268 DR. P. GROOM ON THE LONGITUDINAL 
In all these families, so far as my observations go, the internode-curves of the 
alternate-leaved types are irregular and zigzag, but are capable of analysis into two 
subcurves of greater or less regularity. The internode-curves of the opposite-leaved 
forms are much more regular and show more or less regularity of form. Thus the 
evidence points clearly in the direction of the opposite type of phyllotaxis as being the 
one originally characteristic of the whole cohort, and the alternate type as being derived 
from this by relative longitudinal displacement of the leaves. 
Irregularities in the internode-curves and in the subcurves are due to two causes :-— 
(1) Inherent causes, as in the case, say, of the Chenopodiaceze, where the recurrent 
zigzag is due to leaf-displacement; also in the case of branches where, as proved in my 
previous paper, a double maximum may arise as the internode-curve or subcurve changes 
from the normal type to the descending type by the relative or absolute increased length 
of the more proximal internodes or segments. 
(2) External causes, such as changed supply of water, of light, of heat, and so forth. 
At present no detailed observations have been made on the actual changes in the 
internode-curve occasioned by changes in the surroundings, though the characters of 
such may be inferred by comparing the internode-curves of the main stems, or of the 
corresponding branches, of one species. My published and unpublished observations of 
this nature render it safe to infer that sharp, sudden, and repeated irregularities of the 
internode-curves in the alternate-leaved Centrospermse and other species where opposite 
and alternate phyllotaxis are represented in one genus or species are inherent, and are 
distinguishable from the occasional or more gradual irregularities that are due to external 
changes. 
But I hope in the future to make more detailed observations on the modifications in 
longitudinal symmetry induced by external changes. 
The Centrosperme, apart from the structure of their flowers and phyllotaxis, display 
a number of morphological features common at least to several of the constituent 
families. 
In some families there is a strong tendency for the axes to end in flowers and for the 
branching to be eymose: for instance, in the Caryophyllacez, Phytolaccacee, and 
Portulacaceze. And the partial inflorescences or branching often assumes a dichasial 
form, tending towards a cincinnoid type by strengthening of one of the two branches, or 
may assume a purely cincinnoid form by non-development of the weaker branch. This 
cymose branching with the strengthening of one branch at a node reaches a climax in 
the Nyctaginaeez. In the Chenopodiacexe, Nyctaginacese, and some other families the 
emission of accessory branches is a very general feature. 
In the inflorescence region the precocious development of branches is often associated 
with the “fusion” of branches with the main axis, as is especially seen in the Cheno- 
podiaceze, Phytolaccaceze, and Portulacacez. Where the relatively main axis terminates 
in a flower as in the latter two cases, this fusion is associated with the pushing aside of 
the true end of the stem and the production of leaf-opposed or extra-axillary inflorescences 
recalling those of the Boraginaces. These “fusions” of the branches in the inflor- 
escence-region, together with the leaf-displacements, are responsible for the formation 
