POWER LOSSES IN INSULATING MATERIALS 



lis 



electric constant obtained for the sample. For this reason only 

 mercury electrodes have been found suitable, the sample being floated 

 on a pool of mercury forming the lower electrode and the upper elec- 

 trode being formed by pouring a pool of mercury inside a metal ring 

 on the upper surface of the sample. This introduces the second 

 difficulty. The lower electrode being, of necessity, larger than the 

 upper one, the electric field spreads out considerably beyond the 

 edges of the upper electrode and increases the effective area by an 

 unknown amount. 



Fig. 2 



To determine the magnitude of this error, measurements were 

 made on samples prepared as shown in Fig. 2 ^ is the upper electrode, 

 B is the sample under test and C is a tinfoil guard ring shellaced 

 to the lower surface of the sample but separated from the lower elec- 

 trode Z) by a sheet of paper shellaced over the guard ring. The 

 guard ring covers all of the lower surface of the sample except the 

 area equal to the upper electrode and directly under it. The direct 

 capacity ^ between A and D is measured using the guard ring as a 

 shield to intercept the flux which spreads out around the upper 

 electrode, diverting it away from the measuring circuit. This measure- 

 ment is made at audio frequency on a completely shielded substitu- 

 tion bridge. The difference between this capacity and the capacity 

 o{ A to D without the guard ring is approximately the correction due 

 to this edge effect. Using samples 6 inches square with an upper 

 electrode 5 inches square, this correction was found to be about 7 

 per cent for samples }4 inch thick and 14 per cent for samples ^ 

 inch thick. All values of dielectric constant given above have been 

 corrected. The phase difference is not affected appreciably since it is 

 dependent only on the ratio of resistance to reactance and does not 

 involve the area of the sample. 



The radio frequency generator used consists of a vacuum tube 



oscillator having a maximum output of 250 watts. The coupling 



between the generator and the measuring circuit was very loose and 



^ Direct Capacity Measurement, George A. Campbell, The Bell System Technical 

 Journal, July, 1922. 



