418 



Electrical Conductivity of a Vacuum. [Feb. 28 r 



The discrepancy can be made to disappear by assuming a small tern* 

 perature variation of b. This assumption is shown to be consistent 

 with the general nature of b. 



The work required to drive an ion through the surface layer is 

 calculated, in each case, from the value of b, to which it is propor- 

 tional. Dividing by the charge on an ion this yields the discontinuity 

 of potential at the surface of the conductor. The values found for 

 this are : for sodium, S<£ = 2'45 volts, for platinum 8(f> = 4'1 volts,, 

 and for carbon 8<j> = 6*1 volts. These numbers are inversely propor- 

 tional to the cube roots of the respective atomic volumes. This leads 

 to the conclusion that the work required to force a corpuscle out of a 

 metal varies, approximately at any rate, inversely as the cube root of 

 the atomic volume of the metal. 



In all these experiments, the current when the hot wire is charged 

 positively is small compared with that obtained with the metal 

 negatively charged. Only in the case of sodium was the positive 

 current large enough to deflect a sensitive galvanometer. 



The results which have been obtained are shown to furnish a com- 

 plete explanation of the phenomenon known as the Edison effect. 



The fact that such enormous currents are obtained at such very low 

 pressures confirms the conclusion that the ions are not produced from 

 the gas by its interaction with the metal. Calculation shows that to> 

 obtain the currents registered with carbon, each gas molecule would 

 have to give rise to 25 ions each time it collided with the hot metal 

 surface. 



The energy lost owing to the escape of the corpuscles is compared 

 with the energy emitted in the form of ordinary electromagnetic 

 radiation. The former is shown to be smaller than the latter at the 

 temperatures at which measurements have been made, but it increases 

 more rapidly with the temperature. 



