We have launched a new website and are reviewing this page. Find out more
Open daily 10.00 to 17.45 Admission free Menu

A Weft-Beater from Niya: Making a Case for the Local Production of Carpets in Ancient Cadhota (2nd to mid-4th century CE)

Swati Venkatraman Iyer

Graduate, V&A/RCA MA in History of Design


3rd-4th century texts from the abandoned Silk Road oasis of Niya contain references to valuable carpets that were traded for vineyards and slaves. By focusing on a wooden weaving implement, this discussion explores the social, cultural and economic connections between the carpets in these documents and the ancient textile fragments now in the Stein Loan Collection at the V&A and British Museum.


In October 1906, an undecorated wooden comb-like implement was rediscovered by the celebrated explorer-archaeologist Sir Marc Aurel Stein (1862-1943) during the second of his three landmark Central Asian expeditions.(1) For the object, this event marked the start of an arduous journey beginning in the present day Xinjiang Uyghur Autonomous Region of China (Fig.1) and ending halfway around the world in London; ‘a total journey of some 8,000 miles, including transport through high mountain ranges and across glacier passes, on camels, yaks, and ponies, and subsequent travel by cart, rail, and steamer’.(2) Currently held at the Victoria and Albert Museum in London, this artifact was first introduced to me in 2012 by Helen Persson, the V&A curator responsible for the Stein Loan Collection. As a practitioner of weaving, I had been interested in studying this wooden implement, which was believed to have functioned as a weaving tool (fig. 2). The following discussion focuses on this tool and explores its multiple meanings in the society and culture of a 2nd- to 4th-century oasis settlement, which had served as an important stop along the southern arm of the ‘Silk Road’.

Weft beater

Figure 2. Weft beater, unknown maker, excavated from Niya, China, 100–350 CE, carved wood. Museum no. LOAN.STEIN.530. © Swati Venkat/Victoria and Albert Museum, London

Detailed map showing the areas explored by Sir Aurel Stein

Figure 1. Detailed map showing the areas explored by Sir Aurel Stein

This seemingly ordinary object has played its part in a remarkable story of the international race to exploit an uncharted archaeological goldmine in a desolate backwater of China and, thereby, recover a forgotten chapter of history. In the 20th century, Xinjiang attracted foreign explorers and archaeologists, propelled by intellectual hunger and professional rivalry, and backed by competing world superpowers. Theirs was a story marked both by pioneering discovery and shocking destruction.(3) Since then, the Chinese have vigorously laid claim to their stolen heritage, today scattered amongst numerous foreign collections, while also adding their own considerable efforts and scholarship towards expanding our present knowledge of this material. But what remains of this heritage in China has turned out to be no less contentious, considering the implications it has had for competing ethnic and political claims in an ongoing separatist conflict.

The rich harvest of artifacts unearthed and brought back by Stein from this region between 1900 and 1916 included murals, sculptures, coins, documents, textiles, pottery, furniture and miscellaneous implements from numerous sites and spanning a wide swath of the region’s history. Most of this material was eventually divided up between his two main sponsors: the British Museum and the Government of India. Of the latter corpus, some 700 (primarily) textile objects from the second and third expeditions now at the V&A, were formally received by the Museum on loan from the Government of India and the Archaeological Survey of India during the 1920s and 1930s.

This arrangement was made possible largely because of the initiative of two prominent members of the V&A’s Department of Textiles, who were interested in these specimens ‘as early exemplars of complex weaving techniques’.(4) Most notable among these advocates of the Stein textiles was A. F. Kendrick (V&A Keeper of Textiles 1897-1924) who was especially keen that this material be made available for scholarly investigation ‘at no very distant date’.(5) Encompassing material ranging in date from 200 BCE to 1200 CE (6), the V&A’s Stein holdings undoubtedly offer a unique opportunity for tracing the development of textiles through this period. The Museum’s early outlook has led to the development of a substantial body of scholarship around the collection, as informed by related collections in other institutions, albeit with a predominantly technical focus.(7) Notable contributions from within the V&A include works by Donald King, Verity Wilson and Helen Persson, while externally Krishna Riboud, Anna Muthesius, Zhao Feng and Helen Wang have published studies that also attend to materials from the Stein Loan Collection.(8)

The immense technical and art historical value of ancient silks unearthed in this region is abundantly evident, but this material is also important because of what it has revealed about the larger dynamics of trade, diplomacy, technology and the religio-cultural processes that shaped these features over a significant span of time. Most notably, work such as that of Elizabeth Wayland Barber has firmly demonstrated the considerable potential of wool textiles from Xinjiang as a rich source of information in addressing key questions of historical interest.(9) This essay aims to go some way towards extending that line of research, responding to the V&A’s growing interest in moving beyond a  technical approach, towards exploiting the Stein Loan Collection textiles as ‘a valuable resource for […] economic, social and ethnographical studies’.(10)

While it is undoubtedly rewarding to investigate the textiles themselves, much more might be revealed through the study of the actual tools used to produce them. As Persson has observed, ‘finds of wooden weft beaters linked with pile making or tapestry weaving undoubtedly show that these types of textiles were produced locally’.(11) This study begins by probing that connection in some depth: by investigating whether the wooden comb-like object in the V&A’s Stein Loan Collection might have been used to produce pile textiles such as the extant specimens discussed here. I then consider these findings in relation to the evidence provided by contemporary documents in order to ascertain their significance. I shall draw mainly upon material from the site that Stein referred to as ‘Niya’ (2nd to mid-4th century CE, fig. 1), working primarily with Stein artifacts at the V&A and the British Museum. Niya has yielded a substantial body of written documents besides human remains and artifacts, offering the rare opportunity to assemble a richly detailed picture of a Silk Road oasis during the early centuries of the Common Era. By grounding my research in this local historical context, I adopt textiles as a lens through which a new dimension can be brought to our knowledge of life in ancient Niya. This investigation has benefited from a well-developed pool of Silk Road scholarship, limited in this instance to English language sources.

‘Implements of home industry’(12)

One of the non-textile items in the V&A’s Stein Loan Collection is a plain, comb-like wooden object (LOAN.STEIN.530, fig. 2) from the archaeological site of Niya.(13) Wedge-shaped in profile, it has a solid body provided with short, blunt teeth at one end and topped off with an elongated knob at the other. The thick body tapers down to a curved edge bearing eleven short teeth and a stump at one extreme indicating where the twelfth would have been. Measuring 9 cm wide and 24.2 cm long, this object has been carved out of a single piece of wood which is probably that of one of the poplar varieties native to the region.(14)

Comb beater

Figure 4. Comb beater, unknown maker, excavated from Niya, China, 3rd century CE, carved wood. Museum no. MAS.554. © The Trustees of the British Museum, London

Comb beater

Figure 3. Comb beater, unknown maker, excavated from Niya, China, 3rd century CE, carved wood. Museum no. 1907,1111.97. © The Trustees of the British Museum, London

Three comparable objects from the British Museum’s Stein holdings – 1907,1111.97 (fig. 3), MAS.554 (fig. 4) and MAS.727 (fig. 5) – are all dated to approximately the same period as the V&A piece.(15) The former two are from Niya and the third is from Loulan L.B., a roughly contemporaneous site situated at the eastern extreme of the ancient kingdom of Kroraina to which the settlement at Niya belonged during the latter part of its occupation (fig. 1). These specimens represent a range of sizes for what is essentially the same object type, showing variations primarily in the number of teeth and the spaces between them.(16) Even the largest can be held comfortably in an adult hand, although it quickly becomes clear that the knob was not meant to serve as handle. In all four cases, the knob is too small to allow sufficient grip and its neck is too slender to offer leverage on the bulky body of the implement.

Figure 5. Comb beater, unknown maker, excavated from Loulan, China, 3rd century CE, carved wood. Museum no. MAS.727. © The Trustees of the British Museum, London

Figure 5. Comb beater, unknown maker, excavated from Loulan, China, 3rd century CE, carved wood. Museum no. MAS.727. © The Trustees of the British Museum, London

Stein seems to have briefly entertained the idea that this kind of object was a currycomb used for grooming horses before eventually concluding that it was a weft-beater employed in weaving carpets or tapestry.(17) While the latter theory has since been generally accepted, this essay begins by seeking definite evidence in order to conclusively confirm it before proceeding to consider the context of the object’s use.

A brief overview of the mechanics of weaving offers a useful starting point for a close reading of the object. Any kind of weaving requires two sets of elements, normally yarns. The first is a set of parallel yarns known as the warp and this usually remains fixed and under tension. The second set of yarns, the weft, is introduced one at a time and perpendicular to the warp direction resulting in an interlaced surface. To incorporate the weft into the warp, the warp is divided into two layers between which the weft is then sandwiched. Next, the layers of warp are interchanged, bringing the formerly lower layer to the top in order to lock in the first weft while simultaneously creating a space for receiving the succeeding weft. Repetition of this process produces the simplest sort of woven fabric. Each time a weft is introduced, it must be packed down with the preceding wefts to create a compact, even surface. This function is performed by an implement known as the ‘beater’ which especially in the weaving of rugs, carpets and tapestry usually takes the form of a toothed handheld tool, allowing the weaver to attend to small sections of the weft (or knots) at a time. Its teeth fit between the unwoven warps and the beater is pulled (or pushed downwards/sideways, depending on the arrangement of the loom and the weaver’s position relative to it) with sharp strokes towards the newly inserted wefts in order to force them closer to one another and create a densely woven surface.

Weft beater

Figure 6. Weft beater, unknown maker, excavated from Niya, China, 100–350 CE, carved wood. Museum no. LOAN.STEIN.530. © Swati Venkat/Victoria and Albert Museum, London

Described by textile expert and master weaver Peter Collingwood as ‘extreme textiles’, rugs are designed to endure rough use.(18) Accordingly, the procedure used to produce them generally involves some extreme treatment. Ideally, although not always in practice, the warp for weaving a rug should be maintained under high tension and the weft must be ‘beaten’ down thoroughly. The function of the beater is well accomplished when it is heavy and sturdy enough to allow the weaver to create the requisite momentum without causing or sustaining damage.

