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



the values found for A and B were 3*25 and 0*80 respec- 

 tively, the logs being to the base 10. 



Experiments were also made on the value of k when 

 the pressure was falling. The condenser was charged up to 

 about 5000 volts, and allowed to leak down to the required 

 potential and then discharged. A decided rise in the value 

 of k was observed, evidently due to a hysteresis effect. This- 

 was not a permanent change caused by the high pressures, 

 as, on again charging the condenser up to some voltage and 

 then discharging, the same value was obtained for k as- 

 before the high voltages had been attained. The conduction 

 current appeared to be about the same whether the pressure 

 was stationary or falling ; this could not be accurately deter- 

 mined, as the galvanometer Gr only measures this current 

 correctly when the pressure is stationary, or nearly so. 



As some doubt was felt as to the constancy of the ley den- 

 jar E, an air-condenser was constructed. This consisted of 

 three co-axial copper cylinders, of which the inner and the 

 outer were earthed, and the intermediate one supported 

 midway between them on vulcanite insulators. As the 

 length of the insulated cylinder was about 26*4 cm., and its 

 mean diameter 21*3 cm., the air-gap being about 0*625 cm., 

 the capacity ought to be about 450 cm. or 5xl0~ 4 micro- 

 farad. The value found by experiment was 5'25 X 10~ 4 . 

 This is about 3 per cent, less than the capacity of the- 

 ley den- jar. 



A set of readings obtained with the air-condenser are 

 shown below. Here V is the voltage, obtained by multi- 

 plying the throw of H by the proper constant, Q is the 

 charge on the mica condenser, similarly obtained from that 

 of I, k is the dielectric constant thus found, X is the potential 



Y 

 gradient in megavolts per cm.= jYohi C is the conduction 



current in microamperes per sq. cm., and C the current 

 calculated from the formula log G = 3'286 + 0'794 . X -f- log X; 

 D is the difference between the last two columns. 



The first two readings are rather low. The first one was 

 obtained by charging with a battery in the course of the 

 voltage calibration, and it is possible that the steadiness of 

 the voltage may have caused k to have a lower value. It 

 appeared, both from these and the previous experiments, 

 that, when the voltage was falling, the value of k was in- 

 creased, probably due to a hysteresis effect. With the 

 machine it w r as hard to keep the voltage quite steady, and 

 to this some of the variations observed in k must probably 



