The "Linear-No-Threshold (LNT)-Hypothesis" assumes carcinogenic effects of radiation to be strictly proportional to dose without threshold. Three arguments are commonly used as justification:

Radiation hits to cells occur stochastically.

Cancer develops monoclonally, that is, from a single transformed cell.

The repair of DNA-lesions is prone to error.

These arguments seem to be correct. The stochastic character of radiation hits undoubtedly makes the number of hit cells proportional to the dose (at low doses only a small number of cells are hit), monoclonality is very probably true, and error-proneness is definitively true. But logical thinking forbids us to conclude "linearity" from this.

The logical problem thereby is similar to the following example: The observation: "Dogs are larger than mice" is certainly true, but we must not conclude therefrom that "dogs are the largest animals". This conclusion is flawed, because it ignores additional possibilities, namely the possibility that larger animals exist.

Principally the same holds for radiation and cancer. When concluding "linearity", two effects are erroneously ignored:

The existence of other sources of cancer and the possible interaction between radiation and the number of cancers caused by these sources.

The possibility of interactions between the effects of radiation hits to individual cells.

Cancer is the final consequence of malignant transformations in DNA-molecules of individual cells. Even without radiation, numerous injuries to DNA-molecules occur. Most of them are successfully repaired by very efficient repair systems. Without that repair, we could not survive. Only if the repair systems fail, can cancer develop.

DNA-lesions are -- amongst a multitude of different causes -- generated by ionizing radiation. But radiation causes more: It also stimulates the defensive response of cells and tissues to carcinogenic challenges. Numerous experiments show a reduced number of cancers, if cells or whole animals are exposed to low doses of radiation prior to being exposed to high doses of carcinogens. This "adaptive response" is a fact; it's principal existence is proven, only its extent can be disputed. It can reduce the number of cancers, of all cancers, but most important in this discussion, the number of those cancers that originate from other sources. Therefore, while radiation produces cancers, it also reduces cancers. The overall effect is just the balance between these two partial effects. This balance could well be an overall reduction of cancer frequency. The laws of logic force us to accept this as a possibility, and many experiments show us that it really occurs, at least under appropriate conditions. Thus, the existence of other cancer sources and the possibility of an adaptive response forbid the conclusion of "linearity" from the above arguments.

Furthermore, even when radiation hits to cells occur independently, the effects of these hits need not to be independent likewise, i.e. they can interact with each other. But ever when such interactions occur, the effect of n hits cannot be the n^th multiple of the effect of one hit (without interaction). This is non-linearity! Such interactions could be generated, for example, by transfer of the adaptive response from a hit cell to neighboring cells by transmitters, creating "stimulated regions" which overlap each other depending on dose, or when several hit cells trigger a response of the tissue or whole body (e. g. stimulation of the immune system). Any such interaction excludes linearity.

To judge the validity of the pro-arguments, it is sufficient to know that interactions of radiation with other cancer sources, or between radiation effects, can exist. Whenever they can, the conclusion of "linearity" is not allowed. For all practical reasons, however, it is very important whether these interactions really exist, and how strong they are. This indicates where additional research is required.

Besides the above two forms of possible interactions, additional arguments against the LNT-hypothesis include:

human epidemiological studies showing a threshold

animal experiments with different species showing a threshold for both, external and internal exposure

biological and theoretical models of cancer generation explaining or at least tolerating non-linearity and a threshold

general reflections and analogies to other evolutionary adaptations of life to hostile environments, which generally show a threshold (after all, life developed under irradiation at substantially higher levels than today, and it is generally adapted to an optimum of well-being or health at "prevalent" environmental conditions and is equipped with reliable defence mechanisms against frequent deviations).