The Cortlandt complex, near Peekskill, New York, is formed of rocks that range from peridotite and pyroxenite through norite to diorite. G. S. Rogers mapped 14 species of these rocks and added syenite and sodalite syenite to the list. Robert Balk made a structural survey of the complex and showed it to be a funnel-shaped pluton within which three smaller funnels are developed. If these funnels mark channels through which magma ascended, the various species ought to be grouped more or less symmetrically about the funnels; yet Rogers’ petrographic map shows little correspondence with Balk’s structural map. The explanation lies in the indefinite character of Rogers’ species, which grade into one another in every respect. Progress in understanding the complex requires the abandonment of “species petrology” in favor of “phase petrology” or the mapping of critical mineral phases. The most instructive phase is hornblende, which occurs in two forms, one primary and the other a poikilitic replacement of plagioclase and pyroxene. The first is confined to the marginal zone; the second is found all through the central region, extending far beyond the boundary of Rogers’ “hornblende-norite.” Balk’s central funnel falls entirely within this enlarged hornblende field. The marginal zone of the pluton is characterized by primary hornblende, by the frequent appearance of biotite in large poikilitic plates, and by the occasional development of poikilitic garnets, orthoclase, and micropegmatite. The occurrence of sodalite syenite is not confirmed.