422 BOTANICAL GAZETTE [JUNE 
the resemblance in such cases is mainly external; the metamorphosis is 
revealed by differences in internal anatomy. Now in this study of lilia- 
ceous seedlings we are dealing with anatomical details which are often 
uniform, or very nearly so, among allied genera of the most varied external 
form. Thus, among the Tulipeae some species of Lilium have hypogaeic, 
other species epigaeic cotyledons; the cotyledon of Fritillaria is of the 
common green rush-like form; that of Tulipa resembles Fritillaria, but 
the whole structure of the seedling is transformed by the conversion of the 
stem bud into a dropper. Yet the same vascular symmetry is found in the, 
cotyledon and hypocotyl of all these divergent forms. On the other hand 
seedlings belonging to almost every tribe of the family possess the green 
rush-like cotyledon, but it masks a great variety of vascular structure. 
Thus the reappearance of definite vascular symmetry in several lines of 
descent cannot in this case be put down to the action of external conditions 
moulding distinct types to one pattern. The one simple and adequate 
explanation of the facts is that the various lines of descent started from a 
common ancestor with vascular symmetry of this persistent pattern. 
Now the whole argument from this class of evidence depends on the 
fact that this vascular type is bisymmetrical. Two equivalent and quite 
distinct bundles traverse the elongated cotyledon; two traces in the hypo- 
cotyl unite to form a tetrarch root. A single type which is symmetrical 
about two planes is connected with several symmetrical about one. There 
is no difficulty in supposing all these unisymmetrical types to be descended 
from the one bisymmetrical type; but it is incredible that the descendants 
of distinct unisymmetrical types should all become bisymmetrical struc- 
tures of precisely the same kind. The argument may be neglected, but it 
cannot be read backwards. It cannot be used to demonstrate the forma- 
tion of two cotyledons from one. 
I do not mean to assert that the evidence quoted cannot be reconciled 
with the hypothesis of a primitive monocotylous angiosperm. The bilat- 
eral cotyledon may conceivably represent a terminal member which becomes 
modified in one fashion or another into the likeness of a lateral one. But 
this interpretation of the facts explains only how an apparently lateral 
member may be descended from a terminal one. It is a study in the 
derivation of various monocotylous types from a primitive monocotylous 
form; it gives no clue to the descent of a race with two cotyledons from 
that form. 
The same criticism applies to that class of evidence which I have called 
general. My reasons for considering monocotyledons as a race specialized 
from a dicotylous ancestor by adaptation to the geophilous habit have 
