Windspeed data from GZ is biased in that short-duration, relatively high windspeeds responsible for most of the resuspension of dust are obscured by more frequent and longer-duration lower windspeeds because of the long-distance integration length of the anemometer employed. Inspection of the MRI windspeed data over the intervals used did not reveal the higher-velocity gusts of wind one would expect at the site. This might explain why the prediction equation for dust flux is relatively insensitive to this parameter over long time intervals. Another explanation for low windspeed sensitivity in the prediction equation may be due to the small number of observations used in estimating diurnal windspeed averages, or that the subset used was not representative of windspeed events responsible for the bulk of particle resuspension. Future analysis of the entire data base may answer some of these questions. The dominant effect of precipitation both individually and as part of a cross-product term in predicting dust flux (eqs. 5, 6, 7) is difficult to explain without further study. It is possible that wind-driven soil events are associated with gusts of wind that surround a precipitation event in addition to particle pickup by raindrop impaction, water erosion into the lower collectors of the Bagnold sampler, or possibly through dissolution of crustal layers that form at the site. The latter might make soil particles more resuspendable after a given rain event until the layer would reform. The breakup of crustal layers by saltation events would also be enhanced under these conditions. Good fits for the prediction of dust flux fractions above the ground surface (eqs. 10, 11) using Bagnold sampler collections in each compartment over the entire study period, together with the good fits between soil creep and saltation events (eqs. 12, 13, 14), seemingly make possible simpler data collection schemes for predicting dust flux at GZ site. For example, one can obtain an efficient prediction equation for GZ site through the use of eqs. 5, 10, 12, which requires only that precipitation accumulation within each sampling interval be known: C, = 4.515 x 10*PSg9Z,*+7), (16) The simplicity of this equation may be compared to eq. 15, which contained the windspeed term. However, one should use caution in attempting to generalize such a relationship unless checked at other sites with different topography, climate, and soil characteristics. Also, the substitution of solar radiation input for Sg9 would probably improve eq. 16. The prediction of plutonium flux at GZ site was estimated using the product of the mean plutonium concentration with a dust flux predictor because analysis of variance did not show significant trends in concentration between different time intervals or with height of Bagnold sampling compartments above the ground. The- fractionation of soil particles in a similar manner from ground level to 75 cm above the ground is puzzling, as is the almost constant soil fraction of particles greater than 53 um 693