35 
Fundamental Units of Vegetation . 
aspect of the formation is on a very different level from the 
physiognomical one or even the “ ecological ” one on the basis of 
water-content alone. 
Whilst the goal reached by the two methods—by the ftoristic 
method advocated by Gradmann and by the habitat method advo¬ 
cated in this paper—must in all cases be the same, the latter method 
would appear to be the more appropriate, and indeed the more funda¬ 
mental, from the point of view of the study of vegetation as distinct 
from the study of the flora. The study of vegetation is not a depart¬ 
ment of taxonomy. Each is a separate department of science, 
although it is true that a knowledge of the latter is essential to a 
study of the former. The view advocated by Gradmann should not, 
in my judgment, supersede the view that the formation must be 
determined primarily by an investigation of the habitat; but 
Gradmann’s method furnishes an auxiliary and confirmatory test of 
the formation in all cases of doubtful habitats. I fully agree that 
no formation can be properly described without giving a full list of 
its species. I would go further, and insist that such lists must 
include the characteristic “elementary species” of the formation, 
for these are frequently much more characteristic of formations 
than the well-defined species; but to insist that thefloristic compo¬ 
sition of the formation is more important than the habitat is to 
maintain that effect is more fundamental than cause. 
Moss (1907) followed many previous authors in delimiting 
formations primarily by habitat, and then subdividing the forma¬ 
tions into associations. Applying the test of habitat to the natural 
and semi-natural woods of Somerset, he distinguished habitats which 
are characterized respectively by woods of oak ( Quercus Robur = 
Q. pedunculata), of ash ( Fraxinus excelsior), and of oak and hazel 
(Q. Robur and Corylus Avellana). More extended observations in 
other parts of the country (cf. Moss, Rankin, andTansley, 1910) 
prove that the “ oak-hazel woods” of southern England belong to 
two different communities, one characteristic of non-calcareous 
clays and derived from woods dominated by Q. Robur, and 
another characteristic of calcareous clays and marls, and related 
to woods whose dominant tree is Fraxinus excelsior. The 
latter woods are characteristic of many kinds of calcareous soils in 
England, but do not appear to have been described by continental 
writers. In the Peak District of Derbyshire, woods of Quercus 
sessiliflora are sharply marked off from woods of Fraxinus excelsior : 
the former are strictly confined to siliceous hill slopes, and the 
