1 A Knotted-Pile Mat from the Roman Fort at Vindolanda near Hadrian‘s Wall John Peter Wild I have spoken at three NESAT symposia about textile items from the Roman fort at Vindolanda1, and I apologise for returning there again: but the site never ceases to produce surprises. he fort lies 1.5 km behind Hadrian’s Wall, but on the Stanegate, a patrol road with garrisons established in the late irst century AD, some thirty years before Hadrian’s Wall2. he fort’s claim to fame is the exciting collection of writing tablets and wool textiles which accumulated in rubbish deposits preserved in the permanently damp conditions on part of the site3. he surprise ind to be discussed here came from the illing of a Period I defensive ditch and so can be closely dated to c. AD 85–924. It consists of two pieces of coarse disintegrating dark brown wool tabby, with lots of loose threads (ig. 1, 2). he Vindolanda research team had looked at them at least twice without noticing anything remarkable. But last year, fresh from having analysed some heavy cotton pile fabrics at Berenike in Egypt, I suddenly spotted that the Vindolanda fragments have a knotted pile, scarcely visible amid the yarn debris. hey took on a new importance. I will begin by describing them briely, then review the parallels, and inally make some suggestions about the textile’s original function. he technical analysis he warp is all 2-ply wool, S-plied from medium Z-spun yarns – but alternate warp threads are thick and thin, with a combined count of 5–6 threads per cm. he weft consists of pairs of medium Z-spun single yarns, about 2 pairs per cm, wide spaced. Part of a selvedge survives (ig. 3): the weft pairs simply return into the next shed, but there is a special sequence of warp in the selvedge zone. Reading from the outer edge, there are 8 thick, 8 thin, 4 thick, 4 thin warp threads, and then the ground weave proper with alternating thick and thin warp threads (ig. 4). he pile is formed of symmetrical knots («Ghiordes» or «Turkish» knots),5 which are bound by the thin warp threads only. he rows of knots are separated from one another by one shot of paired weft, and the knots of each successive row are bound by the same two thin warp threads as their predecessors. he pile consists of pairs of yarns, now very weak Z-spun, projecting at least 13 mm from the ground weave surface (ig. 5). here is pile on one side only, and while one might expect it to cover the entire surface, the selvedge zone (29 mm wide) is bare and one Fig. 1. Mat fragment from Vindolanda (VIN 639) with selvedge (warp runs horizontally). Fig. 2. Mat fragment from Vindolanda (VIN 644). Fig. 3. Close-up view of selvedge of mat fragment from Vindolanda (VIN 639). 2 A Knotted-Pile Mat from the Roman Fort at Vindolanda near Hadrian‘s Wall of the smaller loose fragments may have an even wider pile-free area. he pile was probably cut at the time of insertion. hree knots, randomly distributed, consist of three, not two, yarns and a fourth knot has yarns of a completely diferent shade of brown from the rest. However, in conventional carpet weaving symmetrical knots are inserted around two adjacent warp threads on a single warp sheet; here at Vindolanda a shed has been opened so that alternate warp threads, the thin ones, anchor the knot, and the thick warp threads at the back play no part directly. his is a diagnostic feature. Hoo (Sufolk) contained fragments of wool textile with rows of assymetrical («Sehna») pile loops – presumably an import from Europe12. he pile-loop tradition in Europe in fact may go back to the Villanovan Iron Age, as inds from Verucchio suggest13. More relevant is the second-century Roman 2/1 twill from Chalons-sur-Saône, which has rows of loops at intervals and can be paralleled in the eastern Roman provinces14; but there is no sign so far of cut Ghiordes knots anywhere in the Roman West. he home of heavy textiles with pile-knot decoration – carpets, in other words – is Central and Western Asia15. he famous carpet from Pazyryk in the Altai mountains – this is a detail Fig. 4. Structure of the Vindolanda mat. Wool and Dye Penelope Walton Rogers has examined the wools and looked for traces of dyestufs (p.