116 Mr. H. H. Poole on the Dielectric Constant and 



desiccator. By carefully watching the condenser in a per- 

 fectly dark room, a faint flash was seen surrounding the 

 tinfoil disk at the instant of discharge, indicating that the 

 effective area was increased by a brush discharge over the 

 surface. To prevent this the mica and the tinfoil disks were 

 covered with a thick layer of shellac dissolved in absolute 

 alcohol, and the whole gently warmed for several days. 

 Great care was necessary not to overheat the condenser, as 

 this softened the seccotine. Both sides of the mica were thus 

 treated as indicated in fig. 2. The iron ring which formed 

 the support was placed on an earthed metal stand in a 

 desiccator. The base of the latter was formed of paraffin- 

 wax, through which the various wires passed. As the 

 diameters of the disks were from 600 to 1.000 times the 

 thickness of the mica sheet, the edge correction was very 



small, so that K= ^r, where K is the capacity of the con- 

 denser in E.S.U., k the dielectric constant, d the diameter of 

 the disks, and t the thickness of the sheet, both in cms. 



The resistances P and Q were not used in the earlier ex- 

 periments, but it was found that the mica was sometimes 

 pierced by sudden connexion to the high-pressure source, 

 even though the pressure of the latter was, if gradually 

 applied, insufficient to cause breakdown. This was pro- 

 bably due to a surge, which might cause the pressure in the- 

 condenser to rise considerably above that of the source. Ta 

 prevent oscillations in a condenser circuit, it is necessary 



4L 



that R 2 > -^-, where R is the resistance, L the self-inductance,. 



and K the capacity. It was not very easy to estimate the 

 value of L in this case, but, as K was so small, it appeared 

 that a considerable value of R was necessary to prevent risk. 

 The resistances used were carbon lamps whose resistances- 

 when cold were about 1200 ohms. Wires were soldered to 

 the contact pieces of the lamps, and paraffin-wax w r as then 

 run into moulds surrounding the lamp caps to prevent the 

 risk of sparks passing across from one terminal to the other. 

 At first these were inserted between the high-pressure source 

 and the key. This prevented surges at charge, but it w r as 

 found that the act of discharge pierced a condenser which 

 had stood the applied voltage for over half a minute. When 

 this happened, the throw of the galvanometer I was much 

 greater than it ought to have been, in fact off the scale. 

 Apparently the condenser was pierced by being charged in 

 the reverse direction to a somewhat lower voltage than it 



