glycerol estus, did not cause as great a degree of 

 antigen reaction when used in serological tests as did 

 the usual gonococcus antigen; but by saponif ication of 

 the fats with subsequent .isolation of pure nitrogen- free 

 fatty acids, the serological tosts gave better results 

 when such fatty acids were used as antigens. Continuing 

 along this line of investigation, Warden (21) determined 

 that the neutral fats of the gonococcus were of little 

 value when used as antigens in complement-fixation tests, 

 but very excellent and specific results were obtained by 

 using the fatty acids, or still better, the alkaline 

 salts of the fatty acids. Due to such findings he con- 

 cluded that the important factors in such reactions were 

 not only the empirical chemical constitutions of the 

 respective antigens, but that the molecular configuration 

 was as significant and as specific. It has been shown, 

 furthermore, by the same author (22) that if the fats are 

 added to a "solution" of colloidal cholesterol the degree 

 of dispersion of the molecular aggregates of the antigens 

 is greatly increased, thus presenting more surface and 

 producing a more active and sensitive antigen. This 

 finding, of course, adds confirmation to our ideas con- 

 cerning the colloidal reactions in immunology. By simi- 

 lar methods the same author demonstrated the antigenic 

 nature of the fats from Bacillus typhosus, Bacterium 

 anthracis, pneumococci, streptococci, and red blood cells 

 (23) (24). These fats were used not only as antigens in 

 experiments "in vitio", but were injected into suitable 

 animals, with subsequent production of specific anti- 

 bodies. Moreover, the antibodies obtained were not only 

 specific, but were also protective against a dose of 

 homologous bacteria which was fatal for a control animal. 



The careful work of Warden sheds a vast amount of 

 light on the question of chemical composition of antigens. 

 His results have: not yet been confirmed, nor have they 

 been disproved. No one will deny that certain fat- free 

 pure proteins cannot be used as antigens. But no one, 

 on the other hand, can maintain that all antigens are 

 protein. 



The object of this work is to study the chemical 

 nature of the antigen or antigens, of Bacillus coli; to 

 determine whether such antigens are protein, fats, carbo- 

 hydrates, salts, or a mixture of two or more of these; 

 and to analyse, if possible, the antigen in order that we 

 may know of what it' is composed. The experimental 

 methods and data are set forth in the succeeding pages. 



