CONDITIONAL PROBABILITIES

There is much more. The report (as its title shows) is not one for researching the consequences of reactor accidents at all; it searches for the best criteria in comparing possible sites for reactors. For this it uses conditional probabilities, and please do not be frightened by the concept. They are simply probabilities of events that would occur under circumstances that are assumed, rightly or wrongly, to have set in. You can calculate the probability of Sen. Kennedy being blown away by the wind, assuming that he is slimmer than Twiggy, or the probability that those rocks will reach the top of the mountain without breaking, assuming that they roll uphill. It does not matter how improbable or even absurd the assumed condition is; mathematical theorems don't care and computers are patient.

Over and over again, the report makes it clear that its figures deal with conditional probabilities, assuming that there has been a meltdown, plus a breach of the containment building, plus an unfavorable weather situation, etc. That is the difference between the probability of living to age 148 in general, or of living to age 148 assuming you have already reached age 147.

Do you believe Prof. Kendall does not know the concept of conditional probability?

The report, incidentally, does not derive these basic probabilities ("source terms" in its jargon), but accepts them from previous studies, although these hypothetical values are now being abandoned. Recent research (see special issue of Nuclear Technology, May 1981, particularly article by Levenson and Rahn), shows the absence of radioiodine and other radionuclides in several nuclear accidents and planned experiments, though by older theories they should have been present. This has opened a new chapter in nuclear safety: It now appears highly plausible that due to physical laws (plating out of radioactive toxins in a meltdown) a public catastrophe of the type envisioned by the Rasmussen study (with low probability) and the panicmongers (with higher probability) is quite out of the question, and that loss of life and property damage, even in a "catastrophe," must be limited to the immediate neighborhood of the plant. This newer assessment foresees no mechanism by which more than some 100 lives could ever be lost, and even that extreme case has insignificantly small probabilities.