502 THE AMERICAN NATURALIST.  [VOL. XXXIIL 
the species to variation, as described by Miiller and others, and 
to which they attribute the multiplicity of forms of flowers. 
The high degree of zygomorphy and bright coloration have 
evidently been produced by the agency of insects, since the lip, 
the petal most modified in form and color, is the segment most 
visited, while the other segments of the perianth often retain 
their primitive form and plain green coloring. Insects have 
not, however, been able to induce particular colors, but their 
work has been the fixation by selection of those naturally pro- 
duced by the flowers. This is well shown by the rarity of blue 
among the Orchidaceze. This color is very attractive to bees, yet 
its development has not necessarily followed the high specializa- 
tion of the flowers in response to their visits. Its presence 
or absence is rather dependent upon the chemical constitution 
of the nutritive fluids or other internal conditions. There may 
even be no bright coloring, as in the mellifluous plain green 
species of Epidendrum mentioned above. 
The monocotyledonous families were probably very early dif- 
ferentiated, and their subsequent development has proceeded 
along parallel lines, but without any connection with each 
other. With few exceptions, the families possess a perianth, 
though it is frequently rudimentary, and in the majority of 
species non-petaloid. In the opinion of Engler the floral 
envelopes have not been induced by insect pollination. Their 
very general occurrence points to their early development, and 
their first office was undoubtedly protective in its nature, simi- 
lar to that performed by the scales of the gymnosperms. In 
anemophilous families the perianth has remained unmodified 
or, where the protective office has been assumed by glumes or 
bracts, as in the Graminez, has nearly disappeared ; while in 
the entomophilous families insect pollination has caused it to . 
be enlarged and s@ecialized. When the expenditure of produ- 
cing petals, nectar, and color is considered, as well as the vast 
number of individuals blooming at a time when they would be 
brought into competition with many entomophilous flowers, it 
cannot be doubted that the Graminez and Cyperacez are more 
efficiently pollinated by the “wind than could possibly be the 
case by the existing number of insects. 
