66 
flowers except Lyonia, Phyllodoce, and Vaccinium Vitis-idea, 
must be to serve as an aid in cross-pollination; the insect 
will touch them and thereby shake the anthers so that the 
pollen falls out. The fact that in the same species either the 
anthers, or the appendages, or both these organs are covered 
with small protuberances, must have the same purpose, viz. to 
Oppose the proboscis of the insect, and thereby increase the 
resulting shock. That these appendages are, on the other hand, 
wanting in species with open flowers which are usually more 
or less erect or are, at most, horizontally projecting but not 
drooping (Pirola, Loiseleuria, Ledum, Rhododendron), may easily 
be understood. Vaccinium Oxycoccos also belongs to those 
species in which the appendages are wanting, but its filaments 
are covered with hairs which perhaps serve the same end. 
This may apply also to Vaccinium Vitis-idea. 
The fact that in many species the base of the filament is 
very slender compared with the part just above it, appears also 
to be useful; it is thereby much easier for the insects to put 
the filament in motion and shake it, than if its base had been 
thick. 
Arctostaphylos alpina occupies a peculiar position by 
reason of the fact that its appendages are short, thick and 
glabrous or even entirely wanting (aborted). This should be 
correlated with the fact that the throat is very narrow, narrower 
than in A. Uva-ursi, so that the insect must inevitably touch 
the filaments and shed the pollen. This is also the case in 
Phyllodoce. On comparing the above two species it will be 
seen that while A. alpina is far the better adapted for self- 
pollination, A. Uva-ursi is the better adapted for pollination 
by insects. In the structure of its flower Andromeda polifolia 
approaches nearest to Arctostaphylos Uva-ursi, especially in 
that the interior of its corolla is covered with erect hairs as is 
the case in the two species of Arctostaphylos. We may be 
justified in concluding that these hairs are connected with the 
