Ashmolean Eastern Art Online, Yousef Jameel Centre for Islamic and Asian Art

Before looking at the designs of the textiles and at their function and distribution, certain technical characteristics need to be identified and discussed. As the fragments are of cotton woven in India, they are inevitably part of the discussion on the early distribution of cotton fibre and fabric between the subcontinent and Egypt. For contacts during Pharaonic times the evidence is obscure and in bits and pieces, while for Hellenistic and Roman Egypt there are numerous literary references, some of them very reliable as historical sources. The international demand for Indian cotton depended on three aspects: on the limited availability of cotton elsewhere, on the comfort and aesthetic pleasure experienced from wearing a well-spun and woven cotton garment, and on the unique understanding of the Indian producer for the manipulation of dyes, which produced results that could not be matched easily. These three points will be taken up in this and the following chapter.

What follows started out as a single chapter. However, as it grew it seemed unkind to the reader to keep all the material together under a single heading. Textile technology and terminology are difficult matters to deal with, not only for the author, but even more so for the reader. However, it must be an essential aspect of this particular study, and, in order to make it more bearable for the reader, I have decided to split the practical discussion into two parts. The first begins with a brief survey of the distribution of cultivated cotton followed by a definition of the structural characteristics shared by all the Indian textiles in this collection. That includes a consideration of the spin and weave. The next chapter will focus on the differences of dye techniques employed in the production of the fabrics. As has been mentioned in the Introduction, the division of the catalogue sequence is based on distinctions of techniques. The reason for adopting this method needs elaboration, which Chapter 6 tries to provide. There has been considerable confusion and disagreement among textile specialists about the use of terminology, in particular regarding the correct terms to adopt for resist- and/or mordant-dyed fabrics. The term ‘block-printed’ is sometimes used to include both resist- and mordant-dyed textiles, or there may be a distinction made between that term and resist-printed textiles. The latter terminology was adopted, for example, by Irwin and Hall (1971). The opportunity is taken here to look at the problems of terminology, and an attempt is made to come to a basic definition [1]. This provides the introductory discussion to Chapter 6 as the second of the technical chapters.

Cotton as textile fibre

The sources for natural textile fibres can be divided into four broad categories: animal coats, vegetable bast fibre, silk, and vegetable seed-hair [2]. Wool is from the first category, and may be supplied by, for example, sheep, goat, or camel. Flax is the best-known example of the second group, as it is the fibre from which linen cloth is woven, but ramie (nettle) and banana fibres are also used in Asia [3]. For all the bast fibres, parts of the stalk of the plant are made pliable enough to be spun into a thread. Silk, of course, is the fine thread the silkworm produces to make its cocoon. Cotton belongs to the fourth category, as the hairs that develop in the seed pod are the fibres that are spun into thread. The majority of textiles in our collection are of cotton; the few exceptions that were woven from a flax fibre are not of Indian origin and have to be treated separately. Wool and silk are not present among the woven cloths, although some silk thread was used for sewing.

All four types of fibre apparently have been used in Indian textile products, as can be discovered from the Vedic sources. Cotton (karpasa) is not actually mentioned earlier than the fifth century BCE, although it was of course used much earlier. Wool (urna) was and is the source for textiles in the Himalayan regions of India, and is mentioned as a gift in the Mahabharata. Silk (kauseya) is mentioned in lists of textiles in early Buddhist texts (Gopal 1961: 61). Apparently there were two types in use, one from the cocoons of silk cultures, the other a wild silk with much shorter fibres. It is possible or even likely, though, that high quality silk weaving in India also depended to some degree on the import of Chinese silk yarn. The name ksauma might refer to flax, but it could also mean a different grass fibre, of a type that was used in China during Han times and is still woven in the hills of Assam. It is certain from the literary sources that flax was used in India for fibre production, although its specific purpose is unclear to me at the moment [4]. It never assumed a promi­nent role in the textile industry, and none of the historical export fabrics were linen. Despite a silk-weaving tradition that has produced extremely fine cloths, such as the double-ikat patola textiles, and regardless of the history of Kashmir shawls, which were dependent on the fine wool fibres of Himalayan goats, it is in particular for its cotton industry that India has been famous.

