294 BOTANICAL GAZETTE [october 



viable megasporej with three small potential ones above. The two 

 upper spores may either lie side by side (fig. 5); or one over the other 

 {fig. ja)j agreeing in this respect with Eichhornia (Smith). The 

 division line between the upper two cells is very obscure, and as the 

 cells soon show a sign of disorganization the cell wall separating 

 them soon becomes invisible. In Eichhornia, Smith describes two 

 methods of mcgaspore formation from the mother cell; one as I have 

 just described for Hetcranthera, and another in which ^*the elonga- 

 tion of the ovule is accompanied by an elongation of the primary 

 sporogenous cell without division of the latter. . . . The subsequent 

 divisions in the elongated sporogenous cell take place in quick suc- 

 cession.'^ He further says, " From a study of a large number of cases, 

 similar to figs. 14 and 15, I have been led to believe that the four ^ 



mother cells when formed in this w^ay are seldom, if ever, separated 

 by w^alls." I have noticed nothing in Heteranthera to indicate that 

 there are two methods of megaspore formation. Walls are always 

 formed in both first and second divisions, and the divisions occur in 

 all cases as I have described. 



Almost as soon as formed, the large lower megaspore- commences 

 to elongate and encroach on the ones above it, and after destroying 

 them it pushes on through the tapetum and reaches the inner edge 

 of the epidermis (fig. 6). Division of the nucleus now takes place 

 and the embryo sac is soon organized (figs. 6 and 7). The structure 

 of the sac is perfectly normal at this time; the s}aiergids are rather 

 sharply pointed and are decidedly denser than the egg; the antipodals 

 are quite distinct, though small. Smith states that in both Pontederia 

 and Eichhornia the antipodals are ephemeral, disappearing early, but 

 I have found this not to be the case in either Heteranthera or Ponte^ 

 deria. The lower end of the sac where the antipodals rest develops 

 scarcely at all in the subsequent growth of the ovule, and remains as a 



{fig 



The 



and conspicuous, being 



In fig. 8 is represented a sac at about the ti 

 definitive endosperm nucleus is very large 

 in fact as large as the egg or synergids. 



At the stage represented in fig. g a cell wall is present^ cutting off 

 the lower end of the sac into a separate chamber. In this chamber 

 are a number of endosperm cells in addition to the three antipodals 



