The job colour bar, white labour and its erosion¶
Who created and defended the colour bar (mine owners, white unions or the state), and what eroded job reservation from the 1960s?
Works engaging the debate: 42
Positions¶
- White labour drove colour bars (liberal: Hutt, Horwitz; Lipton)
- Capital structured racial division (Johnstone, Davies)
- Erosion driven by white skill upgrading and economic incentives (Mariotti) vs skills shortages, pressure and Wiehahn
- 'Floating the colour bar' flexibility
Evidence base¶
The debate rests on two largely separate evidentiary traditions. The first is archival and documentary, and it dominates the question of who created the colour bar. Johnstone (1971) builds his 'wage colour bar' versus 'job colour bar' argument on Gold Producers' Committee statements, Chamber of Mines annual reports, the Low Grade Mines Commission (UG 34/1920), the Mining Industry Board (1922), the Economic and Wage Commission (UG 14/1926) and the 1918 Status Quo Agreement. He uses these selectively, as illustrative figures on working costs per ton and African wage indices. The union-centred literature draws on organisational archives held at the William Cullen Library: TUCSA, Ironmoulders' Society, AEU and Garment Workers' Union minutes. Lewis (1984b) adds Board of Trade and Industries reports 94, 282 and 286, Wage Board and Industrial Tribunal reports, Department of Labour ARB files and Union Statistics for Fifty Years. Coupe (1990) and Coupe (1995) extend this material into the 1950s and 1960s with press, employer journals (Industry and Trade, The Manufacturer) and interviews. Crankshaw (1990) adds the 1942-50 Censuses of Industrial Establishments and regional Industrial Tribunal reports on reservation of work in building. Studies of the state and state corporations (Clark (1987), (Duncan (1990a)), (Duncan (1990b))) use SABA departmental files (MNW, ARB, ECB, NTS, GNLB) and commission evidence. The Marxist account of erosion in Davies (1979a) and Davies (1979b) instead rests on Financial Mail and Rand Daily Mail reporting, April 1977 official vacancy figures, ILO occupational-structure comparisons and Economic Development Programme projections. Legassick (1974) reads Hansard, the Van Eck and Fagan reports and BTI reports. The source readers (Houghton and Dagut (n.d.)) and Houghton and Dagut (1973) reproduce extracts of many of these commissions.
The second tradition is quantitative and mostly post-1980. Crankshaw (1996) reclassifies the biennial Manpower Survey data tapes for 1965-1990 (about 600 occupations by race) to trace race shares in professional, routine white-collar and artisan work. Mariotti (2008), (Mariotti (2009a)) and Mariotti (2009b) use three sources together: Manpower Surveys for 1969-85 (271 occupations, ranked on the Ganzeboom SEO index), 1970 and 1980 population census microdata on white and African education and occupation by language and cohort, and the manufacturing census for 1950-85. Mariotti (2010) turns the same census into a 3-digit industry-by-year panel to estimate a CES elasticity of substitution between African and white production workers. She also compiles the 28 job reservation determinations from the Government Gazette and SAIRR. Moll (1991a) decomposes white-coloured earnings gaps using the 10 per cent 1970 and 5 per cent 1980 census public use samples. Lucas (1985) estimates mine labour demand by sector from unnamed annual series for 1946-80.
Contested evidence¶
On mining before 1926, the sides share the same commission material but read the Chamber's record in opposite ways. Johnstone (1971) treats the 1904 Ordinance, the 1911 Regulations and the Status Quo Agreement as evidence that owners accommodated the job colour bar while defending a wage colour bar. Crush, Jeeves and Yudelman, as reported in Roth (1995a) and Roth (1995b), read Chamber behaviour from 1906-7 as attempts to modify the bar. On their account the Mine Workers Union and the Council of Mining Unions were its principal defenders from 1945 to 1987. Neither review identifies the underlying records, so the dispute cannot be adjudicated from the evidence ledgers. Pogue (2006) only asserts that managers tolerated the 1926 re-imposition. Turrell (1986) relocates the statutory origin to the 1885 Kimberley blasting rules, also without citation.
On complementarity, the two sides rest on incompatible evidence. Lucas (1985) finds white and nonwhite mine labour to be complements, with white labour quasi-fixed, from sector time series. The Porter-Lundahl model (Lundahl (1982b)) assumes complementarity without data. Mariotti (2010) instead finds manufacturing production workers to be strong substitutes (elasticity roughly 7-14), with no clear decline. Her estimates depend on production wages derived from census wage bills and on a marginal-product assumption that sits uneasily with administered industrial council wages. The teaching deck (Labor Markets in South Africa... (n.d.)) restates a complements story that her own estimates do not support.
