246 SCIENTIFIC METHOD AS 



to the fact that this is precisely suited to the treatment of 

 a body of facts, all of which are unique in space and time. 

 These facts, as we have seen, are outside our own experi- 

 ence : they cannot be presented as the result of a deductive 

 process : and the only way in which we can assure ourselves 

 that such and such an event occurred at a certain place 

 and time is by accepting the testimony of some one who 

 knows. Human testimony is uncertain : in recent years 

 we have had a controversy as to the exact mode in which 

 Charles I was beheaded ; this year there is an animated 

 discussion as to the details of Nelson's tactics at Trafalgar. 

 We can never be absolutely sure of our testimony, but it 

 is the best thing we have. 



If it be contended that this is a sceptical view of history, 

 calculated to destroy much of our confidence in it, I would 

 venture to urge that this contention arises in consequence 

 of the prevalence of a false ideal of proof in history. In 

 the mathematical and physical sciences proof is complete 

 and final ; the conclusions are free from ambiguity or from 

 doubt. But the possibility of such proof depends abso- 

 lutely on the conditions in which these sciences work : 

 their ideal belongs to them in virtue of the character of 

 the facts they investigate and the abstract method pur- 

 sued : it cannot be transferred to any other situation. 

 The physical sciences deal with aspects of physical events 

 which are continually increasing in the degree of their 

 abstractness : the confused and detailed fact of daily 

 experience is further and further from the purview of 

 physical science ; but history, as we have 'Seen, has to do 

 with the ordinary fact of daily experience, and cannot 

 even abstract it from the relations of time and space. 

 Historical facts are susceptible of a proof that is suited 

 to them, and no more ; and the proof will always look 

 painfully precarious and inadequate to one whose expect- 



