KNOWLEDQE OF THE MUTATING OENOTHERAS. 45 



show the range of types represented *. The other rosettes approximated more or less 

 to these, though no two were quite identical. The variability, both of this and an 

 independent outdoor culture the following year, was far greater than DeVries's race of 

 O. LamarcTciana exhibits. Pig. 59 represents a rosette (No. 5) belonging to the 0. lata 

 race already described (p. 36) as originating from this locality. Its leaves differ 

 constantly from the 0. lata of DeYries {cf. PL 3. fig. 42). PL 4. fig. 44 shows another 

 rosette of this O. lata race as grown in outdoor culture. The greater elongation of the 

 blades and petioles is the feature distinguishing both from the race of DeVries. The 

 buds of the two have already been contrasted (PL 3. fig. 43). The individual (No. 5j 

 represented in PL 6. fig. 59 produced a somewhat fasciated stem with straggling, 

 irregular branches, probably a result of the tropical environment. Its rosette closely 

 resembled No. 6, which was elsew^hcre described (Gates, 1912, pi. 2) as continuing to 

 form rosette-leaves without internodes. No. 5 was self-pollinated, and its Pi grown 

 out-of-doors in the usual way yielded the following interesting series of types. Three 

 of the offspring produced rosettes like their parent, i. e. they were Za^a-like, but the 

 flowers were as small as in 0. biennis and the anthers surrounded the stigma in the 

 bud. The buds bore numerous muricated hairs, and the bud-cones were somewhat 

 barrel-shaped (type 1). To the next type belonged 7 plants, a very distinct type 

 represented in PL 5. fig. 66, having very narrow, almost elliptical, whitish leaves of 

 uniform width (type 2), This rosette agrees with DeVries's 0. elli^tica {cf. * Mutation 

 Theory,' i. 394, fig. 83), but the flowers were small with the anthers surrounding the 



stigma. Tw^o other plants (type 3) resembled most those jast mentioned, but were 

 dwarfs. Pour plants belonged to another type (PL 5. fig. 67) having small flowers in 

 which the anthers were almost or quite touching the base of the stigma-lobes (type 4). 

 This unexpected appearance of short-styled flowers w^ould indicate an ancestral cross 

 with O. biennis. I have, however, never discovered O. biennis in my cultures, numbering 

 several hundred plants, from this locality. Of course, there remains the possibility that 

 the tropical conditions under which the fertilization and development of the seeds of the 

 mother plant took place may have served to call forth " latent " characters in the race, 

 which it had acquired through previous crossing. 



However, I grew outdoors an independent culture from a second sowing of Lancashire 

 seeds, and these revealed (among many types) one constant type with flowers inter- 

 mediate between O. biennis and O. Lamar chiana , so this furnishes very good evidence 

 that 0. biennis has, at a previous time at any rate, formed a part of this colony. The 

 behaviour of three races which belonged to 0. grandijlora, though diifering somewhat 

 among themselves, may be summarized here. One race (54x32) was typical rather 

 broad-leaved O. grandiflora, agreeing closely with the Alabama form. The Pi contained 

 11 plants and the Pa 82 plants which bred perfectly true. Another race (No. 40), the 

 Pi from one plant, contained 43 individuals. This race also belonged to 0. grandifloray 

 though differing in minor points from the last, and it was much less uniform. Several 

 plants omitted the rosette stage and 5 were dwarfs. One of the dwarfs produced in the 



* It will be observed that iu hothouse cultures the number of rosettc-leaves is always creator than out-of-doora. 



