320 WILLIS ; MORPHOLOGY OF THE PODOSTBMACEÆ 
noticed by Tulasne) has led to several errors in the descrip- 
tions of these plants. Thus Wight divided Law’s material 
into two species (pedunculosa and foliosa) on the differences 
of the apices of one plant, and Warming has described as 
L. foliosa material collected by Goebel at Khandala, in 
which I have found the majority of the apices to have well- 
developed cupules. Hence, as I have already mentioned 
in a preceding paper, the species L. foliosa (Wight), Wmg., 
is quite untenable. 
The more southern Indian forms (L. z. malabarica), collect- 
ed by Mr. Barber, show very similar cupules, on the whole 
deeper and more like those of the Ceylon forms (PI. XIII., 
fig. 3). 
Looking back to the structure of the vegetative growing 
point already described, it is easy to see how these cupules 
may have arisen by an invagination of the apical point 
during the formation of the flower. The flower, in Ceylon, 
only emerges from the cupule after the fall of the water has 
exposed it to the air, but in India I commonly found the 
flowers growing upwards through shallow water, and in one 
case at least I found flowers which being submerged at some 
depth in a pot-hole could not grow stalks long enough to 
bring them into the air, and had become fertilized under 
water, though not perhaps strictly cleistogamously, as the 
perianth segments had separated. Many ripe fruits had been 
formed in the pot-hole. True cleistogamic flowers occur in 
Podostemon Barberi, as will be described below. 
The flower emerges — in the Indian forms often with a 
nodding stalk — on a pedicel, which at anthesis is about 2-5 
mm. long, usually erect, whatever may be the slope of the 
rock, and which has a clear pellucid cortex through which 
the central vascular bundle is visible (PI. XIII., fig. 3). At 
the base the pedicel is continuous with the thallus tissue. 
Although the thallus and the cupule are so very dorsi- 
ventral, and though the dorsiventrality of the thallus appears 
at so early an age, yet the flower itself shows radial symmetry, 
in spite of its having developed in a more or less horizontal 
