peter harris, tapestryweaver

Recently I was asked to scale up the design of an antique Kashmir shawl motif, to make it more visible and effective in modern shawl production.  The original motif, represented by 100 talim lines (each woven in 2 picks) by 33 nals, at a threadcount of 80 threads per inch, would result in a woven image about 1.7 by 2.5 inches.  Doubling the size of the design at the same threadcount would produce an image 3.3 by 5 inches.  My undertaking throughout this exercise was to make the design no more detailed, requiring no more bobbins (“kani”) than before to do the weaving.

I realized that this proposal provides an opportunity to compare the different strategies, old and new, for getting the most pleasing woven image of a design visualised on a square grid.  In recent years, a carpet design program based on the square grid has been widely adopted for preparing shawl designs and the row-by-row weaving instructions (“talim”).  In a hand-knotted carpet, each unit of the grid represents one knot or tuft, and the knots line up not only in horizontal rows but also in vertical columns, one directly above another, so the square grid provides a detailed and properly aligned representation of the design.

The twill fabric structure of shawlweaving consists of weave-units of 4 warp threads and 4 weft threads.  At each treadling 2 adjacent warp threads (called a “nal”) are raised, two lowered, and 1 weft thread (or pick) is passed between.  The raised warp threads may be named in this sequence as: 1 and 4, 1 and 2, 2 and 3, 3 and 4, and then the sequence is repeated.  While it is true that the whole weave-units, each consisting of 4 by 4 threads, align in straight rows and columns, so they can be represented by the square grid, they are different in two important respects from the single carpet knot.  Each weave-unit shows 4 stitches (or “seeds”, a tapestryweaving term) of the weft thread, in the colours of the design, and these 4 stitches line up in a diagonal row.  So each weave-unit is more detailed, and has a peculiar diagonal orientation, that the square grid does not show.

Important note: In my illustrations, each row of “brick-grid” units represents two picks of weaving, so two rows of “brick-grid” units represent one row of 4-by-4-thread weave units, and give a partial impression of the diagonal shift of weft stitches.  The straight-twill diagonal rises to the right on the back of the fabric as it is woven, but to the left on the front of the fabric, as shown.

As I said, the antique shawl fabric motif I am using for this article, like many others, shows 2 picks of weaving for each different line of instructions.  To double it in size, I would now be weaving 4 picks for each line of instructions (and across 2 nals for every 1 nal previously), just as if I were following instructions from a square-grid design like shawlweavers today.

Because the 4 weft stitches in each weave-unit follow the diagonal orientation, my woven motif will have a slanted, jagged appearance.

To improve on this, many shawlweavers today weave the first two picks of each line of instructions (left-to-right, then right-to-left), then subtract 1 nal from the beginning of the line before repeating the same instructions for the 3rd and 4th picks.  This is an automated approach to counter the diagonal orientation and place the 3rd and 4th stitches more directly above the first 2 stitches in each weave unit.  And already, the line of instructions for the 3rd and 4th picks of a weave-unit is different from the line for the 1st and 2nd picks, but only slightly.

This previously published illustration (June 2012) shows every thread of a motif detail: first as if woven following square-grid instructions, on the left, each 4 picks by the same instructions; at center, one nal subtracted after the first 2 picks; then on the right, following the edited version of brick-grid units, as it appeared in the original 17th c. shawl.

A major improvement in the appearance of the woven motif will result if the line of instructions for the 3rd and 4th picks of the weave-unit is carefully edited for the best placement of those stitches, even without introducing new colours or bobbins.  This process takes place at the design stage, and is based on representing each row of a square-grid design by two rows of brick-grid units that show the diagonal progression of the twill fabric stitches.  Each brick-grid unit, representing two stitches, can be assigned the same colour as an adjoining unit in the same row, either right or left, to produce a clearer design or smoother outline, without requiring more bobbins.  It is true that the talim line of instructions for the 3rd and 4th picks is different from the line for the 1st and 2nd picks.

