262 THE EVOLUTION OF SCIENTIFIC THEORIES 



of that hierarchy of theoretic propositions noted in Chapter VII. 

 For, indeed, if nothing could be held comparati\'ely well established, 

 science would promptly collapse into chaos. When some wild idea 

 of mine has failed an experimental test, certainly I am not disposed 

 to argue (nor my colleagues to hear the argument) that my idea is 

 perfectly sound, and that what has actually failed is one or more of 

 the substantive principles necessarily assumed in making the test. 

 Thus Feigl correctly stresses that some ("established") propositions 

 will and must always be maintained far more tenaciously than other 

 ( "hypothetical") propositions. 



A view that maintains that the whole body of a scientific theory (if 

 not of all science) confronts experience and that modifications may 

 be required in any part of the system if it does not "fit,"— such a view 

 obscures dangerously what is of the greatest importance for the prog- 

 ress of science: the successive testing and securing of parts of science 

 —at least in the sense of an approximation. Naturally, no part can be 

 considered as established with finality— but this insight which im- 

 presses the pure logician should not blind him to the recognition of 

 the method of successive confirmation. 



\\ holly inadequate as an account of ordinary scientific practice, 

 Quine's statement correctly depicts the situation in logical principle. 

 And on rare occasions some such situation is found in practice. Very 

 seldom do we propose to jettison a well-founded relation to save one 

 less well-established— though Marignac was apparently quite pre- 

 pared so to jettison the proposition that masses are additive, thereby 

 to preserve Front's highly speculative hypothesis in the face of con- 

 tradictory data. But not uncommonly we will propose to jettison one 

 relation in order to preserve another not much less firmly founded— 

 as when Mendeleev seeks to preserve his periodic classification by 

 questioning the general applicability of Dulong and Petit's law, 

 which indicated for beryllium an atomic weight incompatible with 

 that classification. And always, finding it almost literally unthinkable 

 that there is some fundamental unsoundness in a major theory to 

 which we can conceive no alternative whatever, we will jettison 

 practically anything and everything to preserve that theory. Thus, to 

 sa\'e classical electrodynamics in the face of the result of the Michel- 

 son-Morley experiment, Fitzgerald proposes that the length of, say, 

 a meter stick is not constant but systematically variable with its 

 velocity. 



