A theoretical rationale is given for the simulated abyssal circulation in the North Pacific detailed in Part I. A simple theory clarifies the importance of the vertical change in area of the horizontal cross section of the basin owing to the existence of bottom topography, or basin shape. With general upward vertical transport in the basin, a water column horizontally diverges and vertically shrinks as it rises in spite of the general upwelling, resulting in the tendency to produce an anticyclonic circulation. This hypsometric effect is detailed by Rhines and MacCready. In the present case, this effect is present in the lower deep layer (3250–4250 m) and in the bottom layer (below 4250 m) of the North Pacific. In the lower deep layer, a single anticyclonic circulation owing to this effect appears explicitly. In the bottom layer, however, this effect almost balances the cyclogenesis owing to the bottom water influx, resulting in essentially eastward interior flow. On the other hand, an anticyclonic gyre in the upper deep layer (1750–3250 m) is maintained by the vertical shrinking of the water column owing to the outflux from the North to the South Pacific, resulting in a reversed pattern of the Stommel-Arons circulation.