THE EVOLUTION OF SCIENTIFIC THEORIES 279 



has functioned successfully. Where in the theory's youth we were 

 happy with small displays of capacity, we now "prove" it by de- 

 manding the utmost of it. No longer fearing the challenge of "anom- 

 alies" instead, as Poincare observes, we actively seek them. 



. . . after the rule is well established, after it is beyond all doubt, 

 the facts in full conformity with it are erelong without interest since 

 they no longer teach us anything new. It is then the exception which 

 becomes important. We cease to seek resemblances; we devote our- 

 selves above all to the differences, and among the differences are 

 chosen first the most accentuated, not only because they are the most 

 striking, but because they will be the most instructive. 



Incredible as it may seem, a great theoretic pattern like Newton's 

 long sustains all these stresses with little or no manifestation of 

 strain. By such stressing, in eflFect we court strain. We find instead an 

 uninterrupted sequence of brilliant successes extending over cen- 

 turies. Bit by bit we lose sight of the principle of corrigible fallibil- 

 ity: no longer do we suppose that the theory ever can fail. And when 

 the ( invariably equivocal ) first appearances of strain materialize, as 

 sooner or later they always do, the general reaction is likely to be 

 very much that Polanyi describes: 



Just as the eye sees details that are not there if they fit in with the 

 sense of the picture, or overlooks them if they make no sense, so also 

 very little inherent certainty will suffice to secure the highest scientific 

 value to an alleged fact, if only it fits in with a great scientific generali- 

 zation, while the most stubborn facts will be set aside if there is no 

 place for them in the established framework of science. 



Even in the unlikely event that these are conceded puzzling facts, 

 the vast majority of scientists will still remain blithely confident that 

 future developments will certainly resolve all such puzzles— if only 

 in some way ( s ) not yet imaginable. Ripening into honored old age, 

 the theory now begins imperceptibly to^ drift toward senescence. 



Old age. Strain sometimes develops simply from the accumulation 

 of precisely the kind of data the theory was designed to handle. Thus 

 the astronomical data collected over centuries produce stresses the 

 Ptolemaic system can sustain only by an ever-increasing elaboration 

 of ever-creakier epicyclic machinery. In other cases the strain de- 

 velops less from accumulation than from refinement of the focal 

 data. Thus Brahe's data subject both the Ptolemaic system and the 