●●$$–$$). he thick ground warp and the paired weft were spun from Hairy leece types; the thin warp is of Hairy Medium wool, and the pile of Medium wool. (I will examine the implications of this later.) Only the weft has any natural pigmentation. It would have appeared mottled brown and white, but in fact was concealed by the warp. he warp was white, but the pile was dyed blue with indigotin, from woad or an indigo plant. he alternating stripes of thick and thin warp against the selvedge might be decorative, but there is no indication that they were in contrasting colours. he Comparisons What are the parallels, and what - if anything - can they tell us about the source of this textile? Its pile structure is quite diferent from most of the types of surface enhancement known in northern Europe. Only two neolithic linen textiles from Charavines (F) and Zürich-Mythenquai (CH) have been made with a inserted turkish knot during weaving6. More familiar examples are the bast-ibre pile hats of the Alpine Neolithic7 and the Bronze Age wool caps, which have a pile added with a needle after weaving8.he technique lived on into the Viking period: a hood from Hedeby has pile formed of unspun animal hair9, and there is a similar tenth-century piece from Heynes in Iceland10. here are some other examples of Anglo-Saxon and Viking date11. However, the seventh-century royal Anglo-Saxon burials at Broomield (Essex) and Sutton Fig. 5. Close-up view of Ghiordes knots on mat fragment from Vindolanda (VIN 639). – has been radiocarbon-dated to 328-200 BC16.. It has symmetrical Ghiordes knots, while a second contemporary carpet from Bashadar has asymmetrical Sehna knots17. Earlier still is a Ghiordes-knotted carpet fragment from the same region, now in Antwerp, that has a radiocarbon date of the seventh century BC18. Archaeological exploration of Roman-period - or, more correctly, Han-period – sites around the Tarim Basin in NW China has brought to light numerous fragments of polychrome wool carpets, showing a variety of types of knot19. More scraps of knotted-pile fabrics have been found in the burial caves at AtTar near Kerbala on the Euphrates in disputed territory between the Roman and Parthian Empires20. An important study of the types of knot represented there highlights the position of At-Tar at a major cultural interface21. While the published distribution map22 may be an oversimpliication of a complex situation, it certainly testiies to the strength of the knotted-pile tradition across a broad zone in the East. Sites within the eastern Roman provinces have also yielded carpet-like fabrics; but, whereas most knots recorded at At-Tar, at A Knotted-Pile Mat from the Roman Fort at Vindolanda near Hadrian‘s Wall sites in Roman Syria23 and in Central Asia are inserted around adjacent warp threads in a single warp sheet, many of those in the Nile Valley can be seen to have been inserted after a shed has been opened, as seems to have been the case with the Vindolanda fragments24. However, almost all the Nile Valley knots are asymmetrical, and may have begun life as loops, cut later, like an Egyptian