64 WHAT IS SCIENCE? 



This mode of expressing the matter is probably not 

 quite correct ; for closer examination would show that 

 it is difficult to regard the assertion that density (unlike 

 colour) is an invariable property as a true law. It would 

 be better to say that there are certain associations (such 

 as that of density or melting-point with the other pro- 

 perties of a substance) which, if they occur at all, we expect 

 to be invariable. In other words, we expect to find laws 

 of certain forms, and if we find an observation which 

 might be a particular instance of a law of one of these 

 forms, we are much more ready to jump at once to the 

 conclusion that this law is indeed true than we should 

 be if the law, of which the observation would be a 

 particular instance, is not of one of these forms. And 

 one of the reasons why we expect such laws is that 

 we have previously found a large number of them ; 

 however, as we shall see presently, this is not the only 

 reason. 



THE ELASTICITY OF LAWS 



But our answer is not complete yet. If this were all, 

 I think we should feel far more uncertainty than we 

 actually do feel about most of our laws. However many 

 favourable instances we had observed, if we felt that a 

 single unfavourable instance, if it occurred, would destroy 

 the law, we should never be free from uneasiness. The 

 contrary instance might occur ; we might go to our 

 laboratory one morning and find that the density of some 

 substance which we had measured the day before was now 

 quite different. Our confidence in the law is largely based 

 on the fact that such an unfortunate incident would not 

 necessarily destroy our belief in the law. 



This statement may be surprising. Surely if a law 

 states that some relation is invariable ; and if, as we pro- 

 fessed in Chapter III., we are going to be really strict in 

 our interpretation of invariability, then a single contrary 



