VI. In every study of successive facts it is necessary to provide a number of halting-places, to distinguish beginnings and ends, in order that chronological divisions may be made in the enormous mass of facts. These divisions are _periods_; the use of them is as old as history. We need them, not only in general history, but in the special branches of history as well, whenever we study an extent of time long enough for an evolution to be sensible. It is by means of events that we fix their limits. In the special branches of history, after having decided what changes of habits are to be considered as reaching deepest, we adopt them as marking _dates_ in the evolution; we then inquire what event produced them. The event which led to the formation or the change of a habit becomes the beginning or the end of a period. Sometimes these boundary events are of the same species as the facts whose evolution we are studying--literary facts in the history of literature, political facts in political history. But more often they belong to a different species, and the special historian is obliged to borrow them from general history. In general history the periods should be divided according to the evolution of several species of phenomena; we look for events which mark an epoch simultaneously in several branches (the Invasion of the Barbarians, the Reformation, the French Revolution). We may thus construct periods which are common to several branches of evolution, whose beginning and whose end are each marked by a single event. It is thus that the traditional division of universal history into periods has been effected. The sub-periods are obtained by the same process, by taking for limits events which have produced consequences of secondary importance. The periods which are thus constructed according to the events are of unequal duration. We must not be troubled by this want of symmetry; a period ought not to be a fixed number of years, but the time occupied by a distinct phase of evolution. Now, evolution is not a regular movement; sometimes a long series of years passes without notable change, then come moments of rapid transformation. On this difference Saint-Simon has founded a distinction between _organic_ periods (of slow change) and _critical_ periods (of rapid change). CHAPTER III CONSTRUCTIVE REASONING I. The historical facts supplied by documents are never enough to fill all the blanks in such schemes of classification and arrangement as we have been considering. There are many questions to which no direct answer is given by the documents; many features are lacking without which the complete picture of the various states of society, of evolutions and events, cannot be given. We are irresistibly impelled to endeavour to fill up these gaps.