xii AFRICAN POLITICAL SYSTEMS
divided into some hundreds of separate tribes, each with its own language, organization, customs, and beliefs; but an examination of a sufficient sample shows that beneath the specific diversities there are such general similarities that we can constitute and describe in general terms an Australian type.1 The type is of course an abstraction just as 'carnivore' ox 'ungulate' is an abstraction, but it is an abstraction only a little way removed from the concrete reality. When a number of such types have been adequately defined they in turn can be compared one with another and a further step in abstraction can be made. By such a process, obviously requiring the labour of many students over many years, we may reach classifications and abstract concepts more precisely defined and more exactly representing empirical reality than the concepts indicated by such phrases as 'primitive society', 'feudal society5, 'capitalist society', that occur so abundantly in contemporary writing.
In attempting to classify human societies, difficulties are met with of a kind that do not exist in other sciences, such as zoology or chemistry. Two societies or two types may resemble each other in one aspect of the total social system and differ in another. It is therefore necessary to compare societies with reference to some particular aspect or part of the whole social system, with reference, for example, to the economic system or the political system or the kinship system. Thus the present volume presents materials for the comparison of certain African societies with reference to their political organization alone.
This, of course, involves making an abstraction of a different kind. For in any social system the political institutions, the economic institutions, the kinship organization, and the ritual life are intimately related and interdependent. In science there are right and wrong ways of making abstractions; the right ways are profitable in that they lead us to important additions to our knowledge; the wrong ways are not merely unprofitable, but are sometimes obstructive. If we are to study political institutions in abstraction from other features of social systems we need to make sure that our definition of 'political' is such as to mark off a class of phenomena which can profitably be made the subject of separate theoretical treatment
The successful use of the comparative method depends, not 1 Radcliffe-Brown, Social Organization of Australian Tribes.