120 Mr. H. H. Poole on the Dielectric Constant and 



X. 



C. 



C. 



1-0 



o-oi 



o-oi 



1-5 



04 



0-05 



2-0 



0-15 



0-15 



2-5 



050 



0-45 



2-83 



1-06 



97 



The figures in the first column are the potential gradients 

 in megavolts per cm., those in the second column the current 

 density in microamperes per sq. cm., while those in the 

 third column are the current densities calculated from the 

 formula 



logO = logX + 3'286 + 0'794.X, 



which was found to agree with the later results. It will be 

 seen that the agreement is good at the lower gradients, but 

 that the observed current is too large at the higher gradients. 

 This is almost certainly due to the brush discharge increas- 

 ing the effective area at high potentials. 



This condenser stood a pressure of 4750 volts or 2*83 

 megavolts per cm. ; it was eventually pierced by being inad- 

 vertently connected to the high-pressure source when the 

 pressure of the latter had risen too high. The thickness of 

 the mica sheet was subsequently measured optically and 

 found to be 0*0168 mm. The figures given above are cal- 

 culated on this value. 



A new condenser was then made out of a sheet 0'012 mm. 

 thick. The results obtained agreed fairly well with the 

 previous ones, both as regards dielectric constant and con- 

 duction current. The latter seemed to be a function of the 

 potential gradient only, as the voltage required to produce a 

 given current was directly proportional to the thickness of 

 the sheet. Further experiments on this point are, however, 

 desirable, as the sheets employed did not differ very much 

 in thickness. The sparking potential through an insulator 

 is known not to be a linear function of the thickness, so 

 possibly a similar effect might occur with the conduction 

 -current. 



This condenser was in use when the brush discharge was 

 discovered. The mica surfaces were then covered with a 

 thick layer of shellac dissolved in absolute alcohol, and the 

 •condenser gently heated for several days in front of an 

 electric radiator. The upper part of the jar E was also 

 coated with shellac. The capacity and the leakage conduc- 

 tivity were both found to have increased by about 35 per 

 «ent., while the capacity was now almost independent of 

 the voltage throughout the entire range. This increase in 



