164 
MR. II. TOMLINSON ON THE INFLUENCE OF STRESS 
throughout, there must be unequal heating wherever two dissimilar parts of the wire 
join, and therefore consequently there must be developed at each such junction an 
opposing electromotive force. Indeed, may not the large resistance of such alloys as 
German-silver and platinum-silver be in a great measure due to a similar unequal 
heating at the junctions of the molecules of the several metals forming the alloy : 
with this difference, however, that here there would be an equalization of tempe¬ 
rature the rapidity of which would be more and more approached in the case of a 
compound circuit ma.de up of several pieces, say of iron and copper, as the distances 
between the consecutive junctions of these pieces became less and less ? 
Further, besides the error likely to arise from the “ Peltier effect/’ we must 
expect to encounter another, inasmuch as the heat generated in a wire by the passage 
of a current by no means necessarily produces the same alteration of potential at the 
two junctions. This may be from two causes: either different rises of temperature 
may be produced because the terminal at one end carries off more heat than that at 
the other, or because the metals which appear to be identical at the two junctions are 
not so. 
That the “ Peltier effect ” does come largely into play sometimes, we can convince 
ourselves by passing a current from a single cell of Daniell for five or ten seconds 
through a thermopile, and then, after disconnecting the cell from the pile, putting 
the latter in circuit of a galvanometer; in such case a very considerable deflection can 
be obtained with a delicate instrument, and, indeed, we can even make the warmth of 
the hand, pressed against the face of one thermopile, generate such a “ Peltier 
effect ” in a second pile connected with the first, that on disconnecting the two from 
each other, and then connecting the second with a reflecting galvanometer, a deflection 
may be obtained which can be rendered visible at a considerable distance. The 
following experiments will show that such errors as those above-mentioned are by no 
means merely theoretical. 
Experiment LXXVIII. 
A silk-covered German-silver wire, -g^-th of an inch in diameter, was soldered at its 
two extremities to two copper terminals, ygtb. of an inch in diameter and 6 inches in 
length, the whole forming a fairly accurate resistance coil of half an ohm. The coil 
was put for 20 seconds in the circuit of one Grove’s cell, and then, after the cell 
had been disconnected, and when a further period of five seconds had elapsed, 
connected by a mercury cup with a galvanometer; a deflection of 200 divisions was 
obtained, whereas, previously, there had been no perceptible deflection. After a rest 
of five minutes there was only a deflection of some 20 divisions. The battery was 
now reversed, and after a connexion with the coil for 20 seconds, a deflection of 
170 divisions was obtained in the opposite direction. 
