THE EVOLUTION OF SCIENTIFIC THEORIES 271 



tion. By it chemists are led to seek and find netc elements, with prop- 

 erties almost exactly those predicted by Mendeleev. When a theoreti- 

 cal construction thus guides us to major new knowledge, whether 

 of old or new relations, we obviously get out of it something more 

 than the ingenuity of the most perspicacious of creators could have 

 put into it. Have we not then attained conclusive evidence for the 

 fundamental soundness of that construction? And have we not also 

 identified, at last, the critical factor in judgment of scientific theo- 

 ries? The decisive criterion is heuristic power, and some particularly 

 striking display of such power may then be taken, as Pascal took the 

 Puy de Dome experiment, to be an experimentum cruets. But even 

 yet our story is far from fully told. After all, for many of Pascal's con- 

 temporaries his was not a crucial experiment, and the theory of 

 horror vacui long survived Pascal himself. 



When we draw from a theory new questions that lead us to im- 

 portant discoveries, we do not thereby establish the validity of the 

 theory. Polanyi notes that some quite erroneous values for the ratio 

 of the atomic weights of hydrogen and oxygen prompted Urey's 

 search for a heavier isotope of hydrogen, for the discovery of which 

 he was awarded a Nobel prize. Of a notable theoretical advance 

 Fierz remarks : 



... it may well be that on the basis of apparently wrong and irra- 

 tional speculations a physicist discovers something that is correct and 

 important. This was the case with Dirac's theory of the electron 

 where the dogmatic postulate (which is also quite wrong) that his 

 equations should be of first order, finally led to the correct theory. 



Observe, too, that the discoveries made with a given theory may 

 derive less from its intrinsic virtues than from the superior talent ( or 

 energy or facilities ) of a man or men who adopt it. Newton accom- 

 plished far more with his clumsy geometrized calculus than Leibniz 

 did with his. Observe further that th6 root of heuristic success may 

 lie not in the explicit content of a theory but in some implicit attitude 

 associated with it: thus it is, I think, that teleology has led to some 

 substantial discoveries. Observe finally that, in De Morgan's termi- 

 nology, "wrong theories" can be "rightly worked": today we reject 

 the phlogiston theory that led Priestley to discover photosynthesis 

 and Scheele to the first discovery of oxygen. Thus even the gaudiest 

 displays of heuristic power cannot coerce our acceptance of a theory^ 



