METHOD OF DISCOVERING CAUSAL LAWS 167 



that whatever can be eliminated without interfering with a pheno 

 menon is not causally connected with the latter. 



There is another principle of equal importance, at which we 

 arrive by the following simple consideration. No variety of 

 positive instances, i.e. instances in which the phenomenon occurs, 

 no matter how great the number and variety ought to satisfy 

 the investigator that he has successfully segregated all the es 

 sential conditions of the phenomenon from its accidental accom 

 paniments, if he can make an experiment whereby he will be able 

 to remove the supposed causal conditions from a positive instance, 

 in order to see whether by so doing he will thereby remove or 

 eliminate the phenomenon itself: thus changing the whole into a 

 negative instance, i.e. one in which the phenomenon does not occur. 

 For, if he can thus successfully change a positive into a negative 

 instance (or vice versa} by removing (or introducing) the supposed 

 cause, he will by this mode of elimination have secured greater 

 certitude about the accuracy of his hypothesis than any variety of 

 positive instances could give him. The principle underlying this 

 mode of elimination is that whatever cannot be removed or elimin 

 ated without interfering with a phenomenon is causally connected 

 with the latter. 



In those two principles we use the terms &quot; cause&quot; and &quot;causal 

 connexion &quot; in the strictest sense, i.e. as reciprocating. They are 

 simple principles in theory, but often not easy to apply satisfac 

 torily in practice. This we shall see presently, from an examina 

 tion of the various experimental &quot;methods,&quot; or &quot;rules,&quot; or 

 &quot; canons,&quot; which are merely so many ways of attempting to apply 

 the principles just formulated. 



In the course of our observations, whether simple or experi 

 mental, an instance or instances may occur in which the effect is 

 absent though the supposed cause is present, or vice versa. Such 

 instances are called exceptions, or exceptional instances? If on care 

 ful examination these turn out to be not merely apparent but real 

 exceptions, they will, of course, oblige us either to modify our 

 hypothesis or to abandon it for a different one. For this reason 

 they are of the greatest importance in inductive research. One 

 single real exception to an hypothesis, i.e. to a supposed law, is 

 sufficient to disprove the latter as it actually stands. 2 But, at the 



1 The term &quot; negative instance &quot; is sometimes applied, in a restricted sense, to an 

 instance in which the phenomenon does not occur though the supposed cause is 

 present. 



The legal aphorism, Exceptio probat regulam, is sometimes applied in a confus- 