On erosion, the Manpower Surveys are read differently. Mariotti (2008) uses the 1969-85 Manpower Surveys and the two census cross-sections to date semi-skilled desegregation to the late 1960s. She then excludes unrest, sanctions and capital lobbying on timing grounds alone, without direct evidence on negotiations. Crankshaw (1996) uses the same surveys to stress a 'floating' colour bar in which African artisans rose from 2 to 19 per cent only after Wiehahn. He attributes this to a union-employer-state compromise that he asserts rather than demonstrates. Davies (1979a) infers a capital imperative after the mid-1970s from vacancy figures and business statements, but supplies no profit-rate series. Coupe (1995), working from Horrell's figures (white operatives falling from 24.5 per cent in 1959), reads job reservation as already failing in engineering. That reading fits both camps.
The small formal reach of job reservation (28 determinations, about 2 per cent of workers) is accepted by Mariotti (2010) and Legassick (1974). It is read either as evidence of flexibility or as a threat that disciplined wider compliance (Labor Markets in South Africa... (n.d.)). Capital's attitude is argued from thin surveys: Webster's 1980 foundry poll and Bernstein's 45 per cent figure in Freund (1987a) and Freund (1987b), against Lipton's opinion data.
Underused and missing sources¶
The sources that would most directly settle who drove changes in reservation are almost never analysed systematically. These are the individual industrial council agreements, Industrial Tribunal and section 77 determination files, Department of Labour files, and union and employer-association minutes. Mariotti (2008) and Mariotti (2009a) explicitly defer coding them. Hansard is cited by ten works but only quoted, never coded for who lobbied. Chamber of Mines and company records on wage bills, complements and costs by mine, which would test (Johnstone (1971)), (Lucas (1985)) and the Crush et al. thesis, are absent.
African wages in homelands and on the periphery remain poorly measured in official sources (Kerby (2014)).
Strength of key claims¶
| Basis | Share of key claims |
|---|---|
| Systematic data | 11% |
| Partial data | 53% |
| Illustrative or anecdotal | 19% |
| Asserted | 17% |
Comparative figures for all debates are on the Literature overview.
Best-evidenced works¶
Mariotti (2010), Mariotti (2008), Crankshaw (1996), Lewis (1984b), Lucas (1985), Moll (1991a)
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 |
|---|---|
| Budeli et al. (2008) | Attributes the colour bar and the dual system to white workers' fear of cheaper black competition and to employer and state use of that division, culminating in the 1922 Rand Rebellion and the 1924 ICA. |
| Christopher (1994b) | Quotes the Minister of Labour, Schoeman (1954), as showing that job reservation was tied openly to preserving white rule rather than to economic efficiency. |
| Clark (1987) | Argues the corporations paid scant lip-service to civilised labour and replaced whites with cheap black workers. This corrects Kaplan and Morris and Christie, who she says miss the implications of the shift for the 'national capital' argument or still see Escom as promoting white labour. |
| Coupe (1990) | Rejects a simple 'divide and rule' reading of racial segregation in unions and workplaces. Argues that political and ideological traditions shaped the form of worker alliances, and that state-capital-labour relations need empirical investigation. |
| Coupe (1995) | Rejects both the free-market doctrine (Hutt) and the structuralist 'fractions of capital' approach (Davies). He follows the Simons in stressing the independent agency of organised white labour in demanding colour bars, and calls the fractions-of-capital theory one that has not stood the test of time. |
| Coupe (1995) | Accepts Webster's deskilling argument but contends that Nationalist government interventions have been underrated as a source of the decline of apprenticeship and of deteriorating race relations. |
| Coupe (1995) | Rejects Davies's claim that white workers were part of a 'new petit bourgeoisie'. |
| Crankshaw (1990) | Disputes this reading. The Act, though it looks racist and protective, bypassed craft union control of training and wages and served capital and state housing needs. |
| Crankshaw (1990) | Shows white craft unions were undercut by cheaper African and 'Coloured' labour and became dependent on the state. The racial division of labour was contested, not simply imposed by white labour or by capital. |
| Crankshaw (1996) | Finds substantial but uneven African advancement, which partly vindicates the liberal view that the colour bar eroded. Still endorses the revisionist 'floating colour bar' idea: whites moved up ahead of Africans, and the racial hierarchy was fractured rather than dismantled. |
