118 Prof. H. A. Wilson on the Diffusion of 



over which it passes. This introduces factors of which no 

 account is taken in the above investigation. 



The velocity of sound in ether at 0° calculated from the 

 above expression is 1787 metres per second, using the value 

 of P n given by the equation P^I^p. The value found by 

 experiment is 1145 metres per sec, which is somewhat 

 smaller than the calculated value. Since V v , according to 

 equation (9), may be expressed at corresponding states as a 

 multiple of its critical value, a much better agreement with 

 the facts would probably be obtained if we write V s = SiV r , 

 where Y s denotes the velocity of sound and Si a quantity 

 which has the same value for all substances at corresponding- 

 states. The value S x will probably not vary much with the 



T 



temperature, at a temperature — ? it is from the foregoing- 

 equal to "639. l ' 7 



V s thus varies with the nature of the liquid in the same 

 way as Y v given by equation (9), which, it should be noticed, 

 holds for matter of any density. If the ether behaved as a 

 perfect gas the velocity of sound would be 302 metres per 

 sec, a very much smaller value than found in practice. 



The various formulae deduced in this paper will be farther 

 tested in a paper that will appear in the Proc. of the Camb. 

 Phil. Soc. vol. xvi. part vii. 



Cambridge, 

 March 20, 1912. 



VIII. The Diffusion of Alkali Salt Vapours in Flames. 

 By Prof. H. A. Wilson, F.R.S., F.R.S.C* 



THE experiments described in the following paper were 

 undertaken with the object of finding the coefficients of 

 diffusion of alkali salt vapours in a Bunsen flame. The light 

 emitted by flames containing alkali salt vapours appears to 

 be due to atoms of the alkali metal probably electrically 

 neutral f but possibly positively charged . Probably an atom of 

 the metal under the influence of the violent collisions with other 

 atoms loses negative electrons, and so becomes a positive ion. 

 Since, however, many free negative electrons are present in 

 the flame even in the absence of salt, it seems certain that 

 such a positive ion must very soon be neutralized by recom- 

 bination with electrons. The metal atom therefore will be a 



* Communicated by the Author. 



i Smithells, Dawson, and Wilson, Phil. Trans. A. vol. exciii. p. 89 

 (1899). 



