THE AMERICAN NATURALIST [Vol. XL VIII 



little by environment, and number of leaves and size of 

 corolla were the only ones that satisfied this requirement. 

 Such character differences as height of plant and size of 

 leaf, while undoubtedly transmissible, are influenced so 

 strongly in their development by nutrition that work with 

 them is exceedingly difficult. For example, if a certain 

 variety of Nicotiana tabacum is grown under the best of 

 field conditions, the longest leaves are about 28 inches and 

 the total height about 6 feet, but a portion of the same 

 seed fraternity may be grown to maturity in 4-inch pots 

 without reaching a height of over 16 inches or having 

 leaves longer than 4 inches. On the other hand, several 

 experiments conducted in the same manner have shown 

 no difference between the frequency curves of variation 

 in number of leaves or of size of corolla, whether starved 

 in small pots or grown under optimum conditions. The 

 character complex number of leaves was chosen for this 

 investigation rather than the size of corolla because vari- 

 eties that differ greatly in number of leaves are common. 



TABLE I 



(Compare with frequency 



bution under normal field com 

 usetts, in Tables VII and XI) 



Previous Work of the " Havana' 1 X ''Sumatra" Cross 

 Several crosses have been made between varieties of 

 tobacco that had a mean difference of seven or eight 

 leaves, but the majority of the data reported here were 

 collected from the descendants of a cross made by A. D. 

 Shamel between the types known in Connecticut as 

 ' ' Havana ' ' and 1 ' Sumatra. ' ' The ' 4 Havana ' ' parent was 



