1 8 Saunders. — On the Structure and 
and they vary considerably in size ; owing to this irregu- 
larity, and to the position of the nucleus (generally 
central), it is difficult to determine the average number 
of meshes in a cell. In those cases where the meshes are 
large and few in number (Fig. 9), the cells in all proba- 
bility represent a slightly more advanced stage of activity, 
and one through which those containing a large number 
of smaller vacuoles will have to pass before reaching the 
next and final stage ; for it is not at all uncommon, as I 
shall have occasion to point out later, for the cells at the 
same level to exhibit different stages of activity. 
( b ) In the sub-epidermal cells (Fig. 8) the appearance is 
even more striking, for since the position of the nucleus 
is most frequently parietal, the meshwork stretches un- 
interruptedly throughout the cell. Such at least is 
generally the case when the dehiscence of the anthers 
is complete ; during the period of dehiscence the regu- 
larity of the meshwork is somewhat broken by the 
starch, which is still fairly abundant, and the disap- 
pearance of which takes place later here than in the 
epidermal cells ; so much so indeed, that some may 
still be found in almost every sub-epidermal cell, when 
the epidermis contains none. The breaking down of 
the starch begins in the superficial cells, and extends 
gradually to those lying deeper. The average number 
of meshes in the long diameter of the cell is about 
twelve. 
A good deal of the starch has disappeared from the un- 
modified parenchyma of the septum, but there is still a con- 
siderable amount in the cells of the central axis. 
E. Pollen mostly shed, and perianth often beginning to 
wither (Fig. 10). 
(a) (a) The external wall presents the same appearance as 
in the previous stage. 
( 8 ) The nucleus, which may be much flattened, is 
generally parietal. 
