E/CN.4/2005/21 page 6 23. Determining the extent of discrimination was a challenging task, owing to both conceptual limitations and inadequacies in the collection of data. During the first half of the twentieth century Brazil was portrayed as an accomplished example of racial democracy. The relatively disadvantaged position of non-Whites was ascribed to the legacy of slavery and to class prejudice, not to racial discrimination. However, mulattoes were perceived as having better social mobility than Blacks, as prejudice against “mixed blood” persons was considered to be weaker. 24. As of 1990, the category “indigenous” was also introduced, thus changing the country’s racial classification system to one based on both “colour” and “race”. The five categories presently used are: branco, pardo, preto, amarelo (yellow, which captures the descendents of Asian immigrants) and “indigenous”. Pardos and Pretos together constitute the “Negros” (Blacks) who represent the majority of the population in most states of the Brazilian Federation. This racial classification is used both by the national census, which is carried out every 10 years, and the “household survey”, which is an annual survey. The national census also inquires about nationality and place of birth. 25. In 2000, “Blacks” represented 45.2 per cent of the total population (5.7 per cent were Pretos and 39.5 per cent were Pardos), of 75 million. There were significant socio-economic inequalities between Whites and Blacks in Brazil and a clear hierarchy in the earning structure, with white men at the top, black women at the bottom and white women and black men in the middle. These disparities reflect: the considerable differences in human capital between races; the different distribution of occupations among races, a higher proportion of Blacks being employed in precarious, unskilled and low-paid jobs and beginning work at an earlier age; and discrimination in remuneration. Blacks (both men and women), regardless of the years of education, earned around 70 per cent of what Whites with similar qualification levels earned, thus suggesting the existence of discrimination. 26. The 1991 census of the United Kingdom introduced for the first time a question about the ethnic group affiliation of respondents and multiple choices were accepted. Nine categories had been identified: “White”; “Black Caribbean”; “Black African”; “Black Other”; “Indian”; “Pakistani”; “Bangladeshi”; “and Chinese” and “Any other ethnic group”. The census covered England, Wales and Scotland. 27. The 1994 Fourth National Survey covered only England and Wales, since only a limited proportion of minorities lived in Scotland. It used two questions to identify ethnic minorities: one was similar to the census’ question and the other asked about family origin (White/British/Irish, Black Caribbean, Indian Caribbean, Indian, Pakistani, Bangladeshi, Chinese, Other and Mixed). “Family origin” was considered to be more precise and reliable than group membership: family origin was a fixed fact, while “ethnic identity” might change in the life time of an individual or from one generation to another. Interestingly, the vast majority of people provided the same answer to both questions. 28. The Labour Force Survey of the late 1980s and the 1991 census reported some progress in the relative employment and earning levels of ethnic minorities compared to the early 1960s, when Caribbean and Asian people, irrespective of their class and levels of qualifications, were mostly clustered into low-paid manual work.