channel at the point of discharge into the drain. It is worth adding that a latrine with the same water channel running round the four sides of the floor, occurs at Timgad,[1] and that a similar arrangement is to be found at Pompeii.

The latrine appeared to have been in use subsequent to the construction of the rampart. Indeed its south wall had every sign of being nothing more than a retaining wall built against it. On the exterior next the rampart the surface was very rough, with yellow clay adhering in the interstices of the stone-work. That the building was comparatively late was shown by the discovery of a lower level beneath it. On removing the flags of the latrine floor, a second floor, covered with flags, was found fifteen inches beneath it. The appearance of this second floor is shown in Plate XIV., Fig. 2. The south wall of this apartment, two and a half feet in thickness, had been utilised as the inner wall of the latrine pit above. What function the room itself had originally served could not be ascertained, but the wall on the south had probably formed the exterior wall of an early enlargement of the bath-house. Apparently it had connected with the wall shown on the plan, bounding the west and south-west angle. This last, however, was traced only from its foundation trench.

The two levels noted at the latrine had their parallel in the main block of the bath building. The floor of the caldarium had been raised by placing upon it a double stratum of freestone blocks, arranged in rough herring-bone fashion, with a line of clay between the layers and a clay floor above. This covering of the lower floor was about twelve to fourteen inches in depth, and extended into the small recess or apartment on the west, covering the surface of the wall dividing it from the caldarium. Evidently, then, at the time the floor level was raised, this wall had been already reduced to the level of the soil. It is probably to the same period that one may assign the construction of a deep drain, running from near the middle of the tepidarium through the caldarium, and then in a north-westerly direction towards the Tweed. Where it passed beneath the main wall of the bath building at the point of exit, it appeared to have merely been tunnelled through. Plate XIII., Fig. 3. There was no trace of any arch or lintel to support the wall above. To the same period also probably belonged a wall, running from east to west, dividing the caldarium and tepidarium into two portions.

1 Boeswillwald, etc., Timgad, p. 14, fig. 7.