GLEASON: A NEW GENUS OF RAPATEACEAE 149 
the latter is prolonged into the midvein of the blade, the back 
of the sheath becomes the margin at the point of constriction 
and is thence continued as one margin of the blade, which is 
itself slightly inequilateral. The blade contains numerous 
longitudinal ribs, arising from the midvein, running almost 
parallel to it, and connected by numerous transverse veinlets. 
rom the axils of some of the leaves arise clusters of three 
to nine erect peduncles as much as 6 dm. long, more or less 
fastigiate at the base, and subtended by a few narrowly lanceo- 
late, colorless, scarious, one-nerved, acuminate scales 30-55 mm. 
long. ther members of the family have normally solitary 
peduncles, although K6érnicke mentions that two may some- 
times occur. They are strongly angled, at least when dry, and 
double or treble their diameter toward the apex. The inflores- 
cence proper is a capitate umbel, bearing ten to twenty flowers. 
The lowest of these is not subtended by a bract, and is borne 
on the longest pedicel, which may be 12 mm. long and appears 
almost as a continuation of the stem. On the opposite side of 
the umbel is a scarious, ovate-lanceolate, acute bract 10-15 
mm. long. When the lowest flower and lowest bract are re- 
moved, a second bractless flower is discovered, situated opposite 
a second bract. Each of the remaining flowers in turn is on a 
shorter pedicel, -until the central flowers of each umbel are 
almost sessile, and the respective bracts also decrease regularly 
in size. In the bracteal characters, Windsorina guianensis 
differs from other members of the family. Two bracts only 
occur in five of the genera: in three genera these are separate, 
in Rapatea they are barely connate at base, and in the little- 
known Saxo-Fridericia they are connate throughout into an 
enclosed cone surrounding the flowers and broken by them at 
or before anthesis. In one genus, Spathanthus, there is a single 
bract, which is adnate to the axis of the spicate inflorescence 
and closely simulates the spathe of an aroid. In Siegolepis the 
involucre is completely lacking. If the bracts of Windsorina 
are regarded as appertaining to each flower, and not as con- 
stituting an involucre to the whole umbel, differences from the 
other genera are again apparent, for such bracts have been 
observed, according to Kérnicke, only in Rapatea, in which the 
true involucre is particularly well developed. In the pedi- 
celled flowers, Windsorina agrees only with Rapatea. 
