248 SCIENTIFIC METHOD AS 



favour of the man who requires no conjectural authors or 

 other hypothetical entities. 



I have suffered in this lecture from a limitation which 

 has affected all who have taken part in this course the 

 necessity of forcing into a limited space matter that should 

 occupy a wide series of lectures. And in one respect it 

 seems to me that the discussion of historical evidence 

 must necessarily be somewhat disappointing as compared 

 with the other forms of scientific knowledge. The other 

 lecturers have shown how successfully their science has 

 proved something : I have been bound to dwell on the 

 extreme precariousness and limited range of certainty 

 belonging to history. The logical type of a proved his- 

 torical fact is not a clear-cut argument moving to an 

 inexorable conclusion, but a complicated exposition of a 

 great variety of inconclusive considerations, out of which 

 certainty emerges gradually by a process of accumulation. 

 For many facts recorded no unassailable proof is possible 

 at all ; but those facts which we may regard as proved, 

 from the variety of their evidence and the wide area of 

 their effect, do extend a kind of indirect certainty over 

 others, unprovable otherwise, which are linked with 

 them and depend upon them. And behind all this lies 

 the question of the value of human testimony. We know 

 it to be uncertain : any of our witnesses may be untruthful 

 or himself deceived ; and it is difficult to determine the 

 truth about them. No laws of probability will protect us 

 against involuntary error : we can only correct that by some 

 a posteriori critical process. But on the whole it is prob- 

 ably true that men do not, except for motives which are 

 more or less calculable and transparent, aim at deceiving 

 their fellows. And if you ask why we should believe any 

 statement which we cannot verify, it appears to me that 

 you must fall back on this fact which is in truth only 



