230 ANGIOSPERMAE DICOTYLEDON ES 



of this species are developed in the ground, and the axis of the inflorescence is 

 therefore bent round at its base, and the flower-buds are closely covered by the 

 small bright-violet bracts, which overlap one another like tiles. Immediately after 

 emerging from the earth, the inflorescence is still bent, but it straightens as the 

 flowers become sexually mature, and is completely vertical only when the uppermost 

 one has developed. Even then the unilateral inflorescence only projects a few 

 cm. above the ground, and, as the plant is hidden beneath the leaves of asso- 

 ciated species, it is not very conspicuous. It is, nevertheless, easily found by 

 pollinating humble-bees, which, flying from plant to plant, work their way from 

 bottom to top of each inflorescence. By following the humble-bee, other stocks of 

 the plant are easily found. When the toothwort grows in an exposed situation, it is 

 rather conspicuous, for the distichous bracts on the back of the inflorescence are 

 red-violet in colour, with whitish margins. The front of the inflorescence, on the 

 other hand, is rendered conspicuous by the closely-crowded flowers. Each of these 

 possesses a violet calyx and a corolla with a red upper and a white lower lip. To 

 this it must be added that the large yellow capitate stigma projects from the flower 

 in the first stage of anthesis, and the hairy, whitish upper anthers in the second 

 stage. The undivided upper lip serves as a kind of roof. The somewhat shorter, 

 three-lobed lower lip lies close to the upper ; a gutter is formed by each of its three 

 divisions, the middle one being continued into the corolla-tube, and leading to the 

 large, roundly-triangular, and somewhat lobed nectary situated at the base of the ovary. 

 This gutter corresponds to a groove on the ovary and on the lower part of the style ; 

 this also reaches to the nectary, the abundant secretion of which collects as a drop in 

 the angle between the gland and the ovary. 



The filaments of the four stamens are still curled round during the first (female) 

 stage, the anthers lying inside the lower lip, and not yet visible from outside. The 

 stigma, on the contrary, projects from the upper lip almost before the flower opens. 

 The stigma was yellow in the flowers I observed near Kiel, the flower mechanism of 

 which I am describing, but according to Behrens it is red. At this stage cross- 

 pollination must be eff^ected by insect visitors, as the stigma can only be dusted with 

 pollen from a flower in a later stage. The flowers are hermaphrodite for a short 

 time in the transition stage which now follows, the stigma being still shining yellow 

 and receptive, while the filaments have elongated and the dry pollen has been shed 

 from the pollen-receptacle formed by the anther-lobes. This is so tightly closed by 

 a thick growth of hairs, that the pollen cannot fall out until the short, blunt point of 

 an anther receives a blow : this is given by humble-bee visitors, which are therefore 

 sprinkled with the dusty pollen when they probe for nectar. During this transition 

 to the second (male) stage, self-pollination becomes possible as the humble-bee 

 creeps back, but it cannot be automatic. During the second (male) stage the style 

 shrivels and the stigma becomes discoloured and dry. The corolla-tube, originally 

 3 mm. long, has grown to a length of 6 mm., and the upper and lower lips, which 

 were at first 4 mm. and 5 mm. long respectively, have each gained one mm., while 

 the style has not elongated, so that the stigma is covered by the upper lip. In 

 consequence of the elongation of the filaments, the pollen-receptacle is now situated 

 in the mouth of the flower, and is, as previously described, struck by humble-bee 

 visitors and thus caused to scatter its pollen. Lateral dispersal is prevented by 



