64 A. D. HERSHEY AND RAQUEL ROTMAN 



virus, and the heterologous cross yielded recombinants in the same proportion 

 as found with the parental stocks. These tests show not only that the recom- 

 binants contain the same genetic markers as the corresponding progenitive 

 types, but also that the region between the markers is unchanged. 



MIXED YIELDS CONTAINING ONLY ONE PARENTAL TYPE OF VIRUS 



Only four mixed bursts lacking one of the parental types of virus were found 

 among the experiments reported in this paper. One, from the cross hXrl, 

 contained 86 percent h + r and 14 percent h r. A second, from the cross hXr7 

 contained 99.6 percent h r+ and 0.4 per cent h r. A third, from the same cross 

 with low multiplicity of infection, contained 83 percent h r + , and 17 percent 

 h r. The fourth, from the cross h r7X wild type, contained 23 percent h r + and 

 77 percent h r. In this case the yield of h r + , which appeared to be homogenous, 

 formed atypical plaques and proved on isolation to differ from any known 

 mutant of T2H. It seems reasonable to suppose that these exceptional bursts 

 contained progeny stemming from mutants contaminating the parental stocks 

 of virus. On the other hand, the one exceptional burst following low multi- 

 plicity of infection, together with the failure to find similar bursts among the 

 crosses involving h and rl3, suggest that a different interpretation should be 

 looked for. Genetic tests which might have clarified this point are lacking. 

 For the present we conclude, as a first approximation, that recombinants arise 

 only in those bacteria in which both parental types of virus succeed in multi- 

 plying. 



It may be added here, because the question arises in connection with these 

 exceptional bursts, that no correlation can be seen between the proportion 

 in mixed bursts of the total virus containing the h allele, and the proportion 

 of the recombinant virus containing the h allele. We have therefore omitted 

 this datum from the tables. 



DISCUSSION 



In collecting and analyzing the data just described, we have had in mind 

 the following questions. Does genetic exchange occur in the course of matings 

 between viral particles, or is it the expression of a mechanism of growth such 

 as that visualized by Luria (1947), according to which the multiplying units 

 in the cell are not phage particles, but simpler structures derived from them? 

 Can the linkage relations represented in fig. 1 be interpreted in terms of linear 

 chromosome-like structures? Are the genetic exchanges reciprocal, as one ex- 

 pects for simple cases of crossing over, or must one look for an alternative 

 mechanism more intimately connected with the mode of reproduction of 

 the virus? 



It was soon apparent that the data for crosses between linked and unlinked 

 factors tended to give different answers to these questions, and we were led 

 to consider a model based on two distinct mechanisms of exchange. The neces- 

 sity for this arises from the following facts. 



First, the linkage data indicate a limitation at about seven percent to the 

 proportion of wild type found in crosses between linked factors, which is dif- 



258 



