95 



tlie gaspressnre in millimetres mercurj'. For llie evaluation of 

 these measurements the distribntion of the election velocities was 

 first measured in vacuo by means of a variable retarding Held with 

 the result, that, in consequence of the initial velocity of the electrons, 

 the potential gradient at tlie filament and the Volta-potenlial difference 



ZOVoU, 



between filament and grid, 0.7 volt had to be added to the applied 

 accelerating potential, in order to obtain the true velocity of the 

 electrons. For a series of electron velocities the logarithm of the 

 plate-current was registered as a function of the pressure in neon 

 and argon. A similar character of the curves in neon and argon 

 is to be expected, assuming that the method is correct, in such away 

 that for each velocity the proportion of corresponding pressures in 

 neon and argon (i.e. pressures giving equal plate-currents) is constant. 

 This is in fact the case for all electron-velocities up to 16 volts. To 

 show this, the curves so obtained for a number of velocities are 

 reproduced in fig. 6. The evaluation is simplified by the fact that 

 the first part of the curves is straight. From the slope of these 

 straight portions we can obtain directly the ratio of the corresponding 

 pressures and so also the ratio of the mean free paths of the 

 electrons. 



A condition for the correctness of the method here applied is, that 

 all collisions between electrons and atoms are absolutely elastic. By 

 reason of the very low efficiency of the non-elastic impacts below the 

 ionization potential in the rare gases this is no doubt the ease for 



