A NOTE ON ''THE SCIENTIFIC METHOD^ 



enquiry is the mirror image of this one, to accept nothing 

 which is demonstrably false. 



No hypothesis is admissible in science that accounts only for 

 the facts it was expressly formulated to explain. Such an hypo- 

 thesis is inadmissible not, as we have seen, because it cannot be 

 verified but because it cannot even in principle be proved 

 untrue. Nothing can be done with an hypothesis that has no 

 'extra-muraP implications, and its acceptance and rejection 

 are equally acts of faith. 



'Testing an hypothesis' is the act of examining these extra- 

 mural implications. If they are true, the hypothesis is in some 

 recognizable but obscure way strengthened; if they are false, 

 the hypothesis is false. All fish have gills, but Socrates proved 

 to lack them; we must therefore think up some other explana- 

 tion of his having died. In practice, of course, we are not often 

 lucky enough to deal with such crisp disjunctions. An hypo- 

 thesis is less often outright false than merely inadequate, and 

 not beyond the help of running repairs. 



The act performed to test an hypothesis may be called an 

 *'experiment\ It is best to use the term in this simple and clean- 

 cut way, rather than to follow common use in restricting its 

 terms of reference to some sort of active messing-about with 

 nature. A ""mere observation** may in this sense be an experi- 

 ment, and if activity is insisted upon as a criterion, it may be 

 answered that even the merest observations cannot be made 

 from a supine position. The hypothesis which predicted the 

 existence of a planet Neptune was tested by the experiment of 

 directing a telescope towards a certain predicted region of the 

 sky. The existence of an experimental method in this generalized 

 sense is what distinguishes the scientific method from that of 

 any other sort of scholarly enquiry, and it is to this method that 

 science owes its power. It will be noticed that I have said 

 nothing here about the virtues of metrical analysis or the con- 

 trolled experiment, or all the many other things that to the 



77 



