216 Campbell: The embrvo-sac of Pandanus 



forms cannot probably be considered a primitive condition, this 

 does not imply that the embryo-sac structures themselves may not 

 be primitive in character. It is a significant fact that in the great 

 majority of angiosperms the structures of the embryo-sac are not 

 in the least affected by the number of divisions in the sporogenous 

 tissue that precede the formation of the embyro-sac. The great 

 variability in this respect shown by plants of near affinity, as in the 

 often cited case of the Liliaceae, indicates that this is a matter of 

 secondary importance. It certainly seems hardly likely that a 

 structure made up of four megaspores should behave exactly like 

 one derived from the germination of a single spore, and the generally 

 accepted view that in such cases as Peperomia and Lilitim the em- 

 bryo-sac is a single megaspore formed without previous division 

 from the mother-cell can hardly be admitted to have been disproved 

 by these recent speculations. If we admit Coulter^s views as to 

 the compound nature of the embryo-sac of Lilium^ we have still 

 to explain why in Peperomia^ where the course of development up 

 to the beginning of the division of the embryo-sac is precisely the 

 same as that of Liliiim^ there should later be such an extraordi- 

 nary departure from the usual angiospermous type. And if the 

 sixteen-nucleate embryo-sac of Peperomia is derived from the eight- 

 nucleate one of a form like Heckeria^ it remains to be explained 

 why there are sixteen nuclei instead of eight, and why the subse- 

 quent history is so different. In the Liliaceae, where there is an 

 entirely analogous condition of things, the structure of the mature 

 embryo-sac is not in the least influenced by the fact that in some 

 cases there is a division of the mother-cell and in other cases the 

 division is suppressed, and it is hard to see why, in the case of 

 Peperomia^ the development of the embryo-sac directly without 

 preliminary division should result in such remarkably different 

 structures, unless we admit that the sixteen-nucleate stage is an 

 earlier type of embryo-sac which has persisted in a few existing 

 genera like Peperomia and Gwuiera. 



In Gnnnera maerophylla^ Ernst found the early development of 

 the embryo-sac to be very much like that of Peperomia, but there 

 was later developed a definite egg-apparatus of the normal type and 

 a group of conspicuous antipodal cells. In G, //i^;/////^;///, accord- 

 ing to Schnegg, these structures are much less evident. Ernst 



