970) 
and on the hypothesis that in the liquid two gas molecules are 
rigidly connected it gives 11 per mclecule of two atoms. 
From 4 (T+ A’) = 38600 (the mean of the numbers in the table) 
with A’ = 71° one finds for 72957 K. 
Hoos == 106.0 TOE 
This is very close to the value for gaseous oxygen at 20° C found 
by Werss and Piccarp 5), from which follows 7 magnetons for each 
of the oxygen atom assumed to be rigidly connected. 
Seeing that above 20° C. gaseous oxygen follows Curin’s law ’”) it seems 
to be by some chance that our formula with 4' = 71° gives that figure. 
The graphic representation of '/y as a function of 7’, if our for- 
mula actually remained true up to 20° C. would consist of two 
intersecting lines that have their point of intersection just at the 
temperature at which the value quoted is determined, which cer- 
tainly would be a curious coincidences. 
Another possibility which Prof. Werss suggested,in a kind private 
communication, is that there might be discontinuity in the region 
between 0° C. and — 188° C. which has not been investigated, by 
which it remains accidental that the continuation of the line for 
liquid oxygen cuts that for gaseous oxygen just at 20° C. There 
is much to be said for this explanation. It is quite possible that the 
change of density between liquid oxygen and gaseous oxygen 
makes A’ into 0. This would be in accordance with what was 
deduced in § 10 for the influence of the water molecules upon the 
value of 4’ for manganese sulphate, and moreover quite in accor- 
dance with Weiss’s idea that the molecular field essentially depends 
upon the density. 
We can further observe, that the change of density, which takes 
place discontinuously with evaporation, can take place continuously 
by an indirect transition. In the above line of thought, if we assume 
that the divergence for liquid oxygen from Curir’s law may be 
defined by a 4’ and pay attention to the change of the number of 
magnetons which must be assumed in that case, the graph which 
represents ‘/, for oxygen of a given density as a function of the 
temperature would be as in magnetite a succession of straight lines 
perhaps connected by rounded off pieces. The magnetic equation of 
state which expresses the susceptibility as a function of density and 
1) P. Weiss et A. PrccArp, ‘C) R.1155, 9p, 1234, 1912. 
*) Prof. Weiss who has particularly investigated this question, kindly tells us 
that the experimental results of Currie agree so well with Curtn’s law within the 
limits of observation errors that a’ could not be more than + 8° or—8°, 