example now in New York25. So far I have only discovered one near-parallel to the Vindolanda fabric’s pile structure: it comes from the Roman mining site at Mons Claudianus in the Eastern Desert of Egypt (MC 1093), and Lena Hammarlund and Lise Bender Jørgensen have very kindly brought it to my attention. It has Ghiordes knots inserted after a shed has been opened, so that the ends of the knots project between pairs of the paired warp on just half the warp sheet. It is a long way from Vindolanda to Mons Claudianus, and I know of no Roman Ghiordes-knotted textiles in between. he fact that all the yarn in the Vindolanda fabric is Z-spun would, as a rule of thumb, be taken to indicate an origin in Roman Europe rather than in the Eastern provinces. Moreover, Penelope Walton Rogers has pointed out that Hairy wool, technically 3 speaking, is not known in Britain before the Viking period, and Medium wool is quite rare (●●p.$$-00), so the textile probably came to Vindolanda with its owner, presumably a soldier serving in the First Cohort of Tungrians from Northern Gaul. But where he obtained it, I do not know. It could have been in Gaul. Why did he need it? What was its function? At the Roman port of Berenike on the Red Sea a large number of Indian cotton textiles have been recorded, many of which were carried there as the personal possessions of the incoming sailors. here were at least seven heavy pile fabrics. One (BE99 1527) has 4-ply warp, multiple weft threads and Ghiordes knots on both sides26. hey can be interpreted as sleeping mats, used by the sailors on board ship between Berenike and India, where beds would have been impractical. here is documentary evidence that rush mats were in demand in Berenike (O. Petr. 228, 233, 257, 262), perhaps used in conjunction with pile sleeping mats, as in modern Japan. Certainly, in burials at At-Tar and at Kerma on the Upper Nile, the dead were laid to rest on reed mats and pile textiles27. However, one member of the Roman leet at Alexandria writes that he had bought a little bed, a grabatulum28, the same word as is used in the New vestament for this kind of Fig. 6. Distribution map of finds of Roman-period textiles with carpet knots (after Fujii 1980). 4 A Knotted-Pile Mat from the Roman Fort at Vindolanda near Hadrian‘s Wall bed29. He presumably preferred it to a sleeping mat. here is direct evidence, however, that Roman veteran soldiers did sleep on the loor, on mattresses. he occupant of a former barrack room in the Roman colonia of Colchester abandoned his mattress in the corner when the Boudiccan rebels burnt the town in AD 6030. It had a diamond twill cover in wool and some kind of stuing. In conclusion, therefore, I suggest that a Tungrian soldier brought his blue pile mat to Vindolanda to sleep on. A recent barrack reconstruction, rather optimistically, gives the cavalry beds31, but it is more likely that the ordinary soldier was equipped with a mat. A thick mat might have made all the diference between a good night’s sleep – and none at all. Acknowledgements I should like to thank Pat and Robin Birley for permission to publish the Vindolanda pile mat, Jane Batcheller for help with photography, Penelope Walton Rogers for much technical advice, the Pasold Research Fund for supporting the analytical work, Lena Hammarlund and Lise Bender Jørgensen for information about the Mons Claudianus inds and to W. D. Cooke, C. R. Cork and Lucy Campbell, the Vindolanda textile team, for their support over the years. Bibliography Adams 1977 – Nettie Adams, he Vulgar Latin of the Letters of Claudius Terentianus (P. Mich.VIII, 467–472) (Manchester 1977). Barkova 1999 – Ludmila