Large mainshocks in the Uniform California Earthquake Rupture Forecast (UCERF) catalog (1850–2011) appear to have been nonrandom in time in the northern California region. Magnitude (M)≥6.4 earthquakes clustered significantly in the half‐yearly solar declination cycle. The most likely explanation appears to be natural earthquake periodicity, because the semiannual bias (1) persists in historical and instrumental catalog subsets, regardless of foreshocks or aftershocks, (2) is not contradicted by pre‐1850 data for northern California, and (3) is replicated in southern California subregions, other plate‐tectonic settings, and global seismicity. Alternatively, the UCERF catalog has overlooked many M≥6.4 mainshocks that occurred outside of the observed semiannual clustering phase or contains systematic errors that bias the distribution of large earthquakes in the semiannual cycle. The semiannual clustering in northern California shifted abruptly after the great 1906 earthquake from a phase during the month after equinox to a phase near solstice, with large earthquakes returning to the postequinox phase since about 1980. In some other regions where semiannual clustering is evident, a similar phase shift occurred after a great earthquake. Correspondence between the semiannual phase and the tectonic regime of the clustered earthquakes suggests those two parameters may be related. I speculate that the apparent periodicity is a response of the regional fault network to the semiannual solid‐earth tide and explore possible explanations for the shift in its net phase after a great earthquake.
Online Material: Case studies for various regions, tables of earthquake parameters, and figures showing maps and distributions of mainshocks in various regions.