| Crankshaw (1996) | Attributes the wide divergence in estimates (e.g. 121,948 vs 1,315,800 for 1970) to flawed class theory and coarse official occupational categories. Offers a finer reclassification that yields far lower figures than Wolpe's. |
| Crankshaw (1996) | The apartheid state expanded African semi-professional jobs by staffing segregated services. White unions, with state and employer compromise, restricted African entry to the skilled trades until the 1980s. |
| Davies (1979a) | The shortage of skilled labour was real from the mid-1960s, but floating the colour bar and foreign capital made it tolerable until the mid-1970s. |
| Davies (1979a) | Skeptical. He sees it as restructuring in the interests of capital, with new controls over African unions and a concealed colour bar, not as real emancipation. |
| Davies (1979a) | He rejects it as a myth: even on optimistic assumptions only a small minority of Africans would benefit. |
| Davies (1979b) | Both are needed. Skill shortages alone were insufficient, since they were mitigated during the boom. The reform became imperative only with the combined economic crisis and the political crisis after the 1976 uprisings. |
| Davies (1979b) | Most white wage earners are 'new petty bourgeois' and politically bound to the bourgeoisie. Economic conflict over job reservation is therefore unlikely to change their political polarization, although it may strain political alliances. |
| Davies (1979b) | Rejected as a myth. Even optimistic projections give fewer than one in six economically active Africans any gain. |
| Duncan (1990a) | Denies a Labour Department conspiracy to split the working class. Racial segregation under the 1941 Act was a concession to Afrikaner Nationalist pressure, not a settled policy of dividing workers. |
| Duncan (1990b) | Denies a conspiracy by the Labour Department. Section 51 segregation in the 1941 Factories Act is attributed to Afrikaner Nationalist pressure and prevailing social values, while the department preferred non-discrimination and incorporation of Africans into the industrial conciliation framework. |
| Freund (1987a) | Agrees with Lipton that white workers had a material interest in racial boundaries and acted autonomously. He rejects her view that capital was indifferent to the colour of its workers. |
| Freund (1987b) | Agrees with Lipton that white workers had an autonomous material interest in racial boundaries. He cites Greenberg and Webster, and says she ignores left scholars who explain this. |
| Houghton and Dagut (1973) | Represents both: Low Grade Mines and Mills-Clay-Martin oppose a legal colour bar; Andrews-Lucas-Rood and the Carnegie Commission accept temporary white reservation; editors note civilized labour was framed by standard of living. |
| Houghton and Dagut (n.d.) | Reprints the Mining Industry Commission's statement of both opposed policies and the 1913 Select Committee evidence, so both sides are represented; the editors take no position. |
| Indicus (1909) | Argues the real barrier to white manual labour is colour, not wage levels, and that restrictions on coloured labour drive up the cost of living and block development. |
| Johnstone (1971) | Rejects the liberal thesis (Hutt, Doxey, Horwitz) that mine owners were rational modernisers attacking an irrational white-labour colour bar. Holds that the owners depended on the wage colour bar and only contested the degree of protection white workers had won. |
| Kerby (2014) | Builds on Mariotti (2012), who finds economic incentives rather than changed white preferences. Extends the geography to the periphery and suggests some industries relocated to the homelands rather than lobbying to reform job-reservation laws. |
| Legassick (1974) | Colour bars did not restrain growth. They were limited, flexible tools that mattered mainly when growth slowed. |
| Lewis (1984b) | Rejects the liberal view (Horrell, van der Horst) that white unions' racial prejudice caused colour bars and union division, calling racism an 'ahistorical' explanation. Builds on Johnstone's class analysis of mining but argues that secondary industry had a different division of labour from the mines. |
| Lewis (1984b) | Argues the white working class was not 'neutered' in 1924. Incorporation was uneven and lengthy, and was completed only with the break-up of the TLC and the formation of SATUC; craft and industrial unions kept militant, non-racial practice for two decades. |
| Lewis (1984b) | Adopts Braverman and Hyman's emphasis on labour process but adds historical change, in which closed craft unions become 'pseudo-craft' unions. Critiques Greenberg for neglecting change over time. |
| Lewis (1990) | Following Johnstone and his own labor-process work, he treats racial division as neither inevitable nor unchanging. Employers and the state helped structure it, and white workers in state-run industries or mines most readily adopted exclusion. |
