118 BOTANICAL GAZETTE [ FEBRUARY 
more than the octants. I shall attempt to show that the middle 
cell of fig. 69 contributes to all the primary tissues of the 
embryo, but to change the accepted definition of the embryo 
cell on this account would result in needless confusion. 
A third possible origin for fig. 69 is that the vesicular cell of 
jig. 67 may have divided first. This is not exactly in line with 
jig. 68, but should be recognized as a possible, if not a necessary 
interpretation. According to this view, the terminal cell of figs. 
67 and 68 would be the same as that of fig. 69, and Silphium 
would agree with other Compositae in having the embryo cell 
marked out by the first division of the oospore. In the lack of 
positive evidence on this point itis perhaps best to let this stand 
as the correct interpretation, at the same time recognizing the 
possibility that the real embryo cell may be but one of the 
daughters of the terminal cell of fig. 68. 
The first wall in the cell producing the octants, or as we may 
now call it, the embryo cell, is longitudinal, and in the radial 
plane of the head (figs. 70, 77). As the cotyledons of the 
mature embryo occupy an antero-posterior position in the ovary, 
it will be seen that this wall does not separate them, but rather 
cuts each one intwo. The cotyledons are not separated until 
the second series of divisions, in the tangential longitudinal plane, 
has occurred ( fig. 72). fig. 73 shows an embryo in which the 
transverse walls followed the first longitudinal, or may even have 
appeared first; but of this latter there is no proof. 
The formation of the octants is soon completed ( jig. 74): 
usually by a series of transverse walls in the quadrants of fig. 72: 
The upper octants divide more rapidly than the lower four, and 
as the two sets differ considerably in their mode of development, . 
it will be easier to treat them separately. We may then go 
back to trace the development of the cells below the octants. 
The first walls in the upper octants are, in most cases, a set 
of anticlinals. Fig. 75 shows an embryo with this first anticlinal 
wall in one of the upper octants. The other three are not yet 
divided. All of the figures up to and including this one are of 
whole embryos, with all of the nuclei shown. The remaining 