On that score, the four wooden objects considered here would appear to fit the bill. The provision of a large, solid wooden body for the necessary impact is one expected feature, although, on account of the desiccation that these objects have no doubt suffered, it is only MAS.554, the largest, which feels suitably heavy when lifted. If such an object were indeed a currycomb, the provision of so substantial a body for a task primarily performed by the teeth would be hard to justify. In the case of the weft-beater, on the other hand, body and teeth are both equally crucial to accomplishing the task effectively.(19)

Weft beater

Figure 7. Weft beater, unknown maker, excavated from Niya, China, 100–350 CE, carved wood. Museum no. LOAN.STEIN.530. © Swati Venkat/Victoria and Albert Museum, London

The most satisfactory evidence in support of the weft-beater theory is to be found in the ample spaces between the teeth of the V&A specimen. A number of even black spots arranged in a roughly linear fashion are clearly visible in the ‘valleys’ between the teeth: on average three between each pair of teeth (fig. 6). The regularity in spacing and size of these black spots indicates that the causative agents were accordingly regularly spaced and uniformly sized, much as warp yarns would have been. The spots in the central valleys are clearer than those in the outer ones, being completely absent at the extremes. Taking into account the curvature in the toothed edge of the tool, this feature suggests that the causative agents were arranged more or less along a single plane (as we might expect warp yarns to have been) such that the central valleys would have come closest into contact with them and, thereby, suffered a higher degree of abrasion. The same abrasion produced long, dark striations along the sides of the teeth (fig. 7), indicating that the causative agents were linear in form, lay along a common plane and were roughly perpendicular to the plane of the object itself. This is consistent with the position warp yarns normally occupy relative to the teeth of a beater. Thus, we may take these marks of use as substantial proof of the object’s career as weft-beater.

However, several questions still remain. Had this tool been employed on multiple occasions to weave different kinds of fabrics with warp yarns of varying thickness spaced differently each time, it is remarkable that the spots should remain so clearly defined. Are we then to assume that the beater was used just once? Or was it used on multiple occasions, but to weave cloth of the same quality each time?(20) There also remains the question of the curvature to the toothed edge, as seen in three of the four weft-beaters.(21) This feature appears intentional but it is not clear what advantage, if any, it might have afforded.(22)

The knob’s primary purpose seems to have been to support the angle between the weaver’s thumb and index finger while they held its body firmly along both flat faces (see fig. 2).(23) Such a grip would facilitate the requisite wrist movement in beating the weft, particularly as expected on a horizontal loom.(24) This observation can only have limited value, as I have observed that experienced weavers are often accustomed to using tools, whose design would not immediately seemideal in terms of ergonomics. Therefore, any inference as to the configuration of the loom based solely on an ergonomic assessment of the weft-beater could prove misleading. Further, as the specialist on prehistoric textiles and weaving Elizabeth Wayland Barber has remarked, ‘since looms are merely clever aggregates of plain sticks, they are seldom identifiable in the archaeological record’.(25) Yet, as her own work has demonstrated, ‘reading’ surviving textile fragments can provide insights into looms and how weavers might have worked.

‘Pile’ usually consists of yarn tufts, loops or bunches of fibres introduced into the structure of a fabric in such a way as to protrude prominently outward from the flat surface of the textile, creating a raised surface as in a knotted carpet. The presence of pile on both surfaces of some textile fragments from Krorainic sites suggests that such fabrics were woven on upright looms which would have allowed comfortable access to what would otherwise be the ‘underside’ of the warp. Such a setup may have been worked either by solitary weavers or by pairs of them attending to the front and back simultaneously to create pile on both surfaces of the textile. It is possible that the ‘tufts’ on the reverse of knotted pile pieces were inserted individually as pre-cut lengths of yarn, in which case it is not inconceivable that a horizontal loom may have been employed. However, the arrangement of these tufts on some pieces recalls a pattern consistent with that produced by the cut-loop pile technique, suggesting that this rather more efficient method was known. The cut-loop pile technique appears to have been employed in producing pile not only on the face of certain pieces but also on the reverse of some knotted pile fragments. The latter fact again implies the use of an upright loom. Yet, textiles with pile only on one surface, whether composed of knots or cut loops, could well have been woven on a horizontal loom arrangement. Thus, it is possible that more than one type of rug loom was in use in ancient Niya. Fortunately, some specimens of carpets have survived largely intact and these can offer a sense of the width of the looms used to produce them. This aspect is considered in greater detail further on in the essay.

Evidence from surviving Krorainic textiles

The weaving of carpets or tapestry rarely calls for loom technology capable of producing anything more complex than tabby or twill constructions, making it difficult to state with absolute certainty what sort of loom the V&A beater was used with. However, surviving textiles can help ascertain the range of products it was possible to weave using a weft-beater like the V&A’s example.

Textiles recovered from Krorainic sites represent an exciting array of patterns, materials, structures, textures and qualities. The well-known resist-dyed cotton textile found at Niya probably came from west Asia or India.(26) Fine, figured silks found at Niya and Loulan, particularly a spectacular group bearing auspicious woven inscriptions, have unanimously been attributed to China. This is hardly surprising considering the strategic location of Kroraina. While it is unclear whether the entire spectrum of colours in some woollen taqueté, tapestry and carpet fragments would have been locally dyed,(27) we can be fairly certain that the ubiquitous felts and woven wools of coarse to medium quality were products of local manufacture and constituted the functional class of textiles used and produced by the general populace, if not the more affluent sections of society as well.(28)

The Chinese Buddhist pilgrim-monk Faxian, travelling through Kroraina (or ‘Shan-shan’ as he knew it) in around 400 CE, observed that ‘the clothes of the common people are coarse […] some wearing felt and others coarse serge or cloth of hair’.(29) Written records from Niya also reveal that the population kept sheep, goats, horses and camels, with the discovery of spinning tools and hanks of yarn at Niya strengthening the case for the local production of textiles from wool and other fibres.

The present discussion is limited to woven textiles, but even within this category apparently ‘indigenous’ fabrics are made from a variety of plant and animal material, and display considerable diversity. Examples survive of multiple variants of the tabby (such as balanced, weft-faced, ribbed and basket weaves), taqueté, twills, tapestry techniques, diverse pile structures and even twining. Significantly, some of these could certainly have been woven using beaters such as those discussed above.

Textile or carpet

Figure 9. Textile or carpet, unknown maker, excavated from Loulan, China, 200–400 CE, woven and knotted wool. Museum no. MAS.693. © The Trustees of the British Museum, London

Textile or carpet

Figure 8. Textile or carpet, unknown maker, excavated from Niya, China, 200-400 CE, knotted wool. Museum no. MAS.540.a-c. © The Trustees of the British Museum, London

Returning to the marks on the V&A’s weft-beater, although it is difficult to arrive at an accurate figure, the spots appear to be arranged about five or six to the centimetre, with each measuring less than a millimetre across on average. This gives a rough sense of warp yarn count (fineness) and distribution (expressed henceforth in terms of ‘warp ends per cm’). These findings can be compared with the specifications of surviving textiles using information gathered from the British Museum’s catalogue entries, Vivi Sylwan’s notes on textiles unearthed by the Sino-Swedish expedition, Wu Min’s study of specimens from Krorainic sites and thorough examination of pieces in the Stein Loan Collection. Since this tool would have been better suited to carpet weaving, it makes sense to focus on heavier fabrics, like pile textiles, and eliminate lighter and finer specimens, which would have been unable to withstand the beater’s impact.(30)

Carpet fragment

Figure 10. Carpet fragment, unknown maker, excavated from Loulan, China, 200–400 CE, knotted wool on tapestry weave. Museum no. LOAN.STEIN.534. © Victoria and Albert Museum, London

An example of pile-woven textile from Niya (MAS.540.a-c, fig. 8) has about 8 warp ends per cm(31), whereas one from Loulan (MAS.693, fig. 9), found by Stein in a refuse heap, is relatively coarse with 3 warp ends per cm.(32) Even allowing for the distortions in textile structure that these specimens must have developed over the centuries, we already have a range within which the beater’s marks may be comfortably situated. Three pieces from the Loulan area come even closer. One, now at the V&A, is a coarse rug fragment showing an interesting combination of pile with tapestry ground that resembles taqueté on the flat side (LOAN.STEIN.534, fig. 10).(33) This specimen has coarse warps arranged at roughly 4.5 warp ends per cm. Next is a group of fragments of a pile carpet, numbered 34:76-79, recovered from a gravesite; Sylwan tells us that these are composed of warps 2-2.5 mm thick arranged between 5.4 to 5.8  warp ends to the cm.(34) This group apparently displays a distinctive ‘latch-hook’ pattern similar to another colourful carpet fragment (LOAN.STEIN.647, fig. 11), also recovered from a gravesite and now at the V&A, which has about 6.5 warp ends per cm.(35)

Of the seven pile textiles surveyed by Wu Min, four in particular show features consistent with the pattern we are seeking. 89BYYM1:12 from Yingpan is a nearly complete carpet with the figure of a lion occupying the field. Each warp yarn measures 2 mm in diameter and these are arranged at 5 warp ends per cm. A rhombus-patterned fragment from Niya, 59MNM1:52(a) has warps measuring 1.5 to 2 mm in diameter and arranged at 4 warp ends per cm. Another rhombus-patterned fragment from Niya, 59MNM:52(b) comes closer to the mark with warp yarns measuring 1.8 to 2 mm in diameter and distributed at 5 warp ends per cm. A pile fragment recovered from grave B2 at Gutai (in the Loulan area) and numbered 80LBMB2:92 has warps 1.3 to 1.5 mm thick and arranged at between 6 and 7 warp ends per cm.(36) Finally, an intact carpet, M3:28, recovered from a burial at Niya (where it had originally been used to wrap the corpse) is composed of warps arranged at 5 warp ends per cm.(37)

. Carpet fragment

Figure 11. Carpet fragment, unknown maker, excavated from Loulan, China, 150–60 CE, knotted wool on wool warp and weft. Museum no. LOAN.STEIN.647. © Victoria and Albert Museum, London

This limited survey serves to reinforce the theory that carpets and other pile textiles were produced in Niya (and possibly Loulan) using weft-beaters like the V&A’s.(38) Some of the textiles considered here were recovered from cemeteries, indicating that they may have been woven well before the date ascribed to the V&A’s beater. For instance, the specimen LOAN.STEIN.647 (fig. 11) has been radiocarbon dated to between 150 BCE and 60 CE.(39) Does this mean that the quality of carpet weaving in the region changed very little over the space of a few centuries? The evidence here is too scanty to permit such a conclusion but, considering the signs of prolonged use generally present in textiles found in Loulan cemeteries, it is possible that these older dated specimens survived in use long enough to serve as models for succeeding generations before they were finally interred with their deceased owners.(40) As we shall see below, a possible explanation for the persistence of certain standards of textile production over long periods of time could be ascribed to the use of textiles as standardised media of exchange.

