and mortality that might result from different types of accidents and also indicates the tremendous complexity of making such estimates.
The table reveals that the consequences of a single accident may be very large but that such accidents are improbable, and the consequences are small per year when averaged over a period of years. Thus, if the chance of an accident is 1 in 1 billion years, a total of 3000 deaths averages only 1 death per 333,000 years. Such averaging, however, does not consider that the social disruption of a catastrophe can be much greater than that associated with a distributed set of small accidents producing the same number of deaths.
It is of interest to compare the hypothetical catastrophic risks of Table 9–9 with those of routine power plant operation. Core meltdown with aboveground breach of containment (the dominant risk) carries a probability of once in a million reactor-years and leads (in round numbers) to less than 10,000 deaths from all causes. Over the same hypothetical period of 1 million reactor-years, the routine operation of the nuclear plant would be associated with 200,000 deaths, that of an oil-fired plant also with 200,000, and that for coal with 2 million (Table 9–1).* If we suppose that the nuclear catastrophe is underrated tenfold, the risks of routine operation and those of catastrophes become approximately equal.
WASH-1400 has been criticized for (inter alia) underestimating some hazards owing to the use of median rather than mean estimates,139–141 and for underestimating the uncertainty of its results. The recent report by the Risk Assessment Review Group to the Nuclear Regulatory Commission142 and the Ford/Mitre Report143 both make the latter point. The review group of the NRC points out both overly conservative and inadequately conservative assumptions in the probability estimates, and the group concludes that the uncertainty is seriously understated. Both groups note that reactor experience provided an upper bound at the times of their reports.
The effects of pollutants on plant life are judged in terms that are quite different from those applied to judge the effects of pollutants on human health. Cancer, for example, is not a problem in the case of crop plants, nor are the late effects of exposure that take years to develop. While genetic effects might be of interest in special cases, the rare occurrence of a mutant in a wild or cultivated population is not important. Major interest
Statement 9–11, by J.P.Holdren: Table 9–1 includes only the deaths caused by industrial accidents, mostly to workers except in the case of coal trains. The comparison is meaningless.