parti} | THE MECHANISMS OF FLOWERS. 349 
longer and more radiating, and their reproductive organs diminished 
 eapitula were complete. 
_ Varying in the opposite direction, the corolla in some of the 
_ marginal florets got smaller, and gradually the anthers became 
_ brown and devoid of pollen; the same changes proceeded step by 
step towards the centre of the flower, until now we find the whole 
_capitulum consisting only of female florets, and measuring only 
 15to18mm.indiameter, When this state was reached, the corollz 
of the marginal florets began again to increase in size and to radiate 
outwards, their anthers vanished completely, and their stigmas ceased 
to unfold. 
_. The capitula on the same plant are always much of the same 
form. In any locality where capitula of the primitive form occur 
along with female capitula, Centaurea Jacea may be said to be 
& mnodicecious. But since in the outset of the last described series 
of changes the outer florets become purely female, before the 
 conspicuousness of the head is visibly diminished, it is clear that 
my explanation of the origin of gynodicecism suggested in the 
eases of Thymus and Glechoma cannot hold in this instance. 
It seems much more likely here that the reduced size of the 
: corolla in the female flowers is the result of a tendency to 
abortion, spreading from the stamens to the corolla. This is 
_ Darwin’s explanation of all cases of gynodicecism (No, 167, 
| chap vii.; 611). 
| ’ _ 258. CENTAUREA Scapiosa, L., agrees in the whole structure 
| of its flower with the previous species, but differs only in its honey 
' being more easily accessible. For while the tubular parts of the 
_ disk-florets are longer than in the previous species (11 to 12 mm.), 
_ the expanded part above, which becomes filled with honey, is 
“shorter (32 to 4 mm.), and while in the primitive form of C. Jacea 
_ the marginal florets and the disk-florets were alike, I have always 
- found in C. Scabiosa the marginal florets devoid of stamens and 
| style, devoid of a wide superior portion (or throat), but notably 
larger (16 to 22 mm. long), with their tubes entirely directed out- 
| wards and dividing into five linear segments, differing little from one 
p another and about 2 mm: longer than those of the disk-florets. 
| __ The less variety of visitors that I have observed is simply due 
to the fact that I have only watched this species on a few occasions 
1 4 
