RECOMBIXATION ANALYSIS IN MICROBLVL SYSTEMS 65 



cannot really distinguish between c, f, and h. Further, there is one 

 obvious error in the order, which can be ascribed to the extremely high 

 error in the determination of the recombination frequency of / with 

 wild-type DNA. Mutant d covers mutants g, j, and /, yet in the above 

 size order, mutant / comes out bigger than d. Clearly, either the link- 

 age relationship of / is wrong, or the determination of the wild-type re- 

 combination frequency of / by wild-type DNA must be wrong. The 

 latter is far more hkely, since the linkage relationship is consistent in 

 all crosses of mutant by mutant involving /. Thus /' may be presumed to 

 be smaller than d, is certainly larger than g, and possibly larger than c. 

 We may assume the size order to be as follows, taking these considera- 

 tions into account: g<c = / = /2</<tf<i<e. 



Let us consider next what is involved in obtaining wild-type re- 

 combinants when diflPerent mutated sites are present in the donor and 

 the recipient cell. Let us start from the simplest situation, in which a 

 point mutation is introduced into the donor DNA molecule. If it is in 

 position 1 of Figure 8, all sequences beginning to the left of the donor 

 mutation will contain the mutant site and will not yield wild-type re- 

 combinants. The effect of this will be to eliminate only very few of 

 the total recombination events that would yield wild type had the 

 donor mutant site not been present. If we place the donor mutation in 

 position 2, again only a few of the potential wild-type recombinants 

 will be eliminated: those starting just to the left of the mutated site of 

 the recipient cell and extending beyond the mutated site of the donor. 

 Thus the effects of a mutant site at either position will be small and 

 tend to diminish only sHghtly the wild-type recombinants arising from 

 insertion of long sequences. 



As the position of the mutant site of the donor approaches that of 

 the recipient cell, the reduction of the recombination ev^ents capable 

 of yielding the wild type will become increasingly severe. It is evident 

 that such a model cannot lead to proportionality between the dis- 

 tance separating two mutant sites and the frequency of wild-type 

 recombinants. 



If, instead of bringing the mutant sites closer, we separate them, 

 the mutant site in the donor molecule will cease to limit the frequency 

 of wild-type recombinants. Their number will approach that observed 

 when a mutant is transformed by wild-type DNA. Thus there will be a 

 limit to the distances which can be measured by double cross-over 

 events of the type assumed. 



Let us now examine what happens if one of the two mutant sites 

 has an appreciable linear dimension. As stated above, by a site of ap- 

 preciable linear dimension we shall mean one in which several points of 

 attack or return are possible within the mutant sequence. Figure 9 

 shows a pair of reciprocal transformation experiments involving a large 



