《地震的力学》

After the 1906 San Francisco earthquake, the commission under Andrew Lawson catalogued the damage and traced the fresh fault scar that ran for hundreds of kilometres along the San Andreas. Writing the second volume, Reid asked the deeper question: not how badly the ground shook, but what physical process produced the shock at all. His answer came less from the ruins than from the surveyor's transit.

The U.S. Coast and Geodetic Survey had fixed the positions of triangulation stations across the fault region in the decades before 1906 — in epochs roughly around 1851–1865 and 1874–1892 — and re-measured them just after the earthquake. Differencing the epochs, Reid found that points far from the fault on its two sides had drifted past each other by several metres over the preceding half-century, gradually, and in the very same right-lateral sense as the sudden offset, while points at the fault trace had scarcely moved. In the earthquake itself the ground along the fault jumped sideways — in places by as much as about 6.4 metres (some 21 feet near the head of Tomales Bay) — as if catching up to a motion the deeper crust had been making all along.

From this Reid reasoned out his picture. The crust on the two sides of a locked fault is slowly sheared by the larger movements of the Earth, but the fault is stuck; so the rock around it bends elastically and stores strain energy, like a steel spring being wound. When the stress at last exceeds the strength of the fault, the rock ruptures and springs back toward an unstrained shape — the elastic rebound — and the stored energy escapes as the seismic waves we feel. The displacement of an earthquake, he argued, is therefore not created at the instant of the shock: it has been accumulating quietly for decades or centuries, and the rupture merely lets it go.

Two consequences follow. First, earthquakes on a fault should recur, as the strain re-accumulates after each release — the seismic cycle. Second, because it is the rock that stores the strain and not the fault surface, the slow build-up could in principle be watched by repeated surveys. Reid hoped this might one day make earthquakes predictable; a century on, that hope is only partly fulfilled — instruments now reveal which faults are loaded, but not the day they will fail.

The full Volume II — 192 pages with the triangulation tables, fault maps and Reid's analysis — is at the source below; Volume I (the Lawson Report) holds the field descriptions of the 1906 rupture.