502 THE AMERICAN NATURALIST [Vol. LII 



were compared, it was found that the former was 

 significantly larger than the latter, in the indicated direc- 

 tion, for only one pair of characters (width of leaf and 

 number of teeth on the involucral bracts) and in this 

 case the difference was only 3.2 times its probable error. 10 



It was sought to throw further light upon this problem 

 by determining the coefficient of correlation in the hybrid 

 upon the basis of progenies rather than of individual 

 plans. The means, for leaf index and boll index, of 

 eleven P 3 hybrid progenies were used for this purpose. 

 These progenies comprised from 8 to 44 plants each and 

 the means were based upon measurement of one leaf and 

 one boll on each plant. Since the Pima parent has the 

 smaller leaf index and the larger boll index, the correla- 

 tion in the hybrid, if determined by the parental relations 

 of the two characters, should be negative. The coefficient 

 obtained was in fact negative, but was no larger than its 

 probable error (r— — .17 + .18). 11 



The balance of evidence is therefore strongly against 

 the occurrence of coherence of characters in these hybrids 

 between somewhat closely related, although distinct, va- 

 rieties of cotton. It does not, of course, follow that the 

 same result would have been obtained in the case of 

 hybrids between less closely related types, especially if 

 these differ in allelomorphic characters rather than in 

 the variable size and shape characters which chiefly dis- 

 tinguished Pima from Gila. 12 



12 Instances of coherence of characters in hybrids of Egyptian with Up- 

 land cotton have been reported by Cook 1909, pp. It], 17 and 191.",. p. 53;. 



