(1) That next the stokehole arch is 10 ft. square, and contains twenty-two pile of tiles, some of the pile on the east side having been laid upon a slanting structure built with chalk. The west wall of this compartment was broken down, but as the pile on that side reached to the outer face of the wall, there was probably a passage here for hot air into No. 35.

(2) Compartment, 19 ft. 8 in. by 10 in., on either side was one of those curious slanting chalk structures, each 6 ft. 8 in. long, 3 ft. 8 in. wide, and 14 in. high at the junction with the wall.

(3) Compartment, 7 ft. by 10 ft., fitted with pilæ. In both corners of the south wall were smoke flues of imbrex tiles. In order to keep one of these flues in position, tiles were built up from the floor of the hypocaust to support it, sufficient space being left between them to enable the fumes to ascend.

On referring to Plate G the remains of a wall with a division in the centre 16 in. wide, for the passage of hot air, will be seen between compartments Nos. 2 and 3. The wall was of flint and tiles, and doubtless built to relieve the pilæ from the weight of the floor that was originally above. A similar precaution was taken between Nos. 1 and 2, where two rows of large tiles had been inserted. The length of the floor would have necessitated more substantial support than mere single columns of pilæ could have given.

The curious slanting ledges of chalk in the centre are a novel feature and difficult to explain. A practical manager of a cement factory, who saw them, suggested that they might possibly have influenced the draught through the hypocaust; a suggestion we are disposed to agree with.

It is almost needless to say that so extensive a hypocaust required great heat to keep it going. The appearance of the walls of the furnace fully proved this.

34. Room, 10 ft. 4 in. by 8 ft. 7 in., partially paved with red tesseræ. When this chamber was cleared out the hollow sound of the western portion of the floor, which was paved with tiles, induced the writer to investigate the cause. On lifting the tiles and the bed of mortar on which they lay a quantity of tile rubbish was found underneath, which on being removed disclosed at a depth of 1 ft. 5 in. the floor of a room paved with red tesseræ, the walls being coated with fine hard reddened plaster. Here was another of the many alterations to the villa. When the steps were made into the great bath (No. 36), this older room was destroyed, the wall dividing No. 34 from No. 36 passing over it. Having slightly digressed in order to

describe this alteration we return to No. 34. In the south wall of the apartment, level with the floor, existed a drain pipe communicating with the gutter outside the house; the pipe was precisely the same as the modern one used in draining land. The entrance to the chamber was in the north wall.

35. This chamber, 26 ft. 7 in. by 5 ft. 5 in., seems to have been connected, as before observed, with the hypocaust (No. 33) adjoining, which is indicated on the Plan by the position of the pile at the point where the west wall terminates. Against the end of this wall a flue tile remained in its original vertical position. Upon the north wall the remains of three flue tiles were found which had been fixed to the face of it, at a height of 11 in. from the floor, and there were marks also where others had existed. Against the west wall nine tiles had been placed vertically, these and the walls at this end of the chamber bore evidence of having been subjected to the direct action of flames. Over the vertical tiles, at the spot indicated by an arrow in the Plan, a V-shaped opening had been cut through the wall, with an imbrex tile at the bottom serving for a gutter which discharged into an open drain outside. The floor of the chamber was partially paved with tiles. At the south-west angle a flue tile was inserted obliquely in the west wall, which conveyed the smoke to the outside of the house.

In the opposite wall, at the southern end, there was an oblong opening at the floor line 5 ft. wide and 2 ft. 3 in. high, which had been blocked up with rubbish when the general filling in of Nos. 32, 33, and 35 took place, as previously stated. On removing the loose material from the aperture the skeleton of a dog was discovered.

We have already seen that heat was transmitted into this chamber at the northern end from the hypocaust No. 33, and it seems reasonable to suppose that the aperture just mentioned was made for the purpose of admitting the hot air to the southern end also. In short there was a circulation of heat under all the apartments situate above Nos. 32, 33, and 35.

We wish it to be clearly understood that the opinions expressed with regard to these three chambers are given with reserve, as the dilapidated condition of the area they cover prevents any accurate conclusions being arrived at.

36. The great bath, 39 ft. 6 in. long, 9ft. 11 in. wide, and 4ft. deep. (Plate H.) The walls were built of flint with a lining of tiles laid in courses, faced with a thick coating of fine plaster mixed with pounded tile. There were four steps leading down into the bath, constructed with tiles and covered with plaster, the edges of the steps being