548 
MR, J. S. TOWNSEND ON MAGNETIZATION OF LIQUIDS. 
’105 grm. of iron per cub. centim., and the 4th from an alcoholic solution of ferric 
chloride containing '127 grm. of iron per cub. centim. 
The curves show that there is a great diminution in k as the temperature rises, 
amounting to more than ’5 per cent, per degree centigrade at the lower temperatures. 
The curves show that k = k Q ( 1 — at), where a is a function of the temperature, 
and is very approximately independent of the acid radical. 
ddie diminution in density of the salt due to expansion contributes such a small 
amount to the rapid fall in k that it is not necessary to make any correction for it. 
The solvent itself did not show any variation due to temperature which could be 
detected by this method, and as the alcoholic solution gives the same temperature 
coefficient as the solutions in water, it is probable that the change due to temperature 
is a property of the molecules of the salt itself. 
The first conclusion that the above results leads us to is that the magnetization is 
entirely due to the iron, and is accurately the same for all acid radicals connected 
with it, so long as the iron remains either in, the ferrous or the ferric state. The 
only other electromagnetic phenomenon we know of that also possesses this property 
is the atomic charge, so that possibly the polarity of the metallic atom may be due to 
its rapid rotation carrying with it its atomic charge. The ratio of the atomic charge 
in the ferric to that in the ferrous state is 3:2. There is a variety of suppositions 
as to the nature of the rotations of the iron atom in the molecule which will account 
for the ratio 266 : 206 of the two values of k. 
A simple case presents itself by supposing the intensity of the polarity proportional 
to q x oj 1 and q 2 o). 2 in the two cases, q Y and q, 2 denoting the charges, and Wj and on the 
rotations, the axes of which coincide with the axis of polarity. In this case, if H is 
the applied force, we get the displacements proportional to and q 2 co 2 H, so that 
the ratio of the values of k is : q 2 u 2 — 266 : 206, so that cj 2 = l , 32w 1 . 
The next consideration is the controlling force, which acts in such a way that the 
induced magnetization is proportional to the applied force. It cannot arise from any 
action of the solvent, since the magnetic properties of the salts are the same in the 
dry state as when they are in solution. Also, since k is accurately proportional to 
the density, we see that each molecular magnet behaves in exactly the same way 
whether the neighbouring magnets are near it or not. 
It is, therefore, highly improbable that the surrounding molecules contribute to the 
controlling force, as there is no variation in it when the mean distances undergo large 
changes. 
We are, therefore, led to the conclusion that there is no controlling force due to 
external bodies, and that the magnetization is a phenomena due to the perturbations 
in the angle of inclination of the axis of polarity to the direction of force H. 
Let 6 denote this angle, which is a function of the time, 0 = 0 O + 80 where 0 O is the 
value of 0 when H ■= 0, and 80 the perturbation at any time due to H, and is propor- 
