P.L. 2007, SPR, 74-94 Public Law 2007

RECONFIGURING DISCRIMINATION LAW Aileen McColgan. Copyright (c) 2007 Sweet & Maxwell Limited and Contributors Cases: DeGraffenreid v General Motors 413 F. Supp. 142 (D (US)) Bahl v Law Society [2004] EWCA Civ 1070; [2004] I.R.L.R. 799 (CA (Civ Div)) Subject: HUMAN RIGHTS Keywords: Canada; Discrimination; Race discrimination; Sex discrimination Abstract: Considers the issue of whether the existing discrimination model, in which special "grounds" are utilised to regulate forms of differential treatment, should be continued. Examines the grounds-based approaches to discrimination, noting the complex nature of social identity. Assesses US problems in this area and highlights difficulties categorising discrimination claimants. Discusses, with reference to case law, the Canadian treatment of discrimination claims.

Introduction In August 2004 the UK government finally succumbed to a groundswell of pressure to engage in a comprehensive overhaul of domestic discrimination legislation. The Equality Act 2006 has put in place the framework for the establishment of the Commission for Equality and Human Rights, while Equalities and Discrimination Law Reviews are under way with the intention, respectively, of investigating the "causes of persistent discrimination and inequality in British society" and considering "the fundamental principles of discrimination legislation [and] ... the opportunities for creating a simpler, fairer and more streamlined legislative framework in a Single Equality Act". The focus of many campaigners has been on streamlining and simplifying the existing legislative morass, as well as on rationalising the current unprincipled hierarchy of protection and, in some cases, extending the grounds upon which protection from discrimination is afforded. These are very important issues. But there are fundamental questions to be asked about some of the concepts which underpin that existing scheme. These include questions as to the current categorisation of discrimination into its direct and indirect forms; the separation of questions of justification from those regarding whether discrimination (at least in its direct form) has occurred; the generally symmetrical approach of discrimination legislation [1]; and whether discrimination law should retain the existing model whereby particular "grounds" are selected upon which differential treatment (or differentially impacting treatment) is regulated. All of these questions are worthy of exploration. But my intention here is to focus on the last of the issues raised above, that is, whether discrimination law should continue to protect only in relation to particular "grounds". Exploration of this, perhaps most fundamental, question touches to a greater or lesser degree on the second and third questions raised above. *75 Grounds-based approaches to discrimination Domestic legislation currently regulates (to a greater or lesser degree [2]) discrimination based on "sex", "colour, race nationality or ethnic or national origins", [3] "sexual 1

orientation", "religion or belief". In addition, legislation regulates discrimination based on the fact that a person is married or in a "civil partnership", [4] "intends to undergo, is undergoing or has undergone gender reassignment", [5] or is disabled within the meaning of the Disability Discrimination Act 1995 (see further below). The standard approach is symmetrical and grounds-based (people are protected from sex, race, sexual orientation and religion or belief discrimination regardless of their sex, race, sexual orientation, religion or belief [6]), while in other cases protection is grounds-based but asymmetrical in that it applies only to those who fall within a pre-determined protected group (those who are married or in a civil partnership, who intend to undergo, are undergoing or have undergone gender reassignment, or who are recognised as "disabled"). The groundsbased approach is common to many discrimination regimes ranging from Canada's Constitutional Charter of Rights (s.15 of which prohibits discrimination "in particular, without discrimination based on race, national or ethnic origin, colour, religion, sex, age or mental or physical disability") to Ireland's Employment Equality Act which regulates discrimination in relation to gender, marital status, family status, sexual orientation, religion, age disability, race and membership (or non-membership) of the traveller community, [7] and Title VII of the US Civil Rights Act 1964, which prohibits "discrimination based on race, color, religion, sex, or national origin". The most obvious questions which arise in connection with grounds-based approaches to discrimination include (a) what grounds are regulated and (b) how these grounds are defined. These questions are not independent of each other, narrow and/or exclusive approaches to (a) creating pressure for expansive interpretive approaches to (b). In Britain, for example, when statutory prohibitions on discrimination extended only to sex and race, the Sex Discrimination Act 1975 (SDA) and the Race Relations Act 1976 (RRA) came under pressure to accommodate, respectively, challenges to discrimination which might more obviously have been categorised as relating to sexual orientation and religion. The courts acceded to some degree in their application of the RRA, arguably stretching the notion of "race" to accommodate claims by Jewish and Sikh complainants, but holding firm against expansive pressures *76 in their application of the SDA. [8] The point here is not that purposive, expansive interpretation of discrimination statutes is a matter for criticism. But the grounds-based approach may set in stone institutional solutions to problems of a previous era and the courts may be unwilling, or perceive themselves as unable, to shape interpretative outcomes so as to make such legislation fit for current purpose. The SDA and RRA were passed at a time when sex and race discrimination were regarded as significant social problems, but discrimination associated with religious identity or affiliation had received little attention in Britain while that associated with disability and sexual orientation would scarcely have been recognised as such. In Mandla v Dowell Lee, which became the leading case on "ethnic group", the House of Lords adopted an approach to the concept which was not wholly based on assumptions relating to the biological underpinnings of "race" and which was sufficiently wide to accommodate claims (such as, for example, those by Sikh and Jewish converts) which could be understood as bringing into play a wider cultural sense of "ethnicity". [9] But while the approach adopted in Mandla was adequate to extend protection against "race" discrimination to groups characterised along these particular religious lines, and thus to vent the pressures then giving rise to litigation by these particular culturally defined groups, the Mandla criteria for identifying "ethnic groups" did not apply to groups defined by reference to "world religions" such as Christianity and--of significantly more practical importance--Islam. A judicial approach which had, in 1983, been adequate to extend the protection of the RRA to the "ethnic groups" then recognised as in need of legal protection became ossified as House of Lords precedent and was too wedded to notions of "race" (understood as biologically determined) to accommodate groups whose needs did not 2

