lying near the surface, with the exception of Fig. 21, which was taken from one of the outer ditches of the extended fort, and could not therefore be as early as the first occupation. The signs of degradation, to which attention has been drawn, probably indicate that the majority of them should be assigned to the Antonine period. Figs. 19, 21, 22 and 23 are enamelled. The enamel has disappeared from Fig. 20. On Fig. 19, which has had a penannular brooch fastened to it, perhaps to take the place of a broken pin, blue lozenges are inlaid against a red ground, while in Figs. 21 and 22 the bars of colour are alternately blue and brown, though possibly the latter may originally have been red. The excavation of the Lochlee Crannog furnished an excellent example of a brooch belonging to this group found on a purely native site.[1]
The most characteristic feature of Figs. 24 and 25 (Plate LXXXVI.) is the expanded circular ornament on the bow, possibly in itself evolved from the smaller stud to be seen on Figs. 19 and 20. Both are decorated with enamel. 
FIG. 47. FIBULAE FROM ILCHESTER
AND HOD HILLThe expanded foot of Fig. 24 has been filled with small triangular patches of yellow and blue. A similar brooch from Heddernheim is dated by Professor Schumacher to the second half of the second century.[2] The same type occurs at Camelon.
Figs. 26 and 27 (Plate LXXXVII.) obviously go together. Fig. 26 belongs to a class that is not uncommon in England; several specimens found in London are preserved in the British Museum. The hollow foot at the end of the stem appears to be reminiscent of a ring, and in some specimens the ring is actually to be seen. In the subsequent evolution of the brooch, the pierced ornament on the bow, with its spiral-like expansion on either side, becomes solidified, the tradition of the earlier open-work. being preserved in the enamel decoration, as in a specimen from Ilchester, Somersetshire, in the British Museum, illustrated in Fig. 47, No. 1, and also in an imperfect specimen from Camelon.[3] Fig. 27 represents a still later type. The trumpet-shaped head of the fibula has disappeared,
1 Munro, Ancient Scottish Lake Dwellings, p. 130.
2 Mittheilungen über romische Funde in Heddernheim, Heft. ii. Taf. ii. Fig. 17.
3 Proceedings of the Society of Antiquaries of Scotland, vol. xxxv. p. 402, fig. 39.
324