Cancer epidemiologists then tried to establish what proportion of cancers these might be. Higginson suggested 70 to 80 percent of all cancers could be prevented; others said as many as 90 percent. In 1981, the Oxford epidemiologists Richard Doll and Richard Peto published the seminal work on this subject: a 120-page analysis in the Journal of the National Cancer Institute that reviewed the existing evidence on changes in cancer incidence over time, changes upon migration from one region of the world to another, and differences in cancer rates between communities and nations. (Colon cancer, for example, was ten times more common in rural Connecticut than in Nigeria; breast cancer was diagnosed eight times more often in British Columbia than in the non-Jewish population of Israel.) Based on this evidence, Doll and Peto concluded that at least 75 to 80 percent of cancers in the United States might be avoidable with appropriate changes in diet and lifestyle.

In the quarter-century since Doll and Peto published their analysis, it has been cited in nearly two thousand journal articles, and yet the fundamental implications have been largely lost. The two most important conclusions in their analysis were that man-made chemicals—in pollution, food additives, and occupational exposure—play a minimal role in human cancers, and that diet played the largest role—causing 35 percent of all cancers, though the uncertainties were considered so vast that the number could be as low as 10 percent or as high as 70 percent.

Higginson had repeatedly remarked on these two points during his tenure as director of IARC. In early reports, Higginson and the World Health Organization had referred to “extrinsic factors” and “environmental factors” as the cause of most cancers, by which they meant lifestyle and diet. The public and the environmental movement had perceived this to mean almost exclusively “man-made chemicals”—the “carcinogenic soup,” as it was known in the 1960s and 1970s. “It appears that only a very small part of the total cancer burden can be directly related to industrialization,” Higginson wrote. The release of industrial chemicals into the environment could not explain, for example, why the nonindustrial city of Geneva had more cancer than Birmingham, “in the polluted central valleys of England,” or why prostate cancer was ten times more frequent in Sweden than in Japan.*61

Nonetheless, this focus on carcinogenic chemicals as the primary cancer-causing agents in the environment also carried over to nutrition-related cancer research in the laboratory. It was assumed that whatever components of diet were responsible for cancer worked the same way that chemicals did: by inducing mutations and genetic damage in cells. When cancer researchers from around the world met in September 1976 at the Cold Spring Harbor Laboratory to discuss the origins of human cancer, the talks focused on those chemicals shown to be carcinogens in animals, and the possibility that they might be found in infinitesimal or greater amounts in human diets, drinking water, or pharmaceuticals.