19 
of the anthers are constantly much in advance of the others. In this 
genus they become much longer, and instead of being, in the bud 
state, imbedded in the substance of a transparent wingot the filament, 
they are attached thereto by abundant glutinous hairs belonging both 
to themselves and to the filament. Here, at least, the horns appar- 
ently have some part in cross-fertilization. In A. glauca they are so 
far doubled backward that their hairy and glutinous tips are approxi- 
mated to the anther pores, so that being jarred by departing insects, 
a quantity of the pollen is shaken out upon them, to be held until the 
arrival of the next visitor. Futher evidence of the flowers being pro- 
terogynous was found in the presence, among the filaments and upon 
• the viscid stigma, of a considerable quantity of pollen before the an- 
thers of those flowers were yet mature. A large number of flowers 
was found punctured near the base with a large, angular aperture. 
Careful and long continued observation placed the mischief to the 
account of a humming-bird, which passed a large part of its time 
among the shrubs. 
. Casstope does not differ essentially from Arbutus, Its horns are 
like those oi Arbutus in slenderness, but are not quite so long, and are 
rather straighten ^ 
Fig. 8. Fig. 9. 
^ The stamens of all the above genera agree in two important ])ar- 
ticulars: the filaments are attached quite near the apex of the anther, 
and the latter possesses backwardly projecting* horns. Both these 
characters, as we have seen, are useful in overcoming the reversion of 
the anther. Now in Gaultheria^ of which I have but a single bud to 
pxamine, and that an old one, we see the first step toward a condition 
in which both these characters are wanting, and where the anthers 
^re never reverted. For in Gaultheria we have the " dorsal " awns, 
c^ut the filament is not attached near the apex of the anther. (Fig. 5.) 
In Cassandra {Fig. 6) the mechanism is still farther lost by the in- 
sertion of the filament near the lower end of the anther, and the 
^niost straight direction of the stout horns, which thus become beaks. 
Here we are obliged to entertain the question as to the dorsal horns 
^f Arbutus being the homologues of the terminal beaks of Vaccineae 
Und Cassandra, etc.) which have become narrowed, recurved, and 
broken through at the base on the inner side to form the ventral pores. 
As above stated, in neither of the tvyo last genera was reversion 
clearly seen, and it was clearly excluded in all the others examined 
up to Pyrolinese, But, in passing, one would call attention to the 
