Aristotle considers all examples, as far as they are used as rhetorical arguments, to be inductions . On the other hand, he finds it worthwhile to distinguish different kinds of examples . Moreover, comparisons and fables are said to be “made” by the orator, whereas historical examples are not . The present article attempts to explain this opposition.A close examination of what can be meant by “induction” reveals that this term applies differently to historical examples and to comparisons and fables. (...) In a broader sense, all examples can be called inductions, insofar as a certain logical relation, different from the syllogistic one, obtains between conclusion and premises. This relation, however, is unable to explain why examples can serve as arguments. Applied to historical examples only, the term “induction” preserves its original meaning of checking samples in a number sufficient to yield generalization. In its narrow sense, “induction” gives a satisfactory account of how historical examples work. As for comparisons and fables — to which “induction” applies in a broader sense only — Aristotle fails to explain what makes them arguments. The explanation, missing in Rhetoric 2, 20, can, however, be given according to the general lines of Aristotelian dialectics and rhetorics. The main task of comparison and fable is to elaborate a so far unconsidered middle term meeting the following conditions: the hearer is prepared to agree that M applies to the subject of the conclusion and that the predicate of the conclusion applies to M. All examples make use of a general proposition from which their conclusion can be inferred. Historical examples produce instances, in order to confirm a proposition previously admitted as suitable premise, provided that its truth can be proved. Comparison and fables, by introducing a new middle term, try to form a premise susceptible of being admitted without proof. (shrink)