two iron nails. This is the modern practice, except that copper or galvanized nails are used. Slate roofing occurs at Segontium fort from the third century, not apparently before,15 although the material was exported to Chester as early as the time of Trajan.16 It may be that the rectangular form of the slates, and the edgewise method of hanging, was copied from the familiar tegulae which they replaced at the fort, or else from shingle roofing. The ridge of the Mithraeum roof was clad in imbrices, portions of which occurred in the rubble. The narthex, 6 ft. by 18 ft., had been almost totally destroyed, partly by natural denudation long before the sewerage trenches were carried through it. No sign of any fitting-bench, pit, hearth, or closet-was noted. The floor itself had also vanished, but (as is proper) appears to have coincided in level with that of the benches and sanctuary of this phase.17 Outside the entrance there was a patch of cobbling it is possible that some kind of rough shelter- pronaos would be far too grand a term18-was erected over it, since two or three groups of cobbles, placed as if to hold the butts of wooden uprights (PI. xi (d) ), were found in line with the west wall. This, however, is far from certain, and the groupings could well be accidental. The chapel measures 35 ft. by 18 ft. the axial, rectangular niche at the north end was 8 ft. wide and about H ft. from front to back. The nave and benches were 30 ft. long their respective widths are uncertain owing to the construction of timber colonnades on either side in Phase II (p. 147). The nave was probably about 8 ft. wide, thus leaving 5 ft. for each of the podia dimensions approximating to those of subsequent phases. The nave-floor was of brick-mortar, 2 in. thick, utterly rotted. The steps leading down from the narthex had been destroyed by the sewerage trenches. The podia extended 15 Segontium, pp. 102, 104. 16 As shown by the Trajanic inscription from Chester R. P. Wright and I. A. Richmond, op. cit. in note 2, no. 15. The small fragment from Caerhun (Arch. Camb., 1936, pp. 210 (fig.), 214) may also be a case in point, although, despite the apparently Trajanic rebuild, it would be hazardous to expand its two letters, however well cut, to Trajan)OA[ug. 17 The narthex was in an utterly mined state--so much so that the possibility of its having been pulled down in Roman times was considered (cf. Rudchcster, p. 196). No positive sign was forthcoming, and since such anterooms are so general a part of the Mithraic plan, the possibility is best ignored. 18 As, e.g., at Heddernheim, no. I T. & M., II, mon. 251.