510 BOTANICAL GAZETTE [JUNE 
The first case to attract our attention was that of Agapanthus 
umbellatus L’Hér., one of the South African Liliaceae. Although 
ordinarily monocotyledonous, a seedling was found with two well 
developed cotyledons. This discovery led us to hope that it might 
be used in determining the relation between monocotyledony and 
dicotyledony. The two conditions shown by the seedlings of 
Agapanthus are illustrated in fig. 1. If the seedlings of the same 
species are thus indifferently monocotyledonous or dicotyledonous, 
there must be some evident relationship between the two conditions. 
In the following account no attempt will be made to cite the 
literature of the subject. It is sufficiently well known to students 
of Angiosperms, and most of it is available in CoULTER and CHAM- 
BERLAIN’S Morphology of Angiosperms. When the remaining mate- 
rial that has become available is investigated, a more detailed 
account of the whole subject of the development of cotyledons will 
be given. 
The structure of the monocotyledonous seedling of Agapanthus, 
a photograph of which is reproduced in fig. 1, is shown by the series 
of transverse sections given in figs. 2-13, and by their diagrammatic 
reconstruction in fig. 14. The section through the cotyledon and 
first leaf above the cotyledonary sheath (fig. 2) shows three vascu-, 
lar strands in the cotyledon, arranged in a triangle, with the xylem 
directed toward the center. If the middle strand had not been 
laid down, the two laterals would show the inverse orientation 
that has suggested that the monocotyledonous condition has 
arisen by a fusion of two cotyledons. The cotyledon above the 
sheath is cylindrical, but as the sheath is approached it begins to 
invest the first leaf (fig. 3). Soon the sheath becomes a complete 
ring in transverse section, which increases in thickness around the 
leaf as the transition region is approached, until, at the junction 
with the first leaf, the investing sheath becomes of almost uniform 
thickness (fig. 9), that is, the side of the sheath away from the 
cotyledon is as thick as the side which is continuous with the coty- 
ledon. In all the many seedlings sectioned, the cotyledonary 
sheath is a simple ring in transverse section, without any infoldings. 
e three strands retain their positions with reference to one 
another until just above the transition region, where they approach 
