528 Holden and Bexon .— Observations on the 
Another difficulty is found, not so much in Cheiranthus itself as in the 
application of the theory to seedlings of other genera. Thus Hill and 
de Fraine ( 10 ) describe a seedling of Silene Schafta which, although 
dicotylous, is yet triarch throughout, the larger of the two cotyledons 
possessing two double bundles. The same feature, though not in so pro¬ 
nounced a form, has been met with by one of us in polycotylous seedlings 
of Centranthus rubra. These seedlings are obviously of type / 3 , and the 
structure of the cotyledons may possibly be due to a fusion or partial 
fusion of the parenchymatous tissues of the lamina having followed the 
division of thecotylar apex. Such a fusion is quite probable when one con¬ 
siders the close proximity of the two cotyledons during their development, 
and the plastic character of the tissues during the early stages. A second 
possible explanation is that whilst the plerome and the outer half of the 
periblem and dermatogen have retained their full activity subsequent to 
apical division, the activity of the inner half of the periblem and dermatogen 
has been completely suppressed, since a less complete reduction in the 
activity of these tissues is often apparent in cotyledons which show incom¬ 
plete division. 
A difficulty of another kind arises in connexion with Lotus cornicidatus 
and Carmichaelia australis , two Leguminosae described by Compton ( 3 ). 
Lotus corniculatus is of considerable interest, since the dicotyl exhibits 
a trimerous symmetry due to the persistence of one of the lateral strands 
from each cotyledon in the hypocotyl, these fusing to form a third pole ; and 
a similar state of affairs is found in Carmichaelia australis . In the former 
species Compton describes a bemitricotyl in which root and hypocotyl 
showed triarch symmetry, but which obviously corresponds to our type / 3 . 
The tricotylous specimens of Lotus cornicidatus and Carmichaelia also show 
a trimerous symmetry ; but we are inclined to regard this as not homologous 
with that of the dicotyl and hemitricotyl, but rather as an instance of the 
complete suppression of the relatively feeble pole of the triarch system 
derived from the fused laterals, and its replacement by the robust median 
strand derived from the midrib of the third cotyledon. 
This view is supported by an examination of the normal seedling 
structure of the genus Lotus , and in fact of the whole tribe Loteae. In 
Lotus the cotyledons each contain a median double bundle and two laterals, 
from which in L. tetragonolobus and L . Requieni tetrarchy results, while in 
L. cornicidatus triarchy is characteristic of the root. This is evidently due 
to the suppression of one of the ‘ paired laterals’, and it is noteworthy that 
the other is in process of suppression. Other members of the tribe Loteae 
show similar features, e. g. Anthyllis tetraphylla is tetrarch, while A. val¬ 
uer aria and A. Barba-Jovis are diarch, and another member, Dorycnium 
hirsutum , may be either tetrarch, triarch, or diarch [( 3 ) pp. 35-9]. A 
similar tendency to the suppression of laterals is also found in tribe Galegeae, 
