G2 WHAT IS SCIENCE ? 



fact that it is uniform ; for, as we have seen, it is made up of 

 invariable associations concerning which universal agree- 

 ment can be obtained. Any part of experience that is 

 not uniform would not consist of invariable associations 

 and would be at once excluded from this closely regulated 

 nature. Indeed the problem before us is simply that 

 of how we distinguish the uniform from the non-uniform 

 parts of the nature of common sense, for that is our task 

 in establishing the relations which are asserted by laws. 

 To attempt to base a method of making the distinction 

 on the assumption that all nature is uniform is simply to 

 misunderstand the problem that is to be solved. 



AN ATTEMPTED SOLUTION 



After this clearing of the ground, we can attack the 

 problem. What is the feature of our previous experience 

 which makes us so certain that the " law " of the rising 

 and the setting of the sun asserts a truly invariable asso- 

 ciation and, consequently, that the sun will rise to- 

 morrow ? In answer every one would say that our belief 

 is certain because we have observed the association an 

 immense number of times without observing any failure. 

 And doubtless this is the reason in this particular case, 

 but other instances suggest that the answer is not funda- 

 mental or complete. For there are instances in which 

 an association which has been found as invariable has 

 at length been broken ; and there are instances in which 

 a law is asserted confidently as the result of a single 

 observation, so that there has been no chance of proving 

 any invariability. The instance of the first kind that is 

 always quoted in these discussions is that of the black 

 swan. Until Australia was discovered, swans had been 

 found invariably to be white in a very large number of 

 observations, and natural historians would have been 

 justified in asserting, according to the principle suggested, 

 the law that all swans are white ; and yet the law was 



