THE TRIUMFETTAS OF AFRICA. 241 
distribution of the fruits by animals, should have given rise to the geniculate 
instead of vice versa. 
As it is obviously undesirable to propose subsections of Lappula without 
taking the numerous extra-African species into consideration, it seems best to 
indicate the principal African groups, one of which may be illustrated by 
means of a diagram. This group of species includes sefulosa and its allies. 
T. Antunesi and delicatula have uncinate prickles, tomentosa, abyssinica, and 
orthacantha geniculate, and the remaining species stellate prickles. 
The diagram serves also to indicate what species may be confused one with 
another. 
A ntunesii —— —— —— Dekindtiana 
| 
dubia — —— —— mierantha —— tomentosu 
| LG | 
| | N 2 | 
| / | N | | 
/ | | "I 
„setulosa | abyssinica — — orthacantha 
/ \ | | 
x | / | 
| / 
\ | / 
angolensis N д buettneriacea —— dubia 
\ 
\ 
paradoxa X | 
| 
УА BN | 
delicatula a intermedia 
The species allied to f{avescens form a much better marked group: they are 
trigona, benguelensis, heterocarpa, and two Indian ones, glabra and rotundifolia. 
All are characterized by a more or less spike-like inflorescence, 20 or more 
stamens, and globose or ovoid, indehiscent fruits. 
T. rhomboidea, eriophlebia, and pentandra come near the Jlavescens group, 
but the inflorescence is not spike-like, and the stamens are 15 or fewer. 
T. semitriloba and Lappula belong to an American group, characterized by 
retrorsely pilose prickles, and probably neither of them is truly indigenous in 
our region. 
The most critical group of all is that which includes scandens, cordifolia, 
macrophylla, tomentosa, pilosa, and effusa; and it has been found almost 
impossible to key these species satisfactorily, 
LINN. JOURN.— BOTANY, VOL. XXXIX. S 
