THE EVOLUTION OF SCIENTIFIC THEORIES 263 



The crucial experiment in practice. The possibility of falsification, 

 so simple and definitive in principle, seems to have dissolved into 

 hopelessly complex equivocality as a matter of scientific practice. 

 Yet the crucial experiment is not ivholly a myth: on some historical 

 occasions one or a very few observations or experiments have been 

 widely accepted as rendering a decisive verdict. This is just the 

 dramatic efi^ect we expect of that species of crucial experiment the 

 logical impossibility of which Duhem and Quine emphasize. How 

 then is any such eflFect produced? 



An experiment may function as crucial simply because its result is, 

 quantitatively and/or qualitatively, so completely out of line with 

 anything foreseen by the upholders of one theory, though perhaps 

 predictable on another. After the fact that first theory might be 

 "saved" by any of the many expedients we have indicated. But the 

 brute blow of the totally unexpected can stun resistance and paralyze 

 any such rescue operation. The crucial eflFect derives then not from 

 logical force but from psychological impact. 



Consider too the eflFect of the sociological context. All the partici- 

 pants in a modern scientific dispute generally agree that a very great 

 deal is to be taken as of principle, insusceptible to change. Hence, 

 though many other relations and theories are involved in the test of 

 some one disputed proposition, even those on the losing side of a 

 crucial experiment may not seek to deny its force by impugning 

 other parts of scientific knowledge. Arago's proposed experiment was 

 made crucial when all concerned, in eflFect, agreed in advance simply 

 to rule out of court a great variety of expedient assumptions that, 

 after the fact, might yet have been used to "save" whichever theory 

 the experiment discredited. In part producer of the unity of organ- 

 ized science, the crucial experiment is thus also in part product of 

 that unity. 



Beyond the logical, the psychological, and the sociological con- 

 texts, we come at last to the most important of all: the scientific con- 

 text. We can no more suppose a major development in science the 

 eflFect of a single crucial experiment than we can suppose World War 

 I the result solely of the assassination of the Archduke Ferdinand. A 

 successful scientific theory renders indispensable services in the 

 correlation and explanation of what we already know and, as a heu- 

 ristic device, in supporting our search for new knowledge. Until we 

 find some alternative theory that promises to perform these services 



