1895.] Flowers and Insects. 109 
appendages. These, with the ten flexuous tips of the crown 
divisions and the five erect petals, give the flower a soft, 
within, while they also render a little more evident the pas- 
sage which leads to the nectar. 
en a corpusculum with its pair of pollinia is withdrawn, 
it shows an unusually short retinaculum, which from its at- 
tachment to the corpusculum curves outward and a little 
downward and is inserted a little below the apex of the pol- 
linium. The apex of the pollinium thus forms a very con- 
spicuous knee, which stands at right angles to the axis of the 
corpusculum, and this knee is the part which is caught by the 
anther wings and thus causes the insertion of the pollinium. 
I find no evidence whatever that the original appendage to 
which the corpusculum becomes attached ever again enters 
the slit, or that the pollinia are introduced in pairs. When 
the pollinia are thoroughly dried, their planes are commonly 
directions, and this increases the chances of one of the pol- 
linia being inserted into the stigmatic chamber. There is 
nothing to render it probable that the bee’s proboscis will be 
introduced in the same relative position, and so there is no 
advantage in both of the knees, or either of them, turning to 
the same side. In Asclepias and Acerates, in which the cor- 
Puscula are usually attached to short hairs on the legs, or 
other parts of the body, as in Acerates longifolia,*® it is im- 
portant that the knees should turn away from the part to 
which they are attached, for this is the only side on which 
they are likely to be caught by the anther wings. In large 
flowers, like Asclepias Sullivantii, in which the corpuscula are 
attached to the bee’s claws, the bees commonly clasp the 
flowers so that the legs are guided upwards between the 
oods. The movement of the knees which brings them near 
together results in turning them inwards, in which position 
they are more likely to be brought tothe stigma. In Euzs/enia 
the corpuscula are attached so near to the end of the pro- 
boscis that there does not seem to be any advantage in turn- 
Ing in any particular direction, though they are slightly 
turned towards the side on which the corpusculum is attached. 
Miiller and Corry erroneously supposed that the movement 
*Bot. Gaz. 12: 245. 1887. 
