154 BOTANICAL GAZETTE [AUGUST 
comparatively heavy spirem (fig. 20), and in some instances at the 
beginning of segmentation fragments of it still remain (fig. 21). 
In cases where the separating nuclear membranes do disappear 
early, the limits of the nuclei are readily followed (fig. 19). GuIG- 
NARD (7), in his first report of double fecundation, says that the 
chromatin of the sperm enters into more or less of a spirem before 
fusion with the two polar nuclei, after which, at times, he is still 
able to recognize the triple origin of the secondary nucleus of the 
sac. None of his drawings are later than fig. 18 of the writer, and 
apparently the chromatin is in the same condition. In a later 
paper (8) he describes a similar condition in Narcissus. 
In Fritillaria, Sax (16), after stating that the chromatin of the 
male nucleus frequently passes into a spirem previous to the 
breaking down of the separating membranes, and in some few 
instances observing the beginning spirem in the polar nuclei, con- 
cludes that there is a complete fusion of the chromatin contributed 
by the three nuclei, and that this is further proved by finding no 
incomplete fusions in later stages. 
In not a single specimen showing the formation of the primary 
endosperm nucleus was the writer unable to distinguish between 
the chromatin of the various nuclei that have contributed to this 
nuclear complex. From the view obtained in fig. 22 it could readily 
be concluded that a complete intermingling of chromatin has previ- 
ously occurred, but when the next section of the same primary 
endosperm nucleus is examined (fig. 21), such a conclusion is seen to 
be groundless. 
Whether or not the separating nuclear membranes have entirely 
broken down by the time of segmentation, the spirems remain dis- 
tinct, and, following segmentation, three groups of chromosomes 
collect upon the three arms of the tripolar spindle (figs. 23, 24). 
In none of the literature examined has such a stage been shown 
or reported, for, if such had, the idea of complete fusion or inter- 
mingling of chromatin material could not have been adhered to up 
to the present time. 
Since the chromosomes are very long and quite numerous (36), 
the writer was unable to follow definitely their final arrangement 
upon the bipolar spindle. From the appearance of fig. 25 it seems 
