AFFINITIES. 267 
Habenaria and Cynorchis flavescens, is, that in the latter the 
caudicula is a long flat rostellum ; in Hebenaria, it is a long 
. canal formed by.the convolution of the rostella. "The divi- 
sions of Mr. Lindley are certainly unequal in value, conse- 
quently they cannot fairly hold the same rank. Thus Epi- 
dendree, Bolbophyllez; and Vande, are much more united 
with each other, or differ much less from each other than 
they do with those tribes represented by Orchis, Spiranthes, 
and Epipactis. In fact they differ only from each other in 
the degree of development and suppression of the separated 
portion of the stigma, i. e. the gland and caudicula; while 
they differ from the others very much in the organs of vegeta- 
tion, and in a deficient degree of consolidation of the grains 
of pollen into distinct waxy masses. 
The claims to separate rank of the three first among each 
other, can only amount at the most to their being sub-divi- 
sions of a division. Vandez probably are the most complex. 
The cause of incumbent pollen-masses requires to be ex- 
plained ; accumbence is in accordance with what we know 
of the direction of division of an anther; but in incumbence 
the principal secondary longitudinal septum is wanting. 
Whatever amount of imperfection may seem to us attributa- 
blé to the necessity of foreign agency to ensure fecundation, 
is sufficiently compensated by the admirable adaptation of 
the pollen masses, and the whole apparatus to that agency, 
by the immense number of ovula, and by the powers of the 
life in the individuals, which in very many is certainly great. 
We may fairly attribute more importance to this last circum- 
stance, since there is no instance of a formation requiring 
foreign agency occurring in any plant of distinctly limited 
growth, except perhaps Rafflesiacee ? 
What is perfection among one set of plants, may be imper- 
fection or aberration in another ; the most striking instance of 
this occurs in Leguminose, Sect. Minoseæ ; in which we not 
only have regularity of corolla, but gamopetalism. 
The approach to regularity of corolla in the Family alluded 
