478 M. R. Kohlrauscli's Theory of the 



On the right-hand side of the differential equation, therefore, 

 we place a power /'" of the time, and as, by trial, it was found that 

 the exponent n on this side must necessarily be unity*, we have 



I. ^AP^lpJl = -bt-[p(^,-r^. 



Integrating between the limits /=0 and t = t, and bearing in 

 mind that when t = 0, »',=0 and Q^ = 0, we have 



II. \o^p3lZIi= ^ 



and 



III 





Provided the principles from which this equation has been 

 deduced are correct, and proper values be given to the con- 

 stants p, m and h, we may calculate residues which ought to 

 agree pretty well with the observed ones as recorded in § 4, 

 Tables a", b" and c". 



It is not difficult to find approximate values for these con- 

 stants. In the first place let us determine p. When the expe- 

 riment has already continued for a considerable time, the state 

 of equilibrium will be nearly reached, that is to say, the residue 

 will not difiier much from the limit which it is possible for it to 

 reach with the charge then present. Thus approximately we 

 shall have 



and p will not be much greater than -^. For example, the last 



determination in the Table i" gives 



i\ _ 0-4888 



Q, ~ 1-0154' 

 and we may assure ourselves that the value jo = 0-5 is not far 



/)Q( — r^ in the above, he has represented by x, and hence, according to him, 



dxz^ — hx^dt. 

 An attempt to construct the curve from the equation 



rf«= — bxf^dt 



gave me less correct results. This attempt, however, might possibly have 

 been more successful had the constants been otherwise determined. 

 Without repeating the calculation I will not venture to decide the point. 



* It would lead us too far to explain this, nor is it nucessai-y, seeing that 

 it does not interest us to know what equations are inappUcable. 



