Abstract
This paper summarises and discusses the three previous parts of this series. Variation in vegetation is multidimensional and vegetation cannot therefore be classified in an hierarchy. Although variation is continuous, it is usually possible (especially in a limited region) to find communities which occur frequently in almost the same form. It is convenient to describe and characterise these as points of reference. Constancy is often the most valuable criterion for characterisation, because there is then no ambiguity in the unit characterised. This is not true if "fidelity" is used. Succession is considered equivalent to the other lines of variation in vegetation. The principal value of phytosociological methods lies in their use for the solution of ecological problems by a "process of successive approximation". By extensive observation the possible factors responsible for the distribution of a community can be progressively narrowed down until a situation is found which may be economically submitted to experimental confirmation.