of the Electrical Conductivity of Mica. 201 



of (a) will be that k will rise with X, but, as (b) acts the 

 opposite way, a maximum will be reached, after which k will 

 fall again. The higher the temperature the more readily will 

 the brush-discharge occur, especially as ions from the flame 

 almost certainly find their way into the oven, so That the 

 maximum value will occur with smaller values of X; thus a 

 greater proportion of the readings will be in the neigh- 

 bourhood of the maximum, and the average value of k will 

 be higher. It is obvious, if the above is correct, that 

 nothing is to be gained by bringing the guard-ring closer to 

 the central disk; in fact, it might be advisable to increase 

 the width of the gap between them. 



From the above considerations it is apparent that the 

 observations on the dielectric constant only show that it does 

 not undergo any large variations up to 230° 0. The values 

 obtained for k are, however, chiefly of use in enabling an 

 approximate correction to be made of the leakage current 

 density. If we assume that k is really constant, then the 

 effective area in any given case is proportional to the appa- 

 rent value found for k. If we make the further assumption 

 that the effective area for conduction current is the same as 

 that for electrostatic capacity (which is probably not strictly 

 true), we can use the apparent values of k to correct the 

 observed values of the current. "With small values of X at 

 13°*5 C. k was found to be 7*70; this value was taken as the 

 true value of & corresponding to an area 2*02 sq. cm. In 



every other case the area was assumed to be — ? , and 



the conduction current density calculated accordingly. The 

 figures for the conduction current at 159 0, 5 C, which have 

 been given above, have been corrected in this way. Owing 

 to the assumptions involved, the correction is not a very 

 satisfactory one ; but, as the variations in the current are 

 very large compared with the correction, it does not seriously 

 affect the form of the results. 



If C = aXe bX or log C = A + log X + BX, the values of 



log ^ ought to lie on a right line when plotted against X. 



The results of the observations at ten different temperatures 

 from 50°-5C. to 229°'5 C. are plotted in this way in fig. 3, 

 the logs being to the base ten. It will be seen" that they 

 lie very approximately on ten parallel right lines, corre- 

 sponding to the different temperatures. To avoid confusion 

 where the lines are close together, the results are plotted as 

 Phil. Mag. S. 6. Vol. 34. No. 201. Sept. 1917. P 



