LINKAGE 



ferent mutant combinations show different de- 

 grees of reduced viability which may be deter- 

 mined by relative numbers in counts of segre- 

 gating fraternities. 



Linkage estimates are ordinarily made from 

 double backcross data. Breeding out virgin 

 Habrobracon Fl females gives ratios of the type 

 obtained from. a double backcross. The formula 

 for estimation of linkage is derived by using 

 the method of Maximum Likelihood (Mather, 1938; 

 Fisher, 1941). This method leads, in the theory 

 of large samples, to an estimate having the 

 smallest standard error which the data will 

 allow. If P equals recombination fraction, r 

 equals number of recombinations or crossovers, 

 s equals number of straights or non-crossovers, 

 then P = r/(r + s). It must be remembered that 

 each double recombination represents two cases 

 of a single recombination and must be added to 

 the single recombinations in each region to ob- 

 tain the total amount of recombinations in that 

 region. The recombination fraction can never 

 exceed 50 per cent since of the four available 

 chromatids only two can produce detectable re- 

 combinations. A higher recombination fraction 

 indicates that the genes are relatively far 

 apart; a low recombination fraction, that the 

 genes are close. 



A variation of the formula, P = r/(r + s), 

 was presented by P. W. Whiting and Benkert 

 (1934). They let AB, aB, Ab, and ab represent 

 the frequencies of the four phenotypes of off- 

 spring expected from heterozygous females a.b/+ 

 (coupling test) or a/b (repulsion test). It may 

 be supposed that wild-type, AB, is the most vi- 

 able, but somatic overlapping may increase the 



113 



