SEROLOGICAL VARIATION 255 



further established by the demonstration that the serum of ferrets or mice 

 given repeated inoculations of the PR8 strain developed neutralizing anti- 

 bodies to the swine virus, and vice versa (Francis and Magill, 1935; Francis 

 and Shope, 1936). It was also noted that the serum of some persons contained 

 neutralizing antibodies to swine influenza virus, indicating further a relation- 

 ship between the viruses of the human and swine influenza (Andrewes el al., 

 1935; Shope, 1936), while convalescent human serum reacted equally well with 

 both viruses in the complement fixation test (Smith, 1936; Hoyle and Fair- 

 brother, 1937). 



At this time it was demonstrated that serological differences existed 

 between strains of human origin (Magill and Francis, 1936). The epidemic of 

 1936-1937 provided a number of additional strains, and a series of studies 

 towards antigenic analysis ensued (Magill and Francis, 1938; Francis and 

 Magill, 1938; Smith and Andrewes, 1938; Burnet, 1937). It became apparent 

 that a number of antigenic components existed in varied patterns and with 

 extensive overlapping among the available strains. Under the conditions of 

 study, however, it was noted that some strains presented greater serological 

 individuality than others; some were poor antigens; some reacted as well or 

 better with serum against heterologous strains than with homologous 

 serum. It was clearly demonstrated that antiserum taken from the test 

 animal early after inoculation had a high degree of strain specificity, while a 

 later specimen or hyperimmune serum demonstrated by its broader effect 

 the multiple relationships between the various strains. The results emphasized 

 the importance of reciprocal cross testing, since the serum against one strain 

 might neutralize the homologous and a given heterologous strain, but serum 

 to the latter would have little effect upon the first, so that different impres- 

 sions would be gained if only one of the sera was used. Although some 

 suggestions were offered for grouping these strains as types (Burnet, 1937; 

 Smith and Andrewes, 1938), the limited knowledge scarcely seemed to warrant 

 more than rough groupings, since the characteristics of one group tended to 

 merge into those of another. In general, the strains from one epidemic period 

 in different parts of the world closely resembled one another; those from 

 1935-36 seemed to differ somewhat from those of 1936-37, but exhibited 

 more relationship with the PR8 strain of 1934 than with the original WS 

 strain of 1933. One of the English 1936-37 strains (Gatenby) resembled the 

 1935 strains. Two other strains closely resembled WS. The relation of swine 

 influenza virus to the human strains was more distant. The observations 

 made by neutralization tests in mice agreed quite well with those from 

 reciprocal cross immunity tests in vaccinated mice with the various strains, 

 although the latter procedure tended to bring out a broader group immunity. 

 It was suggested by Magill and Francis (1938) that the virus particle of a 

 strain might well have a surface composed largely of the dominant antigen, 



