Summary: By means of appropriate adsorption tests it has been shown that Landsteiner's theory as to number and allocation of human isohemagglutinins and agglutinable elements is probably correct. Evidence suggesting the possible existence of sub-groups is discussed, and other data are submitted in support of the recommendation that both the typing of bloods and direct tests between recipient and prospective donors be adopted as routine procedures preliminary to transfusion. Attention is drawn to a number of observations which arouse suspicion, at least, that blood-plasma rather than cells may be the source of certain post-transfusion reactions. There is probably no etiologic or concurrent antigen-antibody plasma reaction following primary transfusions because we were unable to demonstrate any isoprecipitins or alexin-binding antibodies in cross-titrations of representative group sera. Both from the clinical standpoint and because of the information which would derive to the subject of antigenic specificity, serologic methods deserve more thorough application particularly to the problem of those reactions which follow repeated transfusions.