Delineating the nation: South African censuses 1865–2007¶
A.J. Christopher (2009). Delineating the nation: South African censuses 1865–2007. Political Geography 28. Elsevier.
| Type | journal article |
| Scope | Wider apartheid literature |
| Method | archival narrative |
| Period studied | 1865–2007 |
| Regions | South Africa, Cape Colony, Natal, Transvaal, Orange Free State, Bantustans (Transkei, Bophuthatswana, Venda, Ciskei), South West Africa |
| Themes | methods data sources, politics ideology elections, demography urbanisation, bantustans removals influx, origins segregation, transition post apartheid |
| Citation key | christopher2009a |
Argument¶
The census delineates the nation in statistical terms, and in South Africa the state and nation were redefined several times, so the roughly forty censuses give a contradictory picture of the population. Imperial and post-1994 censuses were inclusive, while republican, Union (selectively) and apartheid-era censuses restricted who was counted and how results were published. Group inclusion tracked access to the franchise. Unlike elsewhere, there was no continuity between colonial and post-colonial census practice, since white minority rule broke the link until 1994. The argument rests on a chronological reading of census reports and questionnaires, set against the literature on census, identity and state classification.
Evidence and method¶
Published census reports and questionnaires from the Cape, Natal, Boer republics, Union, Republic, bantustans, South West Africa, Southern Rhodesia and Basutoland, 1865–2007. Also colonial and post-apartheid statistical agency documents, a UN Special Rapporteur report, and secondary literature on censuses and nationalism. Summary tables of population by race group are compiled from the census reports.
Place in the debates¶
- Continuity of census practice between colonial and post-colonial states (see Census, race classification and demographic data). Argues South Africa is an exception to the continuity noted elsewhere, because white minority rule from 1910 to 1994 broke the link; it was restored only nominally after 1994.
- Census as neutral measurement vs. state construction of identity categories (see Census, race classification and demographic data). Follows Scott, Anderson and Kertzer & Arel: the census projects the state's national image and creates the categories it claims to describe.
- Success of apartheid's demographic and territorial engineering (bantustan citizenship) (see Census, race classification and demographic data). Census evidence shows exclusion failed: the white share could not be disguised and 93.9% of Africans were returned as South African citizens by 1991.
Findings¶
-
Between 1918 and 1941 several Union censuses counted only the white population, making other groups statistically invisible. (p. 4)
This expedient was to be repeated in 1926, 1931 and 1941, eliminating the census visibility of the other groups
-
Seat apportionment in the Union depended on the white population alone, linking the census to the exclusive franchise. (p. 4)
Notably this process was related to the distribution of neither the total population, nor the number of registered voters, but to the White population alone.
-
From 1921 the census was placed on a racial basis with differentiated questionnaires and reports, abandoning uniform imperial enumeration. (p. 4)
the continuing imperial aim to gather uniform information for the entire population was abandoned in South Africa. The census was placed on a racial basis
-
The 1970 census statistically separated homelands from 'White Areas' and organised the African population by home language into national units, publishing few national totals. (p. 6)
the African rural areas and urban townships, assigned to each of the nine ‘Bantu homelands’, were separated statistically from the rump South African state
-
The 1980 census, by excluding independent bantustans and asking about homeland citizenship, produced a minority of African South African citizens, which the author calls the peak of census manipulation. (p. 6)
As a result only 21.6 per cent of the African population enumerated was returned as being citizens of the Republic of South Africa.
-
By 1991 the attempt to exclude citizens of self-governing states was abandoned and the census showed near-universal African South African citizenship, indicating policy failure. (p. 6)
93.9 per cent of Africans enumerated were returned as South African citizens in 1991, an indication that government policies of exclusion had failed.
-
The white share of the population shrank despite exclusion; the 1991 share was below the 1970 share for the whole of South Africa. (p. 6)
By the latter date the White population only constituted 16.8 per cent of the population, compared with 17.5 per cent for the whole of South Africa in 1970.
-
Comprehensive enumeration of all groups was revived only after 1994; group inclusion tracked franchise access. (p. 8)
There was a close relationship between group access to the franchise, or the possibility of gaining the franchise, and the degree of inclusion in the census.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Census of the Colony of the Cape of Good Hope 1865 [G20-'66] | population census | 1865 | Shows the first modern scientific census, with a comprehensive enumeration and a four-fold race division following imperial guidelines. |
| Cape of Good Hope censuses 1875, 1891 and 1904 (including 1891 report G6-'92 and 1904 report G19-1905) | population census | 1875-1904 | Documents the extension to six race groups and cross-tabulation by race, and the pragmatic construction of race categories. |
| Report on the British Bechuanaland census 1891 | population census | 1891 | Shows the 1891 Bechuanaland census followed Cape and British classification practice. |
| Census of the colony of Natal 1891 and 1904 | population census | 1891, 1904 | Shows that the 1891 Natal census enumerated only Whites and non-indentured Indians, with African and indentured Indian numbers taken from departmental figures, while 1904 followed imperial precedent. |
| Orange Free State census 1880 and 1890 (Verslag van den census 1890) | population census | 1880, 1890 | Shows the republican emphasis on the White citizenry and the count of men qualified for commando, which rose 30% in 1880-1890. |
| South African Republic census 1890 (Verslag van de Volkstelling 1890) | population census | 1890 | Shows a count of Whites only, with data on the franchise and UK-born share of the Witwatersrand White population. |
| De locale wetten der Zuid Afrikaansche Republiek 1849-1885 (Jeppe & Kotze 1887) | legislation gazette | 1849-1885 | Establishes the legal rule that anyone whose great-great grandparents were not White could not be classified as White. |
