Hepatic glucose output can be readily estimated at different stages of fasting by accepted procedures involving primed continuous infusion of [6-3H] glucose to steady state. In the longterm fasted state in which liver glycogen is essentially depleted, there is no difficulty in estimating gluconeogenesis, since this is equal to the endogenous glucose output. However, with shorter fasting periods, hepatic glucose output will be derived both from glycogenolysis and gluconeogenesis. Based on earlier studies in vitro with 3HHO [1], I describe an approach involving tritium incorporation onto C-6 of glucose in vivo to estimate relative contributions of gluconeogenesis and glycogenolysis to hepatic glucose output in the fasted state. In previous experiments with isolated hepatocytes, in a medium containing 3HHO, tritium was incorporated extensively on to all six carbons of the glucose formed, when L-lactate, pyruvate or L-glutamine was the gluconeogenic substrate employed [1]. Tritium incorporation on the two positions on C-6 of glucose was nearly 900% of the specific activity of the medium 3HHO. However, when glucose was formed in a 3HHO medium from glyco-genolysis, under anaerobic conditionsto prevent any gluconeogenesis, there was negligible tritium incorporation on C-6 of glucose, although extensive incorporation on to C-2 was found, as expected [2]. During gluconeogenesis from L-lactate, tritium was also extensively incorporated onto carbons 2, 3, 4 and 5, and this was true also when gluconeogenesis occurred from substrates that entered the gluconeogenic pathway at the triose-phosphate level, substrates such as fructose and dihydroxyacetone. When glucose was formed from fructose in hepatocytes from fasted rats, the tritium specific radioactivity on each hydrogen of C-6 of glucose was about 8% of that of the medium 3HHO [1], and this incorporation was reduced to about2% when thephosphoenolpyruvate carboxykinase inhibitor, mercaptopicolinate, was added (my unpublished results). Thus lactate formation from fructose and a degree of gluconeogenesis from this lactate may have produced much of this C-6 labelling when fructose was the gluconeogenic substrate. It seems likely, therefore, that tritium incorporation on C-6 is largely confinedto those gluconeogenic substrates that enter the pathway at the pyruvate level, or at the level of a Krebs cycle intermediate. Glycerol produced from lipolysis will also contribute carbon for gluconeogenesis, al-though this source has been estimated to beless than 10% of the total hepatic glucoseproduction [3, 4]. As glycerol enters the gluconeogenic pathway at the triose-phosphate level, we assume negligible tritium incorporation on C-6 of glucose from this source in vivo. Glycerol gluconeogenesis must be independently determined from steady state infusion of 14C-labelled glycerol [3, 4].

The difference between net hepatic glucose outputand glycerol gluconeogenesis can bepartitioned between glycogenolytic glu-cose production and non-glycerol gluconeogenesis. I am neglecting pentose cycle flux, which should be a small net outflow of carbon in the fasted, non-lipogenic and gluconeogenic state, where the stoichiometry of the cycle is 1 glucose 6-phosphate-. 6 CO2 [5]. To determine the extent of tritium incorporation from