142 ATKINSON— TWIN HYBRIDS. 



established in the F^ zygote which produces the pycnocarpa twin 

 type, is in a very high percentage of cases stable in the F, and the 

 following generations. But occasionally other reaction systems, 

 usually dormant, are activated, resulting in lamarckiana, and other 

 forms. The lamarckiana twin type presents also a minimum of 

 fluctuating variations in the rosette stage, though it appears to me 

 probable that it is a physiological homozygote and carries in a 

 latent, or inactive condition the other factors of both parents which 

 are not manifest in the phsenotype. 



In the reciprocal crosses of Oenothera franciscana with CE. pyc- 

 nocarpa, only one of the twin hybrids {franciscana type) is fixed in 

 the first generation.'* The other {pycnocarpa type) having a 

 " hybrid constitution " splits in the second generation into two types 

 which are like the twins of the Fj. The pycnocarpa type with great 

 probability would continue to split in the same way in succeeding 

 generations. When the reaction systems of CE. franciscana and CE. 

 pycnocarpa meet in egg or F^ zygote, a new reaction system is es- 

 tablished combining the factors of the two parents. In the unions 

 certain factors of one or the other parent preponderate but their 

 influence is modified more or less by the homologous factors. But 

 the new reaction system established in the F^ zygote is not the same 

 in all the zygotes. As the two dififerent parent reaction systems 

 meet in the &gg, one or the other of two new reaction systems is or- 

 ganized, and chance seems to determine which one of these working 

 systems is established in a given zygote. The reaction system of the 

 lamarckiana twin is stable, that of the pycnocarpa twin is unstable 

 (and heterozygotic). The twin hybrids in these crosses display in 

 the organization of their reaction systems in the F^ zygote what I 

 have termed "selective dominance" (Atkinson, 1917, p. 253). 



I wish here to express my appreciation of aid given by Mr. H. 

 E. Stork, assistant in botany, in writing notes in the field from my 

 dictation, and for making some of the photographs. 



4 The franciscana twin probably carries the pycnocarpa factors also, but 

 in a subordinate or permanently latent condition. If so, it is a physiological 

 homozygote. If it were possible to introduce a splitting factor into the 

 franciscana twin by an appropriate cross, and cause the pycnocarpa character 

 to reappear in some of the progeny, it would indicate the fundamental hetero- 

 zygotic constitution of the franciscana twin. 



