exposure, and behavior. Identification of these groups was based upon research of historical records, technical reports, or correspondence. For this purpose, a military or naval unit may, therefore, have consisted of several groups, or several units may have comprised a single group. This method was useful for personnel whose activities were confined to a ship and in situations where such activities could be assigned to the entire group under consideration. Using proven statistical methods, the badge data for each group were examined to determine if they adequately reflected the entire group and were therefore valid for use in statistical calculations, or if the badge data indicated, by such characteristics as a bimodal distribution, that the group should have been subdivided into smaller groups where the distribution of readings was more normal. Only when the group data met the above tests were the mean dose, variance, and confidence limits used for assigning doses to unbadged personnel. When using this method, an estimated dose equal to 95 percent probability that the actual exposure did not exceed the estimaté was then assigned to unbadged personnel. This high-sided, but statistically sound, procedure ensured that the assigned doses were much higher than the average or mean for the badged group. 7.5 RECONSTRUCTION OF RADIATION DOSES. The general methodology for dose reconstruction consisted of character- izing the radiation environments to which participants, through all relevant activities, were exposed. The environments, both initial and residual radiation, were correlated with the activities of participants to determine accrued doses due to initial radiation, residual radiation, and/or inhaled/ ingested radioactive material (3; 4). «Because of the variety of activities, times, geometries, shielding, and weapon characteristics, as well as the normal spread in the available data pertaining to the radiation environment, an uncertainty analysis was performed. This analysis quantified the uncertainties due to time and space variations, group size and available data. An automated (computer-assisted) procedure was often used to facilitate handling the large amounts of data and the dose integration, and to investigate the sensitivity to variations in the values of parameters used. The results of the calculations were then compared with film badge data as they 169 LC