900 



niodilicalioii of oxv^eii to llie ()|):i(|iie, which lakes phice at iiboul 

 — 225° C, is nol (((•(•nmi>an,i.i'<l Inj niui conspicuous nHU/nclic inodi/i- 

 Ciition., so Ihnl il probably does nol iii\()l\(» any iiiiporlaiil change 

 in inolecnUir sli-nclni-e. Fnrtlier, that in Iho wiiolc range IVoni the 

 free/.ing point of oxygen down to al)ont - 240° (J. Ihe snsce|)tibility 

 is markedly less than in the li(piid stale. In this region it changes 

 little Avilli the leinpn at lu'c, in lad increases slightly on cooling. 

 This does not seem to agree with Ihe ninch smaller valne which 

 the table gives foi' 253° (J. bnt lieic a pli(;nomcnon intervenes 

 vvliich has not been ob.served before, vi/. that at about — 240° C. 

 the susci'plihililii swhh'nlji j'olls to ulioiil liolj' its value. The exact 

 value of the temperatnre must still be more accnrately determined; 

 it is also not settled whether the transition is actually discontinnons 

 or is completed in a very small range of temperature ^1 or 2 degrees). 

 The susceptibility falls in this iran-^ition to about the values that are 

 found in li(pii(l hydrogen. 



It will be seen that these last values agree with the values of 

 J9T1, within the limits of accuracy of the latter. The results of 

 'J 1)10 arc thereby, as we expected, condennKMl, and the cause to 

 which wc altribnl(Ml the discrepancy between tlu^se residls and 

 those of J910. gains in probability at the same lime. 



As regards the change in the susceptibility below - 252° C. with 



the leni|)eralure, although, it is slight, yet il allows no doubt that 



it is a decrease. 'V\\v obserxations permit Ihe compai'ison of the 



susceptil)ility at dilferent temperatures at constant density without 



the intervention of any correction. We have here, therefore, a 



new example of a sui)stancc ihal follows Ci'rik's law at a higher 



temperature, and on ap|uoachiniz the absolute zero completely 



deviates from il. As the ratio of the susceptil)dities observed above 



and below the transformation point is 



115.8 



X, = - = 2.10, 



54.2 



it becomes jirobable that the exact value of the jump is precise!}' 



2. For above 20° K, the magnetisation must still inci-ease dislinctl}'. 



Kor the ratio of the susceptibility of litpiid oxygen and solid 



oxygen at — 230° C. we find 



3i3.3 



115.3 



This value is less certain than that of X^, for it contains the 

 uncertainty concerning the density of solid oxygen and that of the 

 change of susceptibility between — 220° C. and — 230° C. 



Fi'om our j)revious investigations we had inferred a sudden change 



