Soames's account is based on the old idea that borderline cases involve truthvalue gaps. He says rather little in justification of this idea, merely that it seems arbitrary where to locate the cut-off point for a vague term and that it is unclear what semantic mechanism could determine that point. The book shows no awareness of work on the epistemicist alternative that borderline cases involve irremediable ignorance of truth-values. Since I have defended that alternative at length elsewhere, I will not labour the point here. Surprisingly, Soames combines his anti-epistemicist commitments with the following two claims (216-17):