Manchester Memoirs, Vol. Hi. (1908), No. 10. 31 



carrier for the one united with it. Without some such 



intermediate transient stage of combination a dissociation 



of a complex into three molecules must proceed to an end, 



for the chance of an equilibrium through recombination 



would be negligible. Where an equilibrium is found to 



become established, either the reaction must occur in 



binary stages, or else it must take place in contact 



with solid or liquid boundaries where the molecules 



form a denser layer in which each is alwaj^s in 



relation with others. This consideration reinforces 



the importance of the study of reaction in pure gases, 



as a means of disentangling the intermediate stages of 



chemical combination and the durations of the products 



formed in them. The fundamental importance of this 



kind of knowledge for the adequate interpretation of 



banded spectra has already been alluded to. It appears, 



indeed, to be commonly recognised that direct trimolecu- 



lar combinations occur seldom : the inference from the 



present line of argument is that in gaseous reactions 



they do not occur at all. Recently I have learned that 



Mendeleef had always maintained that gaseous reaction 



occurs in monomolecular or bimolecular stages in all 



cases : there seems to be strong presumption in favour of 



such a view. 



It would require the instincts of a chemist to venture 

 on any attempt to apply this principle to special cases, to 

 discuss why, for example, the presence of a foreign sub- 

 stance sometimes promotes the occurrence of the necessary 

 intermediate binary reaction, and in other cases pre- 

 sumably destroys its product, and so inhibits the final 

 transformation. The recent results obtained by Bone and 

 Edmunds for the thermal dissociation of HoO agree with 

 the conclusions drawn in 1884 by Dixon with regard to 

 the explosion of a mixture of CO and O^, in assigning 



