One of the mysteries of earthquake mechanics is why earthquakes stop. This
process determines the difference between small and devastating ruptures. O
ne possibility is that fault geometry controls earthquake size. We test thi
s hypothesis using a numerical algorithm that simulates spontaneous rupture
propagation in a three-dimensional medium and apply our knowledge to two C
alifornia fault zones. We find that the size difference between the 1934 an
d 1966 Parkfield, California, earthquakes may be the product of a stepover
at the southern end of the 1934 earthquake and show how the 1992 Landers, C
alifornia, earthquake followed physically reasonable expectations when it j
umped across en echelon faults to become a large event. If there are no lin
king structures, such as transfer faults, then strike-slip earthquakes are
unlikely to propagate through stepovers >5 km wide.