202 Atkinson . — A Monograph of the 
where by a more or less free branching the number becomes 
quite variable. Since the branching of the generative fila- 
ments bears an important relation to their function, and to 
the variation of species, I have examined it in detail. 
Subgenus S acker ia. The period of fertilization is the most 
favorable for determining the branching of the generative 
filaments, and fresh material should be used. A longitudinal 
section of a sexual segment should be made and placed, con- 
cavities upward, on a glass slide, and mounted in water. Pro- 
tected by a cover-glass the branching can be studied with a 
high power. If the sexual shoot is of the right age the walls 
will be translucent, and the demonstration is easy. For 
about two-thirds of the distance between the ray-cells and 
the adjacent antherid-zones, the generative filaments are quite 
regularly six above and four below ; but beyond this point a 
more or less free branching occurs. The amount of this 
branching is dependent upon the greater or less growth in 
size attained by the sexual shoot, and varies in individual 
plants collected in the same place. Fig. n shows the formation 
of additional short generative filaments. This is only a mild 
case. I have found a greater number both above and below 1 . 
Subgenus Lemanea. Beside the normal number of gene- 
rative filaments, redundant branches arise, but those which 
arise near the ray-cells rarely reach the antherid-zone. This 
branching is more profuse in the more robust specimens of 
the larger species. Fig. 44 represents one half of a segment 
of well-developed portions of Lemanea australis: a is the 
central axis, b 1 and b 2 are the ray-cells of the removed half, 
b and b 3 are the ray- cells connected with the preparation. 
1 Wartmann (loc. cit.) in the main correctly observed the arrangement of the 
normal number of generative filaments in the species studied by him ; so has 
Bornemann (loc. cit.), but he incorrectly translates Sirodot, and charges him with 
an error. Sirodot (loc. cit.) states the normal number of six ascending and four 
descending correctly, but, by making cross sections of the sexual shoot near the 
antherid-zone, did not interpret the branching correctly. Ketel (loc. cit.) also 
describes correctly the normal number of generative filaments, but the branching 
of the generative filaments in the region of the antherid-zone in the subgenus 
Sacheria and the redundant branching in the subgenus Lemanea seem to have 
escaped close observation. 