Cotton distribution from India

The earliest cotton find so far known was recently reported from Dhuweila in Jordan (Betts et al. 1994). The site was established as a seasonal hunting camp in the Early Neolithic period; it was never particularly important as a settlement. Yet from it comes the earliest evidence of an international transfer of textiles: lime plaster fragments had remains of cotton fibre attached to them, as well as clearly identifiable impressions of weaving. The weave was tabby, and, as far as can be seen, the spin was in the z-direction (see below for an explanation of spin direction). The fragments have been radiocarbon-dated to 4450-3000 BCE. Previously, the earliest cotton remains had come from a Nubian site dated to the mid-third millennium BCE, where seeds and cotton lint had been found. However, for the Nubian sample there was no evidence of a textile purpose: the plant had apparently been used as animal feed.

The cotton fragments were carried to northern Arabia from a distant source, most likely from India. The textile would either have been brought along the trade route from the southern coast of Arabia, or via Mesopotamia and along the northern rim of Arabia. It is certain that the Harappan civilization experienced a rapid rise in urbanization between the middle of the fourth and the third millennium BCE, with an expansion of contact and trade links with both southern Arabia and Mesopotamia. The earliest spun and woven cottons up to this date were from Mohenjo Daro, with a later third-millennium BCE date.

The history of cotton distribution to Pharaonic Egypt is by no means clearly established, and in the literature one can come across statements that are not always based on sufficient evidence. It is certain that flax was the source of most Egyptian cloth in Pharaonic times. Very fine linen textiles were woven from the fibre by the third millennium BCE, and the overall importance of flax for the textile industry in Lower Egypt, at least, continued well into Islamic times. Textiles from Pharaonic tombs are usually of linen, although it has been claimed that mummy wraps were also of Indian muslin, i.e. a cotton cloth (Asthana 1976: 165). I have not been able to find proof for this assertion [5]. It may be true, but unless the reports are substantiated by a fibre analysis, the claim has to be treated with caution. In the same context, it is also mentioned that the tomb textiles are sometimes dyed with indigo from India. The same assertion was made by Forbes in his survey of ancient technology, where he said that indigo was used on a small scale in early Egypt and had been found on clothing dating back to the fifth dynasty (c.2500 BCE), as well as on later mummy wrappings (Forbes 1956: 110). However, there can be no scientific proof for the use of Indian indigo, or even of indigo as such: a dye analysis can only identify the dye agent indigotin, not the actual species used for the dye. Indigotin is present in several plants that are otherwise unrelated, and it is even chemically close to the purple dye obtained from Murex and Purpura shellfish (Balfour-Paul 1992: 99). It is possible or even likely that indigo was used at an early date in Egypt, but chemical analysis cannot provide the proof [6]. By Hellenistic times the import of Indian indigo is more clearly established, as it is referred to in written sources (Pliny XXXIII. 163; XXXV. 43, 46).

A classification of the species and varieties of cotton is difficult and for the non-expert confusing to follow. For the early contact and trade between the Indian subcontinent, Arabia, and Egypt it is a relevant but still very obscure topic. Apparently cotton was originally harvested from a tree version of the plant Gossypium arboreum L. As the plant only grows in a hot climate without hard winters, its distribution remained limited. Another cotton plant, the annual Gossypium herbaceum L., was more suited to climates with hot summers, but cold winters. Watson considers it possible that this cotton plant was not cultivated until Islamic times, but archaeological evidence for cotton seeds of the Gossypium herbaceum L. type have been found in a Late Han context from Central Asia [7]. G. arboreum and G. herbaceum are often referred to as tree and shrub cotton, respectively. However, Watson (1977: 363) points out that both species have perennial as well as ‘herbaceous’ annual forms. The latter he assumes was a late development [8].