Barkova, he Pazyryk – ifty years on. HALI 107 (London 1999) 64–69, 110. Bechwith 1958 – John Beckwith, he Andrews Diptych (London 1958). Bender Jørgensen 1992 – Lise Bender Jørgensen, North European Textiles until AD 1000 (Aarhus 1992). Bergman 1975 – Ingrid Bergman, Late Nubian Textiles. he Scandinavian Joint Expedition to Sudanese Nubia 8 (Lund 1975). Birley 2002 – Anthony R. Birley, Garrison Life at Vindolanda: A Band of Brothers (Stroud 2002). Bowman 1994 – Alan K. Bowman, Life and Letters on the Roman Frontier: Vindolanda and its People (London 1994). Broholm/Hald 1935 – Hans C. Broholm, Margrethe Hald, Danske Bronzealders Dragter Nordiske Fortidsminder 11 (5/6) (København 1935). Cardon 1998 – Dominique Cardon, Neolithic Textiles, Matting and Cordage from Charavines, Lake of Paladru, France. In: Lise Bender Jørgensen/Christina Rinaldo, Textiles in European Archaeology, Report from the 6th NESAT-Symposium, 7–11th May 1996 in Borås (Göteborg 1998) 3–21. Crawford/Reynolds 1977 – Michael H. Crawford, Joyce M. Reynolds, he Aezani copy of the Prices Edict. Zeitschrift für Papyrologie und Epigraphik 26, 1977, 125–151. Crowfoot 1974 – Elisabeth Crowfoot, Textiles. In: Paul W. Lapp, Nancy, L.Lapp, Discoveries in the Wadi ed-Dâliyeh (New Haven 1974) 60–77. Crowfoot 1976 – Elisabeth Crowfoot, he textiles and leather. In: James F. Barfoot, David Price Williams, he Saxon Barrow at Galley Hills, Banstead Down, Surrey. Research Volume of the Surrey Archaeological Society 3 (Guildford 1976) 68–71. Crowfoot 1983 – Elisabeth Crowfoot, he textiles. In: Rupert Bruce–Mitford, he Sutton Hoo Ship Burial III (London 1983) 409–479. Crowfoot 1948/49 – Grace M. Crowfoot, Textiles from a Viking grave at Kildonan on the Isle of Eigg. Proceedings of the Society of Antiquaries of Scotland 83 (Edinburgh 1948–49) 24–28. Crummy 1984– Philip Crummy, Colchester Archaeological Report 3: Excavations at Lion Walk, Balkerne Gate and Middleborough, Colchester, Essex (Colchester 1984). Dimand 1933 – Maurice S. Dimand, An early cut pile rug from Egypt. Metropolitan Museum Studies (New York 1933) 151–162. Emery 1966 – Irene Emery, he Primary Structures of Fabrics: An Illustrated Classiication (Washington 1966). Fluck 2000 – Caecilia Fluck, Knüpfen. In: Heinrich Beck, Dieter Geuenich, Heiko Steuer (Hrsg.), Reallexikon der Germanischen Altertumskunde 17 (Berlin 2000) 58–65. Fujii 1980 – Hideo Fujii (ed.), Al-Raidan I: A Special Edition on the Studies of Textiles and Leather Object from AlTar Caves, Iraq (Tokyo 1980) 44–76. Fujii/Sakamoto 1990 – Hideo Fujii, Kazuko Sakamoto, Hill C. Al-Raidan XI, Textiles from At-Tar Caves, Part II- (1): Cave 16 (Tokyo 1990) 45–65. Fujii/Sakamoto/Ichihashi 1989 – Hideo Fujii/Kazuko Sakamoto/Miziko Ichihashi, Textiles from the At-Tar Caves, Part I: Cave 12, Hill C. Al-Raidan X (Tokyo 1989) 109–166. Fujii/Sakamoto/Ichihashi 1991 – Hideo Fujii, Kazuko Sakamoto, Miziko Ichihashi, Hill C. Al-Raidan XII, Textiles from the At-Tar Caves, Part II- (2): Cave 16, (Tokyo 1991) 157–165. A Knotted-Pile Mat from the Roman Fort at Vindolanda near Hadrian‘s Wall Geijer 1938 – Agnes Geijer, Birka III: Die Textilfunde aus den Gräbern (Uppsala 1938). Geijer 1979 – Agnes Geijer, A History of Textile Art (London 1979). Hägg 1991– Inga Hägg, Textilfunde aus der Siedlung und aus den Gräbern von Haithabu. Ausgrabungen in Haithabu 29 (Neumünster 1991). Hodgson/Bidwell 2004 – Neil Hodgson, Paul T. Bidwell, Auxiliary barracks in a new light: recent discoveries on Hadrian‘s Wall. Britannia 35, 2004, 121 –157. King 1982 – Donald King, Appendix B. Report on the woollen pile rug from tomb 193.105 at Qasr Ibrim. In: Anthony J.Mills, he