86 The Ohio Journal of Science [Vol. XVI, No. 3, 



are two other possibilities. Number two, a possible creepage 

 or leakage from one terminal to another over the surface of the 

 porcelain. 



Number three, a current passing through the hot gases 

 surrounding the body. 



The third type is present if we raise the temperature to a 

 point where the interior begins to glow and apply a sufficiently 

 high potential. When an appreciable current passes through 

 the gas it becomes luminous and is readily observed. It is 

 characterized by appearing suddenly when a critical potential 

 is attained and is large compared with the Number One type 

 of current. The effect can be avoided only by operating over 

 potentials which should be kept below 350 volts. This places 

 a decided limitation in the form of apparatus shown in Figure 2. 



Anticipating the results slightly it may be said that leaving 

 out of account the gas discharge, the current passing through 

 the ammeter and through the porcelain appeared to be separable 

 into two parts, one which decreases rapidly with time and one 

 which if it changes at all changes at a much lower rate. It was 

 thought that this might be due to some alteration in surface 

 condition and this was borne in mind while making the tests 

 in porcelain. It did not seem feasible to alter the form of 

 apparatus to prevent the possible surface leakage in the case of 

 porcelain. It may be remarked also that while the values of 

 current strength differed somewhat from different cylinders, all 

 results were of the same general character. 



The next step in the experiment was to make a glass cylinder 

 of the same form as the porcelain cylinders previously described. 

 Only one such cylinder was used and while it showed appreciable 

 conduction at lower temperatures than did the porcelain, the 

 general characteristics of current-time were the same, namely, a 

 large initial current which decreased with time. 



Next a glass cylinder of the form shown in Figure 3 was 

 used. This was designed to eliminate both the possibility of 

 gas conduction and leakage over the surface. The glass cyl- 

 inder was continued at the upper edge by a glass tube which 

 extended nearly the entire length of the copper rod. A guard 

 ring shown in the figure (R) was wound around this tube and 

 earthed. The rod inside the tube rested on the glass cylinder 

 as before and was terminated with platinum. In this case if 



