additional time and money, the administrator or person who is actually going to use the data should decide on the precision required. Furthermore, this should be decided before the study begins, In transuranic field studies, the precision needed to meet objectives may be difficult to define, as, for example, when a decision regarding cleanup of a contaminated area must be made. Decisions based on highly accurate and precise estimates of contamination levels are obviously desirable. However, stringent requirements on precision may increase sampling and analysis costs considerably. Choice of Measurement Instrumentation and Laboratory Procedures A variety of measurement methods have been used in transuranic field studies depending on objectives, budget considerations, variability over space and time, and other factors. Some examples are field im situ instrument surveys, aerial surveys, and wet chemistry or Ge(Li) scan analyses on environmental samples. The choice of instrumentation will also affect our perception of reality. For example, in situ field gamma counters scan a large amount of soil relative to a 10-gram aliquot being scanned in the laboratory, Hence, the variability between adjacent in situ readings may be considerably less than that between two 10-gram aliquots from the two scanned areas due to an averaging effect in the field. Each method gives us a different view or perspective on the contamination as it exists in the field. Care should be taken so that the most appropriate instrumentation is used to fit the objectives of the study. For example, an in situ measurement may be appropriate for establishing general levels of contamination in soil, but inappropriate for detailed study of the relationship between soil and vegetation concentrations. Division of Population Into Sampling Units Before sampling begins, the population should be divided into sampling units that do not overlap and which cover the population to be sampled. The samples actually collected are drawn from this collection of sampling units. As a simple example, the ground area of a study site might be gridded (conceptually or on paper) and certain of these grid cells chosen at random from which the vegetation is collected and analyzed for Pu. If changes over time are expected to occur, as, for example, with resuspension measurements, then the sampling unit might be a unit of space over a specified time period. The totality of sampling units is called a "frame." Conceptualizing the environment as composed of distict units that fill up the environment but do not overlap can be helpful in the design of field studies because it tends to reduce a complex situation into more manageable distinct sampling units. Selection of Samples There are many ways of selecting particular sampling units for collection. These include simple random, stratified random, or systematic designs, as, for example, on a grid. The method of choice depends in part on the patterns and trends that are likely to be present over space (or time). For example, if soil is uniformly contaminated in all parts of the study site, then the variance of the estimated average concentration is likely to be similar for any of the above-mentioned plans (assuming equal sampling effort). However, if trends in 578 concentration do exist, then either the stratified random or possibly a systematic design is likely to result in a smaller variance estimate. The choice between stratified random and a systematic grid system depends on the correlation structure between samples’ various distances apart. If a considerable amount of data is available, perhaps from a preliminary study, it may be possible to estimate this correlation structure in order to make an objective choice between sampling plans. A knowledge of spatial correlation can also be used to obtain better estimates of averages or totals using an estimation technique termed Kriging (Davis, 1973, pp. 381-390). This technique is also mentioned below in connection with sampling for cleanup. REVIEW OF TRANSURANIC SAMPLING DESIGNS This review is divided into two parts; studies at nuclear detonation or safetyshot sites where the spatial distribution of the contaminants is of considerable importance, and ecosystem-type studies where movement of transuranics between ecosystem components is of primary interest. These two parts are not mutually exclusive, however, since ecosystem studies are usually an important part of studies at nuclear (fission) or safety-shot sites. Local Contamination at Nuclear Detonation or Safety-Shot Sites Radionuclide contamination at safety-shot and nuclear detonation sites is generally highest near ground zero (GZ) with the pattern of contamination determined in part by wind speed and direction at the time of detonation, and whether the assemblies of Pu were covered by a steel plate, buried in the soil, etc. Considerable information may be available from in situ instrument surveys conducted following the shot or prior to the main soil-sampling effort. These advance surveys may serve to define the general level or spatial distribution of radionuclide concentrations over the area. At safety-shot sites (studied by the Nevada Applied Ecology Group (NAEG): see Dunaway and White, 1974; White and Dunaway, 1975, 1976), the FIDLER instrument was used to measure 24lam, which at these sites gives a general idea of the 239>240py present in surface soil. Aerial surveys have also been used to estimate spatial distribution. Examples here are the gamma surveys for ®°co, 13’cs, and *"lam over the islands of the Enewetak Atoll (Stuart and Meibaum, 1973), gamma surveys for !37Cs and ?4lam over the Hanford Reservation in 1973 (Bruns, 1976), and an aerial radiological survey of the Nevada Test Site (NTS) (EGG, 1972). (See Burson, 1974, for recent research progress in airborne surveys.) In situations where “Am and 239°240py concentrations in the same aliquot are related, surveys for 24!Am hold promise for estimating the spatial pattern of Pu in soil. Survey information may in some cases be useful for obtaining at least a rough estimate of the Pu inventory. To do so requires estimation of a “calibration" relationship between the in attu or aerial survey readings and Pu concentrations obtained in soil samples. Gilbert and Eberhardt (1976) evaluated the suitability of estimating Pu in soil aliquots taken from safety-shot sites using 579