2 XCVIII. ACANTHACEA (BURKILL AND CLARKE). 
in some strobilate spikes, the upper bracts may contain a solitary flower each, 
while lower “bracts” on the same spike may contain 2 or 3 flowers, and (though 
beyond question homologous with the upper bracts) be termed, therefore, floral 
leaves. 
The characters taken from the pollen are exceedingly useful in determining whether 
a single flower in a scrappy specimen is a Fuellia or a Dyschoriste. But it must not 
be supposed that the characters taken from pollen are more “‘ absolute” than other charac- 
ters. Thus, Lindau calls the ellipsoid pollen of Lepidagathis ‘‘ honey-combed,” but he 
adds a caution that it is in many species only reticulated ; and, in some species, 1 
should say that this reticulation is so broken and imperfect that it cannot be 
differentiated from the granulated patches very generally present. In other words, I 
consider that there is a perfectly graduated series of pollen in the Acanthacew from 
honeycombed through reticulated to smooth ; and above all in the Hujusticiew, I con- 
sider that there is a complete gradation from the “tubercled” pollen characteristic 
of Justicia proper to that of Micoteba (Lindau), Duvernoia, or Lhaphidospora. 
The difficulty, however, of subdividing Justicia, so that it may be possible to locate 
a species definitely in one of its subgenera, is so enormous that any character 
which assists to such a desired consummation should be welcomed. When, however, 
Lindau throws over inflorescence, bracts, anther-cells and anther-tails, capsule-dehis- 
cence, number of seeds, nature of seed-coat, and forms a genus Micoteba or Duvernoia 
on pollen-characters only, we find the group contrary to nature itself ; it appears to me 
not a genus, but a handful of species taken at random from every part of the genus 
Justicia (in a very wide sense). Nor am I able, by such pollen-characters, to refer a- 
plant to its genus. 
Throughout the Order, the most valuable characters for forming the larger groups 
appear to be the number of ovules (either 2-1, or 3-several, in each cell), the number of 
stamens, the capsule and seeds. The combination of species in the Genera Plantarun’ 
made by Bentham, on a mixture of these characters, appear to place like with like 
much better than giving the pollen-characters a predominating value. 
Both Bentham and Lindau follow their predecessors in giving a high value to the 
elastic rising of the placenta in the fruit of Rungia, &c. Ihave here followed them ; 
but the giving this character a prepotent influence appears to me to separate like fromf 
like nearly as flagrantly as the pollen-character when similarly applied as of pre# 
dominant value (see the observations in table of genera upon Rungia, Macrorungia§, 
Dicliptera). if 
The large strong retinacula holding up the seeds usually suffice to refer a plant td 
this Order. The first 3 genera, where this character fails, are so strongly marked i 
other ways that they are easily recognised. A special eye should be kept on Nelsoni@) 
and Synnema, as these plants have been sometimes, on a hasty inspection, sorted int. 
Scrophulariacee. 
Excluded genera. 
Cornocutamys, T. Anders., is Loganiaceous. 
CyciocueiLon, Oliver, is Verbenaceous. 
; TRIBE I. Thunbergieze.—Corolla-lobes contorted in the bud (no lobe wholl® 
within or wholly without theothers). Twocollateral ovules in each cell. Seeds globose | 
or orbicular, without retinacula. 
Fruit fleshy with one cell only —. . 
1, AFROMENDONCIA. 
Fruit dry with a distinct beak, 2-celled 
2. THUNBERGIA, 
Trize Il. MNelsonieze.— Calyx in the African species 4-partite. Corolla-lobt 
imbricate, the posterior lobes outermost. Stamens 2. Ovules in each cell man 
superposed in two series, Seeds small, globose, without retinacula. 
Spikes dense, slender, covered with bracts below the flowering 
__ portion . . . . se . . 3. Etyrrarra. 
Spikes dense, more or less ovate, sessile or with a naked 
peduncle. . : . 
. . . . . 4, Nexsonia, 
