1010 
then (9) is transformed to: 
af eV dy — a e-¥ dy +... ad inf. 
/; V; 
l re 3l a 
4 
M= N— oD 
Vx 
hence: 
=. 3 al? gal? 
P(t) dt =2l Ed ej t—38e t +... ad inf} dt . (10) 
7 
This series is convergent for all values of ¢. 
If we now put: 
al? ; 9al? : ‘ 25al? 
Chern ==: En 
; MS Tet ene A. etc. 
then we get: 
P(td=— pee dz, — e—*" dz, +... ad inf}. » (LI) 
Our purpose is to determine a. 
1 
From (10) we can calculate the mean value of —. 
We tind for this: 
Si (= Pad int) : 
a 
If we represent the sum of the series between brackets by /, 
and put: 
1 1 
DAs 
then : 
BE z th 
and 
pe Be (12) 
— of th . . . . . . . . . 
The 
index A annexed to ¢ denotes that we mean the times of 
displacement. f is to be approximated with an arbitrary degree of 
accuracy. For our purpose suffices f = 0.916. 
The observations give ti, hence A? can be calculated. 
3. I shall briefly describe the apparatus which I used for my 
observations.') I assume the method of MirrikaNn and EHRENHAFT 
to be known. 
1) Compare for a full description my thesis for the doctorate, which will 
shortly appear. 
