Version classiqueVersion mobile
OpenEdition Books

Territories and urbanisation in South Africa

Frédéric Giraut
Céline Vacchiani-Marcuzzo

Part III. The spatial and territorial dimensions of urbanisation

Texte intégral

  • 1 This refers to the racial classification in force under apartheid, which divided the population int (...)

1The urbanisation of South Africa occurred in rather unique circumstances, linked first of all to the different phases of economic development which gave the main role to port cities on the periphery, and later to mining cities in the interior. The processes of urbanisation were then subjected to a shaping process in the form of the spatial and ethnic apartheid policies, on national scale and on intra-urban scale. It is for this reason that the South African town and city system is often presented as being specific or unique, and difficult to compare with any other, all the more so because the different populations have not had the same rates of growth1, so that the whole system is very heterogeneous.

A town and city system halfway between developing country and new emerging country: advanced urbanisation

Colonisation, urban frontiers and the occupation of the territory (generation of towns and cities)

  • 2 This chronology is derived from that established by R.J. Davies in 1972, which we have reshaped and (...)

2Towns in the pre-colonial period were rare, but there were some, so that it is justified to refer to a “pre-colonial stage”2 which more widely concerns southern Africa, and only the margins of the present South Africa. The urban area of Mapungubwe situated on the northern border appears as the first instance of an isolated urban agglomeration with political and economic functions. The Tswana “agro-towns,” some of which set up and developed within present-day South African territory (Dithaking, Kaditshwene), probably over several centuries from the middle of the second millennium, are certainly the most notable pre-colonial South African agglomerations for their size and longevity, and they were indeed in existence at the time of the first colonial incursions (Freund, 2007). Two capitals of warrior kingdoms should also be mentioned, Ulundi for the Zulus and Thabo Bosiu for the Basothos, also in existence as native agglomerations at the time of the 19th century colonisations.

3Cape Town was the first place to see urban concentration of colonial origin from 1652 and the establishment of the Dutch East India Africa trading company. This is the “colonial pre-industrial phase” extending from 1652 to 1806. No town, in the sense Europeans would give the word, existed on the territory before this date, and it was the implementation of the Western colonial model (Coquery-Vidrovitch, 1993) that marked the beginnings of modern urbanisation. If across the African continent generally, the African colonial town was not a white town, South Africa constitutes an exception, with Cape Town being the only white city in black Africa until as late as the end of the 19th century. Thus the originality of the establishment of the South African urban system and its evolution over time highlight its marked dependence on, and openness towards, the outside world. The first clusters of urban population, established by successive urban frontiers, were totally dependent, in particular in economic and trading terms, on the outside world. At this stage in the process of urbanisation, the main centres of urban growth were the coastal towns, the ports, and the towns situated in the immediate hinterland: Cape Town first of all, since up to the start of the 19th century ten of the thirteen established towns were situated around Cape Town, the others being Durban, Port Elizabeth, Grahamstown and secondarily East London. The arrival of the British settlers in Durban in 1824 emphasised the prime role of the ports in the process of urbanisation (the “Colonial transition stage” 1806-1833). These various urban areas maintained marked autonomy, and each developed relatively dense communication networks with their immediate influence zones, but there were few links from one to the other. There was little integration among these areas of urban growth. Before 1860, urban development thus remained linked to the growth of regional economies, entirely created by Dutch and British settlers.

Discoveries of mineral wealth: acceleration of the urbanisation process on the interior

4The discovery of diamonds in Kimberley and gold in Johannesburg (the “Mid and late colonial stage”) between 1867 and 1883 was the starting point of a different configuration of the urban structure. This change led to the appearance of a new centre, Witwatersrand, in the present-day province of Gauteng; its position in the interior of the territory and its increasing weight, rapidly reaching a dominant position, led to the reorientation of inter-urban relationships and communication networks.

5It was between 1870 and 1930 that there was the most marked acceleration in the growth of new towns in the whole history of South Africa. The combination of this phenomenon and the political unification of the country by way of the Union Act in 1910 led to a continuous reinforcement of the degree of territorial cohesion in the course of the 20th century. Thereafter, situation in relation to the main existing communications, and in particular the railways, was a determining factor (figure 8), with the region around Johannesburg becoming a railway centre that was essential in the process of urban growth. Another more secondary event occurred in the 1920s in the form of a prolonged drought, which led to population migrations from the country to the towns.

