Electrical Conductivity of Mica in Intense Fields. 121 



capacity was probably due to overheating, which may have 

 caused the seccotine to soften and creep along the surface, 

 thus increasing the effective area. Only a few deter- 

 minations were made, as the condenser was pierced by the 

 discharge from a pressure of 2680 volts, or 2*23 megavolts 

 per cm. The mica may possibly have been injured in some 

 way during the heating as the condenser had previously 

 survived 2800 volts. This sheet of mica was not measured 

 optically, as it had been covered with shellac before the 

 optical method was devised. 



A third condenser was then made out of a sheet of mica 

 0*0173 mm. thick, measured optically. It was coated with 

 shellac as before, but was only gently warmed for several 

 days over a carbon-filament lamp. The resistances P and Q 

 (fig. 1), which had previously been connected between the 

 key and the high-pressure source, were transferred to the 

 position shown in the figure, thus preventing surges at 

 discharge. 



Since the tinfoil disks were 1*15 cm. in diameter the 

 capacity K of the condenser 



l*15 2 x k 

 = ~T7. — A ^/n^o — n — ttt* = 5*315 X 10 " 3 X k microfarad, 

 lb x 0-00173 x 9 x 10° ' 



where k is the dielectric constant. Hence 



Q 

 v 



k = 1-883 x 10 4 x ^ 



where Q is the charge on the condenser in microcoulombs, 

 and V is the pressure in volts. Q and Y are obtained by 

 multiplying the throws of the galvanometers I and H by the 

 appropriate factors, which have already been given. 



Several sets of observations were taken using the leyden- 

 jar E. Each day's readings agreed well amongst them- 

 selves, but small variations, amounting to a few per cent., 

 occurred from day to day. The value found for k was about 

 9*0 ; it was apparently independent of the potential gradient 

 within the limits of experimental error up to a gradient of 

 3*145 megavolts per cm., which must be very near the 

 sparking value. At this gradient the leakage current 

 amounted to 1*8 microampere per sq. cm. This current 

 agreed well with the formula 



C = aXe lx , 

 or 



logO = A + logX + BX, 



where C is the current density, X the potential gradient, 

 ■ a, b, A, and B are constants. From these determinations 



