178 WHAT IS SCIENCE? 



form of regularity which science regards as pure chance. 

 If that step can be taken, the scientific problem is solved, 

 for pure chance, like strict law, is one of the ultimate 

 conceptions of science. But there is often (as in 

 meteorology) a long interval between the first step and 

 the second, and in that interval all that is known is a 

 regularity in the long series which is usually called 

 a " statistical " law. 



This procedure, like most scientific procedure, is 

 borrowed and developed from common sense ; but 

 and this is the reason why it is mentioned here it is 

 here that modern common sense lags most behind 

 scientific method. It is a familiar saying that statistics 

 can prove anything ; and so they can in the hands of 

 those not trained in scientific analysis. Statistical laws 

 are one of the most abundant sources of popular fallacies 

 which arise both from an ignorance of what such a law 

 means and a still greater ignorance of how it is to be 

 established. A statistical law does not state that some- 

 thing always happens, but that it happens more (or 

 less) frequently than something else ; the quotation of 

 instances of the thing happening are quite irrelevant 

 to a proof of the law, unless there is at the same time a 

 careful collection of instances in which it does not happen. 

 Moreover, the clear distinction between a true law and 

 a statistical law is not generally appreciated. A statis- 

 tical law, which is really scientific, is made utterly 

 fallacious in its application because it is interpreted as 

 if it predicted the result of individual trials. 



For reasons which have been given already, many of 

 the laws that are most important in their practical 

 application are statistical laws ; and anyone, with a little 

 reflection, can suggest any number of examples of them. 

 Those which are generally familiar are usually entirely 

 fallacious (e.g. laws of weather and of heredity), and even 

 those which are true are habitually misapplied. The 



