262 
POPULAH SCIENCE REVIEW. 
stout filament, which is inserted into the mouth of the tube, as 
represented in fig. 2, and which thence runs backwards, so as 
to have its upper end completely hidden from view in the hood 
formed by the upper lip of the corolla. The short filament ter- 
minates in a connective which is developed to such an extent as 
to be actually longer than the filament itself, and at the two 
opposite ends of the connective are the two anther cells. The 
portion of the connective uniting the upper anther cell with the 
filament, is much longer than that portion which unites the 
filament to the lower cell. The upper anther cell is also itself 
much larger than the lower one, this latter being, in fact, almost 
sterile, and furnishing little or sometimes no pollen. It is not, 
however, so completely transformed into a barren structure as 
is the case in many other species of salvia, to be spoken of pre- 
sently. The upper cell, on the other hand, furnishes pollen in 
great abundance, and allows it to escape by dehiscing in the 
front longitudinally. The filament is attached to the connective 
externally by an excessively movable joint, so that a very slight 
pressure on either anther cell will cause the connective to turn 
on the filament with what the French would call a ‘^mouvement 
de bascule.” The joint is so strong that you may actually cause 
the connective to rotate four or five times in complete circles in 
the same direction on the filament before it is twisted off. The 
lower or sterile cells of the two anthers are adherent to each 
other, so that a slight pressure on one anther will produce rota- 
tion, not only in its own connective, but in both. This descrip- 
tion will be more intelligible on looking at fig. 3. In fig. 3 
the stamens are represented as seen when looked at in front ; 
in fig. 4 they are seen from the side when at rest ; in fig. 5, as 
seen from the side when the anthers are made to rotate on the 
filaments. 
On looking at the front of the flower, the only parts of the 
anthers which are visible are the lower cells, with their portion 
of connective (fig. 6). These stick out into the gaping opening 
of the tube, projecting like an uvula into the throat. The rest 
of the anthers is hidden in the upper lip of the corolla, which 
forms a hood, closed more or less completely in front until the 
flower begins to wither. The lower anther cells project to such 
a distance in the mouth of the corolla as to render it quite im- 
possible for a bee to get at the nectary without pushing directly 
upon their upper surface. No sooner does the bee exercise this 
pressure than the connective rotates. The upper anther cells 
emerge from the hooded receptacle in wliich they are hidden, 
and are seen to perform a circular movement forwards and down- 
wards, until their dehiscent surfaces are brought into close 
contact witli the back of tlie bee, one anther cell rubbing it on 
either side. This movement may be artificially produced] by 
