Camelon and on Roman sites in England. Four very good specimens were found at Newstead. Plate LXVIII., Fig. 2, which is from the ditch of the early fort, shows some signs of wear. It is 478 inches long and has nine teeth, and but for the incised lines, which form a triangle at the base of the teeth, it is without decoration. Fig. 1, from Pit XXXVII, is in perfect preservation. It measures 6 inches long and has eight teeth. The handle terminates in a cross-bar which is decorated with a double set of incised diagonal lines. It has a hole in the centre, probably intended for a cord for suspension. Fig. 4, from Pit LIX, has originally been of the same shape. It is 4¾ inches in length. One end of the cross-bar has been broken off. The main part of the handle is divided into two panels by means of double lines incised across it. In each panel are two circles, inside each of which are seven dots. At the upper end is a hole for a cord. Fig. 3, which represents a common type, came from the inner ditch of the East Annexe. The long-handled combs are of common occurrence among the brochs of Northern Scotland, and Dr. Joseph Anderson[1] has shown how they must have been employed to press the woof on to the web,—the teeth being inserted between the threads of the warp,—and has pointed out that a similar implement is still used for this purpose in the East. In England, long-handled combs have come to light, not only in immediate association with Roman relics, but also on sites such as the Lake-village of Glastonbury and the camp at Hunsbury, near Northampton, which appear to belong to a period of pre-Roman civilisation. On the other hand, the long-handled weaving comb is almost unknown in the finds from the German Limes forts. It would, therefore, appear that at Newstead we must class these combs as things belonging to the native population, and associate them with the characteristic fibulae and horse trappings of Late Celtic design.

Other objects suggestive of cloth-making are spindles. Two of them were found in Pit LIV. They are neatly tapered at each end. One measures 8½ inches, the other (Plate LXVIII., Fig. 7) 6¼ inches. Whorls were some­times of sandstone (Plate LXVIII., Figs. 13, 14 and 15) and sometimes of bone (Plate LXVIII., Fig. 12). Objects cut from pieces of red and black ware (Plate LXVIII., Figs. 8 to 11) probably served the same purpose. A small, neatly made saw from Pit XVII, with its handle of deer horn, the whole only 5½ inches long (Plate LXVIII., Fig. 6), is just such a tool as might have been

1 Notes on the Evidence of Spinning and Weaving in the Brochs or Pictish Towers?' Proceedings of the Society of Antiquaries of Scotland, vol. ix. 548.