considered in Relation to the Phyllode Theory. 331 
the largest and most widely spread genus — has an ensiform leaf. It is also 
significant that Market is rhizomatous, whereas the other two genera, 
Trimezia and Cypella , are bulbous. The differences between Iris, Moraea , 
and Marica seem to be of quite subordinate importance, and I think we 
may reasonably regard these genera as the three modern representatives, 
differentiated in the Northern Hemisphere, Africa, and South America 
respectively, from one primaeval world-ranging Irid type, 1 and still retaining 
many of its generalized characters, including the ensiform leaf. 
(ii) The progression from the ensiform leaf -type. 
Under the ensiform type we must include not only the typical equitant 
phyllodes, such as those illustrated in Figs. 1-13, p. 304, but also the 
forms, such as Sisyrinchium junceum (Fig. 27, p.311), which only differ 
in the matter of proportion from the ensiform leaf of other members of the 
same genus — the width being greater relatively to the distance between 
the dorsal and ventral margins. How easy the passage from the truly 
ensiform to the cylindrical types can be, is seen in the genus Acacia , where 
there are ensiform phyllodes such as A. pendnla (Fig. 14, p. 304) and 
A. uncinella (Fig. 42, p. 318), and, at the other end of the scale, cylindrical 
phyllodes such as that of A. teretifolia (Fig. 28, p. 31 1). In the genus Iris 
we get a similar progression from the ensiform types, through the limb of 
I. Xiphium , with its oval section (Fig. 32 B, p. 313), to the cylindrical limb 
of I. xiphioides (Fig. 31 D, p. 313). 
From the ensiform leaf — using the term in the wide sense — the course 
of evolution shows signs of having proceeded in two directions, the petiolar 
limb being either reduced in all degrees down to the point of actual 
elimination, or else elaborated until it sometimes forms a highly complex 
pseudo-lamina. 
The leaf-bases of equitant leaves in many cases show a marked capacity 
for development : for instance, in Micranthns alopecuroideits , Eckl., the leaf- 
sheaths may be extremely long. In certain genera the leaf-sheath is so 
far developed as to play the more conspicuous role, while the petiolar limb 
either remains as a well-marked organ or is reduced to a mere vestige. 
In a previous paper 2 I have discussed the occurrence of such leaves in the 
Liliaceae ; in the Iridaceae instances are found in the genera Iris (. Xiphium 
Section) (p. 312), Moraea (pp. 314-15). Homeria . Hexaglottis , and G ataxia? 
In the literature, these cases are not referred to in the terms of the phyllode 
theory which I have here employed, but the leaves are merely described, 
non-committally, as bifacial with a monofacial apex. 
In the Juno Irises reduction has gone still farther and the leaves are 
1 For an exposition of the view that all the great families of flowering plants have passed 
through a period when they existed in an undifferentiated world-ranging form, see Guppy, H. B. 
(1919). 2 Arber, A. (19201), 3 Ross, H. (1892-3). 
