Rubiales. 
229 
A further climax is reached in regard to the ovary; in the 
flower of Dipsacaceae the two sterile carpels of Valerianaceae have 
entirely disappeared, the net result being a unilocular ovary 
containing a single pendulous ovule; but according to Payer 1 a 
second carpel makes its appearance in the ontogeny of the flower, 
so that the ovary is developmentally bicarpellary. 
The andrcecium in Dipsacaceae consists usually of four stamens, 
diandrous species being of rare occurrence (Scabiosa spp., Morina 
spp.), and is probably in all cases oligomerous. 2 Progressive 
reduction of the androecium is observable within the genus Morina. 
The calyx in both Valerianaceae and Dipsacaceae displays 
considerable specialization in relation to fruit-dispersal. In the 
latter family it consists of two whorls (three in Triplostegia ) which 
persist in the fruit as prongs, wings, and so on, thus aiding in 
its locomotion. In Valerianaceae the calyx is obsolete in the flower 
(except in Nardostachys), but reveals itself in the fruit in the form 
of bristles, or a feathery crown. This calyx-specialization is 
intimately associated with the dense aggregation of flowers, as is 
especially apparent in the case of Compositse. In a close inflores¬ 
cence the mere crowding of the florets provides a source of mutual 
protection among them ; and the calyx, being no longer required for 
purposes of protection, is pressed into the service of fruit-dispersal. 
Dipsacaceae and Valerianaceae represent ultimate branches in 
the evolutionary tree of Inferae, and we must retire to lower 
branches for the purpose of tracing their connection with the other 
groups (see diagram, p. 230). In our backward course we are 
confronted with the problem of the archichlamydeous stock from 
which Inferae have been derived; and first let us enquire into the 
origin of that group with which we have already dealt, the Rubiales. 
Origin and Affinities of Rubiales. 
We have hinted already that Umbelliflorae is the group which 
is the most closely allied to the archichlamydeous ancestors of 
Rubiales; and a comparison between these two cohorts goes far to 
substantiate this suggestion. 
The essence of both groups is epigyny on the one hand and the 
“ umbellifloral” inflorescence on the other; both these characters 
’ Payer: L’ovganogenie compares de la Fleur, p. 630. 
2 Eichler (Bliithendiagramme, I, p. 280) maintains that tetramerou 
corollas in Dipsacaceae are the product of fusion of tvv 
members of a primarily pentamerous whorl. 
