43 8 Thiselton-Dyer . — Morphological Notes. 
rather less or more in individual cases. And they began 
to flower in May, 1903, when, therefore, about two years old. 
The appearance they then presented is shown in Plate XXI. 
This, however, gives only an imperfect idea of the singular 
appearance of the plants : the leaves were strictly decussate, 
i.e. each successive pair was set on the stem at right angles 
to those above and below. And the divisions were so uniform 
and symmetrical that they exactly corresponded when looked 
down upon from above. 
The flowering of the hybrid was looked forward to with 
much interest, and when it occurred was a complete surprise. 
The majority of the plants flowered, and in every case the 
colour of the flowers was a clear rosy pink, recalling the tint 
of those of Erythraea Centaurium. In the case of a sudden 
variation or ‘ break 5 of this kind it is usual for the colour of 
the flowers to become highly variable. But in this case it 
seemed absolutely uniform, and I could not persuade myself 
that there was any difference between one individual and 
another. 
I must confess that I was completely at a loss to explain 
how a bright pink could arise from a cross between an orange 
and a white. The explanation, however, occurred to my 
friend Dr. Lotsy, who, while staying at Kew, had been much 
interested in the hybrid. The orange colour of the flowers of 
K . jlammea is due to deep yellow chromoplasts immersed 
in a pink cell-sap. In K. Bentii both chromoplasts and 
cell-sap are colourless. The hybrid has inherited the white 
chromoplasts of one parent and the coloured cell-sap of the 
other. 
The foliage, however, exhibits in the adult plant no trace 
of the influence of K. jlammea. But it is widely divergent 
from that of K. Bentii. K. laciniata, which extends from 
Tropical Africa through India to Java, has deeply pinnatifid 
leaves with sometimes linear segments. The leaves of 
K. 'Schweinfurthii from Abyssinia are also similarly divided. 
The conclusion seems irresistible that we have in the case 
of the hybrid a reversion to an ancestral character which 
