178 CYTOLOGY chap. 



this would give a proportion in Fg of 



177 blue, long ; 15 blue, round ; 15 red, long ; and 49 red, round ; 



or, out of the total of 5600 in the above F,, the numbers : 



3871 blue, long ; 328 blue, round ; 328 red, long ; and 1073 red, round. 



This we see is a very close approximation to the numbers obtained in 

 the experiment, and it is to be noted that the gametes produced in excess 

 are those which exhibited the same combinations of characters as the 

 original parents {BL and bl). Similarly, if the gametes in the second 

 experiment were formed in the proportions 



1 BL : y Bl : 7 bL : 1 bl, 



the numbers which would be given in F2 are 



210 blue, long ; 103 blue, round ; 103 red, long ; and 1-5 red, round ; 



again a sufficiently close approximation to the numbers actually obtained. 



Thus, we conclude that the hybrid produces about seven times as 

 many gametes with the parental combinations of characters as with the 

 reciprocal combinations. 



The explanation of this, on the " crossing over " hypothesis, is that 

 the factors for colour of flower and shape of pollen grain lie in the same 

 chromosome.' To take the first case {BL xbl) : during syndesis in the 

 hybrid, the BL and the bl chromosomes come together, and separate 

 again in the diplotene stage, during which they become twisted round 

 one another and Hable to exchange segments as described above. There 

 is, however, a greater chance of the BL and the bl factors remaining 

 together than of the combination having been broken up, and conse- 

 quently of B and / being found in one of the separating chromosomes 

 and b and L in the other ; for (i) the chromosomes may have separated 

 without any crossing over of the particular segments containing the 

 flower colour and pollen grain shape factors respecti\'ely ; or (2) if crossing 

 over has taken place in this region, a length of chromosome sufficient 

 to include both factors may have crossed over. It is plainly only when 

 one of the factors has crossed with its mate, and not when both or neither 

 have done so, that a recombination of the B and b with the L and / will 

 have taken place. It is also plain that the nearer to each other the two 

 factors are located in the chromosome, the more Hkely they are to be 

 included in a segment behaving as a unit in crossing over — or, in other 

 words, the less likely is a cross over and rupture of the chromosome to 

 take place between them. Acting on these considerations Morgan and 

 his colleagues (1915 b) have mapped out the arrangement of a large number 

 of factors in the chromosomes of Drosophila, calculating their relative 



1 Such factors are said to be " linked." 



