THE influence of acid and alkali on proteins in relation to the antigenic properties of these proteins has been studied by many authors. Wells [1909] noted the marked loss of antigenic power when crystalline egg-albumin was treated with sodium hydroxide, whereas little loss occurred following similar treatment with hydrochloric acid. Ten Broeck [1914] found that racemised egg-albumin (prepared by Dakin's [1912] method) was non-antigenic, and similar results with other proteins were obtained by Landsteiner and Barron [1917] and by Kahn and McNeal [1918].Evidence of restoration of antigenic properties to alkali-treated proteins was furnished by Landsteiner and Barron [1917], who showed that when the non-antigenic alkali-treated serum-proteins were treated with nitric acid, antigenic xanthoproteins were formed; these xanthoproteins prepared from alkali-treated proteins gave precipitin and complement-fixation reactions with antisera to xanthoproteins from untreated serum-proteins and also acted as true antigens when injected into rabbits. These results, indicating some reversal of the process or processes which originally caused loss of antigenic power, are very significant, and it is possible that a further study of this phenomenon would add considerably to our knowledge of protein antigens.In view of the importance of this relationship between the action of alkali and the loss of antigenic properties, it was decided to attempt to define more precisely the relationship between PH and loss of antigenicity, and also to try to obtain, by some method other than that used by Landsteiner and Barron [1917], confirmatory evidence of the restoration of antigenic power to alkali-treated proteins. For the latter purpose, it was considered that iodination of the alkali-treated proteins would be most suitable, since iodination can be effected in neutral or faintly alkaline solution, and, although it replaces the original species specificity of serum proteins by a specificity determined principally by the 3: 5-diiodotyrosine groupings [Wormall, 1930], iodination does not lead to any appreciable loss of antigenic properties.In the investigation described here, the proteins of horse-serum were rendered non-antigenic by treatment with sodium hydroxide for approximately 24 hours at 180 in one series and 300 in another series. Fractions of these "alkali-proteins " were nitrated and others iodinated, and precipitin