Abstract Many experimenters have endeavoured to isolate diphtheria antitoxin from fludis containing it. In 1893 Brieger and Ehrlich (1) investigated the properties of milk obtained from animals which had been immunised against diphtheria toxin. This milk contained an appreciable amount of antitoxin which could be precipitated between 27- and 38-per-cent. ammonium sulphate. By this method they were able to increase the antitoxic value of their fluid about five hundred times. But in milk only two proteins are present which differ widely in their physical properties—caseinogen and lact albumin—and their results could not be extended to horse serum, which is the main source of diphtheria antitoxin. Their experiments did not show that antitoxin is not a protein.