Reconstruction of the tablet-woven band E32036 from Louvre
As soon as I discovered the huge collection of tablet-woven bands in Museé du Louvre, which was recently made public online, I knew that I would have plenty of interesting things to work on in next few weeks (maybe months).
As I wanted to start with something simple, my first choice was the fragment E32036. Beside the simple design, it was the only one with the photos of the front and back, which I hoped would make the reconstruction easier for me. If I only knew, how much time and failed trials this seemingly simple pattern would take me...
Trial 1: 8 tablets, each with four threads (2 with all threads blue, 6 with one white and three blue threads)
Luckily, I did not underestimate the pattern that much, to embark immediately on a fine wool and linen reconstruction, and I used cotton threads for my first trial. At the beginning, I assumed that this was a typical diamond pattern, with the "tabs" on the sides caused by the opposite threading of the second tablet from the centre. The obverse matched quite well, but on the reverse, the light threads formed an X in the centre, instead of resembling a TIE starfighter from Star Wars. That meant that the second tablet from the centre had to be threaded in the same direction as the central one (on both sides). And lo! - if I beat the weft really firmly, the sides of the diamond are not smooth and the pattern is close to the original band!
For a moment I thought the mystery is solved, but then I carefully compared my band with the photo on the museum website and found out that there are two blue threads between the white thread at the edge of the band and the weft showing in the middle between the diamonds in the original band, while three in my version. This could have meant two things: Either this section of the band (both selvedges are missing) was made on six tablets instead of eight, as I originally assumed, or some of their holes were empty. However, the first option did not seem to me likely due to the wide dark gap between the central rhombus and the white wavy line on the sides (only on one side in the original band, but it was easier to work on the symmetrical band while looking for the right pattern).
Trials 2 and 3: 8 tablets with two threads only (2 with both threads blue, 6 with one white, or yellow, and one blue thread) (+ border tablets!)
The second attempt was a waste of time, with a lot of unnecessary complications. At first I forgot that it is (with some exceptions) impossible to weave with two-hole tablets without four-hole borders and had to add them. Then, after having struggled to distinguish the crucial details in the 6mm narrow ribbon, I decided to use a knitting yarn for further experiments. The third trial woven in thick wool confirmed my suspicions gathered while squinting at the cotton micro-version. Firstly, the light yarn was visible on the surface where it should not be, and secondly, while the threads are 90° off in the first, second and fourth tablets from the centre, the third tablet (the one with both threads blue) must have the threads in the same position as one of the adjacent tablets. I deliberately put the third tablet on either side in different positions to compare what the results would look like:
Trial 4: 6 tablets, all with three blue and one light threads (+ borders)
I did not succeed in having the jagged sides of the diamonds since the first trial, even if I beat the weft as firmly as possible.
After thinking for a while, I took a needle and dissected the blue thread next to the yellow one from the finished sample. Behold! The mystery is solved! (Removing the edge threads yielded no results compared to weaving without the edge tablets.)
Trial 5: Tablets as before, except that there were only one light and two blue threads in the second tablet from the centre
I wove a short sample in this way to make sure I had removed the correct thread from the second tablet, and then I started thinking: Is it possible that only the second tablet from the centre contained three threads, while other tablets contained four? There are only two blue threads visible above the weft in the central tablets. What would happen if I remove the last thread?
At first glance, nothing has changed. But on closer inspection, in the new version (in the left half of the photo) the blue thread adjacent to the weft visible between the diamonds (which look more like hexagons now) is shorter than the one closer to the edge, while before removal both were the same size (as they are also in the original band). The change is more obvious on the reverse side. Where one pick of weft was visible before the thread was removed from the central tablet, there are now two picks visible. The photo of the reverse side of the original band is very dark and of poor quality, but nevertheless, it seems to me that there is only one pick visible. This examination of the reverse also drew my attention to the fact that the light dots are evenly spaced across the entire width in my sample, while in the original band there is a gap between them, which originally led me to believe that the band had been woven on eight tablets. I can't imagine that removing another blue thread will widen this gap. Moreover, which thread should it be? The threading of the central tablet in this last version was ABC- (the empty hole was opposite the yellow thread) and A-CD in the next tablet (the empty hole was adjacent to the yellow thread). Therefore I have no clue, which thread should be removed from the last tablet. There seems to be no logic in it.
I would have never believed that such a simple pattern can give me such troubles. If there was only the photo of the front side available, I would believe I have the correct pattern now, but photo of the reverse side tells me, I don't have it. After more than ten hours of work, I admitted a defeat.
...
Few days later, however, I decided to continue.
Trial 6: 6 triangular tablets, each with two blue and one yellow threads (+ square tablets forming selvedges)
I hoped that by using triangular tablets I would avoid the question I couldn't answer at the end of the fifth attempt (which thread to remove from the edge tablet). Eventually, this experiment split into three sub-trials.
