HALL AND CAMPBELL. — MAGNETIC EFFECTS IN SOFT IRON. 057 



electric couple, and if the action of the magnet should make a differ- 

 ence of temperature amounting to 0.01 degree between the junctions 

 (1) and (H), the thermo-electric action of this copper-iron couple 

 would produce as great a deflection of the galvanometer as that given 

 above, which was taken to indicate a change of resistance. Accord - 

 inglj'^, it was necessary to consider the possible magnitude of this 

 second longitudinal effect, which will presently be considered. 



It is to be noted that Plate 1 shows in a field of 10,700 an increase 

 of resistance which we may represent, on an arbitrary scale, by 16, 

 while Plate 2 shows in a field of 5400 a decrease of resistance which 

 we may represent on the same scale by 3. Zahn says, " In ferromag- 

 netic metals there is in weak fields an increase of resistance, in stronger 

 fields a decrease. According to the results of Grunmach for iron and 

 cobalt and a preliminary publication of Blake on measurements in 

 nickel, the initial increase is perhaps due to longitudinal components 

 [of magnetism]." When it is remembered that in one of our plates, 

 (1), the resistance is measured at right angles with the fibres or grain 

 of the iron, while in (2) it is measured in the direction of the fibres, it 

 need not, perhaps, be considered strange that the change of resistance 

 caused by magnetic action is opposite in direction in the two cases. 

 In both cases the change is very small. 



We now proceed to the other longitudinal effect which is so inti- 

 mately connected with the one just discussed. 



Loiigitiidinal Electromagnetic Temperature- Difference. — With the 

 magnetic field and the EE current as in Figure 1, we may look 

 for a difference of temperature, A6'', established by magnetic action 

 between two points d cm. apart along the plate. We have spent 

 much time in looking for this effect, especially in Plate 1, being led on 

 by accidental phenomena which simulated it rather persistently, 

 though with discrepancies that were suspicious, until a slight change 

 in the disposition of the plate between the poles of the magnet wiped 

 out the apparent temperature change which we had been studying and 

 possibly gave a slight apparent change in the opposite direction. 



With Plate 2 we gave only one day, a whole day^ to this particular 

 question. The method used was the following : Copper-constantan 

 couple (1), having one junction at point (1) on the plate and the other 

 in a glass tube placed in a water-bath, was put in circuit with one coil 

 of our galvanometer. Couple (3), with one junction at point (3) on 

 the plate and the other in the water-bath, was put in circuit with the 

 other coil of the galvanometer, opposing the action of couple (1) and 

 nearly balancing it. Then the magnetic field was brought into action 

 and the consequent deflections of the galvanometer were observed. 



VOL. XLVI. — 42 



