280 PHENOMENA DEPENDENT ON MOLECULAK PATHS 104 



It is easy to give an idea of the matter contained in 

 these mathematical theories without repetition of the calcu- 

 lation, because the analogy of the problem before us with 

 the theories earlier discussed is obvious. Diffusion consists 

 in a transference of mass effected by means of the molecular 

 motions and viscosity in a transfer of forward momentum 

 caused by the same means ; conduction of heat is likewise 

 a transfer of energy, which is effected as before by the 

 motion of the molecules. This similarity goes so far that 

 the propagation of heat may be directly looked on as a 

 diffusion-phenomenon in which the warmer and colder 

 particles diffuse among each other. For, since the mole- 

 cules as they pass from an upper and warmer layer to a 

 lower and colder one retain their energy till a collision, the 

 process of conduction of heat is completely identical with 

 that of diffusion ; and we have no further difference to take 

 into account than this, that we have now to find, not the 

 number of the diffusing particles, but the sum of their 

 energy. 



In order to form a distinct idea of the arrangement of 

 the experiment that shall correspond as nearly as possible to 

 that chosen before, let us consider a gas enclosed between 

 two unlimited, or, at least, very widely extending, parallel 

 plane walls which lie horizontally with the distance between 

 them equal to the unit of length ; and consider the lower to 

 be kept at the constant temperature C., and the upper at 

 the temperature 1 C. Under these circumstances a distri- 

 bution of temperature is produced of itself between the 

 walls which is independent of the time, and such that at the 

 height x above the lower limiting plane the temperature is 



A constant flow of heat in the direction from above to below 

 takes place in the gas, and this is such that through each 

 imaginary horizontal plane in the space occupied by the gas 

 there flows an equal amount of heat in unit time. In 

 accordance with the usual definition we denote as the 

 coefficient of conductivity of the gas [or simply its conduc- 

 tivity] that amount of heat which in unit time passes 



