ABSTRACT
I f we were confined to deductive reasoning we should be gravely inconvenienced. To say this is indeed to speak too mildly. We should not be able to reach any conclusion con cerning matters of fact that went ‘beyond the present testimony of our senses, or the records of our memory5.']* Generalization (i.e. going beyond the evidence) is essential to carrying on the affairs of our daily lives; it lies at the very foundation of all the empirical sciences. All the sciences except logic and mathematics are empirical; they are based upon observation, experiment and generalizations from ex perience. Generalization from a number of observed instances of a certain class, which are assumed not to constitute all the instances of the class, is called ‘Induction by simple enumera tion5. Its logical foim is: All the observed S's are P's; therefore all S's are P's. This inference is clearly not valid, for, in inferring from a premiss about some S's a conclusion about all S's, there is an illicit distribution of S. Consequently, the premiss may be true although the conclusion is false. This is an essential characteristic of inductive reasoning. All valid reasoning is deductive, but it does not follow from this that inductive reasoning is unreasonable, unworthy of a clear thinker. What does follow is that we must find other criteria with which to check and control our reasoning than the criteria provided by the rules of deductive reasoning. It is far more difficult to discover these criteria, to make them explicit, and to formulate rules than is the case with deduction. To do so
constitutes one of the main problems of what is known as the ‘methodology of science5, i.e. a systematic investigation of the logical character of the methods employed in the empirical sciences. It must be admitted that this investigation is still in a stage that may be described as rudimentary.
