'To define the indefinable': population classification and the census in South Africa¶
A J Christopher (2002). 'To define the indefinable': population classification and the census in South Africa. Area, 34(4). Royal Geographical Society (with the Institute of British Geographers) / Wiley-Blackwell.
| Type | journal article |
| Scope | Wider apartheid literature |
| Method | archival narrative |
| Period studied | 1865–2001 |
| Regions | South Africa, Cape of Good Hope, Natal, Transvaal, Orange Free State |
| Themes | origins segregation, methods data sources, politics ideology elections, demography urbanisation, bantustans removals influx, transition post apartheid |
| Citation key | christopher2002 |
Argument¶
Population classification in South African censuses, from the colonial Cape through Union and apartheid to the post-1994 era, was a state instrument tied to legal and political rights, including franchise, land ownership and place of residence. The categories shifted in line with the state's changing purposes, moving from complex taxonomic schemes to a simplified colour-based hierarchy that matched segregationist legislation. The Population Registration Act of 1950 made classification vital to enforcing the Group Areas Act. The census did not merely observe the groups it counted: it helped consolidate and 'lock in' identities, and the apartheid categories persist after 1994 through self-identification and affirmative-action requirements. The argument rests on tracing category changes (Table 1) through census reports, statutes and property deeds.
Evidence and method¶
Published census reports for the Cape (1865-1904), Natal, Transvaal, Orange River Colony and Union/Republic (1911-1996), South African statutes (South Africa Act, Population Registration Act 1950, Employment Equity Act 1998), government gazettes, Cape Town Deeds Office title deeds with racial exclusion clauses, and secondary literature on censuses and identity (Barrier, Suzman, Martin).
Place in the debates¶
- Whether censuses merely describe or actively create group identities (see Census, race classification and demographic data). Sides with Barrier: census administrations create and reinforce the identity of the groups they were meant only to observe, and the South African case supports this, especially for the Coloured population.
- Whether race and ethnic classification can be scientifically defined (see Census, race classification and demographic data). Agrees with Suzman and colonial census officials that no scientific basis exists. Race definition is an administrative and political construct that shifted with state purposes.
Findings¶
-
The 1875 census introduced a much more elaborate classification based on physical, cultural and linguistic criteria, retained in 1891 and 1904. (p. 3)
a far more complicated system of six major categories and many sub-categories based on a combination of skin colour, facial features, degree of assimilation, origins and language
-
The quinquennial censuses linked to parliamentary seat allocation counted only Europeans, so several interwar and wartime censuses omitted other groups. (p. 4)
Thus the censuses of 1918, 1926, 1931 and 1941 only enumerated Europeans.
-
The Population Registration Act and Group Areas Act of 1950 made official race classification central to residential control under apartheid. (p. 5)
-
Race was assigned by the Population Registration Board using appearance and social acceptability, including demeaning tests. (p. 5)
demeaning tests for the curliness of hair and skin colour were integral parts of the practical assessment process
-
Classification was part of a system of control, with successive refinements serving franchise, land, housing and forced-removal purposes. (p. 6)
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 | Establishes the first modern census's fourfold classification (European, Hottentot, Kafir, Other) and group shares. |
| Census of the Cape of Good Hope 1875 (G42-'76) | population census | 1875 | Shows the elaborated six-category classification based on skin colour, features, assimilation, origin and language. |
| Census of the Colony of the Cape of Good Hope 1891 (G6-'92) | population census | 1891 | Supplies enumerator instructions and the rationale for listing Fingo and Hottentot separately. |
| Census of the Colony of the Cape of Good Hope 1904 (G19-1905) | population census | 1904 | Source of the Cape census report's criticism of the existing groups as unscientific and of its recommended threefold colour classification. |
| Census of the Colony of Natal 1891 | population census | 1891 | Shows Natal merely estimated the indigenous population and enumerated mixed-race people with Europeans. |
| Census of the Transvaal Colony and Swaziland 1904 (and Orange River Colony 1904 census) | population census | 1904 | Shows the 1904 census enumerated all groups as Europeans, Natives, Asiatics and Mixed, with coordination across colonies. |
| Orange Free State censuses 1880 and 1890 and South African Republic census 1890 | population census | 1880-1890 | Shows the Boer republics counted only Europeans/Coloureds or only Europeans; the sources are described but not cited in the reference list. |
| Census of the Union of South Africa 1911 (UG32-1912) | population census | 1911 | Evidences the simplification to three categories and the acknowledgement that few pure-bred Hottentots remained. |
| Census of the population 1921 (UG15-'23) | population census | 1921 | Evidences the introduction of the fourfold classification and the Asiatic category, linked to repatriation monitoring. |
| Population census 1936 (UG21-1938) and 1946 (UG51-1949) | population census | 1936, 1946 | Used for the category names in Table 1 and the 1946 definition of European tied to the South Africa Act and Supreme Court decisions. |
