Operation CASTLE Commander's Report (1954).
'The Bravo event of the CASTLE series yielded 15 megatons, unexpectedly the most ever exploded in atmospheric testing by the U.S.. A scientific miscalculation based on a then-unknown "tritium fusion bonus" of highly enriched lithium-6 contributing the the detonation yield caused the yield to be about two and a half times of that what was expected.
Reports indicate that Bravo was the single worst incident of fallout exposure in all of the U.S. atmospheric testing program. Despite winds in an acceptable flow zone, the greatly unexpected yield caused a much larger mushroom than originally calculated, causing a much wider and massive accumulation path of fallout danger. Fallout was scattered over more than 5,000 square miles of ocean and islands, resulting in the contamination and exposure of military, civilian U.S. personnel working on the shot, and people of the islands who were earlier moved to a supposedly "safe" island but received large amounts of radiation. Acute radiation effects were observed among some of these people.'
Hacker, Barton C.: Elements of controversy: the Atomic Energy Commission and radiation safety in nuclear weapons testing, 1947-1974. University of California Press, 1994 - 614 pp.
Military Effects Studies on Operation CASTLE (1954).
United States Nuclear Tests, July 1945 through September 1992.
U.S. Department of Energy Nevada Operations Office, DOE/NV--209-REV 15, December 2000.