Figure F6. A. Schematic diagram drawn after Barnouin-Jha et al. (1997) showing results for the upper 50 km in a dynamic model of buoyancy-driven three-dimensional (3-D) mantle flow beneath a slow-spreading ridge. Red = flow vectors in the horizontal plane, yellow = flow vectors in the vertical ridge-axis plane, blue = flow vectors in the vertical ridge-normal plane. This illustrates along-axis flow in the shallow mantle from segment centers to segment ends. Note spacing between upwelling centers is ~400 km and the region of melt generation is almost as long as the ridge segments. B. From Ceuleneer (1991), illustrating ductile flow vectors and shear sense inferred from peridotite fabrics in the mantle section of the Maqsad area, Oman ophiolite. Map area is ~17 km long x 14 km wide. Approximate location of inferred paleoridge axis is shown as a red line. C. Schematic diagram from Jousselin et al. (1998) showing their vision of mantle flow, loosely based on observations from the Oman ophiolite, with a narrow zone of upwelling and a thin region of corner flow feeding a ridge segment that is three times longer than the diameter of the mantle upwelling zone. This model requires extensive subhorizontal ridge-parallel flow of residual mantle peridotite from the segment center to the segment ends. Although this geometry seems somewhat extreme and has not been produced in any 3-D dynamic model to date, it illustrates the type of highly focused solid upwelling that could produce the observed along-axis variation in crustal thickness on the Mid-Atlantic Ridge via 3-D focusing of mantle flow. Dynamic models such as that illustrated in A do not have sufficiently narrow zones of mantle upwelling and cannot reproduce the lengths of observed magmatic segments (~30–100 km). MOHO = Mohorovicic seismic discontinuity.