198 Mr. H. H. Poole on the Temperature Variation 



as the thickness of the condenser used in some of the expe- 

 riments described in the last paper, but the similarity is only 

 an accidental coincidence. 



The general electrical connexions are shown in fig. 2, the 

 letters having the same significance as in fig. 1 of the 

 previous paper. It will be noticed that the arrangement 

 has been modified by bringing the galvanometers H and I 

 to the earthed sides of their respective condensers, and 

 obtaining readings by earthing the source by means of the 

 key D. This prevents the occurrence of a large P.D. between 

 the central disk of the mica condenser and the guard-ring, 

 thus rendering possible the use of a condenser of the form 

 already described. A further advantage is that a much 

 simpler form of key can be used. This is a modified Morse 

 key, the moving part of which is permanently earthed. It 

 is held in the position shown by means of a trigger, thus 

 keeping the galvanometers H and I short-circuited. On 

 releasing the trigger a strong spring opens the galvanometer 

 shunts, and almost immediately afterwards brings a stiff 

 wire down on a conductor insulated on sealing-wax and con- 

 nected to the source, thus earthing the latter. As the 

 leakage current passes through I during the interval between 

 these two operations, it will have the effect of reducing the 

 throw of the latter on discharge, unless this interval is so 

 small as to render the error negligible. It was found im- 

 possible to measure this interval by the method adopted for 

 the mercury key described in the previous paper, as the 

 rebound of the wire prevented good enough contact being- 

 made, so that the results were apparently meaningless. As 

 some doubt was felt as to the possible effect of this error, the 

 key was modified during the course of the measurements so 

 as to greatly increase its speed. As this made no appreciable 

 difference in the results obtained, it was assumed that this 

 source of error was negligible. 



R, and S are known resistances used to reduce the sensi- 

 tiveness of G and I. R was made infinite for the smaller 

 leakage currents and varied, as required, for the larger 

 currents. S remained at 4000 ohms throughout, its effect 

 on the sensitivity of I as a ballistic was directly found. The 

 rest of the electrical arrangements remained as before. The 

 leyden-jar was used as the standard condenser E, as it bad 

 proved quite satisfactory and was more convenient than the 

 air-condenser. As before, the Wimshurst machine A was 

 driven by a permanent magnet motor, and the voltage was 

 adjusted partly by varying the speed of the latter and partly 