In the left part of the picture, the tips of triangular tablets pointed downwards, so that two warp threads ran above the weft and one below it; in the middle and on the right, their tips pointed upwards, so that one warp thread ran above the weft and two below it. In the left section and in the centre, the tablets were rotated a full circle before changing direction, similarly to the previous experiments with square cards. However, in the case of the triangular tablets, it meant only three turns, and as a result, the yellow diamonds are immediately next to each other, not separated by a blue space. When tablets were turned 4 times forward and 4 times backward (the right part of the photo), the diamonds and ornaments on the reverse looked the same as in the original band, but due to the more than 360 ° rotation, the yellow dots on the sides between the diamonds changed into small triangles. All this has proved that the original band was not woven on triangular tablets, but on square tablets with 4F4B turning. The shape of diamonds also confirmed the necessity for an empty hole next to the light thread in both central but one tablets to create hexagons/diamonds with "tabs" (whichever you want to call them).
Trial 7: 8 tablets (4 containing 4 threads and 4 containing 2 threads )
I continued to ponder the difference between the reverse sides of the original band and my copies. Also, it seemed unlikely to me that there was only one tablet with three threads (in either half of the band) among all the other tablets with four threads. A combination of tablets with two and four threads seemed more likely to me. That's why I decided to replace the tablet containing three threads with two tablets containing two threads each. The tablet closer to the centre was threaded with one yellow and one blue thread, and the tablet closer to the edge contained both threads blue. Trial 1 with eight tablets was the only one, in which the arrangement of white lines on the reverse resembled the original band. I hoped that by using two tablets with two threads, I would avoid that surplus thread, which was a problem in Trial 1.
The result was very similar to the bands woven with six cards (trials 4 and 5), but I was not able to distinguish small nuances due to different pulling of the weft and due to the fact that I omitted the edge cards this time.
Trial 8: Repetition of trials 7 and 5, but with different yarn
After weaving few centimeters on the tablets strung as in Trial 7, I removed both tablets with two blue threads and threaded one thread of each into the adjacent card with two threads. This returned me to threading from Trial 5, but repeating both trials one after another on the same warp guaranteed that the resulting appearance would not be affected by different pulling of the weft and other factors.
The most significant difference between the two versions is on the reverse side, in the side triangles between the central rhombuses. In the left part of the photo (6 tablets as in Trial 5) there is one blue thread in the middle of these triangles, while on the right (8 tablets as in Trial 7) there are two threads. But even this did not lead to a more pronounced gap between the central five white dots and the pair at the edge, which was observed in the original band. On the obverse, "tabs" are more distinctly formed by a tablet with three threads (left) than by the one with two threads (right). Another difference lies in the wefts between the pattern and selvedge tablets being more pronounced in Trial 5 arrangement. As there are no selvedge cords preserved in the original band, this difference is useless. However, because of the diamonds with distinctive "tabs" on the obverse and one thread in the triangles on the reverse, I am inclined to believe that the version with six tablets (Trial 5) is more likely correct.
Trial 9 - the final reconstruction:
At the beginning of this post, I mentioned that it was easier for me to work with a symmetrical motif, although the white thread in the edge tablet appears on one side of the original band only. After (probable) deciphering the central part of the pattern, it was time to focus on this and other details. Not only is the white thread missing on the other side, but the tablet that should contain it is threaded in the opposite direction. On the same side, approximately in the middle of the fragment, there are short segments of two other cords, which contain white threads, but those are not located symmetrically to the opposite side, but in the same positions as the white threads forming diamonds with "tabs". This means that the whole pattern of the original band could have looked like this:
However, it is also possible that only a small part of much wider band has been preserved. I used brown selvedges not because I think they had this color, but because "archaeological brown" can represent any colour. (However, the weft in the original band is also brown, and therefore choosing this color helped to make selvedges neat. 😉)
The variation in lengths of the motives was caused by use of different wefts. At the beginning (on the left) I used the same thread as for the warp, but even though I beat the weft quite firmly, I could not get the correct proportions of the motives. Therefore I unplied the 2-ply woolen yarn I was using to obtain a thinner weft. The band looked much better with it, but single woolen thread was shredding and breaking too easily. Linen threads I've got available were not thin enough. The weft in the original band was used double, but I could not distinguish, whether it was spun or not. Did Egyptians used horse hair as Central and Nort European weavers did? However, horse hair used double would not be much thinner than threads I already used. What else can I use before resorting to modern syntethic sewing thread?
There is always plenty of loose hair in my surroundings. After having it woven or knitted in so many projects inadvertently, I decided to use it intentionally this time. My hair is quite fine, and therefore two hairs formed too thin weft for the purpose of this reconstruction, but the development in its appearance clearly shows, that the weft (whatever material it should be) distinctly thinner than the warp is essential for this pattern.
After having read about the ammount of work behind deciphering this simple pattern, I hope you now understand, why some authors do not like their patterns, which they published for free, being used by other people for commercial use.