346 



We could propose a subdivision into species of which the 

 leaves possess a simple and those of which they have a stellate 

 texture and it would be a rather natural one. This would be 

 a subdivision which would also be fit for Geunsia, in this way, 

 that Callicarpa would have developed polyphyletically from 

 Geimsia; but this subdivision gives not the same results as the 

 one we used in our table of determination, where we started 

 from the two different forms of anthers, the long and narrow 

 ones, and the short and the broad ones. 



Yet there are, according to the leaves (since the flowers arc 

 very conform), some groupes more distinctly limited e. g,: 



1°. the rufceZ/a-group which may contain C. rubella, C. pilo- 

 sissima, C. caudata, and to which are related C. pedunculata, 

 C. mollis and C. Havilandii, all probably derivable from Geunsia 

 Cumingiana. 



2". the longifolia-gioup with C. longi folia, C. glabra, C. pur- 

 purea, C. elegans and C. japonica, perhaps derived from 

 Geunsia pentandra and in which C. longifolia var. P forms a 

 transition to the: 



3°. /anafa-group, which contains C. lanata, C. longipetiolata, 

 C. ramiflora, C. micrantha, C. formosana, C. Blancoi, C. cana 

 and as most differentiated form C. laciniafa (with deeply divided 

 corolla). C. cana, however, shows also a close relation to C. 

 pedunculata; C. subglandulosa and C. subalbida may be derived 

 from te §-alba variety of Geunsia farinosa. 



Schizopremna, though placed among te CallicarpEj* for its 

 nearly regular flowers, had perhaps better be placed among 

 the ViTiCExE, near Premna with which it seems (according to 

 the description) to be so nearly related that we hesitated to 

 retain it as separate genus; for the same reason it seems not 

 impossible that it has never been found back since Baillon 

 discovered it in 1892, as the distinguishing characteristic (the 

 style) is hidden. Moreover it seems doubtful, whether the fruit 

 is exactly examined. The same may be said for Pygmaeopremna, 

 which differs from Premna only in its small size ' a phenomenon 

 which occurs also in Clerodendron, viz.: C. sarawakanurn). 

 Tectona may be considered as a 5-merous Callicarpa with 

 terminal inflorescence. Besides there are some more differences, 



