Tapestry Project

I wanted to experiment with a knotting technique used in Turkey for making carpets. I also considered the Persian knot, but the Turkish knot looks like it has better integrity. Click on any of the photos to enlarge for better detail.

In working with a small rigid heddle loom on a scarf, I was able to chain together the ghiordes knots using a stick as a gauge for keeping the pile a consistent length. Once each pick (horizontal row) was completed, I cut the pile off the stick to form a pick of individual ghiordes knots. The same knot is used in Scandinavia to make rya rugs and a sailor would recognize it as a lark’s head knot.

I learned several things from this experiment. There is a very good reason rug makers in the middle-east secure each pick of knots with a couple of passes of plain weave. In my experiment, I staggered each pick so that the pair of warps used in one pick were split in the next pick and paired with the warps on either side. This served to keep the warps bound together, but didn’t keep the knots from pulling out with little effort so the scarf tends to shed. I resolved this in the tapestry as described below.

For my tapestry experiment, I wanted to use my interpretation of the ghiordes knot to make a series of chained knots but leave them intact so that instead of individual short lengths of fiber knotted into the fabric, it would be one continuous chain with looping pile. Some simple experiments gave me the basic formula of two staggered picks of chained knots secured in place by a double pick (across and back) of twining weave. This results in a beautifully full pile locked in place very thoroughly.

A side benefit I didn’t realize until I was well into the project is that the back side of the tapestry has a nice finished look to it as well.

I had never done a completely weft-faced weaving, so I set up a very small test to be sure the technique would work with the materials I had chosen.





Because I was unsure of how well some of these techniques I was experimenting with would work and not wanting to add too much complexity to it, I chose a Texas flag design so that I could do some color transitions, but nothing too complicated. In retrospect, I think that the basic technique is time consuming enough that complex patterns wouldn’t add any appreciable trouble to the project. In practice, the warps are much closer together than the knot illustrations above. Estimating the yarn required for the weft is very difficult. What I ended up doing was tie a few inches of tapestry using my gauge and the intended yarn.

I wanted my tapestry to have four selvedges (finished edges) without the normal fringe found at the top and bottom of most tapestry weavings, so I needed to build a special loom. The Texas flag is two units tall by three units wide and I chose to rotate this and make my tapestry twenty inches wide by thirty inches tall. A trip to the local hobby and craft store scored a set of canvas stretchers in 22” and 32” lengths to make my basic frame.

To accommodate selvedges at the top and bottom, I needed a row of nails on each end to secure the warp. Small scale experiments with the braided nylon plumber’s cord I chose to use for the warp showed an eight dent warp would be a good place to start. Since the warp would wrap around each nail and be doubled on the loom, this meant a nail spacing of ¼ inch. To avoid layout marks on the finished loom, I covered the face with blue painters tape to mark for pre-drilling the nail holes. After drilling and removing the tape, I tapped in the nails and began warping.



I was not sure exactly how to proceed with the design, I simply took a black magic marker and drew the basic layout on the warp threads. As you can see below, this wasn’t a great idea. As weaving progressed, the warps were pulled unevenly by the tight knotting and this resulted in as much as a quarter of an inch in deviation for what was initially a clean horizontal line.

Each side of the weaving is finished with three or four warps of twined weaving because the ghiordes knots are not really secure enough to be used on the edges.





Once I got started, the weft yarns kept getting hung up on the little nails at the bottom of the frame, so I covered them with blue painters tape. The rows of nails at the top of the frame make a convenient place to “hang” lengths of yarn for later use.

Also visible in the photo to the right is the dowel rod used to gauge a consistent pile length. The dowel was rounded and polished on each end so that the weft could be slipped over the end while working. It was also sanded smooth and polished by rubbing it down with a scrap of leather. I thought any kind of finish on it would be likely to transfer to the fibers.

I needed something to beat the weft down as I went along. The first pick was done using the needle to beat between each weft. The wood I used for all my beaters was poplar because it is the only variety the local home store sells in hardwood less than ¾ inch thickness. The first version of my beater was carved from a single piece of poplar. It works fine, but it is difficult to shape attractive teeth this way.

My second beater was overly complicated and still didn’t look as nice as I wanted. The teeth are white 3/16th inch birch dowel stained black with vinegroon dye I make by dissolving steel wool in vinegar. This results in a more or less clear liquid that reacts with the tannin in the wood (also works with leather). The resulting stain is actually a part of the wood and so it will not leach out or transfer to the fibers I am working with. The teeth were individually shaped in a lathe, staining them and finally burnishing them with a scrap of leather so they will not snag the fibers. The body is 3/8 inch thick and the blade at the top is a 3/16th inch thick piece of the same wood glued into a slot cut in the body. The blade is only a separate piece from the body because I thought it might look good to stain it with vinegroon as I did with the teeth. Later, trying to touch it up, I got a bit on the body (visible in the upper left-hand corner of the photo at right). The important thing about the blade end is that it needs to taper to a rounded edge and be rounded on both ends in order to prevent snagging on fibers. This blade is used to beat the ghiordes knots down behind the dowel rod where teeth would not be effective.

