Experiments have been done which simulate the modal metasomatism of spinel Iherzolite by partial melts of the subducted slab. The experiments were designed so that the metasomatizing melts were generated during the experiments by partial melting of a slab analog (basaltic composition amphibolite). The melts are thought to be representative of hybridizing melts in that they are derived by high-pressure partial melting under conditions appropriate to a hot slab geotherm. During the experiments, the melts infiltrate into and metasomatize a model depleted peridotite. Chemical modifications to minerals in the peridotite are of the same nature and extent as those found in naturally metasomatized spinel lherzolites. Modal metasomatism produced pargasitic amphiboles in runs at 1.5 GPa and in all but the highest temperature run at 2.0 GPa. The amphiboles are indistinguishable from amphiboles found in amphibole-bearing peridotites from supra-subduction zone environments. Systematic variations in amphibole composition suggest that the melt infiltration process in the experiments involved continuous modification of the composition of the infiltrating melt as observed around inferred quenched melt (i.e., amphibolite or amphibolite/clinopyroxenite) veins in xenoliths and massif peridotites. The compositions of the initial and final mineral phases in the experiments and those of the metasomatizing melts are used to derive amphibole formation reactions at 1.5 and 2.0 GPa that are similar in form to those inferred in studies of natural amphibole-bearing peridotites. The metasomatism reactions show that the extent of amphibole formation in peridotite at 1.5 and 2.0 GPa will, in general, be limited by clinopyroxene and spinel abundance.