94 



elecii-oii ('iirreiil fVoin llie tiltiineiit was calculated. As llie tempe- 

 rature of tlie filament was alwajs low, this quotient was independent 

 of the intensity of the electron emission of the filament. 



This (piotient, multiplied by a constant is |)lotted in the curves ot 

 the figs 2 ami 3 for a series of pressures in neon and argon as a 

 function of the potential difference between ^r and .Yj. The numbers 

 near the curves show the gas pressure in m.m. mercury. Wo see 

 immediately the extra-ordinary difference in the behaviour of both 

 gases. While in neon an increase of pressure foi' all velocities 

 reduces the plate-current in about the same degree, argon shows at 

 10 volts a remarkable deoi-ease of current at pressui'es, where at 

 1 volt practically no influence is observed. As the observed decrease 

 of current can result oidy from the collisions between the electrons 

 and the atoms of the gas, we can deduce from these measurements 

 rpialitatively, that the mean free path of electrons in argon varies 

 strongly with the velocity of the electrons, while in neon this is 

 not the case, or at any rate only to a small degree. A quantitative 

 calculation in the sense of the above consideration can only be 

 taken from these measuiings for slow electrons up to about 10 volts; 

 at higher velocities the electrons produce secondary electron emission 

 from the metalwalls. To retain these secondary electrons, the second 

 grid N^ was introduced a retarding potential equal to \\ of the 

 accelerating potential between G and iVj being applied between 

 JSf^ and P. The result of such series of measurements is shown in 

 figs. 4 and 5 wherein the numbers near the curves again show 



< ) Voctjum 



ZOVo'.k. 



