Finall}' we may remark that the supposition 



V,=../(7') + iVo (3) 



combines the various hypotheses which we have explained above 

 (except that ot" polymerisation"), and agrees with our experiments 

 if f{T) is independent of the density, and A' of the temperature ^). 

 It accounts for all phenomena which are expressed by a })arallel 

 displacement whatever value is ascribed to f{T). if only it remains 

 independent of the density. 



[One can imagine tliat ƒ (7') is in reality of such a form that in 

 the temperature-raiige of our experiments it gives for x of the oxygen 

 in the li(pu(l mixtures of oxygen and nitrogen a value which, while 

 being small for the large densities of -oxygen, increases in passing 

 to more and more high dilutions, and approaches in the limit to a 

 value qreater than that which with Cukik's Uiw is obtained from 

 Weiss and Piccard's experiments. If furthei" experiments which may 

 be undertaken with a view to throwing light upon this point should 

 give a positive result, one could explain without relying upon expe- 

 rimental error the slope of our lines for ^jy^ being a little small 

 (given by 6'= 0,0315, see tig. 2). it would then be of interest to 

 compare that result with what the investigation by Kameri.ingh Onnks 

 and OosTERHUis on the susceptibility of gaseous oxygen at low tempe- 

 ratures will teach when extended to lower temperatures. It is not 

 entirely excluded that the nitrogen plays a different part from a 

 vacuum. Added in the translation.] (Of. iiote 1 p. 91 5. Added in 

 going to press). 



Since then Dr. Keesom has communicated to us that by calculating the values of a 

 in giving to f <,T) [see formula (3)] either Oosterhujs's form, or the form wh'cli 

 Keesom has developed in Suppl. N\ 32a, whereby the values of L are a little 

 changed, one obtains a satisfactory representation. The values of a are then in 

 agreement with the function ',.-/j for large values of c, which he arrives at (see a 

 paper of his to be published shortly). In his calculations Keesom accepted the 

 number o( magnetons of Weiss and Piccard. His theory explains also that the 

 curve i/V. ~fi) of lig. 4 does not pass through the origin. [Note added in 

 going to press.] 



1) f{T) represents^the energy of rotation as a function of the temperature. (3) 

 is only applicable to the range where the magnetisation is strictly parallel to the 

 field. Formula (3) includes inter alia the vanishing of the Curie point in ferro- 

 magnetic substances caused by zero energy deduced by Keesom (Leid. Com. Suppl. 

 N'. 32 a and & (1913). 



[Regarding the influence of the density on f {T) in Keesom's theory see his next 

 paper, cf. also note 2 p. ü of Suppl. W. 32a. Added in going to press.] 



