458 Small. — Pollen-presentation Mechanism in the Compo sitae. 
Australia. He regarded the Helianthoideae as nearest the primitive type, 
and evidenced the free anthers in the female flowers of a sub-tribe, Petrobieae, 
in addition to the presence of paleae on the receptacle and the paleaceous 
pappus of the tribe, in favour of this view. The capitula of the Petrobieae 
are, however, dioecious, and this sub-tribe of four species seems in several 
other characters to be far from primitive. From his remarks on the com¬ 
parative antiquity of the tribes, one gathers that he considered there had 
been three main lines of development, the first giving the Eupatoriaceae, 
Vernoniaceae, Cynaroideae, and Mutiseae; the second giving the Helian¬ 
thoideae, Helenioideae, Anthemideae, Asteroideae, Senecionideae, and 
Inuloideae ; the third giving the Cichoriaceae. 
There has been a number of efforts made to elucidate the phylogeny 
of the Cichorieae, mostly by French anatomists. Vuillemin ( 20 ), in 1884, 
after his elaborate research, concluded that anatomy was of very little use 
in the classification of this order. Col ( 4 ), in 1899, from his researches on 
the secretory canals and laticiferous tissue in the Compositae, concluded 
that the Liguliflorae were derived through the Arctotideae and Calenduleae 
from the ‘ Radiees \ Lavialle ( 11 ), in 1912, from his study of the develop¬ 
ment of the fruit, and in particular the structure of the mature pericarp, 
related the Liguliflorae to the Cynareae, on the one hand, through 
Cichorhim , Catananche , and Scolymus, and to the Mutiseae through Mos- 
charia on the other. Dufour (6), in 1907, and Lebard ( 12 ), in 1913, after 
studying the cotyledons of many seedling Cichorieae, applied their researches 
to the phylogeny of the genera in that tribe. 
The classification of the family suffered many changes before the 
publication of the * Genera Plantarum ’ ( 2 ), but since then Hoffmann ( 10 ), in 
the ‘ Pflanzenfamilien’, has adopted Bentham’s classification in all the chief 
points, and Wettstein ( 21 ) follows Hoffmann entirely. Small ( 19 ) raises 
the family to the rank of an order, which he names the Carduales and 
divides into three families—Ambrosiaceae, Carduaceae, and Cichoriaceae. 
The separation of the few abnormal genera included in the Ambrosiaceae 
seems rather unnecessary, and Hoffmann’s classification is adopted in the 
following analysis of the variations in the characters of the essential floral 
organs, which forms the first of a series of studies undertaken with the 
purpose of elucidating the inter-relationships of the tribes. 
Pollen-presentation Mechanism. 
In order to arrive at a natural classification of this very homogeneous 
family, many attempts have been made to utilize characters which in more 
varied families are regarded as negligible. Even the achenial hairs have 
been studied for this purpose ( 15 ), but although the structure of the hairs of 
an order may prove valuable sometimes, as in the Cruciferae, in most cases 
