202 BOTANICAL GAZETTE [SEPTEMBER 
undergo a second division, but this time no cell walls are formed. 
The resulting four nuclei again divide, and the partition wall between 
the two sets of nuclei disintegrates enough to allow a nucleus from the 
lower set to pass through and unite with one from the upper set. This 
fusion nucleus is the endosperm nucleus. If this account is correct, 
we have at least a partial absorption of a cell wall in the formation 
of the embryo sac and the utilization of all four nuclei of the double 
division, as will be described below for Smilacina stellata. 
WIEGAND’s account is contradicted, however, by SCHNIEWIND- 
Tutes (21), who reports that in Convallaria majalis the mother cell 
divides to form a row of four cells, one of which develops into the 
embryo sac, while the other three disintegrate. As a possible explana- 
tion of the difference which exists between her account and WIE- 
GAND’s, she remarks that’ greenhouse material rarely shows normal 
development. WrEGAND, however, does not mention the use of such 
material in his investigation. That both Wrecanp and SCHNIE- 
WIND-THIES are correct is possible, but not very probable. It 1s 
very much to be desired that this species be reinvestigated to clear 
up this confusion. 
ERNST (9) reported that in Paris quadrifolia the lower of two 
daughter cells develops into the embryo sac. The upper daughter 
nucleus divides once, but the resulting nuclei degenerate. In the 
same paper he reports that the lower of two daughter cells of T rillium 
grandiflorum develops into the embryo sac. The upper daughter 
cell is smaller from the first and rarely divides. For T. recurvatum, OP 
the other hand, CHAMBERLAIN (6) has reported that the embry° sac 
develops from the lower of four megaspores. 
Though scanty, the literature on the order Convallariaceae pes 
to suggest that there exists in the group great variation in the condu 
of the cells and the nuclei resulting from the reduction divisions, an | 
that we may here find transition conditions between the 
normal type of embryo sac formation and that found in the lily. 
The materials for this study were collected in the vicinity of Be 
Wis., in May 1906 and 1907, in November 1907, and in June uae 
Flemming’s strong solution gave the best fixation for the early at 
but worked badly for mature embryo sacs. With these ™ ee 
embryo sacs the following chromacetic fixative gave good res 
oit, 
