326 Arber . — ZI60 Leaf Structure of the Iridaceae, 
of the first plumular leaf of a seedling of this species. Near its attachment 
to the axis the sheath region is closed. This closed basal region, to which 
attention was long ago drawn by Irmisch, 1 being extremely short, may 
easily be overlooked, and it is sometimes, indeed, not typically developed ; 
in serial sections through two shoots of Crocus speciosus , Bieb., I found 
that inextricable confusion was produced by extreme telescoping of the 
internodes, associated with a fusion of the leaf-bases both with one another 
and with the axis. The vascular anatomy of the sheath in a plumular leaf 
of Crocus Tomasinianus is shown in Fig. 56 A ; the median bundle (m.b.) 
is so slightly differentiated as to be scarcely visible, consisting at this stage 
of a few procambial cells only. The main laterals, on the other hand, are 
much larger, and their xylem and phloem can already be distinguished. 
A second smaller pair of bundles occurs, which we may call the external 
laterals (e.l.). A little higher up two invaginations (gf come into existence 
between the main lateral and external lateral bundles. The further history 
of these grooves can be better followed in Figs. 56 E-I, which were drawn 
from a series of hand-sections through the base of a mature leaf of the same 
species. The invaginations, which are first seen in Fig. 56 F, have 
developed farther in Figs. 56 G and H, and the lateral wings, thus 
separated from the main part of the limb, increase greatly in length by 
intercalary growth, and become coiled. In Fig. 56 I the final stage is 
reached — the unrolling of the two wings having produced the curious 
transverse section which characterizes the mature Crocus leaf. I have also 
studied the development of the leaf in Croats bijlorus , Mill., C. speciosus , 
Bieb., and C. vernus , All. 2 (Fig. 57), and I find that in each case the history 
is closely similar to that just traced for C. Tomasinianus . In Crocus vernus 
I followed the structure to the leaf-tip and found that at the apex the wings 
become reduced and the invaginations gradually die out (Figs. 57 B and c), 
so that there is an eventual return to something approaching the unmodified 
form of the limb. 
When we turn to the comparison of Crocus and Romulea , we find that 
the leaves are really built on the same plan, with, however, two differences, 
which — though non-essential — result in a conspicuously different final form. 
In botlj genera we get a pair of grooves or invaginations (g. x ) arising between 
the main laterals (m.l.) and the external laterals (e.l.), but in Romulea there 
is an additional pair of grooves ( g. 2 ) lying between the median bundle and 
the main laterals ; this pair is unrepresented in Crocus . Again, in Romulea 
the external laterals (e.l.) approach one another and finally fuse to form 
a single bundle ( v.b . in Figs. 61 D and 63 B), whereas the process of 
invagination and winging in Crocus carries the two external laterals far 
1 Irmisch, T. (1850). 
2 Pax, F. (1888), figures the leaf of Crocus vernus as possessing no midrib : this, according to 
my observations, is an error. 
