34 
C. E Moss. 
Clements’ treatment of the subdivision of the formation is also 
on the main line of development, for his primary division of this 
unit is the association, termed by him the “ consocies.” 
The floristic view of the formation put forward by Brockmann- 
Jerosch (1907: 237, et seq.) has recently found an exponent in 
Gradmann (1909); and their views may therefore be considered 
together. Whilst these authors recognize that formations should 
be subdivided into associations, they think that the best way of 
delimiting formations is by their floristic composition. “ One begins 
to record,” wrote Gradmann (1909 : 99), “ as completely as possible, 
and in as many places as possible, the natural closed 1 plant- 
communities as one finds them in the country. It soon appears 
that certain single associations have much more in common with 
each other than with all the others. They can be united into 
groups, and each of these groups makes a formation of closely allied 
single associations. The formation comprises in this case all the 
species of the single associations belonging to it, and thus appears 
as an abstraction, in that it is not easily realized in a single 
narrowly limited locality, but only by considering the whole extent 
of all the associations belonging to it.” 
Whilst nearly, if not quite all modern ecologists and plant 
geographers will agree most cordially with this ultimate concept of 
the formation as composed of allied associations, it does appear 
that, in cases where formations are characterized by well-marked 
habitats, the method of delimiting the formation by the method 
outlined by Gradmann is, in some cases, unnecessarily circuitous. 
It is surely superfluous to have to draw up full lists of species 
in order to decide whether one is examining a sand-dune, or, 
peradventure, a peat moor ! Even Clements, than whom no one 
has more strongly insisted on the necessity of examining the factors 
of the habitat, states (1905 : 293) that “this test of a formation is 
superfluous in a great many cases, where the physiognomy of the 
contiguous areas is conclusive evidence of their difference.” 
Gradmann does not overlook the possibility of determining 
formations by habitat; for he says (p. 96) that “ since necessarily 
each formation must correspond with definite habitat conditions, it 
is quite conceivable that we can give adequate diagnoses simply with 
the help of habitat conditions for all the formations of a region. 
This would be an ecological diagnosis of which one certainly could 
not deny the scientific value.” Thus, as Gradmann agreed, this 
1 Why should the investigation be limited to closed associations? 
