42 WHAT IS SCIENCE? 



invariably associated. We cannot, by placing ourselves in 

 such circumstances that we observe some of the proper- 

 ties (for instance short stature, black hair, and a sallow 

 complexion), make sure that we shall observe the other 

 properties of Napoleon. On the other hand, iron is a 

 kind of object suitable for the contemplation of science, 

 and not a particular object, because, if we place ourselves 

 in a position to observe some of the properties of iron, we 

 can always observe the other properties. Now, having 

 found an A and B invariably associated in this way, and 

 therefore defining a kind of object, we seek another set 

 of associated properties, C and D, which are again 

 connected by a law, and form another kind of object. 

 We now discover that the kind of object which consists 

 of A and B invariably associated is again invariably 

 associated with the kind of object which consists of C and 

 D invariably associated ; we can then state a new law, 

 asserting this invariable association of (AB) with (CD) ; 

 and this law marks a definite step forward in science. 

 If it is in such a way that science builds up new laws 

 from old it clearly becomes of great importance to decide 

 what are the most elementary laws on which all the others 

 are built. It is obvious that the analysis which we have 

 been noticing cannot be pushed backwards indefinitely. 

 We can show that in a law connecting X with Y, X is 

 the expression of a law between A and B, and Y of a law 

 between C and D ; we may possibly be able to show again 

 that A is the expression of yet another law connecting 

 some other terms, a and b. But, in the last resort, we must 

 come to terms, a and b, which are not resolvable into other 

 laws, and which, therefore, are not proper subject-matter 

 for science by themselves, but only when they occur in 

 the invariable association (ab). What, then, are the 

 terms at which we arrive at length by this analysis ? 

 What are the irresolvable laws which must lie at the basis 

 of all science ? 



