Ultraviolet HST observations of the Galactic halo star HD 93521 with the GHRS echelle have been analyzed to give interstellar column densities of Fe, S, Si, Mn, and Mg in their dominant ionization state, and also of C+ in the excited J = 3/2 fine-structure level and of S++. The terrestrial absorption line at 1305 angstrom, absorbed by O0 atoms in the excited J = 1 fine-structure level, was used to determine the FWHM of the instrumental profile (1.05 diodes) and to calibrate the wavelength scale precisely. Most interstellar lines show nine clearly resolved components, in general agreement with earlier ground-based work by Albert; the LSR velocities found here range from -65 to +9 km s-1. Least-squares solutions for each component of each species give column densities N(X(i)), centroid velocities v, and values of b, the two-dimensional velocity dispersion. Measures at 21 cm toward this star by Danly et al. give N(HO) for each component, as well as v and b. Ratios of column densities for the dominant species (H0, Fe+, S+, etc.) give an approximate determination of the logarithmic depletion D(X) in the warm gas of mostly neutral H which produces each component. N(S+)/N(H0) shows no dependence on v and gives D(S) = 0.0 dex. Since N(S+) is more accurately determined than N(H0), we determine D(X) for other elements from their abundance relative to S+. All these elements show a marked decrease of Absolute value of D as Absolute value of v increases to 20 km s-1. Thus, D(Fe) = - 1.20 dex for Absolute value of v less-than-or-equal-to 10 km s-1 and averages -0.62 dex for Absolute value of v greater-than-or-equal-to 35 km s-1. Variations of D between components are highly correlated between elements, with numbers of atoms of each species removed from grains in nearly constant ratios one to another as Absolute value of D decreases. In particular, about two Fe atoms leave the grains for every one Si atom, as though these atoms were mostly bound in Fe2SiO4 molecules within the grain cores; in these warm clouds the grain mantles are apparently missing. Other atoms show somewhat similar behavior. Column densities of C+ in the excited state yield average values of the local electron density, n(e), ranging from 0.11 cm-3 at Absolute value of v < 20 km s-1 to 0.04 cm-3 at Absolute value of v > 35 km s-1. Analysis of the Ca+ K line, observed toward HD 93521 by Albert, gives about the same n(e) values in ionization equilibrium, if we assume D(Ca) = D(Fe) in warm clouds. This tends to confirm that electrons, rather than H0 atoms, excite C+. The average of these n(e) values agrees with that found by Reynolds for several lines of sight through the halo. This agreement suggests that halo electrons toward HD 93521 are clumped in the same concentrations, or ''clouds,'' as is the warm H I gas, with the clumping factor f equal to 0.25, the mean of Reynolds's values. The partial ionization of H and S+ in these clumps can probably not be produced by ultraviolet photons, in view of the appreciable H0 column densities (some 10(19) cm-2) but might result from successive shocks, from X-rays, or from energetic particles, perhaps accelerated near these moving gas clumps. Sputtering or evaporation of grains (reducing Absolute value of D) might also be caused by such effects. The line of sight to HD 93521 is unusual in the large number of high-Absolute value of v clouds intersected; data from other halo stars may indicate which of the present results have some generality.
