Conductivity of. some Dielectrics. 495 



The results obtained at 14 o, C. are plotted as small circles 

 in fig. 3, where the ordinates represent the logarithms o£ 

 the conductivity (K) and the abscissae the corresponding 

 potential gradients (X). Here and elsewhere the unit of 

 current density is taken as one microampere per square 

 centimetre, and the unit of potential gradient one megavolt 

 per centimetre, so that the unit of conductivity employed is 

 10 ""(ohm)" 1 (cm.) -1 . It will be seen that the results liealong 

 a smooth curve. The larger circles represent the original 

 high-pressure observations at 9 o, C. and the short right line 

 near the origin represents the results of electrometer obser- 

 vations at 15°'5 C. Allowing for differences of temperature 

 the latest results agree well witb both previous sets. The 

 curvature of the isotherms is however here clearly brought 

 out for the first time, as in the previous work the available 

 range was insufficient, and the isotherms treated as right 

 lines. It might have been inferred from the discordance 

 between the electrometer and galvanometer results (Phil. 

 Mag. July 1916). Moreover, inspection of fig. 3, p. 202, 

 Phil. Mag. Sept. 1917, will show that the results plotted 

 therein could be better represented by slightly curved, 

 rather than by straight isotherms. 



Further tests were carried out on another specimen (B) 

 which had been previously prepared but never tested. This 

 was taken from the same batch as that described in the 

 second of the previous papers, and hence, as was to be 

 expected, its conductivity was only about one hundredth 

 part of that of specimen A. The surfaces had been silvered 

 by Carey Lea's process, the arrangement being the same as 

 that used before. The area of the central disk was 5*1 sq. cm. 

 and the thickness of the sheet 0*0115 mm. The specimen 

 was enclosed in a small oven and tested at various tem- 

 peratures up to 128°C. The results are plotted in fig. 4. 

 Small corrections have been introduced in plotting to allow 

 for any small accidental temperature differences that oc- 

 curred. The isotherms are similar curves which at moderate 

 gradients become parallel right lines. The slope of these 

 lines is 1*18, which is almost identical with the corresponding- 

 slope for specimen A, although the conductivity is so dif- 

 ferent. It will be observed that the conductivity of B at 

 109° C. is very similar to that of A at 14° 0. for all gradients. 

 The upper parts of the curves correspond fairly well with the 

 previously given straight line isotherms for a specimen from 

 the same batch, though the conductivity of the present 



