134 



three times i6 ; 28 and four times 16 ; and 28 and five times 16. 

 The atomic theory assumes that the weight of the nitrogen atom 

 is represented by the number 14, and that of the oxygen atom by 

 16; when combination takes place it must take place between 

 whole atoms ; two atoms of nitrogen, whose weight is represented 

 by the number 28, unite with one of oxygen, weight 16, to form 

 nitrous oxide, and with two of oxygen, weight 32, to form nitric 

 oxide, and so on. Of course, this is all hypothesis ; but the law 

 of multiple proportion, and all the other laws of chemical 

 combination, are facts. If they are not, it is either because there 

 has never yet existed a trustworthy balance, or because a single 

 accurate chemical analysis has never yet been made. The practice 

 of chemistry is based on the facts, and not on the hypothesis which 

 has been formed to explain the facts. 



Mr. J. D. KENDALL'S REPLY. 



That there would be an absolute agreement on all the points 

 involved in such a discussion as this, none of us I think expected. 

 Nor are we disappointed : for although we all arrive at the same 

 conclusion, yet you cannot fail to have observed that, between 

 Mr. Kitchin and myself, there are considerable differences by the 

 way; differences, which, considering our ultimate agreement, it 

 seems to me ought not to exist. In this reply — provided for in 

 the arrangement of the discussion — I will, therefore, address 

 myself to the consideration of those differences, with a view of 

 showing how they have arisen. In the first place Mr. Kitchin 

 objects to my definition of theory as an explanation of natural 

 phenomena founded on /ads, and suggests that it is rather an 

 explanation founded on inferences drawn from facts. But surely 

 that is a distinction without a difference. Can any one say that it 

 is not a fact that the three angles of a triangle are equal to two 

 right angles ? But we know also that it is an inference drawn from 

 facts. We see, therefore, that an inference may be a fact In 



