Gradual, Change , and by Guppys Mel hod op Differentiation. 625 
calyx large and convolute in Dictyandra , large and imbricate in Keenania , 
breaking open irregularly in Pelagodendron , with calyculus in 
Retiniphyllum , with one large sepal in Mztssaenda, &c., 5-merous in 
$ and 2-merous in ? in Phyllis ; 
aestivation descending in corolla of Posoqueria and Molopanthcra ; 
stamens united in Capirona, Bikkia, Chiccoccus , unequal in Hippotis , 8-12 
in Praravinia , opening by pores in Rustia , by valves in Tresanthera, 
two, with 5-merous corolla in Sylvianthus , &c.; anther multilocular 
in Dictyandra, with pollinia in Randia acuminata ; 
stamens often long and short (heterostyly) ; 
ovary superior in Gaertnera, and in spp. of Oldenlandia ; 
ovaries united in pairs in Morinda ; 
ovary i-locular in Acranthera, Casasia , &c., 3-5-locular in Cuviera, &c. 
4-locular in Euosmia, 6-io-locular in Praravinia , 00 -locular in 
Timonius ; 
stigma io-lobed in Mesoptera ; 
capsule septicidal for loculicidal; schizocarp in Diodia , Richardsonia, 
Xanthophyllum, &c.; several with circumscissile capsule ; some with 
berry ; 
no endosperm in Abbottia , Guettarda, &c. ; ruminate endosperm in 
Polysphaeria ; 
radicle of embryo sometimes curved ; 
no cotyledons in embryo of Guettarda. 
There is no escape from the conclusion that characters of all kinds, 
however important they may be in classification, may be acquired over and 
over again by single genera, and therefore that they can be easily acquired 
without needing an immense period for the acquisition. 
It is also very striking that characters are more constant in small 
families than in large. This of course is well known, and I shall doubtless 
be accused of talking platitudes in bringing forward such a point. But it is 
not what one would expect upon the Darwinian theory. If the common 
ancestor of the Rubiaceae had the characters that are most widespread 
throughout the family, such as opposite leaves, regular flowers, free stamens, 
inferior ovary, capsule, endosperm, straight embryo with cotyledons, &c., 
why are members of the family found that depart from all these characters, 
when this is a large and ‘ successful ’ family, that presumably owes its success 
to its characters? These ‘abnormal’ characters do not occur in the most 
nearly related families to the Rubiaceae, so that to imagine them as 
survivors when the Rubiaceae were isolated from others by destruction of 
the intermediates, will not help the matter. If they connect the Rubiaceae 
to any families, it is not to the families usually regarded as their allies, such 
as Caprifoliaceae. Further, why do the small families, so often supposed to 
be relics, show such constancy? One would incline to expect more variety 
in them. 
In what has been said so far I have left entirely out of consideration 
what seems to me the greatest proof of all for the probability that evolution 
T t 2 
