282 F. M. BURNET 



with a strain with, little or no virulence v for that host. The strains will also 

 need to be differentiated by some other characteristic if results are to be 

 intelligible, so that if this is differentiated as A and a, we have a virulent 

 strain AV and an avirulent one av. Obviously, recombination experiments 

 cannot be carried out in host H if we are to avoid the possibility of more 

 virulent mutants of av being selectively favored. All such experiments require 

 what can be called a neutral host (N) on which both virulent and avirulent 

 strains can grow freely. In other words, both should be actively and, pre- 

 ferably, equally virulent for the neutral host. Fortunately, recent technical 

 developments in virology have made it relatively easy to provide a neutral 

 host in the form of an appropriate tissue culture or the chick embryo. If we 

 are interested in mouse and monkey virulence of poliovirus, the neutral host 

 will be a HeLa cell tissue culture, for mouse virulence of influenza viruses the 

 allantoic cavity of the chick embryo, and for poxvirus studies the chorio- 

 allantoic membrane. 



If our experience with influenza viruses can be generalized, some special 

 precautions are needed in characterizing recombinants. In many instances, 

 when interaction occurs between virulent and avirulent strains and a number 

 of LD fluids are tested at once for virulence, it is found that there is a con- 

 siderable range of virulence. As a hypothetical example, suppose strain AV 

 has virulence which can be given a value 8, while av for the same host has the 

 value 0, after interaction, 10 LD fluids with the second character are found to 

 have virulence against H of 0, 0, 0, 0, 1, 1, 2, 4, 4, 4. One of the fluids with 

 virulence of level 4 is titrated and 10 LD fluids from this titration similarly 

 tested against H. The result is likely to be 1, 1, 1, 1, 2, 2, 3, 4, 4. If one takes 

 a primary fluid with virulence 1 and similarly prepares a set of descendant 

 LD fluids the values will be perhaps 0, 0, 0, 0, 0, 0, 1, 1, 1, 1. In each case there 

 is a downward drift in virulence, especially shown when we are concerned with 

 the "transfer" of virulence to a strain with the other qualities of the avirulent 

 parent. From the nature of the phenomenon it is extremely difficult to be sure 

 that we are not dealing with mixtures in such experiments and it is always 

 advisable to have a serological difference available by which at least one check 

 for purity can be applied. 



IV. Genetic Interactions within the Myxovirus Group 



A. Historical 



The first evidence that recombination could occur between animal viruses 

 was published by Burnet and Lind in 1949. They were able to obtain from 

 mouse brains inoculated with mixtures of a neurotropic influenza virus strain 

 NWS (Stuart-Harris, 1939) and non-neurotropic strains of different serological 



