reality approaches a limit with time, a finite distribution remains around the mean total dose. Distributions for each ship are reported at the 90-percent level, i.e., 11.650 (Sth to 95th percentile). Although exposure below deck to fallout makes some contribution to the mean total dose, it is not used in generating a topside dose distribution because its minor contribution involves an averaging of topside readings (for geometrical reasons). Despite the simplified calculation of mean dose starting on the third day after burst, the uncertainty analysis continues to reflect three intervals (taken equal) per day of on-deck exposure at random positions. The value for the fraction of time spent on deck is estimated to be accurate within a factor of 1.2 with 90-percent confidence. For the typical (non-shot) day, this corresponds to 8 to 11% hours on deck. The systematic uncertainty in the time on deck is considered to be greater than its random variation from day to day and ship to ship. The uncertainty in mean total dose is reasonably high-sided by treating the uncertainty in time on deck as a systematic error; as such, the factor of 1.2 applies to the on-deck contribution to the mean total dose as well. Not only the means, but also the distributions as discussed above (minus the below-deck contribution) are directly proportional to the time spent on deck. The below-deck contribution introduces a small, ship-dependent perturbation to the factor of 1.2. The ship-shielding factor reduces the below-deck crew exposure to fallout to a minor contribution to dose, thus any realistic error in that parameter has only a fewpercent effect on the total dose. For example, for a typical day (60 percent below deck) and a ship-shielding factor of 0.10, with an error generously assumed to be +0.05, ‘ ‘ . 0.60(0.05) . . the fractional error introduced is 0-60(0-10)20.50(1) = 0.065. Such values negligibly increase the uncertainty in dose resulting from uncertainty in time spent topside. For doses resulting from fallout onbdard ships or islands, the calculated dose distribution for typical personnel (except as noted) and the uncertainty in the mean (based on time topside) are as follows. The bounds on each represent the 5th and 95th percentiles. 153