386 
DUNCAN S. JOHNSON 
of this micropylar group soon become surrounded by denser masses 
of cytoplasm and by cell walls, to form the egg and the single synergid. 
The other 2 nuclei of this group and the 12 of the other three groups 
later form a compact group near the middle of the short flask shaped 
embryo sac. This is the mature sac ready for fertilization. There 
are no 8-nucleate mature embryo sacs in this species as Coulter 
(1908, p. 363) has misunderstood me to state in my preliminary 
account of this species (Johnson, 1907). After fertilization, the 
occurrence of which is indicated by the presence of 2 nucleoli in the 
egg, the 14 nuclei of the central group fuse completely to form a 
huge, lobed, endosperm nucleus. Thus while the egg and synergid 
both come from one of the 4 megaspore nuclei the descendants of 
all 4 megaspores enter into the formation of the endosperm nucleus, 
making the latter the only really composite element in the embryo-sac. 
From what has just been said it is evident that the striking 
peculiarities of the development of the embryo sac here are the 
occurrence of but 2 divisions of the megaspore nucleus instead of 
3, the participation of all megaspores in the formation of a single 
sac and the organization of the endosperm nucleus by the fusion 
of so large a number of nuclei derived from megaspores. It seems 
perfectly clear, since the first 4 nuclei formed in the embryo sac are 
certainly megaspore nuclei, that the mode of megaspore germination 
here shown is not a primitive one but a specialized type reduced from 
one with 3 divisions such as occurs in other angiosperms. In the 
second place the single multinucleate prothallus or embryo sac 
formed by the cooperation of 4 megaspores has no analogue among 
lower seed plants or the pteridophytes, and therefore cannot possibly 
be regarded as primitive (see Campbell, 1899, p. 455, 1901, pp. 113-117, 
1902, p. 781), or as recalling the many-celled prothallus of the gym- 
nosperms and pteridophytes, which is everywhere developed from a 
single megaspore. Finally the formation of the endosperm nucleus by 
the fusion of 14 nuclei from 4 megaspores has no homologue outside of 
a few other peculiar angiosperms {Gunner a, Penaeaceae) and is pretty 
certainly a later specialization rather than a primitive type from which 
the typical endosperm nucleus, formed by two nuclei from a single 
megaspore, may have been derived. There is nothing about the de- 
velopment of this endosperm to suggest the possibility of its being a 
modified second embryo as suggested by Miss Sargant and others. 
It is to be recalled that Porsch has pointed out that no resemblance 
