618 Willis .— The Origin of Species by Large , rather than by 
given case. If one were to decide that characters of embryo were the most 
important, and start upon the classification of the Orchidaceae, one would 
soon be in difficulty. One cannot take single characters, nor can one be 
sure, till one has actually worked with a family, what characters will prove to 
be of most importance in that family. Grasses, sedges, palms, duckweeds, 
aroids, rushes, gingers, bananas, buckwheats, water-lilies, pitcher-plants, sun¬ 
dews, begonias, &c., &c., are recognized by their leaves, others by the stem, 
by leaf-veining, or by other vegetative characters. One may find a character 
in one place a most important family diagnostic, like rumination of the 
endosperm, which is the only certain character of distinction between 
Anonaceae and Magnoliaceae; in another family, as in palms, it may occur 
over and over again in pairs of closely allied genera, one having it, the other 
not; whilst in some palms, like Euterpe , it may occur in some species and 
not in others. Asclepiadaceae can only be divided with certainty from some 
Apocynaceae by the occurrence of translators to the pollinia—what one 
would have thought, a priori , a character of trifling importance. There is 
almost no character of a family or genus that may not at times be generic 
or specific. So good an account of this fact, which is so well known that 
people have ceased to think about it, is given by Lindley ( 7 , p. xxi) that it 
is worth quoting at some length: 
‘ All Rubiaceae have opposite entire leaves . . . but in . . . Fuchsia , in 
which they are usually opposite, species exist in which they are not only 
alternate, but both one and the other upon the same plant ... in Com- 
bretaceae and Leguminosae, orders usually having alternate leaves, they 
are occasionally opposite ... in Aceraceae, Aurantiaceae, Geran^aceae, 
Rutaceae, and Sapindaceae, both simple and compound leaves are found. .. . 
Myrtaceae are distinguished by these glands (pellucid dots in the leaves) 
from Melastomaceae . . . in . . . Phytolaccaceae, Labiatae, &c., there are, 
however, genera with and without pellucid dots . . . The orders Cistaceae, 
Saxifragaceae, and Loganiaceae are among the . . . cases in which genera 
exist both with and without stipules. 
‘ The number of sepals is sometimes a character of importance, as in 
Cruciferae, in which they are always 4 ... in Malvaceae they are 3-4-5, in 
Guttiferae from 2 to 6 . . . Malvaceae have the calyx exclusively valvate . . . 
but in Penaeaceae both valvate and imbricate aestivation exists ... it 
frequently happens that both regular and irregular calices coexist in the 
same order, as in Rosaceae, Labiatae ... in Melastomaceae all degrees of 
cohesion take place between the calyx and the ovarium, and in Saxifragaceae 
this uncertainty of structure is still more remarkable. 
‘ If the corolla is present, a plant is said to be dichlamydeous, and much 
importance is attached to this peculiarity ; far more, I think, than it deserves. 
It constantly separates plants having much natural affinity ... in the poly- 
petalous orders of Crassulaceae, &c., &c., there are many monopetalous 
genera . . . Compositae are essentially distinguished by their valvate 
aestivation ... an exception existing in the genus Leptadenia . . . Echium in 
Boraginaceae is irregular. 
