White electoral politics, delimitation and party systems¶
Did delimitation rig white elections, what explains white voting shifts, and how should South African party systems be classified?
Works engaging the debate: 20
Positions¶
- Gerrymandering made a 'mockery of democracy' (Gudgin and Taylor) vs limited bias
- Organisational vs ideological causes of the 1948 result
- Racial census voting and one-party dominance after 1994
Evidence base¶
The delimitation question rests on one core series: the reports of the fourteen Delimitation Commissions, 1910-1980. Several are themselves in the corpus as primary documents (Laurence et al. (1913)) (Delimitation Commission (1932)) (Lansdown et al. (1943)) (Murray et al. (1948)) (Brink et al. (1953)) (Potgieter et al. (1966)) (Cillié et al. (1980)). Union of South Africa, Government Printer (1958) is only a gazette issue carrying a clerical correction to the Eleventh Commission's boundary descriptions. The reports share a common unit of observation: registered (white, plus Cape and Natal non-white) voters per electoral division, measured against a quota. The quotas give a sense of scale. In 1913 it was 2,891, from 349,837 European adult males in the 1904 census divided by 121 seats. In 1952, 1,539,802 voters were spread over 150 seats. In 1965 the quota was 11,503 (1,840,530 voters over 160 seats). By 1980 provincial quotas diverged sharply, from 12,626 in the Cape to 15,433 in the Transvaal. Provincial seat entitlement came from census counts of European adult males or adult Union nationals (1904, 1911, 1931, 1941, 1946). Each report tabulates per-division loads and unloads within the ±15 per cent band, and from 1966 wider bands for very large divisions. They supplement this with Surveyors-General boundary descriptions, party submissions and the commissioners' own inspection tours. Christopher (1983a) and Christopher (1983b) are the only works that read the whole series comparatively. They tabulate provincial over- and under-representation and deviation distributions. They also re-aggregate the 1960 referendum returns from Schoeman (1977) by 1958 constituency, showing that a 52.1 per cent republican vote became 65.3 per cent of seats. That is the one quantified estimate of seat bias in the debate. Price (1953) aggregates the official 1953 constituency returns by province, finding a 45 per cent government vote, a 29-seat majority and 87.8 per cent turnout. He sets these against the 1953 quotas. White (1990) adds United Party archives (branch counts showing 674 of 1,696 moribund in 1947, finance minutes, Smuts and Hofmeyr papers), roll data and interviews to the 1948 problem.
Contested evidence¶
The 'mockery of democracy' charge and the 'limited bias' reply draw on the same commission reports. They disagree over what the reports measure. Christopher (1983a) (Christopher (1983b)) accept that loading urban and unloading rural seats was systematic and that commissioners treated the permitted deviation as an injunction. Yet the conclusion that this bias was smaller than the effect of language and race is asserted, with no counterfactual using equal-sized constituencies. The Commissions themselves offer two readings. Brink et al. (1953) claims that loading was not simply urban versus rural, and that several rural seats were loaded more often than not. Lansdown et al. (1943) and Murray et al. (1948) tabulate deliberate rural unloading of 11-13 per cent against urban loading. On 1948, (Price (1953)) attributes the result to a platteland resurgence magnified by delimitation, and estimates Afrikaans United Party support at about thirty per cent without voter-level data. White (1990) blames organisation, under-registration and HNP deletions, and recalculates the Afrikaans UP-LP vote at 39.5 per cent against earlier estimates of 18-25 per cent. That figure rests on his unpublished thesis, so the ideological and organisational accounts depend on mutually unverifiable language estimates. The Coloured roll is a further fault line. Christopher asserts that Coloured voters rarely exceeded a fifth of any electorate. Lansdown et al. (1943) and Murray et al. (1948), by contrast, show that 43,093 Cape Coloured voters inflated Cape quotas, and White (1990) cites counterfactual Cape seat shifts from their removal. For the late apartheid shifts, (Lemon (1982)), (Sisk (1989)) and Guelke (1988) read swings to the HNP and CP as economic and class-driven. Their evidence is press reports, Official Yearbook aggregates (Afrikaner NP vote falling from 85 to 63 per cent) and by-election snippets. Peele and Morse (1974) instead uses a 462-respondent Cape Town survey to find ethnicity dominant and class effects spurious. Engel (1999) warns that the Official Yearbook series contains errors that cannot be reconciled with Heard or Noller, which undercuts vote-share comparisons across the board. After 1994, (Fox and Lemon (2000)) reads voting-district results (14,535 districts) against the 1996 census as a racial census with one-party dominance. Botha (1996) uses Reynolds's estimates to argue parties were becoming less racially structured, citing the NP's 66.7 per cent of the Coloured vote. Both depend on ecological inference without a formal method, and Botha's dominance-versus-hegemony classification uses Sartori categories without computed indices.
