Last word on LNT

LNT is not only nonsense; it's boring. I've been beating this dead horse for two years. I'm sure the choir is as tired of this sermon as I am. I promise this is my last word on LNT.

I've been misleading the choir. I show people the radium dial painter cancer data, and the Karunagappally cancer data, Figure 1, and point out that cancer incidence does not increase linearly in total dose as LNT predicts. Not even close.

Figure 1 LNT versus reality. SNT stays within the confidence limits up to about 2 mSv/d.

One problem with this kind of graph is that it indicates that LNT and SNT converge as we approach zero dose. In fact, in relative terms they are diverging quite rapidly, as the log-log inset shows. At 0.01 mSv, the difference in cancer incidence is a factor of 50,000. This is critical when we start combining extremely large populations with tiny doses as we did at Chernobyl. In fact, it forces even defenders of LNT, such as the WHO, to say do NOT use LNT in this situation. So we are using a harm model to predict the impact of a release which we cannot use in a release.

But the big problem with this kind of figure is we are looking at the same dose rate pattern. For all the thousands of subjects in Figure 1, the dose was received more or less evenly over ten years, or at least that is what we are assuming. The only difference between the left end and the right end of Figure 1 is the total dose.

This is a nearly universal myopia. Either we compare a single dose and focus on the differing totals. Or as in Figure 1, we compare evenly received dose patterns and focus on the total dose. That's why we can show cumulative dose on the bottom axis of Figure 1. But cumulative dose does not determine the harm.

If we really want to see why LNT falls apart, we must compare different dose rate profiles with the same total dose. Figure 2 shows two profiles both with a total dose of 600 mSv. In one case, the 600 mSv was received all at once, as occurred in the atom bomb blasts. In the other, the 600 mSv was received evenly over 10 years, as in the Karunagappally cohort. Unfortunately, the numbers are such that I have to plot this on a log-log figure. Otherwise the dose profiles would be indistinguishable from the axes.

Figure 2. Two 600 mSv dose profiles. Red: dose received in one repair period. Blue: dose received evenly over 3650 repair periods.

For LNT, these two profiles, so different that I cannot show them together on a normal figure, are the same. They both result in an increase in cancer incidence of 5.7%. For SNT, the increase in cancer incidence from the red profile is 10.4%. The increase in cancer incidence from the blue profile is 0.0017%, over 3000 times smaller. The observed increase for the red profile at Hiroshima and Nagasaki is about 5%. The observed increase for the blue profile in Kerala is about -5%.

The dial painter data is even more damning. At 100,000 mSv, the dial painters showed zero bone cancer incidence. LNT says we should have seen an impossible 10,000% incidence. SNT does a pretty good job up to about 2 mSv/day; but prematurely jumps up to nearly 100% at about 7 mSv/day. The actual data does not make the jump until about 20 mSv/day, and never gets above 40%.

LNT is qualitatively wrong because it claims dose rate is irrelevant. In fact, the probability that the DNA damage is repaired is critically dependent on dose rate. Our repair systems are remarkably accurate in repairing isolated damage. But the repair takes some time. If the dose rate is high enough, the repair system can't keep up with the production of Double Strand Breaks (DSB) in the DNA helix. The inventory of unrepaired double strand breaks builds up. The probability of two unrepaired DSB's close together skyrockets. Closely spaced DSB's are the one thing, our repair systems have difficulty with. The probability of unrepaired damage and cancer skyrockets as well.

SNT recognizes our repair ability and its limitations. It gets the shape of the curve right. But quantitatively it ends up over-predicting the harm once dose rates get above 2 mSv per day. For people who want to make sure that our harm model dose not under-estimate risk, this should be comforting. For the rest of us, it is a small price to pay for getting rid of LNT.