232 APPENDICES 



not satisfied that he knows its properties until he has examined the 

 element in isolation and worked out its formula. We cannot do 

 this with 'horse'; instead we perform the same process mentally, 

 by abstracting the type or species from the particular horses 

 examined, thus constructing an 'image' of an 'ideal' horse, such 

 as is frequently found pictured in dictionaries. ^ 



Thus, Morris R. Cohen makes the concept o^ relations central to 

 scientific method: 'This search for laws of nature as invariant rela- 

 tions rather than statistical correlations is at the basis of the faith 

 that where our readings persistently show variations which cannot 

 be explained as within the "error" of our instruments, the varia- 

 tions must be due to the fact that what is measured is not homo- 

 geneous. '4 'But modern mathematical logic has taught us to avoid 

 the old form of the issue between nominalism and (the older) 

 realism by recognizing the relational character of unity.'' ^ 



What are some of the dangers and virtues of such abstraction 

 of 'invariant relations' in science? Something is inevitably lost in 

 the process. We no longer have individuals, with all their specific 

 determinations: we have left behind Man O'War, the iceman's 

 dray horse, the gangling colt, and the wild Arabian steed. 'For in 

 isolating abstract elements we are apt to forget how they function 

 in the actual totality from which they are abstracted,' ^ This is the 

 gist of Bergson's critique of analysis and abstraction in meta- 

 physics: '. . . the great error . . . lies in believing that while 

 remaining on the same level we can find behind the word a thing. 

 Such has been the error of those philosophers who have not been 

 able to resign themselves to being only psychologists in psychology, 

 Taine and Stuart Mill, for example. Psychologists in the method 

 they apply, they have remained metaphysicians in the object they 

 set before themselves. They desire an intuition, and by a strange 

 inconsistency they seek this intuition in analysis which is the very 

 negation of it.' ^ John Dewey's 'Instrumentalism' tends to link the 

 concrete with the practical and criticize 'false' abstractions as 

 lacking pragmatic, verifiable consequences.^ 



Further, the concepts we use in abstracting may be arbitrarily 

 chosen and overlook the very clue needed for further progress 

 in knowledge. Mill puts the requirements as follows: *. . . our 

 general conceptions must be "clear" and "appropriate" to the 

 matter in hand' 9; and he analyses these criteria in terms of the 

 correspondence of concepts to a real 'agreement' among the 

 phenomena. 10 The importance of clarity has been stressed by 



