A correlation between the crossite component (NaM4) in Ca-amphibole and pressure of metamorphism has long been recognized (Shido & Miyashiro, 1959), but only recently has the reaction been identified which buffers this aspect of amphibole composition (Brown, 1974): Ca-amphibole+iron oxide+albite+chloriteI+H2O (±stilp, qtz) = crossite+epidote (±muscovite, qtz). The exact stoichiometry of the reaction depends on compositional variables in the minerals, especially Fe2+/Mg and Fe3+/Al. Ca-amphibole should have fixed NaM4, at any given T and P, where it coexists with iron oxide, albite, and chlorite. Comparison of Ca-amphibole composition with mineral assemblage, in rocks from Otago, N.Z., and elsewhere, supports this hypothesis. In any terrane NaM4 is nearly constant at a particular metamorphic grade where amphibole exists in the buffering assemblage, but varies widely outside of this assemblage. Variations in Fe2+/Mg and Fe3+/Al in the amphibole have relatively little effect on NaM4, but in high pressure amphiboles NaM4 varies inversely with Aliv. Ca-amphiboles from high pressure areas have substantially more NaM4 (Otago, 0.6 of 2.0) than those from lower pressure areas (Sierra contact aureoles, 0.1). These relations suggest that in the buffering assemblage, the NaM4 content of Ca-amphibole should be a useful relative barometer for low to medium grade metamorphic rocks.