358 MR. G. BENTHAM ON COMPOSITJE. 
Generally speaking, the modifieations of the pappus, however 
inconstant in genera or even sections, are very constant in species ; 
the exceptions are chiefly in the coroniform and other very much 
reduced pappi, whieh may be present or absent in different indi- 
viduals or varieties of one and the same species—as, for instance, 
in those forms of Chrysanthemum leucanthemum so elaborately 
worked out by Fenzl, as well as in several Ma£ricarie and some 
species of Centaurea, several Helenioides, &c. 
5. Differences in the Andrecium. 
In all essential points the androecium of Composite is as uniform 
throughout the order as the pistil; it offers none of those differ- 
ences in number, symmetry, position, divection, or structure which 
in so many orders are called in aid of the discrimination of genera. 
The anthers, equal in number to the corolla-lobes, are united, or 
perhaps, in a few cases, only closely connivent in a cylinder round 
the style, and their cells open inwardly and longitudinally without 
any variation ; so also in regard to the filaments, they are always in- 
serted in(adnate to)the corolla-tube, and attached to the base ofthe 
connective. These filaments vary slightly in the height to which 
they are adnate to the corolla-tube (a question of degree affording 
no available characters), and also in their being free or monadel- 
phous after quitting the tube, and glabrous or papillose-hairy. 
Both the latter characters have been made use of in Cynaroides 
as generie—the former to separate rather too artificially three 
monotypic genera from Carduus and Cnicus, the hairy or non- 
hairy filaments rather more successfully applied to the distinction 
of a few large genera where it proves constant. There is also, in 
many Senecionides, for instance, an abrupt dilatation or change of 
texture, and almost an articulation at some little distance below 
the insertion of the anther. But it remains to be ascertained 
how far this is constant even in the genus Senecio; and my own 
observations are insufficient to establish it as a generic character. 
The anthers, however, are sometimes provided with certain 
appendages apparently of little or no functional or homological 
importance, but which nevertheless, from the remarkable con- 
stancy of their presence or absence in whole tribes, supply one of 
the most valuable characters in Composite if applied with proper 
caution. These appendages are either apical or basal. At the 
top of the anther-tube each connective is produced into a thinly 
cartilaginous erect or incurved membrane or point, which may 
