( 668 ) 
the dividing-engine have shown that the ellipsoids are very accurately 
shaped. 
The iron was obtained by melting pure electrolytic Merck iron 
contained in a magnesia boat in an electrical resistance furnace and 
in an atmosphere of nitrogen. The nickel and cobalt were prepared 
in the same way, starting with the purest possible nickel and cobalt 
powder specially prepared by Merck for these experiments. The 
magnetite was obtained by constructing an approximate ellipsoid 
from a drop of very pure magnetite obtained by melting very pure 
Merck sesquioxide in an oxy-hydrogen flame. Since experiment showed 
that it was only at very high temperatures that the last trace of 
oxygen was driven oat and real magnetite 1 ) obtained an iridium 
cupola was used for this operation. 
Ellipsoids of approximate shape were also constructed from Gold¬ 
schmidt chromium and manganese and Moissan vanadium. As can 
easily be seen it is not necessary for comparative experiments that 
the ellipsoids should be constructed with particular accuracy. This 
was, moreover, experimentally demonstrated for magnetite, of which 
various samples roughly worked to various ellipsoidal shapes were 
used for obtaining curves for the thermal change at high tempe¬ 
ratures, and these curves were in agreement with the theoretical 
curve, and consequently with each other. 
§ 3. Experimental method. As mentioned in the introduction our 
aim was not to obtain absolute values for magnetization in strong 
fields at ordinary temperature and at the temperature of liquid 
hydrogen, but to compare the values at these temperatures; for we 
might expect that the change would be only a small fraction of the 
quantity to be measured. Hence it was an obvious procedure to 
make observations at these temperatures alternately in the same field. 
The change, however, from the one temperature to the other neces- 
sitated operations of such duration as to prohibit the use of this 
method. Hence we usually began with a series of measurements at 
ordinary temperature, in which the field was made the required 
senes of strengths. Then an analogous series of measurements was 
Ztf V° W tGmperature ’ “d after the apparatus had returned to 
ordinary temperature, some individual measurements were repeated 
so as to make sure that the apparatus had not in the meantime 
undergone any change. 
') See also P. Weiss. Arch dos Sc.riivs etna! 
4e S6r. t. IX mars 1910. 
t. fevr. 1910 and Journ. de physique, 
