IMMUNOLOGICAL METHODS IN THE STUDY OF VIRUSES 529 



reported. The process appears to take time and to result eventually in an 

 irreversible union. 



In the opinion of one important group, whose spokesman is Boyd (1947), 

 the lattice theory is inappropriate, the essence of the precipitin and allied 

 aggregation reactions being that the deposition of a spread-out and denatured 

 molecule or mosaic of molecules of antibody on a molecule or particle of 

 antigen will give it relatively hydrophobic quaUties, with resultant loss of 

 solution or suspension stability in the usual ionic environment. 



The question of specificity is also controversial, but from the point of view 

 of relevance to the phenomena encountered in immunological work with 

 viruses, the most important feature is the heterogeneity of all antibody 

 populations. If one has two similar but not identical antigens of any degree 

 of complexity, such as purified serum albumin of human and bovine origin, 

 red cells of mouse and rat or two influenza A strains, appropriate antisera 

 against each member of the pair can be prepared. Each will show a high 

 degree of cross reaction. In each case, however, absorption of a serum with 

 the heterologous antigen of the pair will remove its reactivity with that 

 antigen but leave the homologous activity hardly altered. Where, as is 

 often the case with a group of related bacteria, many heterologous strains 

 will react with a given serum, it is possible by appropriate absorptions to 

 produce a widely different set of reagents, each necessarily corresponding to 

 a reassortment of the molecular species in the population of antibody mole- 

 cules. When absorbed with homologous antigen, however, all types of act- 

 ivity, homologous and heterologous, are removed in parallel. 



Analysis of the significance of this phenomenon has been virtually con- 

 fined to bacterial antigens. Both Furth and Landsteiner (1929) working with 

 salmonellas, and Burnet (1934) with dysentery bacilli, showed that there was 

 no evidence for the presence of a mosiac of antigens on the bacterial surface. 

 Soluble preparations were precipitated as a whole, not in fractions. The 

 diversity is in the antibody population, not in the antigen population. The 

 suggestion previously put forward (Burnet, 1934) can still be supported. This 

 is that any antigenic molecule carries a variety of potential antigenic deter- 

 minants, e.g., ABCD, but that any antiserum will contam antibody mole- 

 cules carrying complementary groupings that are not representative of the 

 whole antigen. If we simplify the conditions by assuming that each antibody 

 molecule carries active patches corresponding to two of the four antigenic 

 determinants, then the population of antibody will be composed of ten 

 molecular species, aa, ab, ac, etc. All of these will react with antigen ABCD, 

 but only a proportion with antigens AEFG or BFGH. 



If this is true, it has important imphcations for one of the classic immuno- 

 logical problems in virology, the significance of serological differences in 

 influenza A viruses. Hirst (1952) foimd that by appropriate absorption of 



