ioo THE COTTON PLANT IN EGYPT CHAP. 



stage of bud-development, whereas the diameter of the 

 boll is partly dependent on the environment in which it 

 exists during the first thirty days after the flower 

 opens. 



It may possibly be owing to this division of boll- 

 maturation into two periods that we find no correlation of 

 seed-weight with any of the boll- characters in pure strains, 

 though such correlation is very evident in heterogeneous 

 F 2 families. 



Seed-weight. The mean weight of the seed is 

 determined by weighing and counting samples of not less 

 than 200 seeds. There is much to be dope yet in deter- 

 mining the causes of fluctuation, by examining weekly 

 pickings. 



The highly irregular " 77 " family of 1911 had a mean 

 at 0*097 gram, with a P.E. of 8 '3 per cent. 



Another family, of twenty-three very uniform plants 

 with a mean at 0'095, had almost as much fluctuation, 

 viz., P.E. 7'0 per cent. 



The weight of a seed is completely determined by the 

 mother plant, and not by the embryo. Thus the weights 

 of single seeds of an F l plant show no modality, nor does 

 the weight of the embryo. Forty such seeds, weighed 

 singly, gave the following fragmentary result. 



Mean seed weight ... (V180 gram P.E. 5'9 per cent. 



testa ... 0-078 ,, P.E. 67 ,, 



,, embryo weight... 0'102 ,, P.E. 5'3 ,, 



Embryo weight = 57 per cent, of seed weight. 



The difference between testa and embryo is slight, and 

 both are equally important in causing fluctuation in seed- 

 weight. 



The causes of this fluctuation are obscure. Correlations 

 have been plotted for all recorded characters, but with 

 very few results. The values for "r" need not be given 

 since a qualitative statement is sufficient for the available 

 data. The only connections yet shown are : Leaf length, 