Underused and missing sources¶
No work estimates seats-votes bias systematically across all general elections. Constituency-level returns for contested seats, combined with the commissions' division voter tables, would allow a counterfactual equal-quota simulation. Matching constituencies to district census language and urban-rural composition would test the language-versus-delimitation and organisation-versus-ideology arguments ecologically. For the 1970s-80s, contemporary opinion polls, the 1985 census and constituency returns matched to occupation are noted as unused by Sisk (1989) and Lemon (1982).
Strength of key claims¶
| Basis | Share of key claims |
|---|---|
| Systematic data | 36% |
| Partial data | 38% |
| Illustrative or anecdotal | 8% |
| Asserted | 18% |
Comparative figures for all debates are on the Literature overview.
Best-evidenced works¶
Christopher (1983a), Price (1953), Peele and Morse (1974), Fox and Lemon (2000), White (1990), Engel (1999)
Most-used primary sources¶
Holdings are matched by machine against the archive's file manifest; a dash means no match was found, not that the project certainly lacks the source.
Works and their positions¶
| Work | Position |
|---|---|
| Anglin (1995) | Describes both camps and finds the mandate left real ambiguity. Commonwealth (COMSA) observers were more proactive and the UN more restrictive, with wide variation between individuals. |
| Anglin (1995) | Sceptical. Deterrence worked at specific events, but the late fall in violence and the parties' acceptance of the result owed little to observers. Their main contribution was peace monitoring in 1992-93. |
| Anglin (1995) | Shows the ANC and churches pressed for hundreds of observers while UN budget constraints kept numbers small. He treats the budget dispute, not Pretoria, as the main constraint by late 1993. |
| Botha (1996) | Holds that Sartori's counting rules and Rae's fractionalization index were unsuitable before 1994 because legislation barred organizations and voters, so the system needs to be analysed as a set of legally constructed subsystems. |
| Botha (1996) | Classes the post-1994 system as predominant-party, with warning signs that the ANC could become hegemonic in Sartori's sense. |
| Botha (1996) | Argues that apparent swings to the right were often voter fall-out caused by the NP's own ideological shift to the left, rather than a rightward movement of voters. |
| Brink et al. (1953) | The Commission denies that previous commissions applied a fixed rural-unload/urban-load rule and rejects the Torch Commando's contention that the rule had eroded the value of the urban vote. It holds that each seat is weighed on its merits against the five section 40(3) factors, while acknowledging that in practice rural seats are usually unloaded and urban seats usually loaded. |
| Brink et al. (1953) | Rejects it: only the lists current at the time of delimitation count, and growth is not one of the statutory factors. |
| Christopher (1983a) | 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. |
| Christopher (1983a) | 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. |
| Christopher (1983b) | Partly agrees: the rural unloading and urban loading gave a persistent bias to a rural-based party, but he argues the bias was limited and the commissions were relatively unimportant compared with language- and race-based politics. |
| Cillié et al. (1980) | Does not argue the point. It records the legal basis (a 15% departure from the quota, or 70% for divisions of 25 000 km² or more) and applies it, noting that high rural unloads allow only average loads in cities. |
| Cillié et al. (1980) | Treats it as an administrative problem for boundary-drawing. It does not question the policy. |
| Delimitation Commission (1932) | The Commission applied the statutory rule on sparsity and density of population, which tilted seats toward sparsely populated rural areas and made urban constituencies larger than quota. It nonetheless gave most new seats to urban areas. It does not discuss the wider political implications. |
| Engel (1999) | The author says documentation is weaker than expected and the official list sometimes contains obviously wrong figures, so competing sources (Heard, Noller) are cross-checked but cannot always be reconciled. |
| Fox and Lemon (2000) | Supports the racial-census view for 1999: the black/non-black divide is clearer than in 1994. Rejects reading regional-party support as primarily ethnic, though it concedes race does not fully determine identification. |
| Fox and Lemon (2000) | Describes a one-party dominant democracy that has dangers (weak, racially bounded, fragmented opposition) but may also bring consistent reconstruction policies. |
