To the Editor.
—Erroneous conclusions will be drawn from the article of Wing et al,1 which described evidence of radiation effects in Oak Ridge (Tenn) National Laboratory workers, if the very large biases and uncertainties in relating effects to radiation doses are not recognized. The problem arises because the radiation-dose records represent no more than lower limits, and the derived radiation effects per sievert are thus upper limits.Why are the radiation-dose records so uncertain and so biased? From the Figure in the Wing et al article, it is clear that most of the relevant radiation dose was recorded before about 1960. In this period, the purpose of the radiationdose monitoring program was to detect large exposures. It was not intended to detect and measure low-dose levels of the magnitude reported by Wing et al, which scarcely exceeded natural background, on the average. In particular, radiation-dose readings less than