146 
H. F. Wernham. 
realised; but the further elaboration of these tendencies will be 
met with in active progress. Thus the androecium tends, in all 
the principal families concerned, to be reduced to two stamens 
only ; the structure of the anthers and arrangement of the stamens 
is frequently specialized in various ways in obvious relation to the 
pollination-method; the shape and structure of the corolla is 
diversely elaborated in the same relation. Further, the pro¬ 
gressive specialization of the fruit foreshadowed in the parental 
apocynal stock (Apocynaceas, Asclepiadaceae), and seen also in the 
Transitional Group (Boraginaceae), plays a prominent part in the 
higher Tubiflorae and finds expression in affording critical characters 
for some of the families. Some indication of this phenomenon and 
of its significance has been given in the last chapter. 
The large majority of the forms which now concern us are 
herbaceous in habit, as is to be expected in view of their high place 
in the evolutionary scale (see chapter IV). Notable extensive 
exceptions to this are afforded by nearly all Bignoniaceae and many 
Verbenaceae. Whether or no the arborescent habit is in these 
cases significant of relative primitiveness remains to be considered 
in the sequel. 
As being the more closely connected, it would seem, with the 
Transitional Group, we consider first the group characterized 
primitively by the presence of an indefinite number of ovules in the 
ovary, and named in the last chapter. 
The Multiovulat/E. 
The characters of the families comprised in this group are 
summarised in the table facing p. 152, and the relations between the 
two modern accepted systems of classification are indicated as in 
previous tables. It will be seen that the essential points of dif¬ 
ference, beyond those dealt with in the last chapter, are not of 
any very serious importance; the chief is concerned with the 
systematic position of Selagineae, and this will be considered in due 
course. In Bentham and Hooker’s system, moreover, Multiovulatse 
and Diovulatse rank as separate cohorts. 
Setting aside for the present the highly specialized Acanthaceae, 
a glance at the table will show that the largest family concerned is 
Scrophulariacese; it comprises considerably more species than all 
the others taken together. These species, again, reveal a wider 
range of difference in floral structure inter se than do those com¬ 
prised in any of the other families; at the same time they seem to 
constitute a more or less natural and well-defined group. The sub- 
