Michael Dummett famously asked how the serpent of inconsistency entered Frege's paradise. He himself blamed the impredicative nature of second-order quantification, while many others focused on the inflationary nature of the axiom. Axiom V is, after all, the denial of a higher-order generalization of Cantor's theorem. Predicativists do not deny this, but they block the derivation of the relevant generalization in predicative fragments of second-order logic. Unfortunately, there is more than one higher-order generalization of Cantor's theorem, and one of them remains a theorem in predicative fragments of higher-order logic. Our recommendation to predicativists is to respond that only one of them supports the cardinality gloss we associate with Cantor's theorem and that it is, in fact, false. The other remains a theorem of predicative fragments of higher-order logic but its derivability seems more closely related to the Grelling's paradox than to cardinality considerations.