884 
For Helium Kamertincu Onnes (Suppl. 21) has found d= 0,15 
for liquid He. Hence s,* < (3,99: 0,15): 6.107%, or s5,° < 44,5 .10-%, 
therefore s, << 3,54.107°. 
If for 6 in the case of He the value 0,0007, given in Suppl. 21, 
is assumed, then 06, is about 0,007 : 1,12 = 0,000625, hence 
s, > (0,00062° 22412) : 6.10, ie. 5 23,35 . 10-74, or 5,5 2,86. 10-8. 
This value is evidently too small; perhaps the d-value, for which 
K. O. first (Comm. 1024) gave 0,00043, and which was later on 
raised to 0,0007, must be raised somewhat more. 
We therefore assume for He the middle value s,= + 3,2.10®. 
PERRIN’s value, viz. 1,7.10-® (loc. cit), is at any rate too low. 
It appears from the above examples, that for three substances 
which differ so much as Argon (mol. weight = 40), Helium (mol. 
weight = 4) and the di-atomie H, (mol. weight 2), the values of 
s, differ very little. 
Also on calculation of other not too complex substances s, appears 
to rise very rarely above 4. 10-#. 
The found values are in better harmony than could be expected 
with the value + 3,9.10-8, which follows from our theoretical 
considerations. Not only is the order of magnitude the same, but even 
the numerical value is almost identical. 
When we bear in mind thats, .(d:s,)*—3,9.10 5, then (d: s,)> = 
—1,11 would follow from eg. s,=—= 8,5. 108 (for Argon), hence 
d:s,==1,04. The diameter of the molecule would therefore be a 
little smaller than the distance of equilibrium d of the two charges, 
which might point to a somewhat elongated form of the molecule, 
because then the mean diameter s, would be somewhat smaller than 
the distance of the centres, in which the charges may be imagined 
fictitiously concentrated. But though this supposition is very plausible, 
particularly for di-atomic gases, yet there is by no means certainty 
on this head on account of the not absolute accuracy of the calculated 
values. The more so as there may also be other influences at work, 
of which we only mention that of the degrees of freedom by which 
the factor f is influenced; and also the influence of the deviation 
from the law of equipartition, through which in (3) the factor / 
would apparently become smaller. But even when we leave the 
factor (d:s,)* in (7a) out of account, the concordance between the 
value of s,, calculated from the assumption of electrical forces and 
the values found for different substances, in connection with VAN 
per Waats’ equation of state of the molecule and the temperature 
coefficient of (,—6,):6, found by me, remains remarkably close. 
Fontanivent sur Clarens, October 1914. 
