In “The Uses of Argument” Stephen Toulmin unleashes a fierce attack against formal logic. For his attack to work, modal qualifiers like “necessarily”, “possibly” and “probably” when occurring in natural language arguments do not mean what formal logicians take them to mean but have another semantics. Toulmin gives a pragmatic account of this semantics in which what these modal qualifiers mean is equivalent to what they are used to do. I will defend Toulmin’s account of “probably” as an account of the pragmatics of “probably”, but uphold Searle’s objection that taking this as an account of their semantics commits a speech act fallacy. Thus (although a full defense will not be essayed here) I claim that Toulmin’s attack on formal logic fails