ISO THE SCIENCE OF LOGIC 



find &quot; facts &quot; discordant with our hypothesis, we must make sure 

 we have interpreted these facts rightly. Then we must see if 

 they can be made to fit in with our hypothesis by such correc 

 tion and modification of the latter as will incorporate in it factors 

 to account for those facts. Only when we fail to effect such a modi 

 fication of our hypothesis must the latter be rejected altogether. 



3. It must be based on some analogy with known causes : it 

 must be capable of yielding exact deductive inferences : it must be 

 verifiable by the submission of those inferences to the control of ob 

 servation or experiment. These are three alternative statements 

 of one and the same requirement. We have seen already that it 

 may not be always possible to conceive an hypothesis which will 

 fulfil this condition: the degree in which an hypothesis does 

 lend itself to such verification is the measure of its &quot; exact &quot; 

 scientific character. 



The combination of conditions I and 2 show how the func 

 tions of reason and sense alternate and aid each other in science. 

 Initial observations suggest an hypothesis. This in turn must be 

 verified : and to verify it the scientist must reason from it, and 

 submit his conclusions anew to the control of observation or ex 

 periment. &quot; Thus,&quot; writes Claude Bernard, &quot; the mind of the 

 scientist is placed between two observations, one which is for 

 him the starting-point of a reasoning process, the other its con 

 clusion.&quot; x 



4. // is &quot; verified&quot; or &quot; established&quot; when it is shown to yield 

 not merely a sufficient explanation, but the only possible explanation, 

 of the facts it purports to account for (cf. 2 1 2). Our success in 

 showing this will vary with the nature of the facts and the scope 

 of the hypothesis. Sometimes the facts are subject to the con 

 trol of experiment, and the hypothesis is comparatively restricted 

 in its scope, so that we are able to eliminate and disprove all 

 conceivable alternatives, and thus attain to the ideal of a rigorous 

 verification. Again, the hypothesis, may be shown to be capable 

 of such extension, by consilience of inductions, that although we 

 may not hope to prove rigorously that it is the only possible ex 

 planation of the facts, yet we are able to show that it does 

 explain a vast field of fact, and does so more satisfactorily than 

 any suggested alternative : in which case we may give it a pro 

 visional assent amounting to moral certitude. 2 



l op. cit., ibid. apud MERCIER, Logique, pp., 344-5. 



3 It is often very difficult to distinguish, and there is often no practical distinc- 





