of Thermo-electric Phenomena. 525 



it aside for a couple of months to see if any thing would occur 

 to show the improbability of this being the case; but as any 

 fresh light has appeared rather to confirm than to weaken the 

 hypothesis, I now venture to publish it. But it is necessary 

 to say that, though the hypothesis does, with the help of a few 

 more or less reasonable assumptions, account for all the known 

 phenomena of thermo-electricity, yet it is quite possible that 

 the assumptions made are not the most likely ones, and that 

 the laws deduced are not exactly the true ones. Nevertheless, 

 since they fulfil the necessary conditions as well as the true 

 ones, they are sufficient to show that a consistent account of 

 the facts can be given by means of a hypothesis founded on a 

 connexion between the molecules of matter and electricity 

 which is called " resistance " and which is known to exist, 

 though it is not yet accounted for. 



I have preferred to continue speaking in terms of the me- 

 chanical model, because discussions concerning molecules and 

 aether are apt to have in appearance the looseness which they 

 very often possess in reality. Moreover one has the advan- 

 tage of being able to leave the precise relation which the cord 

 is to bear to electricity or to the aether, and which the buttons 

 are to bear to molecules, as well as the real nature of the con- 

 nexion between the buttons and the cord, unspecified and open 

 to any definition which the vortex theory of matter may suggest. 

 § 21. In a previous communication (page 353 of the pre- 

 sent volume) the passage of electricity through dielectrics and 

 electrolytes was illustrated by mechanical analogies at some 

 length , but its passage through ordinary metallic conductors 

 (§11) was only touched upon sufficiently to enable them to be 

 compared with the other media, and all the details of the pro- 

 cess were left unconsidered. In the present paper it is intended 

 to enter into these details; and it is to be regarded as a con- 

 tinuation of the first paper. 



The electrical resistance of a metal was represented by the 

 friction to which a cord was subject in passing through a row 

 of nearly smooth buttons fixed to a rigid support (see fig. 3, 

 Plate III.). The cord when moving represented a current of 

 electricity ; and the buttons represented the molecules of the 

 conductor. But, since we were not supposing the conductor 

 to be at the absolute zero of temperature, the buttons properly 

 ought not to have been standing still but to have been exe- 

 cuting some kind of periodic motion. 



Let us then suppose that the rods supporting the buttons 

 are not perfectly rigid, but highly elastic, and that each button 

 is vibrating backwards and forwards in the direction of the 

 length of the cord with a succession of harmonic motions of 