6The speed of evolution was closely linked to the economic upheavals in the last decade of the “Late colonial transition stage,” which was an economic expansion phase, characterised by the healthy mining sector (in the Transvaal–present-day Gauteng– and Orange Free State) and by the intensification of agriculture, in the course of which the foundations of the industrial expansion of the 1930s were laid. Unprecedented rural exodus, mainly made up of Afrikaner farmers, fed the growth of towns in this period. The decisive factor, after the discovery of mineral wealth, was the intense industrial development from 1933 (the “Industrial colonial and apartheid stage”). The rapid economic expansion, which accelerated with the industrial era, progressively affected the major part of the South African territory in a variety of ways.

Source: UMR Géogrophie-Cités.
Figure 8 – The establishment of the railways (1860 to 1960)

7However, the creations of towns after 1933 were concentrated in the most dynamic industrial regions, in particular the mining interior. The succession of creations very well demonstrates that the town system was young and underwent major changes in its spatial layout until very recent years. This is in particular the case with the new entities created after 1960, during the apartheid period, almost exclusively located in the bantustans or in the peripheries of the main urban areas, intended to act as de-concentration poles of attraction (Atlantis, to the north of Cape Town for instance). The establishment of an urban system, seen through the successive waves of creations of towns, provides awareness that the system is very young, that growth has been very fast, and closely linked to economic and later to political upheavals in South Africa.

Recent intense phases of urbanisation

  • 3 The Native Urban Areas Act of 1923 marks the start of urban segregation, well before the official i (...)

8In recent decades, the rates of urban growth partly reflect the political upheavals that have occurred. The influx control laws restricting flows of black populations towards the towns and cities3 which were implemented by way of passes (removed in 1986) and the end of the apartheid regime in 1994 both had repercussions on migrations between the rural and urban spaces. In addition, entry into a post-fordist urbanisation phase brought a change in the economy, the growth of which was then generated by other sectors, mainly tertiary, such as trade or tourism, notably in association with the arrival of foreign capital. These most recent changes show that between the private sector and local government, and between the private sector and the state powers, within the Growth, Employment and Redistribution policy (GEAR), relationships altered fundamentally. The main characteristic of present-day urbanisation is a more marked networking of towns and cities, and the concentration of the urban population and activity in the metropolitan areas, in a process of “metropolisation.” It should be emphasised that overall the urbanisation boom occurred suddenly and very fast, considering that it occurred within a rural civilisation.

  • 4 The primacy index has been calculated since Jefferson (1939) by measuring the relationship between (...)
  • 5 The renaming of municipalities affected the names of most of the metropolitan areas, with the excep (...)

9Thus South Africa today presents numerous spatial disparities. The different features inherited from its history that were mentioned earlier led to migratory flows that contributed to creating considerable variations. In more precise terms, the different phases of urbanisation generated three main zones of urban concentration, although without any phenomenon of macrocephaly, which is rather remarkable for the African continent. The primacy index4 between the first and second urban agglomerations is only 2. In addition, the degree of urbanisation is unequally spread according to the province (nine in number since the presidential election of 1994). Gauteng province, which includes the three metropolitan areas of Johannesburg, East Rand (Ekhuruleni5) and Pretoria (Tshwane) obviously heads the list with an urbanised population rate of 99.6 %, followed by Western Cape province (95.1 %). At the other extreme Northern Province has a proportion under 20 % (11 % according to the 1996 census for this province). Thus the provinces are at different stages in the process of urban transition, and with considerable differences between regions.

10These towns and cities form an urban fabric across the national territory that is hierarchised by the various relationships they entertain one with another. Towns and cities, from the smallest to the largest, are in constant contact and the dynamics of their evolution are complementary and interdependent.

What definition can be given of the city or the urban object in South Africa?

A restrictive definition until 1991

11Defining what is urban and what is not, and outlining the different spaces correctly are delicate matters involving variations over time. Should towns and cities be defined administratively or statistically? Should urban entities be outlined according to morphological of functional criteria? These different options are part of the general consideration devoted to towns, cities and issues of urbanisation across the world today. And the issue is all the more crucial in South Africa because displacements in the urbanisation process have upset relationships between town and country. Thus very high concentrations of population are found in certain parts of the territory, while at the same time these areas show no organisation into urban poles of attraction, nor any availability of the functions and services that belong to a town or city. In this context, the definition and delimitation of urban spaces is a genuine political challenge.

