60 



SCIENCE 



[N. S. Vol. XXXVIII. No. 967 



in the following experiment for whieli the 

 other data are lacking. 



A filament 3 em. long gave 0.099 amp. elec- 

 tronic emission continuously for 2.5 hours. 

 The resistance when hot rose from 4,773 to 

 ■1,787 in arbitrary units. The number of 

 atoms of tungsten lost by the filament in this 

 time was = 5.66 X 10'", whilst the number of 

 electrons emitted = 5.57 X 10"- The number 

 of electrons emitted per atom of tungsten lost 

 was 9.84 X 10^- The mass of the electrons 

 emitted in this experiment was thus very 

 close to three times the mass of the tungsten 

 lost by the filament. 



This tube gave 0.1 amp. electronic emission 

 on the average for 6 hours altogether. By 

 that time the mass of the electrons emitted 

 was approximately 2 per cent, of the mass of 

 the tungsten filament. The tube came to an 

 end owing to an accident: the filament grad- 

 ually became deformed until it touched the 

 copper electrode and broke. The hardness of 

 the tube was then tested with an induction 

 coil and the equivalent spark gap was found 

 to be 3.3 cm. The discharge through the tube 

 gave a bright green fluorescence on the glass 

 around the negative wire, but there was no 

 indication of a glow or the faint purple haze 

 which is obtained when traces of gas are pres- 

 ent in tubes of this kind. There is thus no 

 appreciable accumulation of gas even when 

 the filaments are allowed to emit a large 

 thermionic current continuously for a long 

 time. 



Another tube with a wire 2.7 cm. long, giv- 

 ing 0.050 amp., lost 1.19 X 10" atoms of tung- 

 sten in 12 hours as measured by the change 

 in the cold resistance. The number of elec- 

 trons emitted for each atom of tungsten lost 

 was thus '-1.13 X 10^ and the mass of the emit- 

 ted electrons about one third of the mass of 

 the tungsten lost. This tube ran altogether 

 for about 23 hours, giving various currents, 

 and finally gave out, owing to the local loss of 

 material near one end, caused by the sputter- 

 ing or evaporation. Local over-heating is very 

 apt to occur in these experiments as the ther- 

 mionic leakage causes the heating current in 

 the wire to be bigger at one end than the 



other. The mass of all the electrons emitted 

 by this filament was equal to 4 per cent, of 

 its total mass. Under a low-power micro- 

 scope the filament did not appear to be much 

 changed except in the region where it had 

 burnt out, where it was much thinner than 

 elsewhere. 



There is no known reason for believing that 

 the loss of tungsten is due to anything more 

 profound than evaporation. But, in any 

 event, the fact that the mass of the emitted 

 electrons can, under favorable circumstances, 

 exceed that of the tungsten loss proves that 

 the loss of tungsten is not the cause of the 

 electronic emission. 



4. The only remaining process of a similar 

 nature to those already considered which has 

 not been discussed is the bare possibility that 

 the emission is due to the interaction of the 

 tungsten with some unknown condensable 

 vapor which does not affect the McLeod gauge. 

 This possibility is cut out by the fact that the 

 thermionic emission is not affected when the 

 liquid air and charcoal is cut off and the 

 vapors allowed to accumulate in the tube, and 

 by the fact that very considerable changes in 

 the amount and nature of the gases present 

 (as by the admission of mercury vapor) have 

 no effect on the emission. 



Taken together these experiments prove that 

 the emission of electrons does not arise from 

 any interaction between the hot filament and 

 surrounding gases or vapors nor from any 

 process involving consumption of the material 

 of the filament. It thus follows that the 

 emission of electrons from hot tungsten, which 

 there is no reason for not regarding as ex- 

 hibiting this phenomenon in a typical form, 

 is not a chemical but a physical process. This 

 conclusion does not exclude the possibility 

 that, under other circumstances, electrons 

 may be emitted from metals under the influ- 

 ence of various chemical reagents, a phenom- 

 enon which would be expected to exhibit the 

 same law of dependence upon temperature; 

 but it does involve a denial of the thesis that 

 this emission is invariably caused by processes 

 involving changes of material composition. 



The exiDeriments also show that the elec- 



