With three animals in which the nitrogen metabolism was studied during the period before edema had developed, the observed loss of weight was consistently less than the theoretical loss of weight calculated from the negative nitrogen balance. Reasons are given for interpreting the discrepancy as evidence of increasing retention of fluid during the stage of pre-edema. In general the weight curve does not rise during the transition from pre-edema to edema. However, the weight does increase rapidly when fluid is accumulating in the peritoneal cavity.

The level of serum albumin which is critical for the development of edema varied between 1.04 and 2.17 gm. per cent. The range is sufficiently wide to suggest the existence of other factors of importance in determining the exact time of appearance of edema. During the phase of recovery the level of albumin which is critical for the disappearance of edema may be appreciably higher than the level which was critical for the formation of edema.

Among thirty samples of edema fluid the protein concentration was from 0.02 to 0.72 gm. per cent. The average protein level was 0.230 and the median level 0.165 gm. per cent. A positive correlation is not demonstrable between the duration of edema and the protein content of edema fluid.

A difference in behavior toward fluid retention between subcutaneous tissue and peritoneal cavity is pointed out. Subcutaneous tissue is more resistant to acute stress and less resistant to prolonged or repeated stress than the peritoneal cavity.

The rôle of tissue pressure in the etiology of edema is discussed. It is suggested that the critical level of protein in the serum is the concentration which permits the attainment in the tissue spaces of mechanical pressure great enough to break down the restraining action of the connective tissue boundaries of the spaces.