12The layout of the apartheid city (figures 9 and 10) shows that it is not possible to use morphological agglomeration criteria to define urban areas. Indeed, there are discontinuities and breaks of two types, firstly between the townships and the Central Business Districts (CBDs), materialised by buffer zones, and secondly between these centres and urban fragments that developed some distance from the borders of the former bantustans.

13Neither does the administrative criterion appear as being operational, since the different generations of South African municipalities covered only segments of urban agglomerations in the apartheid era, while in the case of the new municipal system they cover areas extending beyond the urban peripheries of metropolitan areas, or else group several secondary towns. Only the interim structure the Transitional Local Councils (TLCs) remained close to the perimeters of the urban agglomerations and extended to include the townships, the ex-municipalities of the apartheid era made up of CBDs, and white suburbs. However, these transitional structures did not include the urban agglomerations that developed beyond the boundaries of the former bantustans, structurally linked to the cities or towns of which they were projections.

  • 6 The aires urbaines, created by INSEE (Institut national de la statistique et des études économiques (...)
  • 7 The MSAs are made up of a centre (core) of at least 50 000 inhabitants. The peripheral areas or out (...)

14The presentation of the urban object in censuses is in this respect a faithful picture of the fragmented structure of the South African town or city (the Davies model, figure 9). There is indeed no official definition of urbanised spaces that is equivalent, for instance, to the French “aires urbaines”6 or to the Metropolitan Statistical Areas (MSA)7 in the United States (which are functional urban areas). The different definitions presented in the censuses fairly clearly show that throughout the 20th century a legal definition of the city is preferred. This is indeed a satisfactory approach when the issue is to control a territorial entity, but it is nevertheless unstable in the long term, and this is particularly true in South Africa where political action has considerably altered and shaped the territory, whatever the scale envisaged. The definition of the urban object was markedly restricted from 1960, and clearly expresses the establishment of the apartheid system, with the city being above all intended for the white population. It is however instructive to see that, very early on, the statistical departments were not satisfied with the definition, and that they put emphasis on the existence of spaces that could be considered as urban on account of their location and/or their functions, while remaining outside the strict legal definition. This constraint binding statistical departments was overthrown as soon as apartheid came to an end in 1996. The political upheavals of that period generated a new administrative grid, and the definition of the city became a genuine subject of consideration. The question of urban sprawl, and the issue of the reintegration of fragments of territories formerly separate from an administrative viewpoint, gradually came to be envisaged. But only the 1996 census, and above all the 2001 census, offer a definition consistent with reality, a definition that was resolutely functional rather than legal.

Source: R.J. Davis, 1981.
Figure 9 – The model of the apartheid city according to Davies.

Figure 10 – The apartheid city and the succession of municipal systems (Giraut, 2005).

1996: Putting an end to the administrative subdivision of urban territories

15The Transitional Local Councils (TLC), intended as temporary entities established in the transitional post-apartheid period (lasting 5 years), show the beginnings of a change in urban matters in South Africa. In the definition of these entities, the authorities for the first time took account of functional aspects of towns and cities, putting an end, with variable success, to the separations between the white city and the non-white townships or segments of bantustans. The vast majority of the newly-formed TLCs indeed combined a former white municipality and its black township or townships (Giraut & Maharaj, 2003). For instance, the King William's Town TLC (in Eastern Cape province) officially integrated black population zones formerly part of the bantustan Ciskei, such as Zwelitsha or Bisho. Very often, the population of the white city was smaller than that of the townships. However, despite the vast progress achieved, imperfections remained in the TLC delimitation process. In some instances, most often for political reasons, certain townships, despite having links with the city or town, were not integrated into the boundaries of the new entity.

