equations for particle terminal velocity, of the form described by Dallavalle. Such equations consider the variables of particle density, air density, particle diameter, air viscosity, and constants incorporating the effects of gravity and particle shape. (Modified versions of the original Dallavalle equations are presented in Reference 67; data on the Marshall Islands atmos- phere required to evaluate air density and air viscosity are also given in this reference.) The last two steps are simplified, however, by the use of a plotting template, so designed that vec- . tors laid off in the wind direction, to the wind speed, automatically include terminal velocity adjustments (Reference 68). Size lines result from connecting the surface-arrival points for particles of the same size from increasing increments of altitude; height lines are generated by connecting the arrival points of particles of different sizes from the samealtitudes. These two types of lines forma network from which the arrival times of particles of various sizes and the perimeterof the fallout pattern may be estimated, once the arrival points representing the line source have been expanded to include the entire cloud diameter. This last step requires the use of a specific cloud model. The model that was used in arriving at the results of Figures 4.6 through 4.9 and Table 4.3 is shown in Figure 4.11. Particles larger than 1,000 microns in diameter were restricted to the stem radius, or inner 10 percent of the cloud radius, while those from 500 to 1,000 microns in diameter were limited to the inner 50 percent of the cloud radius; all particle sizes were assumed to be concentrated primarily in the lower third of the cloud and upper third of the stem. The dimensions shownin the figures were derived from empirical curves available in the field, relating cloud height and diameter to device yield (Reference 67). Actual photographic measurements of the clouds from Reference 69 were used wherever possible, however, for subsequent calculations leading to results tabulated in Table 4.3. The location of the hot line follows directly from the assumed cloud model, being determined by the height lines from the lower third of the cloud, successively corrected for time and, sometimes, space variation of the winds. Time variation was applied in the field in all cases, but Space variation later and only in cases of gross disagreement. The procedure generally followed was to apply the variation of the winds in the case of the 75- and 100-micron particles and use Shot-time winds for the heavier particles. Wind data obtained from balloon runs at 3-hour inter- vals by the Task Force were used both to establish the initial shot-time winds and make the corrections for time and space variation. Wustrative purposes in Table B.29. The calculations for Shot Zuni are summarized for It is of particular interest to note that it was necessary to consider both time and space variation of the winds for Shots Zuni and Tewa in order to bring the forecast patterns into general agreement with the measured patterns. Vertical air motions were considered for Shot Zuni but found to have little effect on the overall result. It is also of interest to observe that the agreement achieved was nearly as good for Shots Flathead and Navajo with no allowance for space Variation as for Shots Zuni and Tewa with this factor included, in spite of the fact that the fallout from the former consisted of slurry rather than solid particles below the freezing level (Sections 3.3.1 and 3.3.2). Whether this difference can be attributed to the gross differences in the nature af the fallout, is not known. 4.3.2 Sampling Bias. When solid object such as a collecting tray is placed in a uniform air Stream, the streamlines in its immediate vicinity become distorted, and small particles falling into the region will be accelerated and displaced. Asa result, a nonrepresentative or biased Sample may be collected. Although the tray will collect a few particles that otherwise Would not have been deposited, the geometry is such that a larger numberthat would have fallen through the area occupied by the tray will actually fall elsewhere. In an extreme case of small, light particles and high wind velocity, practically all of the particles could be deposited elseWhere, because the number deposited elsewhere generally increases with increasing wind veloc- \ty and decreasing particle size and density. This effect has long been recognized in rainfall sampling, and some experimentaicollectors have been equipped with a thin horizontal windshield designed to minimize streamline distortion 115 .