So far the earliest evidence for a combination of both cotton cultivation and its use for textile production still comes from the Harappan civilization and dates to the third millennium BCE (Watson 1977: 357; Allchin 1969: 325). But as is suggested by the find from Dhuweila, there may have been a much longer history to the Indian weaving industry. However, wild cotton trees also grow in east Africa and Arabia, and it is uncertain whether they are indigenous or arrived from India. The east African use of cotton in textile production is not yet sufficiently investigated. Zeuner (1954) claimed that although the Harappan cotton from the site of Mohenjo-Daro has all the measurable characteristics of modern Indian cotton, the wild relatives of the ancient cultivated Indian cotton are not found among the Indian species, but are related to a group confined to Africa and Arabia. The question posed by him is: ‘Was cotton first grown in southern Arabia and thence taken to the Indus valley long before 2500 BC?’ (1954: 374). The wild cotton tree produces seed fluff that is short and therefore difficult to spin. There is no doubt that these particular shortcomings were overcome at an early date in India, certainly by the time of the Harappan civilization, as the evidence of cotton from Mohenjo-Daro is of a cultivated species.

To sum up, it is certain that the evolution of a cultivated version of cotton was very early. Cotton products from India gained a predominant role in international trade, possibly as early as the fourth to early third millennium BCE, if we accept that the Dhuweila find may be of Indian origin. Hutton referred to a cotton fabric with an Indus valley stamp on it, found at a prehistoric site in Iraq (Hutton 1946: 135). Relatively well known is the attempt of the Assyrian king Sennacherib to grow cotton in c.700 BCE, which suggests that a lucrative cotton harvest was already a desirable objective by that date. However, the cold winters of Assyria would not have been a viable climate for the survival of the perennial variety of Gossypium arboreum. Ultimately the efficient utilization of cotton depended on the cultivation of a variety with long seed fibres and a short maturation period, so that the plants could be widely grown even in regions with harsh, continental winters. The annual shrub of Gossypium herbaceum eventually fulfilled this requirement most successfully.

As linguistic evidence suggests, some time during the first millennium BCE cotton cultivation for the purpose of cloth-weaving, as practised in India at the time, started to move in three directions: north-west into Assyria, westwards past the Arabic Peninsula into Sudan, and eastwards through Burma to South-East Asia (Johnson and Decker 1980: 259). In Central Asia there is archaeological evidence for its presence in Khotan and the Turfan basin in the Late Han period (CE 25 to 220), as well as in Yünnan and Szechwan (Kuhn 1988: 58-9).

The importance of cotton as an Indian textile-producing fibre caught the attention of writers in classical antiquity, whether they were commenting on historical events, as was Herodotus, or on the distribution of plants and natural history, as were Theophrastus and Pliny [9]. During the Hellenistic era, Indian imported cotton was a major aspect of the Roman Red Sea and Indian Ocean trade, and quite detailed locations and sources are given in the Periplus (Casson 1989: 22) [10]. But although Indian cotton was famous for its quality, there were by then other centres of production. Theophrastus (IV. 7. 7-8; II. 5. 5) mentions the cotton grown on Bahrain, and in Nubia the cotton production of Meroë is recorded for the Graeco-Roman period [11]. While Theophrastus depends on hearsay and may not be reliable, from Meroë we have actual textile remains that have been analysed (Thurman and Williams 1979). The Meroitic finds even suggest that cotton was a main staple and source of wealth at the time. However, no continuous picture of cotton growing in the Sudan generally has emerged. It is certain that cotton was produced locally in Upper Egypt, and was known in the Near East and the classical world in Hellenistic times, but it was only cultivated on a larger scale in Meroë and India, and possibly Bahrain. Western Central Asia apparently also was an early producing area.

As the Indian production was most prolific of all, the considerable number of Indian cotton fragments that have survived in Egypt from the medieval Islamic time onwards give us only a relatively small and late glimpse of a much larger, continuous trade network of considerable time depth.

