No. 5S7] SHORTER ARTICLES AND DISCUSSION 



:,non 



The constants are very irregular in magnitude, but are with- 

 out exception positive in sign. In many instances they are large. 

 Thus in these individual shrubs which taxonomically show no 

 differences 13 there is nevertheless a distinct differentiation in 

 respect of the great majority of the characters examined. 



While the probable errors are large the evidence warrants the 

 conclusion that some are decidedly more highly correlated than 

 others. 



ule. For method of computation see Biometrika, Vol 

 VII, pp. 477^78, 1910, and Amer. Nat., Vol. XLVI, p. 480, 1912. 



number of ovules. See citations above. 



Fertility = coefficient of fertility (mean seeds per fruit) (mean ovule: 



!3 I believe one of the Hibiscus shrubs had lighter flowers than the rest 



