524 Holden and Bexon.—Observations on the 
diarch root is produced by the early fusion of two of the double bundles, 
thus producing a further illustration of a transition stage from a type /3 
hemitricotyl. On the basis of our work we propose to put forward the 
following provisional explanation of the origin of the polycotylous condition. 
Polycotyly may arise in one of three ways, namely, either by cotyledonary 
fission, by the division of the growing point of the cotyledon, or by the 
displacement of one of the foliage leaves into the cotyledonary whorl. The 
first type of increase, by fission, we consider to be illustrated by our type a, 
in which each of the half cotyledons possesses a midrib of the collateral type. 
This we regard as derived from a double bundle by the wide separation of 
the constituent halves, with which has gone a loss of the median protoxylem. 
The protoxylem is indeed extremely transient in character even in the 
normal cotyledon, and it may well be that it is occasionally ontogenetically 
obsolete. 
It will be noted that this method of cotyledonary increase owes its 
origin to the qualitative division of the parent cotyledon, and although the 
fission may, though rarely, occur so early that the tricotylous condition 
obtains, yet the bundles invariably behave as the constituent halves of 
a double bundle, giving rise to one pole of the root. 
This type of cotylar increase is extremely common, and a consideration 
of Hill and de Fraine’s work will show that the majority of their half 
cotyledons 1 belong to the same category. It is notable that in two of the 
seedlings described by them, namely, Pinus contorta var. Murray ana, 
series A (cotyledons C and D), and P . sylvestris, series A (cotyledons 
d and e ), their origin is shown by the fact that the median resin duct passes 
to one member of the pair and assumes in the hypocotyl the characteristic 
position found in the whole cotyledon. In our opinion, however, cotylar 
fission of the qualitative type constitutes a finite series, and has led to no 
further developments. 
Cotyledonary increase produced by the early division of the cotylar 
apex—our type / 3 —is essentially different, being quantitative in character, 
and thus contrasting strongly with the division of type a. It is evident that 
this type of increase, which is illustrated by a perfectly graded series in our 
material, would offer an ideal method for the development of the polycoty¬ 
lous condition without calling in the aid of a system of 1 promotions * which, 
to say the least, involves considerable morphological changes. 
The initial member of such a series would be a type /3 hemitricotyl in 
which the normal structure is established either in the basal portion of the 
1 e. g. Cupressus torulosa, series C and D ; C. macrocarpa , series B ; Abies Balsamea, series A ; 
A. pectinata, series A (and the six half cotyledons of series B, C, and D); A . magnified var. 
shastensis, Picea ajanensis, series A, B, C, and D; Cedrus atlantica , series A, B, and C ; C. deodara, 
Pinus Pinea, series A, B, and C; P. gerardiana, P. canariensis, P. contorta var. Murrayana, 
series A (half cotyledons C and D); and P. sylvestris , series A (half cotyledons d and e). 
