RECOMBINATION IN BACTERIOPHAGE 55 



above the mean, as is the case with the yields of a minority infecting type, 

 but shows a single mode slightly less than the mean. 



The last two lines of evidence cited seem to show that the principle of limited 

 participation (Dulbecco 1949) referred to above, operates only during the 

 initial stages of infection, or at any rate does not influence the yields of genetic 

 recombinants arising within the mixedly infected bacteria. They suggest 

 further that the h mutant is superior to r7 or rl3 in excluding power only, not 

 as a competititor during actual multiplication. 



YIELDS OF GENETIC RECOMBINANTS FROM SINGLE 

 MIXEDLY INFECTED BACTERIA 



In order to study the variations in yields of recombinants intrinsic to the 

 recombination process, one would like to exclude as many as possible of the 

 accessory sources of variation. The most important of these are variations in 

 burst size, and variations in the relative numbers of the two infecting viruses 

 adsorbed to individual bacteria. It will be seen presently that effects of varia- 

 tions in burst size can be avoided by the simple expedient of computing propor- 

 tionate yields of recombinants, these being independent of burst size. The effect 

 of variations in relative multiplicity could be minimized either by going to 

 very small or very large total multiplicities. Low multiplicities are uneconomi- 

 cal, because at multiplicities sufficiently small so that most of the mixedly 

 infected bacteria receive only one viral particle of each type, very few of the 

 test cultures will yield a mixed burst. High multiplicities also introduce dif- 

 ficulties (Dulbecco 1949). We have chosen to use total multiplicities between 

 10 and 20, within which range the yield of recombinants is constant. 



As previously described, the elementary viral yields vary considerably in 

 the relative numbers of the two parental types of virus and, as expected, these 

 variations influence in turn the yields of recombinants. A correction for this 

 source of variation was devised as follows. Assuming that the genetic inter- 

 action occurs between unlike viral pairs, and that the composition of the viral 

 yield provides a direct measure of the composition of the intracellular viral 

 population during growth, one computes an interaction coefficient 



k = p(A)[l- P (/0] (3) 



in which p(h) is given by (1), and k expresses the influence of the composition 

 of the population on the number of unlike viral pairs present in the cell, neg- 

 lecting effects of genetic recombination. 



The coefficient k has a maximum of 0.25 when half the viral yield contains 

 the h allele. Dividing the proportions of recombinants by 4k serves therefore 

 as a correction for inequality of yields of the parental viruses. This correction 

 is ambiguous only for bursts in which the yields of the two recombinants are 

 large and unequal, and bursts from which either recombinant is absent. 



A summary of the single burst data is given in table 4, which includes the 

 mixed bursts only. The bursts have been separated into the classes k^0.21 

 and k^0.20, to show the effect of the correction described above. It will be 

 seen that the uncorrected mean proportion of recombinants is larger for the 



249 



