and of other related Primula Hybrids . 
377 
P. floribunda isabellina x P. kewensis (type). 
The hybrid derived from this cross resembles the seed parent, P. flori- 
bunda isabellina, and possesses the 1 8 (2 x) and 9 (x) chromosomes. 
The reciprocal cross, P. kewensis (type) x P. floribunda isabellina , set 
no good seed. 
P. floribunda isabellina x P. kewensis (seedling form). 
An unexpected result is obtained from this cross. The hybrid re- 
sembles that of the preceding cross, P. floribunda isabellina crossed with the 
sterile P. kewensis , in external features as well as in the number of chromo- 
somes, which are accordingly 18 ( 2 x) and 9 (#), notwithstanding the fact 
that one of the parents possesses the doubled number, 36 (2 x) and 18 (x). 
It is difficult to imagine by what regulating process this reduction in 
the number of chromosomes has been achieved; that is to say, how 18 [x] 
uniting with 9 (x) can give the result of 18 ( 2 x). Once more the Oeno- 
theras show an analogy, and offer a possible explanation for this phenomenon. 
O. lata , which has 14 ( 2 x) and 7 (x) chromosomes, crossed with O. gigas , 
which has 28 (2 x) and 14 (x) chromosomes, gives rise to a hybrid which has 
21 (2 x) chromosomes. In the heterotype prophase the chromosomes 
appear unpaired in their 2 x numbers. Subsequently, as in the Drosera 
hybrid ( 34 ), Geerts (20) finds that the homologous chromosomes derived 
from either parent pair with one another and that the supernumerary 
chromosomes remain unpaired. In the Oeiiothera hybrid, therefore, 7 bi- 
valent and 7 univalent chromosomes go on to the spindle. The 7 bivalent 
chromosomes separate normally, whilst the 7 univalent ones disintegrate. 
Some of the fragments may eventually reach the poles, but many may be 
left on the spindle and never enter the reconstructing nuclei. Geerts 
obtains the same results in the hybrid resulting from the cross between 
O. gigas and O. Lamarckiana and in the reciprocal cross of O. Lamarckiana 
x O. gigas. In the polar views of the asters of the heterotype and 
homotype divisions of both these hybrids there are 7 distinct and perfect 
chromosomes to be seen, and amongst them many fragments. In the 
second generation of the cross between O. gigas and O , Lamarckiana 
the number of 2 ^ chromosomes has reverted to the original 14. Thus 
■ wenn die Chromosomen wirklich die Trager der erblichen Eigenschaften 
sind, geniigen also 7 Chromosomen der O. gigas fur die Vertretung aller 
erblichen gigas- Eigenschaften in diesen Hybriden ’ (p. 163). 
It may be that the hybrid Primula makes use of some similar means, 
perhaps at the first segmentation division, whereby it eliminates its super- 
fluous chromosomes. 
Further interesting results have been recorded from crossing O. gigas 
with O. lata and O. Lamarckiana. De Vries (6) has found that when 