arise--or went unnoticed--until later. Thus the position was created whereby Muslims, subject in particular after the 2001 attacks on New York's Twin Towers to abuse, vilification and discrimination, found themselves without the legal protection available to the adherents of other religions. [10] It would of course be possible to adopt a long list of grounds upon which discrimination is regulated to a greater or lesser extent, in a broad or narrow range of contexts. This is the approach adopted, for example, by Ireland's Employment Equality Act. [11] But even very wide approaches to grounds do not avoid problems generated by definitions of particular grounds. The UK's Disability Discrimination Act 1995 (DDA), for example, applies to those who have or have had "a physical or mental impairment which has a substantial and longterm adverse effect on [their] ability to carry out normal *77 day-to-day activities". [12] Very significant difficulties have been created by this triple-pronged definition which requires (a) a "medical" impairment and (b) a resulting functional impairment of a sufficient degree on (c) activities which are recognised as "normal". [13] What is particularly striking about this definition is that it excludes from the protection of the Act those who are discriminated against on the grounds of perceived, rather than actual, medical conditions or functional impairments. [14] Some modifications have been made very recently to remove the requirement for functional limitations in the case of those with HIV, multiple sclerosis and most forms of cancer, but it remains the case that someone sacked because she was incorrectly thought to have cancer, or a heart condition, or because her partner was known to be HIV positive, would not fall within the protection of the Act. Nor are problems over the definition of grounds restricted to those provisions requiring the claimant to establish membership of a protected "group". In the United Kingdom the fact that the RRA protects from discrimination "on racial grounds" rather than, as is the case under the DDA, "for a reason which relates to the disabled person's disability" (emphasis added), means that it extends to discrimination based on the "race" of a person's partner or associates, for example, as well as that of the claimant him or herself. [15] The same approach has been adopted by recent legislation dealing with discrimination on grounds of sexual orientation and religion or belief. But the SDA protects only from discrimination on the grounds of the claimant's own sex (or married/civil partnership/gender reassignment status). Partly for this reason, attempts have failed to rely upon the Act to challenge discrimination on grounds of the actual (or preferred) sex of the claimant's sexual partner(s). [16] The pressure on the category of "sex" could be relieved by the inclusion, as protected grounds, of sexual orientation and gender identity, claims by trans people having presented challenges to the courts in a number of jurisdictions. [17] But no matter how numerous the protected grounds, the grounds-led approach to discrimination gives rise to the fundamental difficulties to which I now turn. In the following sections I will consider the lack of "fit" between a grounds-led approach and the complexity of identity; the difficulties this has given rise to in litigation in the United States, and the shortcomings of the attempts which have been made to overcome these difficulties. I will then suggest that categorisations such as those which characterise the case law in this area are problematic and will look to the approach taken by the Canadian courts to discrimination before suggesting a number of ways in which domestic law might be made more amenable to the multiply disadvantaged claimant.

