oj the Electrical Conductivity of Mica. 197 



mica is greatly exaggerated, and the silvered surfaces repre- 

 sented by heavy lines. The cylindrical piece of brass L, on 

 which the mica rests, is supported on several thicknesses of 

 mica and connected to the high-pressure source. The small 

 piece of brass M, resting on the central silvered disk on the 

 upper face of the mica, is connected to the galvanometers as 

 shown in fig. 2. The brass ring N, resting on the silvered 



Fiir. 2. 



guard-ring, is connected to earth. The effective area of the 

 plates, as far as the galvanometers are concerned, is the area 

 of the central disk. This is about 2*02 sq. cm., the diameter 

 of the disk being about 1*61 cm. The area affected by brush 

 discharge is limited to the annular space between the central 

 disk and the guard-ring. Unfortunately this area amounted 

 to 0*78 sq. cm.; so if we assume half of this to be assigned to 

 the brush discharge area of the guard-ring, we find that the 

 maximum possible increase of area due to brush discharge 

 would be about 19 per cent, of the measured area. It is not 

 very easy to reduce this source of error, as if the distance 

 between the guard-ring and the disk is too much reduced 

 there is considerable danger of a transference of charge 

 across the gap. In the condenser used this gap averaged 

 1*4 mm. wide, and, as will be seen later, there was 

 reason to suspect the existence of this latter error at high 

 temperatures. 



The mica sheet used was 0'0173 mm. thick, as measured 

 by the optical method already described. This was the same 



