THERE is no indication that healthy people taking a diet rich in carbohydrates are especially liable to diabetes ; in fact numerous observations show improvement of carbohydrate tolerance following its greater intake. The Staub-Traugott effect is a classical example of this in acute experiments. As a long-term effect diabetes mellitus is not especially common among the huge and mainly carbohydrate-eating populations of the world-e.g., the Chinese-except the rich and the sedentary among them who partake of large quantities of fat as well and encourage obesity by overeating. Although it is theoretically possible that in such circumstances the islets of Langerhans can be exhausted, there is no experimental proof that this happens. Allan (1923) showed that the relationship between the carbohydrate ingested and the insulin necessary to metabolise it was approximately logarithmic ; hence the strain on the insulin mechanism is relatively small. Moreover it is impossible to improve to any great extent nervous hypoglycsemia by treatment for months with a diet containing a very large quantity of carbohydrates. Even diabetics tolerate a high-carbohydrate diet.