BACTEKIOPHAGE GENETICS 315 



bacteria (Meselson and Stahl, 1958) have lent support to the self-comple- 

 mentary mechanism of duplication originally proposed by AVatson and Crick 

 (1953). 



It is possible (Levinthal, 1957) to construct a model of phage genetics 

 which involves only simple modifications of this self-complementary scheme 

 and which is consistent with the experimental results so far reported for the 

 phage system. Unfortunately, however, one does not know enough of the 

 physical chemistry of the DNA molecule to evaluate the plausibiHty of all 

 the interactions that would be necessary in such a model. The difficulties 

 are not connected with uncertainties of the regular structure given by the 

 crystallographer but, rather, the difficulties concern the question of how the 

 molecule behaves in the intracellular environment in which the recombina- 

 tion processes would be taking place. We can only estimate very crudely, 

 for example, the fraction of the hydrogen bonds connecting the two chains 

 of the double helix which will be opened at any one time. However, making 

 what appear to be reasonable assumptions about the way in which DNA 

 molecules would behave in such an environment, one can obtain a self- 

 consistent scheme which does account for all the observations. The essentials 

 of this scheme are shown in Fig. 7 and are presented primarily to demonstrate 

 that there is no inherent difficulty in producing such a model. From the over- 

 all point of view this model results in the interaction of two DNA structures 

 which together produce one new particle which has a heterozygous region at 

 the point at which the copying changes from one of the parental structures 

 to the other; in this region negative interference can occur in that at least 

 one of the two chains can show several switches. In this scheme it would be 

 assumed that particles can also duplicate without mating, that is, one par- 

 ticle could make two as well as two making three. The model also makes 

 some rather specific predictions as well as accounting for currently available 

 data. For example, it suggests that whenever negative interference occurs, 

 that is, whenever there are several switches in the same small region, the 

 total number of switches must always be odd; the mating which produced 

 the particle of negative interference wiU produce a heterozygous particle with 

 the overlap in the region in which the negative interference occurs and the 

 multiple switches will occur in only one of the two strands of the heterozy- 

 gous particle. These predictions can in principle be tested by purely genetic 

 experiments. 



If all of the impHcations of this model are taken seriously, it follows that 

 high negative interference could occur during the local pair-wise interaction 

 which leads to the production of heterozygotes. On the other hand, group 

 mating would result if a second switch were to occur to a different parental 

 structure after the first had been completed. Experiments of Edgar and 

 Steinberg (1958) indicate that the interaction which produced high negative 



