1968). Alternatively, it may represent the diurnal variation in fasting blood sugar levels in established diabetics shown by Faiman and Moorhouse (1967). These workers reported that during three-day fasts five diabetic subjects showed a diurnal cycle in their blood glucose levels, with the peak levels occurring around 8 a.m.; no such cycle could be discerned in five healthy controls.The inverse relation of obesity, as measured by the ponderal index, with the degree of diurnal variation in glucose tolerance, was significant only in the normoglycaemic group, but was also evident in the hyperglycaemic group. Since the morning glucose tolerance area is itself associated directly with the degree of obesity, it is difficult to decide how much of the loss of diurnal variation to attribute to obesity per se and how much to the associated hyperglycaemia. The analyses of the control group in Tables II and III, however, suggest that obesity is the more potent factor when morning glycaemia is within the normal range and is perhaps an independent one. Diagnostic classification of the tests yields many abnormalities by afternoon testing in the group preselected for normoglycaemia. By one system of classification (Table VIII) all the morning tests were normal, but 10 afternoon tests were abnormal. Table IX lists the paired tests in these 10 subjects, and in all but one (No. 2) the morning tests are unequivocally normal. In some, however, the diurnal swing is so greatfor example, Nos. 6, 7, and 9-as to bring the test results into what most would regard as the conventionally diabetic range. The present and potential significance of this phenomenon can only be hypothetical. This exaggerated swing may possibly represent an earler stage in the failure of glucose homoeostasis than the reduced swing seen in those with morning hyperglycaemia. Obviously, only longitudinal studies can provide the answer, and these are fraught with difficulties, not least logistical ones. A suggested working hypothesis, however, is that afternoon hyperglycaemia is the first stage of glucose intolerance, followed by increased morning glycacmia (and reduced diurnal variation), then proceeding to morning hyperglycaemia, sufficient eventually to warrant the description of diabetes mellitus.The small diurnal difference in mean two-hour blood sugar levels measured under screening conditions agrees with our previous experience (Keen, 1966) and presumably accounts for the failure of others to find diurnal variation in population surveys (Hayner et al., 1965;Marine et al., 1969). The reason for the relative lack of diurnal variation under these conditions is probably related to the lesser degree of standardization obtaining. Though the mean blood sugar levels were similar, the variation was very much greater in the afternoon, so that afternoon screening would be likely to yield more subjects exceeding an arbitrary cut-off level.We are grateful to the many men who volunteered to participate in this study, and to our colleagues at the London School of Hygiene for t...