34 SCIENCE (and common sense) 



Evaluating my data, I usually find them "consistent" internally— 

 perhaps also in comparison with the data of some other observer(s) 

 of the same or similar events. In any large array of data, however, 

 I am likely to find one or a few that fail of consistency, either inter- 

 nally or externally. If, after long eflFort, I am unable to fit them into 

 the class of the others, I create a new class for them— most often I 

 dismiss them as "errors" or "aberrations." The historian, too, selects 

 his data under the guidance of considerations of consistency. And 

 his interest in his data, and his ability to work with them, are both 

 product of his insight that— however apparently unique and irre- 

 producible the events concerned— yet are they recurrent events, i.e., 

 manifestations of "human nature," "economic determinism," "social 

 stratification," "copyists' error," and so forth. In such classificatory 

 judgments the historian guides himself as does the scientist. Either 

 may err, and then the scientist's error may be the more readily rec- 

 tifiable: often he can get a more abundant supply of data on "events 

 of the same kind." But the palaeontologist is a scientist who generally 

 cannot, and Schliemann a historian who could and did. 



The scientist finds the unique absolutely intractable, and thus re- 

 jects as data any findings he cannot envision as members of a recur- 

 rent class. The historian finds himself unable "to make anything of" 

 events wholly unlike any others known to him; he, too, shuns the 

 unique. With all due respect to Russell, modern historians rarely do 

 treat of miracles. To the extent that they do so they dispel the 

 uniqueness of miracles by attaching them to a class (e.g., instances 

 of divine intervention in the aflFairs of men, suspensions of natural 

 law, suspensions of human reason). But most nontheological his- 

 torians are uneasy in writing of miracles because it is not plain that 

 they are recurrent; because their proper classification is uncertain; 

 because often there is but one historical account, and that of imper- 

 fect internal consistency, or several accounts with mutual consistency 

 not wholly beyond reproach. The scientist's rejection of miracles as 

 data fit for scientific study is founded on similar considerations— 

 and they are the same considerations he would use routinely as, say, 

 meteor-observer. 



Uniqueness and identity. We seek general knowledge, general 

 order, general predictability. This quest, frustrated in studies of the 

 wholly unique, can prosper only as we find "sameness." We are then 

 clearly well advised to insist that acceptable as subject matter are 