16The question of the reintegration of whole segments of bantustans into the TLC urban entities was complex on account of the very nature of these spaces, generally located relatively far from the white city, between 20 and 60 kilometres away (Giraut & Maharaj, 2003). In fact, certain parts and in some cases the whole of a given independent territory were able to develop a degree of autonomy by establishing their own activities, and even their own administrations. In addition, the border areas of the bantustans often possessed zones of informal population settlement, often rural in character, with low population densities (Graaf, 1986), which did not really warrant their classification as urban areas. Consequently, very few TLCs integrated displaced urbanisation zones (for instance Mabopane and Temba to the north of Pretoria were not part of the Metropolitan Local Council (MLC) (figure 1), forming the Transitional Regional Councils (TRC) that were independent. The period from 2001 saw attempts to correct these shortfalls in the delimitation of urban entities.

  • 8 The dictionary of South African place-names in its most recent edition (Raper, 2004) opens up widel (...)

17Thus in this post-apartheid transition period the issue of the recognition and reintegration of marginalised localities (that were virtually occulted via certain forms of official geography) became acute. It is known that official South African cartography ignored certain places set aside for African populations (Stickler, 1990; HRC, 1992). While for the main townships, recognised as urban localities in the 1980s, their administrative integration occurred without difficulty at the end of the apartheid period, the same is not true for urban areas displaced beyond the boundaries of the former bantustans, where any integration into urban entities was dependent upon their being recognised as urban agglomerations. The reluctance, whether or not it was deliberate, to recognise these very specific places can still be seen in academic publications, otherwise very thorough and valuable.8

The revolution of 2001: adopting a definition suited to reality

18The work conducted by the statistical departments on the measurement of urban objects is certainly one of the most interesting innovations in the 2001 census. To understand the level at which the rural/urban distinction is set, it can be recalled that in the new administrative organisation (figure 1) the 65 536 census enumeration areas are positioned in accordance with the 15 966 subplaces (urban quarters or villages) which in turn belong to 2 674 main places aggregated into metropolitan area municipalities that are six in number, or into two levels outside the metropolitan areas: 231 basic municipalities and 25 district management areas (for the zones of semi-desert, without permanent occupants) headed by 47 district council areas. Finally, nine provinces cover this whole. In these new delimitations, the main drawback is that they are difficult to compare with those that preceded them because they have evolved, even if it is generally possible to aggregate main places according to magisterial district (354 in number), a division used throughout the 20th century for South African censuses.

19The Statistics South Africa departments drew up a report (Investigations into appropriate definitions of urban and rural areas for South Africa: discussion document) which clearly sets out the issue of the measurement of urban objects, proposing a new method for the delineation of the different types of space. Thus peri-urban spaces, suburbs and spaces not hitherto considered as urban, were viewed as being functionally linked to the town or city and defined as urban. The spaces concerned were those previously excluded either by denying their existence (informal settlement zones around cities) or because they were zones belonging to the former bantustans. It should be noted that in this approach to the urban object, it is on the level of the baseline unit, the enumeration area (EA), or census area, that the distinction between rural and urban operates. Thus for the first time we have a morpho-functional definition of the city, far more satisfactory that all previously implemented definitions in South Africa. In addition, the statistical departments endeavoured to classify as urban those spaces that were effectively urban in 1996, but that had been defined as rural from the administrative viewpoint. Thus it is possible to compare the scale of urban growth between 1996 and 2001 using the same delimitations.

Consideration of the South African urban object in other sources

20Secondary sources available and partially used to compare and validate population figures are also very useful for their analysis of the urban object, all the more so because some, such as the Urban Foundation, have broached the question of the South African city and its transformations directly, as a subject of study in itself. The numerous reports produced by the Centre for Development and Enterprise (1996) show the extent to which a new approach to the urban object and its definition constitute a challenge as early as the transition period, before the national statistics departments tackled the problem. Thus the functional aspects of the city were studied and taken into account far earlier than in the censuses.

The choice of a functional definition in the Dysturb database

21Given these different not very satisfactory “official” definitions of the town/city or the urban object, derived from the censuses, a definition was needed to enable the follow-up of urban growth in the long term. The main difficulty was that of recomposing coherent, functional urban units combining the white city and its townships. Despite the difficulty of choosing variable, non-constant ctiteria for delimitation of urban agglomerations, in particular in terms of comparisons of successive urban growth phases, this choice was essential on account of the South African historical context and realities. This is indeed the stance adopted by any historical database on urbanisation, whether those established by historians for Europe (De Vries, 1984; Bairoch et al, 1988), for France (INED urban database, Pumain, 1982), or worldwide (Géopolis, Moriconi-Ebrard, 1994).

