founded on the Structure of their Seedlings. 71 
ancestor, I was not aware that this possibility had been 
suggested before. But a reference in Bernhardi’s paper of 
1832 (4, p. 584) has recently led me to consult Agardh’s 
text-book (1, p. 197). 
Agardh proposes to class the embryos of all flowering 
plants in four main groups, thus : — 
( Dicotyledones verae (all Dicotyledons except 
Dicotyledones < Nymphaeaceae). 
( Polycotyledones (Coniferae). 
r . , , ( Syncotyledones . 
ryptocoty e ones ^ Monocotyledones (Gramineae). 
Under Syncotyledones he includes Lilieae, Aroideae, 
Naiadeae, Palmae, Scitamineae (p. 197), and afterwards 
mentions as belonging to the same class, Cycades and Nym- 
phaeaceae. 
This classification was probably influenced by the fact that 
Agardh did not distinguish clearly between the endosperm 
and the cotyledon in the Monocotyledonous embryo, and 
still less in that of the Nymphaeaceae. He treats the embryo 
with two seed-leaves as the type, and considers that of the 
Syncotyledones to be derived from it by the fusion of the two 
original seed-leaves into a thick fleshy mass. 
The Grasses are considered as the only true Monoco- 
tyledons because their seed-leaf has become single by the 
suppression of the second seed-leaf opposite to it. Thus 
Agardh derives the structure of the Grass-embryo also from 
a Dicotyledonous type. 
It is remarkable that the Monocotyledonous families men- 
tioned by Agardh as typical Syncotyledones (Lilieae, Naiadeae, 
Aroideae, Palmae, Scitamineae) are precisely those on which 
I have worked. His Lilieae probably include Irids and 
Amaryllids as well as the true Liliaceae. So far I have not 
examined any seedling from the Naiadeae, but with this 
exception we have the same horizon. I cannot therefore 
express any opinion as to the possibility of a distinct origin 
for the embryo of the Gramineae. It has been proposed 
