HALL AND CAMPBELL. — MAGNETIC EFFECTS IN SOFT IRON. 639 



The "arm 6" that is, the temperature of the median cross-line of the 

 plate, was found by connecting either the Cof» circuit or the 62/4 cir- 

 cuit with the galvanometer and then bringing the outer junction of the 

 thermo-electric couple to such a temperature as to make the current in 

 the circuit zero. 



The temperature-gradient ^ along the plate was found by connect- 

 ing the thermo-electric circuit Ci-phte-Cs with the galvanometer. 



The transverse potential difference, the Nernst effect, was found by 

 means of the circuit h- plate -li, the assumption being made that the 

 thermo-electric difference between the iron plate and the iron wires 

 could safely oe neglected, so that no account need be taken here of the 

 fact that a transverse temperature-difference, the Leduc effect, is set 

 up simultaneously with the Nernst effect. It now seems very doubtful 

 whether this assumption was justified ; but as we have good reason for 

 supposing that the error thus introduced into the value oi ^Te was not 

 more than 10 per cent, and as we are not sure of the sign of this error, 

 we leave n Tg for Plate 1 as we have found it. 



With Plate 2 the conditions were better. We found arm 6 and 

 d6i<U in essentially the same way as before. We found the transverse 

 6. m. f,, ^E' let us say, produced in the circuit C^- plate- Ci by mag- 

 netic action and deducted from this the amount, a large part of the 

 whole, which from a study of the Leduc effect we found to be attribu- 

 table to a thermo-electric force set up in the circuit C^-plate-d by 

 magnetic action. If we call this e. m. f. t^iE' and call the true Nernst 

 transverse difference of potential AP', we have 

 AP' = A£" - b^^E. 



Reference to formula (3) defining /, Tg will show that it contains the 

 factor 1 -^Itv X -jjY where iv is the width. Now as the Nernst aP' is 



