TubiflorcE , 
* 5 * 
Group the ovary is almost always septate, and in one important 
branch we have seen (chapter V) that there is a definite tendency 
to secondary septation. The most noteworthy exception among 
the Contortae is afforded by Gentianacese ; while in the Geraniales 
Sapindales plexus the ovary is characteristically septate. 
In Gentianaceae the members of one section have flowers 
with ovary unilocular and placentae not projecting; in another 
section the placentae project; and in a third (Exaceae) the placentae 
appear to meet in the middle, so furnishing a septum. In a group 
of this kind, in which a progressive series leading from one extreme 
condition to another is traceable, the problem is to discover which 
of the two extremes is to be regarded as the relatively primitive 
condition—in other words, to discover the direction of the serial 
progress from the aspect of evolution. It will appear from the 
remarks that we have made in the present connection that the pro¬ 
gress is from the ur.septate to the septate condition ; so that in 
Gentianaceae, and in other families with which we are immediately 
to deal, the principle of the ovary-structure is rather to retain 
unilocularity rather than to acquire it ; whereas in the caryophyl- 
line stock (chapter III), for example, the case is the reverse. 
The unilocular, like the septate state may be the result of more 
than one cause ; in the Selaginaceae-Globulariacese group we have 
lately seen that it is due to the abortion of one of the carpels—a 
cause very different from either of the two referred to immediately 
above. 
From the closeness of their connection with the Transitional 
Group, and the considerable diversity of their floral structure, 
we are led to regard Scrophulariaceae as a group relatively 
primitive among the Multiovulatae. From this family we have 
already in the present chapter traced certain lines of descent. 
One, determined by reduction in ovule-number coupled with aggre¬ 
gation of florets into dense inflorescences, leads to the Selagineae 
of Bentham and Hooker. A second, determined by the adoption 
of a parasitic habit, leads to Orobanchaceae, which are notable for 
the retention of what is probably a primitively unseptate ovary; 
the latter character seems to link this family with Gesneraceae ( q . v. 
infra). Another line, tending to secondary unilocularity of the 
ovary, leads to the highly-evolved Lentibulariaceae, in which the 
ordinarily highest limit of economy in the andrcecium—invariably 
of two stamens only—is reached. The corolla, moreover, is usually 
highly specialized, as is also the habit. 
