It has been argued by Bernard Linsky and Edward Zalta, and independently by Timothy Williamson, that the best quantified modal logic is one that validates both the Barcan Formula and its converse. This requires that domains be fixed across all possible worlds. All objects exist necessarily; some – those we would usually consider contingent – are concrete at some worlds and non-concrete (but still existent) at others. Linsky and Zalta refer to such objects as ‘contingently non-concrete’. I defend the standard usage of the word ‘exists’, and the view that many objects exist only contingently. I argue that the Linsky/Zalta analysis, and to a lesser extent Williamson’s, suffers not only from a peculiar ontology but also from two related formal difficulties. Their analysis gives either counter-intuitive or ad hoc results about essences, and it fails to accommodate contingently existing abstracta.