If the main objective of anthropology is the study of variation among and within human groups (Gomila, 1976), it is necessary to ‘determine first the structure and distribution of any identifiable units’ (Roberts, 1965). But this task presents difficulties. Such units are fluid. Human populations are groups of living and reproducing entities. Like any living system, they demonstrate a combination of different levels of organisation and integration, primary as cells, secondary as tissues and, yet more complex, tertiary as the whole morphology. Obviously these different levels and their combinations have different functions, the regulation and evolutionary meaning of which have barely been explored.

It is convenient to distinguish two different levels of population organisation and structure (Lefevre-Witier, 1976). A first level of discontinuity is that of the ‘genetic population’. This consists of the mating circle and the circle of fertile offspring. This definition differs from that of isolates of Dahlberg (1948) which concerns mating only; from that of the panmictic mendelian populations of Dobzhansky (1970) which extends the circle up to panmixia but excludes mate selection; and from the isogamic population of Malecot (1966) which also excludes selection. These three have in common a human group where the maximum of gametic transmission and exchange occurs as well as a strong tendency to genetic homogeneity; the principal limitations are the absence of differential fertility and infant mortality generation by generation expressing the action of natural selection. Our definition is more comparable to the endogamic circle of Henry (1968) and the natural effective population of Wright (1946).