° Contamination outside the defined area: Environmental assessment samples (soil, vegetation) suggested that fall-out contamination was not confined to Rongelap and Utirik. e Radioactivity measured on “unexposed” atoll: An instrument measured dose at Ailuk, previously classified as "unexposed" was 0.01 Gy/h (1.0 rad/h) one hourafter detonation. e Plume reconstruction probably was inaccurate: A computer simulation using all available meteorologic data indicated that the initial path of the fall-out cloud followed was in an easterly direction, but then it shifted to a south/southwesterly direction. This computer simulation was criticized later as being based on wind data that was too sparse to provide sufficient support for conclusions as to the trajectory of the radioactive cloud. e The prevalence of thyroid nodules among populations classified as "unexposed": Thyroid nodules ranged from 1-10.6% amongresidents of atolls previously thought to be "unexposed" compared to 2.4% background prevalence in two southern islands farthermost from Bikini, a statistically significant increase. A well-designed retrospective cohort study of thyroid neoplasia in 7,266 Marshall Islanders from 14 atolls of which 2,273 were alive in 1954 and were potentially exposed to Bravo fallout was published in 1987. 3.4.2 Duration of Follow-Up Follow-up of the exposed cohort differed from that of the unexposed group. Follow-up of the exposed group began in 1954, whereas it did not begin until 1957 for the unexposed group, and some deaths among the unexposed group already had already occurred during the intervening three years. 33