ON TWIN HYBRIDS. 477 



the characteristics of their male grandparent. The 0. brevistylis 

 ciiaracter is thereby shown to be independent of the hereditary units 

 which cause the production of the twins. 



The third and fourth generations have been cultivated in only one 

 case, as yet, wMiich was a cross made between 0. miiricata and the 

 pollen of 0. Lamar ckiana in 1901. The 0. laeta in this case was 

 entirely sterile, and so only the 0. velutina could be studied. It gave 

 a constant progeny of small extent, the fertility in artificial self- 

 pollination remaining in a diminished condition. 



Table ill 



Constancy of 0. velutina during four generations 



Cross of 0. muricata x 0. Lamarckiana 



Generation 0. laeta 



Cross, 1901 — 



1st generation, 1902 35 (sterile) 



2nd generation, 1903--4 — 



3rd generation, 1905 — 



4th generation, 1907 — 



All in all 175 plants were observed, all bearing the same characters. 



The phenomenon of twin hybrids seems to be very rare in plants. 

 There can be no doubt that its occurrence in Oenothera Lamarckiana 

 is intimately connected with at least part of the special manner in 

 which this species displays its mutability. The production of some 

 of the mutants seems to be quite dependent upon it, though that of 

 others is independent. But I must reserve an account of my experi- 

 ments on this point for another occasion. 



SUMMARY. 



1. Hybrids between members of the group Onagra or the 0. 

 biennis group of the evening primroses are, as a rule, constant in 

 succeeding generations. An exception is 0. brevistylis, the char- 

 acter of which follows Mendel's law. 



2. The reciprocal hybrids of this group are, as a rule, unlike one 

 another, and mostly patroclinous. 



3. In many cases, where 0. Lamarckiana or one of its derivatives 

 is the father, two forms of hybrids are produced instead of one. These 

 forms may be called twin hybrids. 



4. One of the twins is broad and smooth-leaved and is called 0. 



