regard to the whole of which his own work is to form a part. Thus no historian can feel perfectly safe in adopting the results of another's work, as may be done in the established sciences, for he does not know whether these results have been obtained by trustworthy methods. The most scrupulous go so far as to admit nothing until they have done the work on the documents over again for themselves. This was the attitude adopted by Fustel de Coulanges. It is barely possible to satisfy this exacting standard in the case of little-known periods, the documents relating to which are confined to a few volumes; and yet some have gone so far as to maintain the dogma that no historian should ever work at second hand. This, indeed, is what an historian is compelled to do when the documents are too numerous for him to be able to read them all; but he does not say so, to avoid scandal.
It would be better to acknowledge the truth frankly. So complex a science as history, where facts must ordinarily be accumulated by the million before it is possible to formulate conclusions, cannot be built up on this principle of continually beginning afresh. Historical construction is not work that can be done with documents, any more than history can be "written from manuscripts," and for the same reason—the shortness of time. In order that science may advance it is necessary to combine the results of thousands of detail-researches.
But how are we to proceed in view of the fact that most researches have been conducted upon methods which, if not defective, are at least open to suspicion? Universal confidence would lead to error as