vi COMMERCIAL VARIETIES 107 



"77" are shown in Fig. 42, determined from weekly 

 pickings of five rows apiece, with a P. E. of 10 per cent. 

 for any point in the curves. 



The total yields of these curves work out as follows : 



"77" 



Assili 



Sakel 



Voltos 



(Afifi 



Yield = 25-1 bolls p.p. P.E. 0"6 



19-9 P.E. 0-5 



=18-5 P.E. 0'5 



17'4 P.E. 0-4 



=20-0? P.E. 1-5) 



Thus the three commercial varieties included in the list 

 are practically identical with regard to yield. 



We have already figured the length of the lint of " 77 " 

 under these conditions, and a comparison of this pure 

 strain with similar rows of the two varieties Afifi and 

 Assili should be of interest, one being old and notably 

 irregular, the other new and presumably uncontaminated. 

 The data are plotted in Fig. 49. The P.E. of random 

 single seeds in respect of mean maximum lint-length being 

 6'3 per cent., we find values of 7 '8 per cent, for Assili 

 and 7 5 per cent, for Afifi. 



Thus, on the statistical evidence of lint-length, the 

 newest "variety "is by no means uniform. This con- 

 clusion is abundantly confirmed by other evidence. Thus, 

 the Sakel cotton contains at least two entirely distinct 

 types of seed fuzz ; the presence of these two types might 

 possibly be unimportant, but in all probability it points 

 to a heterogeneous origin for the original plant or plants. 

 Such heterogeneity might occur even in the offspring of a 

 single-plant selection, if the gametes of this plant were 

 not uniform ; in other words, if the plant was not pure- 

 bred. 



Such constitutional irregularity being manifest even in 

 new varieties (see also Fig. 50), our next step must be to 

 examine the cause. So soon as a family becomes 

 heterogeneous, so soon does natural selection begin to 

 operate. Once natural selection has begun, any sequence 

 of alterations is possible. The usual result of such 



