Parliamentary delimitation in South Africa, 1910-1980¶
A. J. Christopher (1983). Parliamentary delimitation in South Africa, 1910-1980. Political Geography Quarterly, Vol. 2, No. 3. Butterworth & Co (Publishers) Ltd.
| Type | journal article |
| Scope | Wider apartheid literature |
| Method | descriptive quantitative |
| Period studied | 1910–1980 |
| Regions | South Africa, Cape Province, Transvaal, Natal, Orange Free State |
| Themes | politics ideology elections, origins segregation, demography urbanisation |
| Citation key | christopher1983 |
Argument¶
Christopher examines the 14 delimitations of white parliamentary constituencies between 1910 and 1980 in response to the claim that the commissioners made a mockery of democracy. He shows that the statutory rules (15 per cent deviation, 30 per cent for Special Electoral Divisions after 1966) and the 'loading' of urban and 'unloading' of rural seats gave a structural advantage to rurally based, Afrikaans-speaking parties. Provincial apportionment was also distorted, first against the Cape and later against the Transvaal. However, he concludes that this territorial bias was limited, and that the main cause of lopsided election outcomes lay in the language- and race-based nature of South African politics, not in delimitation.
Evidence and method¶
Reports of the 14 Delimitation Commissions (1910-1980), constitutional statutes (South Africa Act 1909, 1961 Constitution, amendment acts of 1965 and 1973), census and voter-roll figures compiled in Table 1, electoral histories (Heard, Schoeman, Lemon), and the 1960 republic referendum results by constituency.
Place in the debates¶
- Whether delimitation commissions rigged South African elections (gerrymandering/malapportionment) (see White electoral politics, delimitation and party systems). Partly rejects Gudgin and Taylor's 'mockery of democracy' charge: bias exists and favours rural/Afrikaner parties, but it is limited and not the main cause of unequal contests.
- Sources of electoral disparity: electoral machinery vs. the racial and linguistic basis of politics (see White electoral politics, delimitation and party systems). Argues language and race, not economic or social cleavage nor delimitation, were the major cause of disparities; the plurality system's suitability for the white electorate is questioned.
Findings¶
-
The franchise became entirely white by 1966 after removals of Black and Coloured voters from the common roll. (p. 2)
In 1936 Black voters were taken off the common roll and in 1956 Coloured voters in the Cape Province were removed, leaving a virtually all-White electorate.
-
Initial provincial apportionment under-represented the Cape and over-represented Natal until 1932. (p. 5)
Until the latter date the Cape was under-represented by from four to seven seats and Natal over-represented by from one to six seats (Table 2).
-
After 1961 malapportionment was reintroduced through Special Electoral Divisions, reversing the earlier pattern. (p. 6)
The result was a reversal of the pre-1958 circumstances with the Cape Province over-represented and the Transvaal underrepresented.
-
Pegging the Assembly size denied the Transvaal seats it was entitled to by registered voters in 1980. (p. 6)
Thus the Transvaal, which at the latter date was entitled, on the basis of registered voters, to 84 seats or over half the membership of the Assembly, was awarded only 76 seats
-
The permitted 15 per cent (30 per cent in special divisions) deviation is the source of marked malapportionment, though modest by international standards. (p. 7)
It is the leeway allowed the Commissioners of a 15 per cent deviation from the provincial quota, and since 1966 of 30 per cent in special circumstances, that has resulted in a marked malapportionment.
-
The share of seats deviating by more than 10 per cent from quota rose sharply over time. (p. 9)
Thus in 1923 only 6.7 per cent of seats differed in size by more than 10 per cent from their respective pro vincial quotas. In 1953 the figure had risen to 46.7 per cent
-
The National Party won in 1948 and 1953 on a minority of votes, aided by loading and unloading. (p. 10)
The National Party was able to win elections in 1948 and 1953 on a minority of votes, and the same is possibly true of the 1958 election
-
In the 1960 republic referendum a narrow vote majority translated into a much larger constituency majority. (p. 10)
The Referendum result was reasonably close, with a 52.1 per cent vote in South Africa in favour of the establishment of a republic (Schoeman, 1977). If the votes are translated into constituency terms then 65.3 per cent of constituencies would have returned republican
-
Homeland 'independence' forced redrawing of constituency boundaries in 1980. (p. 11)
Constituency boundaries were redrawn to coincide with the independent Black state boundaries and the states themselves were excised from the South African electoral map
-
Delimitation contributed to imbalance, but race and language politics were the main cause of disparities. (p. 12)
It may be concluded that the Delimitation Commissions contributed to an imbalance in parliamentary representation, but that the peculiar nature of politics in the country
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Reports of the Delimitation Commissions 1910-1980 (First to Fourteenth Commissions) | commission or official report | 1910-1980 | Supplies the figures in Table 1 (seats by province, voters, average seat size, coefficient of variation) and the Commissioners' stated reasoning on loading, unloading and community of interest. |
| Report of the Ninth Delimitation Commission, Government Gazette No. 3931, 13 February 1948 | commission or official report | 1948 | Quoted to show the 1948 Commissioners rejected a general right to a rural unload and aimed at the quota. |
| Report of the Tenth Delimitation Commission, Government Gazette No. 5009, 20 February 1953 | commission or official report | 1953 | Shows that Commissioners took permitted deviations as injunctions, wanted to unload extensive constituencies further, and characterised Port Elizabeth constituencies. |
| Report of the Twelfth Delimitation Commission, Government Gazette No. 1364, 7 February 1966 | commission or official report | 1966 | Quoted for the mandatory loading of dense and unloading of sparse electoral divisions; also cited on registration activity by party workers. |
