92 



cm 



5 



3 

 2 



110 electric lield, (he iiieiiii tVeo palli alone will determine their, 

 movement and distrihiition, so loiifj; as the velocity of the electrons 

 is not larger than that corresponding to the excitation potential, 

 that i8 : so long as the impacts are entirely elastic. If the apparatns 

 is then filled successively with difïerent rare gases, the movement 

 of the electrons in the one gas must he the same as that in the 

 other, provided the pressures are chosen in snch a way that the 

 mean free path is the same. If, on the contrary, the pressnres 

 of both gases has been adjnsted so as to make the movement of 

 the electrons the same, the inverse ratio of the corresponding 

 pressnres will give the reqnired ratio of the mean free paths under 

 equal pressure. This ratio must be found to be independent of the 

 pressure used in the experiments. 



The apparatus used is shown in fig. 1. 6r is a 

 tungsten filament, ^i and JSf^ are grids P is a 

 receiving plate, and H is a metal shield which 

 prevents electrons from coming from 6^ to P by 

 any other way, than through the space between 

 > the two grids. All metal parts were made of copper. 

 ° Before mounting the apparatus they were treated 

 with nitric acid and showed a clean metallic surface 

 after the tube had been exhausted during 5 hours 

 at 400°. The gases used were so pure that no 

 non-elastic impacts, below the excitation potential 

 could be detected even by a very sensitive device. 

 Before the final measurements, preliminary mea- 

 surements were made with a simpler device, which 

 differed from that of fig. 1 by omission of the grid ^j. 

 Though the experiments made in this way do not 

 allow an accurate quantitative evaluation, the results 

 are given here briefly, as they show very simply and clearly 

 the different behaviour of neon and argon. During these prelimi- 

 nary measurements the entire apparatus was at earth-potential, 

 except the filament which was brought at a variable negative 

 potential, so as to |)roduce an accelerating electric field between 

 filament and gi-id. The electron-current passing on to the receiving 

 plate P was measured by a galvanometer. The measurement consisted 

 simply in noting the current as a function of the accelerating 

 potential between G and A^j, in neon and argon under various 

 pressures. In order to be independent of slow variations of the 

 current in the filament, a second galvanometer registered the total 

 electron current, and the quotient of the plate-current and the total 



Fig. 1. 



