Carbon dioxide gas placed within veins of the arm or leg can display intrahepatic structure by filling liver venous channels (1). The phenomenon has been seen in man when, during negative contrast angiography of the right atrium, reflux of gas through the inferior vena cava and into connecting hepatic veins inadvertently occurs (Fig. 1). Before the deliberate use of this technic in human patients, canine experimentation and evaluation of autopsy material seemed indicated. Experimental Methods A freely flowing intravenous drip with an 18-gauge needle, stock sugar solution, and standard plastic tubing was established in a peripheral vein of a 20-to 40-kg dog under barbiturate anesthesia. A 3-way stopcock, 50-ml syringe, and a small tank of medical carbon dioxide were connected to the infusion. The syringe was washed several times with gas from the tank. Following this, a 50-ml bolus of gas was injected rapidly with hand pressure into the vein (2). Conventional roentgenograms of the liver were obtained at intervals of one to five seconds after injection. Rapid-sequence and cinefluorographic technics were also employed to provide supplementary information. The right atrium in man is filled with the patient in the left lateral decubitus position. This permits gas to reflux only into the right hepatic lobe which is uppermost. The experimental animals, however, were placed in many different positions relative to gravity. In addition to the in vivo animal studies, normal and pathologically affected human livers were insufflated after postmortem removal and examined radiographically. Results As expected, the procedure was free of morbidity in that no vascular spasm, excessive bleeding, or sensitivity response was encountered. Both mortality and the effect upon physiological parameters caused by this technic are thought to be negligible (3, 4). None of the dogs died after spaced but repeated injections of volumes as large as those used in human examination. The lethal dose range appeared to be above several hundred milliliters of gas given as a continuous flow directly from the reservoir. Visualization of the venae cavae and large liver veins near the hilus was good, but outlines of small vessels near the periphery of the liver shadow never attained the quality of detail found accidentally in man. Although venous filling dynamics were better demonstrated by cinefluorography, anatomic structure was best shown by well collimated single exposures. Radiographs of isolated human livers with known tumor foci showed distinct evidence of extrinsic pressure upon the veins after air insufflation. This could be interpreted as indicating the site of mass lesions. Artificial masses produced by distended Foley catheter balloons placed at random in a normal liver were easily detectable (Fig. 2)