60 A. D. HERSHEY AND RAQUEL ROTMAN 



These conclusions are substantially confirmed by the distributions of num- 

 bers of recombinants shown in table 6, which are essentially of the Poisson 

 type, with variance only moderately greater than the mean. In other words, 

 the individual particles of any one recombinant must often arise independ- 

 ently of each other in the same bacterial cell, and with equal probability in 

 different bacterial cells. The deviations from the Poisson distribution are 

 nevertheless significant, and can be attributed to a moderate amount of growth 

 of recombinants. 



Table 6 



Distributions of Numbers of Recombinants in Single Bursts from 

 the Crosses hXrl3 and h rUXwild type 



The Poisson distributions show the numbers of tubes in the various classes expected if there 

 were no growth of recombinants. The distribution with mean 3.4 is appropriate to the hypothesis 

 of reciprocal exchange, and the distribution with mean 1.7 to the hypothesis of non-reciprocal 

 genetic interaction. 



NO. OF RE- NO. OF TUBES FOUND 



POISSON 

 COMB IN ANTS 



PER TUBE 







1 



2 



3 



4 



5 



6 



7 



8 



9 

 10-18 

 19-27 

 28-36 



The conclusion that genetic recombination is not suppressed in some of the 

 bacteria is also supported by the results of the other crosses, in which no bursts 

 yielding both parental viruses and lacking both recombinants were found. Only 

 one burst, from the cross hXr7, failed to show one of the recombinants; it 

 contained 95 percent of the h parent. 



INDEPENDENCE BETWEEN PROPORTION OF RECOMBINANTS AND BURST SIZE 



The data of table 4 do not show any significant correlation between burst 

 size and proportion of recombinants, which means that the number of re- 

 combinants must be very nearly proportional to the total yield of virus in 

 single bursts. 



This conclusion must be qualified in view of the following considerations. 



254 



