peter harris, tapestryweaver

Cohen, Steven, ed., “Kashmir Shawls: The Tapi Collection”, The Shoestring Publisher, Mumbai, 2012

For too long I was eager to get my hands on a copy of “Kashmir Shawls: The Tapi Collection”(1), the lavishly-produced catalogue of shawls in the collection recently unveiled as the Textiles and Art of the People of India (hereafter TAPI), founded by Praful and Shilpa Shah, of Garden Silk Mills in Surat, Gujarat state.  As a tapestryweaver and student of the design and weaving techniques of Kashmir shawls, I was looking for new insights and inspiration in essays by the highly-regarded contributors Steven Cohen, Rosemary Crill, Monique Levi-Strauss, and Jeffrey Spurr, and large-scale close-ups of the fabrics themselves.  I was rewarded with a mass of intriguing technically-minded information, an unexpected homage to my mentor, the late Grace Beardsley, and many heretofore unseen loose gems of shawl fragments.

Kashmir shawls have been treated as collectable art objects, although they are relatively numerous, anonymously made, and usually un-dated.  Dealers and historians have had a free hand to speculate and argue about the age of an antique piece, based on its place in generalizations about the stylistic development of imagery and overall design.  One of their major points of agreement is that the set of collectable Kashmir shawls is closed – that the craft of shawlweaving died out when the shawls lost their fashion status in the late 19th century.  But before then, their tremendous popularity was a complicating factor, constantly pressing for designs that were new, or that responded to the tastes of export markets.  The history of the shawlweaving industry demonstrated craft skills pushed to their limits, an innovative digital system for storing and retrieving designs, and a diversification of weaving methods hearkening the Industrial Revolution.  This mix of tradition and innovation continues to the present.

For years, I have searched for any account of shawlweaving published in English that demonstrated the observer’s understanding of weaving.  The most often quoted historical observer, William Moorcroft(2), described the tapestry-style image-making as “loom embroidery”, an ironic reversal for tapestryweavers accustomed to hearing “tapestry” applied to all sorts of techniques and resemblances.  Replies from curators and historians too often begin with disclaimers about their knowledge of weaving.  But published descriptions of pieces sometimes include threadcounts, usually a good indicator of quality.  In his introductory essay, Steven Cohen advocates another objective measurement, the average weight of the fabric, rendered as grams-per-square-metre, or “gpsm” for short, to provide a basis for judging the date and place of its manufacture.

Here was an opportunity to take grams-per-square-metre measurements as part of the expert scrutiny invited for this private collection of shawl fabrics.  Prior to this, the only example of systematic weight measurements was performed by Grace Beardsley in her close analysis of the shawls of the Koelz collection at the University of Michigan, published under the title “Wrapped in Beauty”(3).  From a copy of her manuscript she gave to me, here is her entire discussion of the weights of shawl fabrics:

“Weight.  Although no fiber analysis was run on the Koelz shawls other than an occasional burning test to rule out silk or vegetal fibers, each shawl was weighed to assess relative weight.  Presumably, a pashmina shawl woven of finely spun goat hair would weigh less per square meter than one of sheep’s wool, other factors remaining constant.  Of course, none is really constant in weaving, but what aspects, if any, of fiber content can be illuminated by shawl weights?

“A comparison of weights per unit area is shown in Figure (shawl weights).  The range is considerable.  If the weight of the twill tapestry patka, discussed under turbans, is included and those of the three cotton-lined shawls excluded, the range is found to be 100 to 370 grams.  The median falls between 230 and 240 grams per square meter, about the weight of a square meter of ordinary cotton bed sheeting.

“Among the eight lightest shawls, those of 180 grams or less, UMMA 17337, labeled “pashmina,” rightfully should be included, for even with a cotton lining it weighs only 189 grams per square meter.  Others in the group are UMMA 17313, 17312, 17331, 17326, 17342, 17327, and 17317.  All are estimated as of late eighteenth century to 1830 manufacture.

“At the opposite extreme, the eight heaviest shawls range from 300 to 370 grams per square meter.  None is of the early nineteenth century, and five of them have been designated as of middle to third quarter of the nineteenth century.  So it seems that in Koelz shawls at least unit weight increased as the century advanced.  A comparison of Koelz figures with other shawl weights is elusive, for the single mention in shawl literature of actual weight is made by Watson, who notes “woven shawl made at Kashmere of the best materials and weighing seven pounds.”  (Watson 1866: 121)(4).  However, as he states neither style nor size, there is little basis for comparison.