References to textiles in 3rd- to 4th-century Kharosthi documents from ancient Niya (Cadhota) (41) 

Written documents inscribed in the Kharosthi script have been unearthed in significant numbers from Niya and other sites.(42) The language used in these documents is a particular dialect of Prakrit associated with the Gandharan region (northwestern Pakistan and eastern Afghanistan), but certain peculiarities in usage reveal glimpses of the local language of ancient Kroraina, which was clearly distinct from ‘Gandhari’ Prakrit. Based on a chronology of local kings appearing in these records, as well as evidence from Chinese ones, E. J. Rapson, John Brough, Enoki Kazuo, Lin Meicun and others have been able to arrive at a fairly precise date range for these documents: from the mid-3rd to the mid-4th century CE.(43) Translations reveal that, in its time, Stein’s Niya was known by the name ‘Cadhota’.(44)

This body of royal orders, taxation records, sale deeds and even private correspondence has made it possible for scholars including R. C. Agrawala, Christopher Atwood, Helen Wang and Mariner Padwa, among others, to reconstruct details of Cadhotan economy, administration, social order, industry and even the spatial distribution of kinship groupings.(45) This research has established the multicultural nature of Cadhotan society, revealing that it was composed of the indigenous (possibly ‘Tocharian’-speaking) population, Gandharan settlers, Khotanese ‘refugees’, a certain ‘people of the mountain’, and some native Chinese.

Frequent references to textile products in the Kharosthi documents provide us information that can be compared with material remains to give a clearer sense of the social, cultural and economic status of textiles in Cadhota. The terminology used to refer to textile items in these Kharosthi texts has already received scholarly attention (most notably in the work of Heinrich Lüders) and some of it has even been reliably translated into English.(46) In his translations, Thomas Burrow generally renders the words ‘tavastaga’ (or ‘thavastae’) and ‘kojava’ as ‘carpet’ and ‘rug’ respectively and these terms shall be of particular interest to us.(47) Such interpretations, based primarily on philological evidence, are also borne out to some extent by the contexts in which these terms appear in the Kharosthi texts.(48) For instance, the length of a tavastaga is communicated in terms of ‘hasta’ (hands), a unit elsewhere used to measure woollen cloth (§318), whereas a different system of units is used to quantify grain, wine and ghee.

An analysis of these documents gives every reason to believe that carpets and rugs not only played an important part in Cadhotan economy, but that they were also produced locally and on a regular, organised basis. This provides us with a context within which to situate the V&A weft-beater and the extant textiles. One way to test this is to examine surviving textiles for attributes that might link them to the literary references cited above.

The Stein Loan Collection includes a unique specimen, LOAN.STEIN.534 (fig. 10), which lends itself to this approach. Its weaver took the trouble to create a tapestry ground on a (minimum) three-shaft construction when a simple tabby would have sufficed as foundation for the pile, suggesting that it was the tapestry, rather than the undyed wool pile, which was meant to be visible.(49) Was this textile, then, not a floor covering per se but some kind of bedding or blanket whose shaggy pile was meant to lie against the body and keep it warm?(50) This interpretation agrees with the Pali language translation of the word kojava as ‘a rug or cover with long hair, a fleecy counterpane’. Burrow’s tentative interpretation of the term ‘alena kojava’ (§549) as ‘a rug for lying down in’, if correct, further supports the Pali translation.(51) The use of carpets and rugs as blankets or bedding is also alluded to by their discovery placed upon coffins and wrapped around corpses.

Archaeological and documentary evidence supporting the local production of textiles in Cadhota

Kharosthi records reveal that textiles, especially rugs, formed a substantial portion of revenue in Cadhota: a predominantly rural, subsistence economy largely dependent on agriculture, horticulture, animal husbandry and domestic industry, alongside more specialised industries. Farms and orchards produced wheat, millet, barley, maize, peaches, pomegranates and a variety of other crop besides wine and dairy products like ghee. §207* provides a list of the items usually collected as tax:

The arrears of tax in Ajiyama-av́ana. […] is to be sent to the feet of his majesty. […] Pomegranates […] 1 sap.ǵa, 6 milima of oǵana, 6 cloths, 1 cow, ghee […] The sum of arrears is: The new and the old ghee 18 khi, 1 asamkharaj́iya (?), 6 rugs (kośava), 1 akiṣḍha, 5 pieces of cloth, 16 kamuṃta, 3 sacks, 3 baskets, 1 sheep, 1 milima 5 khi of wine potǵoñena, 14 milima of corn, 1 cow.

Textile products such as tavastaga, kojava, ‘agisdha’, ‘arnavaji’, ‘namatae’, ‘raji’ and ‘thavamnae’ (in other words carpets, rugs, cloth and felt, among other items) are clearly identifiable in documents relating to taxation. Notably, while artisans such as potters, goldsmiths, bow- and arrow-makers appear in the documents, weavers are conspicuously absent.(52) Historian Christopher Atwood has suggested that this could point to textiles being produced ‘in the home by unspecialized labour’.(53) Significantly, no products made by the other artisan groups cited above appear in taxation records. Instead, most identifiable items owed as tax were either products of the land or (very likely) of domestic industry – from which it could also be inferred that tax textiles were produced locally through domestic industry.

Double burials discovered at Niya allow us to speculate on the social relevance of textiles and textile making by testifying to their role in the formation of gendered identities. Deceased couples were furnished with a variety of objects that clearly specified their gender roles and these give us some sense of the different skills that men and women would have been expected to cultivate. How far Cadhotan life conformed to these idealised conventions in actual reality, it is difficult to tell, but the appearance of textile tools, such as needles and spindles, in connection with female ‘mummies’ gives an indication that such work normally fell within the feminine domain.(54) The discovery of spindles, spools of coloured yarn and even thimbles in such high status burials further points to the centrality of textile production and its related symbolism to both life and death in ancient Cadhota.(55) 

Making textiles was an important and culturally significant activity that may have been practiced on a regular basis in homes and, as some evidence suggests, possibly even in monastic establishments.(56) If the extant ruins do represent farmsteads of extended family groups, as argued by Mariner Padwa(57), then the discovery of beaters within such residential enclaves supports the view that textile production was part of Cadhotan domestic life.(58) Certainly, the labour required for conducting the various stages of textile production would have been available in an extended family which, as we now know from the documents, would have included slaves and dependents. In fact, Padwa’s work even allows us to identify the structures N.XIII and N.XXII, which yielded the V&A beater as well as the beater MAS.554, as part of the settlement referred to as ‘Yave-avana’ in the Kharosthi texts.(59) That beaters were found at more than one location in Yave-avana indicates that weaving, as an activity, was dispersed even within the territorial extent of a single avana (roughly, a settlement comprised of several households that might be connected by ties of kinship).(60)

Vineyards for carpets: were textiles used as money in Cadhota?

In addition to its social and cultural significance, textile production, including weaving, was an important economic activity in ancient Cadhota. Records of transactions reveal that payments were made using various goods, including kojava and tavastaga, in addition to, or instead of, coins. The monetary worth of items being exchanged was reckoned in terms of ‘muli’, which appears to have been a form of local currency that functioned more often as a unit of account than as an actual coin.(61) Transactions were governed by a generally accepted convention of conversion based on well-defined units of measure and units of account. For instance, ‘milima’ was a measure equivalent to 20 ‘khi’ and one milima of corn was worth one muli.(62)

Significantly, kojava and tavastaga frequently appear as payment in sale deeds. §549 records the sale of a parcel of arable land for the price of one ‘Khotanese alena rug and 5 milima of corn’ together reckoned as being worth 15 muli. Bearing in mind that one milima of corn was worth one muli, it is possible to deduce that the rug was worth ten muli. Compare this with §222 which reveals that a rug worth ten muli was exchanged for land capable of yielding five khi of ‘adini’ seed. For the same price, one could even purchase a cow (§327) or ten khi of wine (§571). Allowing for the fluctuations in prices that must no doubt have occurred over the roughly 100 years in time that these documents represent, we still have here a relative sense of the worth of such textiles. That carpets and rugs were offered as sole payment for parcels of land (§222, §579) indicates that these textiles could have high purchasing power. In fact, §431-2 contains a reference to a carpet 13 hands long whose monetary worth was equivalent to a quantity of gold in the form of ‘one golden stater’.(see note 61) Further,  §527 records an official inquiry and judicial ruling in the case of a dispute over ‘12 hands (length) of carpet and 6 milima of corn’. This instance clearly attests to the high value of such carpets.

