During 75 cardiopulmonary bypass experiments in dogs, a fall in arterial pressure or a tendency for the animal to take up blood from the extracorporeal circuit was noted at the onset of perfusion. To determine the mechanism of these reactions, the effect of homologous blood exchange in the dog was studied using a variety of techniques, including total and partial cardiac bypass. METHODS In the 75 preliminary total cardiopulmonary bypass studies a vertical-screen oxygenator and rotary pumps were used. The priming volume was approximately 3 litres and blood from four or five donors was required. Subsequent total bypass experiments were performed with a rotating-disc oxygenator and a pump driven by compressed air (Hufnagel, McAlinden, Vardar, DeVenecia, and Real, 1958). The volume of this circuit was somewhat smaller and only three donors were needed in most experiments.The two partial cardiac bypass techniques used were veno-arterial pumping and left heart bypass. Venoarterial pumping consists of transferring the superior vena caval return to the lower aorta without oxygenation (Hamer, Dickson, and Dow, 1959), and left heart bypass involves pumping oxygenated blood from the left atrium to the aorta. A simple circuit with a closed venous reservoir and a capacity of 800 ml. was used for both these procedures, and was usually filled with blood from one donor.Blood exchange with constant volume was performed by placing 200 ml. of blood from each of one to four donor animals in a closed flask suspended above the recipient and connected to cannulae in the femoral artery and vein. When the cannulae were opened pressure in the flask rose 25 to 30 mm. Hg and blood exchange occurred, the volume of blood in the flask increasing by less than 50 ml.Mongrel pressure by a catheter placed in the cranial mesenteric vein through a small abdominal incision. Normal respiration was allowed unless thoracotomy was part of the procedure, when artificial ventilation was maintained through an endotracheal tube. With all techniques blood exchange was begun at between 300 and 600 ml. per minute. Both total and partial cardiac bypass studies were continued for one hour. The effect of intravenous injection of Escherichia coli endotoxin (1 to 7 mg.) was also studied.Pressures were recorded with Statham P23AA strain gauges and a Cambridge direct-writing instrument. Blood samples were obtained from the recording catheters and from the bypass. Oxygen and CO> contents were determined by the method of Van Slyke and Neill (1924), and oxygen saturation was expressed as a percentage of the haemoglobin capacity without correcting for the oxygen in solution. Plasma pH was measured using a Beckman model G meter. The haematocrit was determined and the nomogram devised by Singer and Hastings (1948) for human blood was used to derive the blood buffer-base and C02 tension. Blood lactate (Mitchell and Cournand, 1955) and pyruvate (Friedemann and Haugen, 1943) were measured, and plasma sodium, chloride, and potassium levels determined. RESULTS The incidence o...