Gradttal, Change, and by Guppy s Method of Differentiation. 619 
‘Vitaceae, Gramineae, Cyperaceae . . . contain hermaphrodite and 
diclinous genera and . . . flowers of all kinds stand side by side in the 
Compositae. 
‘ Eschscholtzia has decidedly perigynous stamens, and yet it is... a genus 
of Papaveraceae, the character of which is to have them hypogynous; and 
all kinds of gradations ... are observable in Saxifragaceae . . . the stamens are 
. . . monadelphous in Malvaceae . . . but more commonly this character is 
unimportant, as in Malvaceae themselves, which have sometimes distinct 
stamens ... in Solanaceae, the genera of which have usually their anthers 
bursting longitudinally, the genus Solatium itself opens by pores. 
‘ Cases exist of both forms (apo- and syn-carpous ovary) being found in 
the same natural order ... in Caryophyllaceae and Bruniaceae there are 
genera, the ovary of which contains several cells . . . in Pedaliaceae and 
Styraceae both erect and suspended ovules coexist. . . the genus Conohoria 
(. Alsodeia) offers ... an instance of three kinds of direction in as many 
species. 
‘ Marcgraviaceae, Melastomaceae, Myrtaceae, Ranunculaceae, and 
Rosaceae, &c. . . . contain both baccate and capsular, dehiscent and in- 
dehiscent genera. 
‘ I doubt very much whether presence or absence of albumen deserves 
much attention in orders . . . where the embryo and albumen are of nearly 
equal bulk . . . among Apocynaceae, which have solid albumen, it is ruminate 
in Alyxia . .. There are plants among Dicotyledons with only one cotyledon, 
as Penaea and some Myrtaceae, or several . . .’ 
The diagnostic characters of flowering plants, as regards their classifica- 
tory value, are often'supposed to rank pretty much in the morphological 
order, embryo, seed, fruit, gynoecium, androecium, corolla, calyx, in¬ 
florescence, bracts, leaves, stem, and root. A glance at the characters that 
divide Monocotyledons from Dicotyledons is sufficient to show that this is 
oniy true, if at all, in a very general way, and this impression is soon 
confirmed if one take up the work of monographing any particular group. 
Nothing but experience can decide which is the most important character, or 
the most useful in classifying any family into its genera, &c. In the 
Lemnaceae, for example, the genera are divided by characters of the root; 
in the Cruciferae the hairs, in the Acanthaceae the pollen-patterns, in the 
Umbelliferae the ridges on the fruit, and the oil-passages in the walls of the 
fruit, are of the greatest possible importance in defining genera: and so on. 
The experience must be freshly gained for each family, and the characters 
that prove by experience , in that family , to be the most widespread and 
constant, and to differ in the most distinct way between genera that are 
recognized as such from experience of all their characters, and from 
continuous distribution, will be the most important in that family. If one 
were to sort out all characters in terms of the greatest frequency with which 
they proved to be of importance, one would be quite likely to find that the 
order above mentioned was adhered to in a general way, but that is all that 
can be said. 
