59 
Tinctoriae (l.c. pp. 20, 21); based upon their own observations, 
repeated the fact that the anthers are 2-verticillate. There does 
not seem to be any exception to this; even in such extreme cases 
of reduction as result in the presence of only four anthers it is 
found that at least one of these belongs to a higher and more 
central whorl than the remaining three. 
The scope of the present paper is such as to preclude the con- 
sideration of the interesting speculations themselves. The varia- 
bility of the androecium on which they are based is, however, 
so great as to suggest that this portion of the flower is on that 
account less suitable as the source of a character on which to 
base a primary subdivision of the genus. It is clear, from the 
difficulty experienced by Pax and Hoffmann in dealing with the 
type of Miiller’s section 2, subsection a, that the new serial 
dispositions adopted by them is hardly an improvement upon 
that suggested by Miiller in 1866, and it is satisfactory to find 
that, if we employ for the purpose of primary subdivision that 
to which Miiller gave the second place, but which in the case 
of their section 3, Tinctoriae, Pax and Hoffmann have been com- 
pelled to disregard, we are able to maintain Miiller’s serial dis- 
position and to break up the genus into what appear to be fairly 
natural groups. 
Treating as a distinct section, 1. 7richocarpa, all those forms 
in which the carpels and ripe capsules are stellate-pubescent 
but not lepidote, and as another distinct section, 2. Ledipocarpa, 
all those in which the carpels and ripe capsules are lepidote but 
not stellate-pubescent, we are able further to subdivide Tricho- 
portion of their respective filaments; and (3. Graciles, the group 
treated by Miiller as his section 2, sub-section a, with 2-seriate 
Hoffmann, with purple. usually m e, capsules loosely 
clothed with pectinate scales, and with always tuberculate seeds; 
enegalenses, exactly as defined by Pax nn, 
subentire flat scales, and with smooth seeds 
ecog- 
nised as distinct species. These ‘species,’ it should be under- 
stood, are recorded ‘without prejudice’; the question of the 
have been distinguished by some competent authority, who has 
actually examined and compared representative specimens of 
closely allied plants. In cases where authors of equal authority 
have described a second time some form already adequately 
characterised, only the original name is given. An instance 
