considered in Relation to the Phyllode Theory. 319 
I think that the comparison with other Irids supports this interpretation. 
In the upper part of the limb (Fig. 48 j) the invaginations have become so 
deep that, if we only knew the leaf in this region, its phyllodic character 
might easily be overlooked. 
The sheath of Gladiolus tristis , with its ridged and furrowed dorsal 
surface (Fig. 48 f), is an instance of a type of structure which occurs 
repeatedly in this family, and to which I shall refer again (p. 332) when 
considering the general question of the evolution of the Irid leaf. 
5. Fibrosis in Irids and Acacias. 
An exaggerated degree of development of the fibrous strands in the 
leaf is characteristic of many of the Iridaceae. In certain cases the survival 
of these strands, especially in the basal region, provides a corm-sheath, which 
is sometimes continued upwards as a tuft of fibres. The most striking 
instance I have seen is Antholyza nervosa , Thunb., in which, in the case of 
a specimen at the flowering stage in the British Museum herbarium, the 
fibres are as much as 28 cm. long. 
In the individual leaf, the fibrous strands generally run parallel to the 
bundles, between the phloem and the leaf-surface, bift they may also 
surround and enclose the bundles. In the ensiform leaf-types they are 
most conspicuously developed at the dorsal and ventral margins. The 
fibrous rim thus produced — which is rounded in section and often, in 
herbarium material, bright yellow, polished, and enamel-like — is character- 
istically confined to the limb of the leaf, and absent from the sheath ; it 
occurs, that is to say, in the member which I interpret as a petiolar 
phyllode, but not in the leaf-base. Wat sonia marginata , Ker-Gawl. 
(Fig. 47), is the most striking instance I have seen ; here, both on the dorsal 
and ventral side, the rim begins abruptly at the junction of sheath and 
limb. There are also many noticeable cases of marginal fibres in the genus 
Gladiolus (Figs. 43 and 44), e.g. G. Ludwigii, Pappe,and G. crassifolius, Baker, 
while various Irises (e. g. Iris verna , L., and I. Douglasiana , Herb., Fig. 9, 
p. 304), Moraea Robinsoniana , C. Moore et F. Muell. (Fig. 11, p. 304), Cypella 
coriifolia , Baker, and some Patersonias, show the same characteristic in 
varying degrees. 
From the standpoint of the phyllode theory, the most suggestive 
feature about the fibrosis of the Irid leaf is the way in which it can be 
paralleled among the Acacias. The phyllode of Acacia neurophylla , 
W. V. Fitz. (Fig. 41 B), for instance, has a dorsal and ventral fibrous rim, 
essentially similar to that of a Gladiolus or a Watsonia. The comparison 
of Figs. 40,41, 44, and 47 will show the general agreement of the marginal 
structure in the two families. The margin of a phyllode seems more liable 
to fibrous thickening than that of a lamina, possibly because the latter is 
characteristically attenuated, while the former — not being a true margin, 
Y 2 
