296 BOTANICAL GAZETTE [October 



appear as in fig, 11. Taken in a .channel it would be the same except 

 for the height of the inner layer of the outer integument. At the time 



m 



of the seed, these large cells are found to be multinucleate, with nuclei 

 and protoplasm in apparently normal activity. In fig. 12 five nuclei 



seem 



are shown in a single section; they 



ordinary mitotic division. Note that, contrary to what might be 



expected, they are on the outer side, where the cell walls are thin. 



PONTEDERIA 



As Smith has followed the early development of the embryo sac 

 in Pontederia, I^^have taken up its history at a point just preceding 

 the formation of the endosperm. In fig. ij is shown a sac just before 

 the division of the definitive endosperm nucleus. This nucleus lies 

 very low, just above the antipodal pocket, and in this position it remains 

 until its division. As a result of the first division, a cross-wall appears, 

 cutting the sac into a large upper chamber and small lower one 

 {fig. 14). The endosperm nuclei in each chamber now divide repeat- 

 edly and the protoplasm increases rapidly in amount. As already 

 mentioned, the lower tip of the sac does not increase in size, and as 

 the part above grows outward, and develops downward at the edges, 

 the antipodal pocket comes finally to hang from the center of a con- 

 siderable depression (fig. 77). The antipodals never divide, but they 

 are not ephemeral, as described by Smith. They are easily visible 

 after cell walls have appeared in the endosperm. 



In the upper chamber of the sac the development of the endosperm 

 presents no great peculiarities; cell formation begins only after the 

 sac has increased in size many times and the embryo has become con- 

 spicuous. In the lower chamber the protoplasm increases rapidly in 

 extent and density (fig. 77), and the nuclei multiply by mitotic division 

 until their number becomes quite large, fifteen appearing in the 

 single section shown in fig. j/. Soon after cell walls begin to appear 

 in the upper chamber, they are also formed in the lower, but in a [ 



much more irregular way. From the first the cells cut out of the 

 basal endosperm are often multinucleate, while those above are 

 uninucleate, as usual, when first formed. Another point of differ- 

 ence is that the cell walls in the antipodal endosperm soon become 



