The creep before the quake.

A study of California's San Andreas fault reveals that Earth's crust gets antsy when a tremor is impending-an observation that may help predict some quakes.

The central part of the San Andreas fault is known as the creeping section because land on either side of the fault often shifts a few millimeters without producing any vibrations. Clifford H. Thurber of the University of Wisconsin-Madison wondered whether these small movements along the central region of the fault tended to precede earthquakes measuring magnitude 3.3 or larger.

In periods of heightened quake activity, Thurber found, a statistically significant number of jolts did come fewer than 5 days after the fault shifted. The link is far from perfect, though. Between Feb. 1, 1972, and Jan. 31, 1973, six quakes occurred right after episodes of creeping. Yet three quakes hit without any preceding creep, and seven creep incidents were not followed by a quake, he reports in the April 4 Nature.

Thurber has since extended the study to show that no link exists between creep and quakes when seismic activity is below average, perhaps because stress in the crust is relatively low.

The correlation between quiet shifts and earthquakes may provide a prediction tool for the small subset of faults that creep, he says. If they notice that creep starts to coincide with quakes, it could indicate an increase in stress along the fault, thus raising the odds that a sizable quake will occur in the near future. By this test, the San Andreas near Parkfield, Calif., is not yet ready to pop. Seismologists had predicted that a magnitude 6.0 quake would strike Parkfield by 1993, but it never came. Because creep and tremors there remain uncorrelated, Thurber says the quake is not imminent.