[bowman] fundamental PROCESSES IN HISTORICAL SCIENCE 547 



In testing the actual necessity or correctness of his conclusions, 

 there is even more need of this "honesty with one's self" in the his- 

 torian than in a detective, or in investigators in other fields of science. 

 If a detective seeks to impose an unnecessary, i.e., merely probable, 

 conclusion on others, an opposing attorney, and a judge, and jury are 

 at hand to check him. Or again, if an investigator, e.g., in the mathe- 

 matical and physical sciences, includes in his published results an 

 unnecessary conclusion, the material and opportunity to test the 

 point will always remain. But in many cases an historical investi- 

 gation, once made, will not be repeated, or repeated only with reduced 

 material and with poorer opportunities. Correction by others being 

 thus uncertain, the self-discipline of the historian should be the greater. 

 He must be detective, and the friendly and the opposing attorney, 

 and judge and jury in one — the trustee of scientific truth. If his 

 investigation is thorough, he is in a better position than any other 

 historian is now, or may ever be, to separate the necessary conclusions 

 from the unnecessary, the scientific result from the unscientific, in 

 respect of the points investigated. He ought not to mix with his 

 scientific results the unscientific in the shape of ingenious hypotheses 

 and conclusions of a seductive probability, and then, from his vantage- 

 ground, impose the whole on readers and others less favorably situa- 

 ted. All his advantage and all his ingenuity should be used to stand 

 between them and this imposition. In this efifort reasoned probability 

 has an important part, as a negative test, both in the primary con- 

 clusions directly drawn from records and in the secondary conclusions 

 indirectly inferred from the primary. 



Reasoned probability in primary conclusions from records. The 

 exemplification or non-exemplification of the requisites of trustworthi- 

 ness is not necessarily uniform even in the same record. Ordinarily 

 records are not wholly trustworthy or wholly untrustworthy, but in the 

 trustworthy there are untrustworthy portions, passages or points, 

 and in the untrustworthy there are also trustworthy portions, passages 

 or points. No record as a whole, or in any such portion, passage or 

 point should be accepted as trustworthy unless, so far as accepted, the 

 exemplification of the five requisites is beyond reasonable doubt, 

 i.e., unless reasoned probability, employed as a negative test, cannot 

 suggest a defect in this exemplification as being reasonably probable. 

 A professor of church history, who is as able a scholar as any that 

 this country has produced, once remarked to the author seriously 

 that he almost despaired of any truth in history when he read in a 

 church paper, at the end of the obituary of a very aged clergyman, 

 that the deceased would be "greatly regretted by his fellow clergy;" 

 the fact was he had long been so increasingly troublesome in their 



