[Vol. 1 



392 ANNALS OF THE MISSOURI BOTANICAL GARDEN 



denticulate, blade narrowing gradually to a very short petiole, 

 midribs and petioles bright red dorsally and ventrally; lower- 

 most bracts 17 mm. in width by 9 cm. in length, upper bracts 

 11 mm. wide by 58 mm. in length. The buds most resemble 

 those of (E. grandiflora, being nearly devoid of long hairs, slender 

 and somewhat rounded, with setaceous sepal tips and some red 

 on the sepals; length of petals 32 mm., hypanthium 43 mm., 

 sepal tips 9 mm., ovary 10 mm. 



In 1912 three famihes of F2 offspring, numbering in all 236 

 plants, were grown from the plant just described. All three 

 families agreed in containing several types exhibiting a remark- 

 able degree of variability. 



An attempt was made to place the plants in five classes, but 

 the categories overlapped and made classification for the most 

 part impossible. The majority of the plants resembled the 

 parent individual in their main features but they varied enor- 

 mously in width of leaf from broad (21 mm.) to very narrow 

 (8-6.5 mm.). These conditions were connected by interme- 

 diates, and, moreover, there were considerable variations within 

 the indixddual, one branch with very narrow leaves being found 

 on a plant with broad leaves. In addition to these variants, 

 the three families contained 35 dwarfs, or 14.8 per cent, and 

 the latter varied in leaf-width in the same remarkable manner. 

 The dwarfs agreed only in having short internodes. Two of 

 them are shown in pi. 21 figs. 13, 14, the former having narrow 

 leaves and extremely short internodes, the leaves of the latter 

 being quite Unear. The plant would never be taken for an 

 Oenothera. 



The advent of a large percentage of dwarfs in this family is 

 similar to their occurrence in other CE. grandiflora races from 

 that locality (see Gates, '14, p. 246). The precise manner in 

 which this capacity for producing dwarfs is inherited, is a diffi- 

 cult question which need not be considered here, particularly 

 as it has been discussed elsewhere (Gates, '14). 



Plate 22 fig. 15 represents one of the LamarcHana-like 

 rosettes from this source, grown in 1909. Others approached 

 de Vries's race more closely, to the point of identity. Plate 22 

 figs. 18, 19 represent selected rosette-leaves taken from this cul- 

 ture to show the range of types exhibited. Such leaves as the 