What, then, was done with textiles collected as tax? Tax items, such as grain, wine and ghee, were collected and placed in stores. From here, they were disbursed as the need arose: in the form of salaries for officials, as ‘honorific gifts and provisions’ carried by diplomatic missions or envoys travelling to other kingdoms (§22, see also §214) or in the form of ‘clothing, food, and wages’ offered as payment for services rendered to the state (§19). Textiles from Cadhota would even have travelled as far as Loulan since we know that the revenue (or part of it) was regularly conveyed to the capital.(63) The maintenance of royal herds was also supported by these provincial reserves (§55) and tax grain and wine were even used to purchase, among other items, carpets and rugs (§431-2, §448, §622). Other channels through which tax goods returned into circulation were not, strictly speaking, legal. Tax grain, ghee and wine were routinely loaned out by officials for personal gain and were sometimes brazenly appropriated for private use. Tax textiles might conceivably have circulated through much the same means as these other tax items.

As has been established elsewhere, textiles were portable and durable forms of wealth, as well as intrinsically useful objects, making them particularly well suited to serve as money along the Silk Road. (64) Much of the silk that lent its name to this network of travel and trade routes was, in fact, textile money in the form of bolts of standard dimensions and qualities that the Chinese administration pumped into the ‘western regions’ for the maintenance of its outposts and interests in that area. Carpets and rugs are fundamentally different from such textile yardage. Standard bolts of cloth could be divided into sub-multiples or even turned into a suit of clothes, but this is not normally the case with a carpet.(65) Both as material objects and as stores of value, carpets and rugs offer greater stability than fabric yardage, since putting them to use would not normally preclude the possibility of trading them off at a later date. This might explain why in other contemporary Silk Road communities, such as Turfan, rugs regularly served as a medium of exchange.(66) According to Helen Wang, Curator of East Asian Money at the British Museum, ‘the use of carpets as media of exchange in the Khotan and Turfan regions indicates that these were well established forms of money in Eastern Central Asia’.(67)

The use of textiles as an efficient medium of exchange presupposes a fair degree of standardisation and there is evidence to support this in the context of Cadhota.(68) Conspicuously, bolts of silk appear only infrequently in the Kharosthi records from Niya and, then too, seldom feature in economic transactions in the way that carpets or animals do.(69) The appearance of bolts of silk in the penalty clauses of legal documents or as fines for various transgressions within the monastic community (where this would have been expected to act as a strong deterrent) indicates that silk in Cadhota was a relatively rare and precious commodity, circulating within a limited social and economic sphere. If Cadhotans did in fact use textiles as money then it seems that locally produced textiles rather than Chinese silk would have been preferred.(70)

Obaidullaii Khan, with his Sons and Villagers, Mirigram

Figure 12. Obaidullaii Khan, with his Sons and Villagers, Mirigram. Marc Aurel Stein, Ruins of Desert Cathay, vol.1 (London: Macmillan, 1912), plate 19. © Toyo Bunko Copy Services

One source of evidence for standardisation among the products of the indigenous textile industry is in the form of a ubiquitous feature of ruined Niya dwellings: what Stein has referred to as the ‘raised sitting platform’. Stein and his local workmen noted that, in several aspects, ancient Niya dwellings closely resembled traditional Uyghur dwellings where low platforms covered with carpets and other furnishings served both for sitting and sleeping on (see fig. 12). The persistence in this region of ancient forms and practices well into the modern era allows us to speculate that the sitting platform occupied as much of an important position in Cadhotan domestic life as it did in the local homes Stein visited in the region and that textiles, particularly carpets, would have been used in much the same way in ancient times to cover and decorate the platforms and floors.(71) In his 1964 monograph on Khotan carpets, Hans Bidder has suggested that ‘it is to this platform that one must look for the original carpet sizes’, since ‘the shape and measurement of the Khotan carpets […] are regulated by the size of the “Aivan” platform’.(72) A survey of the dimensions of the sitting platforms found by Stein at Niya shows that these measured between 3.75 and 4.5 feet in breadth. This range comfortably accommodates the dimensions of a well-preserved carpet specimen from Niya, numbered M3:28, which measures approximately 4.2 feet in width.(73)

Kharosthi texts provide additional evidence for the standardisation of textiles in Cadhota. §207 refers to ‘6 cloths […] 6 rugs (kośava), 1 akiṣḍha, 5 pieces of cloth’ expected in fulfilment of tax arrears and §173 states that ‘in Suḡ́iya's hundred one rug (kojava) is to be given’.(74) That units of such textiles could be anticipated in advance indicates that specific qualities and sizes were assumed.(75)

We might also approach this subject from another angle. It was usually in instances of transactions or disputes that there was a need to record in writing the specific dimensions or value of these textiles. Even in these documents, however, kojava are never accompanied by measurements, whereas tavastaga nearly always are.(76) Some kojava were priced at five muli and others at ten (§327, §222, §549) but it is unclear what determined the difference.(77) As for carpets (tavastaga), their length seems to have served as sufficient indication of their monetary worth.(78) For example, the sale deed §590 records the lengths but not the prices in muli of the carpets offered in part payment for the purchase of a slave woman. It does, however, mention the prices in muli of the camels also being offered in payment. A comparison of this document with another sale deed, §579, allows us to deduce that one hasta (along the length) of a carpet was worth roughly nine-tenths of a muli in both instances.(79) This close agreement in price-per-hasta of carpet between two documents separated by a temporal span of eight years is remarkable.(80) It would appear, then, that the price of one hasta of carpet was fairly stable during this period at least and that the worth of a tavastaga was reckoned by multiplying this fixed value by the total length in hasta units. This would imply that the width, if not the actual knotting or weave quality of such textiles, usually conformed to a fixed standard.

This theory receives further support from the remarkable group of complete (or nearly complete) pile carpets and woollen blankets more recently recovered from Niya, Loulan, Yingpan and Shanpula. At Yingpan, woollen blankets with fringed or tasseled edges were mostly found. A description by Li Wenying (Deputy Director of the Xinjiang Institute of Archaeology) of woollen blankets with fringed or tasselled edges found at Yingpan reveals that these were generally ‘naturally colored, coarse […] in plain or twill weave, 80-85 cm wide and about 150 cm long’.(81) Clearly, such blankets came in a fairly standard quality and size.(82) Another noteworthy group includes three carpets featuring lions and measuring roughly 100 cm by 260 cm on average.(83) Two in particular, one from Yingpan and the other from a gravesite near the Loulan site L.E. are so closely matched in terms of size, pile technique and even the treatment of the lions that Li Wenying believes these were ‘undoubtedly woven following the same design’.(84) She further observes that 'the fact that these carpets were cut along the selvages indicates that carpets with the same designs and colors were probably produced in bolts at the same time and place'.(85)

It is very likely, then, that indigenous textile production in Cadhota, and possibly in the general region, conformed to a fair degree of standardisation.(86) Acknowledging some standardised production does not, however, preclude the possibility of textiles being made to meet special requirements or according to individual tastes. A spectacular carpet featuring a majestic, sinuously rendered lion, as discovered in the 3rd- to 4th-century grave of the exquisitely attired and evidently wealthy ‘Yingpan man’, is an unusual piece that may have been specially commissioned.

Further, the diversity represented by surviving specimens of such textiles, particularly the sheer variety of pile structures employed, serves to complicate and challenge any simplistic notion of ‘standardisation’. It is essential, therefore, to also compare patterns, structural features and the contexts of these finds in order to arrive at a more complete understanding. Yet, if the overall theory presented here is generally correct, it offers a compelling explanation for the remarkable clarity of the spots on the V&A’s beater, as a tool used consistently to weave a specific kind of textile.


Upon discovering a pair of richly hued carpet fragments at Loulan, Stein remarked that ‘it was the first ancient specimen of an industry attested in the Khotan region from very early times […] which I had so far succeeded in bringing to light’, before cautiously admitting that, ‘whether these carpets actually came from Khotan it is, of course, impossible to assert’.(87) This article has considered wooden weft-beaters from the collections of the V&A and British Museum, surviving textiles and other archaeological finds, in conjunction with contemporary manuscripts, to confirm Stein’s speculations about the local origins of these textiles.

Unique marks on the V&A’s weft-beater have helped to connect the kojava, tavastaga and other textiles in these documents with surviving objects preserved by extraordinary natural conditions, highlighting the probability that these were locally produced. In order to further nuance these findings, a comparative analysis of early pile specimens recovered from sites such as Palmyra, Dura-Europos, Berenike and the At-Tar caves with those from Xinjiang would prove illuminating, particularly with regard to shared characteristics, such as pile occurring on both sides of the textiles.

While Kharosthi texts have allowed us to appreciate the socio-economic worth of textile production, use and exchange in Cadhota, the appearance of carpets in burials attests to their cultural and symbolic value, linking them to a more widely occurring practice in evidence at other early sites, as well as to funereal traditions that have survived into the modern era.(88) Thus, we might view the weft-beaters not merely as utilitarian implements, or as representatives of a broad and diverse ancient material culture, but also as tools employed in generating economic, social and cultural capital.


This research was first undertaken in 2012 during my MA in History of Design at the Royal College of Art and the Victoria and Albert Museum. It would not have been possible without the generous support of curator Helen Persson, who not only enabled me to study objects in the V&A’s Stein Loan Collection, but also shared her unpublished work. I am also immensely grateful to curator Clarissa von Spee for granting me the opportunity to study Stein materials at the British Museum and to Iestyn Jones for facilitating that process. Furniture conservator Andrew Thackray at the V&A and Jennifer Wearden (former V&A curator) have both been very generous with their time and expert inputs. This paper has also benefited from my discussions with Dr Zhang He (William Paterson University), who has been working on a related subject and who has very kindly pointed me to certain very helpful publications. For their constructive comments and suggestions, I am particularly thankful to curator Dr Helen Wang at the British Museum, Dr Christine Guth (Senior Tutor, V&A/RCA History of Design), Dr Suchitra Balasubrahmanyam (Ambedkar University Delhi) and my two anonymous reviewers. I wish especially to register my sincere gratitude to Verity Wilson (former V&A curator) for her kind encouragement and for offering me the benefit of her expertise on the V&A’s Stein Loan Collection. My thanks also to Chetan Shastri for his assistance in locating research material. Finally, a special note of thanks to the entire team at the V&A Online Journal and especially my editor Elaine Tierney for giving me the opportunity to develop this research into a paper for the V&A Online Journal and for all of her efforts towards preparing it for publication.


