Stephen Selby recommended the article in Wenwu, Vol. 1, 2000 on an excavation of a tomb at Niya in Xinjiang to me some time ago and I only recently saw a copy. The article described a burial of the Han to Jin dynasties, in which the bodies and artifacts were remarkable well preserved (the title of the English summary is "Excavation of Tomb Coded M8 of Cemetery 95MNI at the Niya Site in Xinjiang", The Institute of Archaeology Xinjiang, pp 4-40).

The bow is both drawn and photographed with its arrows and bow case/quiver. In the 1500 to 2300 years since it was buried, the strung bow has reversed and the quiver has been damaged by insects. The whole assembly resembles Illustration 131 in "Archaeological Treasures of the Silk Road in the Xinjiang Uygur Autonomous Region" (1998). In both cases, four arrows were in the quiver and some were fitted with wooden blunts.

While very informative, the quality of the drawing is not up to the usual level of Wenwu, but there does not be a standard way to depict bows. They are usually drawn in random positions with little reference to their structure and measurements. This article is no different. The bow is described as being 132 cm long, the arrows as being from 80 to 81 cm long. I will leave out the bow case/quiver for a later discussion. There was a bracer in the tomb.

Luckily , there is a photograph taken of the bow and arrows together in profile. Both bows in the abovementioned Xinjiang book are photographed at unusual angles, which would require altogether too much mathematics to extract proportions (and too much guessing). This bow, however, was quite easily analysed. I rapidly came to the conclusion that the "length" of the bow quoted in the article was a point-to-point measurement, not the actual length of the object measured along its edge. By comparing the length of the straightest arrow in the photo with the distance between the tips of the bow, I came up with a ratio of the photograph to the real objects. Using a program to calculate the lengths of curved and irregular lines, I derived a set of proportions. They follow.

I have rounded the inches, because there is no way that the precision of the metric figures could be justified. The bowstring probably stretched over the centuries. The arrow length was used to generate the ratio. The 5 mm difference between the length of the bow in the article is reasonable considering how the information was derived. The arbitrary nature of the beginnings and endings of sections limits the usefulness of these figures. I have taken the limbs to be the same length as the horn belly, which appears to run from the angle of the siyah to the angle on the belly of the grip. This means that at the grip end there is a non-bending portion of the limb, which might more properly considered part of the grip. The same is true of the siyah end. I believe the length of the horn is a useful piece of information and so have measured this way. The main purpose of these figures is to help planning a reconstruction of the bow. The cross sections of the bow can only be guessed at this stage. The grip appears narrow and deep, the limbs flattened and the siyahs narrow. The bow from tomb 3 (see reference above to Illustration 131) has limbs that are broader in the middle than at the ends where they join the grip and siyahs respectively.

I have even more subjective measurements for the angles of the parts of the bow. I had to guess the tangents to the curves, so there is even more chance of error. The precision of the figures does not represent the precision of the measurements and they can be rounded up with no problem. They are as follows.

The first angle is meant to give an idea of the curvature of the siyah. It is concave towards the back of the bow. I do not feel confident that the real curvature of the siyah is this great and I think this may be an artifact of the measurement system. As I mentioned before this is highly speculative. The photograph is not that clear and the extreme tips appear damaged. The other angles are more reasonable. A strip of cloth wraps around the upper limb in a spiral, obscuring its exact contour.

The text of the article implies that the arrows are barrelled. The bracer suggests the archer did not use a thumb ring. Since instances of the Sassanian two-fingered draw are found in murals in Dunhuang to the east of Niya (see illustration 43 in "The Art Treasures of Dunhuang", 1981, for an example), I think it reasonable to assume that the Sassanian release was used with this bow, even though it could equally have been used with a thumb ring had the archer so chosen. The twin quiver/bow case combination is known from carvings in Palmyra (see "Roman Archery Equipment", J. C. Coulston, 1985, Fig. 33) and bone buckles from the Orlat necropolis in Uzbekistan (see "Heirs to the Silk Road Uzbekistan", Johannes Kalter and Margareta Pavaloi, 1997, Thames and Hudson, page 37, Ill. 41-42). The bone buckles clearly show the soft nature of the bowcase and the uneven heights of the quiver tubes. The forward tube is longer with a cap, just like the one in Wenwu and in the abovementioned Xinjiang book. The Wenwu bowcase appears to have been of woollen cloth while the Xinjiang book quotes leather as the material. The Wenwu example appears to have a leather reinforcement at the base to take the siyah of the bow.

The Sassanians in Iran did not use this type of bowcase/quiver, though they did use a full length quiver worn vertically on the right side. In the Journal of Archer-Antiquaries ("The Sassanids", Vol. 12, 1969, pp 29-32), W. F. Paterson derived the length of the component parts of a Sassanian bow from pictures of archers on various objects. His figures are:

The Paterson estimate has shorter siyahs than the Niya bow and does not really agree with in other ways. However, both bows have upper siyahs longer than the lower ones and upper limbs longer than the lower ones. I believe that the larger selection of Sassanian silver plates with archers depicted on them, available now, would support both a longer bow and slightly longer siyahs.

If the release was Sassanian, with the forefinger extended, my experiments show that the arrows may have been shot from either side of the bow because the forefinger can steady the arrow against the right side of the grip with little effect on the accuracy the shot. I have no evidence that this was ever done. I think the position of the middle finger and ring finger on the string below the arrow allowed the fletching to come right up to the nock. This was an advantage of the thumb ring too. The forefinger could have simply been raised out of the way and the arrow shot off the left hand side of the bow, but I prefer the security of the forefinger on the arrow and the arrow on the right side of the bow. This could be simply the prejudice of someone who has shot with a thumb ring for decades. It is also true that a thumb lock with the forefinger and the little finger extended looks like a Sassanian release from a distance, but it is unlikely that it was ever used.

The bows from Niya contain a great deal of information that could be extracted by careful measurement and plans drawn up from the original objects. I think these weapons are very historically significant for bow design. They are obviously related to later Magyar bows, which are probably only a subset of a Sino-Iranian design continuum. Whether the eared bow was developed at the eastern end, the western end or in the middle is hard to tell at the moment. However, if these Xinjiang bows were dated more precisely and their origins more closely identified, they may provide some clues.