466 



^P=fAg-^yM ....... (ir) 



in which equation 11 1 must be valid for tubes the diameter of 

 which is large with respect to tlie free path of the corpuscles, which 

 is actually the case in Rüttenauer's experiments. It is seen that 

 equation 111 is identical with Rüttenauer's empirical approximative 

 formula. 



\ 4. On the influence of the potential gradient on the pressure e (j^ect. 



The "empirical" introduction of the potential gradient by Rüttp:n- 

 AUER in the pressure effect equation rests on the testing by three 

 kinds of observations : 



a. the dependence of the observed values of the pressure effect 

 on the potential gradient with one and the same current tube and 

 the same gas with different current densities. 



b. the dependence when the diameter of the discharging tube is 

 varied ; 



c. the dependence when the nature of the gas is changed. 



With reference to a, we must remark that a critical consideration 



of the values published by Rüttenauer shows that the variation of 



the gradient is irregular, and besides smaller than the deviation of 



p A p 

 the values found for -7 — , ' ^. interse, which values should be appro- 



Ag.ly/M ^' 



ximately constant. Thus in table 4 p. 272 of Rüttenauer's [)ublication 



the gradient for argon varies e.g. between 1,87 and 2,36, while the 



p A p 

 "constant value" — -^r. as a maximum varies between 3,70 and' 



Ag.tVM 



4,98. For the rest the uncertainty in the determinations of the 

 gradient seems to have been considerable. Where in our Thesis the 

 gradient decreased with increase of current density, it appears to 

 increase in a slight degree in the investigations recorded by Rütten- 

 auer in table 3, whereas it decreased in a great degree in table 5. 

 For this category of cases a constant value had, therefore, better 

 be substituted for g, and the empirical formula becomes identical 

 with our theoretical equation (/). 



This is, therefore, in harmony with the statement expressed in 

 our former publication (loc. cit. p. 390) that "in the case that the 

 nature of the bearers is not modified" (hence for a definite gas) and 

 "with not greatly varying tension" (potential gradient) <7 is a constant. 



With reference to case b we remark that the experimentally determ- 

 ined influence of g is unmistakable. So far as the consequences of 



