A city levelled at dawn
The claim is that earthquakes on a fault like the San Andreas are not random violence but the release of strain that has been accumulating quietly for years. Harry Fielding Reid, examining the 1906 San Francisco earthquake, argued that the two sides of a locked fault keep moving relative to one another even while the fault itself does not slip, bending the surrounding rock elastically. When the accumulated strain finally exceeds what the rock can bear, the fault ruptures, the rock snaps back toward its original shape, and the stored energy is released as ground shaking. This is the elastic rebound theory, and it reframed earthquakes as the endpoint of a slow, measurable build-up rather than an unexplained sudden event.
Reid’s survey data
Reid’s evidence came from geodetic surveys of the ground around the San Andreas Fault made at different times over roughly the fifty years before 1906, which showed about 3.2 metres of accumulated bending near the fault line. By comparing the shape of the land before and after the earthquake, he could see that the surface had been deforming steadily in the decades leading up to the rupture, and that the earthquake itself had undone much of that deformation in a matter of seconds. The 1906 event was investigated formally by the State Earthquake Investigation Commission, chaired by Andrew Lawson of the University of California, Berkeley, whose 1908 report documented the rupture’s extent and the fault responsible.
The elastic rebound model
The Lawson Commission traced the rupture along a long stretch of the San Andreas Fault and found that the ground either side had moved past itself by a significant distance, consistent with Reid’s picture of built-up strain being released in one motion. The commission also established that shaking intensity depended heavily on the ground beneath a given site, with sediment-filled valleys shaking far harder than nearby bedrock, a finding that became a basis for later seismic-zoning practice. Reid’s core proposal, that active but locked faults store elastic strain and release it in sudden rupture, has since been supported by GPS measurements of present-day fault motion, which show the same pattern of gradual strain accumulation punctuated by abrupt slip.
The Lawson Commission
What the theory does not do is say when a given fault will fail. Elastic rebound describes the mechanism by which strain accumulates and is released, but it does not by itself predict the timing, size or location of the next rupture on a given segment of a fault. It also does not explain, on its own, why some faults or fault segments creep steadily and release strain gradually while others remain locked for long periods and then rupture catastrophically. These questions about the conditions that determine locking, creep and failure have occupied seismologists in the decades since Reid’s original proposal, without being resolved by the model itself.
Checked against GPS
The practical stakes of the theory go well beyond curiosity about the ground. Recognising that faults accumulate strain that must eventually be released underpins how seismologists think about seismic hazard along fault systems worldwide, and it shaped early instincts toward building codes and insurance practice after 1906, when insurers began treating earthquake damage as a distinct category of risk. The 1906 earthquake was also unusually well documented for its time, with photographic and early film records, which meant the scientific response to it set patterns for how earthquakes would be investigated afterwards. The idea that a fault’s silence is not safety but stored energy has shaped public understanding of seismic risk in California and other fault zones ever since.
What the model does not say
This is worth an hour for anyone who has wondered why earthquakes seem to come from nowhere, because the answer here is that they largely do not: the ground gives warning of a kind, just not one visible without careful measurement over years. The story also shows science working in an unglamorous way, through repeated land surveys and patient comparison rather than a single dramatic observation. It will disappoint readers hoping for a prediction method, since elastic rebound explains mechanism, not timing, and the gap between the two is still where the hard problem sits. As a case study in how a single disaster produced a durable, testable scientific idea, it holds up well.