902 



neiglibourliood of the boiling point or tlie Ireezing point of liydrogen, 

 they deviate from tlie hiw in various degrees'^), Init n/inn/s in the 

 direction of decrease of the susceptihiJitij. 



We thinlv it advisable to recall here in a few words some of the 

 views and hypotheses which guided ns in the above-mentioned 

 researches, in order to make clear the connection of the former ex- 

 periments to each othei-, and to the determinations we shall here 

 deal will), and to elucidate the object of these last. 



When we began our magnetic investigations at low temperature 

 (in 1908) we wished, amongst other things, to test Langevin's recently 

 published theory of paramagnetism which leads to Curie's law, at 

 low tempeiatures. 



On this account (and on others, see Comm. N". 116, § 1) it was 

 natura", we should begin witli oxygen. The deviations that we fonnd 

 in oxygen together with what we observed \\\ other substances, 

 gave us ground to suppose phenomenologically that there would be 

 a law of corresponding conditions for the deviations from Curie's 

 law. This again gave rise to the question of how L.\>gevin's theory 

 would have to be supplemented in view of the new phenouiena. 

 Our tirst idea was" the possibility of polymerisation, which might 

 take the form of association in oxygen. As the degree of association 

 of liquid oxygen if diluted wiih a non-active substance, would be 

 changed, and as dianiagnetic nitrogen could serve as such a substance, 

 we thought (see Comm. N". 116, § 5) that by experiments on mix- 

 tures with this substance we should be able to ascertain whether it 

 WAS, a case of polymerisation or not. 



An experiment made with a less satisfactory apparatus than that 

 which we now use, gave an indecisive result, and even led us to 

 a wrong conclusion, as it seemed to support our assumption of the 

 same change with temperature in the specific susceptibility of oxygen 

 independent of the distance that separated the molecules from 

 each other. In other woi-ds it still appeared possible to us that the 

 specific susceptibility of oxygen vapour at the boiling-point might 

 be the snme as that of liquid oxygen, and that therefore gaseous 

 oxygen at this temperature would de\iate from Curie's law to the 

 same extent as liquid oxygen. Our intention soon to make further 

 measurements on mixtures of oxygen and nitrogen, came to nothing, 

 as one of us left Leiden. It was a considerable time before we were 

 able to continue our experiments. 



In the mean time, Kameri.ingh Onnes and Oosterhuts's investiga- 



'*) Kamerllngh OiNNEs and Perrikr, Comm. Nos. 116, 122a, 124a. 



