South Island, New Zealand is composed of a magmatic arc of early Mesozoic age. The Fiordland portion of the belt exposes high-pressure migmatitic intermediate to mafic granulites (14-16 kbar) in addition to layered mafic to intermediate intrusions all which have formed in the lowermost to middle crust during the early Cretaceous. The northern Westland area exposes the middle to upper crustal portion of the same arc (Klepeis et al., in press). The magmatic units are sodic and have high Sr/Y ratios ("adakitic") and have been suggested to originate from partial melting of lowermost mafic crust (Muir, 1998). We are currently testing this hypothesis by performing fluid-absent partial melting experiments on lowermost crustal samples collected from the base of this arc. We report preliminary results from these experiments on an alkalic, meta-basaltic dike that has been interpreted to represent underplate material emplaced during arc generation. The meta-basaltic assemblage is hbl + pl + czo with minor biotite. Experiments were performed at 14 kbar and 825^oC to 975^oC. The solidus is likely just below 825^oC due to breakdown of biotite. Hbl + pl +/- czo reacting to form melt + gnt is the dominant melting reaction. Granitic to granodioritic melt compositions are observed as a function of increasing temperature. Preliminary LA-ICPMS results of granodioritic glass from the 975^oC sample show HREE depleted, LREE enriched compositions, with high Sr/Y ratios. Experimental results from partial melting of lowermost crustal meta-diorite suggest that partial melting of material that already has a signature of a basaltic source will carry that signature through to daughter melts. Melt volumes are low. Direct partial melting of basaltic underplate, in addition to that of dioritic lowermost crust, may provide a source for the observed middle to upper crustal magmas. These results also bear directly on current magmatism observed in a narrow oceanic trench offshore of southern Fiordland. Solander Island is composed of young (<2 Ma) andesitic volcanics, also adakitic in nature (Anderson et al., 1993). Future work will use these experimental results to help determine the source of these young volcanics and constrain the potential range of thermal regimes for the subduction initiation process.