Jonathan Salisbury

It is claimed to be a crucial advantage of supervaluationism over other theories of vagueness that it avoids any commitment to sharp boundaries. This thesis will challenge that claim and argue that almost all forms of supervaluationism are committed to infinitely sharp boundaries and that some of these boundaries are interesting enough to be problematic. I shall argue that only iterated supervaluationism can avoid any commitment to sharp boundaries, but on the other hand that is the model that Terrance Horgan has recently argued is a form of transvaluationism and thus logically incoherent. I shall first argue that infinitely higher-order vagueness gives rise to an infinite number of boundaries. I will then argue that an infinite number of these boundaries are, in the case of the vague term ‘tall’, located over a finite range of heights. I will argue that because of this, these boundaries must be infinitely sharp. I shall argue that on every plausible non-iterated supervaluationist model, some such boundary will mark a sharp boundary between heights that would make someone ‘more tall than not tall’ and heights that would not. Finally I shall argue that this is the sort of interesting sharp boundary supervaluationism must not admit.