“One might expect the heavier shawls to be more compactly woven or to show coarser yarns, or both.  Differences in Thread [sic] counts are neither great nor uniform.  Shawls of both the lightest and heaviest groups have warps set anywhere from 30 to 42 per cm.  The only exception is the Balti long shawl with an extremely low warp count of 22.  Its weft count of 30 per centimeter is within the 30 to 58 range of all other shawls of the two groups, the heavier group being somewhat stronger in the higher weft counts.  But for the most part, the heavier weights must be attributed to thicker and only somewhat more compactly beaten up weft yarns.  Thicker wefts, of course, allow more area to be woven in less time than does the use of finer yarns under comparable conditions.  Disparity between warp and weft diameter is usual, especially in the later woven shawls, and may run as high in extreme cases as wefts eight times the diameter of warps.  As a device for singling out pashmina shawls, weight measurement can be described only as a way of identifying shawl groups in which pashmina shawls are likely to be found.”

TAPI 12A fragment, 134 gpsm

TAPI 12 whole shawl and detail, 66 gpsm

Grace Beardsley’s weight measurements were conducted as part of a thorough but general assessment of the Koelz Collection shawls using commonsense methods available at the time.  More recently the measurements of the TAPI Collection shawls were performed in the hope of establishing a new scientifically objective basis on which to detail the provenance of individual shawls.  Yet, though the need is obvious enough, it does not seem possible to weigh representative portions of a fabric, apart from taking the average of the whole fabric.  Especially in the case of whole shawls, anyone who has handled or even just viewed one would expect the density of a heavily decorated border or palla portion to be much greater than the plain-coloured central field.  The most likely explanation for the disparity in gpsm calculations for TAPI Plates 12 and 12A is the huge proportion of plain field included in Plate 12.

Tapestry-woven and plain-woven areas in the same structure (twill), may weave up at different rates.  The slack introduced by interlocking multiple tapestry wefts may tend those areas to beat up more compactly.  Long, continuous weft passes introduced by shuttle across a plain field come under tension to bend around successive warp threads, and require more force to beat into place.  In tapestry-woven areas extra weft yarn lengths taken up by interlocking and floats and skips, will add to weight.  Shawl tapestry wefts have been cited as singles thicker than the 2-ply warps (5) for better coverage of the background warp colour, and tensile strength in wefts is less important.  Thicker threads will build up faster.  Less weight of material, but more finely-spun, is required to produce a fabric of equivalent area at a higher sett, and with increased design resolution and detail.  At best, differences in “gpsm” are essentially differences in quality to meet market expectations, within the limits of manageable weaving practice, cost of material, and pressures to increase productivity.  To say that higher gpsm is an indicator of date or place of manufacture, is to imply merely that industry-wide standards of quality declined, or that the output of certain weaving-centres was inferior.

On the other hand, because design information was recorded and reproduced using the talim system, as precisely as digital media can record and playback music, the same design can be woven stitch for stitch by different workshops or different generations of weavers, at the same or different standards of quality.  If the design can be indistinguishably duplicated then the physical properties of the copy become important.  Truly comparable measurements of gpsm may provide useful commentary on production standards and market conditions, but they cannot trump the experienced eye of a design historian.  One can move from accumulations of particular evidence toward generalizations, but to determine the particular identification of an intricately figured piece of weaving just based on this technical characteristic, is reductionist.

There is an air of second-handedness to Steven Cohen’s knowledge of weaving.  He keeps referring to “2 & 2 weft-faced twill” as if it were one concept, when in fact it is two: first, the 2-up-2-down twill weave structure, usually woven with a balance of equal numbers of warp and weft threads per inch.  Important for the repeatability of the design is consistency in the weaving practice, because what would be a perfect circle in balanced weave becomes a flattened oval if the actual weft count goes higher.  This would be termed “weft-predominant” or at most “weft-faced” fabric structure where an increasing proportion of the warp is covered.  In shawlweaving the design is displayed by the different-coloured weft threads, so this desirable feature can be emphasized by using slightly thicker wefts, carefully matched design instructions, and a degree of weft-predominance not so much as to make the fabric thick and stiff.  The warp colour always remains visible as demonstrated by shawls with striped warps, and sometimes to marvelous effect as in the illusion of transparency in “moon” shawls.
Cohen’s remarks about the origins of shawl weaving, the association of twill weave with wool for thickness and warmth, plainweave with cotton or silk in more tropical lowland regions for thinness, were helpful.   Even if there was an earlier Kashmir shawl industry producing natural-coloured or dyed-in-the-cloth shawls – and a lot of the patkas in paintings described as showing plain shawls, have decorated ends – to attempt to introduce tapestry-woven pictorial decoration is still a remarkable leap.  The demand this elicited for a luxury product spurred all its refinements: vanishingly fine threads, design artistry and complexity, and the technologies of precise weaving instructions to manage tapestry imagery in a weave structure that keeps shifting diagonally.