among those elements ... heighten ... differences between members and non-members of a category [and] suppress ... any similarities that some members might share with nonmembers". [19] Further: "Once a characteristic is created as intrinsic to a group, and becomes its identifier, it is regarded as wholly constitutive of that group's social identity... [and] the social identity constructed on the basis of this now 'intrinsic' difference is exactly the same for every member of the group... In this cartoon drawn from the perspective of the categorizer not from that of the subject of categorization, one social characteristic assumes gigantic proportions." [20] The identity which is thrust upon a class defined by a single identifier ("race", for example, "sex" or "sexual orientation") is not neutral. There is no such thing as an "unsexed" black person or "homosexual" or an "unraced" woman or lesbian. Rather, according to Iyer, "other aspects of social identity are rendered indistinguishable from the background norm", that is, "[t]he particular set of social characteristics of the dominant social identity and its ideology ... against which categorizations of difference are made in anti-discrimination law". [21] The purportedly "neutral" is in fact not so, but conforms in all respects (except that of the single identifier) with the "background norm": white, male, heterosexual, temporarily able-bodied, [22] Christian or irreligious. In making this point Iyer draws upon Martha Minow who has pointed out the way in which "difference" becomes identified with the less favoured in each set of binary oppositions ("male"/"female", "black"/"white", "heterosexual"/ "homosexual", "able-bodied"/"disabled", etc.) which is created by the "conception of equality and difference that poises a 'norm' and an 'other"', rather than as a quality of the relationship between persons, objects, etc. [23] Iyer points out that "differentiation ceases to be an explicit comparison *79 [in the analysis of discrimination/equality], and the non-dominant expression of the characteristic appears to inhere in the 'other,' the non-dominant group". [24] Men pass as "unsexed" for the purposes of a race discrimination claim, whites as "unraced" for the purposes of a sex discrimination claim, and heterosexuals as "neutral" in terms of sexual orientation. A grounds-based approach to discrimination requires "splitting" by those who wish to challenge discrimination connected with more than one ground. Angela Harris remarks that the "essentialism" inherent in this approach serves: "to reduce the lives of people who experience multiple forms of oppression to additional problems: 'racism + sexism = straight black women's experience,' or 'racism + sexism + homophobia = black lesbian experience.' Thus, in an essentiality world, black women's experience will always be forcibly fragmented before being subjected to analysis, as those who are 'only interested in race' and those who are 'only interested in gender' take their separate slices of our lives." [25] Those who fail to conform to the dominant social identity along more than one axis can hope, at best, that their multiple non-conformity is overlooked in the context of a claim based on a single axis. [26] And claimants who are discriminated against for more than one reason, or for a reason specific to their combination of factors by which they are differentiated from the "unstated norm", will struggle to establish, as they are required to do by the "unidimensional" approach, that the treatment of which they complain can be attributed to any one of these factors. The unidimensional approach in practice

*78 The dilemma of unidimensionality Nitya Iyer suggests that no list of grounds, however long, can avoid "obscur [ing] the complexity of social identity in ways that are damaging both to particular rights claimants, and to the larger goal of redressing relations of inequality". [18] She points out that "placing elements in a category tends to suppress differences and emphasize similarities

Unidimensionality generates a blindness towards to the complexly situated nature of all potential claimants (rather than solely those who differ from the "unstated norm" in more than one respect). This blindness is illustrated by the oft-cited decision in Degraffenreid v General Motors [27] in which a US district court summarily dismissed a claim by black women that they (but not white women or black men) were discriminated against. The

