1912] STEVENS—HETEROSTYLOUS PLANTS 301 
DIAKINESIS.—In either case the diakinetic gemini arise by the 
continued shortening of the paired loops which make up the strep- 
siténe stage. As shown by fig. 49, diakinesis may not arise simul- 
taneously throughout the nucleus. The stage is of considerable 
duration, however, and presents a very characteristic appearance 
(figs. 50 and 51). The chromosomes of a geminus are generally 
united only at one end and often diverge widely from one another. 
The nucleolus at this stage appears vacuolate, and shows in section 
an outer dark staining region surrounding an inner almost color- 
less portion. Commonly at this stage one or two gemini appear 
clinging to the nucleolus (fig. 51). Their appearance suggests the 
condition reported by DARLING (5, p. 186) for Acer Negundo, where 
five chromosomes appear to arise directly from the nucleolus. 
There is nothing in Houstonia, however, to indicate that this con- 
dition is anything more than a clinging of the gemini to thenucleolus. 
Another characteristic appearance during diakinesis is that 
shown in figs. 52 and 53, where several gemini appear clinging 
together in a single row. This condition very much resembles that 
found by Gates (15, p. 12) in Oenothera. GATES, however, con- 
siders that in Oenothera this condition is previous to true diakinesis, 
and represents a single continuous spirem constricted at regular 
intervals to form a chain of chromosomes. Some of these chro- 
mosomes afterward pair to form typical gemini, but a considerable 
number of them are apparently taken up by the heterotypic spindle 
without having previously paired. An essentially similar method 
of formation of the diakinetic chromosomes has been reported by 
GEERTS (16, p. 610) and by Davis (12, p. 559) for Oenothera, and 
by YAMANOUCHI (41, p. 186) for Fucus. 
That such is the case in Houstonia, however, seems improbable, 
as such an interpretation makes it necessary to regard the condition 
shown in figs. 46 and 47 as a precocious split which afterward closes 
up. Moreover, the two members of each geminus in Houstonia are 
almost always found attached at one end, while in Oenothera such 
a condition is the exception. The condition shown in figs. 52 and 
53 seem to be best explained as a temporary union of independent 
gemini, an interpretation first suggested by MivaKE (25, p. 96) 
for a similar appearance in Galtonia candicans. 