If Jeffrey Spurr had provided or been able to refer his readers to a full explanation of the talim system, the ease with which imported design ideas and customer preferences can be incorporated in Kashmir shawl designs would be obvious.  A text in G. W. Leitner’s compilation (6) explains that designs, as simple line-drawings, could be drawn by artists anywhere, in anybody’s royal atelier, and sent to the Kashmiri shawlweaving workshop where a highly skilled and experienced weaver, the “tarah guru” (“colour-caller”) converted it into line-by-line weaving instructions, called “talim”, that specified the colours and scale in weaving.  That would be the “Kashmiri input…incorporating foreign, sometimes peculiarly Sikh, motifs”, described by Janet Rizvi in her review of Frank Ames’ “Woven Masterpieces of Sikh Heritage” (7).  There she says Ames is “undoubtedly correct in associating them (“innovative designs”) with the establishment of Sikh rule”, and in her review of the present volume (8), she says Spurr’s “suggestion that it (“variety of new motifs”) originated in Europe and reflects the European fad for orientalism is persuasive”.  She could be right both times.  Both imports could be converted into talim with easy technical facility, much as European producers could draft newly-arrived Kashmir shawl designs, for their Jacquard-woven copies.  Not only that, but archived component designs for palla, field, vertical and transverse borders could be mixed-and-matched, and colours substituted through the simple use of a colour-card as a reminder to the weavers.  Moorcroft’s account lists 88 varieties of shawl goods and the markets they were traded to.  Spurr erects a structure of umpteen divisions of “style”, “mode”, “phase” and “type”, in an attempt to pigeon-hole individual shawls, and while his allusions to different evidence and influences are informative, precise categorization is probably too limiting.

More broadly, aside from specific European or Gothic imagery on the one hand, or Sikh iconography on the other, a weaver might observe that emerging 19th c. European weaving technology, particularly the Jacquard mechanism, heavily influenced later Kashmir shawl style: reduced numbers of colours due to the limited number of shuttles a semi-automated European loom could handle; tiny, dispersed details everywhere, answering the need for weft tie-downs in a shuttle-woven fabric; the monstrous vegetative growth of the palla motifs invading the central field reflecting the runaway thousands of cards in the Jacquard chain.  If European shawl customers were used to seeing these effects, and always looking for something fashionably new, and the expositions were awarding prizes for new, more elaborate designs, Kashmir shawl designers would feel compelled to produce these kinds of designs, even though they were not a natural expression of tapestry technique.
In her contribution on embroidered shawls, Rosemary Crill notes that a corner ornament matching a row of upright palla motifs was too difficult to weave, therefore embroidered.  In fact it could have been woven, but would have required a specific new talim to take into account the tilt in its orientation.  On the other hand, I think human and animal figures might be better embroidered because they are likely to be closely judged by additional criteria of recognizable and expressive gesture, not just the symmetry of leaf and flower.  Monique Levi-Strauss attributes the reduced variety of colours to the invention of synthetic dyes in mid-19th-century, certainly a moment that led to the appearance of distinctive new colours like intense pink.  When viewed in the context of technologically competent craftsmanship, these examples become practical and artistic choices, not constraints.

As a craftsman looking at a fragment showing one or two copies of a beautifully intricate motif, I’m not stricken with regret that the piece is incomplete, not collectable.  The inspiration, discipline, and skill of the maker shines through, even if the mirror is a fragment.   How the mirror got broken is another story.  The catalogue of pieces complete and incomplete, even recovered from garments for which they were cut, is fascinating, the format generous, the colour printing unexceptional.

It was a revelation to me to see the production of shawls in Kashmir described as an “industry” in the pre-Industrial period (9)  – a large-scale production involving what we would now call division of labour, proprietary technologies, supply and demand.  Though it had elements of rivalry and secrecy, shawlweaving is not an hermetic, miraculous, forgotten process.  The weaving of any type of cloth is a relationship among many parameters of material and structure, understood by experience and established by practice.  Unlike the chemistry of ceramic glazes for example, the attributes of cloth are tantalizingly visible, even in photographs, but they cannot be taken for granted.  Weight of a representative piece of fabric, if it can be meaningfully measured, is just one simple characteristic.  The Kashmir shawl is a complex product, probably driven as much by what its patrons demanded as what its weavers could achieve.  As one of the most successful stories of craft in the service of art, there is much that needs to be understood and appreciated about it.

