( 653 ) 
cobalt have not been brought to a conclusion. It is difficult to know 
exactly the degree of accuracy of these results. Experimental work 
in every branch was carried out so that an accuracy of 1 in 1000 
or even higher could be expected. But when one considers the 
disturbing influences which made themselves felt in the experiments 
upon cobalt, it seems rather incautious — and this is particularly 
the case with the magnetite measurements — to ascribe to the 
results an accuracy greater than 0.5 °/ 0 , even though the occurrences 
which have thrown suspicion upon the cobalt measurements were 
nearly absent in the case of the other substances, and though in all 
its properties and particularly in its extraordinarily large magnetic 
hardness cobalt stands evidently alone. Since our experiments indicate 
these causes of uncertainty, they show how a higher degree of 
accuracy may be reached if so desired. The present accuracy is quite 
sufficient for the treatment of various problems. 
The experiments with iron and magnetite were carried to 14°,0 K. 
The change of magnetization between 20’ K. and 14° K. is too small 
to be expressed in figures. These experiments, therefore, only extend 
down to 14 K. the temperature region within which the diminution 
of the kinetic energy and the approach of the molecules to each 
other do not occasion the appearance of a single new phenomenon. 
The portions of the curves for nickel and magnetite which have 
been newly obtained are given by broken lines in fig. 1, Plate l. 
Magnetite is of particular importance on account of the perfect 
correspondence between observation and theory over the greatest 
portion of the region between the CuRiE-point and the absolute zero, 
and on account of the occurrence of a deviation of observation from 
theory only at low temperatures. Here, theory gives for the ratio 
between the magnetizations the value 1.139 instead of the value 
given above, 1.057. The result that theory and experiment clearly 
differ at these temperatures is corroborated by earlier experiments 
upon four samples of different kinds of magnetite, two obtained from 
natural crystals, the third from a fused natural crystal, and the 
fourth from artificial magnetite. These gave the following values for 
the ratio between the magnetizations at the temperature of solid 
carbon dioxide (— 79° C.) and ordinary temperature: 
1.033 ordinary temperature 16° C. 
1.042 23°.2 
1.043 24° 
1.037 21°.5 
mean 1.039 21°.2 C. 
while theory gives 1.053 for the same temperature. 
