RECOMBINATION IN BACTERIOPHAGE 59 



if some of the single absences result from erroneous recognition of one recom- 

 binant, but is otherwise subject only to its sampling errors. Taken as a measure 

 of failure of reciprocal exchange, the fraction 4/125 implies an average of 

 3.4 exchanges per bacterium. If, on the other hand, exchanges are not recipro- 

 cal, the fraction (20 percent) of the bursts lacking any given recombinant cor- 

 responds to 1.7 genetic transfers per bacterium. 



These estimates may be too low if conditions vary from bacterium to bacte- 

 rium in such a way that genetic exchange is suppressed in some bacteria. 



Table 5 



Distribution of Bursts with Rtspect to Absence of Recombinants 

 See legend table 4. The proportions of recombinants have not been corrected for dispropor- 

 tionate yields of parental types of virus. 



Yields with small k and small burst size must have this effect, but are evi- 

 dently not very important in the data of table 5, since the different classes are 

 very similar with respect to k and burst size. In short, it is necessary to con- 

 clude that there are at least two or three genetic exchanges per bacterium, 

 independently of the mechanism by which recombinants arise. 



A different kind of estimate of the number of exchanges per bacterium is 

 obtained from the number of recombinants actually found. The data are sum- 

 marized in table 6 in the form of distributions of numbers of the several re- 

 combinant types. One finds on the average 3.4 recombinants of any one kind 

 per bacterium. This evidently furnishes an upper limit to the number of ex- 

 changes per bacterium, insofar as exchanges yield viable and countable pro- 

 geny. This result, taken in conjunction with the preceding estimates, leads to 

 several remarkable conclusions. 



First, since the two methods of estimation, one minimal and one maximal, 

 yield about the same result, there must be in fact only two or three exchanges 

 per bacterium in the crosses between closely linked factors. 



Second, the recombinants must undergo little multiplication after they 

 arise in the cell. 



Third, the conditions of viral growth in different bacterial cells must be 

 equally favorable to genetic recombination; otherwise a larger proportion of 

 bursts would fail to show recombinants. 



253 