1. A fourth Central Asian expedition (1930-1) ran into serious opposition from the Chinese authorities, forcing Stein to abandon it partway through.

2. Aurel Stein, Serindia: Detailed Report of Explorations in Central Asia and Westernmost China, 5 vols (Oxford: Clarendon Press, 1921), vol. 3, 1317.

3.  For a popular account, see Peter Hopkirk, Foreign Devils on the Silk Road: The Search for the Lost Treasures of Central Asia, 2006 edn (London: John Murray, 1980).

4. Verity Wilson, ‘Early Textiles from Central Asia: Approaches to Study with Reference to the Stein Loan Collection in the Victoria and Albert Museum, London’, Textile History 26. 1 (1995): 45.

5. Wilson, ‘Early Textiles from Central Asia’, 45. See also Anne Godden Amos, ‘The Stein Loan of Textiles in the Victoria and Albert Museum’, V&A Conservation Journal 26 (January 1998), http://www.vam.ac.uk/content/journals/conservation-journal/issue-26/the-stein-loan-of-textiles-in-the-victoria-and-albert-museum/, para 6 of 14. For more information on how this material came to be at the V&A, see Helen Wang, Helen Persson and Frances Wood, ‘Dunhuang Textiles in London: A History of the Collection’, in Sir Aurel Stein, Colleagues and Collections, ed. by Helen Wang, British Museum Research Publication 184 (2012), http://www.britishmuseum.org/pdf/19_Wang-Persson-Wood.pdf. Regarding conservation history and current storage conditions see the following publications: Joan Joshua, ‘The Restoring of Ancient Textiles’, Embroidery (September 1933): 15-18; Amos, ‘The Stein Loan of Textiles in the Victoria and Albert Museum’; Verity Wilson, ‘A New Resource for the Study of Central Asian Textiles:  Stein Textiles in the Victoria and Albert Museum’, Oxford Asian Textile Group Newsletter 30 (February 2005), http://www.oatg.org.uk/Newsletters/N30.pdf;
Helen Persson, ‘Stein Mellon Textile Project at the V&A’, V&A Conservation Journal 55 (Spring 2007), http://www.vam.ac.uk/content/journals/conservation-journal/issue-55/stein-mellon-textile-project-at-the-v-and-a/; Thórdis Baldursdóttir, ‘Storage of the Stein Loan Collection’, V&A Conservation Journal 55 (Spring 2007), http://www.vam.ac.uk/content/journals/conservation-journal/issue-55/storage-of-the-stein-loan-collection/. For a more detailed overview of the range of materials and sites represented, see ‘The Stein Collection’, Victoria and Albert Museum, http://www.vam.ac.uk/page/s/stein-collection/. The V&A’s Stein material has been digitised and is accessible on the Museum’s website using the ‘Search the Collections’ feature.

6. Helen Persson, 'Ethnicity Mobility and Status – Textiles from the Taklamakan Desert', in Textiles as Cultural Expressions: Proceedings of the 11th Biennial Symposium of the Textile Society of America, September 24-27, 2008, Honolulu, Hawaii (Honolulu: The University of Hawaii, 2009), http://digitalcommons.unl.edu/tsaconf/235, 2.

7. Wang, Persson and Wood, ‘Dunhuang Textiles in London’, 5-6.

8. For a fuller bibliography of research drawing on the V&A’s Stein textile holdings, see Helen Wang and John Perkins, eds, Handbook to the Collections of Sir Aurel Stein in the UK (London: The British Museum, 2008), 2nd edn, ed. by Helen Wang, www.britishmuseum.org/pdf/Stein%20Handbook%20final(3).pdf, 26. For an overview of research conducted on the Dunhuang textiles at the V&A, see Wang, Persson and Wood, ‘Dunhuang Textiles in London’, 5-6.

9. See Elizabeth Wayland Barber, The Mummies of Ürümchi (New York: W. W. Norton & Company, 2000).

10. Wang, Persson and Wood, ‘Dunhuang Textiles in London’, 6.

11. Persson, 'Ethnicity Mobility and Status’, 4. Interestingly, the earliest archaeological indication as yet discovered for the production of pile textiles comes not from actual textile specimens but in the form of a tool: a distinctive type of bronze knife that is thought to have been used for cutting pile yarn. E. J. W. Barber, Prehistoric Textiles: The Development of Cloth in the Neolithic and Bronze Ages with Special Reference to the Aegean (Princeton: Princeton University Press, 1991; repr. 1992), 171, note 10.  

12. This phrase was used by Stein to describe wooden objects of the kind that we shall consider in this essay. Whether in fact this object was an implement of ‘home industry’ is a question we shall consider in subsequent sections of this essay. Marc Aurel Stein, Ancient Khotan: Detailed Report of Archaeological Explorations in Chinese Turkestan, 2 vols (Oxford: Clarendon Press, 1907), vol. 1, 377.

13. ‘The Stein Collection’, Victoria and Albert Museum, http://collections.vam.ac.uk/item/O97766/the-stein-collection-weft-beater-unknown/.

14. Cultivated white poplar and the wild black poplar were both easily available at Niya, as testified by extant material remains. The former was preferred for construction and writing stationery at Niya. Other varieties such as mulberry also appear to have been cultivated at the site and it is also possible that one of those was the source of wood for the beater.

15. For 1907,1111.97, see ‘Comb Beater’, The British Museum, http://www.britishmuseum.org/research/collection_online/collection_object_details.aspx?objectId=229669&partId=1&searchText=1907,1111.97&page=1. For MAS.554, see ‘Comb Beater’, The British Museum, http://www.britishmuseum.org/research/collection_online/collection_object_details.aspx?objectId=228854&partId=1&searchText=MAS.554+&page=1. For MAS.727, see ‘Comb Beater’, The British Museum, http://www.britishmuseum.org/research/collection_online/collection_object_details.aspx?objectId=228690&partId=1&searchText=MAS.727+&page=1.

16. It would be appropriate at this juncture to mention other relevant specimens of this type of object. Three similar objects, now at the British Museum, numbered in the series 1925,0619.73-75, were acquired by Sir Clarmont P. Skrine in the same general region in the 1920s. Since these are currently undated, it is difficult to say whether they are contemporary to the Niya-Loulan pieces. The first two of the Skrine series share certain features with another British Museum piece, MAS.472, retrieved by Stein from Mazar-Toghrak and dated approximately to the 8th century CE. Of the three Skrine specimens, two seem to have been acquired in Keriya. See Daniel C. Waugh and Ursula Sims-Williams, ‘The Old Curiosity Shop in Khotan’, The Silk Road 8 (2010): 76. This might be significant in view of the fact that the third piece in the Skrine series shares certain formal attributes, particularly the triangular shape of the ‘handle’, with another such object recovered from the Keriya region, now held at the Xinjiang Uyghur Autonomous Region Museum in Urumqi, and dated to between the 4th and 6th centuries CE. See Yue Feng, Xīnjiāng lìshĭ wénmíng jícuì [Best Collections of Xinjiang History and Civilization] (Xīnjiāng měishù shèyĭng chūbǎnshè, 2009), 150. I am grateful to Dr Zhang He for alerting me to this publication. This last mentioned specimen was also featured in the exhibition: Victor Mair (curator); Secrets of the Silk Road. Philadelphia: Penn Museum, 2011. A comparison of the information obtained from the exhibition catalogue and gallery label reveals some discrepancies regarding the place of discovery of this object. Dr Zhang He has also alerted me to the presence of yet another such wooden comb-like object at the Khotan Museum. According to the museum label, this piece was found at Niya and has been dated to the Han period. The shape of this specimen is akin to that of another such object also from Niya, severely cracked and desiccated, featured in Feng Zhao and Zhiyong Yu, eds, Legacy of the Desert King: Textiles and Treasures Excavated at Niya on the Silk Road (Hangzhou and Urumqi: China National Silk Museum and Xinjiang Institute of Archaeology, 2000), 45. The Sino-Swedish expedition also uncovered one such object in the Lop-nor region. See Folke Bergman, Archaeological researches in Sinkiang, especially the Lop-Nor region (Stockholm: Bokförlags aktiebolaget Thule, 1939), Pl 27:12. Finally, there is a broken fragment of a similar wooden comb-like object, L.M.II.iii.03, recovered by Stein from the site L.M. See Aurel Stein, Innermost Asia: Detailed Report of Explorations in Central Asia, Kan-su and Eastern Īrān, 4 vols (Oxford: Clarendon Press, 1928), vol. 1, 203, also vol. 4, pl. XXVI.

17. The first such object discovered by Stein at Niya is listed as a ‘currycomb’ in the index of Ancient Khotan and its apparent resemblance to one is noted on multiple occasions. Yet, both in the descriptive catalogue, as well as the main text, this observation (intended as a visual reference) prefaces a clear description of the object’s function as weft-beater. See Stein, Ancient Khotan, vol. 1, 377, 413. On another occasion, Stein likens the object to a sizing brush based on its resemblance to one depicted in a painted panel from Dandan-Uilik. See Stein, Ancient Khotan, 260. The author of the descriptive catalogues accompanying Stein’s expedition reports, F. H. Andrews, though acknowledging the similarity in form was clearly aware of the differences in function between a weft-beater and a sizing brush. (Compare pages 300 and 413 in Stein, Ancient Khotan). Such inconsistencies are absent in Stein, Serindia and Stein, Innermost Asia (Stein’s reports of his second and third Central Asian expeditions respectively), where objects of this sort are identified throughout and unambiguously as weaving instruments. Further, Andrews explicitly draws a connection between a comparable object of significantly later date discovered at the site of Mazar-Toghrak and the weft-beaters then still in use in ‘Turkestān’. Stein, Serindia, vol. 1, 210.

