The LNT curve is a bit confusing. I thought LNT was supposed to be a linear fit to the data at large doses. I'm looking for a possible replacement for the radon example in our Fear of Radiation article. https://citizendium.org/wiki/Fear_of_radiation
LNT cancer incidence is linear in total dose, but in Figure 1, the x-axis (dose) is logarithmic which transformation bends a straight line into a curve. LNT is also a power law with an exponent of 1.0 and thus LNT shows up as a straight line in log/log space (Figure 2). SNT becomes a power law at low dose but the exponent at the low end for the current fit is 2.18, a bit more than quadratic.
It isn't the logarithmic curvature that is confusing, but the fact that the LNT curve doesn't pass through the high-dose data points. Maybe it was a fit to some other data. The bomb survivor data shows a good fit for LNT above 100mSv. https://citizendium.org/wiki/Fear_of_radiation/Debate_Guide#LNT_Controversy
LNT (like SNT) bills itself as a model that works for ANY radiation exposure. But the BEIR7 fit is based mainly on the bomb survivor (RERF) data which is a single acute dose. It is not a good fit to the RERF data. No smooth curve could be since the RERF data bounces up and down in a very weird way., which should tell you something about the quality of that database.
But the point is that LNT it is off by orders of magnitude when you try to apply it to chronic doses received over protracted periods such as the dial painter data. A regulatory model cant change every time you get a new exposure data. We must have ONE model that does a reasonable job both on large acute doses and large chronic doses. LNT cant do that. SNT can.
OK, that explains it. I was missing the fact that LNT assumes the cumulative dose is the same as one massive exposure. Sorry for being so slow. Minor point: can we use just one unit, Sieverts, and avoid some confusion among our non-technical audience?
In a NPP power plant release, essentially all the harm to the public if any is from photons and electrons for which the ICRP has decreed the RBE = 1.0. So numerically grays = sieverts. So in this situation there is no need for two units.
But your first point is far more fundamental. LNT is the theory that the only thing that counts is total dose. Whether it's received in a few minutes or spread over a few decades makes no difference. Linearity follows from this assumption. Any one who focuses on total dose however adjusted is an LNTer. The focus must be on the dose rate profile. That's what everybody, technical or non-technical needs to understand.
The other point that must be front and center is the difference between damage and harm. Harm in this context should be reserved for unrepaired damage. In this situation, the repair process is more important than the damage, which is going on all the time. We must direct attention to the repair process and the time it takes. That's the key to understanding radiation harm.
You are highlighting bone-cancer-incident rates vs cumulative dose. Assuming an evenly distributed dose rate over 10 years is a good back-of-the-envelope assumption, and probably nothing better can be estimated today.
Where do you have the bone-cancer-incident rates from?
I always thought that tongue and lip cancer was the main problem with the dial painters?
Do you have such data?
What about other types of cancer, or are there any mortality rates known that could be used?
Painting dial faces was continued over many decades, probably till the end of the 50ies or even into the 60ies in some places. There may be more and more recent data available.
I am afraid that focusing on just one cancer type may not be enough to make a conclusive statement that would be accepted by all.
Continue your good work, and I am still standing that a) SNT is way better than LNT by orders of magnitude, but b) still not good enough.
The bone cancer data is from Evans, Radium in Man, Health Physics, 1974, Vol 27, Figure 3, redrawn.
The dial painters were studied in great detail and at great expense to the taxpayer from the 1960's all the way thru the mid 1990s. Here's a quote from Rowland, 1996
This study has demonstrated that, when the radium tumor deaths are removed, the average survival of the dial worker population is indistinguishable from estimates of the survival of contemporary white females of the same age. This is a remarkable result, for it implies that, to the precision obtainable with a population of some 1000 persons, the life expectancy of the remaining population was unaffected by the radium burden.
END QUOTE
Frankly, I've had it with people saying SNT is far better than LNT, but it is not good enough, without offering a real alternative. In other words, we should stick with LNT.
It is incumbent on the SNT-is-not-good-enoughers to come up with a better model, where model means, not handwaving about hormesis, but a computer program which like LNT and SNT can convert a radioactive plume into Lost Life Expectancy for the purposes of regulation and compensation.
