Polarity items (PIs) license the ellipsis of their non-polarity counterparts and vice versa: if ellipsis is subject to a syntactic identity constraint, then we must conclude that pairs like some any are syntactically identical in the relevant sense. One technical way of cashing this out is given, and further consequences are discussed. * Many thanks to the organizers of the Diagnosing Syntax workshop at Utrecht and Leiden, Lisa Cheng and Norbert Corver, as well as to the audiences there for excellent feedback. For comments on the ms, thanks to Anastasia Giannakidou, Marcel den Dikken, and Hedde Zeijlstra, and the two reviewers (not necessarily a disjoint set) for their excellent suggestions and questions; both reviewers rightly point out that the present note seems to be merely the tip of the iceberg of this line of reasoning.