TWIN HYBRIDS OF OENOTHERA HOOKER! T. AND (i. 289 



among the seeds from the cross and 40 for velutina in the offspring 

 of self-fertilized laeta. The splitting is, in the latter case, unilateral, 

 since there are no empty seeds and since the laeta has the same 

 hereditary qualities in its male and its female sexual cells. This is 

 shown by its crosses with velutina giving the same splitting (46 per- 

 cent velutina). 



4. Oe. Lamar ckianaxOe. Hooker i and the reciprocal cross, which 

 produce the twins laeta and velutina, and whose laeta also split into 

 laeta and velutina in their succeeding generations have no constant 

 laeta in their pedigree. 



5. Oe. franciscana Bartlett, a new species from California, closely 

 related to Oe. Hookeri, repeats in its crosses with Oe. grancliflora and 

 Oe. Lamarckiana the splitting phenomena described for the crosses 

 of Oe. Hookeri. Here also the laeta repeat the splitting after self- 

 fertilization, whereas the offspring of the velutina is constant. 



6. There is an almost complete analogy between the twins of Oe. 

 Lamarckiana and those of Oe. grancliflora and this points to a mass 

 mutation into a velutina-produc'mg mutant for Oe. Lamarckiana. 



7. The same analogy is found for the corresponding crosses of Oe. 

 Lamarckiana mut. velutina (Oe. mut. blandina). 



8. The mass mutation observed for Oe. grandiflora and assumed 

 for Oe. Lamarckiana must lead to a production of their twins in 

 equal numbers and does not explain the splitting of the laeta after 

 self-fertilization. The experimental results give often about equal 

 numbers for the two types, but still more often a deviation from 

 this proportion is observed, which almost always consists in an 

 excess of velutina. This shows the splitting capacity of the laeta 

 gametes to be present already in the original species. 



9. The fecundation of Oe. Hookeri by Oe. biennis gives uniform 

 hybrids of a type called rubiennis, which, in their progeny, split 

 off constant hybrids of the type of Hookeri and repeat this in the 

 succeeding generation. No constant specimens of rubiennis have been 

 found. The heterogamic condition of Oe. biennis seems to afford a 

 sufficient explanation for this unilateral splitting. 



10. If the uniform and constant hybrid races Oe. biennisxsyrticola 

 and Oe. syrticolax biennis are fecundated by Oe. Hookeri or another 

 species, the same type results as from the corresponding binary 

 crosses. This shows that the paternal hereditary qualities are not 

 transferred to or handed down by the maternal side in these 

 crosses. 



11. Oenothera suaveolensxOe. Hookeri behaves just like Oe. Hookeri 



19 