18. Peter Collingwood, Rug Weaving Techniques: Beyond the Basics (Loveland: Interweave Press, 1990), 10.

19. Later examples of similar objects demonstrate a clear evolution of this form into one more easily identifiable with weft-beaters still in use today. MAS.472 at the British Museum is dated approximately to the 8th century and was retrieved from Mazar-Tograkh, another site along the southern rim of the Tarim basin. With its fine and closely arranged teeth, this later model appears to fit more readily with the present conception of a weft-beater than the four pieces discussed above. The entry for this object in the descriptive catalogue of Serindia reads as follows: ‘wooden comb, used in weaving (now called panje in Turkestān)’. Stein, Serindia, vol. 1, 210.

20. As far as we can confidently tell, the same reed/warp-spacer or warp arrangement may have been used consistently. This does not imply uniformity in designs, colours and weaves. 

21. An image of another such beater discovered by the Sino-Swedish expedition in the Lop-nor area shows that it too has a curved edge. Bergman, Archaeological researches in Sinkiang, Pl 27:12. A gentle curvature along the toothed edge seems to be present also in the 4th- to 6th-century specimen now at the Xinjiang Uyghur Autonomous Region Museum in Urumqi. See note 16 above.

22. It is possible that this curvature is a skeuomorphic feature that might indicate the use of a different material for making weft-beaters in an earlier period. For instance, Arthur MacGregor writes that 'many cetacean bone beaters from Norway have curved blades, a feature which Hoffmann (1964) attributes to the natural shape of the cetacean bones from which they were fashioned. This curvature was of no particular benefit during use, so that the occurrence of wooden weaving swords displaying the same shape, also catalogued by Hoffmann, is a further instance of the morphology of the skeletal raw material conditioning the development of tools in other media’. See Arthur MacGregor, Bone, Antler, Ivory & Horn: The Technology of Skeletal Materials since the Roman Period (New York: Routledge, 2015), 188.  

23. F. H. Andrews, in describing the first such object found at Niya, observed that ‘the thick edge […] [is] rounded into a form to fit comfortably into the palm of the hand, and furnished with a projecting knob to give firmness to the stroke in using’. See Stein, Ancient Khotan, vol. 1, 413.

24. I am grateful to Vijay Paul Punia for this insight. Email correspondence with Swati Venkat, 5 December 2012. Many traditional patterns of weft-beater known to us today incorporate an angle between the handle and teeth to allow the weaver to exert effort in a direction that is natural and comfortable especially when working at a vertical loom. By this token, although perhaps anachronistically, the V&A’s beater seems better suited to the horizontal format. But this observation cannot, of course, be treated as a conclusive indication mainly because we are dealing with a period for which little else is known regarding weaving methods in this region.

25. Elizabeth Wayland Barber, ‘More Light on the Xinjiang Textiles’, in Reconfiguring the Silk Road: New Research on East-West Exchange in Antiquity, ed. by Victor H. Mair and Jane Hickman (Philadelphia: University of Pennsylvania Museum of Archaeology and Anthropology, 2014), 35. The discovery of objects (tentatively) identified as loom weights at some sites suggests that warp-weighted looms were used in the region. While it is possible to weave pile on vertical warp-weighted looms, we can be fairly certain that the V&A’s beater was not associated with this sort of loom. In a vertical warp-weighted setup, weaving usually grows downwards, employing an upward beating motion. In such a scenario, the very bulk of such beaters would have rendered all but the smallest entirely ineffective (although horizontal warp-weighted looms are not unknown). Interestingly, we do have a specimen of the kind of comb that might have been better suited to this task than the large, heavy ones already discussed. Described in Serindia as a ‘wooden weaver’s comb, long truncated triangle. Very short teeth’ and accompanied by the suggestion ‘perhaps for carding’, MAS.557 from Niya (and now at the British Museum), deviates widely from the other beaters here discussed in terms of size as well as form. It might have been intended for use in weaving finer fabrics such as the woollen taquetés and tapestries. It may even have been used in conjunction with a warp-weighted loom. See Stein, Serindia, vol. 1, 257. See also ‘Comb’, The British Museum, http://www.britishmuseum.org/research/collection_online/collection_object_details.aspx?objectId=228851&partId=1&searchText=MAS.557&page=1.

26. Valerie Hansen, The Silk Road: A New History (New York: Oxford University Press, 2012), 38. A number of blue resist-dyed cotton fabrics unearthed from the Late Roman levels of the Egyptian port site of Berenike have been convincingly identified as Indian in origin. It is possible, therefore, that the Niya specimen, too, originated in India. John Peter Wild and Felicity Wild, ‘Rome and India: early Indian cotton textiles from Berenike, Red Sea coast of Egypt’, in Textiles in Indian Ocean Societies, ed. by Ruth Barnes (New York: Routledge, 2012), 14-16.

27. Grave M15 at Yingpan, dated to between the 3rd and 4th centuries CE, contained the body of a male dressed in exquisite woollen fabrics, displaying such technical complexity and superlative execution that it seems unlikely that these were products of local manufacture. Similarly, some fine tapestries, such as those recovered from the Loulan cemeteries, seem to have been imported.

28. Graves at Niya and Loulan have yielded significant quantities of textiles, including clothing. While wool and cotton have been found, silk is particularly well represented. Certain features in the construction of clothing, either made of silk or trimmed with it, suggest that although the silk fabric came from China, the actual assembly of these garments was done locally and according to local styles. See Ma Yong and Sun Yutang, ‘The Western Regions under the Hsiung-nu and the Han’, in History of Civilizations of Central Asia: The development of sedentary and nomadic civilizations: 700 B.C. to A.D. 250, ed. by János Harmatta et al, 2 of 4 vols (Paris: UNESCO Publishing, 1994; repr. 1996), 224. For an overview of textiles recovered from the burials at Niya of what seem to be members of the local royalty or nobility, see Zhao and Yu, Legacy of the Desert King. For a survey of textiles unearthed from the Yingpan burials, see Li Wenying, ‘Textiles of the Second to Fifth Century Unearthed from Yingpan Cemetery’, in Central Asian textiles and their contexts in the early Middle Ages, ed. by Regula Schorta, Riggisberger Berichte 9 (Riggisberg: Abegg-Stiftung, 2006).

29. Fâ-hien, trans. by James Legge, A Record of Buddhistic Kingdoms (Oxford: Clarendon Press, 1886), 13.

30. It is possible that a variety of finer fabrics such as the patterned taquetés and tapestry flatweaves were woven using smaller and lighter weft-beaters such as MAS.727 with closer set teeth than the three others which are clearly too heavy and broad for that task. 

31. ‘Textile / Carpet’, The British Museum, http://www.britishmuseum.org/research/collection_online/collection_object_details.aspx?objectId=228868&partId=1&searchText=MAS.540.a-c&page=1.

32. ‘Textile / Carpet’, The British Museum, http://www.britishmuseum.org/research/collection_online/collection_object_details.aspx?objectId=228725&partId=1&searchText=MAS.693&page=1.

33. ‘The Stein Collection’, Victoria and Albert Museum, http://collections.vam.ac.uk/item/O93095/the-stein-collection-carpet-fragment-unknown/.

34. In Sylwan’s chosen terms, the warp arrangement is described as follows: ‘54-58 threads to 10cm., in the selvage about 20 threads to 5 cm’. Bergman, Archaeological researches in Sinkiang, 131.

35. ‘The Stein Collection’, Victoria and Albert Museum, http://collections.vam.ac.uk/item/O93006/the-stein-collection-carpet-fragment-unknown/.

36. Wu Min, ‘Study on Some Ancient Wool Fabrics Unearthed in Recent Years from Xinjiang of China’, Al-Rāfidān 17 (1996): 2, 15-16. 

37. Zhao and Yu, Legacy of the Desert King, 80.

38. A similar comparison with the other complete (or nearly complete) carpets recovered from roughly contemporary sites at Niya, Loulan, Yingpan and Shanpula would, no doubt, constitute the logical next step in this study. For a list of some of the complete carpets discovered thus far, see: Zhang He, ‘Figurative and Inscribed Carpets from Shanpula, Khotan: Unexpected Representations of the Hindu God Krishna. A Preliminary Study’, Journal of Inner Asian Art and Archaeology 5 (2010): 70.

39. Persson, 'Ethnicity Mobility and Status’, 9.

40. Stein, Innermost Asia, vol. 1, 227, 231.

41. The following discussions draw upon information from Kharosthi documents from Niya and other sites as transcribed in A. M. Boyer, E. J. Rapson, E. Senart and P. S. Noble, Kharoṣṭhī Inscriptions Discovered by Sir Aurel Stein in Chinese Turkestan, 3 parts (Oxford: Oxford University Press, 1920-9) and translated in T. Burrow, A Translation of the Kharoṣṭhī Documents from Chinese Turkestan (London: The Royal Asiatic Society, 1940). I have also referred to the latter’s 2nd electronic edition (London: International Dunhuang Project, 2013), http://idp.bl.uk/database/oo_cat.a4d?shortref=Burrow_1940&catno=001. Henceforth, the sign ‘§’ will be used to indicate numbers assigned to Kharosthi documents in the aforementioned publications. Where document numbers are followed by the ‘*’ sign, translations have been amended with reference to the transcriptions. 

42. Kharosthi inscriptions have been found not only on wooden tablets and leather documents, but also on stone, on some wall paintings and even on textiles at a number of sites. Of these, documents collected from Niya represent the largest and most significant corpus of such material.

