RECOMBINATION IN BACTERIOPHAGE 67 



recombinants would show some individual bursts containing no recombinants 

 and others containing a very large proportion, especially at low multiplicity 

 of infection. Instead one finds a comparatively uniform yield of recombinants, 

 and the distribution of their proportions is not affected by the multiplicity of 

 infection. 



Also if exchanges occurred freely throughout the period of multiplication of 

 the virus, one would expect considerably greater variations in yields of recom- 

 binants than we have found. For instance, in the cross hXrU, in which there 

 are only two or three exchanges per bacterium, the variations in yields of re- 

 combinants are not much greater than those expected to result from a random 

 variation in the number of exchanges alone. Moreover, most of the bacteria 

 yield only a few recombinants, so that little growth can have occurred sub- 

 sequent to exchange. The conclusion is unavoidable that the exchanges are 

 limited to the terminal phase of multiplication, or at any rate that recombi- 

 nants are prevented from multiplying appreciably in most of the bacteria. 



It is remarkable that the variations in proportions of recombinants are so 

 little dependent on the degree of linkage (as between hXr7 and hXrU), or on 

 the postulated mechanism of exchange (as between the above and hXrl). The 

 coefficients of variation in proportions of individual recombinants among 

 single bursts are, for hXrl, about 40 percent; for hXr7, about 60 percent; and 

 for hX rl 3, about 100 percent. This circumstance also supports the inference 

 that the exchanges are limited to a late phase of multiplication. 



The hypothesis stated permits one to examine further the structure of the 

 linkage units. Since crosses between rl3 and any of the mutants belonging to 

 the group closely linked to r7 yield about the same proportion, seven percent, 

 of wild type (Hershey and Rotman 1948), it might be supposed that one 

 crossover between the distant markers is always accompanied by several 

 others, so that 50 percent of the progeny of synapsed pairs of the units r7 and 

 rl3, for example, would be recombinant types. If this supposition is correct 

 the terms of equation (6) can be evaluated by setting c = 0.5 and p(h + r + ) = 0.07. 

 This gives 0.28 for the average fraction ms of virus descending from unlike 

 synapsed pairs. If this fraction is assumed to be the same in other crosses in- 

 volving the same linkage structure (the cross h r7Xrl3 reported in this paper 

 suggests that it is), one can write for them 



p (wild type) = 0.14 c, (7) 



where c is the appropriate crossover frequency and the proportion of wild type 

 is experimentally measured. The data for the three point crosses involving 

 r2, r3, and r6 (Hershey and Rotman 1948) are examined from this point of 

 view in table 9. The proportions of wild type have been calculated for random 

 crossing over between unit linear structures, using the crossover frequencies 

 given by (7). It will be seen that the data are entirely compatible with the 

 hypothesis tested. Additional tests of this kind are needed, however. 



It will have been noticed that the average yield of recombinants in crosses 

 between distant linked factors is very nearly half that found for unlinked 

 factors. Dr. M. Delbruck has pointed out to us that this relationship can be 



261 



