SELF-FERTILIZATION OF PLANTS. . B51 
saying that it is most probably self-fertilizing.. This is quite in keeping with the 
profusion of siliquas it bears. | 
An exception to the above rule appears to be afforded by Cakile maritima. I found 
this growing at Felixstowe, and it was certainly proterandrous; for the stigma (which, 
however, is capitate) was mature, while the anthers had quite shrivelled. It matured 
only about 20 p. c. of seeds. The order of development was—calyx, all the stamens 
together, pistil, corolla. Hence it seemed to represent an intermediate condition. 
VionACEA.— Herman Müller has well described the self-fertilizing form of Viola tri- 
color; and I would regard it as an instance of a “ species" having been derived from an 
intercrossing form by adaptation to self-fertilization. It may not yet differ sufficiently 
from the typical form to be worthy of the title of ** species; " yet it would seem to be at 
least in the first stage towards the formation of one. 
Viola tricolor is constructed for fertilization by insects, and, as Müller has shown, very 
self-sterile if they do not visit it. It, however, may accidentally become self-fertile by 
the corolla withering, and so pressing the collected pollen-grains into the stigmatic 
chamber, as represented in Tab. XLIV. fig. 6 d. I have found, moreover, specimens in 
which the so-called “lip” was much elongated, and the stigmatic tissue which lines the 
* throat" extending outwards to the tip, which was highly glutinous. The consequence 
was that this ** tongue," as I should prefer to call it, had licked up the pollen, the tubes 
from which were abundantly penetrating the throat, and the ovary much enlarged 
with fertilized ovules. (Tab. XLIV. fig. 6c.) i 
I have met with other small self-fertilizing individuals of Viola tricolor in which there 
was a curious development of conducting or stigmatic tissue, not recorded by Müller. The 
globular ** head ” of the pistil of Viola is usually quite hollow, but containing some fluid, 
the pollen-grains passing into the ' head; but in the cases I am alluding to a pillar-like 
structure issued from the top of the style, and, curving forwards, protruded slightly from 
the orifice of the * head," and the extremity was there somewhat enlarged. This was 
highly glutinous, and a large quantity of pollen-grains were adherent to it, with quite a 
considerable bundle of tubes passing down the “ pillar,” they having fallen directly upon 
it from the anther-cells below. (Tab. XLIV. figs. 6a and b.) 
The existence of cleistogamous and highly self-fertile flowers in the genus [Viola is well 
known. A brief description may be adv isable. The five sepals are normal, but very 
minute. The petals are rudimentary purplish-green scales or absent. The stamens 
are from 2 (V. canina)* to 5 (V. odorata) in number. The filaments are slender, but the 
orange “flaps” large and rounded, while the anther-cells are small, oval at the basal 
part. The anthers are all alike, with no appendages, and are closely adpressed upon the 
summit of the ovary. The style is short, curved, and bears a truncated stigmatic end, 
which lies concealed amongst the anther-cells. The pollen-tubes penetrate, while the 
grains are still in situ, within the cells. As the ovary swells, the filaments are detached, and 
the anthers, becoming raised, form a stellate mass on the summit. (See Tab. XLIV. figs. 
b, c, d.) In * double” Violets I find the cleistogamous flowers become double also t. 
$ Three stamens are sometimes present as petaloid staminodia. 
+ This description was written some time ago; and Mr. Darwin, in his latest work, ‘ Forms of Flowers’ 
mm nee Lee corroborates all I have seen, but adds further details, which should be n a 
