Discursive practices involve systems that allow statements to emerge as 'events' and to be used or ignored as 'things.' Foucault proposes to call these systems of statements, collectively, the 'archive.' Thus, the archive is not just a collection of texts that define a culture, nor even a set of institutions that preserve texts. The archive is 'the law of what can be said' and the law of how what is said is transformed, used, preserved, etc. Thus, the archive is defined as 'the general system of the formulation and transformation of statements.'
Our own, contemporary archive is impossible to describe clearly because it is the very thing that gives what we say its mode of emergence and existence. It is 'that which, outside ourselves, delimits us.' The archive thus becomes more sharply defined the further back we go chronologically, as we are increasingly separated 'from what we can no longer say.' In this apprehension-through- distance, the analysis of the archive shows us that our own identity as discursive beings is defined throughout by differences. 'Difference is this dispersion that we are and make.' The distance manifested in the archive is what justifies Foucault's naming of his project as an 'archeology.'
In these final chapters of Foucault's description of 'The Statement and the Archive,' he turns from the description of statements and discursive formations in and of themselves to the description of these elements as historical material. Now that the analysis of statements has been distinguished from other ways of analyzing language, and now that the analysis of statements has been linked to the analysis of discursive formations (which is set out in Part I), Foucault moves to give us a clearer picture of how these analyses delimit a specific approach to the historical archive.
The description of statements, in their existence solely on the level of the enunciative function, sought to show that Foucault's analysis, even on the micro-level of 'things said,' dealt only with the least obscure, least speculative, most 'positive' aspect of historical documents. In fact, this level of existence, in which statements interact with other statements, is such a basic, given level of existence that it constitutes a historical a priori, something upon which other aspects of language (like grammar or even reference) rest. In the last chapter, Foucault brought this positivity of statements back into play with discursive formations, which are composed of statements as they relate to each other in clearly describable ways.
With this work done, Foucault can now consider the large-scale form of the historical archive whose elements he has so rigorously delimited. If we are to describe statements as 'events' with specific conditions of emergence and as 'things' with specific relations and transformations, how are we to understand the archive? According to Foucault's method, it can no longer be a mere collection of printed material, hanging around in libraries solely by virtue of its material inertia. Nor can it be thought of, on this method, as a record of cultural meaning or a transcription of a set of individual consciousnesses. Any of these descriptions would violate the attempt to describe historical statements solely in and of themselves, at the level of the statement; such descriptions stray beyond the level of the historical a priori.
Foucault's answer is to consider three ways in which the historical archive is usually understood, and to replace each of these descriptions with one that is more positive (i.e., less dependent on hidden continuities or abstractions). First, the archive can no longer be seen as an 'infinite transparency' in which a historical consciousness or 'spirit of the times' can be seen through all of the various statements. Statements, again, are the level at which Foucault's analysis must proceed; they can never be analyzed in the interest of something on a different level. Foucault therefore rejects the model of a transparent archive, in which a 'plethora' of spoken and unspoken statements reveal a single historical 'totality' beneath them. His method will deal with an archive of 'rarity,' in which the historian analyzes the statement in its conditions of uniqueness. Each statement is 'rare,' because only that statement can occupy its specific position in a discursive formation.