The HOW collections in general contribute little information regarding the variation of ground factor with bias ratio, since the three observed bias ratios differ only slightly. This only set of platform-ground data is too limited for extrapoletion to other s.w. system. Little differences are shown by the ground factors of Shots A and D as expected; there is no reason to believe Na was otherwise. MULTI-WIND SYSTEM Multi-Wind Relative Bias In the case of milti-winds the variation in sampling is further complicated by an air-flow pattern that varies in orientation and intensity with the different winds. To study the sampling bias of the complex m.w. system, it has been assumed that the system is the summation of several s.w. systems and the bias effects are cumulative. This assumption is basei on the analysis of m.w. collection data and the success of a vector system, described below. The collection data shows that the m.w. collection curve is very similar, if not identical, to the s.w. curve and it is likely that this similarity is due to the resemblance of s.w. curves to sine curves. The addition of several s.w. curves is analogous to the summation of several sine curves of identical period but varying phase angles and amplitudes whereby the resulting curve is another sine curve with the same period. In the case of uniform collection, relative bias does not exist; however, the problem of ground bias remains and therefore platform values are not ground values. This unique situation occurs when the relative winds rotate uniformly around the platform an integral number of times or when there occur two opposing winds with equal fallout amourts and equivalent combination of falicut variables (vind speed, particle size and éensity).* A vector system has been developed to aid in the analysis of m.w. rela- tive bias. Representing each constituent s.w. system is a bias vector whose direction is the wind cirection and whose magnitude is proportional to the relative amount of fallout that occurs within the particular time-increment. In general, wind speeds, which account for the intensity of the flow pattern, must also be considered; however, since the wind speeds encountered were relatively uniform, this complication is avoided in this study. One important application of bias vector summation is the correlation of the observed bias direction relative to the many wind directions involved since the *Consider the analogous summation of two identical sine curves 180° out of phase. 16