MancJicster JlJ'cinoirs, Fo/. Hi. (190S), A^^. 10. 53 



For if we do not admit this postulate, then the equations 

 of statistical equilibriiim will contain more than two terms 

 as exemplified above ; and that aspect of chemical equili- 

 brium will be at variance with the usual thermod\'namic 

 theory, which expresses an independent equilibrium for 

 every type of reaction that is formally possible. And the 

 reason has been already indicated, viz., the usual expressions 

 for the thermodynamic entropy and available energy of 

 gaseous s\-stems, and through them of dilute solutions, in- 

 volvethe implication that onlybinary molecular encounters 

 need be considered. The two points of view will agree 

 only if all reactions take place in binary stages ; and it 

 becomes a question whether this is a universal rule under 

 all circumstances, or only one prevalent in the prominent 

 cases which are naturally those governed by simple 

 recognisable relations. 



An actual case in which these distinctions may make 

 theoretically a difference is worked out from the thermo- 

 dynamic side in Planck's TJicrniodyiimnics, §247, under 

 the heading of graded dissociation, viz., that of hydriodic 

 acid HI into H,, I,, and I. 



Another question in which such considerations may 

 have scope is that of Ostwald's law of equilibrium of 

 ionisation. If only two ions can arise, they must be equal 

 in number ; thus \{ c is their concentration (dilute) and c 

 that of the non-ionised part, c'-jc may be expected to be 

 constant at each temperature, the ionisation proportional 

 to c being balanced by the recombination proportional to 

 c'-. But this assumes that all the ionisation is spon- 

 taneous, whereas in the cognate phenomena of gases the 

 encounter of an ion (in rapid motion) with a molecule 

 has been shown by Townsend to be a potent cause of 

 further ionisation. This suggests the question whether c 

 should not be replaced by c-^kcc or <;• (i +/r'), which may 



