The problem of vagueness is often conceived as the problem of generalizing a formal theory of meaning applicable only to precise languages to a formal theory of meaning applicable to vague languages too. A Procrustean method of generalization would be to make the vague language precise, then apply the original theory. It is open to several obvious objections. First, a vague language can be made precise in more than one way. Second, the ‘can’ is only in principle; in practice we cannot make our vague language fully precise in even one way. Third, if a vague language ismade precise, its expressionschange in meaning, so an accurate semantic description of the precise language is inaccurate as a description of the vague one. These objections seem at first sight to have little in common with each other. However, one line of thought suggests a revision of the Procrustean method that promises to answer them all.