ELECTROPF. EEEECTS 569 



and almost beyond ihc limit of accuracy of the measurements. How- 

 ever, a careful analysis of the results (omitting the ^" phenol fibre 

 sample which will be discussed separately) seems to indicate that the 

 power factor values obtained with the two 43^" circles agree slightly 

 better with those obtained with the guard ring electrode than do 

 those obtained with any of the other electrodes. 



In the case of the ^s" phenol fibre sample the variations with 

 different electrodes are much greater than the probable inaccuracy of 

 the measurements. Apparently, they can only be attributed to non- 

 uniformity of the material in different parts of the sample. Therefore, 

 a special test to determine this fact was made on this sample. By 

 interchanging the connections to the guard ring and the 4^/2" center 

 electrode, it is possible to measure the capacitance and conductance of 

 the outer part of the sample without including the center part. The 

 sum of the values of the two parts checks well with the value for the 

 entire 6" square. These results show that while the inner part of the 

 sample has a power factor of 1.97%, the power factor of the outer 

 part is 3.17% or approximately 60% higher. The corresponding 

 dielectric constants are 5.08 and 5.48, respectively. The reason for 

 this difference probably is that since the material is of a laminated 

 nature moisture penetrates more readily from the edges than from 

 the sides of the sample and thus causes a progressive variation of the 

 electrical characteristics from the edges to the center of the sample. 

 It is obvious that when the two 43^^" electrodes are used without 

 guard rings, some of the outer part of the sample is included due to the 

 fringe effect and that when the 43^" upper and 6" lower electrodes 

 are used still more of the outer part is included for the same reason, 

 and the values obtained are increased accordingly. 



As previously mentioned, it is probable that this non-uniformity is 

 responsible for the different shape of the edge effect curve for phenol 

 fibre as compared with those of hard rubber and glass and it is almost 

 certainly the cause of the point representing this particular sample 

 being exceptionally high. Since there are wide variations in the 

 power factors of the different samples of both phenol fibre and glass 

 it is possible that there are minor variations through the sample due 

 to causes other than moisture and that these may account for some 

 of the other apparent irregularities in the results. 



Method of Applying Corrections 



While percentage values are the most convenient for discussion of 

 the relative importance of the various corrections involved in the use 

 of a given type of electrode, the absolute values of these corrections 



