Muscovite granite (13.8% muscovite, 4.8% normative corundum) was reacted, with varying percentages of H2O, in cold-seal vessels at 2 kbar and in piston-cylinder apparatus between 10 and 35 kbar. The diagrams illustrating melting/crystallization relationships are: P-T sections with both excess H2O and with no H2O added (0.66% H2O in rock); T-XH2O sections at 15 kbar and 25 kbar showing H2O-undersaturated conditions: the H2O-undersaturated surface for the crystallization of quartz/coesite (small amounts of aluminosilicate minerals persist to higher temperatures). Glass compositions measured by electron microprobe from samples with 5% H2O at 15 kbar confirm that liquids are syenitic through at last 100 ¿C above the solidus, as predicted from the effect of pressure on the Residua System. Results are explained successfully by phase relationships involving muscovite, quartz, and orthoclase in K2O-Al2O3-SiO2-H2O, with reactions depicted in a Ptotal-PeH2O -T model, with special reference to the divariant surfaces in the region PeH2O <Ptotal. With reduced PeH2O (or aH2O), produced either by small amounts of H2O (and H2O-undersaturation) or by CO2+H2O mixtures, subsolidus dehydration reaction temperatures decrease, vapor-present solidus temperatures increase, and muscovite stability in presence of liquid increases. In general, muscovite, biotite, and amphibole can be precipitated from magmas containing only a few tenths per cent H2O (although the H2O-undersaturated liquids coexisting with crystals may contain 3% or more dissolved H2O). This particular granite cannot be a primary magma from mantle or subducted oceanic crust. It is a possible product of partial fusion of pelitic rocks between about 20 km and 40 km depth given sufficient H2O, and xenocrystal muscovite or sillimanite from the source rocks. The phase relationships are consistent with the idea of S-type granites, but not sufficient to prove the origin of this rock. Additional tests require phase relationships of other associated granite rocks, and details of geochemistry, geophysics, and field relationships.