has a certain extent; if the number of collaborators is too great and the part of each too limited, the liberty and the responsibility of each are diminished or disappear.
Histories, intended to give a narrative of events which happened but once, and to state the general facts which dominate the whole course of special evolutions, still have a reason for existence, even after the multiplication of methodical manuals. But scientific methods of exposition have been introduced into them, as into monographs and manuals, and that by imitation. The reform has consisted, in every case, in the renunciation of literary ornaments and of statements without proof. Grote produced the first model of a "history" thus defined. At the same time certain forms which once had a vogue have now fallen into disuse: this is the case with the "Universal Histories" with continuous narrative, which were so much liked, for different reasons, in the Middle Ages and in the eighteenth century; in the present century Schlosser and Weber in Germany, Cantù in Italy, have produced the last specimens of them. This type has been abandoned for historical reasons, because we have ceased to regard humanity as a whole, bound together by a single evolution; and for practical reasons, because we have recognised the impossibility of collecting so overwhelming a mass of facts in a single work. The Universal Histories which are still published in collaboration (the Oncken collection is the best type of them), are, like the great manuals, composed of independent sections, each treated by a different author; they are publishers' combinations. Historians have in our days been led to adopt the division by states (national histories) and by epochs.
B. There is in theory no