— 212 — 



bilities, and most of them, probably all, we find realized in nature 

 if we examine a sufficiently large number of individuals. But 

 if we remember that the group also varies with regard to the 

 length of the spur, the shape of the petals (narrow or broad), 

 the size of the angle between the two upper petals, the presence 

 or absence of a point upon the spur-bearing petal, the shape of 

 the pollen, (with 3-, 4-, 5- or 6-edged equatorial-belt), the growing- 

 form of the plant, pubescence, leaf-characters, stipule-characters 

 — and probably in many other respects, we get a slight idea of 

 the huge variety displayed by this critical group. 



When we consider the great number of combinations, which 

 are realized in nature, we find that Wittr o ck' s and Becker's 

 divisions can hardly be considered satisfactory. As my investi- 

 gations have proved that the multitudinous forms cross with 

 each other, and in this way constantly reproduce themselves and 

 also produce new forms, we cannot go on applying the test, which 

 Wittrock (1897, p. 51) and many others have applied in order 

 to prove, whether a plant-form is an independent one or not, 

 (that is whether the form in question in sowing seed only repro- 

 duces itself), as only the homozygous self-fertilizing forms would 

 do that. It is also impossible to name such hybrids from their 

 origin, where the variety is so large. Further, the same form 

 can be realized from different origins (have heterophyletical 

 origin, as pointed out by Raunkiær [1918, p. 240]). 



Kristof f erson's two small treatises on Viola (1914 and 1916) 

 afford striking evidence of the difficulties in using this antiquated 

 terminology in the investigation of elementar species. Kristof- 

 fer son crossed two lines of Viola, which he names 2 and 10. 

 The line 2 he determines in the following way: "Sie nahm eine 

 Mittelstellung zwischen V. arvensis Murr, subsp. communis Wittr. 

 und V. arvensis Murr, subsp. patens Wittr. var. scanica Wittr. 

 ein". The line 10 he determined as V. tricolor L. subsp. genuina 

 Wittr. /. versicolor Wittr. In regard to the outcome of the crossing. 

 Kristofferson says: "In F 2 trat eine ungemein grosse Spaltung 

 ein. Es gibt tatsächlich nicht zwei blühende Individuen, die ein- 

 ander oder einen der Eltern gleich sind." According to the old 

 method of determination, all those different forms ought to have 

 been named: (Viola arvensis Murr, subsp. communis Wittr. x Viola 

 arvensis Murr, subsp. patens Wittr. var. scanica Wittr.) x Viola 

 tricolor L. subsp. genuina Wittr. /. versicolor Wittr. When we read 



