1908] _ + GATES—REDUCTION IN OENOTHERA 25 
their elements were separated in the second mitosis instead of the 
first. This view is scarcely admissible for several reasons. In the 
first place, on this hypothesis transverse segmentation of the spirem 
must have taken place not only between the (bivalent) chromosomes 
but also in the middle of each chromosome, in order to give a chain 
of fourteen bodies. Such a segmentation seems unlikely. Another 
possible explanation would be that the chromosomes have lost their 
identity during synapsis, and that the bodies we are dealing with now 
are new arrangements of the chromatic material, irrespective of the 
somatic chromosomes. Many considerations, however, strongly sup- 
port the belief that these bodies really represent the somatic chro- 
mosomes. The facts so far educed in Oenothera, in the opinion of 
the writer, all favor the hypothesis of the separate existence and 
genetic continuity of the chromosomes from one generation to another. 
In this connection may be cited certain plants from the F, of O. lata 
XO. gigas, which as stated elsewhere (14) have 21 chromosomes as 
somatic number, 10 of which regularly go to one pole of the hetero- 
typic spindle and rr to the other. Occasionally, however, the segre- 
gated numbers of chromosomes are 12 and g, one chromosome having 
gone to the wrong pole of the spindle. In this hybrid 7 of the chro- 
mosomes are maternal and 14 paternal. If in this case there were a 
pairing of maternal and paternal spirems, it is difficult to see how it 
could be accomplished and result in the distribution of chromosomes 
in the heterotypic mitosis already stated. 
It will be instructive to compare the chromosome history in this 
cross with the often-quoted condition found by ROSENBERG (23, 24) 
in Drosera longifoliax D. rotundifolia. D. rotundifolia has 10 chro- 
mosomes and D. longifolia 20, as the gametophyte number. The 
hybrid naturally has 30 chromosomes in its sporophyte tissues, 
but in diakinesis 20 chromosome bodies appear, 10 of which are 
double, consisting of a larger and a smaller half, while the remaining 
to are the unpaired (smaller) longifolia chromosomes. The larger 
and smaller halves of the 10 bivalents separate and pass regularly 
ee the poles of the heterotypic spindle, but the unpaired chromosomes 
are irregularly distributed or left out of the daughter nuclei. Later 
the pollen deteriorates. This result is strikingly different from that 
in the Oenothera hybrid, and, while perfectly in harmony with the 
