= 
770 The American Naturalist. [September, 
Taking this, then, as the field, here is the problem present- 
ing itself :—From March to November each month brings a new 
prospect in field and forest, an edition of flower-life distinct- 
ively its own; and each edition seems thoroughly to harmo- 
nize with its own peculiar season. Each offers characteristic 
features, and every careful observer can feel in this succession 
of differing forms a harmony into which any decided change 
would break discordantly. To bring this more forcibly to the 
mind, conceive for a moment the existing series somewhat 
transposed: imagine the prairies brilliant with Solidago and 
dotted with the purple Aster in early May ; Trillium adorning 
the woodlands in September; the vivid Lilium superbum and 
the flaming Asclepias tuberosa in the April meadows; Hepa- 
tica and Sanguinaria in mid-July ; and Epigsea in late autumn. 
Entirely aside from the element of novelty, such a picture 
would be bewilderingly inharmonious, and its features out of 
tune with their position by virtue simply of their character,— 
in this there would be an appreciable and evident lack of fit- 
ness. To say that the fall flowers are not the spring flowers, or 
those of summer are neither, merely because they have chosen 
at random this season or that is neither science nor common 
sense. The truth is forced upon us that the various groups of 
flowering plants are not scattered indiscriminately from one 
end of the seasons to the other, but are regulated by definite 
systematic principles; and that just as relations can be traced 
between physical geography and geographical distribution, or 
between plant-history and geological periods, so is there a 
connection between the relations of season to season and the 
relations of their respective floras. 
In this inquiry we see constantly the impossibility of draw- 
ing sharp lines of demarkation between successive groups of 
conditions; and, as in classification systems, so in serially 
arranging successive floras, it is impossible to follow a linear 
order. Also, exceptional instances are many, and hence, to 
reach conclusions that are broadly true, predominant types 
must be considered, though the unusual cases may be in them- 
selves manifestly significant. Only a broad sweep of data can 
bring out a general connection between the multiplicity of con- 
