VIOLARIEAE 145 



gators — mention must be made of the nyctitropous curvature of the peduncle 

 described by Kerner. During the last three or four days of anthesis the nutations 

 of the petals and peduncle are scarcely perceptible. 



The anthers do not all dehisce simultaneously. When the flower opens, or 

 even a day earlier, the upper anther opens, the two lateral ones do so a few days 

 later, and the two lower stamens are the last to dehisce. 



The pollen-grains are dimorphous or trimorphous ; as seen from the front 

 they are quadrangular or triangular, more rarely pentagonal ; in side view they 

 are ellipsoidal. 



With regard to the functions of the various parts of the female apparatus during 

 pollination, Wittrock arrives at the following conclusions, which differ considerably 

 from those hitherto accepted. — 



The form and structure of the stigmatic valve appear not to have been 

 correctly known before, its function being therefore misunderstood Wittrock 

 describes it as an epidermal outgrowth, short, fan-shaped, and very small, and 

 composed of hyaline club-shaped cells, moderately stiff, and beset wiih papillae. 

 In the middle part of the valve these cells are arranged in four or five layers, those 

 of the middle layer being the longest, while those of the other layers get successively 

 shorter as the surface is approached. The sides of the lobe are made up of three 

 layers, the cells of the middle one being the longest. Wittrock finds that the valve 

 is only slightly flexible, so that when the proboscis of an insect is withdrawn from 

 the flower it neither closes the stigmatic cavity nor presses pollen into it. The 

 valve therefore plays but a subordinate part in preventing self-pollination during 

 the visits of insects. The knee-like joint in the lower part of the style is much 

 more effective for this purpose. Its elasticity enables the well-known upward 

 curvature of the stigmatic head from the pressure of an insect's proboscis. 



Wittrock made observations at Stockholm on the visits of insects to wild plants 

 of V. tricolor Z., var. versicolor Wittr., and from these he draws the following main 

 conclusions : — The blossoms of V. tricolor are both Lepidopterid and Hymenopterid 

 flowers in Central Scandinavia. Most insect visitors effect cross-pollination. Among 

 nectar-thieves are a few of the smaller Hymenoptera, e. g. Odynerus oviventris L. , 

 and also the fly Ocyptera brassicaria Fahr. The beetle Cetonia aurata L. devours 

 stamens and petals. The po-dvg. Thysanoptera may in certain cases effect self- 

 pollination. Legitimate insect visitors are few during the height of summer. 



In V. arvensis Murr. the pollen-magazine is quite open in front, so that there 

 is no sharply-defined pollen-cavity, and the pollen-grains can fall unhindered into 

 the stigmatic cavity. Self-pollination, therefore, regularly takes place. The August 

 flowers are usually entirely devoid of nectar-streaks. In autumn flowers the upper 

 petals are obviously smaller, in proportion to the sepals, than in spring and summer 

 ones. In V. patens, blossoms in which the petals are sometimes greatly reduced 

 may also occur in the earlier parts of the year. These are usually borne on axes 

 of a high order, and possess normally developed and fully functional reproductive 

 organs, even when the petals are reduced to small scales ; in this respect, and also 

 in having an open corolla, they differ from cleistogamous flowers. 



The busiest visitor observed by Wittrock, in the neighbourhood of Stockholm, 

 was Bombus subterraneus L., which, on account of its activity, may be compared 



