the cloud before transport by winds) and the diffusion rate of the cloud as it was moved by the winds.* The primary fallout plot technique initially used on CASTLE was a joint analysis of the hodograph and the 72-hour airborne particle trajectory forecast to define the fallout area for the first 12-hour period and assess the orientation and areal extent of the cloud after the first 12 hours. tempted; For BRAVO, the method of elliptical approximations was first atits use was extremely Limited, however, because of uncertainties in its application to high-yield events. The BRAVO data were used to re- fine this method for subsequent events. The method of elliptical approximations drew ellipses over the hodograph between wind levels. The minor axes of the ellipses were determined by the amount of change in wind direction between wind layers. The radia- tion intensity associated with the ellipses was based on scaling and past experience. The method was based largely on empirical data from the Ne- vada Proving Ground (NPG), and it had been used successfully to predict the fallout fields from low-yield PPG detonations. As the tests began, it became apparent that a more accurate method was needed to clearly define the fallout from high yields. The large differ- ences between NPG and PPG shot yields and cloud heights made the accuracy of using elliptical approximations for CASTLE shots suspect. There was no assurancé that the high-yield fallout mechanisms were described at all. For example, the trapping characteristics of the atmospheric boundary with the stratosphere (the tropopause) could only be conjectured, and even if * Because little was known about the fallout mechanism of high-yield events and because of errors in other parameters of the fallout forecast, some additional safety factor was assumed necessary, even though the value of the factor could not be precisely stated. For CASTLE, the estimate of the diffusion-caused widening of the fallout-contaminated area with distance was increased by the addition of a 15° sector on each side of the calculated fallout area. (Reference 16, p. 61). In previous operations the factor used was 109 lll