273 



which the grooved channel entered 6 inches from the top, were 

 two small openings ; one connected with a channel (E f ) simi- 

 lar to the channel (E e) and which was traced for 40 yards ; 

 but nothing of the kind was found connected with the other. 

 Both the channels (E e) and (E f) appeared to have a fall 

 towards the tank, as if designed for the conveyance of water 

 to that point ; and yet as a reservoir for water it appears to be 

 of very small size. If it be imagined that this structure is 

 modem, the difficulty in conceiving for what purpose the tank 

 could have been constructed is equally great ; and though no 

 doubt the being constructed of flags is not the usual character, 

 yet I should not suppose this fact totally invalidates the 

 opinion that it is Roman work; for among other examples 

 in other parts, where the floors of rooms were formed of flag 

 stones laid regularly over a mass of concrete. Dr. Bruce speaks 

 of flag stones forming part of the hypocaust at Cilurnum on the 

 Roman wall. Nearly in a line with the tank, and to the north- 

 east of it, a large square of foundations (G) formed of the best 

 worked stone that was met with, measuring 10 by 18 yards, 

 was excavated : the west wall of which is marked strongly with 

 fire and displays some remains of a fire-place. The floor had 

 been covered with that coarse kind of pavement called Ruderatiot 

 formed by imbedding refuse pieces of tiles and stones in a struc- 

 ture of mortar. 



Adjoining the north wall of the hypocaust (B) is a third 

 chamber (L) referred to above, of which the foundations alone 

 remain, measuring from north to south 28 feet, from east to 

 west 27 feet, on the same level as the other two rooms. It 

 communicates with the second of these by an opening like a 

 flue (b), and which I imagine has been for the escape of 

 the smoke from the hypocaust ; only a few feet of it existed, 

 so that neither its course nor destination could be traced. 

 As a quantity of large red stone and coloured tesserce were 

 dug out of this room, it had in all probability at one time 

 possessed a tessellated floor. Round the lower part of the wall 

 was a species of ledge, formed by the upper course of masonry, 

 being narrower than the lower. This might have been a sup- 



