Our latest attempts have been directed toward an evaluation of kriging for estimating Pu concentration contours. This estimation technique involves using field data to estimate the spatial correlation structure that may exist at a given study site. This structure is then used to estimate the optimum "weights" to apply to field data points to estimate the concentration at another point. A general introduction to kriging and an account of the first attempts to use kriging for estimating Pu concentration contours at the Project 57 site was given by Barnes et a?. (1977). The present paper also makes use of kriging to estimate plutonium concen- trations over space, but it differs from the approach used by Barnes et al. (1977) in that it relies primarily on the observed linear relationship (in logarithmic scale) between field FIDLER cpm readings for Am, and Pu uCi/m* concentrations in 10-gram aliquots of soil. This linear relationship was first studied by Church et al. (1975). The Pu data are used primarily to adjust the initial estimates of average Pu concentrations obtained using the log Pu-log FIDLER regression. The principal steps in this two-stage estimation procedure are as follows: Stage 1 1. Estimate the linear regression relationship between Pu and FIDLER measurements using data at random locations. 2. Use the above regression equation to convert FIDLER readings into Pu concentrations. 3. Use kriging to estimate the average Pu concentration for each 100- x 100~-foot cell. Stage 2 4. Compute the differences (residuals) between observed Pu soil concen- trations at random locations and those predicted from the Pu-FIDLER regression. 5. Use kriging to smooth these residuals to obtain an estimate of the average correction to apply to the estimated average Pu concentration (from step 3 above) for each cell. The results obtained in this paper should be viewed as another step in an evolutionary process toward the evaluation of different statistical methods for handling the highly skewed and variable nature of transuranic field data. The present kriging approach is an improvement over the approaches tried by Gilbert et al. (1975) and Gilbert (1978) since approximate confidence intervals are obtained on cell there is still much to be learned, particularly about when transforming estimates in logarithmic scale back scale. These and other problems are discussed in the 366 averages. However, the bias introduced to arithmetic following sections.