222 INTRODUCTION TO PHILOSOPHY OF SCIENCE 



probable according as the realm of the clearly given exhibits 

 events more or less like the hypothesis in question. If there 

 are events of a closely similar character, the empirical 

 generalizations descriptive of the behavior of such events 

 may be taken as the foundation for predictions from the 

 hypothesis, and such predictions will have a certain plausi- 

 bility ; but if there are no events of a closely similar character, 

 the only foundation for the prediction is some less closely 

 related empirical generalization which is then imaginatively 

 modified through such activities as abstraction and idealiza- 

 tion. Predictions in such cases will have a correspondingly 

 lower degree of plausibility. 



Reference should be made to a further feature of the pre- 

 dictive act. In order that the verificatory movement may 

 be carried out it is not sufficient simply that implications 

 be drawn from the hypothesis; the consequences must be 

 such as can be either confirmed or infirmed x empirically. 

 Ry this is meant not that the predictions must be such as 

 can be corroborated by data, though such must be the 

 character of the predictions if the theory is to be estab- 

 lished; what is meant is, rather, that the predictions must 

 be of the general kind which have relevance to empirical 

 data, i.e., they must be of the kind which can be unequivo- 

 cally confirmed or infirmed by experience. Hence the draw- 

 ing of an implication which demands the performance of an 

 impossible physical experiment, or the assuming of an im- 

 possible observational point of view, would be illegitimate. 

 Furthermore, the implications must be such as to refer to 

 the range of the more clearly given data; no verification of 

 hypotheses, which by their very nature refer to the obscurely 

 given events, can be made if the implications drawn from 

 the hypotheses do not take one out of the realm of obscurity. 

 From this it follows that verifications of highly abstract 

 hypotheses (e.g., the law of universal gravitation) cannot be 



1 The term "infirm" may be used in what follows to designate negative verifica- 

 tion or negative confirmation. Thus the proposition "Today is hot" would be con- 

 firmed if the day is hot, but would be infirmed if the day is cold. This use is closely 

 similar to that of Nicod, though not identical with it. See Chapter XVI, p. 361. 



