646 PROCEEDINGS OF THE AMERICAN ACADEMY. 



approach to equality, which may be interpreted as meaning that the 

 equipotential lines of the longitudinal electric current in the Hall 

 effect and the isothermal lines of the longitudinal heat-current in the 

 Leduc effect are rotated in the same direction and to nearly the same 

 extent, at certain temperatures. Detailed comparison gives 



(III.) 



It would seem from this showing that, within the range of tempera- 

 ture here considered, the isothermal lines are rotated in the Leduc 

 effect about three-fourths as far as the equipotential lines are rotated 

 in the Hall effect, the strength of magnetic field being the same on 

 both cases. 



Comparison of Tables II and III indicates that, within the range of 

 temperature shown, the static Thomson equipotential lines are rotated 

 farther in the Nernst effect than the isothermal lines are simultan- 

 eously turned in the Leduc effect, but that the excess noted diminishes 

 rapidly with rise of temperature and seems likely to become zero or 

 negative with further rise. 



As we have in the Moreau formula a suggestion at least of some 

 relation between e Tg and ^ Tg, we might expect to find evidence of a 

 reciprocal relation between f^T^ and eT'n ', that is, between the Leduc 

 effect and the Ettingshausen effect. Reflection shows, however, that 

 one link in the chain of reasoning, if it deserves such a name, is here 

 missing. We go from the Hall effect to the Nernst effect by way of 

 the Thomson effect, the known difference of electric potential which 

 accompanies flow of heat or difference of temperature in iron under 

 normal conditions. But no corresponding difference of temperature 

 necessarily accompanying flow of electricity or difference of electric 

 potential is known to us. On the other hand, if we could find by 

 inspection of the various coefiicients before us evidence of a relation 

 between ^7"^ and eT"/,, we might perhaps take this as a clue to what 

 could be called the reciprocal of the Thomson effect, and could be de- 

 scribed as a temperature-gradient due to potential-gradient. The 

 possibility of such a discovery makes it desirable to examine closely 

 all the relations which a study of our various tables of data suggest. 



