336 BOTANICAL GAZETTE [November 



The endosperm-nucleus of Piper divides to twenty or more 

 free nuclei, and then simultaneously cell-walls are formed about 

 all of these. In Heckeria the endosperm-nuclei are separated 

 from the beginning by cell-walls. Both genera resemble Peper- 

 omia in that the endosperm is comparatively slight in amount in 

 the ripe seed and contains no starch, an abundant supply of the 

 latter being stored in the surrounding perisperm. 



In the germination of Peperomia and Heckeria the swelling 

 of the endosperm and embryo bursts the seed-coats, and the 

 endosperm protrudes through the rent as a sac which continues 

 to surround the embryo until after root and cotyledons are dif- 

 ferentiated. The embryo is truly dicotyledonous. The root 

 finally pushes out through the endosperm, but the latter remains 

 about the tips of the cotyledons and imbedded in the seed till all 

 the starch of the perisperm is absorbed. 



The present studies have failed to show, in close relatives of 

 Peperomia, any peculiarities in the development of the embryo 

 sac which are clearly intermediate between that found in this 

 genus and that of typical Angiosperms. 



The striking differences in the mode of formation of the 

 endosperm in these three related genera shows again, as has 

 been demonstrated by Hofmeister (1859) and Hegelmaier (1885), 

 that characters of this kind are often of no value as indications 

 of relationship. 



The position taken by the writer in an earlier paper (Johnson, 

 1900^, p. 9), that these peculiarities in Peperomia are secondary, 

 a view supported by Strasburger (1900, p. 293) and Goebel 

 (1901, p. 806), seems still the most reasonable one. 



The case of Gunnera, where Schnegg (1902) has shown that 

 the embryo sac contains sixteen or more nuclei, and that the 

 endosperm nucleus is formed by the fusion of eight or ten of 

 these, might at first sight seem to indicate that we have here 

 again the persistence of a character which Campbell believes 

 to be primitive. A consideration of the evidently distant rela- 

 tionship of these genera, however, together with the fact that 

 no other Angiosperm known shows any relic of this supposed 

 primitive type, makes it probable that Peperomia and Gunnera 

 have secondarily and independently developed this type. 



> \ 



f 



1 



