This article takes its starting-point from the intuition that, in some sense, the extant Aristotelian corpus was written in paragraphs. I hasten to explain. First – almost needless to say this explicitly – I do not suggest that the Aristotelian corpus was set out in the indented form we are familiar with from modern editions. Second, and more to the point, I do not have anything to say about the question, whether the text was formally set out divided by the paragraphē mark (a horizontal line used to signal divisions in ancient papyrus rolls). The paragraphē was certainly in use in Aristotle's time – he himself refers to it in Rhet. 1409a20 – although how widespread its use was, we do not know. Nor do we know with any certainty how the ancients themselves conceptualised this mark and its significance. It does seem, however, to have corresponded more to our concept of a full stop than to our concept of a paragraph. Be that as it may, my argument is not about how the text was set out on the space of the page, but about how the text was conceived: in the most general sense, how it was written and read. The Aristotelian corpus may well have been originally, in appearance, just a constant stream of undivided columns on papyri. Yet it had a certain discursive structure which, I suggest, is best approached as that of a paragraph. It is upon this discursive structure of Aristotle's works that I focus in this article.