Improved methods of radiocarbon analysis have enabled us to date more precisely the earthquake ruptures of the San Andreas fault that are recorded in the sediments at Pallett Creek. Previous dates of these events had 95% confidence errors of 50--100 calendar years. New error limits are less than 23 calendar years for all but two of the dated events. This greater precision is due to larger sample size, longer counting time, lower background noise levels, more precise conversion of radiocarbon ages to calendric dates, and better stratigraphic constraints and statistical techniques. The new date ranges, with one exception, fall within the broader ranges estimated previously, but our estimate of the average interval between the latest 10 episodes of faulting is now about 132 years. Variability about the mean interval is much greater than was suspected previously. Five of the nine intervals are shorter than a century; three of the remaining four intervals are about two to three centuries long.

Despite the wide range of these intervals, a pattern in the occurrence of large earthquakes at Pallett Creek is apparent in the new data. The past 10 earthquakes occur in four clusters, each of which consists of two or three events. Earthquakes within the clusters are separated by periods of several decades, but the clusters are separated by dormant periods of two to three centuries. This pattern may reflect important mechanical aspects of the fault's behavior. If this pattern continues into the future, the current period of dormancy will probably be greater than two centuries. This would mean that the section of the fault represented by the Pallett Creek site is currently in the middle of one of its longer periods of repose between clusters, and sections of the fault farther to the southeast are much more likely to produce the next great earthquake in California. The greater precision of dates now available for large earthquakes recorded at the Pallett Creek site enables speculative correlation of events between paleoseismic sites along the southern half of the San Andreas fault. A history of great earthquakes with overlapping rupture zones along the Mojave section of the fault remains one of the more attractive possibilities. ¿ American Geophysical Union 1989