3

4

women had been laid off under a "last in first out" policy which had resulted in the dismissal of all black women working at the plant. According to the court: "The legislative history surrounding Title VII does not indicate that the goal of the statute was to create a new classification of 'black women' who would have greater standing than, for example, a black male.[ [28]] The *80 prospect of the creation of new classes of protected minorities, governed only by the mathematical principles of permutation and combination, clearly raises the prospect of opening the hackneyed Pandora's box." [29] The effect of the Degraffenreid decision was, of course, not to maintain parity as regards protection between black women, on the one hand; white women and black men, on the other. Rather, it denied protection to black women from discrimination on the basis of which white women were either protected (sex) or did not require (race). [30] Degraffenreid permitted black women to challenge discrimination that they shared with white women (discrimination on grounds of sex alone) or with black men (bare race discrimination). And black women who suffered from the cumulative (or "additive") effects of sex discrimination and race discrimination could challenge each of these aspects of disadvantage. [31] But Degraffenreid did not permit challenge to intersectional sex and race discrimination: discrimination taking the form, for example, of employment-related detriment connected with perceptions of black female sexual availability, [32] or recruitment discrimination based on perceptions of single black motherhood. The shortcomings of the unidimensional approach in the domestic context are illustrated by Bahl v Law Society. A tribunal accepted that the claimant, an Asian woman, had been discriminated against specifically as a black woman. The Employment Appeal Tribunal overturned the tribunal's decision, Elias J. ruling that the tribunal erred in law "in failing to distinguish between the elements of alleged race and sex discrimination", [33] and the Court of Appeal rejected Bahl's appeal. The court ruled that the tribunal had failed: "[T]o identify what evidence goes to support a finding of race discrimination and what evidence goes to support a finding of sex discrimination. It would be surprising if the evidence for each form of discrimination was the same... In our judgment, it was necessary for the [employment tribunal] to find the primary facts in relation to each type of *81 discrimination against each alleged discriminator and then to explain why it was making the inference which it did in favour of Dr Bahl on whom lay the burden of proving her case." [34] The Court of Appeal's decision indicates some of the difficulties experienced by discrimination claimants who differ from the "norm" (i.e. white, male, non-disabled, heterosexual, of mainstream Christian or indeterminate faith) by reference to more than one characteristic. Had the claimant in Bahl been either white or male (and otherwise "norm-conforming"), the first instance decision would have been immune from interference given the tribunal's finding of less favourable treatment in relation to a number of incidents and the inference permitted from such treatment and a difference in sex or (but not, it appears, and) race. [35] As it was, Dr Bahl would have had to make separate claims under the RRA by reference to the treatment of real or hypothetical white women, and under the SDA by reference to black men. If the discriminatory treatment which she alleged was intersectional, that is, specifically connected with her identity as a black woman, the RRA and SDA claims could each readily be defeated by evidence relating to the employer's non-discriminatory treatment of black men and white women respectively. [36] Not only does the decision in Bahl protect intersectional discrimination from challenge under domestic discrimination law, it renders proof of discrimination virtually impossible for someone who differs from the "unstated norm" in more than one respect, unless she avoids drawing attention to more than one of her deviations from this norm. And while Dr Bahl might at least have been permitted her choice of weapon had she chosen to pursue her claim under either, rather than both, the SDA and RRA (this because an employer is unlikely to defend a claim of sex discrimination by stating that the difference in treatment

