What happened in the 1960s to upset this way of constructing a statistical archive? This is going to be a research question for me and for Len Smith over the next few years, so my answer today is rather limited.
According to Borrie (1975: 455–6), Australia’s statisticians had long defended their very limited practice of Indigenous enumeration by pointing to two constraints. One was practical (the difficulties of enumerating ‘tribal people’), and the other legal (their interpretation of s.127 of the Constitution). By 1967 both of these obstacles had been overcome. However, Borrie also found among government officials the view that it was difficult to define who is an Aborigine and also ‘a vaguely formulated, but nevertheless strongly held view that separate statistics are in some way discriminatory, even if collected in order to make special provision for Aborigines’ (Borrie 1975: 456).
The word ‘discrimination’ is our clue to the crisis in the Indigenous statistical archive. Liberal opinion favouring ‘assimilation’ was obliged, by the mid-1960s, to consider whether there was a positive sense of ‘discrimination’. That is, if Aboriginal people were not doing well and if the state had an obligation to help them do well, might it not be necessary for the state to discriminate in their favour in certain ways? And was it not necessary to ‘discriminate’ (in the sense of distinguish) Aboriginal from non-Aboriginal in order to know how badly or how well Aborigines were doing, so that ‘positive discrimination’ could be soundly based? Some historians now interpret the 1967 referendum as an expression of a widespread (though not universal) conviction that it was time for public policy to discriminate in Aborigines’ favour, at least until they had ‘caught up’ with the rest of Australia in certain respects (Taffe 2005).
The idea that Australia was failing in its duty to improve Aborigines’ conditions of life was given powerful intellectual expression in the Social Science Research Council of Australia’s multi-author project ‘Aborigines in Australian Society’. The Rowley Project (as I like to call it, after its Director, Professor C.D. Rowley) produced 14 books between 1970 and 1980. It was a common complaint of the Rowley Project authors that social scientists had insufficient data on Indigenous Australians. Finding the census inadequate, Leonard Broom and Frank Jones remarked of Aboriginal affairs administration that ‘the management of a rubbish tip is more carefully monitored’ (Broom & Jones 1973: 75). For example, they reported that there were no reliable statistics on Aboriginal mortality other than about 15 years of records for the Northern Territory (Broom & Jones 1973: 63). For W. E. H. Stanner, ‘the very absence of more precise information is itself the best evidence of past indifference’ (Stanner 1970: vi–ix). Peter Moodie, in his study of Aboriginal health, pointed to official ‘caginess’ about quantifying ‘the Aboriginal problem’ (Moodie 1973: 275). However, he feared also that Aborigines themselves might resist distinguishing ‘any self-identifying Aborigines’ in databases (Moodie 1973: 274). He urged Aborigines to consider that by allowing their ‘statistical visibility’, Maoris and American Indians had improved their mortality (Moodie 1973: 122).
The argument for Aborigines’ ‘statistical visibility’, carried strongly by social scientists, was heard by the Whitlam government (and possibly it was influential under the Coalition governments in the formulation of the race question in the 1971 Census). In 1975, Borrie welcomed the terms of reference of the Whitlam government’s National Population Inquiry ‘that it should include the Aboriginal population not only in the total situation, but also as a separate sub-study’. He commented that, ‘There is no doubt that a separate study is needed. In every conceivable comparison, the Aborigines and Islanders, whom it is proposed in general to treat in one group, stand in stark contrast to the general Australian society, and also to other ”ethnic” groups’ (Borrie 1975: 455).
However, social scientists in the Rowley project had had to work with what they could find in the unreformed statistical archive. To illustrate the variety of their responses to its limitations, I will briefly outline what Rowley, Broom and Jones did.
Rowley showed one way that the racial distinctions in the extant statistical archive could be meaningful. He argued that there were two kinds of situation facing Aborigines: ‘colonial’ Australia and ‘settled’ Australia. Though his description of the difference between the two situations drew attention to differences in their characteristic histories of colonisation and in their resulting social institutions (Rowley 1970c: 2), he suggested a numeric index for deciding whether a region was ‘colonial’ or ‘settled’: the relative proportions of the full-blood and the mixed descent components in the Aboriginal population in each Statistical Division, according to the 1961 Census. We retain much of Rowley’s geographic binary when nowadays we compare ‘remote’ and ‘non-remote’ Indigenous statistics using the 2002 NATSISS.
