varrut.] THE MECHANISMS OF FLOWERS. 141 
_ without the stigmas ever comimg im contact with the anthers in 
an expanded flower. But when the flowers close up, on fading, 
_ self-fertilisation abnay® takes place, and seems to result in full 
_ productiveness. 
_ Hypericum miiidaliouiacs, L., is intermediate between J. 
| perforatum and H. hirsutum in regard to the size of its flowers 
and the number of their stamens. Each flower contains between 
_ fifty and sixty stamens, in bundles consisting of from sixteen 
it to twenty-two. In the expanded flower, I have never seen the 
stigmas in immediate contact with the anthers; though such 
; contact. regularly occurs ultimately, when the mart of the flower 
_ draw towards the centre. The want of symmetry in the petals 
is noteworthy; it occurs also in other species of Hypericum, but 
not so strongly marked. Each petal is more expanded on one 
4 
€ 
a Fic. 43.—Hypericum hirsutum, L. 
2 Flower, seen obliquely from above, a, a, a, the three stigmas. 
| “side than on the other, and the broader side bears notches from 
its apex to about the middle, a black gland lying in each notch. 
3 broad and glandular side is sometimes on the right and 
_ Sometimes on the left, but is on the same side in all the petals 
| cof one flower. The flowers are visited by flies, which feed on the 
"pollen (590, 1.) 
Hypericum humifusum, L., has smaller flowers and fewer 
_ stamens (three to five in each group) than any of our other 
) species. The anthers only come in contact with the stigmas as 
the flower closes up. 
_ I have observed no insects on this species, or on #, 
 quadrangulum. 
In all the above-mentioned species, the grouping of the 
stamens in three bundles and the position of the styles between 
