VIII HEREDITY 139 



mediates break in at least two different ways. Probably 

 there are again two allelomorphic pairs concerned. 



Another series of crosses gave an entirely different set 

 of ratios. One of these was derived from an Egyptian 

 parent which possessed an inconspicuous petal spot, and 

 the divergence from expectation was at first ^ attributed to 

 fluctuation. The ratio in Fg was 5 : 9 : 47. The other 

 series was not open to this interpretation, since the 

 Egyptian parent had a fully-developed petal spot, while 

 the F.2 ratio was 27:41 :112. If we assume that the 

 petal spot in the latter cross was controlled by two 

 allelomorphic pairs, with 3:1:1:3 gametic coupling, 

 while the " full spot " class consisted of two groups which 

 were practically indistinguishable, we obtain a theoretical 

 ratio which is very close to actuality, viz., 25 : 40 : 115. 



It is inadvisable to discuss the point more fully, since 

 we require much larger series of numbers than those the 

 author can present, before a decision can be taken as to the 

 probability of this view. The consistently erratic 

 behaviour of the spotted and intermediate forms in F3 

 supports such interpretation, but the figures available are 

 not sufficient to substantiate it. 



The only decisions at which we can arrive with certainty 

 are : — that the presence or absence of the petal spot in 

 these iEgypto-American crosses is not determined by a 

 single pair of allelomorphs, and that there is strong 

 evidence for complication of this mechanism by gametic 

 coupling in some cases. 



Anther colour. — The colour of the central brush of 

 stamens is an important feature in the general colour- 

 effect of the cotton flower. In Egyptians it is bright 

 golden -yellow, in Uplands and Hindi it is whitish, or 

 rather buff in colour, and the Fj is intermediate. 



In F2 we obtain ratios which appear to be genuinely 

 1:2:1, and no exceptions have yet been noted in sub- 

 sequent generations. The characteristic thus appears to 



