20 CALIFORNIA ACADEMY OF SCIENCES. [Proc. 30 Ser. 



cells; but from comparison with other more typical cases 

 (fig. 39), it seems more probable that the three lower cells 

 were derivatives of the inner cell. Another exceptional 

 case is shown in fig. 41, where the archesporial cell has 

 formed vertical walls as well as transverse, and the whole 

 archesporial complex is nearly globular, instead of the usual 

 row of cells. In all cases, it is the lowest cell of the series 

 which finally becomes the embryo-sac. The number of 

 specimens examined was not sufficient to decide positively 

 whether the inner of the two primary archesporial cells may 

 be transformed directly into the embryo-sac, but it is cer- 

 tainly not the case generally. As a rule, the mother-cell 

 divides into two, and the upper cell again into two (fig. 39). 

 Of the three cells thus formed, the lower one is the larger, 

 and its lower end more or less pointed. Soon, by its growth, 

 it destroys the others, so that it finally occupies the whole 

 space filled by them, and only traces of the latter can be 

 seen between it and the tapetal cells (fig. 42). The further 

 development of the normal embryo-sac differs in no wise 

 from the familiar type of the Angiosperms. The primary 

 nucleus divides into two, one going to either end of the young 

 embryo-sac (fig. 42). Each of these divides again (fig. 43), 

 and by a further division there result, as usual, four nuclei 

 at each end. These nuclei stain strongly, and are embedded 

 in granular protoplasm, which is absent from the central 

 part of the embryo-sac. 



Of the four nuclei at each end, which at first are much 

 alike, one grows rapidly in size and forms the polar nucleus. 

 The two polar nuclei then move toward the centre of the 

 embryo-sac, or the lower one may remain nearly stationary. 

 These polar nuclei (fig. 50) have a large and conspicuous 

 nucleolus, but the writer was not successful in clearly dem- 

 onstrating the centrospheres, which Schaffner (1896, p. 

 125) states are very easily seen in Alisma. The antipodal 

 nuclei are small and stain strongly; they become surrounded 

 by definite cell walls, in which respect they differ from 

 Alisma (Schaffner, 1. c), but agree with most of the Mono- 

 cotyledons (Goebel, 1887, p, 386). I could observe no 



