172 THE ROYAL SOCIETY OF CANADA 
two facts is established in the experimental test in the 26 cases; the 
second fact is itself a scientific principle in historical method, which is 
established in turn by the following facts:— 
In scientific treatment of an historical record the scholar must first 
determine whether such record is essentially trustworthy. This deter- 
mination is made according to the demonstrable presence or absence 
in the record of those qualities which, according to the universal experi- 
ence of human intercourse, are necessary in men and records in order 
to produce essential trustworthiness, and which, therefore, if present 
in men and records, necessarily produce such trustworthiness. Accord- 
ingly, in the scientific treatment of records, if a scholar accepts a record 
as essentially trustworthy, his acceptance is for necessary reasons; and 
this general decision being for necessary reasons, all exceptions to it 
must be for reasons at least equally necessary. By the contrary 
course history, as a science, would become bankrupt because if a scholar, 
having accepted a record as necessarily and essentially trustworthy, 
dare arbitrarily, 7. e., for an unnecessary reason, stamp a single state- 
ment in it as erroneous, he dare do the same in succession to every 
single statement in the record; and the net result at the conclusion of 
this process will be that the same record which is accepted as necessarily 
and essentially trustworthy is also arbitrarily stamped and rejected as 
wholly erroneous. 
The prevailing treatment of discrepancy proceeds from the assump- 
tion that discrepancy between two records is prima facie evidence of 
defect in one or other record, 1. e., that the discrepancy, if unexplained, 
establishes such defect; and in the prevailing treatment the test of such 
explanation is the possibility of reasonable harmonization. By the 
experimental test in the 26 cases it is established that this assumption 
and test are alike wrong. Discrepancy has a legitimate and necessary 
place in trustworthy records, and therefore its presence, whether 
explained or not, does not establish defect in either record; and har- 
monization is an illusive test of legitimacy, because on the one hand 
there is no guarantee that the most reasonable and probable harmoniza- 
tion will accord with the facts, and, on the other, a discrepancy which is 
legitimate and even necessary may seem quite impossible to recoacile. 
Discrepancies arising from the categories are legitimate, and only 
upon proven exclusion of such a legitimate origin can a discrepancy 
be held to involve necessarily a defect. Therefore, as a result of the 
principles and facts above enunciated, where discrepancy lies between 
records essentially trustworthy, defect is not to be imputed to either 
record except upon proof that the discrepancy could not conceivably 
have arisen legitimately from the categories. The difficulty of supply- 
ing this proof has been noticed. Not only, as mentioned at the con- 
