126 COLLIGATIVE RELATIONS AND SCIENTIFIC LAWS 



that, due to the action of some "extraneous force," the fall has not 

 been "free." Thus safeguarded from all possibility of failure, the law 

 is then nothing but a convention? Nonsense! We lack absolutely de- 

 iiniti\'e criteria for the recognition in advance of a perfectly freely 

 falling body— just as we do for an ideal lever or pure sulfur— but we 

 do not lack all criteria. We need not then wait until after the fact to 

 rush in with postulated extraneous forces that explain away cases to 

 which Galileo's law should not have been applied in the first place. 

 Instead, as Poincare observes, generally we can recognize, in the 

 world of experience, those real systems sufficiently approximating the 

 ideal systems to which the law applies by definition. 



It would do me no good to have given the name of free fall to falls 

 which happen in conformity with Galileo's law, if I did not know that 

 elsewhere, in such circumstances, the fall will be probably free or 

 approximately free. That then is a law which may be true or false 

 [read "efficient or inefficient"], but which does not reduce to a 

 convention. 



Because we grasp the denotation of "freely falling body" the law is 

 for us no "mere convention": by it we are invested with power to pre- 

 dict what will be observed, and almost always ( though not invariably 

 or precisely) this is just what is observed. Thus, substituting "rela- 

 tion" for Poincare's first ambiguous use of "fact," we may say with 

 him: 



. . . all the scientist creates in a fact is the language in which he 

 enunciates it. If he predicts a fact, he will employ this language, and 

 for all those who can speak and understand it, his prediction is free 

 from ambiguity. Moreover, this prediction once made, it evidently 

 does not depend upon him whether it is fulfilled or not. 



Might one not still object that some colligative relations are so deeply 

 and generally involved, both in our predictions and in all our observa- 

 tions, that they can give rise to no recognizable predictive failures? 

 Russell recalls what at one time seemed a whole group of such 

 relations : 



Kant asserted that [Euclidean] geometi\' is based on an a priori intui- 

 tion of space and that experience coukl never contradict it because 

 space constitutes a part of our manner of perceiving the world. 



