The phenomenon of rigid designation - in particular, de jure rigidity - is typically treated metaphysically. The picture is: reference is gained in a way that puts no constraints on what an object in other worlds, or counterfactual situations must be like, in order to be the referent of that term, other than 'being this thing'. This allows 'pure metaphysical' investigation into, and discovery of 'the nature' of the referent. It is argued that this presupposes a 'privileged' ontology, of a (...) sort that is far from obvious. An alternative picture of rigidity is sketched, terms have associated criteria of identity and criteria of application for counterfactual situations. A term is rigid when these coincide. Aside from being a better account of rigidity, it also undermines the picture of rigid terms allowing us to 'metaphysically' investigate the nature of the referent: matters of identity and essence for the referent are settled by these semantic constraints (though it can be done in a way that allows are a posteriori necessities). (shrink)