A significant sulfate artifact may be produced by the interaction of S
O2(g) with basic particles on filters when the SO2 is not first remove
d by a denuder. Such an artifact could invalidate most historical sulf
ate measurements in regions with alkaline soil. To test this hypothesi
s, we compare measurements from Teflon filters without a denuder to me
asurements from nylon filters following a carbonate denuder from the I
nteragency Monitoring of Protected Visual Environments (IMPROVE) netwo
rk. The data at 11 remote sites in the Desert Southwest show no eviden
ce of a sulfate artifact. Using a normal error distribution, we show t
hat any artifact is much less than 20 ng/m(3). In addition, data from
16 collocated samplers at Meadview, AZ, show that removing the SO2 by
denuders does not change the measured sulfate. The mean difference bet
ween no-denuder and denuder is -4 +/- 10 ng/m(3) for Teflon filters an
d 3 +/- 17 ng/m(3) for nylon filters. Brigham Young University obtaine
d large differences between their denuder and filter pack samplers at
Meadview. Our analysis shows that these differences are explained bett
er by a bias in their denuder sampler than by an SO2 artifact. Our ana
lysis also shows that their measurements are too imprecise to determin
e an artifact less than about 500 ng/m(3).