randomly repositioned for each subsequent interval. A distribution around the mean film badge reading is calculated by assuming a random position, corresponding to an intensity reading, each time a crewman comes on deck. The tails of this distribution indicate, in a general way, the possible error of the mean dose if crew positioning were significantly biased toward the extremes of intensity readings. Note: for personnel moving continuously about the deck, their dose approaches the calculated mean. In order to arrive at dose distributions, it is assumed the reported average intensities used to reconstruct the topside environments in Section 2 were derived from many topside measurements that were normally distributed, and could be characterized by a mean ( uy ) and standard deviation («). For the sixteen ships under consideration, shipboard survey data are not available to substantiate this assumption; however, detailed surveys on the YAG-40 following Shots ROMEO and YANKEE indicate a distribution of topside intensity values that can be approximated by applying a normal distribution to the data. Figure 4-1 summarizes the results of surveys taken onboard the ship on 31 March and 8 May. Each survey consists of 70 topside intensity readings obtained at the same location following each shot (Reference 18). _ The survey data are depicted by histograms while the smooth curves represent normal distribu- tions fitted to the survey data. From Figure 4-1, it does appear that the topside intensities following fallout deposition can be adequately represented by assuming a normal distribution of values. The fractional (of mean) standard deviation (u/o), a measure of the spread in the intensity data obtained during each survey, is determined to vary between 0.52 (31 March survey) and 0.40 (8 May survey) on the YAG 40. A value of 0.50 is chosen as being applicable to represent the spread in intensity data around the average (mean) values reported for the sixteen ships of interest. The normal distribution around the average intensity is integrated throughout each interval on deck to obtain the corresponding distribution in dose. When the dose distributions from all intervals are combined, the square of the standard deviation of the resultant normal distribution is equal to the sum of the squares of the standard deviations of the contributing distributions. As contributions from more intervals are added, the fractional standard deviation of the combined distribution decreases. 151 Because the calculated dose in