But there is more to the failure of these organizing tactics than simply the undermining of unity by the forces of dissonance or transformation. Foucault is actually arguing that unities of discourse are constituted by such differences; when we look, with Foucault's method, for the factors that group a set of statements under a single discourse, what we find is a certain singular space in which these statements relate in a wide variety of ways. Our task then, in our pursuit of individualized discourses, must be to detail these relations rather than look for commonalities shared by all of the statements. Each relation can only be a matter of difference, of the relative size and nature of the gap that separates one statement from another. Thus, we arrive at the puzzling paradox that Foucault points to, in which the search for discursive unities can only proceed through the study of the differences that define the discursive relation.
This is why Foucault calls this chapter 'Discursive Formations,' in contrast to the preceding chapter's 'Unities of Discourse.' With groups of statements individualized by the nature and degree of their differential relations to each other (and to other groups of statements), we are really no longer talking about coherent 'groups' at all, but more properly of 'formations,' a term that articulates an identifiable regularity of relations without unjustly emphasizing relations of similarity over those of difference. The regularities that make a given discursive formation identifiable Foucault calls 'rules of formation.' Again, this is a term meant to encompass a broad range of principles of relation: rules of formation are themselves of many kinds (existence, coexistence, maintenance, modification, and disappearance).
Foucault recognizes that the positing of discursive formations and the rules of these formations does not make a concrete statement, and he does not even promise that discursive formations will take up the slack where false unities of discourse have been rejected. The project remains 'an as yet uncharted land and an unforeseeable conclusion.' We have suspended received notions of grand historical continuity, and now we have even erased the more modest signposts of disciplinary divisions. It remains to be seen whether Foucault's method will reconstitute them.