RECOMBINATION IN BACTERIOPHAGE 69 



between linked factors (Hershey and Rotman 1948). Equation (9) also pre- 

 dicts, in agreement with the data for single bursts, proportionality between 

 yields of recombinants and k. 



The agreement supports the inferences previously drawn that the markers 

 r7 and rl3 are attached to the same linkage unit, and that the frequency of 

 crossing over between them is 0.5. It suggests that the pairing itself is complete 

 without appreciable repetition, and occurs at random except that about 26 

 percent of the units of each kind are effectively segregated from their opposite 

 numbers. The measure of this segregation, m, is on this view the same for 

 crosses between linked and unlinked factors. On the other hand, this inter- 

 pretation cannot be rigorously correct, because one can show by multiple 

 factor crosses (Hershey and Rotman 1948) that repeated exchanges, or ex- 

 changes among three viral particles, occur. An estimate of the amount of 

 repeated pairing has not yet been attempted, except that the considerations 

 just offered suggest either that it is small, or that random pairing is limited to 

 a small proportion of the population. 



It follows from equation (9) that the interpretation in terms of orderly 

 pairing accords with the fact, otherwise very puzzling, that the proportion of 

 recombinants is not affected by size of burst even in crosses between linked 

 factors. 



It has been seen that the linkage data support fairly well the idea of linear 

 structure, but independent evidence for crossing over is meagre. According 

 to any simple model of reciprocal exchange, a correlation between proportions 

 of sister recombinants in individual bursts would be expected. This expectation 

 has been only partially realized, and the question arises whether the linkage 

 data themselves require the crossover hypothesis. The following model, 

 suggested by Dr. A. H. Sturtevant, shows that they do not, and also shows 

 that the question of reciprocity is closely connected with the question whether 

 the exchanges are material transfers. 



Suppose that the replication of linear structures occurs zipperwise along the 

 pattern from one end to the other, but that the partners separate prematurely 

 to yield fragmentary replicas. Additions to the fragments are subsequently 

 possible only after pairing with the same or another homologous structure, 

 which in mixedly infected bacteria could belong either to the same or a different 

 parental line. Genetic recombination in a two factor cross will depend, then, on 

 the contingency that the two marked regions of a given replica be laid down 

 one after the other on homologous structures from the two unlike parents. 

 With simple assumptions, all the consequences of the crossover hypothesis 

 (equation (6)) follow from this model, except that the independent origin of 

 the two recombinants provides an additional source of independent variation 

 in their numbers. 



The complications peculiar to this model have to do principally with the 

 evidence that exchanges occur only during the terminal phase of growth. 

 These complications are not very serious if one assumes that during early 

 stages of growth the probability is great that a fragment will be started and 

 completed on patterns belonging to the same parental line; that is, that the 



263 



