Convection of plasma within the terrestrial nightside plasma sheet contributes to the structure and, possibly, the dynamical evolution of the magnetotail. In order to characterize the steady state convection process, we have extended the finite tail width model of magnetotail plasma sheet convection. The model assumes uniform plasma sources and accounts for both the duskward gradient/curvature drift and the earthward E × B drift of ions in a two‐dimensional magnetic geometry. During periods of slow convection (i.e., when the cross‐tail electric potential energy is small relative to the source plasma's thermal energy), there is a significant net duskward displacement of the pressure‐bearing ions. The electrons are assumed to be cold, and we argue that this assumption is appropriate for plasma sheet parameters. We generalize solutions previously obtained along the midnight meridian to describe the variation of the plasma pressure and number density across the width of the tail. For a uniform deep‐tail source of particles, the plasma pressure and number density are unrealistically low along the near‐tail dawn flank. We therefore add a secondary source of plasma originating from the dawnside low‐latitude boundary layer (LLBL). The dual plasma sources contribute to the plasma pressure and number density throughout the magnetic equatorial plane. Model results indicate that the LLBL may be a significant source of near‐tail central plasma sheet plasma during periods of weak convection. The model predicts a cross‐tail pressure gradient from dawn to dusk in the near magnetotail. We suggest that the plasma pressure gradient is balanced in part by an oppositely directed magnetic pressure gradient for which there is observational evidence. Finally, the pressure to number density ratio is used to define the plasma “temperature.” We stress that such quantities as temperature and polytropic index must be interpreted with care as they lose their nominal physical significance in regions where the two‐source plasmas intermix appreciably and the distributions become non‐Maxwellian.