| Orange River Colony census 1904 and Transvaal Colony and Swaziland census 1904 | population census | 1904 | Shows the post-war censuses enumerated all groups and attributes, with religion but not language. |
| Census of the Union of South Africa 1911, report [UG32-1912] | population census | 1911 | Shows the first Union census, with a full enumeration, simplified three race groups and no language question. |
| Union White-only censuses 1918, 1926, 1931, 1941 (e.g. Census 1918 Part III education [UG1-'20]; 1926 Part VI nationality and parentage [UG35-'29]) | population census | 1918-1941 | Used for the White-only enumerations tied to constituency delimitation, and the language-ability and nationality questions (42% bilingual in 1918, 52% British in 1926). |
| Third census of the Union 1921 (Report [UG37-'24]; Part VIII Non-European races [UG40-'24]) | population census | 1921 | Shows the 1921 racial basis, with separate questionnaires and reports per race group and a new Indian category. |
| Sixth census 1936 (Vol. IV Languages [UG44-'38]; Vol. V Birthplaces, residence and nationality [UG24-'42]) | population census | 1936 | Supplies the bilingualism (64%) and nationality figures (98% South African; Indian 82% South African, 17% British). |
| Population census 1951 (Vol. I [UG42-'55]; Vol. IV [UG34-'54]) | population census | 1951 | Shows that the census formed the basis of the population register under the Population Registration Act of 1950, and the director's statement on Indian classification. |
| Population census 1960, Vol. 11 Part 2, Coloured and Asian families | population census | 1960 | Shows that nationality was omitted from the Indian questionnaire in 1960. |
| Republic of South Africa population censuses 1970, 1980, 1985, 1991 (Reports 02-05-01, 02-80-01, 02-80-12, 02-85-01, 03-01-22, 03-01-26) | population census | 1970-1991 | Shows the Bantu national unit classification, territorial exclusion of independent bantustans, citizenship results (21.6% in 1980, 93.9% in 1991) and the White share of population (17.5% in 1970, 16.8% in 1991). |
| Bantustan censuses: Transkei 1985, Bophuthatswana 1985 and 1991, Venda 1991, Ciskei 1991 | population census | 1980-1991 | Used for the persistence of apartheid classifications and citizenship/titular-language shares (Table 4; 76.0% Bophuthatswana citizens). |
| South West Africa and Namibia censuses 1960, 1970, 1981, 1991 | population census | 1960-1991 | Shows that SWA censuses followed South African lines, and that Namibia dropped the race question after independence. |
| Southern Rhodesia censuses 1921 and 1926; Basutoland 1956 census | population census | 1921-1956 | Shows coordination of neighbouring censuses with South Africa and the removal of the English/Dutch language question after incorporation was rejected. |
| Statistics South Africa census 1996 and 2001 documentation (reports 03-01-17, 03-02-02, 03-02-04, 03-02-43) and 2007 Community Survey questionnaire | population census | 1996-2007 | Shows retention of the four-fold apartheid race classification, the 'others' category and imputation, and nationality handling in post-apartheid censuses. |
| Chinese Association of South Africa v Minister of Labour, High Court (Transvaal Provincial Division) 2008 | legislation gazette | 2008 | Establishes that the Chinese community was legally recognised as 'Coloured' for equity programmes only in 2008. |
| UN Special Rapporteur report, Mission to South Africa (2005) | commission or official report | 2005 | Quoted to show that census categories still follow the apartheid typology and ignore indigenous peoples. |
Inference¶
Quantification: Descriptive tables.
Qualitative comparative-historical reading of successive census reports, questionnaires and director's statements across regimes, supported by summary tables and selected percentages, and framed by theory on census, state and nation.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| South African censuses varied between inclusive enumerations and narrow ones covering only Whites, tracking the political regime and the franchise. (p. 108) | Comparison of Cape, republican, Union and later census reports, summarised in Tables 1-3. | Systematic data |
| Union censuses were tied to the constitutional delimitation of constituencies, which relied on the White population, leading to White-only wartime and later censuses. (p. 104) | Union census reports 1911-1941 and the constitutional provisions as described. | Partial data |
| The 1980 census manipulated citizenship of Africans, so that Whites outnumbered African South African citizens. (p. 106) | 1980 census social characteristics report (Report 02-80-12) with the citizenship question. | Systematic data |
| The attempt to exclude self-governing bantustan citizens failed, as 93.9% of Africans were returned as South African citizens in 1991. (p. 106) | 1991 census social characteristics report (03-01-22). | Systematic data |
| The White nationality shift from British to South African identity was rapid, as measured in the censuses of 1926 and 1936. (p. 105) | 1926 Part VI and 1936 Vol. V census nationality tabulations. | Systematic data |
| The post-apartheid censuses retained the apartheid four-fold race classification for monitoring redress and equity legislation. (p. 107) | Statistics South Africa 1996 and 2001 census documentation, 2007 Community Survey questionnaire, and equity legislation. | Systematic data |
| Unlike other former colonies, South Africa lacked continuity of census practice between colonial and post-colonial states. (p. 108) | Synthesis of the census reports across regimes, compared with secondary literature on other countries. | Partial data |
The author's caveats¶
The author notes that census race categories are unscientific and shifting, that terminology changed over time, that boundaries and enumeration coverage varied between censuses (limiting comparability), that the 1985 and 1991 censuses were affected by unrest and undercount, and that responses on identity may reflect enumerator or state pressure.
Gaps in the evidence¶
The argument reads census questionnaires and published reports but does not use archival files on how directors and ministers decided on questions and categories, or enumerator instructions and fieldwork practice. It has no evidence on how respondents answered or how undercount varied by group, which the undercount adjustment reports and archival census files could address. Those records existed at the time and are now accessible in state archives and Statistics South Africa holdings.
Replicability: Data public and well described.