| Fox and Lemon (2000) | The homeland and CLPA boundaries are still visible in voting, but this reflects race and the urban/rural divide more than ethnicity. |
| Guelke (1988) | Guelke rejects the Schlemmer-style reading of the result as a triumph for the centre and no setback to reform. He treats the Conservative Party's success and the far right's consolidation as a significant blow to the prospects for peaceful change. |
| Guelke (1988) | He judges the state of psephology to be 'relatively primitive'. He criticises the book for asserting opinion without statistical analysis, and for ignoring abstention by white radicals. |
| Lansdown et al. (1943) | Applies the constitutional 15 per cent tolerance, and defends heavier urban loading and lighter rural unloading on grounds of community of interest, sparsity and geography. Treats it as a technical necessity rather than a political choice. |
| Laurence et al. (1913) | The Commission notes that two quota methods leave the Cape with 51 seats despite a larger electorate than a Transvaal quota would justify. It proposes flexibility, but only within its statutory mandate. |
| Laurence et al. (1913) | It says that party-political effects such as South African Party versus Labour were not considerations, and that it confined itself to the statutory criteria. |
| Lemon (1982) | Lemon treats the HNP surge as evidence of a real white backlash against labour and 'petty' apartheid reforms, but credits Botha with holding most of the NP together by facing both ways. He sees the electorate as divided rather than as a monolithic Afrikaner bloc. |
| Murray et al. (1948) | The Commission rejects the claim that rural seats are 'as of right' entitled to a substantial unload up to the maximum at the expense of urban seats. Loading and unloading is a discretionary tool, but in practice it keeps a systematic rural unload and urban load. |
| Murray et al. (1948) | It notes that Coloured voters, though excluded from the apportionment of the 150 seats, count in the Cape's per-division quota, raising that quota. It declines to recommend legislative change as outside its terms of reference. |
| Peele and Morse (1974) | Partly applies: party labels are symbolic and sociologically conditioned, as in the US, but ethnicity is a single dominant predictor rather than an additive class-religion-residence index. |
| Peele and Morse (1974) | Little support for cross-pressures; Barber's thesis holds only for 'ethnic mobility' (ambiguous Afrikaners), not socioeconomic mobility. |
| Peele and Morse (1974) | Disagrees with Lever: Nat-preferring abstainers in Gardens were mainly English and more liberal than Nat voters, so they were not drawn to the HNP; their abstention reflects disillusionment or reversion to the UP. |
| Peele and Morse (1974) | Electoral change within the existing party system is unlikely; change must come from the Afrikaner Nationalist majority or non-partisan labour and business organisations. |
| Potgieter et al. (1966) | The Commission rejects any fixed formula. It follows previous commissions in loading dense urban areas and unloading sparse ones, and it applies the new amendment so that rapidly developing areas carry fewer voters. |
| Price (1953) | Rejects the rumour that ex-servicemen and cost-of-living discontent decided it. Attributes it to a platteland resurgence plus peri-urban seats filled by migrants from the platteland. |
| Price (1953) | Argues that the delimitation system, which loads town seats and unloads country seats, gave the Nationalists a seat majority from a minority of votes. He regards the claim to a mandate as unsupported. |
| Sisk (1989) | Reports Laurie and Kleynhans, who think the CP could win up to 50% of the vote, and Frankel's doubts about its narrow base; Sisk judges that the CP could at least end the NP's absolute majority. |
| Union of South Africa, Government Printer (1958) | Takes no position; it supplies raw administrative and fiscal data (revenue, bank balance sheets, trade figures) only. |
| White (1990) | Stresses organisational failure (registration, deletion, delimitation, party structure) over ideology. Argues that earlier work underestimated UP support among Afrikaans-speakers because of under-registration and roll deletions. |
| White (1990) | Shows mining and financial interests funding the UP, first ad hoc and then through the United South Africa Trust Fund. H.F. Oppenheimer's priority was winning elections, with liberalising race policy second, and the funding exposed the party to 'Hoggenheimer' attacks. |