| Lucas (1985) | Finds that eroding the bar in diamonds reduced white labour productivity and demand, as the Mine Workers' Union feared. In the other sectors black and white labour were complements. |
| Lundahl (1982b) | Finds that job reservation and the colour bar divide whites by sector and skill, with agriculture and unskilled whites gaining from some measures and industry and skilled whites losing. |
| Mariotti (2008) | Rejects each as the main cause. Labour demand is necessary but insufficient. She argues the decisive factor was rising white, especially Afrikaner, education and occupational status, which shifted political power within industries. |
| Mariotti (2008) | Sides with those who stress worker-enforced segregation: discrimination persists while the politically dominant group benefits from it, and ends when its interests change. |
| Mariotti (2009a) | Argues that white workers' economic incentives were the primary cause. It rejects explanations based on political pressure, capitalist pressure or state intervention, and rejects changes in African characteristics. |
| Mariotti (2009a) | Sides partly with Heckman: white workers did act collectively, but their preference for higher incomes overrode their preference for segregation, even with explicit state support for segregation. |
| Mariotti (2009b) | Rejects the conventional wisdom that international and domestic political pressure from the mid-1970s ended apartheid. Argues that labour-market desegregation began earlier, at the height of apartheid, for different reasons. |
| Mariotti (2009b) | Supports Heckman's point that collective worker action can sustain segregation without state backing. Adds that white workers' preference for higher incomes eventually outweighed their preference for segregation, even with explicit government support. |
| Mariotti (2010) | Disputes it for manufacturing: the estimates show continued high substitutability, with no evidence of a decline for production workers. |
| Mariotti (2010) | Sees it as flexible: unions and employers 'floated' the colour bar by recoding and downgrading occupations, so Africans crept up the job ladder, consistent with Lipton and Feinstein. |
| Mariotti and Fourie (2014) | Reporting Lundahl, they say that white unskilled labour benefited most and capitalists were constrained, contrary to the Marxist claim. |
| McGrath et al. (2004a) | Cites Terreblanche & Nattrass and Beinart, who hold that the Second World War made the colour bar economically redundant. It stresses skill as bound up with racial ideology, white labour protection and social control, with employers floating the colour bar upwards from the 1960s. |
| Mohamed (1991) | Supports Lewis: SAEWA's evidence fits the deskilling thesis, with the qualification that racial protection must also be situated in a racially segregated society and cannot be reduced to changes in the labour process. |
| Mohamed (1991) | Argues that skilled workers, though historically conservative, were drawn towards unskilled workers by deskilling and falling living standards, contrary to expectations. |
| Moll (1991a) | He attributes the erosion to rising demand for skilled labour, multinational and political pressure, the Wiehahn reforms and the union wave. He does not test any specific theory of discrimination. |
| O'Meara (1978) | Sidesteps the conceptual status of white workers but implicitly distinguishes productive workers from supervisory 'new petty bourgeoisie' (e.g. white miners), which shapes the contrasting outcomes in the MWU and GWU. |
| O'Meara (1978) | Argues that white wage gains after 1948 were bought by raising the exploitation of black workers, with capital and the NP state cooperating in incorporating white labour. |
| Pogue (2006) | Argues technological change in stoping (Taylorism, drill director, lightweight drills) rather than only capital-labour conflict shaped the mines' stance; mines tolerated de jure restrictions because efficiency gains were secured and removing them was politically costly. |
| Roth (1995a) | The book argues against recent radical scholarship that mine managers supported the Color Bar, and attributes defence of white privilege from 1945 to 1987 mainly to the Mine Workers Union and the Council of Mining Unions. |
| Roth (1995b) | As reported by Roth, the book argues against recent radical scholarship, showing that the Chamber of Mines tried (largely unsuccessfully) to modify the bar from 1906-7 and that the MWU and its allies were the main defenders from 1945 to 1987. |
| Spandau (1972) | Argues tariff conditions, the Fair Wage Clause and the Apprenticeship Act mattered more than high minimum wage determinations. |
| Turrell (1986) | Turrell rejects Jeeves's middle path that the colour bar and migrancy 'evolved together'. He sides with Davies: entrenched migrancy led white unionists to defend the colour bar, a pattern already settled at Kimberley. |