FIDLER data obtained at random soil sample locations immediately prior to taking the soil sample, They found the method (termed “double sampling") to be applicable for higher levels of contamination where the correlation between FIDLER readings and Pu wet-chemistry concentrations was sufficiently high and the cost ratio of FIDLER to wet-chemistry analyses sufficiently low, Once surveys have estimated the spatial distribution Pattern, additional survey readings and/or soil samples are sometimes taken along a transect running from GZ down the "hot line” or center of the fallout pattern. Closer spacing between sampling may be used near GZ in order to estimate more accurately the rapid changes in concentration levels likely to occur in that area. This was the general approach used at the NAEG safety~sho t sites, the Plutonium Inventory and Distribution Program at NTS (Church et al., 1974; Brady and Church, 1975), and at the Trinity site in New Mexico (Hakonson and Johnson, 1973). The purpose of such transect samples is usually to gather information to facilitate the design of more intensive studies. Soil samples collected along a single transect are usually too few and widely spaced to be useful for estimating inventory. If the objectives of the study require additional informatio n on soil concentration, there are several design options open to the investigator depending on objectives. If the in situ or aertal surveys have satisfied the requireme nts for information on the spatial distribution of contaminat ion, then soil samples may be collected primarily to estimate inventory, In this case, the survey information may be useful for stratifying the study site within which soil samples are collected. The idea of stratification is to enclose areas of different concentration levels into Separate strata (subareas) so that the vartability between samples within strata is as small as possible relative to the variation between stratum means. If the sample locations within strata are chosen at random, this design is termed stratified random sampling (Cochran, 1963). Stratified random sampling was used at the NAEG safety-sho t sites (Gilbert and Eberhardt, 1974; Gilbert et al., 1976a) and tended to result in estimates of inventory with smaller standard errors than if stratification had not been used (Gilbert et al., 1975). If in situ or aerial measurements are not sufficiently well correlated with soil concentrations, or if such surveys have not or cannot be done, the design of the intensive soil sampling plan may need to fulfill the two objectives of spatial distribution and inventory. One approach is to arrange a rather fine grid over maps of the study area and to choose at random a number of squares defined by the grid. This approach was used for sampling soil on most islands in the 1972 Enewetak Radiological Survey (Lynch and Gudikson, 1973). The selection process can be performed using a random number table such as produced by the Rand Corporation (1955), or by using a good pseudo-ra ndom number generator on a computer. The locations actually sampled would be the centers of the randomly chosen squares. An alternative design would be to first stratify the area based on 24Jag Ge(Li) scans on soil samples collected on a grid. Once the strata are determined, a grid within each stratum (grid size need not be the same for all strata) could be established. Then two or more locations could be chosen at random within each grid square in each stratum. 580 The above discussion has concerned contamination from local sources. In this situation, it is usually best to avoid sampling at only "undisturbed" locations, since these locations may not be representative of the area (Eberhardt, 1976). However, when estimating total deposition from worldwide fallout, it is common to sample only undisturbed sites (Hardy and Krey, 1971). The distinction is whether interest centers on estimating the existing spatial distribution and inventory or on estimating the total amount deposited over time. Undisturbed sampling locations have also been used to estimate the spread of contamination from the Rocky Flats plant (Krey and Hardy, 1970). The issue involved here is important for other study objectives, also. Consider a study for evaluating the need to clean up an area, In terms of potential health risk, it may be more important to determine where most of the contaminant is presently located than to know only the total amount initially deposited. Sampling to Assess Relationships After information has been obtained about the spatial distribution of transuranics, there may be interest in estimating the amounts of contamination in various ecosystem components and the movement rate of transuranics between components. The design of these studies depends to a large extent on the particular ecosystem being studied. The design of such studies should be based on whatever information is available or can be obtained on where the bulk of the contamination is located, and how this contamination is likely to move through the environment. Three ecosystem-type studies may be briefly mentioned. These include NAEG studies in Nevada at 10 safety-shot sites referenced above. Soil and associated vegetation samples were collected at ail 10 sites according to a stratified random sampling plan. Small mammals were also collected at some sites, and in Area 13 (Project 57), cattle were grazed in the Pu-contaminated areas to obtain information on uptake and tissue distribution of Pu due to grazing a contaminated area (Smith ef al,, 1976). An important objective of these NAEG studies is to evaluate the potential hazard to man from this contamination. A provisional Pu transport and dose estimation model has been developed for this purpose by Martin and Bloom (1976). Gilbert et al. (1976) gave an initial synthesis of the Area 13 Pu data for soil, vegetation, small mammals, cattle, and a "hypothetical man." Other transuranic ecosystem studies include those at the Savannah River Plant and in. some of the canyons at Los Alamos. At Savannah River (McLendon et al., 1976), samples of soil, a resuspendible fraction of soil, vegetation, grasshoppers, and cotton rats were collected and their Pu (238py, 239°248py) concentrations compared in order to study the transport of Pu in a humid climate. The primary objective of the Los Alamos studies (Hakonson et al., 1976) was to estimate the ?38pu, > Pu, and 137¢g levels in canyon ecosystem components (stream sediments, vegetation, small mammals) as a function of distance below waste discharge areas. The design of such studies is complicated due to incomplete knowledge of how transuranics move through the environment, and of the spatial patterns and correlations over time and space. Under the assumption that samples collected near each other are more alike than those from further apart, the collection of adjacent soil and vegetation samples is often practiced. However, Gilbert et al. (1975) illustrate that in desert environments, the correlation between 581