Thermal Electrolysis, 509 



of the rolled metals might give rise to the action; but this 

 was disproved by the following experiments : — 



Some crystals of silver prepared by electrolysis were placed 

 in the open end of a piece of glass tubing slightly constricted, 

 and then immersed in silver chloride heated in a crucible by 

 a Bunsen lamp. In about half an hour the crystals were 

 found to have grown in a net-like mass from their original 

 position to a point about half an inch higher in the tube. 

 This experiment was repeated with crystals of silver which 

 had themselves been deposited from the fused chloride by 

 means of metallic silver. A similar result was obtained. 



We were then led to the conclusion that the change de- 

 pended upon the unequal heating of different parts of the im- 

 mersed metal, or rather of the salt in which it was immersed. 

 It is evident that upon the contact theory of voltaic action, 

 there will be a difference of potential between the metal and 

 the liquid chloride with which it is in contact; and it is in 

 accordance with analogy to suppose that this difference of 

 potential will vary according to the temperature. Now, under 

 the conditions of the experiment, it cannot be supposed that 

 all parts of the fused chloride in contact with the immersed 

 metal were always equally heated; and we have therefore the 

 possibility of a current being established with the consequent 

 electrolysis of the salt. 



In order to test this view, some silver chloride was fused in 

 a hard glass tube and a rod of silver placed in the liquid. 

 On heating the underside of the lower end for 10 minutes, 

 we found a considerable crop of silver crystals in the compa- 

 ratively cool part of the fluid. 



In another experiment some silver chloride was fused in a 

 crucible, and one side of the vessel was more strongly heated 

 than the other. Two long rods of silver were connected with 

 a galvanometer and placed, one in the hotter, the other in the 

 colder part of the chloride. The latter was found studded 

 with crystals at the end of 15 minutes, whilst the former was 

 quite clean. On repeating this experiment, it was always 

 found that the galvanometer gave a larger deflection the 

 greater the difference of temperature between the portions 

 of the fused mass penetrated by the silver wires, and that 

 the current was reversed with a reversal of the rods. Copper 

 wires in cuprous chloride gave similar results. 



In an experiment with an electrometer we obtained a clear 

 indication of a difference of potential between silver rods in 

 hotter and colder parts of silver chloride fused in a small cru- 

 cible, the deflection showing a difference of possibly -^ of a 

 Phil Mag. S. 5. No. 71. SuppL Vol. 11. 2 



