Seedling Structure of Gymnosperms . II. 221 
behaviour of the bundles of the seed-leaves of these polycotyledonous forms 
must therefore vary in the same plant. The statement of our observations 
on the behaviour of the cotyledonary traces, contained in this present com- 
munication and in the first part of this research (Annals, ’ 08 ), very strongly 
brings out the fact that the seed-leaves, as judged by the behaviour of their 
bundles in the transition-region, naturally fall into three categories, viz. 
(a) whole cotyledons, characterized by the bundle of each forming one 
pole of the root ; (b) half-cotyledons, which are recognized by the bundles 
of two of them being required to form one pole of the root-structure ; and 
(c) subsidiary cotyledons, the strands of which have no influence on the 
number of bundles in the root-structure. 
Attention has already been drawn (p. 209) to the conclusions arrived 
at by Chauveaud in this connexion. 
Any or all of these classes of seed-leaves may be represented in 
one seedling. 
The external morphology is not infrequently useful in this present 
connexion ; attention already has been drawn to the occurrence of seed- 
leaves partially split in the longitudinal direction, e. g. Cupressns tondosa , 
Abies sibirica , Pinus montana , var. gallica , and Araucaria Cunninghamii ; 
and also to some cotyledons of a larger growth than their fellows, e. g. 
Pinus contorta , var. Murrayana. Again, seedlings have sometimes been 
seen in which the seed-leaves were obviously grouped together, a fact 
already recorded by Duchartre, 1 who stated that this grouping is more 
evident in the embryo than in the seedling. Finally, Mrs. Tansley 2 has 
described the cotyledons of Torreya as showing a marked tendency to 
lobing, and other observers have recorded the same thing for the Cycads and 
for Ginkgo . 3 
There is thus an abundance of macroscopic evidence partly to justify the 
above conclusion ; but, at the same time, if this hypothesis be correct, the 
examination of a large number of plants should provide some transitional 
cotyledons, in which the bundles should be more or less divided and accom- 
panied by an actual division of the seed-leaf itself to a degree more or less 
corresponding. Such examples are not wanting. 
Pinus contorta , var. Murrayana (Series K), had three entire seed-leaves, 
one very much larger than the rest, and having two vascular strands entirely 
separate throughout the whole length of the member. These two bundles 
rotated towards one another within the axis and together formed one 
pole of the root. This is a case of one whole-cotyledon on the way to the 
formation of two half-cotyledons (Diagram 5). 
Pinus montana , var. gallica (Series B). Two of the three seed-leaves 
were split longitudinally at their apices, and each of them had two vascular 
1 loc. citi 2 Chick : New Phytologist, ii, 1903. 
3 See Part I, Annals, 1908, p. 693. 
