Bot.— Vol. I.] CAMPBELL— SPARG A NIUM. 307 



larger than in S. simplex, and two are placed below the 

 third one, and lie in the depression at the base of the embryo- 

 sac. This is much narrowed, and may account for the fact 

 that sometimes these two lower antipodal cells take no part 

 in the formation of the group of antipodals formed after fer- 

 tilization (fig. 36). In such cases they remain for a long 

 time quite unchanged, and ultimately show signs of disor- 

 ganization. The upper of the three original cells, however, 

 enlarges greatly and assumes a vesicular form, projecting 

 strongly into the cavity of the embryo-sac. The nucleus of 

 this cell, as in the older antipodal cells of S. simplex, en- 

 larges very much, soon undergoes division, and this is 

 quickly repeated, so that four cells, arranged quadrant-wise, 

 result. Where this was the case division-walls could be 

 seen between the nuclei. Sometimes, however, the lower 

 antipodal cells also enlarge (fig. 35) and the development 

 proceeds much as in S. simplex, except for the relative po- 

 sitions of the cells. How far the antipodal cells may finally 

 develop in S. Greenii could not be ascertained, as the next 

 stages were absent from my material. So far as could be 

 judged from the few stages examined, the cells show a 

 tendency to become more inflated than in S. simplex and 

 the granular contents are less abundant, but these peculiar- 

 ities, which seem also to characterize S. eurycarpum, may 

 not be constant, and a study of the later stages in these 

 species will be necessary before the question can be defi- 

 nitely decided. 



V. The Embryo. 



The development of the embryo does not, at first, keep 

 pace with the development of the antipodal cells and endo- 

 sperm, and for some time it remains without any apparent 

 change in appearance. Although a large number of sec- 

 tions were made, it was impossible to find stages between 

 the unicellular embryo and such advanced stages as that 

 shown in fig. 37. It is therefore impossible at present to 

 compare the first divisions in the embryo of S. simplex with 

 the corresponding stages of that in S. ramosum, the only 



