[ ^^7 ] 



therefore it follows, that the evidence of teftimony can approach 

 indefinitely near to certainty ; and can at length exceed the evi- 

 dence of any inference, however cogent, which can poflibly be 

 deduced from perfonal experience, or from, perfonal and derived 

 experience conjointly. 



It is to be obferved, however, that the calculation here flated, ap- 

 plies only to the teftimony of different witnefles, who fimply give 

 their evidence as to the truth or falfliood of a propofed fadl ; or of 

 ■witneffes each of whom has an opportunity of knowing what 

 teftimony the others have given. This, without doubt, is to 

 take the force of concurrent teftimony at the greateft: difadvan- 

 tage ; neverthelefs, even in this cafe we find, that it has no limit. 

 But there are other cafes in which the leaft number of concur- 

 rent witnefles, let the degree of their veracity be however fmall, 

 can afford a probability which fhall exceed any given degree of 

 probability however great ; namely, where the witneffes have had 

 no means of knowing each others teftimony, and the fa£l is at- 

 tended with contingent circumflances, which make a part of 

 their depofition : becaufe the chances of their not concurring in 

 thefe circumftances, may exceed any given chance. In thefe 

 cafes we obferve, that even witnefles who have been obferved to 

 tell fallhood oftcner than truth, may yet produce belief; becaufe 

 here the probability of the truth of their report is not derived 

 from the chances of their coinciding, abftraQedly, in truth or 

 falfhood, but from the chances of their coinciding in circum- 

 ftances contingent in their nature, and which have no apparent 

 connection with each other. As for inftance, if each witnefs 

 Vol. VII. * P 3 fliould 



