68 PROCEEDINGS OF THE AMERICAN ACADEMY. 



contact with each other much of the time, we have suggested at once 

 an explanation of the vanishing electric resistance and the vanishing 

 specific heat observed in pure metals as they approach the absolute 

 zero of temperature. With atoms pressed very close together we 

 should expect some of the electrons, what Rutherford would call the 

 peripheral ones, to be movable from one atom to another with com- 

 parative ease, and with the same conditions we should expect the 

 degrees of freedom of the atoms to be fewer, 3 and the thermal capacity 

 of the metal less, than at higher temperatures. A piece of metal in 

 this condition, immersed in helium gas and subject to the impact of 

 the gas molecules, may be compared to a brick building bombarded 

 by tennis balls. If the individual bricks were free, they would absorb 

 kinetic energy from the balls; but agglomerated in one huge mass 

 they repel the attacks and remain almost unmoved. 



On the other hand, thermo-electric phenomena appear to require 

 the presence of free electrons within metals. In looking for a theory 

 of electric conduction in solids we should not, even at the start, forget 

 the fact that circuits exist in which the electric current is maintained 

 at the expense of heat energy solely. Now, in all cases in which the 

 transformation of heat into work is really understood, it is effected by 

 means of change of dimensions, expansion and contraction of the 

 working substance in which the heat resides and operates as molecular 

 or atomic energy. In a thermo-electric current the electricity is the 

 factor which undergoes a cyclic change; the metals are in a fixed 

 state, though one of non-uniform temperature, and they neither 

 expand nor contract after this fixed state is reached. It would seem, 

 then, that the electricity must expand and contract in its cyclic course 

 and serve as the vehicle and transformer of heat energy. 



Hence my attempt to discover what a combination of the two kinds 

 of electron action might be expected to do in a metal unequally heated. 

 In the course of this undertaking, which has extended over some 

 months, I have been led to change somewhat from time to time my 

 point of view and the particular assumptions of which I have made 

 use. For example, starting with the idea that the free electrons have 

 a heat capacity which is constant, though much less than that of 

 ordinary gas molecules, I later determined to try the experiment of 

 taking this thermal capacity as a variable, increasing with rise of 

 temperature but still, at ordinary temperatures, below that of gas 



3 See Jeans, Phil. Mag., Vol. 17 (1909), p. 794, where the possibility that the 

 atoms of a metal may have very little thermal capacity when locked together 

 is discussed. 



