r ' 



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



it to leak down to the voltages given. The leakage-currents 

 are not given, as, at the pressures at which they could other- 

 wise be accurately read, the fall in potential was so rapid 

 that the lag of the galvanometer prevented any reliance being 

 placed on the readings. 



V. 



Q- 



k. _ 



3878 



1-819 



8-84 



3138 



1-483 



8-90 



2793 



1-338 



903 



2463 



1185 " 



904 



2146 



1-040 



912 



1704 



0-815 



8-98 



1266 



015 



9-15 



807 



0-384 



901 



It would appear as the result of these observations that up 

 to a gradient of 3 megavolts per cm. k is constant to within 

 a few per cent., and also that a small hysteresis rise in k, 

 amounting to about 4 or 5 per cent., is noticeable when the 

 pressure is falling. 



Turning to the leakage-current : all the evidence would seem 

 to point to this being a true conduction current, as distinct 

 from an effect due to increase of polarization. In the case 

 of the high-voltage readings given above, the current through 

 the condenser averaged 1*56 microampere for about ten 

 minutes, during which time the quantity transferred through 

 the condenser must have been about 935 microcoulombs. 

 The actual charge on the plates available on discharge 

 was only 2*4 microcoulombs, and may have increased by 

 0*05 microcoulomb in that time. Moreover, no appreciable 

 reverse current could be observed through the condenser 

 after discharge, as should be caused by a recovery from 

 polarization. 



The agreement with the formula given is within the limits 

 of error over the entire range, that is for currents ranging 

 from 0*01 to 1*5 microampere per sq. cm. ; as the current 

 is extremely sensitive to small changes of voltage near the 

 upper limits, the actual errors are naturally greater in that 

 part of the scale, though the relative errors are probably 

 smaller. 