was by reason of race, or vice versa), this will not always be the case. A claimant may be denied protection from discrimination on a prohibited ground because he has identification by reference to a different (and unprotected) characteristic thrust upon him, and that characteristic assumes Iyer's "gigantic proportions" and obscures recognition of the fact that the discrimination complained of relates to the prohibited ground. [37] So, for example, in Canada v Mossop Canada's Supreme Court rejected a claim under Canada's anti-discrimination statute by a gay man who argued that he had been discriminated against on grounds of his family status (undefined by the Act) by being denied leave in connection with the death of his partner's father. [38] A majority of the Supreme Court deduced from the fact that the claimant was *82 gay that the discrimination of which he complained could only be understood as discrimination on grounds of sexual orientation, which at the time was not covered by the relevant legislation. [39] A similar effect is evident in the decision of the House of Lords in Pearce in which their Lordships ruled that a lesbian subject to gender-specific abuse had been subject only to discrimination on grounds of her sexual orientation, which was not at the time subject to regulation, rather than her sex. [40] And yet further examples of this problem can be found in the US cases dealing with sex discrimination claims by "effeminate" men and "butch" women. The Supreme Court has recognised that same sex sexual harassment can amount to sex discrimination as long as the harassment is motivated by gender. [41] It has also accepted that discrimination on the grounds of failure to conform to gender stereotypes can amount to sex discrimination. [42] Most US jurisdictions have accepted that the requirement for proof that same sex harassment be motivated by gender can be satisfied by evidence that the plaintiff was targeted for a failure to comply with genderbased stereotypes. [43] But it remains virtually impossible for a gay man or a lesbian (or even someone suspected of being either) to convince a court that workplace harassment was because of their sex rather than (or as well as) their sexual orientation. [44] Yet a further shortcoming of the unidimensional approach to discrimination is in the context of indirect discrimination. Even under the most recent, and reasonably user-friendly, rubric, a woman claiming indirect race discrimination must show that she has had applied to her "a provision, criterion or practice which [was] applie[d] or would apply equally to persons not of the same race or ethnic or national origins as that other" but which, inter alia, "puts or would *83 put persons of the same race or ethnic or national origins as that other at a particular disadvantage when compared with other persons". A black woman claiming indirect sex discrimination must show that she has had applied to her "a provision, criterion or practice which [was] applie[d] or would apply equally to a man, but ... which puts or would put women at a particular disadvantage when compared with men". In neither case is it sufficient for the claimant to show that women of her racial group are put at a particular disadvantage by the employer's practice (the same is true when it comes to establishing indirect discrimination on grounds of sexual orientation, religion or belief and, from October 2006, age). A woman of Bangladeshi origin who wishes to challenge discrimination against homeworkers may, in a particular workplace or sector, be unable to establish a sufficiently disparate impact on women (taken together) to satisfy the requirements for indirect sex discrimination, in circumstances such that virtually all the women affected were also of Bangladeshi origin and the practice has a clearly disparate impact on women of Bangladeshi origin. She may equally be unable to establish indirect race discrimination if the number of Bangladeshi women engaged as homeworkers is relatively small by comparison to the overall number of men (or men and women) of Bangladeshi origin employed in the particular workplace or sector. Other practices may impact disproportionately on older women (but not older men or younger women), on gay men (but not straight men or lesbian women), on orthodox Jewish women (but not orthodox Jewish men or on other women). These will not be capable of challenge under the current

5

6

approach to indirect discrimination unless claimants are allowed to combine in law those factors which in fact combine to their disadvantage. Acknowledging complex claimants?: the "sex-plus" approach Degraffenreid was not widely followed in the United States. In Jefferies v Harris County Community Action Association the Fifth Circuit Court of Appeals allowed a "sex plus race" claim by a black woman, relying on Phillips and its progeny and stating that it was "beyond belief that, while an employer may not discriminate against these subclasses of women [those with young children, married women, single pregnant women, etc.], he could be allowed to discriminate against black females as a class. This would be a particularly illogical result, since the 'plus' factors in the former categories are ostensibly 'neutral' factors, while race itself is prohibited as a criterion for employment". [45] The real importance of the Jefferies ruling was in relation to the question of proof, the court ruling that the employer could not defeat a claim that it had discriminated against black women by demonstrating "the absence of discrimination against black men or white women". The "sex plus" approach *84 to discrimination claims is flawed for reasons which are discussed below. But it did constitute an advance from Degraffenreid and from the approach which prevails to this day in the United Kingdom and which was illustrated in stark form by Bahl v Law Society. Having said this, while the Jefferies court had referred to "distinctions ... between men and women on the basis of immutable or protected characteristics", without expressly demanding that any "plus" factor fell into one or other of these categories, the case was referred to by Pamela Smith as the "high-water mark in the judicial affinity for the combinational existences of Black women". [46] In Judge v Marsh the US District Court for the District of Columbia limited "sex-plus" cases to those featuring a single other factor only, taking the view that allowing "sex plus" claims "turns employment discrimination into a many-headed Hydra", resulting in "protected subgroups ... for every possible combination of race, color, sex, national origin and religion". [47] Peggie Smith points out that the "sex plus 1" limitation disadvantages black women who are rendered unable to challenge, for example, discrimination based in addition on marital status or pregnancy, that "[a] Black woman bringing a discrimination suit on the basis of a third factor ... would have already exhausted her 'plus' allowance with her race allegations". [48] Strictly applied, the Judge restriction would prevent challenge to discrimination against a black pregnant woman. Practical difficulties have arisen with the "sex plus" approach in cases in which black women in the United States have been denied certification as suitable representatives for class actions alleging race discrimination and sex discrimination (that is, cumulative or "additive", rather than "intersectional", discrimination). In Moore v Hughes Helicopter Inc, for example, the Ninth Circuit Court of Appeals remarked that "Moore had never claimed ... that she was discriminated against as a female, but only as a Black female" and that, therefore, there were "serious doubts as to Moore's ability to adequately represent white female employees". [49] (A woman who falls into the "sphere within a sphere" which is "sex plus race" is not regarded as properly within the sphere which is (solely) "sex"). This ruling might be seen to follow inexorably from the recognition (in Jefferies) that "race and sex discrimination" is qualitatively different from (cumulative) "race discrimination" and "sex discrimination". But it misses the point that all women are "raced", and that a white woman seeking representative status in a sex discrimination claim is no more characteristic of her black female colleagues than Ms Moore was of her white female colleagues. [50] *85 The Moore problem is at present one which has few practical implications in the United Kingdom where class actions are not as such recognised. But the conceptualisation by the "sex-plus" approach of white women as the default or "unraced" category has profound 7

