i6o THE COTTON PLANT IN EGYPT chap. 



If long and short petals are represented by A and a, and 

 the style lengths similarly by B and b, and there is no 

 mutual interaction, then we obtain 



9 Long petals with long style, AB. 



3 Long petals with short style, Ab. 



3 Short petals with long style, aB. 



1 Short petal with short style, ab. 

 Now if the correlation of the two dimensions within any 

 given flower is almost perfect, we shall obtain such a 

 correlation diagram from the Fj as that shown in Fig. 53 

 (upper). The ab group is almost continuous with the AB 

 group, and is scarcely to be differentiated from it. The 

 aB and Ab groups lie on either side of this diagonal 

 compound scatter. Thus, on working out the value of 

 " r " for the whole table we shall obtain quite a low 

 value, whereas it is in reality a case of almost perfect 

 correlation. Comparing this with the observed result 

 obtained for the King and Charara Fj as shown in the 

 lower half of the figure, there can be no possible doubt 

 that some such grouping is really present. The parents 

 and the Fj are marked on the diagram as A, E, and F. 

 The Fg data of the cross confirm this conclusion, so 

 far as they go, and the probability is high in favour of 

 the view that even the semi-smooth Gaussian curves of 

 error shown by the length of petal, style, column, and 

 filament in the F2 of this cross are (Fig. 60) essentially 

 nothing more than simple 3 : 1 curves, which have been 

 deformed by fluctuation, both autogenous and ordinary. 



With regard to the form of the bract and its width there 

 is nothing definite to be said. Narrow bract is doniinant 

 over wide, just as narrow leaf segments are dominant over 

 wide segments, and narrow bolls over wide bolls. The Fo 

 ranges from one extreme to the other in a modal curve, 

 and pure strains have been extracted, but no details are 

 available by which the stages of this segregation can be 

 traced. 



