270 THE EVOLUTION OF SCIENTIFIC THEORIES 



How strange that a theory should be valued primarily as a source of 

 new questions, rather than as a set of new answers to old questions. 

 Yet no stranger than that the Eskimo values the fish-hook more than 

 the fish that alone sustains his life, or that the pioneer woodsman 

 valued more the flint-and-steel than the fire which alone gave him 

 warmth. The flint-and-steel offers the means of making fires; the 

 hook, the means of catching fish; the scientific theory, the means of 

 winning new knowledge. 



A theory is expressly designed to provide accommodation for a 

 certain number of known relations. Its success in doing so we may 

 regard as testimony for its creator's ingenuity rather than for the 

 theory's own intrinsic soundness. After all, in logical principle we 

 accept the possibility that those same relations might be correlated 

 by some quite different theory, and Joseph Black long ago observed 

 that: 



A nice adaptation of conditions will make almost any hypothesis agree 

 with the phenomena. This will please the imagination but does not 

 advance our knowledge. 



However when, serving as a heuristic guide, a theory does "advance 

 our knowledge," we begin to think of it as more than just "a nice 

 adaptation of conditions." We may so advance in knowledge even of 

 relations already known. Thus, for example, a new theoretic deriva- 

 tion may suggest some large extrapolation of a relation earlier 

 thought limited in applicability or, contrariwise, indicate (by the 

 assumptions and approximations involved) sharp limitations to a 

 relation earlier thought general and exact. Again, suppose we find 

 that the theory can easily be extended to rationalize some relation(s) 

 quite outside its original scope. Thus \\^hewell observes that when 

 the long-familiar precession of die equinoxes was brought within the 

 explanatory framework of the Newtonian synthesis, originally de- 

 signed to deal with quite different motions, this "gave to the theory 

 a stamp of truth beyond the power of ingenuity to counterfeit." 

 Finally, who can fail to be impressed when the theory leads to en- 

 tirely new knowledge? 



Through Gibbs' phase rule we come at last to see a profound rela- 

 tion hidden in data already long available, but first rendered com- 

 prehensible by Gibbs' concept of chemical potential. Consider, too, 

 the striking example presented by Mendeleev's periodic classifica- 



