440 Scientific Proceedings, Royal Dublin Society. 



magnetism to make the corresponding step of showing what sort 

 of a thing an electric displacement is that it may produce these 

 stresses, so that the line I am pointing out goes beyond Maxwell's 

 analogy, but must be considered as only a dim sort of hypothesis, 

 and not as a complete theory. 



In the first place, it is evident that a non-conductor must be a 

 medium that can have different temperatures at different parts 

 without entropy being conducted from one place to another, i. e. 

 without conduction of energy ; and a conductor must .be an engine 

 that transfers entropy undiminished, and always produces heat or 

 work equivalent to the degradation of entropy. 



Now a distribution of molecules of a gas and their velocities is 

 possible by which there would be stresses in the gas similar to those 

 required by electrostatic phenomena, and yet without either cur- 

 rents of gas or conduction of heat taking place. The molecules of 

 a material gas do not get themselves or their velocities distributed 

 in this way by differences of temperature : in all material gases 

 there is certainly conduction of heat, but it seems probably possible 

 to invent a law of action between molecules which shall produce 

 any required law of distribution of molecules and their velocities. 

 A somewhat similar problem was solved by Maxwell when he cal- 

 culated the law of force between molecules, in order that the tem- 

 perature at all points in a column of gas subject to gravity might 

 be the same. As the whole thing is so hypothetical, I have not 

 thought it worth while attempting to work out the law of action 

 between molecules that would account for the distribution of velo- 

 cities required by my hypothesis. 



If the number of molecules going in a direction defined by a 

 direction angle Q be represented by [fx = cos Q) 



m = nio + m-z (,u^ - i) ; 

 and if the velocity of the corresponding molecules be 



then there will be neither currents of gas nor to the second order 

 of small quantities any conduction of heat, but the pressure will be 

 different in different directions, and will be parallel to the direction 

 from which 9 is measured. 



