1882.] 
437 
[Trelease. 
plants can be better studied. A larger insect ought to effect fer- 
tilization immediately. Sprengel supposed that insects frequent 
by preference gibbous flowers, as if nature had indicated the 
presence of nectaries by these irregularities ; the flower of Gold- 
fussia has a swollen corolla, but Kurr has contradicted Sprengel’s 
statement” (2, 17-18). I think, however, that the majority of 
facts are in favor of Sprengel’s theory of the pollination of flowers. 
In closing, a few general considerations and conclusions may 
be of interest. 
With the exception of Lemna, all of the species studied belong 
to Delpno’s Zoidiophilae, and possess means for attracting and 
profiting by the visits of animals. In each there is more than 
the offering of a food supply accessible to all comers, the struc- 
ture in most being sucli as to restrict the favored visitors to lim* 
ited groups, varying with the species, and including four families 
of birds and two orders of insects. Salvia Heerii, Westringia and 
Cystacanthus are instructive as showing how organs originally 
useful in one part of the reproductive process may be made to 
assume secondary roles. The former is also a good example of 
highly specialized plants adapted to two distinct groups of polli- 
nators that are apparently equally efficient. By the complete 
closure of the flower by the connective it would be rendered ex- 
clusively ornithophilous, while the contraction of its tube might 
render it exclusively psychophilous. 
Not only are all but welcome visitors excluded, but the adap- 
tation goes to the extent of utilizing only limited parts of the 
favored ones. This is effected by successively bringing the pol- 
len and stigma into the same small area at maturity, as inProtea- 
ceae, Westringia, Cystacanthus and Diosma ; or by making them 
occupy positions on the line along which visitors approach the 
flower, either above the axis, as in Salvia, or below it, as in Gold- 
fussia. 
It is interesting to note the correlation of the drooping habit 
with the cymose development of the flower cluster in Grevillea, 
as in Mesospinidium sanguineum and other orchids ; and to com- 
pare it with the racemose development of the erect cluster of 
Cystacanthus. In both the flowers are protandrous, and both are 
adapted by these arrangements to crossing in an identical 
