JOHN WESLEY POWEXL DAVIS 



guistic families. Yet his usual method of writing, especially 

 in his later years, consisted in the synthetic exposition of large 

 problems, without the citation of sources or the mention of 

 particular instances, but with abundant imagery and seemingly 

 overabundant reiteration. It may be well believed that pres- 

 sure of work was in large measure responsible for these pe- 

 culiarities of composition. 'He was ever ready to draw off a 

 generous flood from his great reservoir of knowledge, but he 

 had no time to trace the flood back to its spring of supply.' 



Powell's liking for generalization had been early shown in 

 his classification of valleys and in his treatment of the broad 

 principle of the baselevel of erosion; but in these two prob- 

 lems he was dealing with inorganic factors, which behave in 

 the same way the world over. In ethnological problems, on 

 the other hand, no one continent affords a sufficient base for 

 all-embracing conclusions of the kind that one frequently meets 

 in Powell's essays; and hence a reader who did not look far- 

 ther than the printed page might infer that the conclusions 

 there stated were sometimes broader than their foundation. 

 Such a mis judgment would, however, only show that Powell's 

 synthetic style of presentation did not reflect his habitual 

 method of investigation. Most of his essays give no direct 

 indication of the extended observation and the abundant read- 

 ing on which their conclusions rest, for Powell was a profound 

 believer in the scientific method of investigation, which regards 

 the observation of visible facts as the essential first step in the 

 approach to theoretical inferences as to invisible facts, and 

 which' finds in frequent return to observation the only means 

 of verifying the correctness of the theoretical inferences. He 

 had a great confidence in the results thus gained and accepted 

 their guidance wherever they led. 



A reader of the synthetic essays may sometimes feel not 

 only that their author does not adduce a sufficient number of 

 facts for the support of the generalizations to which he rapidly 

 rises, but that he not infrequently passes over from induction 

 of generalizations to deduction of consequences from them 

 Without giving sufficient notice of his passage. For this reason 

 his essays do not necessarily carry conviction to one who is 

 uninformed of the "patient research by the rigorous methods 



67 



