Mr. M. M. Pattison Miiir on Chemical Classification. 189 



two experiments which appear to be carried out under similar 

 conditions are really so carried out. 



Nevertheless we may arrive at a conviction that these two 

 quantities (valency and chemical composition) are really con- 

 nected together, if we notice that " one always varies percep- 

 tibly at the same time as the other"*, although we may be 

 unable accurately to measure the quantities, or to state the 

 quantitative relations existing between them. 



The only way of dealing with the question before us is, I 

 think, to attack it in detail. We must endeavour to decom- 

 pose the quantity " general chemical composition " into parts, 

 and to establish a relation between each of these, in succession, 

 and the valency of the elementary atoms. If this could be 

 accomplished, the further problem would still demand solution, 

 Does the simultaneous variation of several parts modify their 

 separate actions ? f 



20. It would be beyond the limits of such a paper as the 

 present to attempt any more than the briefest outline of the 

 results which have been obtained in the attempted solution of 

 the various parts of the problem under consideration. Start- 

 ing with the hypothesis (shown on other grounds to be a pro- 

 bable hypothesis) that the valency of each elementary atom is 

 a fixed quantity, it becomes possible to deduce the forms of 

 combination of a stated number of different elementary atoms, 

 the valency of each being given. The question is mainly a 

 mathematical question J, although the occurrence of unsatu- 

 rated compounds &c. may complicate it. But the only means 

 which we at present possess of assigning the probability to 

 the results of such a deduction is, studying the general che- 

 mical and physical properties of the various compounds of the 

 elements in question. From a certain assumption concern- 

 ing valency we deduce a number of possible formulae, each 

 of which represents a definite compound ; but we cannot lay 

 our hand on this or that compound ; we have no means of 

 certainly identifying any one of the hypothetical compounds. 

 Each of them may have been prepared and studied by us, and 

 we may yet have failed to recognize any one of them. We 

 make a second assumption, in favour of which there is 

 indeed much evidence, viz. that the general characteristics of 

 compounds are correlated with inner structure, or, in other 

 words, with composition. From what we can actually observe 

 of the general characteristics of the compounds of the given 



* Principles of Science, vol. ii. p. 110. 

 t See ' Principles of Science/ vol. ii. p. 55. 



X For a full discussion of this part of the problem see Die modernen 

 Theorien, pp. 138-170. 



