Aristotle, in the Rhetoric, appears to claim both that emotion-arousal has no place in the essential core of rhetorical expertise and that it has an extremely important place as one of three technical kinds of proof. This paper offers an account of how this apparent contradiction can be resolved. The resolution stems from a new understanding of what Rhetoric I. I refers to - not emotions, but set-piece rhetorical devices aimed at manipulating emotions, which do not depend on the facts (...) of the case in which they are deployed. This understanding is supported by showing how it fits with evidence for how rhetoric was actually taught in the fifth and fourth centuries BCE, in particular by Thrasymachus and Gorgias. The proposed interpretation fits well with Aristotle's overall view of the nature of rhetoric, the structure of rhetorical speeches, and what is and is not relevant to the pragma, the issue of the case at hand. (shrink)