634 



U cannot be denied tliat while on the one hand, one gets tiie 

 strong impression that Langevin's formula is followed (cf. the figure, 

 in which the Langevin curve and the observed points have been 

 drawn), on the other hand the deviations are larger than was anti- 

 cipated. However the^y maj' be explained from the sources of error. 

 Besides all that has been said in the preceding communication as 

 to the accuracy, it must be pointed out that the larger deviations 

 occur especially at the lower field strength values, where the topo- 

 graphical corrections are rather uncertain and also the measurements 

 of the field strength less reliable. Further, the magnetic moment 

 acting at the very low temperatures is so large that the assumption 

 of a rigid distribution of the magnetism on the pole faces (and on 

 this assum|)tion the field measiiremeiitH and the determination of 

 the topographical corrections are more or less based) certainly holds 

 no longer. 



Moreover it must be observed, that errors in a^ and in //.exert 

 on the abscissae an intluence opposite in direction to that on the 

 ordinaten and thus appear greater in the diagram. Taking all 

 these circumstances into account, especially also the uncertainty of 

 the demagnetisation, it may be concluded, that powdered gadolinium 

 sulphate follows Langkvin's formula down to about 1°.3 K ; thus it 

 seems possible to use the magnetic susceptibility of gadolinium sul- 

 phate in thermometry. 



^ 4. Results. The specific magnetisaliou of powdei'ed hydrated 

 gadolinium sulphate has been imesligated for (he temperatures of 

 liquid hydrogen and liquid helium. It appears that though the fun- 

 damental assumptions to Langevin's theory do not apply, yet Langevin's 

 formula is followed. For the parameter n of Langkvin's theory the 

 value 7 has nearly been reached. The highest magnetisations obtained 

 are about 84 '/„ of the magnetisation corresponding to perfect paral- 

 lelism of all elementary magnets. This result is independent of the 

 uncertainties in the temperature and the value of the demagnetising 

 field. vSo it appears that Prof. Ehrenfest's theory is here not 

 applicable without further extension, since this theory (which is based 

 on quanta assumptions and holds, contrary, to Langevin's theory, 

 directly for crystal powders) gives for the saturation magnetisation 

 only 50 7. of the value mentioned. 