In the Destruction of Aboriginal Society, Rowley (1970a) used 1961 Census and 1964 State/Territory estimates to show, State by State (but not Tasmania) the varying density (gross number per shire) of the Indigenous population. Differentiating full-blood from half-caste, he calculated their proportions in States and in capital cities, and he compared their distribution between urban and rural in each State/Territory. Distinguishing ‘full-blood’ from ‘mixed-blood’, he compared the age structures of each jurisdiction. He projected full-blood and mixed-blood population growth for 1961–81. The analytical value of distinguishing ‘half-caste’ from ‘full-blood’ is not clear, as Rowley preferred to account for Aborigines’ behaviour by reference to their socialisation, not by reference to their genetic characteristics. However, the statistical archive that he had to work with was shaped by the genetic terms of an earlier era, so we get table after table distinguishing ‘full-blood’ from ‘half-caste’.
Where he could do without the inherited statistical archive, Rowley did not distinguish half-castes and full-bloods. Researching Outcasts in White Australia (Rowley 1970b) he conducted two regional surveys (NSW outside Sydney, and Eyre Peninsula) with Aboriginal respondents. Without any reference to genetic characteristics, he presented data on Aborigines’ housing quality, school attainments, books in dwellings, institutional background, water supply, bathing facilities, sanitation, garbage disposal, household composition, occupation, previous and current employment, skills and experience, post-primary education and training, and average weekly wage. He looked at their receipt of social security benefits, their ownership of property, their use of hire purchase, and their use of insurance policies. On some of these variables, he tabulated men and women separately.
In a few of his arguments, Rowley compared Aborigines with non-Aborigines. In Destruction of Aboriginal Society, he did so only to the extent of comparing the age structures of the full-blood, mixed-blood, and ‘all Australians’ groups. In Outcasts in White Australia, Rowley analysed his data on ‘weekly income from all sources’ and on ‘membership of clubs and organisations’ by comparing Aborigines in rural NSW and Eyre Peninsula with non-Aborigines in the same regions. He obtained from the Commonwealth Bureau of Census and Statistics and from the Western Australian Government statistics on types of offence, leading to conviction or committal, for 1962–64. From NSW, Victoria and South Australia he obtained comparative data on types of offences for which Aborigines and non-Aborigines were arrested and charged in the second six months of 1965.
In A Blanket a Year (1973) Leonard Broom and Frank Jones had a rather different agenda, enabled partly because, writing later than Rowley, they had the fruits of the first reforms of the statistical archive. The 1966 Census, as well as being the first to claim complete enumeration, was the first to release data on Aborigines in the same terms as the data for other Australians. Broom and Jones were able to compare the educational and workforce status of Aborigines with non-Aborigines. However, we should note exactly what they meant by ‘Aborigines’, for in their efforts to compare Aborigines and non-Aborigines, Broom and Jones were still burdened by the racial classifications of the unreformed archive. Their data on ‘Aborigines’ included only those classified in the 1966 Census as having ‘50 per cent or more Aboriginal ancestry’. They wrote that ‘comparable data for approximately 16 000 to 17 000 identifiable Aborigines of less than 50 per cent have not been released, and they therefore cannot be dealt with here’ (Broom & Jones 1973: 13). Broom and Jones were concerned that by delimiting the Aboriginal population to those with 50 per cent or more Aboriginal descent, their comparison ‘may exaggerate the…dissimilarity between Aborigines as a whole and the rest of the population’ (Broom & Jones 1973: 24). In a section headed ‘selection as a factor in Aboriginal health statistics’, Moodie also worried about the ‘bias’ in the construction of Aboriginal health data, though the distortion that he pointed to was in terms of the institutional rather than the genetic ordering of the ‘Aboriginal population’ (Moodie 1973: 23). That is, data on Aboriginal health tended to be about ‘the more closely supervised government settlements, mission settlements, and the larger cattle stations’. These people were both ‘in a dependent situation’ and enjoyed ‘better access to medical and health services’ (for which they did not have to pay) than the more independent, but less well serviced ‘relatively large fringe-dwelling and metropolitan groups’ (Moodie 1973: 23).
Although they issued such caveats about the bias in their ‘Aboriginal’ categories, Broom and Jones and Moodie can be said to have initiated the research program that is now familiar to us as the comparative study of Indigenous and non-Indigenous labour market status, human capital acquisition and health status. These are among the pioneer works of our contemporary paradigm.
Broom and Jones were pioneers in another way. They constructed a ‘total’ Aboriginal population when they sought to project the growth in Aboriginal numbers. In this endeavour, they benefited from the Commonwealth Government’s interpretation of its responsibilities after the repeal of s.127 in the May 1967 referendum. The referendum of 1967 led to the undifferentiated inclusion of Aborigines in birth, marriage and death registrations, making it possible for Broom and Jones to estimate plausibly Aboriginal fertility and infant mortality in the only jurisdiction where Aborigines were a large minority—the Northern Territory. By using these registration data and the 1961 and 1966 Census data, in two papers (Jones 1970, 1972) and in chapter four of his book with Broom, Jones gave an account of the distribution, fertility and mortality of the entire Aboriginal population—combining the ‘dark’ and the ‘light’ segments.