Cemeteries of Qasr Ibrim (London 1982) 87. Kramer 1995 – Johannes Kramer, κράβατος, κραβάτιον und Verwandtes in den Papyri. Archiv für Papyrusforschung 41, 1995, 205–216. Maik 1994 – Jerzy Maik, he new textile inds from Palmyra. Archaeological Textiles Newsletter 18-19 (Manchester 1994) 11–12. Mamez/Masurel 1992 – Louis Mamez, Hubert Masurel, Étude complémentaire des vestiges textiles trouvés dans l‘embarcation de la nécropole du Caolino à Sasso di Furbara. Origini 16, 1992, 295–310. Monthel et al. 1998 – Gérard Monthel, Marie Schoefer, Odile Valansot, Dominique de Reyer, Witold Nowik, Les tissus gallo-romains de Chalon-sur-Saône: Étude pluridisciplinaire. Bulletin du CIETA 75, 1998, 21–36. Pfister/Bellinger 1945 – Rudolf Pister, Louise Bellinger, he Textiles: he Excavations at Dura-Europos: Final Report IV, Part II (New Haven 1945). Poulter 1998 – John Poulter, he date of the Stanegate and a hypothesis about the manner and timing of the construction of Roman roads in Britain. Archaeologica Aeliana, 5th series, XXVI (Newcastle upon Tyne 1998) 49–56. Rast-Eicher 1992 – Antoinette Rast–Eicher, Die Entwicklung der Webstühle vom Neolithikum bis ins Mittelalter, Helvetia Archaeologica 23/90, 1992, 56–70. Rast 1995 – Antoinette Rast, Le vêtement néolithique. In: Alain Gallay (ed), Dans les Alpes, à l‘Aube du Métal (Sion 1995) 149–154. Reisner 1923 – George A. Reisner, Excavations at Kerma (Cambridge, Mass. 1923). Riefstahl 1933 – Rudolf M. Riefstahl, Ein Knüpfteppich spätantiker Tradition aus Ägypten im Metropolitan Museum zu New York. Mitteilungen des Deutschen Archäologischen Instituts: Römische Abteilung 48, 1933, 127–152. 5 Rubinson 1990 – Karen S. Rubinson, Pre-Islamic carpets. In: Ehsan Yarshater (ed.), Encyclopaedia Iranica IV (London 1990) 858–861. Rudenko 1970 – Sergei I. Rudenko, Frozen Tombs of Siberia (translated by Michael W. hompson) (London 1970). Sakamoto 1985 – Kazuko Sakamoto, Ancient pile techniques from the At-Tar caves in Iraq. Oriental Carpet and Textile Studies I (London 1985) 9–17. Schlichtherle/Wahlster 1986 – Helmut Schlichtherle, Barbara Wahlster, Archäologie in Seen und Mooren: Den Pfahlbauten auf der Spur (Stuttgart 1986). Seiler-Baldinger 1991 – Annemarie Seiler-Baldinger, Systematik der Textilen Techniken (Basel 1991). Sheffer/Granger Taylor 1994 – Avigail Shefer, Hero Granger Taylor, Textiles from Masada: a Preliminary Selection. In: Ehud Netzer (ed.), Masada IV: he Yigael Yadin Excavations 1963–1965: Final Reports (Jerusalem 1994) 151–282. Sheffer/Tidhar 1991– Avigail Shefer, Amalia Tidhar, he textiles from the ‚En-Boqeq excavation in Israel. Textile History 22 (London 1991) 3–46. Stauffer 2002 – Annemarie Staufer, Tessuti. In: Patricia von Eles, Guerriero e Sacerdote: Autorità e Communità nell’età del ferro a Verucchio: La Tomba del Trono. Quaderni di Archeologia dell‘Emilia Romagna 6, 2002, 192–220. Stein 1921 – Marc Aurel Stein, Serindia: Detailed Report of Explorations in Central Asia and Westernmost China (Oxford 1921). Stein 1928 – Marc Aurel Stein, Innermost Asia: Detailed Report of Explorations in Central Asia, Kan-su and Eastern Iran (Oxford 1928). Tattersall 1949 – Creassey E.C. Tattersall, Notes on CarpetKnotting and Weaving (London 1949). Wearden 2003 – Jennifer Wearden, Oriental Carpets and their Structure: Highlights from the V & A Collection (London 2003). Wild 1977 – John-Peter Wild, he Textiles. Vindolanda III (Haltwhistle 1977). Wild 1979 – John-Peter Wild, Roman textiles from Vindolanda, Hexham, England. he Textile Museum Journal 18, 1979, 19–24. Wild 1992 – John-Peter Wild, Vindolanda 1985–1989: irst thoughts on new inds. In: Lise Bender Jørgensen, Elisabeth Munksgaard (eds.), Archaeological Textiles in Northern Europe: Report from the 4th NESAT Symposium, 1–5 May 1990 in Copenhagen. Tidens Tand 5 (Copenhagen 1992) 66–74. 