The Problem. —Carbohydrate tolerance may be studied in terms of either glycosuria or glycemia. Testing the urine for sugar is still of value for physicians who have small opportunity to take blood, as in insurance work, or to have the blood analyzed, owing to remoteness from a laboratory. Blood sugar determinations, on the other hand, have proved more illuminating, especially in diagnosis, and, therefore, in this paper will be considered with little regard for the urine findings or the blood sugar threshold. Standards should be founded not only on careful examination of a few persons but also on statistical analysis of many persons. Series of more than ten normals seem to have been reported from only six clinics, and of these only two have been of satisfactory statistical size, namely, fifty-three described by Goto and Kuno1and fifty-eight by Cummings and Piness.2 The opinions resulting from such limited evidence disagree