278 F. M. BURNET 



Two factors are probably responsible for these discrepancies: (1) For any 

 given host-viruses system, the likelihood of any cell- virus particle interaction 

 initiating demonstrable infection is always something less than unity. Where 

 the host is of relatively low susceptibility, the proportion of fertile contacts 

 may be very small. (2) Any population of virus particles will comprise some of 

 intrinsically lower mfectivity (cf . incomplete virus) and some whose infectivity 

 has been reduced or destroyed by thermal degradation or other environmental 

 influences. 



Under the circumstances, it is not possible to provide any infallible rule for 

 the production of pure clones of virus. If we could deal with a system in which 

 virus particles were 100 % infective, the possibilities could be calculated. 

 Theoretically, when equal numbers of infective particles A and B are mixed 

 and inoculated into fully sensitive hosts at high dilution, one should obtain at 

 the 1D 50 level 17 % of positive harvests containing both A and B. At a further 

 10-fold dilution only 0.5 % will show both types. 



In Liu and Henle's (1951) experiments, mixtures containing equal numbers 

 of viable imits of strains A and B were inoculated at various fractions of the 

 1D 50 and the harvests tested by subinoculation in homologous antiserum for 

 presence of the heterologous type. Their results gave about 30 % of double 

 infections at the 1D 50 level, which is nearly twice the theoretical expectation 

 and suggests that, at some stage after infection has been initiated by a virus 

 particle, cells still uninfected become more susceptible to infection by residual 

 particles which otherwise would fail to initiate infection. 



The practical difficulty in working with influenza viruses is, however, not 

 quite so great as these figures suggest. Most of the double infections in Liu and 

 Henle's experiments were only detectable by subinoculation in antiserum 

 homologous to the apparent type of the harvested virus. There was a great 

 excess of one form and a limit dilution of such a fluid would give a much 

 lower proportion of double infections than is obtained from a starting mixture 

 of equal parts of each virus. If we have a mixture in the proportion 90A : 10B, 

 the theoretical proportion of double infections at the 1D 50 level is reduced to 

 2.8 %. This principle can be made use of in practice by the in vitro character- 

 ization of all limit dilution fluids. If such tests give unequivocal results we 

 can be certain that not more than 10-20 % of virus of different in vitro 

 character is present. In any given experiment the likely types of "contami- 

 nant" will be known and the tests arranged accordingly. By using only 

 fluids in which the type required is shown to be dominant by in vitro tests, 

 the necessary procedures to obtain pure clone fluids are considerably reduced. 

 We have found no exceptions to the rule that if two successive limit dilution 

 titrations with 4-6 embryos per dilution each give three limit dilution fluids 

 with uniform in vitro findings, all descendant clones are uniform in their in 

 vitro qualities. Similar principles would presumably hold for any other virus 



