372 Stephens . — The Embryo- Sac and Embryo of 
type, namely, the absence of close affinities to other types in the same 
region, and the pronounced adaptation it shows to xerophilous conditions — 
an adaptation which extends even to details of the embryonic structure, 
and must have been the result of a prolonged exposure to such conditions. 
Additional weight is lent to this view by the fact that in all three genera 
investigated the embryo-sac shows the peculiar structure that has been 
described ; these genera then would appear to have been evolved since the 
establishment of this type of sac. In considering the morphological bear- 
ing of any deviation from the normal angiosperm life-history shown by 
such a group there are two possibilities to be kept in view ; they may 
be relatively primitive features retained by it, or recent developments of no 
phylogenetic significance. The former view would seem more probable 
on the preceding facts, and it would be quite legitimate to regard 
this sixteen-nucleate embryo-sac as more primitive than the usual eight- 
nucleate type if it were formed, as is usually the case in the latter, from one 
of four megaspores. Indeed, although this is not the case, it is still quite 
possible to consider that the reduction division by which these megaspores 
are formed has here been shifted, and takes place at the germination of the 
spore — that is, the megaspore mother-cell functions directly as the embryo- 
sac, and all the nuclei formed in it belong to the gametophyte. This view 
is held by Ernst, who has discussed 1 this point for the sixteen-nucleate 
embryo-sacs of Gunner a, Peperomia , and the Penaeaceae. In all of these 
the formation of walled megaspores is omitted, and this is a strong argu- 
ment in favour of regarding them as derived rather than primitive forms. 
If they were really primitive in respect of their embryo-sac structure, the 
normal row of four megaspores might be expected. Ernst, however, 
arguing from such cases as that of the Liliaceae, where tetrad-formation 
may be partly or entirely suppressed within the limits of a single order 
without influencing the development of the embryo-sac, considers that these 
processes of development that go on within the embryo-sac are quite 
independent of its origin, and can be considered apart from it. He points 
out that the divisions within the macrospore mother-cell are the last divi- 
sions within a macrosporangium, while those within the embryo-sac repre- 
sent the germination of one of its derived spores, and concludes that these 
two processes can undergo various grades of reduction quite independently 
of one another ; thus if the two divisions forming the spore-tetrad are 
omitted, chromosome reduction will be shifted so as to take place within 
the spore at its germination . 2 This he considers has happened in the cases 
under discussion. 
But on the other hand, sporogenesis is a normal phase in the life- 
history — the culminating point of the sporophyte generation — and some 
1 Ernst, 1908 (B), pp. 26-9. As regards the Penaeaceae, Ernst was misled by my former 
statement (see ante, p. 364) that three megaspores appeared to be formed. 
2 See also Farmer, 1907, pp. 192, 196. 
