1256 
Unfortunately, however, in about 50 out of 100 substances examined 
by Marnuws the rule does not hold good at all; while at least 37 
of the 74 substances investigated by us, hence also about half of 
them, deviate considerably. 
Marnews determined the values of a chiefly in two ways, first from 
the surface tension, reduced to the absolute zero-point, and secondly 
from the critical data. The two series of values of a did not differ 
much, from which M (loc. cit. p. 160) drew the conclusion that a 
is almost independent of the temperature. Tyrur ‘') came to the same 
conclusion. 
We found, however, by means of an accurate calculation that 
the two series of values do differ, indeed, and that the values of 
» are about 16°/, higher than aj (at least when the critical tempe- 
rature is not too low). 
We will not enter here into the details of the caleulations (loc. 
cit. p. 154 et seq.), nor into some theoretical considerations which 
seem very questionable to us (particularly those in the last Paper, 
loc. cit. p. 603 et seq.) *), but only mention that M has found 
1,50 X 10-2 (loc. cit. p. 183) for V/C as middle value, whereas we 
find 1,47 — 10-2 as mean value for those substances in our tables 
a 
for which the rule is more or less valid. 
In the numerous cas 
succeeds in finding means to make his rule hold good. He either 
s in which the rule does not hold, M always 
pronounces the most normal substances to be associated (even still 
at the critical temperature, where water, ethylalcohol ete. are already 
almost normal!), or he applies strange corrections to the valencies, 
and e.g. declares chlorine to be triwvalent in all the cases in which 
his rule does not hold good, while this element again falls back to 
its monovalent rôle in the cases in which his rule does apply. *). 
1) Z. f. ph. Chem. 87, p. 195 (1914). 8 
2) This will be more fully treated in my book on the Equation of state, which 
I hope, will be able to be published alter the war. 
3) In the same way M manipulates some numerical factors, e.g. the constant of 
the formula of Eörvös (resp. Ramsay and Sutetps) in order to establish a non- 
existing identity of the two series of values @ and ax. For the same purpose also 
v. p. Waats’ factor *7/g,.(or corrected by us to °7/¢,,, in which a is =1 for 
ideal substances, and for ordinary substances about = 0,977) was replaced by 
[s°—(s—2)] : s*(s—2) = nee 
AT Re s—2 g2’ 
a 
which is only correct in the limiting cases = 4 
(for substances with very high critical temperature), and N.B. does not converge 
to 27/4, for ideal substances (s = 5/s), but actually to §7/¢,! In the correct expres- 
sion for 27/g4,, viz. (f/—1): 5? he namely substitutes for //—1 the entirely faulty 
expression (s*—{s--2)]:(s—2}, which for s—5/, would not converge to 3, but 
