228 INTRODUCTION TO PHILOSOPHY OF SCIENCE 



gregate are eliminated and replaced by others. The infirmed 

 prediction may have followed from certain of the proposi- 

 tions, but not from others. All that is required, therefore, 

 is to remove the undesirable elements, and replace them by 

 propositions which will have other implications. It is a 

 noteworthy fact that in the history of science no hypothesis 

 seems ever to have been completely discarded; it may have 

 gone through such extreme modifications as to be hardly 

 recognizable, but there is usually a permanent core which 

 persistently resists inflrmation. This is probably that part 

 which has been derived through the method of construction. 

 Around this center there gathers through successive acts of 

 imagination, a variety of fringes, each of which determines 

 certain predictions, to be subjected to verification. Those 

 predictions which receive confirmation become incorporated 

 into the hypothesis as part of its essential core; those which 

 receive inflrmation are rejected and replaced by alternates. 

 In this way the hypothesis expands and contracts, meanwhile 

 growing progressively in such a way as to retain both past 

 successes and past failures. It is still convenient, at times, 

 to explain in terms of the geocentric theory of the heavens, 

 the caloric theory of heat, the fluid theory of electricity, 

 and the homuncular theory of the origin of life. All of 

 these theories did actually explain at one time, in spite of 

 misleading and erroneous implications which can be drawn 

 from them. Progress in science consists not in the complete 

 abandonment of infirmed theories, but in their modification 

 in such a way as to retain their positive explanatory features. 

 In the second place, inflrmation may require not modifica- 

 tion or rejection of the hypothesis itself but merely a re- 

 examination of the reasons for supposing that the hypothesis 

 did imply the proposition which was infirmed. It has already 

 been shown that there are no rules for the drawing of con- 

 tentual implications. When the hypothetical entities are 

 such as have never been observed, one is quite unable to 

 assert with certainty anything as to their consequences. 

 Hence an infirmed prediction may prove to have been er- 



