 1985, 11–16. course common in Coptic collections12 – but there is only one from the pre-ad 75 midden, which must be one of the earliest dated pieces anywhere.13 Recent excavations of some upstanding middens outside irst- and second-century forts guarding roads in the Eastern Desert have revealed classic stratigraphy, and some splendid textiles published by Dominique Cardon.14 The layer sequences (ig 5) have been meticulously recorded by the excavator, though in his report he felt obliged to justify the time and efort devoted to the exercise.15 He has demonstrated that the layers, closely dated by associated pottery, glass and many dated ostraka, did not simply build up at a steady pace, but relect a number of episodes of large-scale clearance within the forts.16 In assessing the textiles one has to take this into account. On the island of Elephantine, an important religious centre with a large dependent population close to Aswan, the results of excavation are even more striking.17 Between the ith and the ninth century, a period which saw the beginnings of Islamic control, Roberta Cortopassi has shown that the popularity of decorated clothing, especially simple stripes and checks, rises markedly at the expense of plain fabrics (ig 6). The dated stratigraphic sequence in other words presents evidence of technical and cultural change. Archaeology, however, contains pitfalls for the unwary. Technically, a layer is dated by the latest object in it before it became sealed. Archaeology฀and฀the฀dating฀of฀textiles ฀ ฀ ฀ However, much of the content of the layer will be much older, but there is no way of saying how much older. Excavators, for instance, have noted that most third-century coins are found in fourth-century layers18 along with a lot of (presumably) third-century rubbish. As a consequence, the latter half of the third century is a virtual dark age archaeologically in many provinces of the Roman Empire. The order in which layers were laid down can also be misleading. At Karanis in the Faiyum it has been suggested that passers-by threw rubbish through the cellar windows of empty houses.19 When the upper loors collapsed, they spread early textiles on top of a layer below containing later textiles – the reverse of normal logical stratigraphy. Some of the site inds from Karanis bring us face to face with the kind of textiles we expect from burials, to which I now turn. Settlement archaeology may have its problems; but the archaeology of Roman cemeteries in Egypt is a disaster area, as is widely recognised. Many thousands of textiles have been recovered from graves since the late nineteenth century, by fair means and foul; they have no dated context now, and many inds cannot be safely attributed to a speciic site. A century ago Egyptology was in the Heroic Age; but there were plenty of villains, too. One of the most notorious is Albert Gayet, who excavated – if that is the right word – at Antinoupolis. By 1902 he had uncovered some 40,000 corpses.20 He kept virtually no site records, but on the boat returning home he wrote some catalogue notes for the exhibitions of selected inds which he organised in Paris each year. On฀methods 22 ig 7 Some oicers in the Roman army of Egypt c. 550–600 ad, as visualised by Graham Sumner on the basis of the documentary and textilearchaeological evidence. © Drawing Graham Sumner. Reproduced by courtesy of Osprey Publishing. Gayet’s contemporary, William Matthew Flinders Petrie, is regarded as one of the heroes, though he was only marginally concerned with cemetery digging. He prided himself on his meticulous recording, but his published descriptions of graves and their contents read to us as mere summaries, though his notebooks sometimes contain more information. Digging at Hawara in 1888, for example, he uncovered a body accompa- 23 Archaeology฀and฀the฀dating฀of฀textiles On฀methods 21 22 23 24 25 Whitehouse (forthcoming). Petrie 1889, 12, 21; Petrie 1893, 101–104; Roberts 1997. Griggs/Kuchar/Woodward 1993; South 1998. Du Bourguet 1964, 22–35; Stauffer 1992; Schrenk 2004, 13–19. D’Amato/Sumner 2005, pl. D. nied by textiles which include purple wool tapestry roundels with lying-needle decoration in lax (they are now in the Ashmolean Museum, Oxford).21 With it was an unworn coin of the House of Constantine, of about ad 340, the importance of which for dating the textiles he recognised at once.22 But for Petrie, Coptic textiles presented a convenient currency for thanking his sponsors: some 20 years ago I found in a museum basement unopened parcels of Coptic textiles, just as Petrie had posted them from Cairo. It is easy to condemn Gayet and Petrie from a modern standpoint; but they were typical of their time, only answering the questions which seemed important to them. Some ten years ago there was a television programme on the excavations by the Brigham Young University, Utah, on the Roman and later cemetery of Fag el-Gamus in the Faiyum. The purpose of the dig was to recover dna samples for comparison with those of the modern population. In one scene a lady was said to have been buried in and with 25 textile items, many complete garments. So far, we only have brief interim reports23 and nothing speciic on dating. It sounds like a glorious opportunity – missed. I have already trespassed upon territory which rightly belongs to other contributors to this volume, and I have no intention of venturing upon a topic particularly familiar to students of Coptic textiles – dating by art-historical comparison.24 In fact it is more satisfactory as a method of relative dating – placing textiles before or ater one another in sequence of production – than as a method of absolute dating, attributing to them dates bc, ad or whatever. In other Roman industries, like pottery, workshop debris provides the starting point in questions of date of production and origin; but for textiles only ind-distribution maps are available, and they have to be treated with extreme caution. When looking at a series of supposedly characteristic workshop traits, for example – can they be explained in terms of developments through time or do they relect contemporary technical and cultural phenomena geographically dispersed? There are so many variables. If we had been in Egypt in the middle of the sixth century ad, we would have Archaeology฀and฀the฀dating฀of฀textiles encountered soldiers at checkpoints along the roads dressed in colourful uniforms – Coptic textiles in fact. Fig 7 is Graham Sumner’s vision of them, based on dated wall-paintings, documents and of course surviving textiles.25 Such vignettes bring home to us the centrality of sound and precise dating in any attempt to combine a range of types of evidence; a ‘radiocarbon revolution’ for textiles, let us hope, is just round the corner. On฀methods 24 Textiles dated฀by the฀archaeological฀ context