43. See the following publications: E. J. Rapson and P. S. Noble, Kharoṣṭhī Inscriptions Discovered by Sir Aurel Stein in Chinese Turkestan Part III (Oxford: Oxford University Press, 1929), 323-8; John Brough, ‘Comments on Third Century Shan-Shan and the History of Buddhism’, Bulletin of the School of Oriental and African Studies 28. 3 (1965): 582-612; Enoki Kazuo, ‘The location of the capital of Lou-lan and the date of Kharoṣṭhī inscriptions’, Memoirs of the Research Department of the Toyo Bunko 22 (1963) [published 1966]: 125-71; John Brough, ‘Supplementary Notes on Third Century Shan-Shan’, Bulletin of the School of Oriental and African Studies 33. 1 (1970): 39-45; Lin Meicun, ‘A New Kharoṣṭhī Wooden Tablet from China’, Bulletin of the School of Oriental and African Studies 53. 2 (1990): 283-91.

44. It is generally accepted that Cadhota was the capital of a minor kingdom called Jingjue that may have fallen under Hun dominion before it became part of the kingdom of Kroraina by around the 3rd century CE. Kroraina, which may have had its seat at what is now the site of Loulan, had been subjugated by the Chinese (who called it Shan-Shan) in the 1st century BCE, remaining since then a vassal state but with what appears to have been a considerable degree of autonomy.

45. Some notable publications are as follows: Ratna Chandra Agrawala, ‘Position of Women as depicted in the Kharoṣṭhī Documents from Chinese Turkestan’, Indian Historical Quarterly 28. 4 (1952): 327-41; Ratna Chandra Agrawala, ‘Position of Slaves and Serfs as depicted in the Kharoṣṭhī Documents from Chinese Turkestan’, Indian Historical Quarterly 29. 2 (1953): 97-110; Christopher Atwood, ‘Life in Third-Fourth Century Cadh’ota: A Survey of Information Gathered from the Prakrit Documents Found North of Minfeng (Niyä)’, in The Silk Road: Key Papers, Part I: Pre-Islamic Period, ed. by Valerie Hansen (Leiden: Global Oriental, 2012), 115-43 [this paper first appeared in the Central Asiatic Journal 35 (1991): 3-4]; Helen Wang, Money on the Silk Road: The Evidence from Eastern Central Asia to c. AD 800 (London: The British Museum Press, 2004), 65-74; Mariner Ezra Padwa, ‘An Archaic Fabric: Culture and Landscape in an Early Inner Asian Oasis (3rd-4th Century C.E. Niya)’ (Unpublished PhD thesis: Harvard University, 2007).

46. See Heinrich Lüders, ‘Textilien im alten Turkistan’, Abhandlungen des Preussischen Akademie des Wissenschaften, Philosophisch-Historische Klasse 3 (1936): 1-38. See also Ratna Chandra Agrawala, ‘A Study of Textiles & Garments in the Kharoṣṭhī Documents from Chinese Turkestan’, Bhāratīya Vidhyā 14 (1953): 75-94.

47. Burrow sometimes translates tavastaga as ‘tapestry’ although there seem to be stronger linguistic grounds for ‘carpet’ as a suitable interpretation. See T. Burrow, The Language of the Kharoṣṭhi Documents From Chinese Turkestan (Cambridge: Cambridge University Press, 1937), 94.

48. Both terms, kojava and tavastaga, often appear grouped together with textile items such as felt and cloth in lists also containing non-textile objects. This placement suggests that kojava and tavastaga were textile products. Additionally, tavastaga are often measured in ‘hasta’ (hands), a unit elsewhere used to measure woollen cloth (§318). kojava are only counted as individual objects with no reference to dimensions - indicating that they may have come in standard sizes. The colour of these kojava is mentioned only rarely (§431-2) and some are specifically identified as ‘Khotanese’ (§549, §583, §592) but no further descriptors are provided, except for an instance where the epithet has not been conclusively translated (see §549 for ‘Khotanese alena rug’). Two other as yet obscure terms, ‘arnavaji’ and ‘agisdha’, are treated by the authors of the documents in a manner comparable with tavastaga and kojava. Arnavaji and agisdha are both grouped with other textile products when they appear in lists that also include non-textile items (§714, §207). Arnavaji, like tavastaga, is measured in hasta in §83 and its length of eight hasta as recorded in this document is comparable with that of a tavastaga in §578. Although this does not necessarily mean that arnavaji was also a carpet of some sort, we cannot as yet dismiss the possibility. The colour of an arnavaji is mentioned on one occasion (§83). Agisdha is mentioned along with other items of ‘clothing and bedding’ in §431-2, leading Burrow to suggest that agisdha is ‘probably some kind of rug or blanket’ or ‘some woven material’. See Burrow, A Translation of the Kharoṣṭhī Documents, 28, 38. Like kojava, these are generally counted as discrete objects without reference to dimensions. ‘Namatae’ (or ‘namataga’) is yet another textile product which is quantified in this way. The term may refer either to felt garments or to felt rugs. See Burrow, The Language of the Kharoṣṭhi Documents, 100. At present, it is not clear whether the terms arnavaji, agisdha and namatae were used to refer to carpets or rugs at all. Therefore, for the purposes of this article, I shall concern myself with kojava and tavastaga which can be interpreted more securely as ‘carpet’ and ‘rug’ respectively on the basis of strong philological evidence.

49. Helen Persson has also identified some soumak detail in this piece. Helen Persson, 'Pile carpets and flat weaves from the Silk Road' (paper presented at the International Conference of Oriental Carpets, Stockholm, Sweden, June 16-19, 2011). I am indebted to Helen Persson for so generously sharing with me the unpublished text of this paper.

50. A textile such as LOAN.STEIN.534 (fig. 10), may have been placed with its pile facing up at night for warmth as bedding and then been turned over during the day to show its brighter flatwoven side in which case the dense pile on the underside would still provide a comfortable cushioning effect while sitting or reclining. Another piece discussed by Wu Min, 80LBMB2:93 recovered from the grave B2 at Gutai in the Loulan area, reportedly displays what appears to be cut-loop pile on both faces. Observing that the pile on the ‘reverse’ is damaged and matted together, possibly on account of friction caused by contact with the body and through the absorption of sweat, she surmises that this textile might have served as bedding. Wu Min, ‘Study on Some Ancient Wool Fabrics’, 4.

51. See Burrow, The Language of the Kharoṣṭhī Documents, 84, 77. In light of this, §431-2 might also be seen as offering additional support to the idea that kojava was an item of bedding. Hans Bidder discusses the terminology used to describe foreign made carpets or rugs in early Chinese literature and suggests that the terms ‘ch’ü-yü’ and ‘t’a-teng’ might represent Chinese renderings of the words ‘kojava’ and ‘tavastaga’ respectively. It is unclear whether the former Chinese term referred to pile textiles at all whereas, on account of a strong philological indication, Bidder is confident that the latter term refers to knotted carpets. Other scholars, such as Ma Yong, have suggested a possible link between the Chinese term ‘ch’ü -so’ and the term ‘kojava’. See Wu Min, ‘Study on Some Ancient Wool Fabrics’, 12.  If that idea is indeed correct, then the following, cited by Bidder from an ancient Chinese source, might enable us to distinguish qualitatively between kojava and tavastaga: ‘the finer form of “ch’ü –so” is called “t’a-teng”’. For a lengthier discussion, see Hans Bidder, Carpets from Eastern Turkestan: Known as Khotan, Samarkand and Kansu Carpets (New York: Universe Books, 1964), 30-34. See also Zhang He, ‘The Terminology for Carpets in Ancient Central Asia’, Sino-Platonic Papers 257 (May 2015). A general survey of sleeping rugs produced by diverse rug-weaving peoples in more recent times shows that such textiles often feature pile that is shaggier and longer than that found in floor coverings, where the pile is usually trimmed. Bearing this in mind, and considering the philological and contextual evidence we already have for interpreting kojava as items of bedding, the difference in qualities between kojava and tavastaga as indicated by the Chinese remark above might be understood as corresponding to a distinction in function between bedding and floor covering respectively. See John T. Wertime, ‘Back to Basics: “Primitive” Pile Rugs of West and Central Asia’, Halı 100 (1998): 86-97 for examples of sleeping rugs produced in various parts of Asia.

52. See §361 for bow-makers, §578 for ‘goldsmiths of the people of the mountain’, §621 for potters and §715 for arrow-makers. Such textual evidence of specialised artisanal production is supported to some extent by archaeological finds.

53. Atwood, ‘Life in Third-Fourth Century Cadh’ota’, 119.

54. Hansen, The Silk Road: A New History, 39-40. See also Zhao and Yu, Legacy of the Desert King, 44.

55. Zhao and Yu, Legacy of the Desert King, 44, 46, 84-5, 87.

56. The beater MAS.727, from Loulan L.B., was discovered inside a particularly large structure adjoining what Stein supposed was a Buddhist shrine. Stein, Serindia, vol. 1, 404.

57. See Padwa, ‘An Archaic Fabric’.

58. All the beaters discovered at Niya by Stein, including MAS.557 (discussed in note 25), were found amidst the ruins of such structures. Both MAS.727 and a broken wooden beater with the Stein number LM.II.iii.03 were discovered in similar settings at Loulan. As for the two other Niya beaters discussed in note 16, I have been unable to locate information regarding their find spots.

59. Padwa, ‘An Archaic Fabric’, 164-6, 170-72.

60. For the purposes of governance and taxation, an avana was also treated as a spatially defined administrative unit. For a more in-depth discussion regarding the significance of avana, see and compare Atwood, ‘Life in Third-Fourth Century Cadh’ota’ and Padwa, ‘An Archaic Fabric’.

61. There are references to ‘golden staters’, ‘drachmas’ and ‘masa’, which may have served as coins, weights and/or units of account; in general, however, these appear less frequently than muli in the Kharosthi texts. For a broader discussion of money in the Kharosthi documents, see Wang, Money on the Silk Road, 65-74. To judge from these documents, Cadhotans were more accustomed to using media of exchange such as grain, wine and textiles rather than coins. Standard bolts of Chinese silk that served as money even as far afield as Loulan and Turfan do not seem to have been as common in Cadhota.

