AFRICAN SPECIES OF CROTALARIA, 247 
and in some species the flowers are sometimes in racemes, sometimes solitary. 
Rarely the racemes are subumbellate. Occasionally the inflorescence, as in 
C. polyantha, Taubert, is paniculate. 
The pods vary from being one-seeded, globose, and sessile, to many- 
seeded, oblong, or oviform, or cylindrical, and are either sessile or stipitate. 
They are invariably turgid, and it does not seem advisable to place in 
this genus any plant whieh has not a distinetly inflated pod. This rule has 
not been adhered to invariably. The pods are either glabrous, or pubescent, 
or tomentose, or pilose. In Sect. Fareta they are filled with wool or silky 
hairs. 
The seed in this genus is also of considerable importance for taxonomic 
purposes. The seeds of a certain number of species have recently been 
carefully investigated by Dr. Louis Capitaine *. They are generally oval 
or ear-shaped, the campylotropy of the ovules being here very accentuated. 
Dr. Capitaine divides the seeds he investigated into two series :—A. Seeds 
in which the radicular prominence is more or less circinate, or having the 
form of an overhanging boss or projection. B. Seeds in which the radicular 
prominence is feeble. In series A come C. retusa, Linn., C. striata, DC., 
C. laburnifolia, Linn., C. incana, Linn. In series B, C. verrucosa, Linn., ete. 
The earliest attempt to classify this large genus was by Linnæus in 1753 
(Sp. Pl. ed. 1. p. 714); he subdivided it into A. Foliis simplicibus and 
B. Foliis compositis, divisions which have been adopted by nearly all 
subsequent authors. De Candolle in 1825 (Prod. vol. ii.) follows Linnæus 
with further subdivisions ; but Endlicher (f Genera’ in 1836-40) relied 
entirely upon the character and shape of the pods for his sections, and as pods 
are not always present, I have not thought it advisable to rely entirely on 
characters drawn from the fruit. A most important contribution towards 
the classification of the genus appeared in 1834, in Wight and Arnott’s 
* Prodromus, the main divisions being founded on the leaves, but there 
are twelve subdivisions founded on characters drawn from the habit, the 
character of the stipules, pod, ete. Another most important paper, as already 
stated, was published by Bentham in Hooker’s ‘ London Journal of Botany ’ 
in 1843, in which the author, following on the lines of Wight and Arnott, 
relied mainly on foliage and habit to characterise his groups. Harvey, in the 
‘Flora Capensis? (in 1861-62), adopted three groups:—(1) Simplicifolize, 
(2) Oliganthz, (3) Racemosie ; but while these groups are fairly satisfactory 
when dealing only with the South African species, they do not seem 
applicable when dealing with all the African species, as we have plants such 
as C. lukwangulensis, Harms, and C. ukingensis, Harms, which sometimes 
have solitary flowers and sometimes have flowers in racemes, and thus the 
same plant would have to be both in Oliganthe and Racemose. My father, 
* ‘Tes graines de Légumineuses,’ p. 28, etc. 
