TubifiorcB . 
1 59 
stock ; but in Eccremocarpus and the tribe Crescentieae (formerly 
separated as a distinct family) the ovary is unilocular and the 
placentation parietal. Septation is, however, foreshadowed in these 
forms, for the protruding placentae sometimes meet, especially near 
the base of the ovary. These exceptions to the bilocular condition 
represent about 7% of the whole family. The Crescentieae, more¬ 
over, differ from the rest of the family in their fruits, which are 
indehiscent, either fleshy or dry. 
Enough has been said, perhaps, to favour the suggestion that 
Bignoniaceae lie nearer to the ancestral stock than any other 
group of the Multiovulatae. Indeed, it will be gathered that their 
advance upon that stock is little, if any, greater than the minimum 
required to justify their inclusion in the last-named group; their 
position is thus indicated on the diagram, p. 152. 
An offshoot of the bignoniaceous line is represented by the 
families Pedaliaceae and Martyniaceae, included in one family, 
Pedalineae, by Bentham and Hooker. Their affinity with Big¬ 
noniaceae is evidenced by their sub-regular flowers and usually 
didynamous androecium, often with a posterior staminode; the 
latter is, however, small, and sometimes wanting. Their advance 
upon Bignoniaceae is seen in their constantly herbaceous habit, and 
particularly in the fruit-character. The tough capsule-wall is 
produced into woody hooks, spines, etc., which may aid in the 
dispersal of the fruit by anchoring it in the wool or fur of animals. 
The ovary may be 1- or 2- locular at first, but there is a distinct 
tendency to chamber the developing fruit by the division and 
protrusion of the placentae; this is coupled sometimes with a 
reduction in the number of ovules and seeds. 
This completes our examination of the Multiovulatae, and our 
conclusions as to the general lines of affinity within the group are 
shewn in the diagram, p. 152. This includes also the Diovulatae, 
and a vertical double line separates the two groups; those members 
of the Multiovulatae in which the number of ovules is reduced 
appear correspondingly near to this dividing line. The latter is 
regarded as a more or less sharp one, and the two groups fall on 
either side of it as the respective products of two lines of descent 
diverging from the apocynal stock (see chapter V) and emerging 
already within the Transitional Group. 
We now turn to the other branch:— 
