http://science.sciencemag.org/content/202/4372/1045
DOI: 10.1126/science.202.4372.1045
First look at tables 2 and 3. The raw radionuclide release from the reactors, measured in curies, are far greater than those from the coal plant. Both reactor types release thousands of curies of xenon isotopes and hundreds of curies of krypton isotopes. The PWR also releases thousands of curies in the form of tritium. The coal plant releases less than 2 curies of all radionuclides combined.
Next look at table 4. The maximum whole body dose commitment, e.g. the dose that somebody might be exposed to if they lived 500 meters from the sources, just beyond the plant boundary, is highest for the boiling water reactor (4.6 mrem/year), then coal (1.9), then the pressurized water reactor (1.8).
But if you look at table 5, average population dose within an 88.5 km radius, you see that the population dose commitment over the whole region is lower for both pressurized and boiling water reactors (given the assumption that 100% of the food people eat is grown in the same region).
Table 6, "Population dose commitments from the airborne releases of model 1000-MWe power plants as a function of food intake", is the most interesting table in the article. It shows what parameter is most key to which power source produces greater population exposures: percentage of food eaten that is grown within the same region as the population. At 50% or more local food consumption, which is assumed in table 5, coal always exposes the regional population to a higher radionuclide dose than nuclear reactors. If people were eating 30% or less locally grown food -- which admittedly does not seem likely -- then the region's population could be exposed to more radiation from a nuclear plant than from an equivalent coal plant.
Why do nuclear reactors initially emit so much more than coal plants, measured in curies of radionuclides, yet generally expose human populations in the region to less of a body burden? Why does the relative population exposure ordering of nuclear power and coal change depending on how much locally grown food those populations consume?
The answers lie in the chemical and biological behaviors of the different radionuclides that respectively dominate emissions from reactors and from coal plants. "Radium-226 and radium-224 are the major contributors to the whole-body and most organ doses from the coal-fired plant. Assuming that the deposited radionuclides could enter the food chain, ingestion is the main exposure pathway for the population dose commitments from this plant (93 to 96 percent for the whole body and most organ doses, 83 percent for the bone dose, and 62 percent for the lung dose). ... Carbon-14 is the main contributor to the whole-body and most of the organ doses from both nuclear plants. Ingestion is the major exposure pathway."
Radium is chemically similar to calcium, so it is taken up by plants via the same pathways that take up calcium. It gets stored in the bones of exposed humans. Carbon of course is a major part of human and plant dry mass and also gets stored in organisms. These key isotopes are chemically available to plants and processed as nutrients in the bodies of both plants and humans.
Those thousands of curies of noble gases initially released by the nuclear plant? They matter orders of magnitude less when it comes to human exposure. Those gases can't chemically react with anything and diffuse throughout the whole volume of the atmosphere. Plants don't concentrate them and the human body can't store them.
The much less significant exposure route, "immersion," basically means exposure to radionuclides in the air all around you. Under most assumptions the immersion route will deliver a lower exposure to a regional population than ingestion via food. But if people in the exposed region are eating 30% or less locally-grown food, then the immersion route can become dominant and can lead to higher population exposure from nuclear power than from coal power.
The Scientific American article does not touch on any of these interesting points. It paraphrases the original in a way that actually introduces mistakes. It does not explain why some reasonable assumptions lead to lower population radionuclide exposure from nuclear power than from coal power. The broad summaries remain similar but the mistakes and simplifications lend the Scientific American article an unfortunate air of "here's a nice simple conclusion to bash your friends with the next time they fret about radiation from nuclear power."