33 
Fundamental Units of Vegetation. 
salt marsh formation ; and his treatment therefore of the formation 
which he so closely studied was strictly in accord with the usage 
which was becoming more and more general. His formation was 
obviously determined by habitat, his associations by floristic 
composition and minor differences of the general habitat. 
Harshberger (1904), in a phytogeographical sketch of a portion 
of Pennsylvania, subdivided formations into associations. He 
determined his formations and associations by the character of the 
land on which they occur (p. 133), i.e., by habitat. However, many 
of his “associations” (cf. p. 145) are undoubtedly units which 
British authors would term only societies. 
Although many earlier writers regarded the formation and the 
habitat as vitally connected, it is to Clements (1905) that ecologists 
owe the most emphatic expression of this view. Clements (1905: 
292) stated unequivocally that “ the connection between formation 
and habitat is so close that any application of the term to a division 
greater or smaller than the habitat is both illogical and unfortunate. 
As effect and cause, it is inevitable that the unit of the vegetative 
covering, the formation, should correspond to the unit of the earth’s 
surface, the habitat.” This view, as has been shown, was by no 
means new; but no one had previously stated with sufficient 
emphasis and in general terms what must be regarded as the 
foundation of the modern treatment of vegetation. The concept is 
much more stimulating and much more scientific than a merely 
physiognomical view of the formation ; and this latter view, useful 
enough in the early days of plant geography, has now been quite 
outgrown. It is no longer possible to regard a forest as a 
“ formation,” nor even a coniferous forest. Such complex pieces 
of vegetation must be resolved into separate associations, and 
the latter rearranged into formations on a basis which shall 
commend itself to those who search after real affinities and under¬ 
lying causes. The rearrangement of associations into formations 
will not be accomplished at once, except in the case of well-marked 
habitats. Where the habitats are less sharply defined, much exact 
and quantitative experimental work remains to be done; and here 
again Clements, in his Research Methods in Ecology, has performed 
useful and pioneer work. Until much work of this character has 
been performed, until certain habitats have been more closely 
investigated, ecologists and plant-geographers must be content to 
refer certain communities simply to their associations, rather than 
hastily build up formations on flimsy foundations. 