As multiple posts on this substack have argued, the policy implications of SNT are night-and-day different from LNT. Under SNT, nuclear power's hazards become no worse than the hazards associated with hydro or chemical plants, and can be regulated in the same manner. That's all we need.
The reference you give and the quote from it are both good and remarkable.
The problem I have with SNT is that it has the same disease as LNT regarding the idea of radiation being always harmful, down to the lowest levels - just with SNT, this is much, much less than what LNT does.
Radiation hormesis at low dose rates has been discussed in many peer reviewed papers, which at least should lead to a cut-off dose (dose-rate), below that SNT should be nulled. You could label that SNT-0
The point I am making is to clearly make a point about at which level to stop worrying about radiation at all.
A further improvement would take radiation hormesis into account at low dose (dose-rate). I know that here the discussion would also be strong - but that should be part of Gordian knot.
To my knowledge there isn't statistically significant evidence for hormesis in large population exposures. The low dose cohort from the nuclear bomb survivors experienced an insignificant decrease in cancers
I believe that hormesis exists but the evidence that would make it useable for regulation does not. The regulatory guide must be conservative to have any real chance of being adopted. Once you have SNT, improvements on it will be marginal to the actual economic effect of reasonable regulation based on it. Any nuclear plant designer would likely prefer to design to SNT as is rather than incorporate hormesis and risk being re-regulated by a future government that disagrees
In this view, rather than improving SNT in some way, it would actually be far more consequential to create a cost benefit guide to how a regulatory body applies SNT, which will likely be an order of magnitude or more than is reasonable. For instance a cost benefit review agency could be given authority over the NRC. The NRC tells Georgia Power to redo their rebar at a cost of a billion, Georgia Power appeals to the oversight agency, the oversight agency overrules
Underwriter's Certification would balance this much more efficiently with a market system rationally balancing the costs and benefits but an oversight agency might be a lot easier to implement at least as a first step
The LNT curve is a bit confusing. I thought LNT was supposed to be a linear fit to the data at large doses. I'm looking for a possible replacement for the radon example in our Fear of Radiation article. https://citizendium.org/wiki/Fear_of_radiation
LNT cancer incidence is linear in total dose, but in Figure 1, the x-axis (dose) is logarithmic which transformation bends a straight line into a curve. LNT is also a power law with an exponent of 1.0 and thus LNT shows up as a straight line in log/log space (Figure 2). SNT becomes a power law at low dose but the exponent at the low end for the current fit is 2.18, a bit more than quadratic.
It isn't the logarithmic curvature that is confusing, but the fact that the LNT curve doesn't pass through the high-dose data points. Maybe it was a fit to some other data. The bomb survivor data shows a good fit for LNT above 100mSv. https://citizendium.org/wiki/Fear_of_radiation/Debate_Guide#LNT_Controversy
LNT (like SNT) bills itself as a model that works for ANY radiation exposure. But the BEIR7 fit is based mainly on the bomb survivor (RERF) data which is a single acute dose. It is not a good fit to the RERF data. No smooth curve could be since the RERF data bounces up and down in a very weird way., which should tell you something about the quality of that database.
But the point is that LNT it is off by orders of magnitude when you try to apply it to chronic doses received over protracted periods such as the dial painter data. A regulatory model cant change every time you get a new exposure data. We must have ONE model that does a reasonable job both on large acute doses and large chronic doses. LNT cant do that. SNT can.
OK, that explains it. I was missing the fact that LNT assumes the cumulative dose is the same as one massive exposure. Sorry for being so slow. Minor point: can we use just one unit, Sieverts, and avoid some confusion among our non-technical audience?
In a NPP power plant release, essentially all the harm to the public if any is from photons and electrons for which the ICRP has decreed the RBE = 1.0. So numerically grays = sieverts. So in this situation there is no need for two units.
But your first point is far more fundamental. LNT is the theory that the only thing that counts is total dose. Whether it's received in a few minutes or spread over a few decades makes no difference. Linearity follows from this assumption. Any one who focuses on total dose however adjusted is an LNTer. The focus must be on the dose rate profile. That's what everybody, technical or non-technical needs to understand.