While this basic process of editing a design will produce improved, clearer lines and edges, it can lead to further refinements.  In my example I tried to retain the same weights of lines and shapes as in the original design.  Lines can be made finer and more delicate, without becoming broken-up and spotty.  Treating each talim line of instructions for two picks of weaving as an opportunity to make the design more clear and precise, is not a new idea.  Until recent years, it was the normal practice.  Shawl talim showed “half” units at the ends of every second line, to relate the talim line for picks 1 and 2, to the talim line for picks 3 and 4.  How to write shawl talim with these half-units, whether by hand or by computer, and relate these instructions to the warp threads as they appear treadled on the loom, is an article for another time.  Productive and convenient as it is to use CAD software for shawl designs, a program that provides only the square grid is not adequate.

It has been impressed on me that traditional shawl weavers are poorly paid for their long hours of work, and reluctant to consider any changes in their practice that could slow their production even more.  I have argued that the amount of weaving is the same, only the number of talim lines of instructions is more.  Admittedly, that leaves a gap where the weaver may have to refer more often to the talim for the instructions he is following, and where he gets less benefit from the unconscious memorization that comes from repeating the same instructions over and over.  Still, if the motif or pattern is repeated a dozen times across the width of the cloth, the memorization is still there.
Some shawlweavers and dealers acknowledge and seem resigned to the poorer rendition of their designs that comes from the current practice of 4 picks per talim line.  To me it seems a tragic absurdity, doomed to be abandoned, to devote so much effort to working at the limits of the fineness of threads and eyesight, and yet follow a method that produces coarse results.  It is up to dealers to recognize and pay more for superior workmanship.  Weavers must be able to take pride in their own work, to uphold the pride they have in the tradition of the Kashmir shawl.

Posted in Kashmir shawl weaving

Two aspects of the constraints I’ve heard mentioned regularly already that kani shawl weavers work and live under, their basic low rate of pay and the competition for the unknowing buyer of look-alike goods made by cheaper technologies, were given an impassioned reading for me tonight.
My perspective as a student of the antique, museum-quality artifacts featured in coffee-table books, has been to advocate the restoration of current design and weaving practices to their historical norms, one way or another connected to following a line of talim instructions for two picks of weaving instead of four.  The revival of these methods can be used both to reconstruct designs from antique fabrics and to create new designs.
Kani shawls have always been expensive, luxury goods, and quality needs to be evident to the discerning eye, to support not only the price of that piece but also the reputation of the craft.  Historians have conjectured that the decline in the fashionability of the “paisley” shawl in the late 19th century was partly due to the proliferation of cheaper versions.  The craft has been pronounced dead by authoritative commentators at intervals ever since.  A further macabre twist to this pessimism is that the recent decades of uprising and repression in Kashmir were good for the preservation of traditional crafts, because the workers were kept at home by disruption and shrinkage of the local economy, unemployment, and curfews.  To the extent that the economy has reopened, workers are shifting to other opportunities.
Now I’m told that the practice of two picks per talim line survived till as recently as the mid-1990’s, and two days before I saw dated samplers from that time in the Craft Museum of the School of Designs that showed some evidence of two picks per talim line.  Maybe the shift to four picks has more to do than I thought, with the introduction of CAD programs for carpet design based on the square grid.
Two years ago I was told that a shawl weaver might expect to receive as little as 4000 rupees a month, about $80.  To receive 6 or 8 thousand was a sign of the weaver’s superiority and the employer’s recognition.  In light of inflation and devaluation of the rupee, what are the numbers today?  Not only is the pay low, but there is a tough winter season to get through that’s only heard-about by the visitors in the temperate summer months, and a lack of social safety net.  A serious illness or family emergency can be a financial disaster, diverting children’s school fees to necessary but futile treatments and unreliable medicines.  Add to this the still uncertain security of the military occupation, the knock on the door in the middle of the night, the not knowing when or whether someone gone out and overdue, will come back.  Many times in the past two weeks I’ve stood in front of casually levelled automatic weapons, reminded of the judge in the recent terrorist incident in Islamabad who, the story goes, was accidentally shot by his own startled bodyguard.
In these circumstances, even if I argue I’m not trying to make shawl designs more detailed or difficult, the full-time weaver, trying to maintain his economic foothold, will object to spending twice as much time reading talim instructions, and getting no benefit from having half-memorized the sequences of steps he’s worked in two picks already.  If the difference in the quality of rendering the design is recognizable it should be worth more, but that may only happen at several removes from when the weaver got paid.  To demonstrate the value of two picks instead of four, I should be editing acceptable contemporary designs, instead of trying to reimpose the old elaborate ones.
I don’t know the terms under which the weavers work.  They should be paid on a time basis instead of piecework.  It should be the dealers commissioning difficult designs who take the economic chances.  Buyers need to recognize the difference and intrinsic value of the goods they are considering – this is not likely to happen with passers-by lured into souvenir shops – at best they will be shopping for price.  The Indian market, knowledgeable and intimate with the living tradition, perhaps not so preoccupied with antiquarian value as the West, is probably more discerning and reliable.  Maybe the biggest part of my benefit to the shawlweaving craft has been thrust on me by the ready audience for my popularizing articles and talks in the West.

