54} DE. E. E. GATES— CO^'TEIBUTION TO A 



similar to the broad-leaved type derived from the reciprocal cross, as will be seen by 

 comparing PI. 6. fig. 78 with PI. 5. fig. 69. 33ut a narrower-leaved type frequently 

 appears and breeds fairly true in later generations. This type has identical characters in 

 O. LamarcMana X O. biennis and in O. nanella X O. biennis. A typical rosette is shown 

 in PL 6. fig. 79. It is markedly narrow^er-leaved than fig. 78, and comes true to its. 

 difference, yet it does not resemble the vehitina or narrow-leaved type derived from the 

 reciprocal cross [cf. PI. 6. fig. 79 with figs. 70 & 72). Whether, therefore, with 0. biennis 

 as the pollen-parent w^e derive two types diff'ering only in leaf-width (and not in other 

 characters, as do laeta and relutina), or wliether this represents a single type with widely 

 fluctuating variability, I am not prepared to decide. Another cross, O. nanella X 

 O. bien?iis, yielded a large offspring with a continuous range of variation in leaf -width, 

 but agreement in other characters. If there are two types in this cross they difi'er from 

 each other, then, only in leaf- width, and not, like laeta and vekiHna, in a series of other 

 foliage characters. 



Another point w ith regard to these hybrids, 0. nanella X 0. biennis, O. Lamarclciana 

 X O. biennis, and (O. nanella y^O. biennis) X {O. Lama?^ chiana X O . biennis), which arc 

 all identical, is that some of the young seedlings are frequently lacking somew hat in 

 chlorophyll, giving them a whitish appearance. These individuals form smaller rosettes 

 with yellowish centre, but otherwise agree with the type. These rosettes finally, 

 however, " catch up " to the others, and the appearance of two distinct types in the 

 seedlings is seen to be due to the presence of weaker individuals which contain less 

 chlorophyll and hence grow more slowly. I have not determined whether this 

 constitutional weakness is inherited. A fruiting shoot of O. Lamar ckiana X . biennis 

 is shown in PL 6. fig. 80, for comparison with fig. 77 and PL 4. fig. 52. 



That the reciprocal hybrids difi'er from each other, even at an early age, is evident from 

 a comparison of PL 6. figs. 81 & 82, which represent respectively O. biennis X O. nanella 

 and its reciprocal. Comparison with fig. 83, which represents pure O. nanella seedlings, 

 shows that both hybrid types difi'er from O. nanella even at this stage. It also shows 



that O. biennis X O. nanella is at this time patroclinous, most resembling O. nanella, while 

 the reciprocal, w^ith its long leaves and petioles, is nearer O. biennis. 



As shown by the table (VII.), in 0. nanella X O. biennis, and also in its reciprocal, I 

 have obtained dw^arfs in the Fj. This is contrary to the results of DeVries, w^ho found 

 the dwarf character recessive in the Fj, reappearing in the F^. Thus (DeVries, 1908) in 

 O. muricata X O, nanella, the F^ gave laeta and velutina, about 50 per cent, of each, and the 

 laeta type continued to breed true while the velutina split in the Fgand later generations, 

 giving about 60 per cent, dwarf velutina (miirinella) and the remainder normal tall 

 telutina. In the paper above referred to, DeVries concludes that laeta breeds true because 

 its pollen carries only tallness, which is dominant over its own heterozygous egg-cells 

 (tall and dwarf) but recessive to pure "dwarf* cells. The latter, however, is contra- 

 dictory to the fact that in velutina tallness is dominant to dwarf ness, and also leaves 

 unexplained why laeta does not split while velutina does split. Further experiments will 

 be required before this contradiction can be cleared up. 



