22 
Saunders. — On the Structure and 
temporary reserve material), and also of granular proteid 
substance (metaplasm). Further, in the epidermal cells these 
processes are accompanied by a change in the character of 
the external wall, a change which continues until it reaches a 
maximum in the fully open flower ; the internal part swells, 
becomes mucilaginous, and loses its clear inner outline, the 
appearance of the internal surface being such as to suggest an 
active destructive influence on the part of the protoplasm. 
As the secretion must pass through both layers of the ex- 
ternal wall before reaching the lumen, it is possible that some 
of the mucilage contained in it may be derived from the inner 
layers of the wall, and may have been carried along by the 
flow of the secretion. When the flower-tube opens the starch 
begins to diminish, the large apparently single grains break 
up into the separate grains which compose them, and these 
rapidly disappear. The question at once arises, are these 
starch-grains converted into some form of soluble carbohydrate 
which is then excreted ? Has their disappearance from the 
cells any direct connection with the presence of sugar in the 
secretion ? Before attempting an answer to these questions, 
it was necessary to determine at what stage in the develop- 
ment of the flower the sugar first made its appearance. With 
this object in view I treated sections from flowers of different 
ages with Fehling’s solution, but with extremely unsatisfactory 
results ; the reduction was always most diffuse, and was 
evidently caused by substances other than sugar. Abandon- 
ing this method as useless in this particular case, I was 
obliged to fall back upon the extremely rough plan of tasting ; 
no sweet taste could be detected, nor was there any liquid 
visible in the flower-tubes of buds, nor in those flowers in 
which the perianth had just opened ; but in those in which 
the stamens were projecting from the open flower-tube, though 
the anthers were still closed, it was possible to detect a sweet 
taste 1 . Now the histological appearances which are charac- 
1 In thus testing for sugar, I used almost exclusively flowers of K. media ; in 
K. nobilis , owing to the greater length of the flower- tube, the stamens do not 
project through the open tube while the anthers are still closed ; consequently the 
