considered in Relation to the Phyllode Theory. 327 
apart, so that they are eventually separated by the entire width of the flat 
leaf. We have already shown that the leaf of Romidea is but little modified 
from the ensiform type— that is to say, it is easily derivable from the most 
widespread form of petiolar phyllode. If my interpretation of the limb of 
Crocus , as precisely equivalent to that of Romulea , be accepted, the Crocus 
leaf must also come into the category of petiolar phyllodes ; indeed, apart 
from the comparison with Romidea , the structure of the Crocus leaf, 
considered in itself, carries indications that this point of view is not too 
far-fetched. Fig. 56 C shows how petiole-like is the stage passed through 
just above the sheath region of the first plumular leaf in C. Tomasinianus. 
A comparison of Fig. 56 F and Fig. 58, again, reveals how closely the 
stage at which the invagination is just beginning may be compared with 
a certain Dicotyledonous petiole, that of Marrubium velutinum , Sibth. et 
Sm. ; 1 we even get a corresponding grooving and the same conspicuous 
pair of main laterals ; but whereas the median bundle in Crocus is present, 
though extremely reduced, in Marrubium it is altogether absent. 
With minor variations, the type of leaf described seems to be universal 
in the genus Crocus , except in the case of two species confined to Spain,. 
C. carpetanus , Boiss. et Reut., and C. nevadensis , Amo et Campo. 2 I have 
not been able to get material of C. nevadensis , but I have cut sections of 
C. car pet anus, which has a leaf differing conspicuously from that of a typical 
Crocus. There is a closed sheath at the base, which passes into a somewhat 
asymmetrical bifacial leaf, traversed by a series of bundles, alternately large 
and small ; the dorsal surface is ridged opposite each of the larger bundles 
(Fig. 59). The median bundle is not, as in the other Crocuses, smaller than 
the main laterals. This leaf I regard as morphologically a leaf-sheath — 
the petiolar limb of the other species of Crocus being unrepresented. But 
it is very different even from the sheath region of the other Crocuses, and 
rather recalls such leaves as that of Moraea bituminosa (Fig. 37, p. 315). 
The curious isolation of the leaf anatomy, in association with certain other 
aberrant features of the plant to which Maw draws attention, arouses some 
doubt as to whether the inclusion of C. carpetanus in the genus Crocus can 
be justified. 
The small genus Syringodea , which is confined to the Cape, resembles 
Crocus and differs from Galaxia and Romidea , in having no above-ground 
stem. Ross 8 describes the structure in 5 . montana , Klatt, as dorsiventral, 
with a convex lower surface and concave above. I have examined .S'. bicolor , 
Baker, and I find that the structure corresponds in general to Ross’s brief 
description of 5 . montana. There is no strict symmetry about a midrib, 
and the main laterals (in /.) are larger than the median bundle ( m.b .), which 
does not occupy precisely the middle line of the leaf (Fig. 60). The only 
Crocus to which such a leaf could be compared is C. carpetanus , and even 
1 Cf. Petit, L. (1887). 2 Maw, G. (1886); 3 Ross, H. (1892-3). 
