THE EVOLUTION OF SCIENTIFIC THEORIES 267 



that go entirely unmentioned by Lavoisier's original theory, though 

 they are perhaps the most striking manifestations of its focal phe- 

 nomena. Consider finally the interpretation of the photoelectric eflFect 

 proposed by Einstein. This interpretation excluded practically every- 

 thing but that one complex of e£Fects. Yet it introduced a major 

 element of discontinuity gravely out of keeping with the theory of 

 classical electrodynamics by which an enormous number of relations 

 were accommodated. The correlative index of Einstein's theory 

 might then be held contemptible; and his work was indeed harshly 

 criticized by Planck, much as Copernicus' was criticized by Bacon, 

 as an irresponsible innovation that leaves entirely out of account 

 much more than it takes in. 



These examples amply suggest that, if certain new theories are 

 ever first to be considered, the correlative index must not figure as an 

 immediately decisive criterion of judgment. A number of quite dif- 

 ferent examples suggest that in fact it is not so regarded. Newton 

 unites astronomy and terrestrial physics in a single system with cor- 

 relative index approached by no other theory of his time. But some 

 of Newton's not least acute contemporaries rejected his theory. 

 Darwin proposes a theory with correlative index of a very high order, 

 at first rejected by many if not most contemporary biologists. Such 

 rejections are more probable when, as in the cited cases, the new 

 theory seems to have cosmologic overtones; but they occur even 

 when no such overtones are present. Arrhenius' theory of ionization 

 handsomely joined the phenomena of electrochemistry with a multi- 

 tude of other phenomena of chemical reactivity, color, osmotic pres- 

 sure, vapor pressure, etc. It represented a major advance in correla- 

 tion, and it too was quite widely rejected on its first appearance. 



EXPLANATORY APPEAL 



Presumably what is for us explained is only what we find derivable 

 in some system of correlation. If so, there can be no explanation save 

 by way of correlation, but this does not at all establish that explana- 

 tion is nothing but correlation. Indeed, when men reject new the- 

 ories that represent substantial advances in correlative efficiency, 

 they generally do so because they find inadequate the explanatory 

 appeal of those theories. Clearly then something more than correla- 

 tive efficiency must enter into the making of explanatory appeal. 

 A theory's explanatory appeal usually depends primarily on the 



