DICELLANDRA AND PHJEONEURON. 489 
stone. All we can do is patiently to seek the red line ofaffinity, 
not by singling out an imaginary more or less absolute character, 
but by viewing our objects from as many sides as possible, not as 
the corpses with which we have to deal in our museums but as 
living organisms. This may be a slow way, but it is the only 
one which promises lasting results. We cannot always wait 
for it for practical reasons, and artificial systems will have 
to be set up for the time to bring some sort of order into the 
chaos; but as our science advances, the excuses grow fewer and 
fewer, and the gain in stability outweighs more and more the 
outlay in time and labour. 
To return from this excursion into the philosophy of systematic 
botany —and the same applies, of course,to systematie zoology— 
to our two Melastomaceous genera: the red line of affinity lies 
evidently not through that comparatively isolated differentiation 
of the connective which determines the homceandry or heterandry 
of the andrecium, but through the much more complicated 
and closely correlated modifications of those parts of the 
flower which ultimately form the fruits and seeds. Their 
character is the same in Phaoneuron and in .Dicellandra 
liberica, and we shall therefore have to refer the latter to 
Pheoneuron. 
As to Dicellandra setosa, I have seen only fruits. These agree 
in every point completely with those of D. liberica; and as Sir 
Joseph Hooker's description very well fits the Liberian species, 
there is scarcely any doubt that both are identical. Both may 
therefore be united as Pheoneuron setosum. Gilg’s original 
Pheoneuron dicellandroides, on the other hand, appears to me 
to comprise two distinct species, and a further species may be 
added, known from specimens collected by Moloney in Lagos. 
From seeds of this, a plant has been raised at Kew from which 
Sir Joseph Hooker has prepared a plate for the ‘ Botanical 
Magazine,’ the species being named Ph. Moloneyi*. Thus the 
genus Dicellandra appears, in the present state of our knowledge, 
monotypic, whilst Pheoneuron, on the other hand, comprises 
four species. I give the technical description of the two 
genera and their species in the second part of my paper, and 
place those of the genera side by side in order to facilitate 
comparison. 
* [t has since been published with tab. 7729. 
