682 Professor J. J. Thomson [April 19, 



metal ; the latter will depend upon the average velocity of these cor- 

 puscles, for if they are moving with very great rapidity the electric 

 force will have very little time to act before the corpuscle collides 

 with an atom, and the effect produced by the electric force annulled. 

 Thus, the average velocity of drift imparted to the corpuscles by the 

 electric field will diminish as the average velocity of translation, 

 which is fixed by the temperature, increases. As the average velocity 

 of translation increases with the temperature, the corpuscles will 

 move more freely under the action of an electric force at low 

 temperatures than at high, and thus from this cause the electrical 

 conductivity of metals would increase as the temperature diminishes. 

 In a paper presented to the International Congress of Physics at 

 Paris in the autumn of last year, I described a method by which 

 the number of corpuscles per unit volume and the velocity with 

 which they moved under an electric force can be determined. Apply- 

 ing this method to the case of bismuth, it appears that at the tempera- 

 ture of 20° C. there are about as many corpuscles in a cubic centimetre 

 as there arc molecules in the same volume of a gas at the same tem- 

 perature and at a pressure of about a quarter of an atmosphere, and 

 that the corpuscles under an electric field of 1 volt per centimetre would 

 travel at the rate of about 70 metres per second. Bismuth is at present 

 the only metal for which the data necessary for the application of this 

 method exist ; but experiments are in progress at the Cavendish 

 Laboratory which it is hoped will furnish the means for applying the 

 method to other metals. We know enough, however, to be sure that 

 the corpuscles in good conductors, such as gold, silver or copper, must 

 be much more numerous than in bismuth, and that the corpuscular 

 pressure in these metals must amount to many atmospheres. These 

 corpuscles increase the specific heat of a metal, and the specific heat 

 gives a superior limit to the number of them in the metal. 



An interesting application of this theory is to the conduction of 

 electricity through thin films of metal. Longden has recently shown 

 that when the thickness of the film falls below a certain value, the 

 specific resistance of the film increases rapidly as the thickness of 

 the film diminishes. This result is readily explained by this theory 

 of metallic conduction, for when the film gets so thin that its thick- 

 ness is comparable with the mean free path of a corjrascle, the num- 

 ber of collisions made by a corpuscle in a film will be greater than in 

 the metal in bulk, thus the mobility of the particles in the film will 

 be less and the electrical resistance consequently greater. 



The corpuscles disseminated through the metal will do more than 

 carry the electric current, they will also carry heat from one part to 

 another of an unequally heated piece of metal. For if the corpuscles 

 m one part of the metal have more kinetic energy than those in 

 another, then, in consequence of the collisions of the corpuscles with 

 each other and with the atoms, the kinetic energy will tend to pass 

 from those places where it is greater to those where it is less, and in 

 this way heat will flow from the hot to the cold parts of the metal ; as 



