292 H. F. Wernham. 
For the rest, the tendency to oligomery of the androecium 
which marks the progeny of Rubiales is absent in Campanulatae ; 
and even when zygomorphy occurs of that type which concerns 
each individual flower, as in Lobelia and its allies, this is unaccom¬ 
panied by any reduction in the number of stamens. This is 
conceivably in relation to the formation of an anther-tube for the 
purposes of pollen-presentation ; the sole exception is afforded by 
the very specialized Candolleaceae, in which the stamens number 
two only and are fused with the style to form an irritable 
gynostemium {infra), comparable in complexity only with Asclepiads 
or Orchids. 
Canipanulacece. The essential floral characters of Campanu- 
latae are shewn in the table in the preceding chapter, p. 219. 
Setting aside Cucurbitaceae for the moment, Campanulaceae, the 
family second only to Compositae in number of species, display the 
widest range of variation. This fact, coupled with the arboreal 
or suffruticose nature of not a few species, leads to the supposition 
that Campanulaceae represent the nearest approximation to the 
stock to which the Campanulatas owe their origin—the Campanal 
Stock. 
In Campanulaceae the anthers of the isomerous androecium 
exhibit every degree of syngenesis, from complete mutual freedom to 
close union ( Lobelioidcce ); in the former case the process of pollen- 
presentation may take place in the bud. The corolla ranges in 
symmetry from actinomorphy to the strong zygomorphy seen in the 
split corollas of Lobelioideae. The inflorescence is typically a 
raceme, conspicuousness being obtained by enlargement of the 
individual flower; but the tendency to aggregation is observable 
and the capitulate inflorescence is attained, e.g., in Jasione, Pliy- 
teuma. The ovary is usually polycarpellary, multilocular, and mul- 
tiovulate; but there is a distinct tendency to a bicarpellary ovary, 
unilocular by reason of the imperfection of the septa. 
An interesting feature of Campanulaceae is the relative position 
of the ovary, which is not infrequently semi-superior ; or, at least, 
not entirely inferior. The tendency to epigyny is thus not com¬ 
pletely worked out in the Campanal Stock ; and this is suggestive 
of its direct derivation from the rosalian plexus, in contrast with 
Rubiales, derived as we have seen from an archichlamydeous branch- 
ftockin which the tendency to epigyny is already fully realized—the 
Umbellifloral Stock. In the Rosales, as we have already seen, the 
tendency to epigyny is actively at work, and we take the measure 
