22 BOTANICAL GAZETTE [JULY 
There are of course minor differences in these accounts, SCHAFFNER 
(28) stating, for example, that in Liliwm tigrinum there is a splitting 
of granules in the spirem, but the linin thread remains single. Differ- 
ences of opinion are also expressed regarding the arrangement of the 
loops of the spirem before segmentation, and their relation to the 
chromosomes formed. ; 
These two general schemes agree that the heterotypic mitosis is 
a reduction division separating whole somatic chromosomes, while 
the second division is longitudinal. The essence of the distinction is 
that the first view regards the chromosome bivalents as formed by a 
side-by-side union of homologous chromosomes through the medium 
of parallel threads, while the second view holds to an end-to-end 
union. It will be seen that, omitting the points which are left undeter- 
mined, the account in Oenothera corresponds more nearly with the 
latter scheme than with the former, though differing in some respects 
from both. RosENBERG (25), from a comparison of forms having 
long and short chromosomes, has attempted to harmonize the latter 
view with the former. He examined Listera, Tanacetum, Drosera, © 
and Arum, and found that, for example in Drosera, which has short 
definitive chromosomes much like those of Oenothera, the spirem 
first segmented into long twisted chromosomes lying in pairs with 
their long axes parallel. Later, as they condensed into the short, 
rounded definitive chromosomes, they frequently swung around end 
to end, so that an observer seeing only the later stage would conclude 
that they had been arranged tandem on the spirem at the time of their 
origin. Similar conditions were sometimes observed in Listera. 
I think my figs. 22-28 make it evident that this explanation will not 
apply to Oenothera. The chromosomes in Oenothera do not undergo 
any such great amount of condensation, but are already thick, heavy 
bodies when first formed from segmentation of the spirem (fig. 24)- 
Their diameter at this time is about the same as that of the spirem 
just previous to segmentation, as is shown by comparing figs. 22 and 
23 with jigs. 24 and 26. The fact that as many as eight or more 
chromosomes may be found forming a single connected chain (fig. 26) 
also renders this explanation impossible. 
MryakeE (18) finds that after the pairing of elements in synapsis 
(the exact method of this pairing need not be entered into here) in 
