CROSSINGS IN MELANIUM-VIOLETS 



253 



lized. The seed ol each /'i-phmt was sowed separately, of course. 

 If a rross was made with an individual of a pure line, already mul- 

 tiplied, selfed seeds were always taken from the mother plant. 



In the first years the seeds were sown in soil collected in places 

 ([uite free from VioUi; later only sterilized soil was used. In no case 

 I have found any intermixture or mutation in the pure lines. 



As mentioned in the ahove the j,'erminalion of the s»'ed was often 

 very poor and the power of ^germination decreases rapidly when the 

 seeds are stored a few years; already the third year it is oidy a low 

 percentage. I have the impression that it is necessary to sow the seed 

 immediately upon the harvesting in autumn. The 

 germination in sterilized and unsterilized soil is ahout 

 the same, and the time of germination is also about 

 the same, or about a forthnight. 



THE CONSTRUCTION OF THE FLOWER 

 AND THE POLLINATION. 



Although WiTTROCK (1897) already has given a 

 detailed description of the construction of the flower 

 in his book I think it suitable to summarize the most 

 important points. 



The biologically most important and morpholo- 

 gically the most differentiated of the petals is the 

 spur-petal. At its base, close to the entrance of the 

 spur, there is a yellow spot, the honey-guide; behind 

 it is the pollen-magazine. This is a furrow lined 

 with fine, one-cellulated hairs. In the posterior part the dark spot on 



it is rather narrow; in the anterior part it widens. *'''' ^^"*^'''?'', ^^^^ 



^ of the style. 



In V. tricolor (fig. 2 A) it is closed, as a rule; in 

 V. (iroensis (fig. 2 C, D) it is more or less open. The style (fig. 1) 

 has a pecular construction. Its lower part is thin and knee-shaped; 

 in the upper part it becomes thick and head like. On its anterior 

 side there is an entrance to the stigmatic chaml)er. Under this hollow 

 an epidermal outgrowth, the labellum (fig. 1 a — b) of the stigma, 

 is seen. In V. tricolor the inlet to the stigmatic chamber is directed 

 down-wards and at the same time somewhat antrorse: in V. arvcnsis 

 it points down-wards and backwards. 



The stamens are also of a singular construction. The filaments 

 are small; the anthers, on the contrary, are large and broad and cling. 



Vi^. 1. Stif^ma of 

 V. tricolor. The di- 

 stance a — b = the 

 width of the label- 

 lum. c is the inlet 

 to the stigmatic 

 chamber and m 