The other point that must be front and center is the difference between damage and harm. Harm in this context should be reserved for unrepaired damage. In this situation, the repair process is more important than the damage, which is going on all the time. We must direct attention to the repair process and the time it takes. That's the key to understanding radiation harm.
Thanks Jack, for this insightful summary.
You are highlighting bone-cancer-incident rates vs cumulative dose. Assuming an evenly distributed dose rate over 10 years is a good back-of-the-envelope assumption, and probably nothing better can be estimated today.
Where do you have the bone-cancer-incident rates from?
I always thought that tongue and lip cancer was the main problem with the dial painters?
Do you have such data?
What about other types of cancer, or are there any mortality rates known that could be used?
Painting dial faces was continued over many decades, probably till the end of the 50ies or even into the 60ies in some places. There may be more and more recent data available.
I am afraid that focusing on just one cancer type may not be enough to make a conclusive statement that would be accepted by all.
Continue your good work, and I am still standing that a) SNT is way better than LNT by orders of magnitude, but b) still not good enough.
Hans Peter,
The bone cancer data is from Evans, Radium in Man, Health Physics, 1974, Vol 27, Figure 3, redrawn.
The dial painters were studied in great detail and at great expense to the taxpayer from the 1960's all the way thru the mid 1990s. Here's a quote from Rowland, 1996
https://osti.gov/servlets/purl/371427
BEGIN QUOTE
This study has demonstrated that, when the radium tumor deaths are removed, the average survival of the dial worker population is indistinguishable from estimates of the survival of contemporary white females of the same age. This is a remarkable result, for it implies that, to the precision obtainable with a population of some 1000 persons, the life expectancy of the remaining population was unaffected by the radium burden.
END QUOTE
Frankly, I've had it with people saying SNT is far better than LNT, but it is not good enough, without offering a real alternative. In other words, we should stick with LNT.
It is incumbent on the SNT-is-not-good-enoughers to come up with a better model, where model means, not handwaving about hormesis, but a computer program which like LNT and SNT can convert a radioactive plume into Lost Life Expectancy for the purposes of regulation and compensation.
As multiple posts on this substack have argued, the policy implications of SNT are night-and-day different from LNT. Under SNT, nuclear power's hazards become no worse than the hazards associated with hydro or chemical plants, and can be regulated in the same manner. That's all we need.
Thanks a lot Jack.
The reference you give and the quote from it are both good and remarkable.
The problem I have with SNT is that it has the same disease as LNT regarding the idea of radiation being always harmful, down to the lowest levels - just with SNT, this is much, much less than what LNT does.
Radiation hormesis at low dose rates has been discussed in many peer reviewed papers, which at least should lead to a cut-off dose (dose-rate), below that SNT should be nulled. You could label that SNT-0
The point I am making is to clearly make a point about at which level to stop worrying about radiation at all.
A further improvement would take radiation hormesis into account at low dose (dose-rate). I know that here the discussion would also be strong - but that should be part of Gordian knot.
To my knowledge there isn't statistically significant evidence for hormesis in large population exposures. The low dose cohort from the nuclear bomb survivors experienced an insignificant decrease in cancers
I believe that hormesis exists but the evidence that would make it useable for regulation does not. The regulatory guide must be conservative to have any real chance of being adopted. Once you have SNT, improvements on it will be marginal to the actual economic effect of reasonable regulation based on it. Any nuclear plant designer would likely prefer to design to SNT as is rather than incorporate hormesis and risk being re-regulated by a future government that disagrees
In this view, rather than improving SNT in some way, it would actually be far more consequential to create a cost benefit guide to how a regulatory body applies SNT, which will likely be an order of magnitude or more than is reasonable. For instance a cost benefit review agency could be given authority over the NRC. The NRC tells Georgia Power to redo their rebar at a cost of a billion, Georgia Power appeals to the oversight agency, the oversight agency overrules
Underwriter's Certification would balance this much more efficiently with a market system rationally balancing the costs and benefits but an oversight agency might be a lot easier to implement at least as a first step
Go SNT!