Recently revised leveling data and a more complete set of gravity data show that the uplift accompanying the later part of the Matsushiro swarm in central Honshu can be explained as due to dilatant expansion of the source region during the formation of a new fault. The local dilatancy occurred between the second and third peaks of seismic activity, although some preswarm dilatancy may have taken place in the region. It is likely that water inflow kept the medium saturated during expansion, except possibly for the times of most rapid uplift, when the drop in seismic activity indicates dilatancy hardening. A reasonable amount of horizontal expansion can explain the observed rate of change of gravity during uplift. The gravity measurements carry uncertainties of the same order as the observed changes, but the ensemble of data indicates clear trends during uplift and subsidence. The volume of land subsidence following the peak uplift agrees very closely with the best estimates of anomalous outflow of warm mineral water from springs during the last part of the swarm and supports the conclusion that subsidence was due to closing of water-filled cracks. The local rate of change of gravity during subsidence is very high, 1 mgal/m, and can only partly be accounted for by density increase due to crack closure.