54 A. D. HERSHEY AND RAQUEL ROTMAN 



and probably do not influence appreciably the yields of recombinants in genetic 

 crosses, as the following discussion will show. 



The two distributions shown in fig. 6 illustrate the competitive relations 

 encountered. In the cross hXr7, most of the unmixed yields contain the h 

 rather than the r parental type. Corresponding to this, there is a tendency for 

 the mixed yields to contain an excess of h virus. The combined effect is to cause 

 a definite increase in yield of h virus at the expense of r, as compared with the 

 input proportions. The cross hXr!3 also shows these characteristics, the ef- 

 fects being evident in table 2. 



In the cross hXrl, on the other hand, the unmixed yields of each kind are 

 approximately equal in number, and the mean proportion of h virus in the 

 mixed bursts, in the total yield, and in the input mixture of viruses is the 

 same. The crosses between wild type and h rl, h r7, and h rl3 are like hXrl in 

 this respect, as shown in table 2. 



It might be supposed that the suppression of one virus by a second is favored 

 by an excess of the second. This is true only in a special sense, as Dtjlbecco 

 (1949) has shown, and we have confirmed. An excess of one virus tends to sup- 

 press a minority type completely in some bacteria, but there is a compensating 

 excess of this type among the mixed bursts, so that the average proportion of 

 the minority virus in the yield averaged over many bacteria is the same as in 

 the input mixture. This identity has been established with considerable pre- 

 cision for proportions of rl between 7 and 50 percent in mixed infection with 

 wild type. The nature of this relationship, which is at first sight perplexing 

 in the case of unequal multiplicity, has been explained by Dtjlbecco (1949) 

 in terms of a limitation to the number of viral particles which can participate 

 in growth in a single bacterium. If all those viral particles in excess of a certain 

 number attached to the same bacterium fail to grow, and if the excluded ones 

 are chosen at random, the result will be precisely the one described, provided 

 there is no selection during the growth of the successful particles. 



It is apparent that with certain viral pairs, the excluding mechanism does 

 not operate at random,' or there is continuing selection during growth. Thus 

 h mutant slightly suppresses r7 or rl3, but not rl or wild type. Wild type 

 suppresses rl3, but not rl or r7 (Hershey and Rotman 1948). There is no 

 selection with respect to either h or r factors when wild type is crossed with 

 h rl, h r7, or h rl3 (table 2). 



The competitive relations discussed above are of immediate interest only 

 in the negative sense that they probably do not influence the yields of recom- 

 binant virus in crosses. The evidence for the latter conclusion, drawn from data 

 presented elsewhere in this paper, may be summarized as follows: (1) the link- 

 age relations deduced from average yields of virus are the same as those de- 

 duced from single bursts selected for equality of yields of the two infecting 

 viruses; (2) in the reverse crosses hXr7 and h r7 X wild type, one gets with- 

 in experimental error equal numbers of all four recombinants in spite of the 

 fact that in one case the infecting pair, and in the other the recombinant pair, 

 have unequal excluding power; (3) in all crosses, the distribution of yields of 

 recombinants among single bursts does not show one peak at zero and another 



248 