  • 9 For derails of the successive delimitations of urban areas in France, see two studies: Guerois M., (...)

22It is essential to escape from administrative boundaries to identify geographical urban entities that are coherent and relevant, and that evolve over time both by the growth of their populations in situ and by spatial extension of the built-up areas. It is because the latter process is almost always one-way in time and space (progress outwards with the problem of the integration of already formed settlements which then change status, as do the inhabitants; no regression of built-up areas) that a “variable geography” outline is the only way to give meaning to the city defined as a geographical entity. In the case of South Africa and the spatial discontinuities that were imposed between the urban employment core (the former centres of the “white cities”) and the workers' residential peripheries (the non-white townships), the concept of the urban agglomeration made up of a continuous built-up area needs to be extended to include the functional urban agglomeration. The issue is, as for the ZPIU (Zones de peuplement industriel et urbain) established in France in 1962, and as for the aires urbaines of 19969, to take into account new forms of urbanisation incorporating commuting patterns, and hence the use of individual transport (cars) and collective transport (taxis, buses and so forth).

23Thus the functional definition of the urban agglomeration implies an urban area made up of a main urban pole (the white city) and its periphery with which it in immediate contact (towns with or without official recognition, black townships, economically linked to the urban pole) the main link being that of the commuting journey from home to workplace. The black taxi lines, an extremely well organised transport system, typify this link. This essential economic relationship between the centre and its peripheries was thus the main justification for the choices made to develop the spatio-temporal database for towns and cities proposed by Dysturb. The continuity of built-up zones is a criterion that was used above all in the case of the largest urban agglomerations (Johannesburg, Cape Town, Durban, Pretoria etc.) to include the white municipalities belonging to the urban agglomeration area, but the built-up area continuity criterion was not used for the townships, since by their very nature they were physically separated from the towns and cities by a wide buffer zones, and sometimes distant from the city (figure 11).

24This gives scope for exploration of the dynamics of the city systems on different scales, and also for working more specifically on the dynamics of the peri-urban spaces which in South Africa entail particular features and constitute a challenge, with the practices of displaced urbanisation inherited from the grand apartheid system. Finally, retrospective data on administrative status enable study of political and administrative urban functions within the urban dynamics at different periods.

Figure 11 – Examples of delimitations of urban agglomerations following a functional definition.

  • 10 For this purpose, one main source was used: The Two South Africas, a people geography (HRC, 1992). (...)

25Establishing the relevant associations between the white cities and the black townships is the only way, particularly from 1970, to obtain a coherent, functional urban area, even if, for the administrations of the time, these inhabited areas were not part of the white city10 (figures 11 and 12).

26The criteria used to form the database for the urban agglomerations are as follows:

  • where possible, continuity of the built-up area between centre and periphery;

  • aggregation of functionally linked spaces: joining of townships and centre (CBD + former white suburbs);

  • minimum population threshold: 5 000 inhabitants.

  • 11 Transkei, Bophuthatswana, Venda, Ciskei.
  • 12 These terrirories were enumerated separately in the censuses, since they were officially independen (...)

27A further difficulty concerns the way in which the bantustans are taken into account (TBVC11, Lebowa, KwaZulu, Gazankulu, Qwaqwa, Kwandebele and Kangwane) for the 1970 to 1991 period. For certain years, there are no data whatsoever (for instance Venda or Ciskei).12 When the data was available, agglomerations of over 5 000 inhabitants were taken into account (on the basis of the threshold chosen). When data for certain dates were lacking, estimates were made on the basis of mean annual growth rates between the known dates. Few towns were in fact concerned, but it was considered to be more accurate to proceed in this way so as not to lose track of any town at a given date.

  • 13 In Géopolis (1994) F. Moriconi-Ebrard made a start on reconstituting the agglomerations of more tha (...)

28From the maps provided, also including road maps and aerial photo-graphs, it was possible to reconstitute urban agglomerations13, although it was a lengthy, difficult task. The minimum hierarchical threshold of 5 000 inhabitants was applied once associations had been established between each former white town and its peripheral units. The boundaries thus formed vary over time and in the different censuses, since new quarters and townships appearing between censuses were integrated into the main towns.