62. Burrow, A Translation of the Kharoṣṭhī Documents, 39.

63. See also §622 for carpets bought in Cadhota being sent to the capital.

64. For a detailed discussion, see Helen Wang, ‘Textiles as Money on the Silk Road?’, Journal of the Royal Asiatic Society 23. 2 (April 2013): 165-74. For specific case studies, see other articles in the same special issue of the Journal of the Royal Asiatic Society.

65. For an interesting exception, see Valerie Hansen and Xinjiang Rong, ‘How the Residents of Turfan used Textiles as Money, 273 – 796 CE’, Journal of the Royal Asiatic Society 23. 2 (April 2013): 297 for ‘6 and a half carpets’. For present-day examples of the dividing up of carpets in the same region, see Bidder, Carpets from Eastern Turkestan, 9. Bidder recounts an instance wherein four brothers fighting over their deceased father’s property decided to cut up a precious carpet into quarters so that they could share it among themselves. 

66. Valerie Hansen and Xinjiang Rong, ‘How the Residents of Turfan used Textiles’. See also Wang, Money on the Silk Road, 78. Of Turfan, Helen Wang writes that ‘from the late 3rd century to the early 6th century, silk and carpets were the main forms of money for purchases, loans and hiring labour’. Helen Wang, ‘How much for a camel? - A new understanding of money on the Silk Road before AD 800’, in The Silk Road - Trade, Travel, War and Faith, ed. by Susan Whitfield and Ursula Sims-Williams (London: British Library Publishing, 2004), 31-2.

67. Wang, ‘How much for a camel?’, 31-2, see also note 31.

68. This seems also to have been the case in Turfan. Valerie Hansen and Xinjiang Rong, ‘How the Residents of Turfan used Textiles’, 284.

69. It appears that bolts of silk were generally used by outsiders and by those Cadhotans who had dealings with outsiders or with the capital. See Hansen, The Silk Road: A New History, 49.

70. Although it has been posited that some may have been produced in Cadhota, it is far more likely that the silk referred to in these documents came from China. The remains of silk cocoons, alongside the seeds and desiccated trunks of mulberry trees found at Niya, suggest that some silk fabrics, probably the plainer varieties, may have been produced there. See Hansen, The Silk Road: A New History, 39.

71. In a passage extolling the practical advantages of the ‘aiwan’, which was ‘a kind of square central hall or Atrium’ found ‘in well-to-do people’s houses throughout the southern oases’, Stein commented that these ‘seemed delightfully adapted to the climate’; on account of this, he ‘felt sure from personal experience that much of the daily life, long dead and buried, must have passed in those ruined Aiwans’ encountered repeatedly in the ancient residences he had excavated. Marc Aurel Stein, Ruins of Desert Cathay: Personal Narrative of Explorations in Central Asia and Westernmost China, 2 vols (London: MacMillan and Co., 1912), vol. 1, 135-6.

72. Bidder, Carpets from Eastern Turkestan, 35, 40.

73. The dimensions were reported in centimetres as ‘128 cm by 295 cm’. It is unclear whether this piece is complete along its length since the publication cited here does not include a full image of the object. Zhao and Yu, Legacy of the Desert King, 80. 

74. ‘Hundred’ here refers to a unit presumably composed of a hundred households or a hundred individuals. For a discussion on the ‘hundreds’, see Padwa, ‘An Archaic Fabric’, 87.

75. It is unclear whether the same was applicable to tavastaga since we know that these came in a wide range of sizes. From the available textual evidence, it appears that tavastaga ranged in length from four to 23 hasta. The document numbered Niya 91NA15+91NA18 contains what appears to be a reference to a tavastaga 23 hands long but unfortunately, I have been unable to access a translation. For the transcription, see Stefan Baums and Andrew Glass, ‘Niya 91NA15+91NA18’, http://gandhari.org/a_document.php?catid=CKD0797.  

76. Admittedly, tavastaga (carpets) appear without reference to measurements in documents like §622 and §633 but in the latter instance even ghee receives the same treatment. We might put this omission down to the nature of these documents as being quite different from tax assessments or receipts. In another instance (§534), the length of a carpet is absent presumably because the document is a list of goods deposited by someone. Thus, in instances where the monetary worth of the carpet is an important consideration, its length is mentioned without fail.

77. By comparing the few instances where it is possible to definitively ascertain the value of kojava we find that they were worth either five muli or ten. The different prices might be indicative of different sizes or perhaps qualities although the Khotanese alena kojava (§549) was worth the same as a kojava of unspecified type (§222). Five and ten muli were fairly large sums and if indeed kojava came in multiple varieties, which was submitted as tax?

78. In all instances the authors of these documents only recorded the length of tavastaga in hasta, as if to say that all other attributes such as width, weave, material, patterns and finishes conformed to a fixed set of standards. This imparts the impression that tavastaga were akin to textile yardage when, in fact, carpets are quite distinct from that class of textiles. Perhaps the length of a carpet was enough to communicate a fixed set of proportions that the transacting parties were already familiar with? This, in turn, might have served as a tacit indicator of the intended use of the carpet and, by extension, of the technique used and labour involved: details that would have had a bearing on the value of these commodities but that the authors of these documents would not have seen the need to specifically record. The official nature of many such documents might also explain the omission of qualitative descriptors.

79. The carpet in §579 is 13 hands long and is priced at 12 muli which amounts to 0.92 muli per hasta. According to §590, ‘Śāṃcā […] arose and sold a woman called Lýipaae to the scribe Ramṣotsa’ for the price of ‘one viyala camel valued at forty, and a second aṃkla < tsa > camel valued at thirty, one carpet twelve hands long, and a second carpet eleven hands long. Also received were eight sutra muli. The total price is ninety-eight. So they agreed on equal terms.’ This tells us that the price of the carpets, measuring a total of 23 hasta, jointly amounted to 20 muli. Therefore we may deduce that each hasta of carpet was worth about 0.87 muli, or roughly nine-tenths of a muli, on average. The latter calculation assumes that there was no particular distinction in value between the two carpets. It is unclear whether the average value we have arrived at represents a fixed rate or whether some amount of negotiation took place between the transacting parties on both occasions. As the documents do not record any such proceedings, it is difficult to judge what factors might have influenced the final value of these articles.

80. Interestingly, the transactions recorded in §579 and §590 are dated respectively in the 9th and 17th regnal years of the king Amgoka. It is from the 17th year of Amgoka’s reign that we receive the earliest confirmation of an event of considerable political significance that is believed to have placed Kroraina (at least nominally) under Chinese control. The Chinese seal affixed to §590 suggests increased Chinese influence even with regard to affairs of local concern at this western extreme of the kingdom. See Brough, ‘Comments on Third-Century Shan-Shan’, 598-9. What effect the altered political status of Kroraina may have had upon the economic situation in Cadhota is difficult to tell, though.

81. Li Wenying, ‘Textiles of the Second to Fifth Century’, 254.

82. A coverlet, M3:24, recovered from the joint burial M3 discovered at Niya in 1995 was evidently made up from two pieces of the same Chinese silk brocade fabric joined together along the length. It survives intact and measures 94 cm x 168 cm. Zhao and Yu, Legacy of the Desert King, 72-3. Interestingly, these dimensions are quite similar to those of the woollen blankets just discussed. Thus, it seems that the silk fabric was received in the form of yardage and that it was later assembled locally into a coverlet.  

83. Zhang He lists a ‘diamonds and tiger’ patterned specimen from Loulan L.E. As I have been unable at present to access any further details, it has not been possible for me to include this piece in this discussion. Zhang He, ‘Figurative and Inscribed Carpets’, 70.

84. The Yingpan specimen, 89M3:1, measures 90-100 cm x 260 cm while its counterpart from a gravesite near Loulan L.E. measures 103 cm x 266 cm. See Li Wenying, ‘Textiles of the Second to Fifth Century’, 256 and Yue Feng, Xīnjiāng lìshĭ wénmíng jícuì, 150. I am indebted to Dr Zhang He for sharing with me the latter publication. Wu Min discusses another lion carpet from Yingpan, 89BYYM1:12, which measures 95-100 cm x 260 cm. Wu Min, ‘Study on Some Ancient Wool Fabrics’, 15. It appears that these pieces, with their common subject matter and matching dimensions, belonged to a specific, rather standardised class of carpets that might have served a specific function. We might, perhaps, see these lion carpets as early representatives of an ancient tradition that has survived into the present day in the form of the tribal ‘gabbeh-ye-shiri’ (lion-patterned gabbeh rugs) produced in the Fars province of Iran. See Parviz Tanavoli, Lion Rugs: The Lion in the Art and Culture of Iran (Basel: Wepf & Co., 1985).

85. Li Wenying, ‘Textiles of the Second to Fifth Century’, 256.

86. There are also references to carpets and rugs that may have originated outside Cadhota. See §633 for kojava and carpet bought from ‘the mountain’. No fewer than three documents refer to ‘Khotanese’ kojava (§549, §583, §592) but this need not necessarily mean that such kojava were actually made in Khotan. We know that Cadhota received Khotanese refugees from time to time and these refugees might have practiced a distinctively Khotanese style of rug weaving in their new home. A comparison of §222 and §549 suggests that there was no difference in price between an ordinary kojava and a Khotanese one. A comparison of the dimensions of some of the slightly later pieces from the looted site at Shanpula suggests that a standard width of about 116 cm to 119 cm had been established in the region. For the dimensions of the five Shanpula carpets recovered in 2008, see Zhang He, ‘Figurative and Inscribed Carpets’, 59, 77-9.

87. Stein, Serindia, vol. 1, 372-3.

88. ‘Even today the Mohammedans bring their dead for burial rolled in Kelims or felt carpets.’ Bidder, Carpets from Eastern Turkestan, 82.