62 Roman Villa at North Wraxhall. 



pipes, like the underground flues, were blackened inside with soot, 

 it is evident that smoke as well as heated air must have passed at 

 times through them, and entered the bath-rooms above by their 

 perforations. It is possible that some means may have existed for 

 cutting oflP the communication between the furnace and the hypo- 

 caust, until the wood fired in the former had burnt down to 

 charcoal or a red braise, as is done in a modern oven. The furnace 

 which heated the two sudatories A and B was at E (see plan) in 

 the chamber adjoining the latter, into which the draught of hot 

 air passed by an arched opening beneath the level of the floor, 

 while the lower part of the wall between the two rooms was 

 pierced by similar openings. On one side of this furnace at (e) is 

 a recess, which was found half-filled with charcoal. It was evi- 

 dently the fuel-store. The other side was banked up with concrete 

 to prevent the escape of heat. The two other bath-rooms, C and D, 

 were warmed in the same way from a second furnace in F, commu- 

 nicating with them in a similar manner by an arched opening in 

 the wall. The partition wall between the chambers C and D, like 

 that between A and B was also pierced with several rudely 

 arched openings or flues ; so that one fire in each case warmed two 

 apartments. 



The two semi-circular bays or recesses in chambers A and B are 

 worthy of particular remark. In A the tubular flue-pipes were 

 ranged round the inner curve next the wall, as well as round the 

 interior walls of the small square chamber adjoining. In B the 

 recess was separated from the square chamber by a wall reaching 

 rather above the floor (but perforated at either end to allow the 

 passage of the hot air) and having two flagged steps upon it 

 ascending from the floor of the chamber to that of the niche itself, 

 which had a solid foundation of concrete down to the base of the 

 building, but with an interval between this semicircular pier 

 and the wall for the passage of the heated air. In A there was no 

 such division, but only square pilasters projecting from the sides 

 marked the separation of the chamber from the recess. The entire 

 arrangement suggests the idea that these semi-circular niches were 

 capable of being shut off" from the adjoining chambers by a curtain 