The fabric structure

There are basically three methods of decorating the surface of a textile: through the weave, by paint or resist and mordant application, and by adding further substances to the woven fabric, such as embroidery or appliqué. We are only concerned with the second method here. For all our textiles the basic fabric itself is used as the vehicle for design, but does not contribute to the patterning. For that reason, it seems best to summarize the general characteristics of the cloth itself, before discussing the textiles in detail. The weave used throughout is a balanced tabby with, in general, an equal weight and spacing of warp and weft [12]. The thread-count per square centimetre varies, but it is rarely below 15 x 15, or higher than 25 x 25. Each catalogue entry gives the thread-count, but the thread diameter was generally not recorded. The average diameter is between 0.2 and 0.4mm., which is completely consistent with that published for the much smaller collection at the Kelsey Museum in Ann Arbor where the thread diameter is given for each fragment (Barnes 1993).

Before the cotton fibres can be utilized, they need to be separated from the seeds to which they cling; this can be done by hand (the least efficient way), but is usually done with a hand mill, the cotton gin [EA1994.718]. Prior to the invention of the gin, the separation was achieved by moving a roller over the seeds and hair; this method is illustrated in a scene from Ajanta. The next step in cotton preparation is a fluffing of the fibres with the cotton bow, which serves two purposes: it removes impurities of dirt and seed pod remains, and it loosens up the fibres in preparation for spinning [EA1994.6].

It is impossible to say from the appearance of a thread whether it was spun with a hand spindle or on a spinning-wheel. The early history of complex spinning devices seems to be linked with sericulture in China (Kuhn 1988) [13]. The spindle wheel was the major time- and labour-saving tool developed for the production of thread. It probably originated in China, where the earliest representation is on Han dynasty stone reliefs; there is no evidence for an Indian origin (Kuhn 1988: 168). The cotton and weaving industry of India apparently did not depend on the spindle wheel, although its eventual introduction must have given considerable impetus to the growth of the industry. It could have been a major contributing factor to the apparent increase in the Indian textile trade in medieval times.

When spinning any fibre, the twist given to the spin can take two directions, either / or \ (Fig. 1 (a) and 1 (b)). To simplify verbal reference to each direction, so as not to have to say each time ‘from top right to bottom left’ or ‘from top left to bottom right’, the / direction of Fig. 1 (a) is called z-spun, the \ direction of Fig. 1 (b) s-spun, each term referring to the direction of the letter’s central bar. The textile fragments in the collection that I identify as definitely Indian are all woven from z-spun cotton. For Indian cotton thread, this is the usual twist. When producing a yarn, two or more spun threads are often plied together; this increases the thickness and strength of the finished yarn. However, the Indian textiles are all woven from single threads without ply.

According to Barber (1991: 65-8), the direction of the spin was originally determined by the type of spindle used. A spindle with the whorl towards the top, which is set in motion by rolling it along the thigh, produces the s-twist, while a spindle with the whorl near the bottom is started with a flick of the thumb, for which the natural direction is from bottom to top (Barber 1991: 67). There has been some discussion about whether the twist direction of spinning a thread was determined by the fibre spun, and, as it happens, an s-twist is more suitable for a flax fibre than a z-twist, while the natural turn of cotton fibres supports a z-spin. But during the historical time under consideration here, a general mix-up seems to have happened in Egypt. The Egyptian cotton and flax fibres identified are spun in both an s and z direction, without regard to the fibre used. As the spin direction found in fibres, whether s or z, seems to be culturally determined to some degree, this is potentially interesting. Who was involved in the spinning industry at the time? It could mean that spinning workshops employed people who came from two different traditions.

Also remarkable is that when the spinning-wheel was introduced, the twist direction was apparently retained. This could be achieved by setting up the tool in a specific way. The placement of the continuous band that connects the wheel with the spindle determines the twist: an open band (a single, open loop) results in a z-spin, a ‘closed’ band (the loop is given a twist and moves in a figure 8) produces the s-spin [14].

