NOTES ON MELILOTUS ALBA 
419 
The primary sporogenous cells become differentiated in each 
lobe of the anther as a single row (7, Plate XIII), probably five 
cells long. Both cross and longitudinal divisions follow rapidly 
and the result is a column of spore mother cells, four to six in 
number in cross section (8, Plate XIII) and ten in longitudinal 
section. The cells average ten to twelve microns in diameter 
but during the ensuing growth period increase in size until at 
the time of synapsis the diameter is fourteen to fifteen microns 
(9, Plate XIII). 
As the synaptic stage is initiated the mother cells begin to 
round off and to thicken their walls (10, Plate XIII). By the 
time that the heterotypic and homotypic divisions have given 
rise to four nuclei the wall thickening has become very pro- 
nounced and ridgelike ingrowths are produced midway between 
the nuclei (14, Plate XIII). Either these thickenings' grow 
inward, meeting in the center and cutting off the protoplasts from 
one another, or plates of material similar in staining reaction 
to the outer wall form between the nuclei and growing out- 
ward attach themselves to the projecting ridges (15, Plate XIII). 
The newly formed cells measure ten microns in diameter. 
The whole tetrad is now encased in the thickened wall of the 
mother cell and each member is individually enclosed in a layer 
which appears to be a condensation of the material of the mother 
cell wall (16, Plate XIII). This is the “special wall”, of Stras- 
burger. Inside this the microspore wall now forms, first as a 
delicate layer, but rapidly thickening (17, Plate XIII). The 
spores are at first spherical but soon elongate and when mature 
measure twenty-three by fifteen microns (18, Plate XIII). A 
characteristic feature of the wall is the possession of three longi- 
tudinal grooves or folds symmetrically disposed along the sides 
of the spore (19, Plate XIII). Each groove has at its middle 
point a thin spot in the wall, the exit pore. 
The division of the nucleus into tube and generative nuclei 
takes place some little time before dehiscence of the anther, and 
the generative cell when formed lies in the end of the grain 
(20, Plate XIII). 
Most of the pollen grains in the older anthers sectioned were 
as if turgid and did not show the wall foldings, but in grains 
which were taken from ripe anthers and examined, in air the 
wall folds were present. This pollen instantly became turgid 
