feudal domain what agents corresponded to each of our ministers of state—as Boutaric did in his study of the administration of Alphonse of Poitiers.
This method of drawing up lists of questions which bases all historical construction on an a priori procedure, would be objectionable if history really were a science of observation; and perhaps some will think it compares very unfavourably with the a posteriori methods of the natural sciences. But its justification is simple: it is the only method which it is possible to employ, and the only method which, as a matter of fact, ever has been employed. The moment an historian attempts to put in order the facts contained in documents, he constructs out of the knowledge he has (or thinks he has) of human affairs a scheme of arrangement which is the equivalent of a list of questions—unless, perhaps, he adopts a scheme which one of his predecessors has constructed in a similar manner. But when this work has been performed unconsciously, the scheme of arrangement remains incomplete and confused. Thus it is not a case of deciding whether to work with or without an a priori set of questions—we must work with such a set in any case—the choice merely lies between the unconscious use of an incomplete and confused set of questions and the conscious use of a precise and complete set.
VI. We can now sketch the plan of historical construction in a way which will determine the series of synthetic operations necessary to raise the edifice.
The critical analysis of the documents has supplied the materials—historical facts still in a state of dispersion. We begin by imagining these facts on the model of what we suppose to be