| Population census 1951 (UG42-1955) | population census | 1951 | Shows the fivefold division with Cape Malay separated, the 62 807 count, and household returns as classification reference point. |
| Population censuses 1960 (RP62-1963), 1970 (Report 02-05-01), 1980 (02-80-13) and 1985 (02-85-01) | population census | 1960-1985 | Shows reversion to fourfold groups, 'Bantu National Units' language tabulations and the shift in naming. |
| Population census 1991 (Report 03-01-06) and 1996 (Report 03-01-19) | population census | 1991, 1996 | Shows the post-1991 shift to home language and that 99.1 per cent identified with apartheid categories in 1996. |
| Census 2001 categories (Table 1) | population census | 2001 | Table 1 lists 2001 categories but no 2001 source is cited in its source line. |
| Cape Government Gazette no. 2174, 29 July 1847, Native Locations in the Eastern Districts | legislation gazette | 1847 | Contrasts the official terms for residential locations with later census terminology. |
| Statutes of the Union of South Africa 1910 and 1911 (South Africa Act) | legislation gazette | 1910-1911 | Provides the 'European descent' definition and the quinquennial census provision. |
| Statutes of the Union of South Africa 1950 (Population Registration Act) | legislation gazette | 1950 | Gives the statutory definition of a white person that became the census definition. |
| Government Gazette no. 19370, 19 October 1998, Employment Equity Act | legislation gazette | 1998 | Shows post-apartheid legal usage of 'black people' as a generic term. |
| Jeppe and Kotze 1887, De locale wetten der Zuid Afrikaansche Republiek 1849–1885 | legislation gazette | 1849-1885 | Source for the South African Republic's one-sixteenth rule for European status. |
| Deeds Office, Cape Town: Deed of Transfer T8135/1924 (Erf 984 Vrede Hoek), CTT4966/1951 (Erf 468 Carnarvon), T16024/1949 (Erf 3043 Paarl) | state archives | 1924-1951 | Illustrates racial exclusion clauses in property deeds and how popular usage converged on the official classification. |
| Malaya 1932, A report on the 1931 census | commission or official report | 1931 | Comparative colonial example of the difficulty of defining race for census purposes. |
| Suzman 1960, Race classification and definition in the legislation of the Union of South Africa 1910–1960 | secondary literature compilation | 1910-1960 | Legal survey drawn on for the lack of any scientific basis of classification. |
Inference¶
Quantification: Illustrative figures.
Descriptive historical case study that traces changes in census classification categories over time (Table 1) against legislation and political context, supplemented by deed examples and theoretical support from Barrier's argument.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| The census classification was repeatedly changed by the state to serve its own changing political purposes, shifting from complex taxonomic schemes to simplified colour-based categories. (p. 402) | Successive census reports 1865-1996, tabulated in Table 1, with the introductory statements and instructions quoted from the reports. | Systematic data |
| Simplified Union-era classification reflected the racial hierarchy in discriminatory legislation and the drive for national uniformity. (p. 404) | 1904 Cape census report, 1911 census report and South Africa Act, read against legislation. | Partial data |
| Formal census categories diverged from popular usage, e.g. in title deeds, which gradually converged with the official classification. (p. 404) | Three Deeds Office documents (1924, 1949, 1951) quoted as examples. | Illustrative or anecdotal |
| The Population Registration Act and the 1951 census tied classification to Group Areas, and the Cape Malay separation was meant to test the need for Malay group areas. (p. 405) | Population Registration Act text, 1951 census report (62 807 Cape Malay) and group area proclamations. | Partial data |
| Census reclassification served administrative ends, such as the 1970 transfer of Hottentots to Coloured, reducing the African count in the western Cape. (p. 405) | 1970 census and the redefinition of Natives as Bantu, with the author's inference about the effect. | Asserted |
| The 1996 census retained apartheid-era categories and nearly all respondents identified within them. (p. 406) | 1996 census report (Report 03-01-19), 99.1 per cent identifying; note 7 compares non-response rates. | Systematic data |
| Census administration reinforces and 'locks in' the groups it was meant only to observe, supporting Barrier's contention. (p. 406) | Theoretical argument from Barrier (1981) and Martin (1998) applied to the South African census sequence. | Asserted |
The author's caveats¶
The author notes that the classification changes and the 99.5 per cent placement imply small differences between censuses, and that omission rates in 1996 (0.92 per cent not stating race) are low. He also notes the 1911 census refinements for Hottentots and Indians were not reported because the results were unreliable.
Gaps in the evidence¶
The argument that the census shaped group identity relies on assertion and secondary theory rather than evidence on how enumerated people responded or how classification affected self-identification. Internal census files, enumerator instructions and Population Registration Board records, plus survey data on identity, would test the 'locking in' claim better; many exist in state archives. Quantitative comparisons of category sizes across censuses, including reclassification effects, could be built from the published tables but are not presented.
Replicability: Data public and well described.