(Reference 72). The sampling of solid fallout particles presents even more severe problems, however, because the particles may also blowoutof the tray after being collected, producing . an additional deficit in the sample. In addition, samples collected in identical collectors located relatively close together ina fixed array have been found to vary with the position of the collector in the array and its height above the ground (References 10 and 72). It follows from such studies that both duplication ang replication of sampling are necessary to obtain significant results. Consideration was given to each of these problemsin the design of the sampling stations. Anattempt was made to minimize and standardize streamline distortion by placing horizontal wind. shields around al] major array platforms and keeping their geometries constant. (The flow characteristics of the standard platform were studied both by small-scale wind-tunnel tests ang ‘ measurements made on the mounted platform prior to the operation (Reference 73). It was found that a recirculatory flow, resulting in updrafts on the upwind side and downdrafts on the downwind side, developed inside the platform with increasing wind velocity, leading to approximately the same streamline distortion in every case.) Similar windshields were used for the SIC on the YAG 40 and the decay probe tank on the YAG 39, and funnels were selected for the minor array collectors partly for the same reason. Honeycombinserts, which created dead-air cells to prevent loss of material, were used in all OCC and AOC collectors. This choice represented a compromise between the conflicting demandsfor high collection efficiency, ease of sample removal, and freedom from adulterants in subsequent chemical and radiochemical analyses. . Retentive grease surfaces, used in the IC trays designed for solid-particle sampling, facilitated single-particle removal. . All total collectors were duplicated in a standard arrangement for the major arrays; and these arrays, like the minor arrays, were distributed throughout the fallout area and utilized | for all shots to provide adequate replication. At the most, such precautions make it possible to relate collections made by the same kind of sampling arrays; they do not insure absolute, unbiased collections. In effect, this means that, while all measurements made by major arrays may constitute one self-consistent set, and those made by minor arrays another, it is not certain what portion of the total deposited fallout these sets represent. As explained earlier (Section 3.1), this is one reason why radiological properties have been expressed on a unit basis wherever possible. Efforts to interpret platform collections include a discussion and treatment of the relative bias observed within the platforms, as well as comparisonsof the resulting platform values with buried-tray and minor array collections on How Island, water sampling and YAG 39 tank collections, and a series of postopera- tion rainfall measurements made at NRDL. Relative Platform Bias. The amount of fallout collected by the OCC and AOC,collectors in the upwind part of the standard platform was lower than that collected in the downwind portion. It was demonstrated in Reference 74 that these amounts usually varied symmetrically aroundthe platform with respect to wind direction, and that the direction established by the line connecting the interpolated maximum and minimum collections (observed bias direction) coin- cided with the wind direction. A relative wind varying with time during fallout was treated by vectorial summation, with the magnitude of each directional vector proportional to the amount of fallout occurring in that time. (Variations in the relative wind were caused principally by ship maneuvers, or by oscillation of the anchored barges under the influence of wind and cur_ rent; directions varying within +15 degrees were considered constant.) The resulting collection pattern with respect to the weighted wind resultant (computed bias direction) was similar to that for a single wind, although the ratio of the maximum to the minimum collection (bias ratio) was usually nearer unity, and the bias direction correspondingly less certain. The variability in relative-wind direction and fallout rate, which could under certain condi- tions produce a uniform collection around the platform, may be expressed as a bias fraction (defined in Reference 74 as the magnitude of the resultant vector mentioned above divided by the arithmetic sum of the individual vector magnitudes). In effect, this fraction represents a measure of the degree of single-wind deposition purity, because the bias fraction in such a case 116