on the Thermoelectrical Properties of Bismuth. 293 



Consequently it would seem to follow, from Grimaldi's experi- 

 ments, that both pure and commercial bismuth are rendered by 

 magnetization negative to the unmagnetized metal, i. e. the 

 thermoelectrical current would flow from unmagnetized bis- 

 muth to magnetized bismuth through the hot junction. 



Grimaldi shows that, in the following respects, the effect of 

 magnetization on the thermoelectrical properties of bismuth 

 resembles the effect of magnetization on the electrical con- 

 ductivity: — 



(1) The amounts of the two effects are comparable with 

 each other. 



(2) Transverse magnetization produces a greater effect than 

 longitudinal magnetization. 



(3) Rise of temperature* diminishes the effect. 



(4) The effect increases in greater proportion than the 

 magnetizing force. 



It is impossible for the author to compare his own results 

 with those of Grimaldi ; but it would seem from the above 

 that, by using high magnetizing forces and lower temperatures 

 at the junctions, the effect of magnetization on the thermo- 

 electrical properties of bismuth might well be found to be 

 some three or four hundred thousand times the effect in iron, 

 supposing both metals to be magnetized to unit intensity. 



This being the case, it is difficult to believe that the altera- 

 tion due to magnetization is a real alteration of the thermo- 

 electrical power of the metal. But, again, how are we to 

 account for it. According to Ettingshausenj, when a plate 

 of bismuth, AB (fig. 2), is arranged, as for experiments on 



Fig. 2, 



HalFs phenomenon, with its plane parallel to the flat faces of 

 the pole-pieces of an electromagnet and perpendicular to the 

 lines of force, whilst a current of electricity is conducted longi- 

 tudinally through the plate, the excitation of the electromagnet 

 produces a difference of temperature at the two points a and b; 

 whilst, on the contrary, if a current of heat be conducted 



* At least as far as 100° C. 



f Annalen der Physik und Chemie Band xxxi. (1887). 

 Phil. Mag. S. 5. Yol. 25. No. 155. April 1888. X 



