422 BOTANICAL GAZETTE [DECEMBER 
cause a segmentation which is so conspicuous that it is seen easily 
with a pocket lens (figs. 5, 6). 
The evanescent walls were, for a time, the most puzzling feature 
in the embryogeny, especially while it was thought that they 
belonged exclusively to some particular mitosis. As observations 
multiplied and it became evident that the evanescent segmentation 
might occur at the sixth mitosis and was increasingly frequent 
at the seventh, eighth, and ninth mitoses, the explanation became 
evident. The early mitoses, up to the sixth, follow in such rapid 
succession that a new mitosis is begun before the preceding one 
could establish a wall. Between the succeeding mitoses the 
intervals are increasingly greater, and consequently the natural 
tendency of a spindle to develop a wall finds a more marked ex- 
pression. 
We regard the appearance of evanescent walls as a reversion, 
indicating that the ancestors of Dioon had embryos which were 
cellular throughout. It can hardly be questioned that the free 
nuclear condition in gymnosperms, whether in the proembryo or 
in the gametophyte, is not a primitive but rather a highly special- 
ized condition. In the small eggs of the pteridophytes, including 
even the heterosporous genera, the first nuclear division is followed 
by the formation of a wall. We suggest that free nuclear division 
in the proembryo of gymnosperms arose in connection with the 
increasing size of the egg, the increased mass of the egg becoming 
too large to be segmented, and at the same time the large mass 
favoring rapidly succeeding mitoses. The nuclei divide simul- 
taneously because exposed to practically similar conditions. As 
mitosis continues and the mass of cytoplasm about each nucleus 
becomes less, the mitoses succeed each other less rapidly, and 
as the relation between the nuclei and surrounding cytoplasm 
approaches that which is found in the ordinary tissues, segmenta- 
tion begins. If this view is correct, we should expect the most 
extensive free nuclear periods in the largest eggs, and such is cer- 
tainly the case. The egg of Dioon edule often reaches a length 
of 5 mm., and the number of free nuclei is about 1000, theoretically 
1024. Zamia, with an egg about 3 mm. in length, has 256 free 
nuclei; and Gingko, with an egg of somewhat approximately the 
SORES ING Tes Vel nt IDE See aaa eset 