By the time I made the third beater, I had put almost a month into learning to tie the knots on the tapestry quickly and consistently. It gave me a much better idea of the proportions and shape of my ideal beater.

This one uses a poplar body with integral blade while the teeth are birch dowels shaped by hand in the lathe. The only finish used on this was to burnish it with candle wax.

I work with weft pieces that are about 6 yards long, so there are an average of 2 knots for every 3/16" of the 30" height of the tapestry. I have experimented with various methods of joining two yarns. Some work really well, others are a complete failure for me. I have never been able to get the Russian join to work for me. The knot I have settled on is one I learned from a mountain climber on the side of Enchanted Rock in the Texas Hill Country back in 1974. He called it a barrel knot. Essentially you make an overhand knot on the ends of each of the yarns you want to join, trapping the other yarn inside the knot. Once each knot is pulled tight, pull the two pieces of yarn and the two knots snug against each other and it will not come apart. You can trim the loose ends very close to the knot because it is not going to slip. Don't use this knot where you might need to untie it later, because it is exceedingly difficult in rope and likely impossible with yarns.

Below is a shot of the work in progress. Based upon counts as I have progressed to this point, once completed this 20" X 30" tapestry will have just over 25,000 knots and around 13,000 twined weaving "stitches".

Because of the way the knots and twined weaving are done, each pick of weft must be completed as a whole before going to the next pick. Other tapestry techniques allow working within an area of a given color and going back to fill in around it. Here I have five weft needles at the same time even though there are only two colors being used. This is because the colors must be contiguous within a pick or the rug will not have good physical integrity. The only part that runs straight across is the binding double pick (across and back) of twined weaving after each set of knots.

Now that I am working with the star, you can more easily see the twined weaving that 'binds' each set of knots in place at the top of the partially completed star below. Also, as noted earlier, uneven pulling of warps during knotting has resulted in the lines on the warps shifting. This wasn't a real problem until I got to the star. As you can see from the photo, I got around this problem by hanging a paper drawing of the star behind the warp. This paper sketch behind the warps is common in tapestry weaving and is called a cartoon.

One thing I noticed once the color breaks shifted from the vertical line of the red and white to the diagonal lines of the star, I realized that it is necessary to see exactly where on the warp, your current pick is. This would not be a problem on a regular tapestry, but with the dowel I am using as a gauge for pile length, this became impractical.

I remedied the problem by taking a short scrap of yarn for each color change on a given line and hooking it into the warp where the color change occurs, then sliding it up out of the way so it will not interfere with the weaving.

As I get closer to the end of the project, many of the things I worried about have resolved themselves as I went along. I delayed beginning the project for a long time because I was unsure how some of my ideas would work together. My advice to someone putting off a project because they have not worked out all the details is dive in. Get an idea of how the first few steps should proceed and when a problem arises, don't push it. Take a step back and just think about it for a week or so if necessary. Don't get into a hurry, and if it doesn't look right, undo it to the point where it did and sort out the problem. I spent many hours undoing hundreds of knots. In (soon to be) retrospect I do not regret removing those problems even though at the time I thought I might just set fire to the entire project in frustration.

I never anticipated a problem at the top of the weaving until I approached it. The closer I get to the pins at the top of the loom, the more difficult it is to work the needles in and out of the warp. The solution came to me as I worked and now I have a dowel inserted behind the warps to lift them higher on the pins as you can see below.

One other possible issue I see is the free end of the warp cord. It was tied around the starting pin as a loop. As I approached it, I looped it around the first and second pins so that I can weave one side of the loop as the selvedge warp that it is while the other side of the loop will be woven with the warp next to the selvedge warp and the running (loose) end of the cord will be woven in with the selvedge warp. I have no idea whether this will help keep things together once it is off the loom or not, but the whole process is an experiment. The other end of the warp received a similar treatment at the beginning of the project, but I did not think to get a photo.

As I approach the end of the project, I am trying to fix things that probably don't need fixing. I was attempting to tidy up the knot at the end of the warp when the whole thing slipped apart. I tried a couple of methods of tying it back together but all of the things I came up with for knotting it would have resulted in a knot that might show up on the back side. At the corner of the project, I just couldn't face that after so many hours. My solution was to sew the running (loose) end of the warp cord alongside the second warp. No bulge at all and perhaps not too much bulk.

Completed Tapestry Below