622 



given and in the last column tlie ditference (in percents) of lO'C 

 with the mean value 2,030. 



.o'H 



Fig. 5. 



?• was determined from experiments on the attraction of two 

 small ellipsoids ') of Swedish Carbon iron placed as well as 

 possible at the same spot as the substances in tlie actual experiments. 

 Use was made of the measurements of Steinhaus und Gummch ') 

 on the relation between field strength and magnetisation when satu- 

 ration is nearly reached, tlie so called law of approach 



s was calculated from formula (5). The values found for r and n 

 have also been smoothed graphically '). 



In these determinations the distribution of magnetism on the pole 

 faces of the magnet has been supposed to be perfectly rigid '). 



1) Masses 30.0 and 32.0 mg., major axis 6.2 mm., minor axis 1.1 m.m. 



') Ber. d. Physik. Ges. 17 (1915) p. 271. 



') This causes the product of the given r and s to be not exactly equal to q. 



*) Cf. P. Weiss, J. de Phys. May 1910 and P. Weiss and H. Kameblingh 

 Onnes, Leiden Comm. N". 114, p 16. 



Strictly speaking: for a magnet current of 15 amp. the distribution of magnetism 

 on the pole faces of the magnet has been supposed to be perfectly rigid and as 

 regards the other current intensities it has only been supposed to be the same 

 for gadolinium sulphate at hydrogen temperatures and for the S.C. iron ellipsoids. 

 In fact, the magnetic moments are of the same order in both cases (though the 

 volumes on which they are distributed are different); in the case of gadolinium 

 sulphate at helium temperatures they are much larger, yet the same values of r 

 and s have been applied (cf. next communication) [Note modified in the translation]. 



