THE EVOLUTION OF SCIENTIFIC THEORIES 283 



are lifted by conceptual refinements, additions will be made to a 

 numerator which will in any case be still more spectacularly en- 

 larged by new discoveries. For example, at the outset only a very 

 few known relations lent themselves to interpretation by Torricelli's 

 new theory, which however led to the discovery of a great many 

 new relations in barometry and elsewhere. These discoveries fit 

 smoothly into the theory that predicted them. On the other hand, 

 they can be accommodated in the theory of horror vacui only with 

 the aid of supplementary postulates which decrease the correlative 

 index of that theory. Thus, given the heuristic power of Torricelli's 

 theory, it is almost automatically assured an ultimately superior cor- 

 relative index. 



Displays of heuristic power also aflFect explanatory appeal. When 

 Newton postulates a universal gravitational attraction he asks us to 

 accept a purely hypothetical force. Guided by Newton's theory. Cav- 

 endish at last detects the action of that force in the laboratory. What 

 Newton proposes is then no longer strictly hypothetical, but only an 

 immense generalization of what we can observe. In somewhat the 

 same way, Copernicus' "absurd" postulate of a moving earth pro- 

 vokes multiple unsuccessful searches for stellar parallax. But these 

 do lead to the detection of stellar aberration, which is at last some 

 reasonably clear indication that the earth does have an annual mo- 

 tion. Such gains in the explanatory appeal of the heuristically power- 

 ful theory are complemented by losses in the appeal of its competi- 

 tor. Copernicus' theory leads men to expect the phase cycle of Venus. 

 That cycle discovered, the Ptolemaic system can account for it only 

 with the aid of supplementary assumptions of no great plausibility. 



Given the above processes, we may not be long in recognizing 

 some difference between the number and character of the supple- 

 mentary postulates required to support a theory in its decrepitude 

 and those that for a time sustain a theory in its earliest youth. Thus 

 we distinguish different kinds of weakness in the infant who pushes 

 a chair in front of him as he learns to walk and in the ancient who 

 drags himself around on crutches. Overloaded with supplementary 

 assumptions, the older theory becomes functionally inadequate. That 

 is, it becomes difficult to handle, its adherents begin to use it in 

 different ways, and communication between them may begin to fail. 

 Worst of all, any lingering remnants of heuristic power are impaired. 

 The theory grows so complex that unequivocal predictions are not 



