THE EVOLUTION OF SCIENTIFIC THEORIES 291 



when computing the fuel costs for a projected motor tour, one pro- 

 ceeds by way of the AristoteHan view that continued motion de- 

 mands a mover; and not at all by beginning with the Newtonian con- 

 cept of inertia, and then proceeding to make corrections for all items 

 of frictional resistance. Even more to the point, observe the notable 

 survival of the geocentric system "falsified" in the earliest of "scien- 

 tific revolutions." The manifold relations originally conceived and 

 expressed in geocentric terms are handled with matchless felicity in 

 that frame of reference. And so to this very day the navigators and 

 surveyors, who are most concerned with just those relations, continue 

 to use the geocentric system. If, as I suppose, the geocentric system 

 is not meaningfully a limiting case of the heliocentric system, then 

 the correspondence principle does not apply. Nonetheless, the older 

 theory remains active in the presence of the "new" theory by which 

 it has been subordinated. Nobody "believes" the geocentric system 

 in any cosmologic sense, nor does anybody so "believe" the caloric 

 theory. Yet do we not still use the latter much as we do the former? 

 Though we never mention a caloric fluid, in calorimetric( !) work do 

 we really conceive ourselves as measuring motion or a substantive 

 heat? A "heat pump" doesn't supply motion: a pump pumps some- 

 thing, if not heat then energy. What we conceive to floiv or pass 

 through a thermally permeable membrane is surely not motion but 

 heat. OflFering a felicitously "natural" conception of the simple ther- 

 mal data it was designed to handle, the caloric theory remains still 

 useful and in covert use. It has not been falsified, but only subordi- 

 nated, by the far more widely applicable kinetic theory, of which it 

 is not meaningfully a limiting case. 



However it may be in these older instances, the phenomenon of 

 incomplete displacement is perhaps the most conspicuous feature of 

 the more recent "revolutions" to which the correspondence principle 

 does apply. The advent of relativistic mechanics leaves classical me- 

 chanics in complete command of the classical domain. For this is, of 

 course, just the domain of all those limiting cases that classical me- 

 chanics—as the limiting form of relativistic mechanics— handles very 

 nearly to perfection. Thus, for example, the accuracy of purely classi- 

 cal celestial dynamics is perhaps best indicated by the magnitude of 

 its one "serious" anomaly: the predicted rate of precession of the 

 perihelion of Mercury fails by a margin that would represent an 

 accumulated error of one degree of arc in 8700 years. Obviously, we 



