50 Mr. E. H. Griffiths on the Influence of 



surround the silver flask. Communicating with this coil 

 were four leads — Nos. 1 and 2 of which were fastened to one 

 end, and Nos. 3 and 4 to the other end of the coil at the roof 

 of the calorimeter, and, passing up through the steel vessel, 

 communicated with the exterior. Nos. 1 and 3 formed the 

 ends of a circuit in which were placed a high-resistance 

 galvanometer and the Clark cells ; Nos. 2 and 4 a circuit 

 which contained a rheochord of special construction (for a 

 description see paper J), reversing-keys, and storage-cells. 

 It was thus possible to maintain the ends of the calorimeter- 

 coil at a difference of potential equal to that of any number 

 of Clark cells ; and I am convinced that the variations in the 

 potential-difference during the course of an experiment rarely 



E 



amounted to j ~ nm . (The grounds on which this somewhat 



bold statement is based are fully given in paper J, p. 283.) 

 The contents of the calorimeter could thus be heated or cooled 

 without any disturbing effects from external causes. By 

 means of the electrical arrangements above described, the 

 supply of heat could be regulated at will and accurately de- 

 termined, while, by the insertion into the flask of a volatile 

 liquid such as ether, it was possible, by adjusting the current 

 of dried air, to regulate the rate of cooling. 



Particulars of the Clark cells have already been published 

 (paper J, pp. 286-288), and the whole 36 maintain to-day 

 almost exactly the same relative values as they had when 

 compared with the Cavendish standard in 1892 : there is, 

 therefore, every reason to believe that their absolute value 

 remains unchanged. I have not yet had time to repeat the com- 

 parison with the Cavendish standard, but I hope to do so shortly. 

 I do not, however, anticipate that any correction will be found 

 necessary. The whole of these cells were contained in a bath, 

 through which the tap-water was turned by means of a regulator 

 whenever the temperature exceeded 15° C. 



The calorimeter previously described differed considerably 

 from that used in my determinations of the mechanical equi- 

 valent. During these experiments the air-pressure in the 

 space between the walls of the calorimeter and those of the 

 steel chamber containing it was reduced to from '5 to *2 millim. 

 as measured by McLeod's gauge; and my observations proved 

 that there was a strangely rapid diminution in the loss by 

 convection, &c, when the pressure fell below half a milli- 

 metre*. I proposed, however, during the experiments for 

 which this second calorimeter was designed to maintain the 



* A similar conclusion was arrived at bv Bottomley ; see Phil. Trans. 

 1887, A. 



