The potential benefits of nuclear power are great, but their achieve­ment involves accepting certain risks. These benefits and risks must be balanced. Evaluations of benefits and risks are made in many ways. One is through the preferences of the public — for example, in choosing among rail, air, and automobile travel. Another is through the governmental processes for determining our goals and the means of achieving them.

Nearly everyone appreciates the difficulty of making such evalua­tions of benefits and risks. There is often disagreement even among rea­sonable individuals about what is beneficial and what is harmful. Even when agreement can be achieved, the difficulty remains of assigning a value to indicate the degree of benefit or harm. For example, how does one assign relative values to the aesthetically pleasing appearance of nu­clear plants and the less attractive features of a fossil plant?

A further problem is how to take account of the statistical probabili­ties involved. For instance, how is account taken of the probabilities asso­ciated with more than fifty thousand deaths a year in automobile acci­dents, and about two thousand deaths a year in aircraft accidents? Then too, how does one balance the health hazards attributable to the coal min­ing industry and to air pollution from fossil fuel stations, with the bene­fits? (See Appendix, p. 221 below.)

In the nuclear power field, we are constantly seeking ways of better evaluating both benefits and risks. We have considered whether an even more quantitative approach than the present one could be used to evaluate the safety of nuclear power plants. We requested a study group appointed by the aec to consider this matter as part of a study of the regulatory proc­ess. The group concluded that with existing techniques and knowledge, the total risks to the public from nuclear power plants, although very small, cannot now be meaningfully expressed in numerical terms. But the group also said that quantification techniques do show promise as a tool in corn —

parative safety evaluation and that efforts should be made to improve the collection of data. (“Report to the Atomic Energy Commission on the Reactor Licensing Program,” June 1969.)

One reason for not being able to express total risks in numerical terms is the excellent safety experience of nuclear power plants, where no meaningful risk experience has been accumulated. The fact is that no deaths or accidents affecting the general public have occurred in any civil­ian nuclear power plants in the United States. There has been only one re­actor accident in the United States reactor program which resulted in fa­talities. It occurred in an experimental Army reactor, the SL-1, at our testing station in Idaho and involved the deaths of three operators.

A growing body of literature reflects the efforts being made to devel­op methods for evaluating benefits and risks more analytically. Such ef­forts were reflected in the proceedings of a recent symposium of the Na­tional Academy of Engineering (Ramey, May 1, 1969). Also worth not­ing in this connection is an article in Science, “Social Benefit versus Tech­nological Risk,” by Chauncey Starr, dean of the engineering school at ucla (Starr, 1969).