Ever since the recognition of diabetes mellitus as a disease entity, considerations of interrelations of the pancreas with liver metabolism have occupied a prominent role. In Minkowski's untreated depancreatized dogs, 1-2 and in other experimental diabetic preparations, 3'4 it was well substantiated that in the liver the glycogen concentration rapidly became low, the fat content slowly increased, 5'6> 7-8 and ketone production increased. 9'10 Early in the history of diabetes mellitus, it was therefore recognized that a deranged liver metabolism was consistently associated in some manner with the symptoms of glycosuria and hyperglycemia. While subsequent experimentation has revealed many of the correlations between hepatic function and regulation of blood glucose, the question whether impaired liver metabolism is a direct result of the lack of insulin remains open. While several investigators11'12> 13 have claimed a direct action of insulin upon glucose production by the liver, many others14'15> 16 have failed to find any such effect. The present paper will attempt to examine critically this large body of evidence. The earlier literature has been reviewed extensively in the monograph by de Duve. 17 The general viewpoint that will be formulated is based upon the assumption that insulin does not have a direct action upon the isolated liver, but that insulin affects hepatic metabolism only indirectly. While it is believed that the available evidence is not in conflict with such a conclusion, it is admitted that the opposite viewpoint cannot be wholly rejected at the present time.

In Minkowski's laboratory in the period between 1890 and 1900 attempts were made to delineate the role of the liver in diabetes by hepatectomy experiments. Kausch18 demonstrated that hepatectomy led to hypoglycemia both in normal birds and in previously depancreatized birds. However, the decisive experiments came