I examine the resolution of the type of stress drop estimates that hav
e been used to place observational constraints on the scaling of earth
quake source processes. I first show that apparent stress and Brune st
ress drop are equivalent to within a constant given any source spectra
l decay between omega(1.5) and omega(3) (i.e., any plausible value) an
d so consistent scaling is expected for the two estimates. I then disc
uss the resolution and scaling of Brune stress drop estimates, in the
context of empirical Green's function results from recent earthquake s
equences, including the 1992 Joshua Tree, California, mainshock and it
s aftershocks. I show that no definitive scaling of stress drop with m
oment is revealed over the moment range 10(19)-10(25); within this seq
uence, however, there is a tendency for moderate-sized (M 4-5) events
to be characterized by high stress drops. However, well-resolved resul
ts for recent M > 6 events are inconsistent with any extrapolated stre
ss increase with moment for the aftershocks. Focusing on corner freque
ncy estimates for smaller (M < 3.5) events, I show that resolution is
extremely limited even after empirical Green's function deconvolutions
. A fundamental limitation to resolution is the paucity of good signal
-to-noise at frequencies above 60 Hz, a limitation that will affect ne
arly all surficial recordings of ground motion in California and many
other regions. Thus, while the best available observational results su
pport a constant stress drop for moderate- to large-sized events, very
little robust observational evidence exists to constrain the quantiti
es that bear most critically on our understanding of source processes:
stress drop values and stress drop scaling for small events.