George Cahill, a former professor at the Harvard Medical School, is a pedagogical example. Cahill had done some of the earliest research on the regulation of fat-cell metabolism by insulin in the late 1950s, and had coedited the 1965 Handbook of Physiology on adipose-tissue metabolism. In 1971, when Cahill gave the Banting Memorial Lecture at the annual meeting of the American Diabetes Association, he described insulin as “the overall fuel control in mammals.” “The concentration of circulating insulin,” he explained, “serves to coordinate fuel storage and fuel mobilization into and out of the various depots with the needs of the organism, and with the availability or lack of availability of fuel in the environment.” When I interviewed Cahill in 2005, he told me it was true that “carbohydrate is driving insulin is driving fat.” But Cahill did not consider this chain of cause and effect to be a sufficient reason to speculate that carbohydrates drive obesity. Nor did he consider it a possibility that avoiding carbohydrates might reverse the process. Rather, he believed unconditionally that positive caloric balance was the critical factor. When it came to weight regulation, Cahill repeatedly told me, “a calorie is a calorie is a calorie.” He acknowledged that the obese ate no more, on average, than the lean, and this is why he believed that the obese must be fundamentally lazy and this was the proximate cause of their obesity.*121 There was no reason to test competing hypotheses, Cahill said, because any competing hypothesis would contradict the laws of physics as he understood them.

When clinical investigators tried to unravel the connection between diet, insulin, and obesity in human subjects, as the University of Washington endocrinologist David Kipnis did in the early 1970s, the results were invariably analyzed in light of this same preconception. Kipnis had fed ten “grossly obese” women a series of three-and four-week diets that were either high or low in calories, and high or low in carbohydrates. The fat-rich diets lowered insulin levels, Kipnis reported in The New England Journal of Medicine in 1971, and the carbohydrate-rich diets raised them, regardless of how many calories were being consumed. Even when these women were semi-starved on fifteen hundred calories a day, a high-carbohydrate content (72 percent carbohydrates and only 1 percent fat) still increased their insulin levels, even compared with the hyperinsulinemia of these obese women on their normal diets.

One interpretation of these results is that we could remove the carbohydrates from the diet and replace them with fat, and weight would be lost, perhaps without hunger, because insulin levels would drop, even if the total calories consumed did not. Kipnis’s results, as the University of Heidelberg clinicians Gotthard Schettler and Guenter Schlierf wrote in 1974, underlined the “necessity of restricting carbohydrates in obesity in order to restore insulin levels to normal, thus hopefully decreasing appetite and fat deposition….”