Sir

Over the past few decades a modern theory of earthquake physics has been developed that is solidly based upon the laws of rock friction1. Ironically, over that same period there has emerged a view that, alone among crustal faults, California's San Andreas fault grossly violates those laws, while still generating earthquakes indistinguishable from those on other crustal faults. This conclusion is based on the absence of a heat-flow anomaly adjacent to the San Andreas fault that would be expected from a conductive model of frictional heating — a conclusion that requires the friction on the fault to be exceedingly low compared to that of any plausible geological material.

In testing this hypothesis with stress data2, I found much of these data ambiguous, but throughout Southern California the data unambiguously indicated that the San Andreas fault is not weak relative to the surrounding crust. This model-independent conclusion is in strong conflict with the heat-flow interpretation. Mark Zoback, in commenting on my results in News and Views3, reiterated the heat-flow interpretation as the strongest argument for a low-strength fault, which indeed it is.

As Sherlock Holmes noted about the dog barking in the night-time, the curious thing about the conductive heat-flow anomaly is its absence. There is, however, a broad heat-flow anomaly associated with the San Andreas fault4, large enough to account for frictional heating on a strong fault. Heat flow in this anomaly shows very high spatial scatter, suggesting, together with its breadth, a heat-transport mechanism other than conduction.

Townend and Zoback5 show that crustal scale permeability, owing to the presence of active faults, is high enough to ensure that pore pressures are ubiquitously hydrostatic in the crust. Such high fracture permeability would strongly favour advection by fluid transport along fractures, which can transport large heat fluxes without generating massive hot-spring activity6. It seems there are cracks in the heat-flow edifice.