186 INTRODUCTION TO PHILOSOPHY OF SCIENCE 



terms: The method of construction permits one to attribute to 

 the symbols such content as they must have by virtue of the data 

 from which they have been derived, and the method of hypothesis 

 permits one to attribute to the symbols such additional content 

 as they may be presumed to have if they are to be successful 

 instruments of prediction. This indicates clearly that the 

 former is essentially a method of description and the latter is 

 essentially a method of explanation; in the former one "re- 

 mains close to the data," in the latter he "leaps into the 

 unknown"; in the former one merely elaborates, in the latter 

 he creates; in the former one is simply making more clear 

 what he has already observed, in the latter he is anticipating 

 what he has not yet observed. 



A very important consequence of this distinction of meth- 

 ods is that a pure construction can never explain. This may 

 appear at first sight to be quite contradicted by the facts. 

 It seems obvious that science is continually explaining in 

 terms of abstractions, series, and complexes. But a moment's 

 reflection will convince one that in so far as these entities do 

 explain they cannot be pure constructions. For a pure con- 

 struction must have only such content as one is entitled to 

 give it from the data; but an explanatory entity — if it is to 

 explain — must contain more content than the data which 

 it is to explain. Explanation consists, as will be seen in the 

 next chapter, in the transfer of properties and relations 

 from the explanatory entity to that which is to be explained ; 

 in explanation one increases the content of the data by add- 

 ing to them the features of the hypothesis. But if a construct 

 has been obtained by direct derivation from the data, there 

 is clearly no possibility of turning about and explaining the 

 data in terms of the construct. The construct does not con- 

 tain that increment of novelty without which explanation is 

 impossible. A pure construct used as an explanatory entity 

 produces an ad hoc or verbal explanation. One cannot use the 

 soporific quality of opium to explain the fact that it produces 

 sleep unless he knows more about the soporific quality than 

 the fact of its producing sleep. In the same way, one cannot 



