Surface volcanic gases may reflect volatile budgets in magma and forecast impending eruptions, and their release to the atmosphere may affect climate. The dynamics of magma degassing is complicated, however, by differences in the solubility, partitioning, and diffusion of the various volatiles, all of which can vary with pressure, temperature, and melt composition. To constrain possible gas outputs, we carried out experiments to determine how Cl partitions between water bubbles and silicate melt, and decompression experiments to examine how Cl behaves during closed- and open-system degassing. We incorporated our findings and those from the literature for CO 2 and S into a steady, isothermal, and homogeneous flow model to estimate fluxes of gases at the vent from ascending water-rich magma, assuming different scenarios for the onset and development of permeability in bubbly magma. We find that, for given permeability scenarios, total gas fluxes vary with magma flux, but ratios of gas species do not change. The S/Cl and SO 2 /CO 2 ratios do change, however, depending on whether the magma is oxidized or reduced. After magma fragments into a Plinian eruption column, gases continue to escape from cooling pumice in the plume, but here the rate of gas release is controlled by diffusion, which varies with temperature. Degassing of pumice and ash was modeled by linking a steady-state plume model, which gives the vertical variation of mean temperature and velocity of particles inside the plume, to a conductive cooling model of pumices, which controls diffusion of Cl, CO 2 , and S in pumice. We find that gas loss increases with column height (mass flux) and initial temperature, because in both cases pumices cool over a longer time period, allowing more gas to diffuse out of the matrix glass. The amount of gas released also depends on the size distribution of particles in the erupting mixture, with less being released for a finely skewed distribution.

We examined the textures, volatile contents, and cooling histories of microlite-defined flow bands in several rhyolitic obsidians in order to test whether textural variations between bands could be ascribed to different degassing and cooling histories, and to assess the timing and location of band formation. Flow bands are defined by variations in microlite number density ( N V ) and size. For each mineral phase examined, smaller average crystal sizes characterize the microlite-rich bands in contrast to microlite-poor bands, which contain relatively larger crystals of lower N V . Magmatic H 2 O concentrations of microlite-rich and microlite-poor bands show no statistical difference between the textures. Calorimetric measurements yield similar glass transition temperatures and cooling rates for adjacent bands. These observations suggest that microlite heterogeneities could not have developed during late stage cooling and degassing during flow emplacement, as such textural variations imply distinct cooling and/or degassing histories. Rather, textural heterogeneities must have formed during flow in the conduit. Hydrothermal annealing experiments were conducted on natural fragmented rhyolite in order to simulate the welding process in silicate melt and to provide first-order estimates of the time scales and deformation required to produce flow bands. Flow bands formed in experiments conducted at H 2 O-vapor pressures of 50 and 100 MPa, and for temperatures ranging from 800 to 850 °C. In each case, bands formed as a result of redistribution of oxide-rich domains and grain boundary coatings in annealed glass powders that underwent viscous deformation. Experiments suggest that bands may form on relatively short time scales (∼7 d) and for small bulk strains (∼1). Band formation may be promoted by high melt-H 2 O concentrations, shear stress, and viscous and frictional heating.