resultant vector direction should be the bias direction. This resultant direction has been designated as the computed bias direction to distinguish it from the observed bias direction. Since sampling variation is reduced or even eliminated by the variance of wind direction, vector analysis aids in the explanation of certain bias effects. Jn a hypothetical sense the bias vector concept regards the final collection variation as the result of a portion of the fallout being deposited biased and the remainder being distributed uniformly.* Thus in the event of o resultant vector with zero magnitude, the entire fallout is uniformly deposited; but in the case of a e.w. system where the vector resultant is equal to the arithmetic sum of the individual vectors, the entire fallout is deposited biased in a given direction. A quantity designated as the bias fraction provides a relative measure of this division of fallout and is defined as follows: bias fraction « final_amount of fallout that is assumed_to be deposited biased total amount of fallout = Magnitude of resultant vector arithmetic sum of vector magnitudes A bias fraction approaching unity indicates a s.w. system is in effect (assuming wind speeds are comparable). The opposite extreme is a value close to zero which indicates uniform deposition and no relative bias. Muiti-Wind Ground Bias Due te the complexity of the m.w. system, equivalent ground value determinations by the bias ratio-ground factor method involve complicated qualifications and therefore will require an exceedingly large amount of data. For instance two m.w. cases of equal bias raties have equal ground factors only if both have reiative correspondence in wind and fallout variation. Until further platform-ground information is available no reliable method of determining ground value is known. If necessary the mean platform value, since it is a weighted mean, may be used as a lower-limit approximtion of the ground value. *As an example, consider the eituation of two opposing consecutive winds A and B of uniform fallout rate and equal fallout variables. Assume the fallout duration of wind A as t and wind B as 2 t. At the conclusion of wind A fallout, wind B fallout begins and after t time, uniform collections exist at this particular intermediate time point. Proceeding further, a wind B fallout of t duration (1/3 of the total fallout) is then added to the uni- form collection. 17