African and Madagascar Flowering Plants. 389 
convey the pollen, and I have often found its back covered 
by it.™ North West Side Lion’s Head, Cape Town. 
Gladiolus pilosus (Figs. 156-159). 
Very similar to above and probably fertilised by the 
same insect. — Houwhoek. 
Gladiolus inflatus, Thunb. (Fig. 162.) 
The perianth has a remarkable similarity in shape to 
that of the Foxglove, and probably is visited by large bees. — 
Houwhoek. 
Gladiolus longicollis, Baker. 
The perianth ends in a tube 3*5 inches long and the 
flowers are visited by the Spurge Hawkmoth (on the authority 
of Mr. Medley Wood). — Natal. 
Antholyza. 
See Annals of Botany, Vol. IV. p. 277. 
Xridaceae [general remarks ) 1 . 
From the preceding it follows that there is really a very 
strong similarity in Homeria , Ixia, and Romulea , and if we 
were to take a sort of generalised Romulea as ancestor, the 
other genera would naturally group themselves into main 
lines of development, which are represented by Moraea and 
Gladiolus respectively. In the first line of development the 
species of Moraea afford a nearly perfect series of transitions 
from a form not very far removed from Homeria to one 
almost as highly specialised as Iris itself. 
The two main features in this evolution of Moraeas are, first, 
the gradual disappearance of the internal perianth-segments. 
Now Heinricher (Sitz. d. Kon. Acad, der Wiss. Wien, Bd. 
LXXXVII. 1883, p. 114,) points out that this inner perianth- 
circle does not appear till very late in the development of the 
flower (after that of the stamens); hence we can perhaps see how 
1 One can scarcely regard Weismannism as anything but a theory unsupported 
by facts. I therefore maintain the following suggestions, based on undoubted facts, 
which may to a certain extent explain the origin of variations. 
