159 BOTANICAL GAZETTE [MARCH 
lected in August to determine the condition of the buds. In 
some of these the carpels appeared as slight protuberances ; 
others were more advanced and showed the carpels outlined but 
with no trace of ovules. In October buds of S. cordata and S. glau- 
cophylla the nucellus of the ovule was quite conspicuous, but the 
integument had not begun to form. As a rule, the integument 
does not form until spring. Early February buds from a small 
plant of S. cordata showed carpels but no trace of ovules. Material 
from the same plant taken three weeks later showed a conspicu- 
ous integument. Staminate buds, collected in October, showed 
the stamens fairly outlined. The gland, or nectary, is frequently 
conspicuous in the winter buds. 
A diligent search was made for rudiments of floral organs 
which might be expected to be found were the flowers of Salix 
reduced rather than primitive, but an examination of early stages 
in several species failed to show the least trace of anything 
which could be interpreted as a petal or sepal, or as indicating 
an earlier ambisporangiate condition. The prominent nectar 
gland has a single terminal pore. There is nothing in its his- 
tory which would allow it to be regarded as a reduced or trans- 
formed floral organ. 
DEVELOPMENT OF THE MICROSPORES. 
Staminate buds of S. glaucophylla, collected early in October, 
showed the condition represented in fig. z. There are here three 
layers of cells between the epidermis and the spore genous cells. 
The three layers appear alike, the endothecium and tapetum hav- 
ing no distinguishing characters. In some cases there were four 
lavers instead of three. Another specimen of S. glaucophylla, 
collected at the same time, had the tapetum somewhat differ- 
entiated. That buds pass the winter in about this state is 
proved by the fact that buds of the same species collected in mid- 
winter showed the same condition. Material of S. sristis, collected 
late in March, showed the tapetum well differentiated, but the — 
endothecium still appearing like the middle layers (/g- 2). 
The number of middle layers may vary from one to four, even 
-* 
