68 WHAT IS SCIENCE? 



answered, but is hardly contemplated at all, in the 

 actual discovery of laws. When we are seeking laws, 

 we are only thinking about the experience that we have 

 actually had ; and the problem which we seek to solve 

 is one that has reference only to that experience. We 

 ,seek to order the experience, to change it from a miscel- 

 laneous collection of apparently unconnected observa- 

 tions to a connected series of particular instances of 

 a few wide principles. These principles by means of 

 which, and in terms of which, weirder our past experience 

 are laws ; they state, as has been said so often before, 

 associations between events and properties which have 

 proved in our past experience to be invariable. It is 

 because the associations have proved invariable through- 

 out this experience that by means of them we can order 

 the experience as many particular' instances of a few 

 principles. When our experience is increased by the 

 addition of observations which were future but are now 

 past, we_seek oncejnore io orderjn the same manner our 

 experience ; but in this increased 



volume all experience is of equal value, that which was 

 future is in no way different from that which was past, 

 for all is now past. It 'may happen that the order estab- 

 lished for the original experience is equally valid for that 

 which we now have ; the portion that is added can again 

 be regarded as particular instances of the laws which 

 were established as a result of the original experience. 

 And if that happens, we have no reason to change our 

 laws. But if that does not happen, if the laws estab- 

 lished for the original experience do not prove valid when 

 the volume of it is increased, then we have two 

 alternatives. We may either . reject altogether the 

 ad3edexperience ancF^miat it is noFproper~subject- 



matter for science, or we may alter slightly (or radically) 

 our laws, so that they now order satisfactory both the 

 and the ng\v. If we adopt the second alternative, 



