an Explanation of Hall's Phenomenon. 259 



and the point F, the current passes from a stretched to a 

 compressed portion ; heat will therefore be absorbed in the 

 region B, For like reasons, heat will be absorbed in C and 

 developed in D. The temperature of the copper plate will 

 therefore not be uniform, the portions A and D being on 

 the whole hotter than the portions B and C. But the resist- 

 ance of a metal increases with its temperature. The resistance 

 of A and D will therefore be greater, and the resistance of B 

 and C smaller, than before the plate was strained. If there- 

 fore Gr H were originally an equip otential line, it is clear that 

 it will be so no longer. An equipotential line through the 

 point will now be inclined to G H in the direction K L, as 

 shown in the figure. 



Supposing the plate to be of iron instead of copper, the 

 Peltier effects will be reversed, and the regions which in the 

 former case were hot will now be cold, and vice versa. The 

 distribution of resistance will be changed in a corresponding 

 manner, and the equipotential line will be rotated in the 

 opposite direction. 



The peculiar thermoelectric properties of copper and iron, 

 discovered by Sir William Thomson, are thus seen to be 

 sufficient to account for HalPs phenomenon in the case of 

 those metals. If the explanation which I have offered be the 

 correct one, it is clear that Hall's " positive " metals, cobalt 

 and zinc, should, when tested by Thomson's method, exhibit 

 the same thermoelectric effects as iron; that his " negative " 

 metals should behave thermoelectrically in the same manner 

 as copper; and that lead might be expected to be thermo- 

 electrically unaffected by strain. It became exceedingly 

 interesting to ascertain if this was the case, and I therefore 

 proceeded to repeat Thomson's experiment upon all the metals 

 mentioned by Hall. In most cases at least two or three diffe- 

 rent specimens of the metal were used, and, except in the case 

 of cobalt, the following simple method of operating was 

 adopted. 



A short piece of the wire or strip of the foil to be examined 

 was held at the ends by two fixed brass clamps, about 5 or 6 

 centim. apart, which were connected with the terminals of a 

 low resistance reflecting-galvanometer. A point near the 

 middle of the metal was then gripped by a pair of pliers 

 which had been previously heated, and one half of the wire 

 was drawn taut, leaving the other half slack. A galvano- 

 meter-deflection immediately occurred, which ceased when the 

 tension was discontinued. The process was then repeated, 

 the pull being made in the opposite direction : this resulted 

 in an opposite deflection of the galvanometer. 



