1903.] 



imparted to a Vacuum by Hot Conductors. 



417 



for a considerable range of pressures. To account for the results at 

 the higher pressures, it is necessary to assume that ions are produced 

 by collisions. 



The variation of current with temperature is examined over the 

 following range : — 



For platinum from 10~ 10 to 10~ 3 ampere per sq. cm. of surface. 

 For carbon ,, 10 -s „ 2 ,, ,, „ 



For sodium „ lO -11 „ 2 x 10~ 2 ampere total current. 



The corresponding ranges of temperature for platinum and sodium 

 are roughly from 1000° C. to 1600° C, and from 100° C. to 450° C. re- 

 spectively. The small currents from sodium were measured with a 

 quadrant electrometer, but as a general rule, a sensitive D'Arsonval 

 galvanometer with suitable shunts was used. 



Perhaps the most striking result of the investigation is the rela- 

 tively enormous currents which have been obtained. The biggest 

 leak measured was 04 ampere from a carbon filament to an electrode 

 placed near it; this corresponded to a current of 2 amperes per sq. cm. 

 of the carbon surface, the potential on the filament being - 60 volts. 

 The pressure in this experiment was only g-Joth mm. This current 

 and some of the largest currents from sodium were registered on a 

 Weston ammeter. 



In all these experiments the potentials employed were too small to 

 maintain a discharge between the electrodes. 



Throughout the range given above, the relation between the 

 saturation current and the temperature was found to be represented 

 by a formula of the type 



C = AOie-W, 



where A and b are constants for each conductor. 



The values which have been found for these constants are — 



For platinum, A = 10 2 ' 5 , b = 4*93 x 10 4 . 

 For carbon, A = 10 34 , b = 7 -8 x 10 4 . ^ 



9-7 x 10 4 . [ 

 11-9 x 10 4 . ) 



For sodium, A = 10 31 , b = 3* 16 x 10 4 . 



The value of A varies very rapidly with the value found for b, so 

 that only its order of magnitude is given. 



The number n of free corpuscles in a c.c. of the metal is calculated 

 from A. For platinum this gives n = 10 21 , whereas Professor 

 Patterson* found 10 22 . In the case of the other conductors, the 

 number found is absurdly great compared with Patterson's values. 



* ' Phil. Mag.' [6], vol. 3, p. 655. 



