Xos. 022-1V23] HYBRIDS IX EGYPTIAN COTTON 497 



than did the corresponding selfed parental progenies of 

 the Pima variety, but wore not significantly more variable 

 than the corresponding Gila progenies. In length of 

 fiber the hybrid F 2 was not significantly more variable 

 than either parent. The hybrid F 2 was signilicantly 

 more variable than the Fj in the leaf characters, but not 

 in the boll characters. This result was so surprising 

 that it was accepted only after repeatedly checking the 

 original data. The averages of the coefficients of varia- 

 tion, for leaf index and boll index, of the eleven F. ; 

 hybrid progenies did not differ significantly from the co- 

 efficients of the F 2 progeny from which they were de- 

 rived. The average variability of the F 3 did not exceed 

 that of the third selfed generation of the more variable 

 parent (Gila) and two of the F : , progenies were not more 

 variable than the corresponding generation of the less 

 variable parent (Pima). 



These facts point to the possibility of obtaining a rela- 

 tively uniform new variety of cotton by hybridization of 

 two varieties belonging to the same general type, al- 

 though hybrids between different types, such as Egyp- 

 tian and Upland, are notoriously difficult to fix. 6 



Distributions 



The distributions, for the important characters leaf 

 index, boll index and fiber length, of the parental and 

 simple hybrid progenies, are shown in Figs. 1 to 3. 



The range of the hybrid F 2 for none of the characters 

 appreciably exceeded the combined parental ranges and 

 for the majority of characters it was more restricted 

 than the latter. The variation was therefore much 

 smaller than in the F 2 of hybrids between less closely re- 

 lated types of cotton, in which the range often greatly 

 exceeds that of both parents. (For example, the Eg] p- 



