316 BOTANICAL GAZETTE [APRIL 
with JuEL’s (12) counts for 7. officinale. The count registered 
by Stork for the species now under discussion is twenty-six to 
thirty. Since the latter takes no cognizance of fission at any time 
before metaphase, it is possible that his higher estimate is due to 
reckoning separated halves as units. It will be recalled that the 
somatic number is about twenty-six, and that there are about 
thirteen pairs of prochromosomes. Fig. 22 shows a normal reduc- 
tion division of thirteen univalents at the homotypic plate. These 
facts all give the necessary assurance that in prophase we have 
the origin of the diploid number of univalents, unpaired, from a 
dual and therefore a split thread. 
To summarize developments thus far, there is first the appear- 
ance of approximately thirteen (the haploid number) pairs of 
prochromosomes. The thread entering synizesis shows in places 
a doubleness unexplainable at present. The thread emerging from 
synizesis becomes very evenly distributed through the nucleus, 
and then shows what is interpreted as non-simultaneous splitting. 
By the time segmentation is reached splitting becomes indubitable, 
and the formation of twenty-six univalent chromosomes occurs by 
the lateral refusion of the two halves previously split apart. 
In contrast with these findings it should be noted that JUEL, 
Osawa, and Stork, working on parthenogenetic species of Taraxa- 
cum, all expressly state that the postsynizetic thread is single 
and remains so, and that the univalent chromosomes are single in 
composition. JuEL and Osawa, working on sexual plants of the 
same or nearly related genera, report an obvious doubleness of the 
thread. Since there is no question of the duplex nature of bivalent 
chromosomes in sexual plants, these investigators conclude that 
the doubleness noted is due to synaptic pairing. The three 
workers cited agree that diakinesis is followed by the greatly elon- 
gated nucleus as mentioned. Stork, however, considers the 
chromosomes here to be unpaired, that is, unsplit because “‘ there 
are certainly not upward of sixty.”” Accepting his maximum count 
of thirty as correct, one would scarcely expect to find more than 
sixty halves. 
Comparison of figs. 12-15 with figs. 36-41 strongly suggests 
that the elongated nucleus is not the outcome of diakinesis, but 4 
