248 BALFOUR—NEW SPECIES OF RHODODENDRON. 
quite different. There are none of the setose hairs of Rh. Dal- 
housiae and the peltate leaf-scales are far apart, leaving a large 
area of leaf epidermal surface visible. As Hemsley and Wilson 
’ point out there is also the absence of scales on the outside of 
the corolla separating the species and that is an emphatic mark 
of divergence in this complex. Rh. excellens by all characters 
is one of the group of Rh. crassum and stands in it between 
Rh. crassum itself and Rh. megacalyx. From the former its 
oblong leaves and larger calyx sufficiently separate it ; from 
the latter it differs in its longer narrower leaves, lepidote pedicels 
and calyx-cup, and in its smaller calyx. It has the indumentum 
of the leaf under-surface of Rh. megacalyx. Through the kind- 
ness of the Director of the Royal Gardens, Kew, I have been 
enabled to examine the type sheet of Rh. excellens and to 
compare it with Ward’s specimens of Rh. megacalyx. 
The prominent and easily observed differences are not the 
only marks of separation from Rh. crassum. Close examina- 
tion of the leaf surface between the scales on the under surface 
of the leaf will show that the papillae of the epidermis are 
altogether different. In Rh. crassum these papillae are little 
low cupolas so that the sides of the papillae are divergent and 
a vallecular interval lies between their tops. Only where there 
is a scale is there lengthening of the papillae. Beneath the 
scale fringe the papillae are elongated and inclined towards the 
stalk of the scale thus increasing as it were their protective 
value. But in Rh. megacalyx the epidermal papillae are vertical 
rods of some length and their sides are parallel. They are set 
close together. All these papillae have a granular wax-coat. 
Further, in Rh. crassum the pits on the under side of the leaf in 
which the scales are set are not very deep: the stalk of the 
scale brings the umbo to the mouth of the pit on a level with 
the surface of the adjacent epidermis and the scale fringe has 
space to spread horizontally as it does. Very different is it in 
Rh. megacalyx. There the pits on the under side of the leaf 
are deep, the umbo which has very thin cells is not brought to 
the mouth of the pit on a level with the epidermal surface, 
there is no room for lateral extension of the fringe of the scale, 
and it is not only narrower but is turned upwards so that the 
scale to surface view has a cuplike appearance. That this 
space relation is in causal connection with the narrowness and 
direction of the fringe I am not prepared to assert, because 
the single specimen of young expanding leaves which I have 
seen suggests that from the outset and before the leaf-surface 
pits are developed the size and direction of the fringe are deter- 
mined. The condition is of interest looking to the modification 
of form of the elements of indumentum in the genus. From 
