For many English speakers, A-bar movement of a possessor requires pied-piping of the containing DP. However, Davis (2021) shows that about half of speakers also permit possessor extraction. In such examples, long-distance wh-movement or topic/focus fronting separates the possessor from the Saxon genitive morpheme [’s] as in (1). Davis uses a variety of diagnostics to argue that this is truly possessor extraction: (1) English possessor extraction (Davis 2021, ex. 2. See that work for many analogous examples.) Mary is the author [CP who1 they said [CP [DP t1’s new book] is good]]. Based on a new study of 17 speakers, I show that such extraction has an un-noticed restriction. Though 14/17 speakers who accept examples like (1) also accepted extraction of a full DP by topic/focus fronting, all rejected extraction of possessive pronouns like "my". I argue that possessive pronouns are immobile because they are portmanteau morphemes which express a non-constituent unit—a possessive D and the possessor in its specifier—via morphological spanning (Bye & Svenonius 2012, Merchant 2015, Svenonius 2016, a.o.). Further, I argue that this finding entails that phase spell-out applies to entire phases (Fox & Pesetsky 2005, Ko 2014, a.o.). See attached PDF for full abstract.