Material velocity changes have been reported to precede the last two magnitude 5 earthquakes along the San Andreas fault in central California. In both cases the anomalies were based on an increase of ∼0.2 seconds in travel-time residuals from small regional earthquakes at one or more nearby seismic stations. A detailed reexamination of the data shows that the changes were more likely caused by differences in the depth and magnitude of the source earthquakes during the “anomalous” periods and were unrelated to any premonitory material property changes. Additional data are presented from sources chosen to minimize such problems. They show that travel times before the two magnitude 5 earthquakes were in fact stable to within a few hundredths of a second for rays that passed within a few kilometers of the hypocenters.
Given the great latitude that can be exercised in the selection of data after the fact to define premonitory changes, such anomalies may not be of any significance unless it is explicitly shown that they are not due to some other change in the sources used or signals measured.