THE UNIQUENESS OF THE INDIVIDUAL 



hypothesis may be false although the inferences drawn from it 

 are themselves empirically true. This combination is by no 

 means disalloAved by logic. Consider that paradigm of empirical 

 facts, the mortality of Socrates. If Socrates is a fish and all fish 

 mortal, it follows with pitiless logical rigour that Socrates is 

 mortal too. This is perfectly straightforward deduction, and 

 Socrates'* mortality is adequately so explained. How then does 

 the scientist prove his hypotheses? The answer, now widely 

 accepted, is that except in certain limiting instances he never 

 does. No concept is so maltreated by lay usage as that of proof. 

 In a strictly formal sense, accepted hypotheses remain per- 

 petually on probation; one does not prove them true, though 

 one may often act exactly as if they were. But what the scientist 

 can often do with complete logical precision is to disprove 

 hypotheses. If what follows from an hypothesis is false, then the 

 hypothesis is false, and false in logic. This consequence of the 

 asymmetry of the process of Smplying** is a central property of 

 scientific method, and it influences experimental design in a 

 direct and conspicuous way: many experimental designs are 

 simply well-laid traps to lure on a so-called 7iull hypothesis and 

 then confound it. The precision of the act of disproof is thus 

 very far from being a formalistic fancy. This does not mean, of 

 course, that the accepted hypothesis is merely 'not disproven"*; 

 there are obviously degrees of certitude of conviction, but these 

 are for the most part informally worked out. It is clear that 

 an hypothesis gains in acceptability merely by its fitting in to 

 a wider theoretical scheme of which it is a part. In this way 

 hypotheses may sustain each other.* At all events, the scientist 

 would soon be beggared by Descartes'* first precept of intel- 

 lectual enquiry- -""c/e ne recevoir jamais aucune chose pour vraie 

 que je ne la con?iusse evidemment etre telle''. His own precept of 



* [I confess here to the fault of having lumped all hypotheses together, 

 as if they were a single logical species; for a careful analysis of their several 

 forms, see Probability and Induction, by W. Kneale, Oxford, 1949.] 



76 



