“The walls were of red and yellow brick in alternate layers composed of 18 in. tiles.” Outside the walls was “a drain of wooden planks, 18 in. deep by 10 wide, running towards the river” (see plan). The walls were erected on piles. The sketches show some of the box-flue tiles which had impressed patterns (see Fig. 25). Some additional information is given on a lithograph by A. J. Stothard (1848). The walls were 3 ft. thick. Above the floor of the south room, which was of coarse red and yellow tesseræ, was a second, about a foot higher in level; this was “a layer of red concrete 2½ in. thick, hard, and the upper surface almost glazed” (compare a floor found in Eastcheap, “concrete stuccoed over and painted red.”—_V.C.H._). This building was doubtless a house; at the time it was found it was called a bath, but it seems too small to have been even a secondary public bath. As Thos. Wright says: “Many writers have concluded hastily that every house with a hypocaust was a public bath” (cf. the plan of a house at Lymne, _The Roman, etc._, p. 160). The stoke-hole of the hypocaust was at F, and there were flues up the middle wall and the western apse. The large room was 23 ft. square; some tiles of 2 ft. square were found here, also window glass and an iron key. The plan lay square with the south City Wall (Fig. 24), and the building can hardly be earlier than this wall. It may thus be accepted as a late fourth-century house, and we may further infer that box-tiles with impressed patterns were a characteristic of this century. On two sides of the house were lanes about 10 ft. wide. As in so many cases modern walls seem to have been laid out on the same alignment as the Roman building. The house just described had two apses, and the Bucklersbury house also had an apse. This was in agreement with general custom. As Thos. Wright remarked: “One peculiarity which is observed almost invariably in Roman houses in Britain is that one room has a semicircular alcove, and in some instances more than one room possesses this adjunct.” In the plan given in _Archæologia_ of the Roman walls and floors found in and about Lombard Street in 1785 two apses seem to be indicated; thus we have evidence for five in the scanty records; altogether there must have been scores in the city.