A strong enough longshore impulse leaving the coast to the left (looking downwind) generates 'full' upwelling in which the thermocline comes to intersect the free surface. This problem is beyond the scope of linearized (small displacement), theory, but it may be treated as a simple geostropic adjustment problem in which the wind stress is supposed to be exerted impulsively on a two-layer fluid. A minimum impulse is found to be necessary for a full upwelling to develop from hydrostatic equilibrium, the mangitude of which is close to the product of top layer and propagation velocity of long waves on the thermocline. When the upwelling-favoring wind impulse is greater than the minimum, the upwelled front moves offshore by a distance proportional directly to the extra impulse (above the minimum) and inversely to top layer depth time Coriolis parameter. Upwelling episodes observed in Lake Ontario during the International Field Year for the Great Lakes (IFYGL) and feasibility studies before IFYGL show a frontal behavior in good quantitative agreement with the simple theoretical model. The offshore component of the impulsively exerted wind leaves the position of the front anaffected. However, a sustained offshore wind increases somewhat the offshore displacement due to a prior longshore impulse. The effect is relatively weak, and an offshore wind alone rarely produces full upwelling. A sustained longshore wind acting on an already upwelled thermocline generates considerable negative potential vorticity. Adjustment following such a 'second' longshore impulse produces a more complex thermocline shape and under certain conditions a frontal countercurrent. In a closed basin, full upwelling following a wind stress impulse occurs over only a portion of the shoreline. Given quiescent conditions, the upwelled front may be expected to propagate alongshore, somewhat as an internal Kelvin wave. Observations in October 1972 in Lake Ontario show frontal motions resembling internal Kelvin waves. However, only one half of the wave propagates, that half in which particle velocities have the same direction as the wave propagation velocity. Furthermore, the propagating warm zone is much narrower than the cold upwelled zone which developed under a succession of previous wind impulses.