913 



which agrees with the law following from Table IT, if we assume 

 that L or ^\'() decreases nnth the densUij. 



This being established, the experiments further permit, and this 

 gives them an additional significance, to account for the uxuj in 

 which the molecidar field changes mith the density or icith the distance 

 of the molecules. 



We remark in the first place, that the accurate calculation of the 

 molecular fields rests upon the knowledge of the deviations from 

 Curie'.s' laiu, and not upon that of the susceptibilities themselves 

 [formula (1)]. It would therefore be necessary to know the specific 

 magnetisations in absolute value down to at least 0.1 7o i^^ order to 

 be able to deduce the fields from them with sufficient certainty ; 

 this is especially the case for the great dilutions where the deviations 

 are extremely small. A determination of this degree of accuracy 

 demands in itself a long and difficult special investigation with 

 perfected apparatus. 



The solution, if not completely found, may yet be brought within 

 narrow limits. The first question that then arises is, whether the 

 field is equal to a particular power of the distance of the molecules 

 i.e. of the density. If we suppose ^=ao", then the molecular field 

 is n'(>"+^ At constant temperature \/y_^=f{Q) is then a parabola of 

 the degree n -\- 1, with the axis vertical and the top on the axis of 

 ordinates. For n = 0, Vz is represented by a straight line. 



ox A'' 



0,J 



0.8 



1,0 



1.1 



1.1. 



Fig. 4. 



In Fig, 4, the curve wiiich our experimental data give for the 

 temperature 77°.'45K. is shown; it deviates from the straight line, 



