376 MR T. C. BAILLIE ON THE 



F, with a fairly long neck was filled to the bottom of the neck with mercury and put 

 into the pot ; and the space, H, between was filled nearly full of mercury. To prevent 

 mercury leaking through the cast-iron pot it was previously lined with pipe-clay, a paste 

 of pipe-clay and water being painted in with a brush and allowed to dry. Into the 

 neck of the flask was fitted a glass piece made as shown in fig. 2. The arrows show the 

 path taken by the gas to reach the burner, and the temperature was kept constant by 

 the mercury cutting off" the gas supply at E on reaching a certain temperature. The by- 

 pass, B, was opened to allow a full supply of gas while the heater was being warmed up 

 to the proper temperature, the mercury in the flask being allowed to run over at D, 

 which was at all other times closed. When the desired temperature was reached the 

 by -pass, B, was nearly closed, enough gas being allowed to pass through it to keep the 

 Argand burner, G, from going out. This gas regulator worked so well that a ther- 

 mometer hung in the same place in the pot of mercury showed no variation exceeding 

 one-tenth of a degree centigrade during a whole day. 



At first a large steel cap was fitted on the end of the bar, with mercury inside it, 

 the idea being to make it at once the heater and the regulator. It showed a steady, 

 slow rise of temperature, and, although there was no visible leakage, in a few days fine 

 drops of mercury were seen on the iron tray placed under the burner to catch the 

 mercury in case of accident. No leakage of mercury could be noticed from it even 

 under greater pressure from the inside while standing cold, and therefore the mercury 

 must have leaked through pores too small to be noticed while the flame played upon 

 them. The variation of the temperature of cut-orT is a very delicate test of such 

 leakage. 



The rate at which heat was given out at the other end of the bar was obtained by 

 measuring the rise of temperature and the rate of flow of the stream of water which 

 played on the end of the bar. A brass cap, M, was fitted on the end of the bar, the 

 water entering the space between it and the bar having its temperature measured at 0, 

 and the temperature of the water leaving the bar was measured at P. The thermom- 

 eters at and P were Anschutz thermometers graduated in fifths of a degree centi- 

 grade. The rate of flow of the water was found by observing the time taken to fill the 

 flask, Q, of known capacity to the fiducial mark. The water was supplied at constant 

 level from a chamber, S, containing the well-known inverted bottle device, R. Distilled 

 water was used, but great difficulty was found in keeping the rate of flow regular until 

 the plan was tried of making the outlet of a piece of glass tubing drawn out fine and 

 broken off at the capillary portion. With this improvement the flow was very uniform, 

 and the temperature of the water (at 0) reaching the bar was also very steady, but the 

 temperature of the water leaving the bar (at P) varied. When the water had been once 

 used it was cooled by being put in the inner chamber of a double copper tank, while 

 cold tap water was circulating in the outer chamber surrounding it. The same water 

 was thus used over and over again. 



The order of taking readings was as follows : — 1°, the thermometers in the bar ; 2 , 



