12 NATURAL HISTORY OF PLANTS. 
narace@ in the complete anatropy of its ovule, und we now know of 
Connaracee in which this anatropy is, as it were, sketched out. The 
same may be said of Rutacea and Simarubee, groups to which Brunellia 
has been successively referred, though they are usually characterized 
either by glands with odoriferous essential oil, or the marked bitter- 
ness of all the parts; while Averrhoa, among Oxalidee, is now most 
closely allied to Connaracee' through Connaropsis, which would be a 
Cnestis were its carpels but free instead of being united into a five- 
celled ovary. As for the Detariee and Copaiferee, they are so close 
to the unicarpellary species of Connarus (Omphalobium), and to Tricho- 
lobus, where the carpel is also solitary, that there is no collection 
where the two groups are not to be found intermixed. There are 
really two points in which these reduced Leguminose differ from 
Connaracee ; they possess stipules and a completely reflexed ovule ; 
all other characters being similar, there is a very close affinity 
between the two groups. One more alliance remains to be pointed 
out—that between this Order and the series Spireee of Rosaceae. 
Nothing can bear closer resemblance to certain plants of this series 
with biovulate carpels than do 4gelea, Manotes, and several other 
Cnestidee ; the perianth, the diplostemonous androceum, the five 
free biovulate carpels, are all identical; and as these last are often 
nearly anatropous in Manotes, which moreover -possesses alternate 
pinnate leaves and a panicled inflorescence, all that we have left to 
separate the two types is that certain Spireee have stipules and that 
their seeds are usually exalbuminous. But as these two features are 
not even constant, the reasons which have led us to place Connaracee 
between Rosacee and Leguminose will easily be understood. 
What then are the characters by help of which we can subdivide 
Connaracee ? What characters are constant in this small order? 
Of the latter there are several, by no means without importance— 
the independence of the carpels, their number (never greater 
than that of the petals), and the number of ovules in each, 
the upturning of the micropyle, the consistency of the pericarp 
(always dry and finally dehiscent), the true diplostemony of the 
androceum, the alternation of the leaves, the absence of stipules, 
1 Its affinities with which were long since demonstrated by R. Brown. 
