416 TECTONICS/Seismic Structure At Mid-Ocean Ridges Figure 11 Average depth of magma lens reflections beneath ridges versus spreading rate Magma lenses lie within two dis tinct depth ranges of km for fast spreading ridges and 2.5 km for intermediate and slow spreading ridges Both shallow and deep lenses are observed at some intermediate spreading ridges The curved line shows the depth to the 1200 C isotherm calculated from the ridge thermal model of Phipps Morgan J and Chen YJ Data from different ridges are labelled: RR, Reykjanes Ridge; JdF, Juan de Fuca Ridge; GSC, Gala´pagos Spreading Centre; CRR, Costa Rica Rift; Lau, Lau Basin; NEPR; northern East Pacific Rise; SEPR, southern East Pacific Rise Figure 12 Thickness of the extrusive crust at the ridge axis versus spreading rate For data obtained from detailed reflection surveys, average thicknesses are shown by black dots with standard deviations where available (solid lines) or thickness ranges (dotted lines) Data derived from other seismic methods are shown by stars Data for the East Pacific Rise are labelled by survey location CRR, Costa Rica Rift; MAR, Mid Atlantic Ridge; JdF, Juan de Fuca Ridge; GSC, Gala´pagos Spreading Centre Figure 13 Crustal thickness versus spreading rate Crustal thick nesses are determined from seismic data obta ned away from fracture zones (Reproduced from Bown JW and White RS (1994) Variation with spreading rate of oceanic crustal thickness and geochemistry Earth and Planetary Science Letters 121: 435 449.) predicted if the density of the magma is equivalent to that of lavas erupted onto the seafloor (2700 kg m 3) Either the average density of magma is greater or mechanisms other than neutral buoyancy control magma-lens depth The alternative model hypothesizes that magmachamber depth is controlled by the thermal structure of the ridge axis In this model, a mechanical boundary, such as a freezing horizon or the brittle–ductile transition, prevents magma from rising to its level of neutral buoyancy The depth of this boundary within the crust will be primarily controlled by the thermal structure of the ridge axis, which is expected to vary with spreading rate The inverse relation between spreading rate and depth to low-velocity zones at ridges apparent in early seismic datasets provided compelling support for this hypothesis Numerical models of ridge thermal structure have been developed that predict systematic changes in the depth to the 1200 C isotherm (a proxy for basaltic melts) with spreading rate that match the first-order depth trends for magma lenses (Figure 11) This model predicts a minor increase in lens depth within the fastspreading-rate range and an abrupt transition to deeper lenses at intermediate spreading rates, consistent with the present dataset The numerical models also predict that, at intermediate spreading rates, small variations in magma supply to the ridge can give rise to large changes in axial thermal structure These models are supported by recent