After spinning the thread, it is then reeled into hanks. Sizing the hanks with starch strengthens them and makes them less likely to break when they are set up on the loom to be woven. For cloth that is to combine various colours in the weave, the hanks would now be given to the dyers; for the purpose of our block-printed textiles, though, the undyed threads were woven into cloth. The early Indian textiles were produced on simple treadle looms with the warp stretched horizontally [15]. The width of a cloth woven on this type of loom is restricted to the width the weaver can comfortably work with as he manually inserts the shuttle with the weft threads; it is not practically possible to weave a cloth that is wider than 80 to 90 cm. A typical cloth to be block-printed would, however, be twice as wide. To achieve this size, two widths of cloth would be sewn together along the selvedge. This produced the so-called piece-good, a single cloth approximately 180 cm. wide and at least 230 cm. long. The length of the fabric could vary, but the width was fairly constant. Most weaving in India nowadays will be done on industrial or semi-industrial looms, but the fabric used for block-printing is sometimes still made up from two lengths sewn together along a selvedge.

Weaving and dyeing are two distinctly different activities and do not take place in the same workshop. The specialists in block-printing and dyeing purchase the ready woven cotton fabrics in bales that are divided and worked as lengths of 10 m. Only a fraction of all dyers’ workshops also produced (and produce nowadays) block-printed textiles; the craft combines two specializations, dyeing and printing, and often adds a third: that of wood-carving, in the preparation of the blocks. To some degree we should move into the present tense now, as the techniques can still be observed, and there may even be a revival of good quality production under way. Resist and mordant methods of textile decoration are still found in India and Pakistan, and it is instructive to follow the present-day production procedures, as they seem at least in part similar to the manufacture of the historical fragments. Varadarajan (1983) and Bilgrami (1990) both present detailed accounts of present-day production: Lotika Varadarajan reports on the techniques and patterns used in Rajasthan and Kutch, while Noorjehan Bilgrami focuses on the still vital printing tradition of Sindh in Pakistan. The craftsmen who do the printing now usually come from families where generations have been textile printers and dyers.

Before the resist or mordant mixture can be applied, the fabric needs to be washed, steamed, and soaked in a solution containing oil, dried dung, and carbonate of soda. The washing and steaming will remove all starch remnants and other impurities from the fabric, and the soaking in an oily and acidic solution (provided by the dry dung) prepares the fabric to uniformly receive the printing (Bilgrami 1990: 55-66).

Contemporary block-printing in India and Pakistan uses wooden blocks, sometimes with the addition of nails driven into the wood. The nail heads would appear as round dots, as in the numerous pearl borders of our fragments. Entire blocks made from metal are generally not used, unlike in Java, where copper blocks are employed to print the wax for ‘stamped’ (cap) batik textiles. The blocks made in the Sindh now are from local trees; Bilgrami gives three types (1990: 37) [16]. I was told by the printer and dyer Mohammad Siddi in Dhamadkar, Kutch, that acacia wood was used for carving blocks. According to Bilgrami, the block carving is a specialized craft, the knowledge passed from father to son. Elsa Sreenivasam also reports this for Gujarat, where wood block carvers work in the small town of Pethapur, to supply the blocks for the printers of Ahmedabad (Sreenivasam 1989: 29). Mohammad Siddi, however, carved his own blocks, and his sons continue the tradition (Ill. 12).

Both Varadarajan and Bilgrami concentrate their descriptions of present-day block-printing on the manufacture of ajrakh, a blue and red cloth that is closely related in its design structure to the Indo-Egyptian textiles. There is also a great tradition in south-eastern India for resist- and mordant-painted, rather than block-printed, textiles; these are referred to as kalamkari [17]. The chintzes imported into Europe from the seventeenth century onwards were predominantly produced with a brush, although occasionally the term ‘chintz’ is used to mean resist- and mordant- dyed textiles in general (Victoria and Albert Museum 1983). The application of resist and mordant, by block or brush, is described in various ways, sometimes creating a certain amount of confusion. This issue will be addressed in the following chapter.