In this case, the historical method was used. Researchers went to the field and studied a special worker population which had high lung cancer incidence. (Supplemental note: in the actual study, the lung tumors were cut open and examined under an electron microscope. The inspection revealed asbestos fibers at the core of the tumors, thus confirming that the fibers induced the cancerous growths.) The statement shows that the lung cancer incidence was traced back to the workers' high industrial exposures to cancer, thus validating the hypothesis that asbestos caused cancer to these WW2 shipyard workers. Because the statement says "studies" rather than "study," we must conclude that more than one study replicated the claimed results, thus giving us a high degree of confidence in the reliability of this knowledge. This appears to be solid science and could serve as a basis to mandate workplace safety standards that greatly reduce workers' exposure to asbestos.

The "...this means that..." statement which follows is an extrapolation of the data yielded by shipyard workers. This latter statement remains untested. What the extrapolation produces is a hypothesis: "67,000 deaths will result to homeowners and office workers annually as result of asbestos exposure." The hypothesis is testable, by the historical method, against its converse: "67,000 deaths will not result to homeowners and office workers annually as result of asbestos exposure." To test these competing hypotheses, we have to study the general populace (homeowners and office workers) with casual exposures rather than a special populace, like shipyard workers, with industrial exposures. Gathering the required data will take a lot of work to build the necessary data base, but given the serious nature of the hypothesis, this work is certainly merited. If we find from our study that substantial numbers of deaths in the general populace are caused by asbestos, we have confirmed our hypothesis and we can use this as a firm basis to begin to regulate asbestos and begin to remove it from our normal home and work environments. However, if we cannot discover substantial asbestos-related deaths within the general populace, our knowledge base supports the converse hypothesis and refutes the extrapolation. We have no scientific basis on which to impose sweeping mandates that affect the general populace. Just as it is risky to extrapolate laboratory data obtained on cells and animals to estimate what happens to a general population of humans, it is likewise risky to take one study done on a special population through the historical method , and extrapolate what that means for a more general population.

You might be thinking "Why do such a study? If the substance is dangerous in high industrial exposures, it may well be dangerous to anyone. So why not pass mandates to protect the entire populace right now?" There are a couple of sobering considerations. The first is that if you used that thinking as a basis for action, your cotton sweatshirt would today be classified as a hazardous substance. Lung disease caused by inhalation of excessive amounts of cotton fiber is well documented from studies done when early cotton mills had no standards for ventilation and air filtration for workers. Damage caused by extreme industrial exposures are not safely extrapolated as a basis to evaluate danger posed to the general populace. The second consideration is cost-benefit. In the early 1990s, the money spent to abate asbestos was about equal to the entire budget of the National Institutes of Health. One must ask whether better health benefits and more lives saved will result from incurring the costs of widespread asbestos abatement, or whether investing the same amount of funds to address other health problems will yield higher benefits.