1 6 Larmor, Physical Aspect of the Atomic Theoiy. 



Yet he has no alternative but to hold that each ultimate 

 grain is itself a self-existing cosmos, of complexity 

 probably beyond any complete analysis on our part, 

 which may indeed to appearance merge itself in combina- 

 tion with another atom or molecule, but is always recover- 

 able unaltered, — that there is no degradation of matter. 

 He holds probably that it is necessary to believe that in 

 the same pure substance the molecules are all exactly 

 alike, or, at any rate, that they are as nearly alike as 

 individuals of a very sharply defined species in the organic 

 world ; though he knows no natural reason which would 

 compel them to be so constituted, except in so far as they 

 may represent the limited number of types of dynamical 

 structures that can be built up from simpler identical 

 primordial elements. It is vastly more suggestive to accept 

 this wonderful inference, which constitutes the Daltonian 

 theory, as our working hypothesis, than to try to refrain 

 altogether from analogical reasoning about unseen mole- 

 cules : moreover, this procedure is almost imposed a priori 

 by the general principle already alluded to, that the 

 simplest theory is probably the most fruitful representa- 

 tion of reality. 



There is one branch of actual observational knowledge 

 in which this identity of the molecules of a substance 

 asserts itself with special strength : if the molecular 

 theory had not been introduced on the evidence of the 

 laws of definite and multiple proportions in chemical 

 compounds, it must have demanded recognition as a 

 result of a study of the crystalline structure of bodies. 

 We call to mind that correspondences are now coming 

 to light by which it is becoming possible to reason 

 regarding the type of the molecule, and the geometrical 

 grouping of its constituent atoms, from measurement of 

 the crystalline aggregate : in such cases the single 



