( 1225 ) 
a es reat ec AR le ORE LN 
8 
For as and f= 4, we find of course again sr = 8, but to 
this a value of a=0 belongs. With s = 3,64 and f= 6,6 we find 
for CO, the value sr=7,1, which value is smaller than I had 
expected. For ether, for which we may put s—= 3,77 and f= 7, 
we find sr little different from 7,1. Small errors in v and /, however, 
have a great influence on the value of this quantity. For r a value 
is found little higher than 1,88. That in my ‘Quasi association” I 
put sr little different from 8 also for substances like ether is, therefore 
owing to a too high value for r. If the value of «a is calculated from 
de Sin or from 1 — B rd a is found to differ 
r oi r? a ie 
little from */,. This result would be in perfect accordance with what 
the theory had predicted concerning the value of a in the approximate 
formula used for spherical molecules. But we find another value of 
a for another value of s and f. 
The relation between a and f is given by the formula: 
and by the aid of (VII) 
Bf—2)?_, Bf) 
ed ie 
ai2—s 
from which we derive: 
iy (eta, 8 € = 
2 CAR ag Oh eee 
s df fi (2 5) 
2—s 
ie 
d 8 
This value of En is equal to 0 for f= 4 and s=—, but for 
greater value of f and corresponding value of s it is always positive, 
as, indeed, might have been expected. It was, namely, to different 
variability of 5 with v that we attributed the different value of f 
and s. But the different value of a is still inexplicable. Is the devia- 
tion from the spherical shape the cause? And is, for the cases in 
3 aes \ 
which oa another cause, a real diminution of the molecule added 
to the cause assumed up to now for the decrease of 5? But the 
assumption : 
