A study of 5 patients confirmed earlier observations on the union of Donath-Landsteiner hemolysin with red blood cells only at low temp., and the greater effectiveness of short chilling over a long period at low temp. The auto-hemolysin and the iso-hemolysin could not be separated by absorption experiments. CO2 could not be demonstrated to act as a substitute for chilling in promoting union of hemolysin and red blood cells. Artificially produced paroxysms were sometimes characterized by hemo-globinemia without hemoglobinuria, frequently by a sudden drop in the number of leucocytes in the blood and the percentage of polymorphonuclear cells, followed by leucocytosis. Elevation of blood pressure and temp. occurred inconstantly in the artificially produced paroxysms. Antiluetic treatment in 3 cases was followed by cessation of symptoms, disappearance of auto-hemolysin from the blood and change of a positive Wassermann to a negative one, in the order named.