AFFINITIES OF DAVIDIA INVOLUCRATA. 313 
seated in small pits which ultimately become raised ; (2) the number varies from one to 
twelve, approximately, emerging chiefly in groups of five or six; (3) the emergence is 
chiefly in irregular circles or ellipses; (4) the vascular bundles of individual groups of 
stamens, with some coalescence, form few-bundled irregular cylinders which merge into 
others in the neighbourhood of the general plexus ; (5) the stamens are not articulated 
and there is no direct evidence that we are dealing with reduced male flowers. This is 
a case which needs a comparison with fairly closely related forms. The apparently 
isolated position of Davidia renders this a difficult matter. There are several natural 
orders with which Davidia is supposed to have some affinity, and two of these, the 
Euphorbiaceze and Hamamelidacez, show reduction series with regard to the male 
flowers. Among the Euphorbiacee there is the well-known Anthostema- Euphorbia 
series. In Anthostema the male flowers are unistaminate, each stamen being 
articulate and surrounded at the point of articulation by a toothed perianth. This 
structure is replaced in Zwphorbia by an articulate stamen. In Euphorbia the male 
flower has almost lost its individuality. We might consider that the inflorescence of 
Davidia is made up of a globular head of similar flowers which have completely lost 
their individuality. That the negative evidence of Davidia is of little value may be 
better realized from a study of the subtle case of Altingia (Hamamelidaces). The 
stamens in this genus are arranged in a spike. This organ affords no clue whatever 
in some of the species, but in others there is some trace of a bract associated with 
individual stamens. If it were possible to telescope-up inflorescences of this kind, it is 
conceivable that a structure might result resembling the male inflorescence of Davidia 
both morphologically and in vascular structure. The above-mentioned evidence may 
be used with equal force in support of either view, so that the case becomes purely 
speculative. 
The solution of the problem depends upon an interpretation of the peculiar morphology 
of the hermaphrodite flower. In earlier discussions on this subject, when the number 
of the epigynous stamens was uncertain and the course of the staminal bundles in the 
ovary unknown, it was difficult to compare or contrast the hermaphrodite flower with 
known forms with the exception of the apetalous, perigynous flowers of Fothergilla. 
In both cases the ovary is surrounded by a ring of stamens pepe number. The 
perigyny of Fothergilla alone is not an insurmountable barrier, since transition from 
perigyny to epigyny may be found within the genus Saxifraga, and the Hamamelidacese 
comprise hypogynous, perigynous, and epigynous forms. Apart from perigyny, however, 
the ovaries of the Hamamelidacez and Davidia are fundamentally dissimilar ded 
respects and the resemblance is a purely superficial one. The formation of a single ring 
of stamens such as occurs in Davidia and Fothergilla is not difficult to explain. Two 
possible processes may be illustrated by the hybrids Gerrye ane and Corylopsis 
spicata respectively. The male flowers of Garrya Thuretu are remarkable. <A 
Single-stalked structure arises in the axil of each bract. "Hon is made up of three 
congenitally-fused flowers. There are five or six petals irregularly situated. The 
stamens vary in number (about ten or eleven). Some of these alternate with the 
petals, whilst four, more or less rudimentary and more centrally apa surround 
AZ 