repercussions for how "sex" discrimination is conceptualised. Diamond Ashiagbor has made the point that the equation of discrimination against part-time workers and discrimination against women holds true for the most part in the United Kingdom only for white women. [51] If black women are always treated as a subcategory of "woman", or as a subcategory of "black", the danger is that their particular concerns will always be underplayed because they will always be seen as minority interests within the wider group (whose perceived interests are, in fact, aligned with those of white women, in the case of sex, and black men in the case of race). It might be possible to reconceptualise "woman" to recognise the raced quality of each individual. This would have the advantage of recognising that a black woman is no less wholly a woman by virtue of her racial identification than is a white woman, or less wholly black by virtue of being also a woman than is a black man. But it fails to take into account the many-faceted nature of the individual. As Katherine Bartlett remarks: "One cannot talk about 'black women' ... without implying that one is talking about heterosexual black women, cannot talk about 'heterosexual black women'... without implying that one is talking about heterosexual able-bodied women. Any category, no matter how narrowly defined, makes assumptions about the remaining characteristics of the group that fail to take account of members of the group who do not have those characteristics." [52] This recognition is not a recipe for paralysis. [53] It does, however, indicate the difficulties inherent in grounds-based approaches to discrimination, even those which set out to accommodate the needs of those who experience disadvantage along more than one axis. The class of women alone is capable of subdivision into infinite subcategories of women along the lines indicated above. The class of men can be equally subdivided, and these myriad subcategories of men and women may require further refinement to capture all those subcategories of person defined according to the specificity of their racial identification, religious affiliation and/or specific disability, etc. The difficulties presented for *86 current conceptions of discrimination are obvious: to whom is an elderly black Christian heterosexual woman, or a bisexual disabled white agnostic woman, to compare herself for the purposes of a discrimination claim? Must the real or hypothetical comparator differ from her by reference to every identified characteristic? Will permitted grounds of distinction be limited, as in Judge v Marsh, to control the "many headed hydra"? Problematising categories Before turning to consider alternative legal approaches to discrimination, it is useful to draw attention to the difficulties inherent in categorising potential discrimination claimants and the grounds upon which they might be permitted to complain. Cases such as Pearce [54] posit as uncontroversial a distinction between "sex", on the one hand, "sexual orientation" or "gender" (in the sense of conformity of given genitals to behaviour and (clothed) appearance) on the other. "Common sense", it appears, has a significant role to play. Yet there is nothing preordained about the categories to which things, or people, are assigned for the purposes of social or, indeed, legal organisation. Simone de Beauvoir famously remarked that "one is not born, but becomes a woman". [55] And racial categories, too, "are fundamentally social in nature". [56] Thus Adrien Wing describes, as Angela Harris did before her, how at the beginning of the 20th century the United States was absorbed with the question whether the Poles, Irish, Jews and Armenians who were finding their way in large quantities to that country's shores were "white enough" to be worthy of citizenship. [57] The ascription of "race", like that of "sex" or "gender", can have profound significance. But this significance is not the consequence of any "truth" it captures about the essential "nature" of those who fall into the category [58]: *87 Jean 8

