116 COLLIGATR-E RELATIONS AND SCIENTIFIC LAWS 



materializes: some of the preliminary measurements may have been 

 faulty, some of the many variables I thought it safe to ignore may 

 actually have affected the outcome of the test, etc. I can still argue 

 that the law of the lever is absolute, exact, and "true" as applied to 

 the hypothetical ideal lever: by definition that retreat position is al- 

 ways secure. But as a colligative relation usefully applicable to actual 

 lever systems the law is clearly contingent ( since prediction depends 

 on potentially fallible antecedent determinations of conditions ) , ap- 

 proximate (since all the possibly relevant conditions cannot be con- 

 trolled or determined), and at most efficient (to the extent that it 

 represents an optimal reconciliation of our conflicting desires for ease 

 and generality of application and for reliability ) . 



That our desires conflict is nowhere clearer than in the last ex- 

 ample. As ordinarily conceived, the law of the lever is a relation of 

 great generality. But, seeking to draw from it predictions of maxi- 

 mum reliability, we had sharply to slirink its applicability. In effect, 

 we limited it to systems approximating an analytical balance under 

 the closely controlled conditions of the laboratory— to systems ap- 

 proximating in their extreme artificiality the extreme abstraction of 

 tlie ideal lever. Meyerson writes : 



Doubtless, if nature were not ordered, if it did not present us with 

 similar objects, capable of furnishing generalized concepts, we could 

 not formulate laws . . . 



In fact, we only attain laws by violating nature, by isolating more 

 or less artificially a phenomenon from the whole, by checking those 

 influences which would have falsified the observation. Thus the law 

 cannot directly express reality. The phenomenon as it is envisaged by 

 it, the "pure" phenomenon, is rarely observed without our interven- 

 tion, and even with this it remains imperfect, disturbed by accessory 

 phenomena. 



The artificiality of the special lever system with which we worked 

 reflects our attempt to isolate it from the rest of the universe, to in- 

 sulate it from the manifold accessory factors operative in the raw 

 phenomenon but omitted in the statement of a law that refers ex- 

 plicitly only to certain weights and distances. Ordinarily sufficient 

 (though not perfect) isolation, insulation, can be achieved: that is 

 the message of the principle of dissolubility. And ordinarily we arrive 

 at acceptable predictions without taking account of more than a very 

 few complicating factors. We do not, cannot ordinarily, even con- 



