An answer is suggested to the question of why plasma and magnetic energy accumulate in the Earth's magnetotail to be released in sporadic events, namely substorms. It is shown that the idea of steady convection is inconsistent with the idea of slow, approximately lossless, plasma convection in a long, closed-field-line region that extends into a long magnetotail, such as occurs during Earthward convection in the Earth's plasma sheet. This inconsistency is argued generally and demonstrated specifically using several quantitative models of the Earth's magnetospheric magnetic field. These results suggest that plasma-sheet convection is necessarily time dependent. If flux tubes are to convect adiabatically Earthward, the confining magnetic pressure in the tail lobes must increase with time, and the magnetotail must evolve into a more stretched configuration. Eventually, the magnetosphere must find some way to release plasma from inner-plasma-sheet flux tubes. This suggests an obvious role for the magnetospheric substorm in the convection process. To probe this process further, a two-dimensional, self-consistent, quasi-static convection model has been developed. This model self-consistently includes a dipole field and can reasonably account for the effects of inner-magnetospheric shielding. Starting with a variety of initial configurations, forcing plasma-sheet flux tubes to convect Earthward results in stretching of inner-plasma-sheet flux tubes, the development of a local minimum in the normal magnetic field component in the near-Earth plasma sheet, and an increasing lobe magnetic field. This behavior occurs generally, independent of the specific magnetopause or far-tail boundary conditions. These results suggest that the magnetospheric substorm is the inevitable outcome of Earthward plasma-sheet convection, in order to release plasma from inner-plasma-sheet flux tubes and the magnetic energy in the tail lobes associated with this plasma's confinement. Also, this inner-plasma-sheet behavior indicates why substorm onset should occur in the near-Earth plasma sheet. Some recent ideas concerning why sudden compressions and northward turnings of the IMF sometimes trigger, and other times do not trigger, magnetospheric substorms are discussed in light of these results.