27] METHOD OF INVESTIGATION 27 



number of editions, which has only very relative significance. 

 A single new edition of a large and costly work may sometimes 

 signify as much as many editions in other cases. Fortunately 

 it is almost always easy to determine the true significance of 

 the different tests, for the simple reason that they serve as 

 checks to each other. 1 



Odin's foregoing criteria apply only in part to the first 

 four categories of men of letters. With respect to patrons, 

 librarians, actors and orators he used other tests. 



The speeches of orators are indeed frequently printed; but 

 it is well known that the reception which the public reserves 

 for the printed address does not always correspond to the 

 popularity of the orator. . . . For the other three categories 

 even this test is lacking. I have therefore been obliged in all 

 these cases to restrict myself to the tests which were only sub- 

 sidiary for the other groups. As for orators and actors, I 

 have considered first of all the impression which they have 

 produced on contemporaries, as it has been reported by wit- 

 nesses worthy of trust. Here I hardly risk deceiving myself, 

 for it is at least as easy to judge impartially of the success 

 attained by an orator or actor as to appreciate exactly the 

 popularity which a written work has enjoyed. I have had 

 more difficulty in deciding which of the " patrons " and 

 " librarians " had a right to appear on the list. For these two 

 groups information was not always as abundant and explicit 

 as could be desired. Thus it may be that I have erroneously 

 omitted more than one person who was really important. 

 Nevertheless, I have reason to suppose that the number of 

 these omissions cannot be considerable. 2 



Following Odin's method as closely as possible, the in- 

 vestigator began his compilation of a list of American men 

 of letters. Five encyclopedias of biography and literature 

 were carefully studied. These were : 



1 A. Odin, op. cit., p. 365. * Ibid., p. 366. 



