Any method of determining the volume of the cardiac output in man which requires intelligent, alert and active cooperation by the subject can scarcely be regarded as suited to the study of this most important aspect of cardiac function in the clinic. Errors are sure to arise from ignorance, lack of training, excitement, apprehension, etc., of such magnitude as to render results worthless. When the ethyl iodide method of Henderson and Haggard (1) was devised, in which the sole requirement of the subject is that he breathe through a mouthpiece, it appeared that a long step had been taken toward the attainment of quantitative information concerning the heart's output under the various circumstances of disease. Impressed by these possibilities Starr and Gamble, at the suggestion of Professor A. N. Richards and with generous assistance from Professor Henderson, learned to use the method and began the study of normal subjects in order to accumulate the experience and the data requisite for its profitable application to patients. Unfortunately discrepancies were soon encountered between their experiences and those which Henderson and Haggard had published.It was soon evident that, by means of the analytical method of Henderson and Haggard, known concentrations of ethyl iodide could not be estimated with satisfactory accuracy. It was therefore necessary to devise new methods which met this requirement (2). Employing these methods Starr and Gamble were unable to verify certain of the conceptions on which the original method was based, namely the absence of ethyl iodide from blood returning to the lungs, and the size of the distribution coefficient for ethyl iodide between air and blood (3). They were therefore confronted with the question of abandoning the project altogether or of so altering the plan of the method that it should yield results in which they could have confidence. The latter alternative was chosen, the distribution coefficient was redetermined and the method was redesigned so that the relatively large amount of ethyl iodide in venous blood could be estimated at each determination of cardiac output. By means of this new method it was found possible to make satisfactory estimations of the flow of blood perfused through the lungs of dogs at a known rate 13