HALL AND CAMPBELL. — AL\GXETIC EFFECTS IN SOFT IRON. G55 



completeness we shall describe two other longitudinal effects which have 

 been observed, though we have not yet made sure of them in iroii.^^ 



No one of the four lougitudinal elfects has any assigned name. 

 Each must be designated by a rather long descriptive title. 



Longitudinal EUrtromagnetic Potential- Difference. — With the mag- 

 netic field and the ii^fc' current as in Figure 1, we may take note of a 

 change of longitudinal potential-gradient, due to transverse magnetic 

 action, without change of strength of the longitudinal current. Of 

 this effect Zahn says, " As its magnitude is found to be proportional 

 to the strength of the primary [longitudinal] current-strength, it may 

 be taken as a change of resistance." 



We shall call this effect positive when it produces an increase of 

 resistance. We take as the coelHcient 



eZ, = A/? ^ RH, (21) 



where R = the normal resistance of a certain length of the plate and 

 AR = the increase of resistance of the same length of the plate. 



In studying this effect in Plate 1 we connected points (1) and (8) of 

 the plate (see Figure 6), by means of the iron wires, with our galvano- 

 meter, introducing into the galvanometer circuit e. m. f from another 

 circuit, so as to get an approximate balance of forces at the start, thus 

 keeping the scale of the galvanometer within the field of view. Then we 

 applied the magnetic field and observed the deflection produced, which 

 was due to an increase or decrease in one of our counter-balancing 

 e. m. f.'s, the other being not affected by the magnet. This method is 

 troublesome, because it is difficult to keep the requisite degree of 

 equilibrium through an extended series of observations. 



This operation gave us for Plate 1 values of eZ^ which are exhibited 

 in the following table. The numbers given in the last three columns 

 are approximate only. 



With Plate 2 a different method was used. Copper wires connected 

 with points (1) and (3) of the plate led to one coil of the galvanometer, 

 while wires from two other points, A and B (see Figure 13), of the 

 plate-circuit led to the other coil. By means of the shunts Si and S^ 



^' But see p. 664, footnote. 



