(121 ) 



susceptibility with iiicreasinp; field, that is a co)inneiiceinimt of 

 satarailoii ^). 



Bismutli. It has been shown bv Curik and lalei- coniirnied by 

 HoNUA that for bismuth between ordinary temperauire and the 

 melting- [)oint the temperature coeflicient of the magnetisation may 

 well be represented by the formula 



The fact that at ordinary temperature we obtained an absolute 

 value of the susceptibility which is somewhat smaller may be due 

 to the presence of a small quantity of free iron which, on account 

 of the intensity of the tields used, will have no inlluence upon the 

 relative results ; indeed, for various kinds of commercial bismuth 

 which were not so pui-e as ours. Curie found the same dependence 

 upon temperature. The values which we obtained with our specimen 

 may quite well therefore serve as the basis of deductions regarding 

 relative values. With .:« :z= 0,00105, the mean of the values obtained 

 by Curie with samples which were not perfectly pure, extrapolating 

 to 20°.3 K. we should obtain /-.his = — 1-64. 10-^ , which is a 

 value somewhat greater than that given by expei-iment. It would 

 therefore appear that the lineai- change of the magnetisation of bis- 

 muth does not go on as the temperature falls, and this result is also 

 in agreement with the experiments of Fleming and Dewar, who only 

 went as far as liquid air. The two values which we obtained in 

 hydrogen boiling under reduced pressure seem to indicate that when 

 the temperature sinks as low as this, the susceptibility becomes pretty 

 well independent of the temperature; if the law of change were still 

 linear at these temperatures then between 20.°3 K. and 13.°9 K. one 

 should have to observe an increase of 0.7 "/„, while what was actu- 

 ally observed was i-ather a decrease. 



Hi/diV(/en. With a view to the determination ot the corrections 

 to be ap|)lied we made a measurement of the suspectibility of liquid 

 hydrogen, the result of which probably deviates less than JO "/„ from 

 the true value. This determination is rendered extremely diflicult by 

 the smallness of the density and of the volume-susceptibility of the 



') Willi rogaid lo the difference that was found belween llie c.onsUuilri fur the 



formula — — for solid and for li(iuid oxygen (Goinm. No. 1 l(i, April I'Jlt», ^ o), 



it may be remarked that cryslallisalion phenonH'na can inlluence llieni. The invesh- 

 gali(jn of the problems referring lo this point will be seen to be further compli- 

 cated when we remark that on freezing oxygen Uansparentsolidoxygen is first formed 

 which, al lower lemperalures, is transformed into an opatjue mass. 



