The free air gravity anomaly and depth are sampled at 2-km intervals along two long, reasonably straight ship tracks across the Atlantic Ocean. The resulting series are then processed as if they were time series, and filters are obtained to perdict the gravity observations from the bathymetry. More than half the energy in the gravity field can be predicted by this means, and that which cannot emphasized unusual structures beneath the sea floor. More information can be obtained by comparing the gravity and the bathymetry after both series have been Fourier-transformed. Isostatic compensation begins when the wavelength exceeds about 100 km and increases with increasing wavelength. The results are compared with predictions from various simple models and agree best with a model in which the topography results from variations of crustal thickness and the plate thickness is little greater than 10 km when the compensation occurs. These observations can be understood if the topography results from large-scale intrusions into the lower crust within tens of kilometers of the spreading center. Though such a model for a slowly spreading ridge differs from most of those which have been previously put forward and must be regarded with skepticism until it is supported by evidence from other sources, it appears to be compatible with the limited information now available.