Ipomoea paniculata. 285 
or four secretory cells on the surface. Here and there, 
however, was an epidermal cell projecting, which had only 
undergone transverse division, and would probably have pro- 
ceeded to form a gland later. 
It would seem that the secretory epithelium lining the 
large gland of the leaf is homologous with an aggregation 
of these small surface-glands, the only difference being that 
the development of the epidermal cells which form the 
invagination in the earliest stages of the gland has been 
retarded, and that when they do become differentiated, each 
cell gives rise to a small gland ; these are consequently 
crowded thickly together in a curved surface, and so give 
rise to the complicated secretory epithelium above described 
(compare Fig. 20 with Fig. 15). This view obtains additional 
support from the reactions of the glands with certain staining 
reagents. The secretions of the large gland and of the small 
superficial glands appear to be identical also. 
Nature of the Secretion . In mature leaves the secretion 
of the glands accumulates in a large drop projecting from 
the depression in which the mouth of the gland lies. I was 
able to get a considerable quantity of it by cutting off some 
large leaves and putting the cut ends of the petioles in water. 
They were covered over with a bell-jar to prevent evapora- 
tion, and in this way I was able to remove the secretion at 
intervals as it was renewed. It consists of a thin, clear, 
transparent fluid, having a taste like dilute saccharine. After 
placing some on a slide with Fehling’s solution and heating, 
an abundant reduction of the copper occurred, demonstrating 
the presence of a sugar. Tannin was also present in con- 
siderable quantity. 
Thick sections were cut in the hope of finding some of the 
secretion in the duct, but the result was always disappointing 
because of its extreme solubility. Nor were the results 
afforded by the secretory cells themselves much more satis- 
factory. At an early stage many granules were present in 
the protoplasm which stained in dilute haematoxylin, were 
coloured brown by iodine, and were presumably proteid in 
x 2 
