390 
AMERICAN JOURNAL OF BOTANY 
[Vol. 8 
Whenever the flower of U. americana has an equal number of perianth 
lobes and stamens, the number is usually 8 (PI. XVI, fig. 22); occasionally 
7, and rarely 9 parts occur in each whorl. In any case the suppression of 
an organ may occur. Figure 23 shows a flower with 7 perianth lobes and 6 
stamens, and figure 24 shows one with 9 lobes and 9 stamens. The lowest 
number of lobes found in U. americana was 7 and the lowest number of 
stamens was 5. 
Small lobes of the perianth frequently exist which are unobserved by 
means of the hand lens. These occur between and within two main lobes 
(PI. XV, fig. 2, a). Anatomically, the vascular supply to this small lobe 
is a branch from a strand to a main lobe. However, there is a possibility 
that the strand to a main lobe is a union of two or three strands which 
originally passed to alternating parts of the perianth, and the branch to the 
small lobe may be the result of a separation of an aggregate strand rather 
than a case of true branching. Transverse sections frequently reveal 
indefinite organization of a perianth strand at its origin which may be due 
to the passing out together of several strands from the floral axis leaving a 
single gap. 
An organized cylinder of vessels continues above the passing off of the 
traces to the perianth and to the stamens; this first suffers diminution, and 
then breaks into four strands. This latter change occurs before the stalk 
of the pistil is isolated from the tissue of the surrounding floral organs (PI. 
XV, figs. 14, 15; PI. XVI, figs. 16, 17). Two of these four strands pass up 
the posterior and anterior edges of the pistil respectively (PI. XV, fig. 3, 
1,1^; PI. XVI, fig. 17, I, and are the dorsal carpellary bundles. The two 
remaining bundles, the lateral strands (PI. XVI, fig. 17, 0), bear towards the 
posterior side, approaching each other as they ascend and apparently form- 
ing one bundle (PI. XVI, fig. 18, 0 + 0). Serial sections (PI. XV, fig. 15; 
PI. XVI, figs. 16, 17, 0) show that these strands in present development are 
apparently a continuation of the axis and not branches of the dorsal car- 
pellary bundles (see discussion, p. 404). Where the pendulous ovule 
originates this composite strand separates into four strands, one passing 
into the ovule, one soon vanishing toward the posterior side of the pistil 
(PI. XVI, fig. 19, i, i^; PI. XV, fig. 3, i, i^), and two passing upward. The 
latter branch again sends bundles into the lateral edges of the two styles 
(PL XVI, figs. 19-21, and 0^). 
This separation of the ovule-bearing strand indicates the probability 
of the former presence of more than one ovule. The branching at the 
apex supplying both styles (PL XV, fig. 3, 0^, 0^) indicates that these are the 
supply of an axillary placenta tissue. The existing ovule is in the anterior 
carpel. The aborted bundle (PL XVI, fig. 19, i^] PL XV, fig. 3, is the 
remnant of the supply to the ovule that was borne in the posterior carpel 
and which still is present in some species (Engler, 9; Bentham and Hooker, 
2; Baillon, i). 
