oct..4«n.i if YAK1 DA. ON THE GENL S I 





time it is almost certain that a lesser degree of differentiation 

 would lead to the production of the bifoliolate leaf of Jeftersonia 

 diphylla, from which, by the fusion of the leaflets at their bases, 

 the bilobed leaf of/, dubia would be formed. 



It is beyond doubt that the ancestors of these reduced types 

 such as Achlys and Jeftersonia had leaves more compound than 

 they are now — possibly hi- or triternate. One is therefore led 

 to the conclusion that, as far as the leaf is concerned, Ranzania, 

 Epimedium, Leontice and Nandina show the ancestral character 

 in a greater degree, and Nandina appears to be most primitive 

 in this respect. 



The Inflorescence. — The inflorescence is a kind of spike, and 

 is terminated by a single perfect flower. The presence of a small 

 pleiochasial branch at the base of the spike of A. japonica is 

 worthy of special notice. It often happens that this pleio- 

 chasium is very feebly developed and in some cases it is reduced 

 to such an extent that it is represented by a single (often more 

 or less imperfect) flower or even by a rudiment of a flower, which 

 can only be recognised by the bract subtending it. It is quite 

 possible that this bract may be completely suppressed; the flower 

 representing a pleiochasium would then appear to belong to the 

 main spike. 



If we go a step further, it is not unreasonable to regard the 

 flower with one or two additional flowers or the side of its 

 own torus as having been derived from a reduced pleiochasium. 

 The fact that the main spike is terminated by a flower 1 ' also 

 favours the view that the spike has been derived from a pleio- 

 chasium. 



Thus, we may retrace the spike of Achlys to a compound 

 pleiochasial cyme such as we see in Nandina. These are reasons 

 for the assumption that the inflorescence of Nandina is the most 

 primitive type amongst Berberidaceae, and that the less com- 

 pound cyme in other genera of this family may have been derived 

 from the Nandina-type by reduction. It is evident that the inflo- 



1) It is difficult to ascertain whether this terminal flower opens earlier than 

 some of the neighbouring flowers. 



