BALFOUR—NEW SPECIES OF RHODODENDRON, 303 
where the umbo passes into the fringe becomes tinted and there 
appears to be a ring of resin-like substance. The fringe itself 
may remain uncoloured and whitish, or it may change to a brown 
colour—the cell-walls here altering. These scales are at first 
whitish all over, both on top of and below the leaf lamina. The 
scales on the upper surface are almost superficial. By that I 
mean they are not sunk in deep pits. The upper leaf-surface 
may be undulate and the scales may come off from the depres- 
sions, but they are always raised above the surface so that the 
fringe spreads more or less horizontally over it. These scales 
may shrivel up on the old leaves remaining as a greyish debris 
making the surface somewhat hoary, or may fall off and leave 
only a trace of their existence in the undulation of the surface 
or in greyish shreds, or they may, if the umbo becomes resinous, 
remain as scintillating disks spread over the surface. As to how 
far the difference is to be regarded as associated with any par- 
ticular environment or as a specific character there is insufficient 
evidence. 
The form and distribution of the scales on the under-leaf 
surface furnishes a mark of much value in the diagnosis of species. 
The scales of the mould already mentioned may be contiguous 
or discontiguous. The area of exposed leaf epidermis where the 
scales are discontiguous is about the same as, or more than, that 
of the scales. There are states when the scales are not quite 
contiguous and there are left narrow chinks between the 
scales. For purposes of description this state is treated as 
contiguous. Where the scales are contiguous the fringes of 
adjacent scales may not merely touch but overlap in an imbricate 
fashion. By this character of contiguity or discontiguity of 
scales we can at once diagnose groups of species. Where the 
scales are contiguous they are essentially superficial ; where dis- 
contiguous they are sunk in pits, and this sinking affects the 
appearance of the scale—because if the pit be deep the length 
of the stalk of the scale may be less than the depth of the pit ; 
the umbo of the scale is then depressed and the fringe may. be 
turned upwards and only slightly overlap the leaf-surface. Such 
sunk scales are readily recognised. The evident function of 
all scales is water conservation through the layer of still air 
they maintain over the leaf-surface. The wax-bearing epidermal 
papillae appear to be much better developed in the discontiguous 
species than in the contiguous ones. 
There are yet other interesting features to notice of this 
indumentum on the under-leaf surface. In some cases the 
contiguous scales are large and all of about the same form, size, 
and colour, and their imbrication gives the effect that one 
observes in such scale-covered leaf-surfaces as occur in Elaeagnus, 
G 
