We have studied 40 earthquakes on seven major transforms along the Mid-Atlantic Ridge. The majority show the expected transform-parallel strike-slip motion on steeply dipping fault planes. Some ''anomalous'' mechanisms, suggesting either unusually shallow dip or a component of dip-slip motion, occur near ridge-transform intersections. The vertical tectonism suggested by transform morphology may thus be restricted to intersections where thermal contrast across the transform is greatest and stress orientations can be oblique to both ridge and transform. Centroid depths, determined by waveform inversion, are limited to a shallow zone corresponding approximately to the area above the 400¿C isotherm of the thermal structure expected from lithospheric cooling. The depth extent of faulting predicted by estimating fault area from seismic moments extends approximately to the predicted 600¿C isotherm. Assuming that the areas above the 400¿C to 600¿C isotherms fail seismically, moment release for six of the transforms accounts for varying fractions of the slip predicted by plate motion, suggesting that some aseismic slip occurs if the time period sampled is representative. In contrast, on the Vama Transform the inferred seismic slip is greater than predicted unless faulting extends to the 650¿C isotherm. Alternatively, the Vema may have an unusually long recurrence time or be mechanically anomalous due to its complex history. In contrast to transform events, centroids for oceanic intraplate earthquakes extend to approximately the 750¿C isotherm. This result is intriguing if depths reflect lithospheric strength as a function of depth. Transforms have much higher strain rates than intraplate areas, so if both had the same rheology the limiting isotherm for transforms should be higher. Alternatively, if a weaker rheology were appropriate, the higher strain rate should cancel the weakening and yield a limiting isotherm similar to that for intraplate regions.

Transforms may thus be either weaker and/or have higher temperatures than expected. In the latter model the required high temperatures, perhaps due to intrusion at depth predict measurable heat flow anomalies. The second possibility, unusual rheological weakness, is reminiscent of the unexplained weakness of the San Andreas fault, where heat flow data exclude amomolously high temperatures.