Impossible Persons, Daniel Harbour's comprehensive and groundbreaking formal theory of grammatical person, upends understanding of a universal and ubiquitous grammatical category. Breaking with much past work, Harbour establishes three core theses, one empirical, one theoretical, and one metatheoretical. Together, these redefine the data subsumed under the rubric of "person," simplify the feature inventory that a theory of person must posit, and restructure the metatheory in which feature theory as a whole resides. At its heart, Impossible Persons poses a simple question of the possible versus the actual: in how many ways could languages configure their person systems, in how many do they configure them, and what explains the size and shape of the shortfall? Harbour's empirical thesis -- that the primary object of study for persons are partitions, not syncretisms -- transforms a sea of data into a categorical problem of the attested and the absent. Positing, innovatively, that features denote actions, not predicates, he shows that two features alone generate all and only the attested systems. This apparently poor inventory yields rich explanatory dividends, covering the morphological composition of person, its interaction with number, its connection to space, and properties of its semantics and linearization. Moreover, the core properties of this approach are shared with Harbour's earlier work on number features. Jointly, these results establish an important metatheoretical corollary concerning the balance between richness of feature semantics and restrictiveness of feature inventories. This corollary holds deep implications for how linguists should approach feature theory in future.