All explosive eruptions are driven by expansion of the volatiles released from the magma on decompression. Gas expansion on bodies without atmospheres is extreme. One consequence is that initial strong mechanical and nearly isothermal coupling between gas and pyroclasts in a choked flow gives way to weak and eventually negligible coupling as the pressure decreases through a series of shocks and eventually enters the Knudsen regime. A second consequence is the extreme fragmentation of the magma into small pyroclasts, and by analogy with the contrast between the sizes of basaltic pyroclasts on the Moon (~30 to ~400 mm) and Earth (tens of mm to at least tens of cm), we infer that the range of pyroclast sizes in terrestrial rhyolitic eruptions (bulk of particles typically in the range 1 mm to a few tens of mm) should map into a pyroclast size range of ~1 mm to 1 mm for rhyolitic melts on the Moon. We previously simulated explosive basaltic eruptions on the Moon and we have adapted out earlier work to apply to rhyolitic eruptions by changing the assumed gas composition from CO to H2O, since this is the commonest volatile in high-silica melts, and including larger amounts of the gas component and hence greater maximum pyroclast ranges. We calculated pyroclast ranges as a function of pre-eruption magma water content, n, given in both ppm and wt.%. Rmono is the implied maximum range of pyroclasts when all of the clasts have the median 300 mm size. Rcoarse and Rfine are the maximum ranges of the coarse and fine fractions when the bimodal size distribution is assumed, as seems applicable here. In a previous analysis by others, CBVC explosion products were treated as a single phase in which all pyroclasts acquire the same speed as the gas, that speed being the result of the conversion of all of the initial internal energy of the (assumed) single-phase fluid to kinetic energy. These assumptions ignore many important components of the processes involved (as outlined above), and result in underestimates of the ejection speeds and dispersal values for equivalent initial temperatures and gas phase compositions. If the 33-44 km radius is the coarse component radius, this implies that the magmatic water content, n, was ~2.5-3.5 wt.%. Similarly the ~300 km maximum extent of the outer pyroclastic deposit implies n = ~2 wt.%. Taken together these results imply an H2O content of ~2-2.5 wt.%. Using solubility data for water in rhyolite of, a magma reservoir top would have to be at a depth of at least ~7-10 km in the lunar crust to retain this amount of water. It has been estimated that the CBVC extrusives contain ~0.55 wt.% residual water, making the pre-eruption total 2.5-3 wt.%. This would require the magma top to be at depths of at least 10-15 km. These depths are very much minimum values, and are entirely consistent with the suggestion that these magma reservoirs were located at the density trap at the base of the crust.