138 BOTANICAL GAZETTE [FEBRUARY 
and the vestiges of three others appear in its lower portion (fig. 3). 
These vestiges suggest that the peduncle of the staminate head once 
possessed a more extensive vascular system. This conclusion is 
further supported by the larger number of bundles in the base of the 
pistillate head. GOEBEL (9), in explaining the existence of the 
two or three bundles in the peduncle of the staminate head of 
Ambrosia, made use of a similar hypothesis of reduction, though he 
presented neither of the above lines of evidence. In Xanthium we 
find that now only three bundles conduct all the water for 150 or 
more staminate flowers. This supply seems all the more meager 
when it is noted how scantily this head is protected against water 
loss. Only a few involucral bracts mature, and these are so small 
as to be of scarcely any service to the capitulum in its older stages. 
The conical receptacle (fig. 1), when young, so closely resembles 
a developing spike that Warminc’s generally accepted theory as 
to the spicate origin of the capitulum is at once recalled. Asin most 
Compositae, the marginal flowers appear first, and the apical region 
maintains its meristematic nature for some time. Each flower is 
subtended by a cylindrical or spinelike floral bract which, though 
furnishing slight protection, probably does not appreciably lessen 
the amount of transpiration. 
The parts of the flower appear in centripetal order. A lobed 
corolla is followed by a whorl of four or five stamens, and finally 
a pair of carpels. About one-third to one-half of the flowers in 
each head have only four stamens, not even the rudiment of a fifth 
appearing. In flowers with five stamens the corolla always pos- 
sesses five equal lobes. In those with four stamens the corolla 
may have either four equal or five unequal lobes. In the last 
case two lobes frequently represent quadrants, while the other three 
make up the remainder. of the cycle. Evidently the primitive 
staminate flower of Xanthiuwm was pentamerous with respect to 
petals and stamens. It follows, therefore, that in some cases 4 
reduction has taken place in the number of stamens and corolla 
lobes, the former yielding more readily than the latter. 
The stamens are always united by their filaments. Though 
RostowzEw (13) and JuEt (11) report that the anthers never fuse, 
in Xanthium commune the adjacent anthers are sometimes joined 
