ABSTRACT
The history of ideas usually credits the discourse that it analyses with coherence. If it happens to notice an irregularity in the use of words, several incompatible propositions, a set of meanings that do not adjust to one another, concepts that cannot be systematized together, then it regards it as its duty to find, at a deeper level, a principle of cohesion that organizes the discourse and restores to it its hidden unity. This law of coherence is a heuristic rule, a procedural obligation, almost a moral constraint of research: not to multiply contradictions uselessly; not to be taken in by small differences; not to give too much weight to changes, disavowals, returns to the past, and polemics; not to suppose that men's discourse is perpetually undermined from within by the contradiction of their desires, the influences that they have been subjected to, or the conditions in which they live; but to admit that if they speak, and if they speak among themselves, it is rather to overcome these contradictions, and to find the point from which they will be able to be mastered. But this same coherence is also the result of research: it defines the terminal unities that complete the analysis; it discovers the internal organization of a text, the form of development of an individual œuvre, or the meeting-place of different discourses. In order to reconstitute it, it must first be presupposed, and one will only be sure of finding it if one has pursued it far enough and for long enough. It167appears as an optimum: the greatest possible number of contradictions resolved by the simplest means.
