Analysis of variance was applied to the 239°240p, concentration and particle fractionation of dust samples in the compartments of the Bagnold sampler assuming equal variance over type and range of treatments used. The results of these analyses show that there was not a significant concentration difference (p < 0.05) across collector height, nor a difference in concentration across sampling intervals (p < 0.05). However, the particle size fractions greater than 53 um diameter was significantly different across sampling intervals (p < 0.05), although not significant across collectors (p < 0.05). In the former comparison, only two of the particle fractions (0.7274, 0.5096) were significantly different from all other fractions, and from each other (p < 0.05). A mean 799?240py concentration of 1. 33 pci/g dry soil was used to estimate plutonium flux using eqs. 9, 10 for any given saltation compartment: Cc, = 1. 91PS99U*Z, ~ 2.21 (15) where C, = Pu flux for ith saltation compartment, pCi/cm*/day 25 = midheight of ith saltation compartment from ground level. Application of this equation over all saltation compartments and time intervals is presented in Table 4. The first compartment, C,, includes the soil creep contribution, and the plutonium concentration has a coefficient of variation of 1.5 for all determinations. DISCUSSION Statistical treatment of GZ data at Trinity Site indicates a number of important factors to consider when attempting to model both dust and plutonium flux at the site. It is observed that a cross-product of parameters rather than the individual ones provides better fits to observed data. Comparison of eq. 1 with those ultimately derived (eqs. 3, 5, 6) for predicting dust flux shows that accumulated precipitation, solar radiation pattern, and diurnal windspeed energy (Healy, 1974) are important interacting parameters when time intervals of about 30 days are used as sampling intervals. Since solar radiation input at the site is related to Sgqg, and also to evapotranspiration rate (Change et al., 1965), it appears to be a better predictor of soil moisture status than either mean diurnal air temperature near the ground or observed soil moisture content at the beginning and end of a sampling interval. It is possible that actual measurement of solar radiation input and/or measurement of soil surface temperatures on a continuous basis would improve the fits. 692