Counting Statistics Are Not the Limiting Error in Radiation Dose Measurement: An Error-Structure Recount

Radiation measurements are routinely counted longer in pursuit of smaller statistical uncertainty. From Poisson statistics, 1% precision requires 10,000 net counts and 0.1% requires 1,000,000. For four stated detector scenarios, counting times and crossovers against systematic floors (calibration 5%, geometry 15%, dose conversion coefficients uncertain by a factor of about 1.5-2, i.e. 21-35%) are computed with background subtraction. Where a dose coefficient is applied, statistics fall below the coefficient uncertainty after 23 counts or fewer; against calibration alone after 400 counts. Counting a whole-body measurement for one hour instead of ten minutes moves total uncertainty from 5.04% to 5.01%. The exception is short-dwell handheld survey work, where statistics do dominate.