Taking seriously the arguments of Earman, Roberts and Smith that ceteris paribus laws have no semantics and cannot be tested, I suggest that ceteris paribus claims have a kind of formal pragmatics, and that at least some of them can be verified or refuted