90 AMERICAN MIDLAND NATURALIST 



as soon as the pollen has escaped the apical membranous appendage 

 of each stamen becomes red. Simultaneously the petiole begins 

 to elongate in the curved portions thus having the two bractlets 

 on the recurved peduncles farther away than the apex of the 

 sepals. A certain variation of these characters is found in sub- 

 terranean and aerial apetalous flowers. In the plant under con- 

 sideration as \\ell as of a number of other acaulescent forms where 

 the apetalous flowers are born under or near the ground the 

 peduncles are recurved below the flower, and the short style is 

 always found curved towards the peduncle. The two stamens 

 also are on the side of the pistil facing the recurved stalk of the 

 flower (See Fig. la). Of all the specimens which I have examined 

 in sufficiently early stages of growth, I have found either well 

 developed pollen grains in the undehisced anthers or when already 

 dehisced, the pollen tubes growing from the grains into the stigma 

 and style. I have not found a single apetalous flower entirely 

 devoid of well developed stamens. The pollen grains are readily 

 seen with the low power of an ordinary compound microscope, 

 one inch ocular and two-thirds inch objective. (Fig. 3.) 



Before dehiscence particularly the pollen grains vary greatly 

 in size. The walls are not as thick as in those of the petaliferous 

 flowers, which average about* 37.5 fi, the variation being only 

 two or three microns. Those of the apetalous flowers range 

 from 23.7 f.1 to 52 ^ in diameter, and have not the peculiar three 

 plate lids on the surface through which the pollen tubes find 

 their exit, characteristic of the pollen of the petal bearing flowers. 



Pollen in stages producing pollen tubes were also found in 

 the few apetalous aerial flowers of Viola rostafa examined. In this 

 the peduncle is not recurved below the flower; the style with more 

 or less abruptly recurved stigma is longer and the spatulate stamens 

 are five in number instead of the two of the acaulescent forms. 

 Should further examination of caulescent and acaulescent violets 

 substantiate this variation in number as constant, a new and 

 important character of distinction of the groups is presented 

 helping to emphasize the fact that they may well be considered, 

 as Spach does, to be diff"erent natural genera. 



Plate 2. — Viola candidula Nwd. (Natural size.) Parts of spring flower 

 separated. Drawn in late spring phase as the apetalous flowers were beginning 

 to appear, one larger summer leaf also shown. Lower right hand corner: 

 ovary. Same with stamens in place, upper petal, lateral petal, lower petal . 