Figure 12 – Example of the association of territories on the scale of an urban agglomeration.

  • 14 Cartographic material in Arcinfo format obtained in 1997 was extremely useful until the issue of th (...)

29The database presented in Dysturb thus plots out the evolution of the population of the urban agglomerations of over 5 000 inhabitants from 1911 to 2001. The task of reconstituting the agglomerations and of checking data from one year to another did not highlight any major inconsistency in the long term. Certain anomalies did however require correction, for instance a township that could not be found at one date while it was present at the previous and later dates. To decide not to take it into account would have led to a very marked decrease in population for the town concerned, which would have been incorrect. The missing data thus led to the estimation of certain figures. In most cases estimates were made on the basis of mean annual growth rates between the two dates either side of the missing year. However, instances in which such estimates were required were very few, amounting to not more than ten cases. Another difficulty encountered was that of homonyms, in particular for the place-names of townships. So as to avoid mistakes in associating these names, and to ensure that the right townships was allocated to the corresponding white town, maps and GIS were used.14

30Between 1911 and 1991, the work of reconstitution was relatively easy. To go further using the data for 1996, the database encountered a problem of correspondence between the former municipalities and the new TLCs. Although the definition of the TLC appeared to correspond to the definition that we had established for functional agglomerations, in all cases it proved necessary to go right down to the place names level so as to find the entities and reconstitute the agglomerations. In all, 60 % of the results did in fact correspond to the figures given for the TLCs, but for the rest there was a discrepancy which was corrected, even if it only represented a small difference in population. The last stage of the constitution of the database was the most delicate, since between 1996 and 2001 not only did the country change its territorial grid with the establishment of the new municipalities, but there was also a more difficult issue to be managed, arising from the new toponymy established for these same municipalities, so that any immediate correspondence between the agglomerations of 1996 and those of 2001 became impossible. To this were added the boundaries to these new administrative entities, which do not cover only the urban areas but also a considerable number of rural areas, creating an extra difficulty in establishing immediate correspondences. The solution adopted was to start from the main places, or even the subplaces, so as to return to the basic entities (towns/cities and townships). These different levels did not undergo changes in name (at least for the most part) so that they were easier to use. However several revisions of the data were conducted so as to complete the year 2001. Finally, at the initial date of 1911 the database has 25 urban agglomerations, in 1951 there are 89 and in 2001 there are 307 urban agglomerations more than 5 000 inhabitants.


1 This refers to the racial classification in force under apartheid, which divided the population into four rigid, sophisticated categories: White, Black (or Bantu) Asian or Indian, and Coloured. Despite the end of apartheid and of segregation policies, these categories are still used today, even if the term “Black” now refers more and more in the new South Africa to all “non-whites” (Gerrvais-Lambony, 1997). See Part 1 of this booklet.

2 This chronology is derived from that established by R.J. Davies in 1972, which we have reshaped and completed. Davies is one of the few South African geographers to have analysed the organisation of the urban network and to have conducted comparative studies on all the towns and cities in the country.

3 The Native Urban Areas Act of 1923 marks the start of urban segregation, well before the official instatement of apartheid. It is crucial in that it underpins all subsequent legislation. The Act signalled the start of residential segregation by giving municipalities the power to create quarters reserved for Africans. In addition, the text established a system of population control (influx control) involving “passes” (these passes carried the name of the employer, and enabled black workers to leave the bantustans, native population reserves, to go to work in town. At the outset, the system was not widely observed. Indeed the periods in which the mining industry developed and industry expanded created considerable needs for labour in the towns and cities, so that influx control lost a large part of its relevance in most towns. The text was later strengthened by an amendment in 1937.

4 The primacy index has been calculated since Jefferson (1939) by measuring the relationship between the population (if the first city and that of the second (P1/P2). F. Moriconi-Fbrard (1993) proposed a generalisation of this notion, distinguishing as metropolises the city or cities situated immediately above the greatest discontinuity threshold in the distribution of city size (Pi/Pj max) and named this the macrocephaly index. It is therefore not the same measure, but in theory if a = 1 in Zipf's law, it can be expected that P1/P2 = 2. In practice, the worldwide mean according to F. Moriconi-Ebrard should be around 4 or 5.

5 The renaming of municipalities affected the names of most of the metropolitan areas, with the exception of Johannesburg and Cape Town.

