172 THE COTTON PLANT IN EGYPT chap. 



diameter, while the Egyptian boll was only 27 mm. in 

 diameter. On crossing the two strains we effectively 

 placed the Egyptian seed inside a boll whose cubic 

 capacity had been doubled, and an increase in seed-size 

 followed. Thus we may regard the Egyptian seed as 

 being constitutionally large, and dominant over the 

 genuinely small Upland seed. At gametogenesis in Fj, 

 the two seed-weight allelomorphs separate from one 

 another, and the 3 : 1 ratio appears in Eg if we clear 

 away the lumber brought in by autogenous fluctuation. 



We have discussed this masking of the difference in 

 seed-weight, and the development of the difference under 

 equal opportunities, in terms of boll- width alone for 

 convenience, though other factors are also involved. The 

 displacement of the means for such dissected graphs, 

 when compared with the probable error of a pure strain, 

 which is obviously too stringent a test for these hetero- 

 geneous groups, gave significant deviations in respect of 

 " discontinuous growth " ; a slight but significant deviation 

 with respect to branching ; a slight indication of connec- 

 tion between extensive shedding and heavy seed, which 

 is probably indirect, since " discontinuous growth " sheds 

 less than the other types ; and a very marked relation 

 between wide or narrow boll and heavy or light seed 

 respectively. In the last case the figures were as follows : — 



36 Widest boUs. Mean seed- weight -f- 9-4% {P.E. x 3-2 = 4-3%) 

 24 Narrowest bolls. ,, ,, - 13-2% {P.E. x 3-2 = 5-3%) 



The simple cross of Sultani x Afifi showed dominance 

 of heavy seed over light in F^ (Fig- 67), and although 

 the difference between the two parents was very small, 

 yet some indication of segregation is shown. This takes 

 the form of two modes in the Fo curve, which appear in 

 both the brother families ; since the mean weights are 

 computed, and hence do not suffer from any subjective 

 error, this coincidence is probably significant, and due to 