Posted in Kashmir shawl weaving

Looking, or imagining you’re looking, at a design drafted on a brick grid instead of the more common square grid, raises the question, what do you do with the half “bricks” found at the ends of every second row? It’s a more important issue than just the half-bricks themselves, because they determine the position of the rest of the bricks in the row and their design information, in relation to the rows above and below.

But this “problem”, or quirk of visualization, would not have arisen historically in the talim system of designing for Kashmiri twill-tapestry shawls, if it didn’t correspond to a feature of the traditional weave structure, 2/2 twill. In this structure, each weave unit consists of four warp threads and four weft threads. Four weft threads tells you that each cycle of the weaving requires four treadlings. In each treadling two adjacent warp threads are raised, becoming a visible pair or “nal”, and the other two lowered and basically disregarded. In the next treadling, the two warp threads that are raised shifts by one thread to the right, and so on through the cycle of four treadlings, so in the fifth treadling the same two warp threads are raised, as in the first treadling. It will be observed that in one, and only one, of the four treadlings, a single warp thread is raised at the left-hand edge of the cloth. Voilà, the “half” brick.
In the old talim system, each line of instructions is followed for two picks of weaving, two complete rows of weft threads, two treadlings. The count of nals in each line is the same, so for the lines which include half-nals, the count is completed by adding 1/2 at the beginning and end to the number of whole nals between. My practice is to make the treadling where the single warp thread appears on the left, the first pick worked according to the talim line of instructions that begins with a half-nal. In this pick, all wefts are moving from left to right, and at the right-hand side the other single warp thread will be found, completing the count as expected. At the next treadling, because the pairs of raised warp threads are advancing to the right, the single raised warp thread at the right-hand side disappears. Begin inserting the weft threads from right to left, subtracting the missing half-nal. When you reach the left-hand side you will find the missing warp thread has reappeared to form a pair with the single warp on the left, providing a count one whole nal more than expected – that is, instead of 3 1/2 when you started that line of instructions, for example, it now looks like four. This is correct.
Talim lines beginning with whole nals proceed exactly as written, in both directions. It is only in the second treadling of the half-nal talim lines, that there is potential for confusion about a mistake in counting.

The Generate Text Summary function of my Stitch Painter program is not sophisticated enough to put in half-unit counts at the ends of the appropriate lines, so I select the grid area to make sure that the odd-numbered lines of the text summary will have counts totaling one nal less than the even-numbered lines, and I remember to add the half-nals mentally to the odd-numbered lines when weaving.

Posted in Kashmir shawl weaving