The viscosity of the Gangotri Himalayan leucogranite has been experimentally determined between 800¿ and 1100 ¿C, 300 and 800 MPa, for meltwater contents of 3.98 and 6.66 wt %. The melt viscosity is independent of pressure and shows an Arrhenian behavior relative to temperature within the range of conditions investigated. We present an empirical relation that can be used to determine leucogranite magma viscosities knowing their meltwater content and temperature. This relation together with phase equilibria experiments constrain the viscosity of the Himalayan leucogranites to be around 104.5 Pa s during their emplacement. These viscosities and the widths of dikes belonging to the feeder system are consistent with the theoretical relationship relating these two parameters and show that the precursor magma of the leucogranite was at near liquidus conditions when emplaced within host rocks with preintrusion temperatures around 350 ¿C. Calculated terminal ascent rates for the magma in the dikes are around 1 m/s. Magma chamber assembly time is, on this basis, estimated to be less than 100 years (for a volume of 150 km3). In addition, the dynamical regime of the magma flow in the dikes was essentially laminar, thus allowing preservation of any chemical heterogeneity acquired in the source. These results constrain the viscosity of melts formed during the first steps of crustal anatexis, those involving muscovite breakdown, to be also around 104.5 Pa s. Thus compaction may not be the only mechanism of melt segregation in partially melted crustal rocks in view of the very short timescale inferred for magma ascent and emplacement. ¿ American Geophysical Union 1996