The analysis of the San Andreas event, reported in this week's issue of Nature, has yielded a surprisingly complex picture of the slow process; in the few cases previously reported, the slow rupture appeared to be rather simple. Carnegie's Alan Linde and colleagues* report that the slow 1992 quake had a total displacement across its rupture surface of only a few centimeters but was slower by two orders of magnitudes than any previously detected slow event. Its complexity, they write, was comparable to that of a regular earthquake.

The borehole strainmeters which picked up the slow earthquake signals are located near the transition between locked and stably sliding segments of the San Andreas Fault in central California. (In stably sliding faults, slip usually occurs through creep and small earthquakes. Locked faults often produce large ruptures; the 1906 San Francisco earthquake occurred in the locked partion of the San Andreas to the north of the study area.) The slow strain changes extended over 30 square km (about 11 square miles) of fault surface area with a total strain rupture corresponding to a magnitude 4.8 earthquake. They included a series of slow events with several episodes of varying slip time. Depth ranged from 0.1 km to at least 4 km, and perhaps as much as 8 km. The small regular earthquakes accompanying the slow earthquake sequence were each no larger than magnitude 3.7.

Together with previous slow earthquake analyses, this new work suggests that a relation exists between the amount of slow redistribution of stress and the size of associated earthquakes. For example, a slow earthquake preceding the great 9.5-magnitude Chilean earthquake of 1960 was very large in extent and had a slip of several meters. A 1978 slow event in Japan had a slip of about one meter and was followed by earthquakes as large as magnitude 5.8. The 1992 California event, displacing only a few centimeters, was accompanied by very small earthquakes.

What this means for understanding the earthquake process is not clear. Several scientists, among them Linde and his colleagues at Carnegie's Department of Terrestrial Magnetism in Washington, D.C., have previously suggested that slow earthquakes are an integral part of the seismic faulting process. Such events, they say, may be important in earthquake nucleation, i.e., they may act as triggers in initiating the process leading to normal earthquakes. Or, they may account for the excess of plate convergence over seismic slip in subduction zones. (The rate at which two plates converge is often greater than the detected rate of seismic slip; therefore, some other mechanism, such as a slow earthquake, must account for the difference.) This particular slow earthquake, located where locked and stably sliding segments of the San Andreas join, may have some bearing on the nature of such fault transitions.

However, it is not easy to detect slow earthquakes. Because rupture takes place slowly, seismic waves are not generated. Thus, the usual techniques for detecting earthquakes (with seismometers) are not available. Additionally, the total surface displacements of slow quakes are often too small to be seen by conventional distance measuring techniques, such as the satellite- based Global Positioning System. As well, the signal amplitudes of slow earthquakes decrease rapidly with increasing distance from the source. With only a few arrays of instruments like borehole strainmeters sensitive enough to detect them, progress in determining their role in the seismogenic process is likely to be slow.

Research in seismology has been a part of the Carnegie Institution's Department of Terrestrial Magnetism for decades. DTM scientists, for example, helped develop the Sacks-Evertson borehole strainmeter, in which a hydraulic amplification system permits detection of extremely small changes in rock strain. A Sacks-Evertson strainmeter was one of the two instruments used in this study. (The other was a Gladwin tensor strainmeter.) Both instruments are cemented under compression into subsurface rock, typically 200 km deep.

DTM, led by Sean Solomon, is one of five research arms of the Carnegie Institution of Washington, a nonprofit scientific and educational organization founded by Andrew Carnegie in 1902. The president of the Institution is the biologist Maxine Singer.

*The authors of the Nature paper are Alan T. Linde of DTM, Michael T. Gladwin and Ross L. Gwyther, both of the Commonwealth Scientific and Industrial Research Organisation, Queensland Center for Advanced Technologies, Australia; Malcolm J. S. Johnston, USGS, Menlo Park, CA; and Roger G. Bilham, University of Colorado, Boulder.