very nearly, if not exceeding, the yield of the spring, there was also another, which I have every reason to think was for the delivery of cold water, and conveyed in a lead tubular pipe of 2¼in. in diameter. A length of 25ft. 6in. of this pipe, in its original position, has been found and laid bare. It is made with a roll along the top, and burnt, as was usual before the invention of "drawn pipes." This pipe is particularly interesting as there are also in it two soldered joints at intervals of 9ft. in the method of making which we have clearly not improved on the work of our Roman predecessors. This pipe starts from the same point in the north-west angle of the hall as the other supply, and is sunk in the lower pavement of the schola, which (wanting the pipe) is continued to the centre of the north side of the bath, where stands a stone pedestal 3ft. 3in. long, 1ft. 6in. wide, and 2ft. 6in. high. This pedestal has small vertical rails, or balusters, at the angles and on the shorter sides, and that towards the bath has some appearance of having once had a tablet of either bronze or marble inserted in it. At the top is a circular hole 3½in. in diameter, through which the pipe previously mentioned must have passed. The upper portion of this pedestal is sculptured, and much mutilated, and appears to me to be the drapery covering the feet of a figure that has perished. It is true that the work bears some resemblance to a small recumbent figure; but if so it is not worthy of the name of sculpture, as it is in the worst taste, and altogether out of keeping with the architecture or the other sculpture we have found.¹⁹ There are several grooves in the schola for branches of this pipe: 1st. The continuation of it to the northern semi-circular bath of 1755. 2nd. From the first soldered joint to baths on the north of the Great Bath. 3rd. Along the western end of the latter to baths on the south, and along the schola to the south circular bath of Lucas's. Beneath the mutilated sculpture is a second pedestal, or plinth, perfectly plain, with the upper surface sunk to a level corresponding with a similar indentation on the third step. Within this must have stood a marble on bronze sarcophagus, the base of which was 6ft. 9in. long by 2ft. 5in. wide. The water flowing through the aperture previously described would run into the sarcophagus (I use the word in its modern sense) and from it into the bath. This water was not poured in sufficient volume to perceptibly cool the bath, but was provided for the thirst of the bathers. In the modern baths of Bath there is no such provision.

The hall enclosing the bath I have already spoken of as 110ft. 4½in. long by 68ft. 5in. wide. It has been completely thrown open since this paper was read at the British and Gloucestershire Archæological Society, in 1884. These excavations are open to the sky, excepting on the east end (over which Abbey Street, at a height of 23ft. is carried on a viaduct, which I have erected).²⁰ The platform, or schola, surrounding the bath (measuring the original surface of the upper floor) is 13ft. 9in. wide on the four sides. This platform was formed by a layer of large freestone 9in. to 10in. thick, laid on the level of the top step but one, on a solid bed of concrete. Above this was another layer of concrete, and possibly on this, when the baths were first erected, a mosaic of tesseræ; but that, if it ever was there, has all disappeared, and its place has been supplied with paving, mostly of freestone also, of inferior thickness to the lower paving. Very little of this remains, and what there is is much fractured and worn; indeed not only is this paving much worn, but the lower paving also where the traffic was the greatest. I have given in the plan (Pl. VIII.) almost every detail of these floors, and shall speak of them again further on. The general appearance of the place is symmetrical, but there are remarkable variations and inaccuracies that point to the fact that the juxta-position of this bath with other buildings, of which we have at present no knowledge, must have rendered these variations necessary, ultimately interfering with the completion, architecturally, of the building.

On either side, north and south, are three recesses, or exedrae, two of which are circular and one (the centre) rectangular. The south rectangular one is 17ft. wide by 7ft. deep; the north one is nearly a foot wider, and one foot less in depth. Greater variations exist in the circular recesses; for, commencing in the western one, on the south side, the width is 17ft. 3in., and the depth 7ft. 6in.; the eastern one is 14ft. 3in. wide, and 6ft. 9in. deep; the exedrae vis-a-vis on the north is 17ft. 3in. wide, and 8ft. 4in. deep; the remaining one, to the west, is 17ft. wide, and 7ft. deep. I give these dimensions irrespective entirely of the pilasters which are attached to the walls on either side the reveil of the recesses, and in the rectangular recesses in the enclosing angles also. Piers are now standing on the margin of the bath, dividing the north and south sides each into seven bays. These piers are built with solid block freestone, but as there are continuous vertical joints on either side of the central division of each pier, it is clear that an alteration was made in the design either previous to its entire completion or subsequently.

I will endeavour to describe the bath as originally designed. Along the margin of the bath, north and south, stood six piers, equally divided (about 14ft. apart), as far as the length of the bath, but allowing a lesser distance from the attached pilaster at either end. These piers are cut out of a block (in plan, 2ft. 10½in. from east to west by 2ft. 8in. from north to south), so as to form a pilaster of three inches projection on either face. As the original pilasters on the north and south walls do not correspond with these piers, I am led to conclude that the schola and exedrae, north and south, were not vaulted at first, and were the only portion of the hall that was roofed, and that the roof was only of timber, supported by an arcade, the arches not exceeding 17ft. in height, and that the eaves of the roof of about 22ft. in height dipped towards the bath. This was a very usual arrangement in the Atrium of a Roman house with the impluvium in the centre. A crypto porticus would thus be formed on the two longer sides of the bath, but the schola on the east and west ends was open to the sky. Practical experience, either on the completion of this plan, or previously to its entire execution, led to its abandonment. At any rate a roof over the whole was found essential to the comforts of the bathers. The piers were accordingly strengthened. Pilasters were erected, projecting 2ft. 9m. into the bath, with smaller pilasters on the other side projecting on the schola, 1ft. 4in. by 1ft. 11in. wide; and vis-a-vis to these pilasters corresponding ones were affixed to the side walls. Unfortunately this brought into prominence the irregularity of the size and position of the exedrae, and the pilasters were affixed correctly with reference to the arcade, as was absolutely necessary, but more or less trespassing on the width of the opening of these