Measured heat flow at Cajon Pass is consistent with predictions based on local site conditions and regional heat flow. With observations now ranging to a depth of 31/2 km, there is still no evidence for significant frictional heating anywhere on the San Andreas fault. The result supports the view, long suggested from heat flow studies, that the fault is weak in spite of estimates based on Byerlee's law, isotropic strength, and hydrostatic fluid pressure that suggest a strength several times larger. Recent evidence (Zoback et al., 1987; Mount and Supper, 1987) that the maximum principal stress might be almost normal to the San Andreas fault would support the weak-fault model and add constraints over and above those imposed by heat flow; e.g., local friction coefficients μ≲0.1 or fluid pressures along the fault greater than lithostatic (λ≥1), compared to μ≲0.2 or fluid pressure greater than twice hydrostatic (λ≥0.74) for the heat flow constraint alone. These constraints are a challenge to existing models of faulting, and they are stimulating promising new points of view. The balance of plate boundary forces around a weak fault depends on the basal traction coupling the seismic layer to the rest of the system; heat flow limits the coupling force across the fault to an insignificant ~1011 N/m. The weak fault also precludes significant near-field basal driving tractions, but it permits a large basal drag force which could result in a highly stressed seismic layer offering appreciable resistance to plate motion through its base.

Such tractions could develop progressively if the fault surface weakens as it evolves; if they exist, they should cause an observable reduction in shear stress resolved in the fault direction and a rotation of principal axes as the fault is approached; if they do not exist, the seismic layer rides passively on the lower crust. Heat flow measurements should detect whether such basal tractions might be associated with basal decoupling and flow. Coupling at the base of the seismic layer is controlled by the rheological profile, the usual representation of which raises three questions in applications to the San Andreas fault zone. First, the linear frictional portion through the seismic layer implies a resisting force on the fault much greater than the heat flow limit permits. Second, the large stresses implied for the temperature-sensitive ductile layer might be unsustainable; they could lead to shear heating and weakening at plate boundary strain rates. Third, in the ductile layer the stress is sensitive to whether deformation is concentrated in narrow vertical mylonite zones, as sometimes assumed in models of the earthquake cycle, or more broadly distributed by bulk flow in a deep-crustal ''asthenosphere.'' Horizontal basal shear stresses are of the same order as vertical strike-slip stresses near the base of the seismic layer, they could result in bulk flow or horizontal detachment leading to a different pattern of long-term stress, strain rate, and dissipation and a requirement for decoupling and basal drag on the seismic layer in the near field. Results from the San Andreas fault taken with long-standing speculation about the orthogonal relation between oceanic transform faults and extensional spreading centers suggest that strike-slip transform faults might be anomalously weak in both continental and oceanic settings. ¿ American Geophysical Union 1992