THE form in which carbohydrate circulates in the body is glucose: it is frequently of clinical importance to determine the amount of this substance in the blood, but estimation of blood-sugar may not be synonymous with determination of blood-glucose. Different methods of estimation give somewhat different results, but this is of little importance clinically, provided the same method is always used. It is, however, of some interest to inquire into the causes of these discrepancies and a certain amount of work has been recently devoted to this subject. After fermentation with yeast, blood still gives a residual reduction with oxidising agents, which is obviously not due to glucose: according to I. M. Rabinowitch (Biochem. Jour., vol. 22, p. 753; 1928) the amount of the non-fermentable reducing substances present in normal or diabetic human blood is about 0.025 per cent expressed as glucose. It is very constant in the same individual, is not affected by insulin or by the administration of glucose by mouth, and is the same in venous as in arterial blood: in all these respects it shows a marked contrast with glucose.