Historical accounts can be pitched at different levels and these will generally change at different rates. ‘Deeper’ strata, such as the histories of sea routes or crop rotation, move more slowly than the ‘surface’ histories of governments and wars, and this means that different kinds of methodological questions are asked. A concern with how to establish causal sequences or whether totalities can be defined from a nexus of relations gives way to questions over what type of strata should be isolated for study, and the periodisation that should be adopted (AK 4, 10). While the focus in history was moving towards patterns on a large scale, specific histories dealing with strands of culture and knowledge (e.g., the history of ideas, of science, or of literature) appeared to move in the opposite direction towards a concern with rupture and discontinuity. The figures Foucault mentions in outlining this second tendency are among those whose work is most clearly a point of reference for the analyses that follow: Gaston Bachelard, Georges Canguilhem, Michel Serres and Martial Guéroult.

Of these, Bachelard arguably made the most influential contribution through his understanding of science as an open and episodic invention of new realities that are not drawn from empirical experience. Although Foucault does not mention them in the Introduction, Bachelard's convictions that philosophy should learn lessons from the mathematical sciences, and that it should not impose on scientific thought a conceptual framework that science itself had left behind, were also both important for the notion of discourse and its analysis that Foucault introduces in this book, as was Bachelard's writing on temporal atomism, or the arithmetisation of time (these themes are discussed in the section on Bachelard above).