298 



C, LEVINTHAL 



information. On duplication in a new host these two subunits separate from 

 each other, and each makes another structure which has the same genetic 

 information as itself [as in the model of DNA duplication suggested by- 

 Watson and Crick (1953)]. Using this type of model, the picture which we 

 draw for the majority of the heterozygous particles is indicated in Fig. 5 (a). 

 The structures which are shown to be improbable by the three-factor experi- 

 ments are indicated in Fig. 5 (b). The average length of the overlap region 



(a) 



(b) 



Fig. 5. The two possible schemes by which the two-factor data on heterozygous 

 particles can be represented. The three-factor crosses indicated that the large majority 

 of the observed particles are of the form (a). 



can be estimated from the frequency of double heterozygosis for closely 

 linked markers, since this double heterozygosis occurs if both of the markers 

 are on the overlap region. It should be remembered, however, that in drawing 

 a picture of the type indicated in Fig. 5 we are doing no more than indicating 

 graphically by two lines the genotypes of the two segregants or two types of 

 daughters which are usually produced by single heterozygous particles. 



The graphic method used here to describe the heterozygotes is sufficient 

 to satisfy all the available experimental data; however, one can certainly 

 invent other models which also satisfy the data. For example, instead of 

 speaking of the average length of the overlap region, with the tacit assump- 

 tion that the distribution in lengths is more or less normal, one could imagine 

 a situation in which there were a large number of very small overlaps and a 

 smaller number of very large ones. This would also satisfy all the experi- 

 mental data mentioned above, but would lead to quite different conclusions 

 concerning the role of the heterozygotes in forming recombinants. 



