The principal results of this study of remotely triggered seismicity i
n The Geysers geothermal field are the demonstration that triggering (
initiation of earthquake failure) depends on a critical strain thresho
ld and that the threshold level increases with decreasing frequency or
equivalently, depends on strain rate. This threshold function derives
from (1) analyses of dynamic strains associated with surface waves of
the triggering earthquakes, (2) statistically measured aftershock zon
e dimensions, and (3) analytic functional representations of strains a
ssociated with power production and tides. The thresholds also consist
ent with triggering by static strain changes and implies that both sta
tic and dynamic strains may cause aftershocks. The observation that tr
iggered seismicity probably occurs in addition to background activity
also provides an important constraint on the triggering process. Assum
ing the physical processes underlying earthquake nucleation to be the
same, Gomberg [this issue] discusses seismicity triggered by the M(w)
7.3 Landers earthquake, its constraints on the variability of triggeri
ng thresholds with site, and the implications of time delays between t
riggering and triggered earthquakes. Our results enable us to reject t
he hypothesis that dynamic strains simply nudge prestressed faults ove
r a Coulomb failure threshold sooner than they would have otherwise. W
e interpret the rate-dependent triggering threshold as evidence of sev
eral competing processes with different time constants, the faster one
(s) facilitating failure and the other(s) inhibiting it. Such competit
ion is a common feature of theories of slip instability. All these res
ults, not surprisingly, imply that to understand earthquake triggering
one must consider not only simple failure criteria requiring exceeden
ce of some constant threshold but also the requirements for generating
instabilities.