PART III. | THE MECHANISMS OF FLOWERS. 435 
Lamnium album with five stamens; the upper lip was absent, and 
the fifth stamen, which stood behind the other four, was well 
formed, but shorter than the others. 
In Scrophularia, on the other hand, the anthers come in contact 
with the ventral surface of the insect. The fifth stamen is thus 
useless, but not injurious; whether it be present or absent is of no 
importance, and it is therefore beyond the influence of natural 
selection. Accordingly, the small black scale-like appendage on the 
upper wall of the corolla in Scrophularva (¢, 1, 2, 3), which repre- 
sents the fifth stamen, shows not unfrequently more or less com- 
plete reversion to its primitive form (4-7, Fig. 146). The more 
completely it approaches its typical form, the more perfect also are 
the pollen-grains which it bears; thus in the anther &, 4, not half of 
the pollen-grains attained the normal size ($$ mm. in diameter) ; 
most of them were much smaller ($¢§ mm.) and shrunken; in the 
anther k, 7, only a few pollen-grains fell short of the normal size. 
In exceptional cases the anther even dehisces, and lets part of its 
pollen escape. 
The flowers of Scrophularia are remarkable for being speci- 
ally visited by wasps. The wide globular corolla is about 
5 mm, in diameter, and in its base, near the superior side, two 
large drops of honey may be seen, which are secreted by the 
yellowish base of the ovary. Wasps visit the flowers in great 
numbers; clinging with all six legs to the outside of the flower, 
with the abdomen applied to it below, they easily insert their 
heads between the upper and lateral lobes of the corolla and reach 
the honey with very little loss of time.t In young flowers they 
- touch the stigma, and in old flowers the anthers with the under- 
side of the head and of the pro- and meso-thorax, and thus 
_ regularly fertilise younger flowers with the pollen of old, as 
_ Sprengel showed. Severin Axell's doubts regarding the possibility 
of proterogynous dichogamy in entomophilous flowers may be 
easily refuted by observing Scrophularia in flower in the open air. 
_ For, as Sprengel showed and as I have repeatedly observed, flowers 
aie constantly to be seen whose stigmas are covered with pollen, 
but whose anthers are still unripe and hidden within the corolla. 
Sprengel is wrong in saying that fertilisation can only be effected 
by insects. The stigma, when supplied with pollen by insects, 
bends down over the lower lip and withers, while the anthers ripen 
- and project beyond the lower border of the corolla ; but in absence 
of insects, as I have frequently seen in plants flowering in my 
1 See Sprengel (702), Title-page, fig. xxv. 
FF 2 
