( 475 ) 



B to bismuth, and in which T' represents the potential, C a positive 

 function of temperature, P and N numbers of positive and negative 

 ions per unit of volume. 



The electric conductivity for bismuth has been represented by 

 RiECKE by a formula of the form 



7 = -Pi? f/) + ^^B Cn 



in which Cp and c„ are again functions of temperature. lu the 

 magnetic field the conductivity decreases. The simplest supposition 

 is DOW that Pjj and Nji decrease proportionally ; a thing which 

 would bo self-evident if Pb were equal to iVg, as in electrolytes. 



Let us further suppose, which is also very probable, that only 

 the quantities relating to bismuth vary in the magnetic field, then 

 a proportional decrease of Pjj and Nb appears to lead to a variation 

 of the difference V^ — F^ with a negative amount, and so to an 

 increase of the thermo-electric difference between bismuth and copper, 

 which is negative outside the magnetic field. If on the contrary 

 w(! should suppose that the variation of A^b is small as compared 

 with that of Pb, then the variation of F/i — F^ would become positive 

 and the thermo-electric difference between bismuth and copper would 

 decrease. 



The observations show that the variation of this thermo-electric 

 difference in the magnetic field corresponds to an increase, and so, 

 in consequence of the proportionality of increase of resistance and 

 longitudinal effect found by me, we should have to prefer the former 

 of the two suppositions. 



(May 12tli, 1899.) 



