considered in Relation to the Phyllode Theory . 329 
of general symmetry which makes it scarcely possible to identify a midrib. 
These conditions can be paralleled in the leaves of some of the Liliaceae, 1 2 
and in certain Dicotyledonous petioles and leaf-bases. For instance, a small 
median bundle and larger main laterals are found in certain species of 
A stelia, Allium, and A rnocrinum (Liliaceae), and in the petiole of Antigonon 
(Polygonaceae), while the leaf-base phyllode of Anemarrhena (Liliaceae) 
and the leaf-sheath of Foeniculnm (Umbelliferae) resemble the leaf of Crocus 
carpetanus in showing no obvious symmetry about a midrib. I do not 
know, however, of an} 7 cases in the Liliaceae exactly corresponding to that 
of the Moraeas in which there is no median bundle (e. g. M. Macleai , 
Figs. 12 A and B, p. 304, and M . polystachya, Figs. 36 A and B, p. 315), but 
the petioles of certain Labiates ( Salvia Verbenaca , Fig. 39, p. 315, and 
Marrubium velutinum? Fig. 58, p. 324) show a corresponding lack of 
a median strand. 
9. The Evolutionary History of the Irid Leaf. 
(i) The primitive character of the ensiform leaf -type. 
In considering the evolution of the Irid leaf, the first necessity is to 
arrive at some working hypothesis as to what foliar type is to be regarded 
as primitive for the family. As I have shown on pp. 302 and 303, the 
typical ensiform (isobilateral equitant) leaf is met with in more than 30 of 
the 57 genera, including members of both the principal tribes, Iridoideae 
and Ixioideae. Not only is this form of leaf characteristic of the majority 
of Irids, but many of the other leaf forms met with in the family — how- 
ever superficially different — prove on examination (as indicated in the 
preceding sections of this paper) to be easily derivable from the ensiform 
type. Among the Liliiflorae, it is not only in the Iridaceae that the 
ensiform leaf occurs — it is found also in the other two great families, the 
Liliaceae and Amaryllidaceae. Its distribution, both within the Iridaceae 
and in the Liliiflorae in general, justifies us in treating it as, in all probability, 
representing the leaf-type of the original Irid stock. The idea is consistent 
with Baker’s 3 suggestion that Hewardia (of the Colchicaceae) forms ‘an 
excellent connecting link between Liliaceae and Iridaceae’ — for in this 
monotypic genus the leaves are ensiform. 
Within the genus Iris itself, there seems good reason to consider the 
ensiform leaf-type as primitive. Such a conclusion certainly follows if we 
accept the deduction from Dr. Willis’s ‘Law of Age and Area’ 4 that, 
within a limited group, the more widely ranging types are likely to be the 
1 Arber, A. (1920 3 ), pp. 449 , 461 , and Figs. 1 and 34 . 
2 Petit, L. (1887) ; in the case of Salvia Verbenaca I have failed to find the small median 
bundle figured by Petit. 
3 Baker, J. G. (1880). 
4 Willis, J. C. (1914). 