6 The aires urbaines, created by INSEE (Institut national de la statistique et des études économiques) for the purpose of describing more adequately the peri-urban extension of cities, are made up of an urban “pôle” (centre of attraction) of more than 5 000 inhabitants to which are aggregated in an iterative manner all the communes (municipalities) in which 40 % of the working population commute to the pôle or to a secondary pole attached to it in a previous stage. This definition, like all definitions of functional urban regions, is thus mainly based on outreach (extent of influence) and the intensity of home to work commuting.

7 The MSAs are made up of a centre (core) of at least 50 000 inhabitants. The peripheral areas or outlying counties are aggregated to the centre if they have 15 % of their workforce commuting to the centre and if the density and the growth of the population are sufficiently high.

8 The dictionary of South African place-names in its most recent edition (Raper, 2004) opens up widely to African toponymy, integrating numerous allonyms; it however continues to ignore part of what the HRC termed “the shadow South Africa” (1992) occulted in the official apartheid cartography: distant townships, townships or concentrations on the bantustan, peripheries, and squatter camps. Thus certain major urban areas do not appear, such as Ozizweni, Ezikhaweni, Esikhaweni, Kwa Msane, Wembezi, Ezakheni, Emondlo, all in KwaZuluNatal, or Kayaletu in ex-Ciskei, or again Ga Luka in ex-Bophuthatswana. And if Bushbuckridge is mentioned (a vast concentration of African population in a pocket of ex-Bantustans a few dozen kilometres from the employment poles of Kruger park and the provincial capital Nelspruit), it is always as the denomination of a small mountain range that gives its name to a village, Bosbokrand! As for Inanda (a very vast urban quarter that developed autonomously on the periphery of Durban in continuity with the township of KwaMashu) the locality appears as a mission post.

9 For derails of the successive delimitations of urban areas in France, see two studies: Guerois M., Paulus F., 2002 (
Brunet R., 1997,

10 For this purpose, one main source was used: The Two South Africas, a people geography (HRC, 1992). This was on the initiative of the Humans Rights Commission in 1990. It aimed to draw up new maps of South Africa, mainly maps of “shadow black South Africa,” i.e. of the black townships close to the white cities. This source distinguishes the white cities, defined as being previously reserved for the white population in the Group Areas Act (1950) and the “black towns” or townships, covering the zones reserved for the black, Indian and “coloured” populations. If previously the authorities completely ignored the functional associations of urban areas, in this report this characteristic was central in the definition of the town/city.

11 Transkei, Bophuthatswana, Venda, Ciskei.

12 These terrirories were enumerated separately in the censuses, since they were officially independent.

13 In Géopolis (1994) F. Moriconi-Ebrard made a start on reconstituting the agglomerations of more than 10 000 inhabitants from 1951 to 1991.

14 Cartographic material in Arcinfo format obtained in 1997 was extremely useful until the issue of the official 1996 census GIS, and then the establishment of SA Explorer by the Municipal Demarcation Board, and finally the GIS that accompanied the 2001 census. From one year to another, these systems improved and are today extremely powerful. It can however be noted that the CD-Rom SA Explorer is a completely new GIS system. The work conducted by the Demarcation Board and Michael Sutcliffe have provided a very versatile tool which takes the new 2001 municipal boundaries into account. The data is available on several geographical scales, from the wards (electoral districts) to the provinces, and including the previous TLCs and the new municipalities. A second edition, reviewed and adjusted, was issued in 2002.

Table des illustrations

Légende Source: UMR Géogrophie-Cités.Figure 8 – The establishment of the railways (1860 to 1960)
Fichier image/jpeg, 1,0M
Légende Source: R.J. Davis, 1981.Figure 9 – The model of the apartheid city according to Davies.
Fichier image/jpeg, 1,0M
Légende Figure 10 – The apartheid city and the succession of municipal systems (Giraut, 2005).
Fichier image/jpeg, 1,0M
Légende Figure 11 – Examples of delimitations of urban agglomerations following a functional definition.
Fichier image/jpeg, 1,5M
Légende Figure 12 – Example of the association of territories on the scale of an urban agglomeration.
Fichier image/jpeg, 913k

© IRD Éditions, 2009

Conditions d’utilisation :