e 
8 oe the deratnge is ‘cut off while. the 
1897 ] THE LIFE HISTORY OF SALIX 161 
might cause one to lose acell wall now and then. I have exam- 
ined over twenty cases of the first division of the embryo in 
Capsella, and classes in the laboratory have thoroughly examined 
this and other early divisions in the same type, but have found 
no exception to the rule that the first division is longitudinal. 
Hanstein does figure the first division as transverse in Nicotiana 
and Viola altaica, but the figures are not convincing because the 
three nuclei in the large upper cell of his figs. 7 and g of pi. 5 
make it possible to apply the usual interpretation. 
In Salix, as a rule, the second division is also longitudinal and 
at right angles to the first, but it occasionally happens that the 
second division is transverse (figs. g8, 57). Both cases may be 
found in the same species, and in S. cordata and S. petiolaris \ have 
found both on a single plant. In studying sections of embryos 
in these early stages, it is very easy to make mistakes. The 
young walls are often elusive, even in good preparations, and it 
is safest to make the sections thick enough to include the whole 
embryo. The nuclei will then enable one to interpret with cer- 
tainty such stages as figs. 48, 49, 53. 
The third division, usually transverse, but sometimes longi 
‘tudinal, brings the embryo into the familiar octant stage (figs 
53,54). The first transverse division separates the hypocotyl 
and cotyledon portions of the embryo. 
After the octant stage one naturally looks for the periclines 
which mark off the dermatogen, and usually they are found, but 
the embryo sometimes proceeds a little further before this differ- 
entiation takes place. Sometimes a pericline cuts off the der- 
matogen in one octant, while a neighboring octant makes one or 
more divisions before the pericline appears (figs. 56, 59). In 
_Capsella the first pericline usually appears in the upper octants; — 
in Salix I can find no regularity, the first pericline appearing in 
one octant as frequently as in another. The entire dermatogen, 
ere = the ee Sek may be cut off while the 
‘Very rarely, a part — 
ge is s shown in in fg 65, — : 
‘yo is still in the 
