1903] Leavitt, — Drosera intermedia 265 
REVERSIONARY STAGES EXPERIMENTALLY INDUCED 
IN DROSERA INTERMEDIA. 
R. G LEVITT. 
THE existing species of Drosera, more than go in number, without 
doubt have descended from a common original stock bearing leaves 
provided with tentacles like the tentacles found throughout the 
genus to-day. What the form, or outline, of the primitive Drosera- 
ceous leaf was, is a question which it is important to answer in 
considering the meaning of certain stages of development which 
make their appearance, under given conditions, in several species 
which I have had under observation. In the absence of actual relics 
of the very ancestors themselves, we must infer the original condi- 
tion from a comparison of the living species, from the facts of their 
geographical distribution, and especially from their comparative 
ontogeny. If the results developed from these several kinds of data 
agree, we may have a good deal'of confidence that our inferences 
are rightly drawn. The evidence can be given only in outline in 
this brief paper. The conclusion may be stated at the outset: the 
original type of leaf was probably not unlike that of our Drosera 
rotundifolia, 
When we compare the Sundews of the world we find that, as to 
leaf-form, they fall into a few classes, with seemingly intelligible 
interrelationships. The transitions between the salient groups are 
marked by intergrading species. There is, first, the group with 
strictly linear, filiform leaves (e. g. Drosera filiformis), embracing 
eight species. The still fewer species with much elongated, narrow, 
spatulate leaves bridge the gap between the linear and the rounded. 
The roundish-leaved species number about 56; that is, species whose 
leaves approach the orbicular form found in Drosera rotundifolia 
much more than they do either that characteristic of D. f/iformis, or 
that of D. binata. The next group, comprising about 15 typically 
Australian species, has the leaf-blades not elongated nor much 
broadened, crescent shaped or orbicular, and peltately attached. 
The affinities of this type are not clear, except from a study of the 
ontogenetic development; from this source, however the indications 
are unequivocal. Some members of the peltate group would afford 
a fairly satisfactory passage from the rounded type to the extremely 
