40 SCIENCE (and common sense) 



must believe in science [which Bernard here equates with detennin- 

 ism] and doubt their means of investigation. 



Spawn of our faith in the principle of determinism, the criterion of 

 regularity here comes in play. 



Nowhere is the need for judgment more apparent than in use of the 

 criterion of regularity. Polanyi remarks that crystallographic analysis 



. . . defines a polyhedron which is taken to be the theoretical shape 

 of a crystal specimen. It embodies only such aspects of the specimen 

 as are deemed regular and in respect to these it is required to fit the 

 facts of experience; but otherwise, however widely the crystal speci- 

 men deviates from the theory, this will be [fiiTnly ignored and] put 

 down as a shortcoming of the crystal and not of the theory. 



Theoretic judgments are necessarily involved whenever the scientist 

 brings the criterion of regularity to bear and, since his theories or his 

 interpretations of them may err, his judgments may err. Yet always 

 he must apply the criterion of regularity, always he must make a be- 

 ginning somewhere— with just those data he judges potentially re- 

 ducible to order. 



The detailed course of a feather falling in air is totally unpredict- 

 able in the view of common sense and, beyond noticing that the 

 feather falls irregularly, common sense does not concern itself with 

 this matter. The scientist judges that the fall of the feather is 

 completely determined in principle, but complexly determined by 

 multiple conditions difficult to evaluate, much more difficult to con- 

 trol. Such falls are recurrent (though not reproducible); at least ap- 

 proximately concurrent descriptions of them can be obtained from 

 diflFerent observers— and wholly unexceptionable descriptions from 

 high-speed stroboscopic photography. But these data seem still so 

 little accessible to reduction to order ( and, given our theoretic views, 

 so little likely to teach us anything new) that they are rejected as 

 subject matter. 



Consider mirages. Irreproducible in detail, they are yet recurrent. 

 Moreo\'er, we get substantially concurrent reports of them from all 

 those present at a given place at a given time. Beyond noting, per- 

 haps, that mirages are frequent in desert conditions, common sense 

 declines to deal with data on mirages: it sees no prospect whatever 

 of reducing them to determinist order. This lack of orderability, not 

 the possible presence of a "subjective" element, is what makes the 



