266 THE EVOLUTION OF SCIENTIFIC THEORIES 



judgment of its correlative efficiency, a brand new theory often fails 

 to shine in this respect. 



The insufficiency of the correlative index. Ordinarily the creator of 

 a new theory has conceived a new interpretation for some particular 

 data not smoothly accommodated in an older system of thought. 

 Often he can see in outline how to extend his conceptions to a much 

 larger body of data, but sometimes he cannot at once demonstrate 

 the applicability of his ideas even to all that the older system took in. 

 He will then exclude from consideration certain whole groups of 

 phenomena. He is the more disposed to this policy because usually, 

 between his theory and at least some of the excluded data, there will 

 be discrepancies he would have somehow to explain away were not 

 the whole group of phenomena excluded. As it is, he will almost cer- 

 tainly have encountered some discrepancies between his theory and 

 a few of even the relatively small central group of data on which the 

 theory turns. These discrepancies he must deal with, and most often 

 he does so by adding to his theory various ad hoc postulates— if only 

 such as argue the inaccuracy of the conflicting data. The supple- 

 mentary postulates, taken together with a very restricted scope of 

 application, may then burden the new theory with a correlative index 

 distinctly inferior to that of the established theory it would displace. 



Though it represented the labor of a lifetime, Copernicus' system 

 displays no clear superiority in correlative index. It drops entirely out 

 of correlation all the physics accommodated with the Ptolemaic sys- 

 tem as part of the scholastic synthesis; and, if it requires somewhat 

 fewer epicycles, it requires rather more numerous ad hoc postulates 

 to explain away the apparent fixity of the earth, the absence of de- 

 tectable parallax, etc. Consider as a second example Lavoisier's new 

 chemical system. Where the phlogiston theory had postulated few 

 elements but multiple states or conditions of phlogistication, Lavoi- 

 sier postulates many elements. Let us suppose that initially the num- 

 bers of postulates were roughhj equal. We will then judge Lavoisier's 

 theory superior for its convincingly tight correlation of the phenom- 

 ena of oxidation and reduction. But we must also observe that at its 

 first appearance this theory accommodated practically nothing else 

 and, even within the focal domain, Lavoisier concedes his theory's 

 incapacity to cope with Priestley's "stubborn facts." On the other 

 hand, the phlogiston theory took in (albeit loosely) a very large part 

 of chemistry— including appearances of colors and thermal eflFects 



