150 BALFOUR—NEW SPECIES OF RHODODENDRON. 
in a transverse section diminishes upwards until towards the 
top there is but one cell. Almost from the base the hair curls 
upon itself in the same way as a tendril curls when it fails to 
find a support, and it is the intercurling of the several hairs that 
gives the peculiar resilient character to the whole superstratum 
of indumentum. Beneath this upper stratum of hairs lies an 
under stratum of stalked peltate scales on every part where 
hair-indumentum is found. They are not confined to the under- 
leaf surface. These peltate scales, at first white and translucent, 
become infiltrated later with red secretion, and they look like a 
series of Peziza-cups spread on the epidermis. And then be- 
neath the peltate scales is to be found a remarkable develop- 
ment of long rod-shaped epidermal papillae, each rod being 
coated outside by large wax-granules. 
form.—The hairs of the upper stratum of indumentum 
are somewhat different. They are fewer, very thick-walled, 
quite fibre-like. Each has a stout multicellular base, and 
then extends with a curious jointed character sooner or later 
branching into long fibre-like threads which do not curl so 
definitely as those in Form A. yet are conspicuously interwoven. 
In the under stratum we miss entirely the peltate scales. These 
are replaced by glands which have multicellular stalks short or 
long and rounded or oblong heads. And then there are no 
epidermal papillae. 
The contrast is interesting. In both a layer of still air is 
secured by a thick covering of hairs, in one case unbranched 
and many, in the other branched and fewer; and then, as 4 
check to the transpiration, are peltate scales and wax-coated 
epidermal papillae associated with the unbranched hairs, glands 
and no papillae with the branched hairs; the latter combina- 
tion perhaps allowing more perfect distribution between fewer 
hair-stalks of the emanation from the glands. 
I am not prepared to maintain that all the plants cited in 
the table under Rh. bullatum are really that species. - Their 
distribution is against identity. It is rare to find a species 
running from the dry Chungtien Plateau south to the moist 
region of Tengyueh, and the specimens themselves show difference 
which lack of material alone makes it difficult to appraise. I 
do no more here than point out the possibility of more than one 
species being included under the name, and leave over dis- 
cussion of the forms until after additional illustrative specimens 
have been obtained. 
Rh. cruentum has not been described adequately, and I 
therefore give a description of it here, bringing in also - 
of its ally Rh. Bureavi, so that their relationships may be 
appreciated :— 
