( 120 ) 



tiires at which the de^'iations become appreciable differ considerably 

 for the different substances. It is notewortiiy that, in the case of 

 gadolinium sulphate, this temperature lies very low, and is probably 

 only 15° K. 



Concerning the behaviour of these substances at temperatures 



below that at which this deviation becomes appreciable, //^?^zö? oxygen 



can, in the first place, give us some information. In fact in this case 



<■ 

 another well defined law, viz, — has been found to obtain over a 



l/T 



pretty wide range of temperatures ''90° K. to 63° K.). At the lowest 

 five tem[)erature8 reached the magnetisation coefficient of ferrous 

 sulphate increases somewhat more rapidly than would be in 

 accurate agreement with this law, bul it undoubtedly approximates 

 to it. At the four hydrogen tem|)eratures at which measurements 

 were made dysprosium oxide follows the same law as liquid oxygen. 

 Taking all this into account we are only justified in assuming that 

 for all substances there exists a certain range of temperatures differing 

 greatly for the different substances, for which this law to a high 

 degree of approxiuuition governs the dependence of susceptibility on 

 temperature. 



c 

 In the neii>hboui'hoo(i of 15° K. solid oxygen did not follow the 



law, but deviated still more st rongly from Curie's law, and even 

 showed a tendency to attain a constant value; further experiments 

 which have not yet been completed, confirm this result. As none of 

 the three other substances have yet exhibited this phenomenon it still 

 remains an open question if it is characteristic of oxygen only, or 

 if one only needs to lower the temperature sufficiently to be able 

 to invoke its appearance as a general property of all paramagnetic 

 substances. 



On the occasion of oui- research uj)Ou oxygen we remarked con- 

 cerning the theoretical explanation of the increasing deviation with 

 falling temperature that it might, amongst other causes, be due to 

 association amongst the oxygen molecules. Some experiments which 

 are still incomplete upon mixtures containing equal quantities of 

 OX}' gen and nitrogen seem to exclude this explanation, but these 

 experiments must be continued further to be quite conclusive. From 

 the experiments which have already been concluded, however, we 

 have sufficient reason for not reverting to this explanation in § 1. 



We must further note that with none of the paramagnetic sub- 

 stances investigated we have been al)le to observe a decreasing 



