may therefore be regarded as belonging to the end of the first century. Apparently, then, the peculiarly British characteristics of the group had been fully developed at a comparatively early date.

Fig. 8 is without any enamel decoration. Here, as in all the members of the group, the foot is well developed. In the centre of the bow is a circular knob, while the head is trumpet-shaped so as to cover the back of the spring. The wire of the spring is looped through a collar, forming a ring on the top by which the brooch could be fastened to a chain, for the fibulae of this group seem to have been regularly worn in pairs. The spring of such a fibula with part of the chain still attached is seen in Fig. 10. It was found in Pit LXV with coins of the first century. Though smaller in size, Fig. 9 closely resembles Fig. 8, showing as it does the characteristic treatment of the knob on the bow. It was found in the Praetentura, and is probably early. Figs. 11 and 12, the latter imperfect, both appear to be earlier than 100 A.D. They were found on the south side of the fort between the ditches cut for the second occupation, and were possibly dropped before the latter were constructed. Both are characterised by the simple and graceful ornamentation of the trumpet-shaped head.

There are no traces of enamel on Fig. 11, but on Fig. 12 the background has been filled entirely with red. An earlier stage of the pattern they display occurs in the decoration of a silver-gilt brooch found at Birdlip, Gloucestershire, now in the Gloucester Museum. This beautiful example of Late Celtic Art has recently been described by Mr. Reginald Smith,[1] who assigns it to the middle of the first century. As we have already noted in dealing with the horse trappings, there can be little doubt that at Newstead the Late Celtic Art was beginning to lose something of its inventiveness and the charm of its wayward designs. On the Newstead brooch the Birdlip pattern has developed into a much more conventional device. The same pattern is repeated in a still later and more degraded form on the well known pair of silver-gilt fibulae, found near Backworth, Northumberland,[2] with coins showing that the deposit cannot be earlier than the year A.D. 139.

Figs. 13 and 14 were found together beneath the cobbled base of the rampart surrounding the Bath Building. Probably, therefore, they were deposited prior to the reduction in the size of the fort in the second century. If so, we shall be safe in placing them before 150 A.D. In shape they do not differ materially from the two immediately preceding. The treatment

1 Archaeologia, vol. 61, pt. 2, p. 341.

2 Romilly Allen, Celtic Art, p. 104.