OENOTHERA LAMARCKIANA. 547 



In order to explain these unexpected figures we may assume 

 that the pollen and the ovules of the race of 0. blandina are more 

 vigorous than those carrying the 0. laeta gametes in 0. Lamarckiana. 

 After fertilization the 0. blandina tubes will come first into the 

 ovary, and the 0. blandina ovules will attract them vigorously. 

 The first copulations will produce almost purely 0. blandina germs, 

 and only later will the remainder embrace the other possible com- 

 binations also. It is evident that such a process may be influenced 

 strongly by external conditions, and therefore that we must expect 

 the deviations to be very variable in amount. 



In analogous crosses of 0. blandina this type not rarely reaches 

 about 90 per cent of the living progeny of the hybrids, and this 

 holds good for almost all the experiments made with this race (4). 

 Whenever the differences between the expected forms among the 

 progeny of a cross are such as to be obvious in the seed pans or in 

 the seedling boxes, the superfluous specimens of 0. blandina may be 

 pulled out before transplanting into the garden, but in other cases 

 they may considerably reduce the number of plants of the desired 

 type on the beds, if space is limited. 



The same predominance of 0. amphivelutina fertilizations is ob- 

 served in those cases where a lethal factor is present, killing the 

 germs within the seeds. The number of barren grains must then 

 be ascertained. 0. mut. oblonga and 0. mut. albida may be chosen 

 as examples. Their formulas are (oblonga' -\-velutina)xvelutina and 

 (albida' -\-velutina)xvelutina. From this we should expect about 

 50 per cent of barren grains in both races, after self-fertilization. 

 The figures obtained from the harvest of 1913 and 1921 for biennial 

 0. oblonga, however, were 57—62—70—82—83, giving an average 

 of 71 per cent barren grains. For the annual plants of the same 

 race (191 1 and 1914) it was still higher. I found 67-75-94-94-95, 

 or together 85 per cent. 0. albida is a race of weaker constitution, 

 and as a rule gives a still poorer harvest. I counted 84—87—89—99, 

 or an average of 90 per cent empty seeds. So it is in numerous other 

 instances, the 0. amphivelutina germs occurring almost always in 

 a greater quantity than would be expected. 



On the other hand, 0. amphilaeta germs are usually produced in 

 too small numbers. It may suffice to give the figures for three of 

 the newest races, the formula of which is (mutant' -\-laeta)x laeta, 

 which would give 50 per cent of living germs reproducing the race, 

 and 50 per cent of laetax laeta, resulting in barren grains. The ob- 

 served figures, however, were for 0. Lamarckiana mut. nitens 20—35 



