Phillips (1966) considered a linear, barotropic, nondivergent model driven by time‐dependent winds in an attempt to explain theAriesobservations. He found that the model greatly underestimated the response as compared to the observations. The model is reexamined by using newer estimates of the wind stress (Bunker, 1975) and of the eddy kinetic energies (Freeland et al., 1975). With these the model predicts reasonably well the time and space scales and the amplitude of the barotropic mesoscale variability south of Bermuda. It is suggested that the low‐frequency temperature fluctuations observed in thePanulirusdata near Bermuda result from the advection of the mean baroclinic field past Bermuda by the barotropic fluctuati