THE EVOLUTION OF SCIENTIFIC THEORIES 275 



world that took it for granted that air is an element. Ideas that do 

 survive such testing by the conceptual environment have then still to 

 show themselves viable in the empirical environment. That is, they 

 must be capable of productive exploitation by the empirical tech- 

 niques available. In the then-contemporary state of science, ancient 

 atomism probably could not have issued in any works of value. Simi- 

 larly, though we detect important elements of soundness in Berzelius' 

 idea of electrostatic bonding, it was rejected in the first half of the 

 19th century because it did not, perhaps could not then, lead to any 

 important new discoveries. 



If it is to prove viable, a scientific theory must, in two distinct re- 

 spects, fit its times. Obviously then ideas that fail at one time may 

 later succeed where formerly they failed (or vice versa). Newton's 

 corpuscular theory of light triumphs in the 18th century; given the 

 phenomena of polarization and di£Fraction, the undulatory theory 

 triumphs in the 19th; given quantum eflFects, a new corpuscular view 

 flourishes in the 20th. The times thus ripen toward the domi- 

 nance of an opinion at one time almost extinct. Large changes in the 

 intellectual climate permit Copernicus to gain for the heliocentric 

 theory a hearing that had been denied Aristarchus; small changes 

 make it possible for Cannizzaro to succeed where Avogadro had 

 failed. Small advances in experimental techniques permit Pasteur to 

 succeed where before him Spallanzani had failed; large advances in 

 observational techniques permit Kepler and Galileo to consolidate 

 the position of the Copernican theory as Copernicus himself could 

 not have. 



A notably inadequate theory may enjoy entirely undue viability if 

 (like the Ptolemaic theory) it for long confronts no serious competi- 

 tor. And two or more competing theories may prove almost equally 

 viable simply because they are all almost equally ( in ) eflFectual heu- 

 ristic guides. We see readily that in such cases— which occur ordi- 

 narily only in the earliest stages of development of a science— what 

 fails is the conditions for natural selection, not natural selection as 

 such. But we may think to observe also some genuine failures of that 

 mechanism. When "rightly worked," quite "wrong theories" may be 

 chosen by natural selection— and quite right theories rejected. Ap- 

 parent miscarriages of this sort may distress us, but should they do 

 so? Conceiving scientific theories as primarily heuristic tools, we may 

 well expect the verdicts of natural selection to be almost invariably 



