July 11, 1913] 



SCIENCE 



59 



of origin of the electronic emission which are 

 enumerated below: 



1. The emission is due to the evolution of 

 gas by the filaments. 



The lamp and McLeod gauge were cut oS 

 from the rest of the apparatus by means of a 

 mercury trap, the volume being then approxi- 

 mately 600 c.c. A filament 4 cm. long giving 

 a thermionic current of .064 amp. was found to 

 increase the pressure from zero to < 1 X 10"' 

 mm. in five minutes. The number of mole- 

 cules iVj of gas given off is therefore 

 < 2.13 X 10". The number of electrons given 

 is N„ = 1.2 X 10°°- The number of electrons 

 emitted for each molecule of gas evolved is 

 thus N„yN^ > 6.64 X 10°. 



In the above experiment a liquid air trap 

 was interposed to keep the mercury vapor off 

 the filament. In another experiment with a 

 filament 8 cm. long this was not the ease and 

 with a current of .050 amp. the pressure rose 

 in thirty minutes to a value which was too 

 small to measure, but which was estimated as 

 less than 10"' mm. The corresponding value 

 of N,yN^ is 2.6 X 10'. In this case the cur- 

 rent was unaffected when the mercury vapor 

 was subsequently cut off by liquid air (a 

 change of 0.4 per cent, would have been de- 

 tected). 



The magnitude of the above numbers ef- 

 fectually disposes of the idea that the emis- 

 sion has anything to do with the evolution of 

 gas. 



2. The emission is caused by chemical action 

 or some other cause depending on impacts be- 

 tween the gas molecules and the filaments. 



In a tube with a filament 1.4 cm. in length 

 and having 1.65 X 10"^ cm^ superficial area 

 the pressure rose to < 2 X 10"° mm. in 5 min- 

 utes with an emission of .050 amp. If the gas 

 is assumed to be hydrogen, which makes most 

 impacts, using a liberally high estimate of the 

 temperature of the copper electrode which de- 

 termines the temperature of the gas, I find 

 that the maximum number N^ of molecules 

 impinging per second during this interval 

 would be < 7.0 X 10"- The number of elec- 

 trons emitted per second would be iVj = 

 3.13X10". The ratio NJJST is thus 



> 4.47 X 10"- If the putative hydrogen atoms 

 simply turned into a cloud of electrons whose 

 total mass was equal to that of the hydrogen 

 the value of NJN^ would be only 8.68 X 10". 

 The data already referred to for the tube with 

 the filament 8 cm. long give an even larger 

 ratio for NJN^, namely, 1.57 X 10'. Moreover, 

 in some of our experiments the changes in gas 

 pressure were much larger than those recorded 

 above, but they were never accompanied by any 

 change in the electronic emission : also the ad- 

 mission of mercury vapor at its pressure 

 (about 0.001 mm.) at room temperature pro- 

 duces no appreciable change in the emission. 

 Thus there is no room for the idea that the 

 emission of electrons has anything to do with 

 the impact of gas molecules under the condi- 

 tions of these experiments. 



3. The emission is a result of some process 

 involving consumption of the tungsten. 



To test this question some of the lamps were 

 sealed off after being exhausted in the manner 

 described. The filaments were then heated so 

 as to give a constant thermionic current which 

 was allowed to flow for long intervals of time. 

 In this way the total quantity of negative 

 electricity emitted by the filament was deter- 

 mined. The wire was placed in one arm of a 

 Wheatstone's bridge so that the resistance 

 could be recorded simultaneously. The cold 

 resistance was also checked up from time to 

 time. 



At these high temperatures the resistance 

 of the filaments increases slowly but continu- 

 ously. This increase is believed to be due to 

 evaporation of the tungsten. It was found 

 to be proportional to the time of heating 

 when the thermionic current was kept con- 

 stant, in the case of any particular filament. 

 In the case of one filament which gave 0.05 

 amp. for 12 hours the increase in the resist- 

 ance of the hot filament was 9 per cent. The 

 accompanying proportionate increase in the 

 cold resistance was slightly lower, namely, 7 

 per cent. The latter may probably be taken 

 as a fair measure of the amount of tungsten 

 lost by the filament. The increase in resist- 

 ance of the hot filament, which is less favor- 

 able for our case, will be considered instead 



