'27 6 PROFESSOR TAIT ON THE 



But we see also how diffusion varies with the relative size of the particles, 

 the sum of the diameters being constant. For the smaller, relatively, are the 

 particles of smaller mass (those which have the greater mean-square speed) the 

 more rapid is the diffusion. 



And further, by comparison with the results of §§ 51, 54, we see how much 

 more quickly a gas diffuses into another of different specific gravity than into 

 another of the same specific gravity. 



When the less massive particles are indefinitely small in comparison with 

 the others, the diameter of these is s ; and for their rate of diffusion we have 



Z— ■=- ?1 1-26 . 

 dt 7rs' 2 Jhj_ 



When it is the more massive particles which are evanescent in size, the 

 numerical factor seems to be about 3*48. Hence it would appear that, even in 

 the case of masses so different, there is a minimum value of the diffusion- 

 coefficient, which is reached before the more massive particles are infinitesimal 

 compared with the others. 



[At one time I thought of expressing the results of this section in a form 

 similar to that adopted in the expression for D in § 51. It is easy to see that 

 the quantity corresponding to A. would now be what may be called the mean 

 free path of a single particle of one gas in a space filled with another. Its 

 value would be easily calculated by the introduction of h x for h in the factor v 

 of the integral 



/ 



while keeping e in terms of h. This involves multiplication of each number in 

 the fourth column of the Appendix to Part I. by the new factor e-^-^* 1 hj/h*. 

 "Rut, on reflection, I do not see that much would be gained by this.] 



