915 



might even see in this a proof that in the two cases influences are at 

 work which are the effect of different causes ^). 



If we continue the idea tiiat the deviations from Curie's law are 

 to be attributed to a negative field, we come with the value 0,03152 

 which we deduce for Curie's constant, to 14.11 magnetons per 

 molecule of 2 rigidly connected atoms, while Weiss and Piccard, 

 from their determination of the constant for gaseous oxygen arrive 

 at exactly 14. An error of 0.2 degrees in the lowest temperature 

 at which we made observations would explain this difference. So 

 there is no reason to take this observation as at variance with 

 the law of magnetons. [At the end of § 4 we have drawn attention 

 to a circumstance that would possibly explain the difference. (Added 

 in the translation)]. 



Although the hypothesis of the negative molecular field is sufficient 

 to describe the phenomena, it is not devoid of interest to consider 

 in how far the other hypotheses can be reconciled to the observations. 



As regards the polymerisation-hypothesis, it is not probable that 

 the association decreases as rapidly with the density at constant 

 temperature as would have to be the case if the experiments were 

 to be explained by it, nor that the result of a given form of the 

 hypothesis, which might give this change, would lead to the set of 

 parallel straight lines in fig. 2. 



OosTERHUis's hypothesis, that the energy of rotation should take 

 the form of 



hv hv 



must he supplemented by a further hypothesis. It is in the line 

 of the deduction to accept that the moment of inertia of the molecule 

 changes considerably by dilution, and increases to very great values. 

 Even in accepting this hypothesis it appears from calculations by 

 Dr. Oosterhuis that the deviations of the curves calculated from the 

 set of the parallel straight lines in fig. 2 are too large to be explained 

 by experimental errors ^). 



1) The function pVx, which was given in a preliminary note (Soc. Suisse de 

 physique, Frauenfeld 1913) was ha^ed upon data which later appeared to have 

 been insufficient in number, and before some subsequently calculated corrections 

 had been applied. The function may be correct for a range of weak concentrations. 



~) In these calculations Oosterhuis started from the values of i. which were 

 deduced immediately from the experiments according t(j Table II, which in partic- 

 ular gives the most probable slope of the straight lines accepted to be parallel. 



