HALL AND CAMPBELL. — MAGNETIC EFFECTS IN SOFT IRON. C47 



A look at the uppermost line of 'liable I shows that, for the tempera- 

 ture 20° C. at least, we have, very nearly. 



If we test this formula tlirough the various temperatures we find 



(IV.) 



The approximate equality here indicated appears to be closer than 

 that of the Moreau formula. But is it anything more than accident 1 

 Going back to the definitions of gZ'e, etc., we find, putting A/-*/ for the 

 transverse potential-difference of the Hall eff'ect and AP^' for the 

 transverse potential-difference of the Nernst effect, 



* ' " ' \ w utj \ w dlj AF2 \dl tvtj ^ ' 



The dimensions of this quotient A are those of a temperature-gradient 

 divided by a current density. 

 We find also 



_ (AB\ . dB\ . (AB\ , C\_ AB\ f C ^d6\ 



The dimensions of this quotient B are the reciprocal of those of the 

 quotient A, and accordingly it appears probable that we have only an 

 accidental coincidence in the approach to equality shown by A and B. 



If we undertake to explain the Ettingshausen eff'ect as due to the 

 magnetic rotation of the isothermal lines of a longitudinal tempera- 

 ture-gradient set up by the electric current along the plate, we see 

 that, as the coefficient gT'h is about 0.09 as large as ^7"^, we must 

 assume our temperature-gradient in question to be about 0.09 of a 

 degree per centimeter in our plate when the electric current density 

 therein is 1. Such a condition, as a normal attendant of a current in 

 iron, could not have escaped the attention of ordinary experience. If 

 it were set up by the act of magnetization, it would be detected in the 

 longitudinal tests presently to be described. 



Is it, then, possible that the Ettingshausen transverse effect is due 



