Bathymetry and gravity data obtained during a detailed Hydrosweep survey of the southern East Pacific Rise from 7¿S to 9¿S are used to investigate isostasy at the axis of a fast spreading mid-ocean ridge. In particular, we examine the manner in which the bathymetric crestal high is supported and how this support varies along the axis within the 160-km-long 7¿12'S-8¿38'S ridge segment. The crestal high stands about 400 m above the adjacent ridge flanks and has a nearly constant minimum axial depth for a distance of 140 km. The summit is broad and flat, and an axial summit caldera is present for the entire length of the ridge segment. However, the width and cross-sectional shape of the crestal high vary systematically along the axis. It is broad with gentle slopes in the center of the segment but becomes progressively narrower and steeper toward the ends of the ridge segment. The ridge crest is marked by a free-air gravity anomaly high about 15--20 km wide with an amplitude of 10--15 mGal relative to the ridge flanks. Mantle Bouguer anomalies vary systematically along the axis with minimum values found near the center of the segment.

The axial mantle Bouguer anomalies thus do not reflect the axial depth but are correlated with changes in the cross-sectional area of the crestal bathymetric high. The effects of cooling and subsidence away from the axis were removed from the bathymetry and free-air anomalies to isolate residual topographic and gravity anomalies associated with the ridge crest. The residual crestal bathymetric high was modeled as a flexural feature resulting from the upward buoyant load of a region of low density material centered beneath the axis. The lithosphere was treated as a broken plate, either with a constant flexural rigidity or an effective elastic thickness Te which grows at a rate proportional to the square root of distance from the axis. The best fitting values of Te for the constant rigidity case are in the range of 0.3--0.6 km. For --the growing plate model, Te increases at a rate of 0.2--0.3 km1/2. The gravity constrains the mass deficiency to extend to a depth of 20--30 km for both lithospheric models.

We interpret this low-density material as a region of partial melt feeding magma to the ridge axis. The best fitting density anomalies imply that a 4--9% melt fraction is present beneath the crestal high. Upwelling of melt to the axis is thus confined to a narrow zone within about 10 km of the axis. The mass deficiency and thus the upwelling partial melt are not distributed evenly along the ridge axis but rather are concentrated in the central portion of the ridge segment. It thus appears that differences between the along-axis gravity and depth patterns observed at slow spreading and at fast spreading ridges are not the result of a change from three-dimensional, focused upwelling at slow spreading ridges to two-dimensional sheetlike upwelling at fast-spreading ridges. Rather, the differences in axial gravity and depth between fast and slow spreading ridges reflect differences in the efficiency of the shallow along-axis magma distribution. ¿ American Geophysical Union 1993