In the strong sense, segregation is a policy for setting aside or relegating a population forming a religious or racial sub-group by legally confining them to one or several quarters of a city (spatial segregation); these populations also undergo political, economic or legal disciminations (social segregation). Among the older segregation models we can cite the relegation of Jews to ghettoes in European cities, or the "untouchable" caste kept apart from villages and cities in India. In American cities, racial segregation of the former black slaves, established in 1870 was only abolished by the Civil Rights law in 1964. In South Africa, the system of racial separation known as Apartheid, in force until 1990, dated back to the first city planning by the British colonisers, and took the form of systematic physical separation into white quarters, black townships, and "coloured" quarters. Today, segregation most commonly results more from actual practices becoming established over time than from legal provision. It take the form of groups of individuals entertaining more or less exclusive relationships among themselves, based on various criteria. In cities, this social segregation, less and less detectable by outward signs among individuals in democratic societies, is often mirrored by a spatial segregation which makes it more perceptible, and sometimes compounds it. We can distinguish active, chosen segregation (for instance the formation of "wealthy ghettoes" (ghetto), or in some instances ethnic ghettoes) and passive segregation, or spatial relegation, which tends to exclude and group populations who cannot choose their place of residence, either because they are underprivileged, or because they are less well integrated or assimilated.
The relatively exclusive spatial distributions of certain social groups are based on criteria of income, ethnic identity, religion of culture, and this leads to the development of markedly distinctive urban quarters. Segregation indices enable measurement of the intensity and the evolution of the phenomenon. These processes were studied in the 1920s by sociologists in the Chicago school, who thus laid the foundations of urban ecology. They were only analysed in the social sciences in France from the 1960s and 1970s, despite pioneering work by sociologists such as Mararic Halbwachs at the start of the 20th century, or Paul-Henry Chombart de Lauwe in the 1950s, or again the very thorough study by the geographer Marcel Roncayolo on Marseille in 1952. The expression "social segregation" appears in the introduction by P.George to a collective work (Etudes sur la Banlieue de Paris, A.Colin, 1952) linking it to the new forms of social relationships arising from industrialisation and property speculation. Thus segregation, first approached as the mere observed differentiations in the social composition of city quarters, came to be analysed in more critical manner with works bearing very significant titles, such as those by the geographer David Harvey (Social justice and the city, 1973) and the sociologists Henri Lefebvre (Le droit de la ville, 1972) and Manuel Castells (La question urbaine, Maspero, 1972).
The models of quantitative geography developed by English-language researchers were called on, as were the Marxist categories of urban sociology, in research on social geography, to measure and explain urban segregations in the light of social class structures – segregation or social division of the urban space was seen as reflecting the social division of labour. «Suburbs» and large urban ensembles in particular were widely explored. Segregation, most often assessed in terms of localisation and residential strategy, has also been studied for its role in engendering inequalities in access to services and amenities, in particular by M.Pinçon, E.Préteceille and P.Rendu (Ségrégation urbaine, Anthropos, 1986). It has however been far less explored on the basis of representations and use of space. While the question of the articulation between spatial segregation and social segregation is the subject of debate among researchers (Jacques Brun and Catherine Rhein, La ségrégation dans la ville, l’Harmatan 1994), on the French political scene there has been the emergence of the notion of "social exclusion" in the discourse of urban policy, particularly after the 10th plan, so much so that the Loi d’orientation pour la Ville in 1992 could have been called the "anti-ghetto law".
The separation of social groups within across urban spaces is mainly observed on the basis of residence, and far more infrequently on the basis of the places that are used and frequented by the population. Residential segregation has different forms of spatial organisation according to whether the populations are differentiated on socio-economic, ethnic or demographic criteria. Socio-economic segregation has strong, stable structures that follow land and property price gradients across the urban area. The general model observed in the largest urban areas in concentric, and distorted by contrasts between rich and poor districts, as observed by Hoyt as early as 1939 in Chicago. There are also distortions relating to differences in income and social status of the residents, higher in the west that in the east of Paris and London (though this does not appear as a geographical parameter, since in Brussels the wealthy quarters tend to be in the east). Strategies to support the exclusive appropriation of the pleasant quarters (les beaux quartiers), including public spaces, by certain categories have been analysed by Monique Pinçon-Charlot and Michel Pinçon (Dans les beaux quartiers). In less extensive cities, for instance in France Lyon and Marseille) the spatial segregation of the rich and the poor is often just as marked, but the spatial configuration is less clear, because the quarters chosen by the wealthy are selected not so much for reasons of general access, as for the amenities provided by the site – for instance a hill with a view or a meander on the river that offer a living environment felt to be particularly pleasant. Distribution is partially reversed between European cities, where the centres are often inhabited by wealthy residents, and North American cities where it is the peripheral areas, the suburbs and city fringes (and also the so-called gated communities) that draw the most wealthy.
Demographic segregation concerns the composition of populations of given areas in terms of household age groups. The general model follows the distribution of the size of homes, and forms concentric zones, tending to contrast central areas housing populations that are older on average, with peripheral zones housing younger populations, with increasingly larger families in the more peripheral areas. This is particularly characteristic of French cities, even those of moderate size.
The ethnic dimension can be marked, more so in the United States than in Europe. The models for ethnic spatial distribution are very varied, as the spaces becoming available to foreign populations, often among the poorest in the case of recent immigration, are relegation spaces, either located in districts with old housing close to the city centre, as is most often the case in the older cities in North America, or in districts of cheap housing estates or tower blocks situated in peri-urban zones where access is difficult, as in the Paris suburbs.
Models were devised at an early date to simulate the processes of segregation. In 1965 Richard Morrill applied Hägerstrand’s Monte Carlo model to reproduce the diffusion of the black ghetto in Seattle, according to an encroachment process close to that described by the sociologists of the Chicago school. Schelling’s model (1978, Micromotives and macrobehaviour) is based on the hypothesis that it is the preferences of individuals that determines urban segregation. He used a cellular automaton to show how individual preference, even slight (of the order of 35%) for neighbours belonging to a category close to one’s own is very likely to generate the formation of homogenous districts, and virtually complete segregation of the populations of a city. This model shows how local fluctuations generate a stable, non-planned structure, on the scale of the overall city. In contrast to what certain specialists in socio-physics maintain, however, this model does not explain urban segregation. The explanation for segregation is not physical, it is social and institutional. Even if we restrict ourselves to an individualistic explanation in terms of "preferences", these preferences are sustained by feelings of insecurity towards others and otherness, and the preferences always originate from, and are fostered by, collective representations. Most often, it is legal provisions that generate these segregations, and contribute to maintaining them.