EVOLUTION OF THE INFLORESCENCE. 533 
Church, in his treatment of the genus *, regards type C, that of H. fetdus, 
as the most primitive, then A as intermediate, and finally B as the most 
evolved, basing his ideas on the progressive reduction in the number of 
flowers. This view does not wholly commend itself to me, for the following 
reasons. In type A the bracts are very foliaceous, making the inflorescence 
ill defined. In the other two iypes the bracts are more reduced and the 
flower or flower-clusters more -sharply marked off from the foliage. The 
flower-arrangement in a species like H. viridis appears to me as perhaps 
the least evolved in the genus. It has, besides the terminal flower, two or 
three laterals ; tertiary floral branching takes place, to some extent, as well. 
The bracts are very leaf-like. Forms like H. torquatus, hort., and H. cyclo- 
phyllus may be still more primitive, as they produce only two secondary floral 
axes and no tertiary ones; or they may represent the first stage in reduction 
towards H. niger, for I regard the solitary or twin flowers in this latter 
species as due to reduction from a three-flower cluster, and so far am in 
agreement with Dr. Church. The non-foliaceous bracts in H. niger and 
the character of the second flower when present, suggest some reduction 
rather than primitiveness. 
In type A, the viridis-group of- species, the true foliage leaves are all 
radical. The shoot which appears above the ground may then be regarded 
as the inflorescence, the bracts of which still largely retain the form and 
colour of foliage leaves. In Æ. niger the geophytic condition is still more 
accentuated. The inflorescence is much less elevated and the bracts have 
wholly lost their foliar character. 
The habit of //. fwtidus and its allies is quite different. These plants 
are caulescent—in fact, dwarf evergreen shrubs. The foliage leaves are 
wholly eauline. These species, exhibiting arborescence, have probably 
retained the more primitive vegetative habit of the genus, the geophytic 
condition in the other members of the genus being derived from it 
(cf. Pwonia Moutan with the other species of the genus). But in regard 
to the inflorescence of //. fwtidus, this would seem to be a considerable 
advancement on a form like that shown by H. viridis. 
An inflorescence like that of I fwtidus can be derived from one re- 
sembling Z. viridis by increasing the number of secondary floral branches, 
by continuing the branching of these latter to a higher degree, and by 
reducing the foliaceous character of the bracts. The inflorescence of 
H. jetidus may have nine lateral axes, and the branching may be carried 
to the fifth degree. The lower secondary axes branch trichasially and the 
upper dichasially. The higher branching in both cases is dichasial, each 
axis bearing a couple of bracteoles. The terminal flower of the main axis 
is the first to expand, then it was noticed that the end flower of the 
* Church, A. H., ‘Types of Floral Mechanism,’ Oxford, 1908, pp. 10-18. 
