The Archeology of Knowledge is Foucault's attempt, after the fact, to describe theoretically the method he used in his first three books of history (Madness and Civilization,The Birth of the Clinic, and The Order of Things). This is, then, not the presentation of a formal theory built logically from axioms, but a description of a specific kind of approach to history (a 'way of speaking' about history). Archeological analysis seeks to describe the history of discourse, the set of 'things said' in all its interrelations and transformations. These processes occur at a very specific level, which is neither the level of the events of history, nor the level of a teleological 'progress' of ideas, nor the level of an accumulation of formal knowledge, nor the level of the popular or unspoken 'spirit of the times.' The analysis of discourse abandons all preconceptions about historical unity or continuity, describing instead the processes of discourse in all their disruptions, thresholds, differences, and complex varieties.
Foucault begins with a polemic Introduction (Part I), noting recent shifts in historical method, relating these shifts to the newly uncertain status of the historical document, and critiquing histories that depend on loose notions of continuity as unhelpful and outdated. He says that these histories are also narcissistic, because what they really seek in forms of historical continuity is the assurance that history depends on the constant present of a transcendent human consciousness.
Part II, 'The Discursive Regularities,' asks what kinds of unities really do exist in the history of discourse. Foucault tries four hypotheses, in which unity is based on the object of discourse, the author(s) of discourse, the concepts used in discourse, or the theories and themes of discourse. Each hypothesized basis for discursive unity turns out to be something more complex than we thought it was, and each turns out not to be the single basis for unity, but one aspect of a discursive unity that can only be described in its variability and complexity. The four hypotheses do yield four specific levels at which discursive formations can be analyzed, however: the formation of objects of discourse, the formation of enunciative positions or modes, the formation of theoretical strategies, and the formation of concepts.
In Part III, 'The Statement and the Archive,' Foucault takes a step back from the level of discursive unities and attempts to describe the discursive field from its smallest elements to its most general totality. The smallest units are statements; although they have no single, stable unit (they change size according to their field of use), they form the most detailed level at which discourse can be analyzed. 'Statement' really refers more to a specific aspect of articulated language than it does to a unit of language. The statement is the level of the active, historical existence of a set of signs. The rest of Part II is devoted to maintaining the rigorous description of the statement as a positive, describable, specific aspect of history as Foucault moves up to the level of the archive, which is 'the general system of the formation and transformation of statements.'
Part IV addresses the difference between Foucault's archeological method and that of the history of ideas. For the four issues of originality, contradiction, comparison, and change, Foucault shows that his method replaces broad continuities and generalizations with specific, describable relations that preserve the differences and irregularities of discourse. The last chapter in this part, 'Science and Knowledge,' deals with the reasons that archeological analysis has focused on the history of the sciences, and with the details of how this focus is carried out. Foucault concludes with an intriguing, often poetic, dialogue between himself and a hypothetical critic of his method. In it, he defends archeology against charges that it is essentially structuralist and that it invests discourse with transcendence over other elements of history.