| Report of the Fourteenth Delimitation Commission, Government Gazette No. 7246, 3 October 1980 | commission or official report | 1980 | Used for the 1980 redrawing around 'independent' Black states, communications criteria and the survival rate of constituencies. |
| Report of the Third Delimitation Commission, UG 58-1919 | commission or official report | 1919 | Shows how the community of interest criterion linked towns to surrounding farmers and led to dividing Bloemfontein. |
| Report of the Second Delimitation Commission, UG 15-1913 | commission or official report | 1913 | Cited as having examined a 20 per cent deviation which the Assembly was unwilling to adopt. |
| South Africa Act 1909 and Republic of South Africa Constitution Act 1961; Constitution Amendment Act 1965; Constitution and Elections Amendment Act 1973 (Select Committee Report SG 5-1973) | legislation gazette | 1909-1973 | Establishes the legal criteria, quotas, deviation margins, Special Electoral Divisions and the pegging of House size and provincial seats. |
| South Africa Act 1909, TG 14-1910 (Transvaal) and 1904 census | population census | 1904-1911 | The 1904 census adult White male population was the basis of the initial provincial apportionment; later censuses (1911, 1946, 1951) underlie claims about the Cape's loss and Afrikaner urbanisation. |
| Colonial redistribution reports: Cape Comparative statistics 1883 (G.108-1883), Cape Redistribution of Seats Commission 1897-8 (G.8-1898), Natal Redistribution Commission 1907 | commission or official report | 1883-1907 | Shows precedent for delimitation criteria and for rural overrepresentation or unloading in the colonies. |
| Assembly Debates, Vol. 8 (1980) | parliamentary debates | 1980 | Cited for parliamentary rejection of proposals to reduce the range of constituency sizes. |
| Schoeman (1977), Parlementêre verkiesings in Suid-Afrika 1910-1976 (referendum and election returns) | election returns | 1910-1976; 1960 referendum | Source for the 1960 Republic referendum result (52.1% in favour), used to test the votes-to-seats discrepancy. |
| Heard (1974), General Elections in South Africa 1943-1970 | secondary literature compilation | 1943-1970 | Cited for the observation that uncontested 'safe' seats confuse the 1958 election result. |
| Ford (1968), The Urban Trek (census 1946 and 1951 urbanisation data) | secondary literature compilation | 1946-1951 | Supports the claim that Afrikaans-speakers became an urban majority between the 1946 and 1951 censuses. |
New data¶
The author compiled Table 1 of delimitation statistics for 14 delimitations (1910-1980): seats by province, number of voters, average seat size and coefficient of variation. He also derived percentage-deviation distributions of constituency electorates (Figures 1-2) and a 1960 referendum result by 1958 constituency (Figure 3).
Inference¶
Quantification: Descriptive tables.
Descriptive, comparative analysis of delimitation reports across the 14 commissions and across provinces. Constituency size deviations from quota are tabulated and plotted, and the 1960 referendum vote is converted into constituency terms to estimate the seats bias. The causes of the bias are then argued qualitatively.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| The first-stage provincial apportionment disadvantaged the Cape until 1932, with the Cape under-represented by 4-7 seats and Natal over-represented by 1-6 seats, and by up to 12 seats in votes terms in 1923. (p. 5) | Delimitation Commission reports (Table 1 and Table 2) and census-based entitlements, which show the Cape's White male loss between 1904 and 1911. | Systematic data |
| After 1961 parliament reintroduced malapportionment, with the Cape over-represented and the Transvaal under-represented (76 seats against an entitlement of 84 in 1974). (p. 6) | 1974 and 1980 Commission reports, the 1973 Act and the Select Committee report. | Systematic data |
| Commissioners used the permitted deviation as an injunction, so constituency sizes cluster at the extremes. The share of seats deviating over 10 per cent from the provincial quota rose from 6.7% (1923) to 46.7% (1953) and 56.9% (1966). (p. 9) | Commission reports, with deviation distributions in Figures 1-2 and the 1953 report. | Systematic data |
| Loading urban and unloading rural seats gave a persistent advantage to the rural-based National Party, which won the 1948 and 1953 elections on a minority of votes. (p. 10) | Secondary election studies (Lemon, Heard) and the delimitation reports; no systematic vote-seat data for these elections are shown. | Partial data |
| In the 1960 referendum, 52.1% voting for a republic would have translated into 65.3% of constituencies returning republican candidates, indicating the bias. (p. 10) | Schoeman (1977) referendum returns re-aggregated by the 1958 constituencies (Figure 3). | Systematic data |
| Territorial bias from delimitation is real but limited; language and race politics, not the Commissions, are the major cause of the unequal contests. (p. 12) | Comparative reading of the delimitation reports and electoral history; no counterfactual quantification of the effect of bias versus other causes. | Asserted |
| Coloured and Black voters on the common roll were of limited influence, rarely forming more than one-fifth of the electorate of any constituency. (p. 12) | Asserted from the Commission reports without presented tabulation. | Asserted |
The author's caveats¶
The author notes that uncontested 'safe' seats confuse any votes-to-seats inference for 1958 and earlier elections. The 1960 referendum gives the only complete set of national voting figures since 1910. He also notes that the first-stage apportionment figures before 1958 did not match the actual electorate.
Gaps in the evidence¶
The argument lacks constituency-level election returns for all general elections (many seats were uncontested), so it cannot properly estimate seat-vote bias or compare delimitation effects with other causes. Comparable data on language group and urban-rural composition by constituency, e.g. from census and voters' rolls, would allow a direct test of the partisan effect. These sources existed then in the Commission reports, rolls and censuses but are not exploited.
Replicability: Data public and well described.