6 A Knotted-Pile Mat from the Roman Fort at Vindolanda near Hadrian‘s Wall Wild 1993 – John-Peter Wild, Vindolanda 1985–1988: he Textiles. In: Robin E. Birley (ed.), Vindolanda III: Preliminary Reports on the Leather, Textiles, Environmental Evidence and Dendrochronology (Bardon Mill 1993) 76–90. Wild 1994 – John-Peter Wild, A hairmoss cap from Vindolanda. In: Gisela Jaacks, Klaus Tidow (eds), Archäologische Textilfunde – Archaeological Textiles: Textilsymposium Neumünster 4.–7.5.1993. NESAT V (Neumünster 1994) 61–68. Wild/Cork/Cooke 1996 – John-Peter Wild, Colin R. Cork, William D. Cooke, he use of image analysis to determine twist level in archaeological textiles. Archaeometry 27/2, 1996, 337–345. Wild/Cork/Cooke 1997 – John-Peter Wild/Colin R. Cork/ William D. Cooke, Analysis and evaluation of a group of Early Roman textiles from Vindolanda. Journal of Archaeological Science 24, 1997, 19–32. Wild et al. 1998 – John-Peter Wild, William D. Cooke, Colin R. Cork, Lucy Fang Lu, Vindolanda: some results of the Leverhulme Trust Programme. In: Lise Bender Jørgensen, Christina Rinaldo (eds), Textiles in European Archaeology: Report from the 6th NESAT Symposium, 7–11th May 1996 in Borås (Göteborg 1998) 85–95. Wild/Wild forthcoming – John-Peter Wild/Felicity C. Wild, he textiles. In: Wilhemlina Z. Wendrich, Steven E.Sidebotham (eds), Berenike, 01: Report of the 2001 Excavations at Berenike and the Survey of the Egyptian Eastern Desert (Los Angeles forthcoming). Abbreviations O. Petr – James G. Tait, Greek Ostraca in the Bodleian Library at Oxford and Various Other Collections I (London 1930) 82–152. P. Mich.viii – Herbert C. Youtie, John G. Winter, Papyri and Ostraca from Karanis VIII (Ann Arbor 1951). References Wild 1992; Wild 1994; Wild 1996. Birley 2002; Bowman 1994; Poulter 1998. 3 Wild 1977; Wild 1979; Wild 1993. 4 Birley 2002, 60f.. 5 Seiler-Baldinger 1991, 118 Abb. 194; Emery 1966, 221, 225–226. 6 Cardon 1998; Rast-Eicher 1992. 7 Fluck 2000, 61–62; Schlichtherle/Wahlster 1986, 70 Abb. 99; Rast 1995; Broholm/Hald 1935, 232–236, 247–250. 9 Hägg 1991, 90 Abb. 43, 4b , 35. 10 Geijer 1979, 199. 11 Crowfoot 1976, 68–71; Geijer 1938,131–132,183, Taf. 37,4; G.M. Crowfoot 1948–9, 24–28, pl. VII, 1a–1b; Fluck 2000, 63; Bender Jørgensen 1992, 39. 12 Crowfoot 1983, 471, ig. 331 (B4); 442–444, ig. 315a (SH10). 1 2 Staufer 2002, 215–220; Mamez/Masurel 1992, 298–300. Monthel et al. 1998, 29–30, igs. 13, 14; Shefer/Granger Taylor 1994, 205. 15 Tattersall 1949; Wearden 2003. 16 Barkova 1999; Rudenko 1970, 298–304, igs 141–145, pls 174–175; Rubinson 1990, 859. 17 Barkova 1999. 18 De Moor, pers. comm. 19 Stein 1921, 250, 433, 438, 445, 711, pls XXXVII, XLIX; Stein 1928, pl. XLIV; Fujii 1980, 76, table 1–6. 20 Fujii/Sakamoto/Ichihashi 1989; Fujii/Sakamoto 1990; Fujii/Sakamoto/Ichihashi 1991. 21 Fujii 1980, 44–76; Sakamoto 1985, ig..0.6. 22 Fujii 1980, 71 ig. 1-54. 23 Dura: Pister, Bellinger 1945, 47-49 (no. 224 is irregular Ghiordes knot); Palmyra: Maik 1994, 11–12, ig..1; Wadi ed-Dâliyeh: Crowfoot 1974, 61, ig.B; ’En-Boqeq: Shefer, Tidhar 1991, 17–18, ig.19. 24 Bergman 1975, 21–22, igs 10, 11; cf Shefer/Granger Taylor 1994, 166, ig. 98, 205. 25 Dimand 1933; Riefstahl 1933; King 1982. 26 Wild/Wild forthcoming. 27 Fujii/Sakamoto/Ichihashi 1989, 128; Reisner 1923, 301, pl.63, 2–3. 28 P. Mich.VIII, 468; Adams 1977, 43; Kramer 1995; Crawford/Reynolds 1977, 146. 29 Mark II, 4, 9; Beckwith 1958, pl. 6. 30 Crummy 1984, 42–47. 31 Hodgson/Bidwell 2004, 135, ig..8. 13 14