Problems in Immunology 143 



stantial evidence that the macrophages were involved, for 

 these cells took up the dye particles and then began shed- 

 ding surface films at about the time antibodies first ap- 

 peared. After McMaster's study of typhoid bacilli injected 

 into the lymph nodes, Ehrich, Harris, Dougherty, White, 

 and others experimented widely and demonstrated a con- 

 nection of lymphocytes with antibody formation, although 

 it was never certain what relation lymphocytic antibodies 

 bore to antibody production as a whole. Finally, a group 

 of Scandinavian workers proposed the plasma cell as the 

 primary synthesizer of antibodies. Following their lead, 

 Astrid Fagraeus devised ingenious experiments that ap- 

 peared to confirm the important role of these cells. The 

 issue, however, is somewhat beclouded by the failure of 

 cytologists to agree upon a definition of the plasma cell, 

 so that this aspect of the formation of antibodies is still not 

 fully solved. 



Neither have we advanced far in our knowledge of the 

 mechanism of the formation of antibodies. According to 

 the theory of Breinl-Haurowitz-Alexander-Mudd-Pauling, 

 an antigen or an active fragment penetrates to the as yet 

 uncertain sites of globulin formation and there, by its 

 presence, so distorts the newly synthesized globulins that 

 they are subsequently able to combine with the homolog- 

 ous antigen, when and if again in contact with it. The 

 Burnet theory, on the other hand, postulates an influence 

 of the antigen on the framework or enzyme systems through 

 which the globulins are built up — a modification which 

 persists as a kind of ''training," and so may explain the 

 more rapid secondary response following the reinjection 

 of certain antigens. I have discussed these theories in 

 greater detail elsewhere and have pointed out that the 

 Breinl-Haurowitz theory satisfactorily covers the behavior 

 of antibodies to pneumococcal carbohydrates in man, while 

 the Burnet hypothesis offers advantages in the description 

 of the behavior of antibodies to a protein such as diph- 



