Richard Wilson, Harvard University
Hormesis, Low Dose Linearity and the Nuclear Industry
Physics Division Colloquium - 3 Mar 2000

In understanding the effects of radiation on people, one depends in the first instance on epidemiological data on the exposure of people to high levels (such as at Hiroshima and Nagasaki) and uses a model to extrapolate to other populations and other doses. In general one wants data where the dose is measured for each individual and the health of each individual is followed. When the dose is high, and the prediction is for the effect of high doses, there is reasonable agreement on the magnitude of the risk. But below a cumulative dose of about 20 Rems (0.2 Sv) direct data are inadequate to tell us whether there is a small risk, a zero risk or whether the radiation dose is beneficial. Under these circumstances some scientists use data of lesser quality - so called "ecological studies" where effects averaged over a large population are compared to doses averaged over the same population. The data on radon by Dr Bernard Cohen are particularly important. When there is a small risk there are many confounding effects. While it is clear that whatever radiation risk exists is small compared to the risk of the confounding pollutant and therefore should not be singled out for attention, it is less clear scientifically what the "true" risk (if any) is. Under these circumstances I pefer to retreat to general arguments that date back to Doll and Armitage, and have been emphasized as being more general by Crump et al and by Crawford and Wilson. These are based upon the general observation that: (1) cancers produced by radiation are not now, and may never be, distinguishable from those thatv occur naturally. (2) that whatever produces the naturally occurring cancers has, by definition, already overwhelmed the appropriate defense mechanisms. I conclude that at doses low enough to cause only a small increase in cancers the relationship of dose and response is almost certainly linear. But this is not unique to radiation. It is not unique to cancer. As studies on automobile accidents tell us it is merely a very ordinary situation. The public is best served by reducing radiation exposure only when it is cost effective, $2000 per man-Rem ($200,000 per person-Sv) for public exposures and 10 to 100 times less for occupational exposures. The nuclear industry would do well to limit their defense by emphasizing these last points.

ANL Physics Division Colloquium Schedule