Abstract

An attractive principle about domains of quantification is the analogue of the Separation Axiom in set theory: restricting a domain by an arbitrary predicate yields a domain. In particular, restricting a domain by a predicate that applies to nothing yields a domain. Thus if there is a nonempty domain, there is an empty domain. But semantics for the empty domain involves some neglected subtleties. Untangling them requires us to revise the usual definition of truth in a model, avoiding the detour through Tarski's notion of satisfaction.