182 INTRODUCTION TO PHILOSOPHY OF SCIENCE 



None of these relations, however, is sufficiently deter- 

 minate to permit the ascertainment of one hypothesis to the 

 exclusion of all alternates. One is obliged to say, therefore, 

 not that any hypothesis is proved by being derived from the 

 data, but that it is given a certain probability by virtue of the 

 more or less definite hints offered by the data. Data do not 

 imply an hypothesis, they merely suggest it. The function of 

 hypotheses in the total scientific situation makes it ex- 

 tremely unplausible that an explanatory conception could 

 be read out of the data in a necessary way. Hypotheses, as 

 will be seen later, serve as instruments of anticipation; they 

 are devices by which the accumulated knowledge of nature 

 is combined with the imaginative envisagement of possibil- 

 ities, in such a way as to predict the character of the still 

 unexplored portions of nature. It is unlikely that the un- 

 explored is exactly like the explored; hence nature as known 

 cannot imply nature as unknown. Yet it is equally unlikely 

 that the unexplored is totally different from the explored; 

 hence nature as known must offer the only available hints 

 as to the character of nature as unknown. These two facts 

 suggest that prediction is possible only by what Tyndall 

 calls a prepared imagination, i.e., an imagination which en- 

 deavors to include all relevant data as guiding factors in 

 determining the creative insight. 



A logic of discovery is possible, therefore, only in a limited 

 sense. It offers no technique of prediction. But it does offer a 

 technique by which the imagination may be prepared for 

 the creative leap. Much that is ordinarily included in the 

 act of discovery is of this preparatory character. Hence in 

 recognizing the principles of a logic of discovery one is in a 

 better position to see that the act is really of a twofold 

 character — one part capable of a certain formalization, the 

 other essentially mysterious and unpredictable. Though 

 this approach does not solve the problem of scientific dis- 

 covery as traditionally conceived, it does reduce it to two 

 more sharply defined problems, each of which is capable or 

 incapable of solution upon its own grounds. 



