544 A. B. DAWSON 
in its early stages the secretion is readily fixed in all fluids, the 
fixed globules then appearing as large, more or less spherical 
granules. Sooner or later, however, it reaches such a condition 
that in Kleinenberg’s, Flemming’s, or other acid fixing solutions 
the large granules disintegrate. In some cases, the cell walls 
disappear before the change in the nature of the secretion occurs 
(fig. 19); in other cases, the secretion reaches the mature con- 
dition while the cell limits are still well defined (figs. 18, 24). 
In figure 13 a large cluster of cells is seen at the upper pole of 
the gland in the region of the future duct. This condition sug- 
gests that cells are migrating downward from the epidermis 
faster than they can be accommodated by the developing gland, 
and reminds one of the epithelial plug described by Drasch (’94). 
In this gland an intercalary region is present. Also several 
flattened cells (fig. 18, cl’) are distributed over the periphery of 
the gland. They appear to be undifferentiated cells which have 
migrated into their present positions after the metamorphosis 
of the original glandular epithelium. In the peripheral regions 
of other glands, in which the epithelial cells have been trans- 
formed bodily into secretion, small gland cells, with distinct 
walls and filled with young secretion granules, are often observed. 
They probably represent later stages of the flattened cells shown 
in figure 13. 
In young glands the intercalary cells appear to form a reserve 
from which granular cells are differentiated from time to time, 
and the supply of cells in this region is maintained either by 
mitosis or emigration of cells from the epidermis. Besides being 
continually reinforced in these later stages by epidermal cells 
which migrate downward by way of the intercalary region, the 
granular cells may themselves multiply by mitotic division (fig. 
7). Binuclear or multinuclear conditions in the large granular 
cells have been described by several writers (Engelmann, ’72; 
Drasch, ’92, 94; Nicoglu, ’93; Vollmer, 93; Talke, ’00; Esterly, 
04; Arnold, ’05; Nirenstein, ’08; Phisalix, 710). Schultz (89) 
has described mitosis as occuring very frequently in the poison 
cells. Drasch (’92, ’94), however, states that he has examined 
hundreds of preparations and has never found a single mitotic 
