362 MR. G. BENTHAM ON COMPOSITAE. 
shortened or quite obsolete; but these exceptional species have 
other characters which leave no doubt as to their affinities, and 
after all they are but very few in proportion to the number of 
species in the tribe. 
Mutisiacez (about 450 species in 49 genera) belong also to the 
tailed class, showing the anther-bases of figs. 8 to 10, without any 
gradual reduction of the tails as in Cynaroidese, but with two re- 
markable exceptions. Schlechtendahlia and Barnadesia have no 
tails at all, even rudimentary, to their anthers. The former, a single 
anomalous species, may possibly hereafter be connected with some 
other tribe; but the ten species of Barnadesia are so decidedly 
Mutisiaceous in every other respect, that they must be accepted ' 
as a striking exception. 
Lastly, Cichoriaceee (above 700 species in 56 genera) are as uni- 
form in their anthers as in their styles and corollas, although there 
may be some ambiguity as to the class of anther-bases to which 
they should be referred. They have been described as tailed and 
as tailless ; and after examining hundreds of species either in the 
dry or in the fresh state, I have been left in doubt as to which is 
the best designation. The anthers are always sagittate at the 
base with pointed auricles ; but how far the fine point is produced 
beyond the end of the cells is a matter of uncertainty ; it is gener- 
ally so produced, although never to any considerable length, and 
seems to vary in that respect (within very narrow limits) in one and 
the same species ; but these niceties are difficult to appreciate, and 
I may not always have been sufficiently careful in my dissections. 
There are some other differences in the anthers, such as the 
proportion occupied by the polliniferous portion, the degree in 
which the anther-tube is exserted from or included in the corolla- 
tube, &c., in which I have been hitherto unable to discover any 
good generic indications. Steetz and some others have also esta- 
blished genera on characters derived solely from the shape of the 
pollen-grains ; but this character has been accurately observed in 
by tar too small a number of species to be as yet made available 
for systematic purposes. It would require the close observation 
and study of years to decide upon its value ; and if really suffici- 
ently connected with other characters to establish it as a natural 
one, it never would practically be very useful,as requiring a high 
microscopical power to verify it. It has been generally said that 
the pollen is angular (usually dodecaedrous) and scabrous in 
Cichoriaces, globular or elliptical and smooth in Mutisiacese, 
