SCIENTIFIC DISCOVERY 187 



explain actual levers in terms of perfect levers, matter in 

 terms of space-time, heat in terms of molecules, or life in 

 terms of a vital force, if in each case the explanatory notion 

 is a mere construction from that which it presumes to ex- 

 plain. It is likely that many explanations in science are of 

 this pseudo-hypothetical character. Mention may be made 

 of potentials, capacities, unconscious purposes, tendencies, 

 instincts, affinities, powers, urges, and causal efficacies. 

 There is reason to believe that most of these notions — unless 

 they are definitely increased through the hypothetical 

 method — are constructs rather than hypotheses, and hence 

 have no explanatory function. 



It follows that the problem of the logic of discovery reduces 

 to two more specific problems: (a) What are the techniques 

 of construction or of elaboration? (b) What are the techniques 

 of hypothesis-formation? Unfortunately no very sharp line 

 can be drawn between these problems. The task of the logic 

 of discovery, presumably, is to formulate the techniques by 

 which nature may be anticipated. But this problem is not 

 likelv to admit of solution. The best that one can do is to 

 list the various techniques which may be employed as prac- 

 tical aids to the performance of the act of discovery. But an 

 examination of these methods reveals the fact that they vary 

 among themselves in determinateness. By determinateness 

 is meant the degree to which the operation permits one to 

 infer the content of the explanatory entity from the data. It 

 has already been pointed out that the data never imply the 

 hypothesis, yet they do seem to determine its content in 

 certain limited ways. The purpose of a classification of the 

 techniques of discovery is to exhibit the most common of 

 these methods in such a way as to disclose their degrees of 

 determinateness. Abstraction, for example, is relatively 

 highly determinate; synthesis and analysis, on the other 

 hand, are not so highly determinate since logic has not yet 

 provided the principles according to which the properties 

 of wholes may be inferred from the properties of parts, or 

 vice versa. 