……………………
Footnotes:
(1) Cohen, Steven, ed., “Kashmir Shawls: The Tapi Collection”, The Shoestring Publisher, Mumbai, 2012
(2) Moorcroft, William, and George Trebeck, “Travels in the Himalayan Provinces of Hindustan and the Panjab from 1819 to 1825”, (H. H. Wilson, ed., London, 1841) Oxford University Press, Karachi, 1979
(3) Beardsley, Grace and Carla Sinopoli, “Wrapped in Beauty: The Koelz Collection of Kashmir Shawls”, University of Michigan, Ann Arbor, 2005
(4) Watson, J. Forbes and John William Kaye, eds., “The People of India….”, India Museum, London 1868-1875; note by PH: this weight is incredibly heavy – even my full-size, plain long shawl of coarse sheep’s wool from Pakistan is only a couple of pounds – 1125 gm. / (1.4 x 2.6 = 3.64 sq.m.) = 309 gpsm
(5) Moorcroft, op. cit., vol. 2, p.168 ff.
(6) Leitner, “An Account of Shawl Weaving…from Linguistic Fragments Discovered in 1870, 1872, and 1879”, Lahore, 1882
(7) Rizvi, Janet, “Woven Masterpieces of Sikh Heritage”, Marg, Vol. 63, No. 2, December 2011; Ames, Frank, “Woven Masterpieces of Sikh Heritage”, Antique Collecters’ Club, Woodbridge UK, 2010
(8) Rizvi, Janet, “Kashmir Shawls: The Tapi Collection”, Marg, Vol. 64, No. 4, June 2013
(9) Pauly, Sarah Buie, “The Kashmir Shawl”, Yale University Art Gallery, New Haven, 1975

Posted in Kashmir shawl weaving

I had two sorts of problems using the toy piano loom for the first time, one simply due to the selection of inappropriately-sized threads, the other suggesting a modification of the loom itself.  In spite of its overall small size, It was not difficult threading the heddles with a fine crochet hook – the knots forming the heddle eyes tended to hold the eyes open.  As suspected, the pipe clamps holding the front and back rollers need some running assistance to improve their grip, or else a complete redesign.  The open-sided rollers allow the selvedge warps to slide off the hump to become too tight or too loose, depending on when the slippage occurs.  When weaving got under way, I found I badly needed an apron area below the fell line to park my accumulating kani/bobbins between uses – I improvised one out of cardboard.

I was disappointed to find that the warp I wanted to use was too heavy.  It was a British-spun tussah silk single, the finest thread in my store of mostly thick tapestry yarns.  It would have been more traditional, and more size-appropriate to use some of the same hand-spun pashmina two-ply that I had for the weft, but I thought it wouldn’t be nearly as strong, or as easy to distinguish the raised pairs (nals) of warps I needed to count for my tapestry imagery.  I wanted 400 warp ends – 100 nals or pixels of my design – in the available width of my loom, about 5 inches at 80 ends per inch (a kani-shawl standard) and the full width of the pattern-repeat I wanted to use.  I found that my initial heading was weaving up at a steep 40 picks per inch.  By removing 10 groups of 4 threads across the width, I reduced the warp to 90 nals, about 72 ends per inch, but found the heading still weaving up at less than 50 picks per inch.  I was reluctant to remove more warps in search of a balanced weave, because too much of the pattern-repeat would be missing.  As the weaving proceeded I had to look out for warps sticking to each other, but with the extra slack of tapestry wefts I was able to weave at about 50 picks per inch.
The second problem was one built into the loom: I had tied the string heddles symmetrically – the eyes midway between the supporting bars – without considering where they would sit in relation to the passage of the warps from back to front beam.  It turned out the warps were deflected downward slightly.  One disadvantage was that the side panels of the loom blocked more than necessary a clear view or access to the shed opening.  Another was that because of the interaction of the counterbalance harness, warps on one shaft would become more slack when others were raised.  My anticipated fix for this is that the vertical arms of the counterbalance harness can be shortened 1/4 inch, or the point where it attaches to the side panel can be raised the same amount.  Both methods would have the effect of raising the whole harness assembly a tolerable amount, without disturbing other features.

Nevertheless, it was possible to proceed with my woven sample, with extra care but no backtracking or significant lost time compared with the inherently slow pace of twill-tapestry.  In a month of steady half-days of weaving, I completed 160 talim lines, 320 picks on 360 warp ends.

The finished sample flattened out in width off the loom, and is approximately balanced-weave (width 5 3/4 inches = 63 e.p.i.; length 5 1/4 inches = 61 p.p.i.).  Close inspection showed that seeming irregularities in the weave structure I had seen in photos of antique fabrics could just be due to differences and unevenness in handspun wefts.  The resulting sample was useful to show, and received more recognition than I expected, simply for its fine threads.  I thought it was a barely workmanlike effort, but showed improvement as the weaving advanced.  It provides a demonstration of the fluid, delicate line that can be achieved with two-picks-per-talim-line (2PPTL), and the beginning of a useful field-repeat pattern.  This same design, which is from an old traditional talim, is now being used for a sample woven at the School of Designs, Srinagar, to try to prove the merits of returning to 2PPTL practice.
The loom worked, but I think sheer portability is its only advantage – more elbow room and better working conditions would be worth more in any situation except where I want to be able to demonstrate the actual weaving process on location.

Posted in Kashmir shawl weaving