Charles de Menezes, who was shot in London in July 2005 by police who apparently mistook him for an Islamist bomber, was not saved from death by the fact that he was, in fact, a Brazilian Catholic, rather than a Muslim (or even, by proxy, an "Arab" or an "Asian" [59]). And just as an infertile woman who is refused employment on the grounds that she is considered likely to become pregnant is no less discriminated against on grounds of sex than her fecund sister would be by similar treatment, so a woman who is sacked in the mistaken belief that she is HIV positive is no less victimised on grounds of disability than she would be if the belief was accurate. The power of the categorisation is no less real by virtue of any lack of "fit" between the person who is "mistakenly" categorised and the assumed "essential" nature of the category. [60] Categorisations have enormous significance in the lived world. But protection against racism, sexism, heterosexism/homophobia and discrimination based on actual or perceived religious belief cannot rationally be provided only to those who fit some externally imposed definition. The question whether Mr Mossop was discriminated against on grounds of his "family status" should not turn on whether his partner was male or female, any more than the question whether a woman is discriminated against as a woman by being subject to sexualised abuse should turn on whether she is lesbian or heterosexual, or the question whether a man victimised because he fails to conform to other men's notions of appropriate masculinity is subject to sex discrimination should turn on whether he is gay, straight or bisexual. [61] The socially constructed nature of categories calls into question the utility of "rationing" of protection from discrimination either (as in the case of the DDA) to those who fall within certain pre-defined groups or, as is more commonly the case in the United Kingdom, to groups defined along grounds- related lines. The former approach may at least have the effect that protection is targeted at those persons or groups who are most vulnerable to disadvantage and/or prejudice, although the grounds which are typically protected at present generally fail to protect other than indirectly from disadvantage and/or prejudice associated with class and poverty. But the latter type of protection applies equally to those who fall on the more privileged side of each line and so the obsession with categorising is particularly misguided.

By calling into question the grounds-driven approach to the regulation of discrimination I do not suggest the replacement of that approach with one which replaces a focus on grounds with an emphasis on "diversity". As Sheila *88 Foster points out, "a left-handed person, an obese person, a genius, and innumerable other persons deemed to be 'different' [would] satisfy the goal of diversity". [62] How might the difficulties associated with our current grounds-driven approach to equality be avoided, without running the risk of a collapse into vacuity? One possibility has been suggested by L'Heureux-Dubé J., in the application of Canada's Charter of Rights, s.15 of which provides that: "Every individual is equal before and under the law and has the right to equal protection and benefit of the law without discrimination and, in particular, without discrimination based on race, national or ethnic origin, colour, religion, sex, age or mental or physical disability." In Andrews v Law Society of British Columbia, Canada's Supreme Court ruled that "discrimination", for the purposes of s.15, consisted in (1) a distinction (intended or otherwise), (2) based on grounds relating to the personal characteristics of the individual or group concerned, which (3) has the effect of imposing disadvantages or burdens not imposed on others, or of withholding access to advantages or benefits available to others. [63] The personal characteristics which could found a claim were those enumerated within s.15 or analogous thereto, determination of the question whether grounds were analogous

to those in s.15 turning, inter alia, on whether those in the group defined by reference to the personal characteristics at issue had been subjected to historical disadvantage, stereotyping and prejudice. The Andrews approach required a focus on grounds as a preliminary stage, before the court could address the substantive question of whether differential treatment [64] amounted to "discrimination". In Egan v Canada, L'Heureux-Dubé J. suggested (dissenting) that Canada's Supreme Court should retreat from the grounds-based approach it had adopted to the analysis of discrimination under s.15 of the Charter. [65] She took the view that a purposive approach to s.15 required a focus on the effects, rather than on the constituent elements, of discrimination, and that s.15 would be breached by (1) a legislative distinction which (2) resulted in a denial of one of the four equality rights (equality before the law, equality under the law, equal protection of the law or equal benefit of the law) on the basis of the rights claimant's membership in an identifiable group, in circumstances such that (3) the distinction was "discriminatory" within the meaning of s.15. For a distinction (direct or impact-related) to be "discriminatory" (the third step of L'Heureux-Dubé J.'s proposed test), it had to be capable of either promoting or perpetuating the view that the individual adversely affected by it was less capable, or less worthy of recognition or value as a human being or as a member of Canadian society, equally deserving of concern, respect, *89 and consideration. This required consideration of both the nature of the group affected by the distinction and the nature of the interest adversely affected by it. L'Heureux-Dubé J. stated that "the social context of the distinction", rather than the "grounds" upon which it operated, were "dispositive of the question of whether discrimination exists". In her view, an a priori limitation on the grounds of discrimination "marks the introduction of the same sort of counter-productive, formalistic and artificial debates that have been conducted under anti-discrimination legislation" on issues such as whether "sex" discrimination included discrimination in connection with pregnancy and sexual orientation. Using the example of discrimination against domestic workers, L'Heureux-Dubé J. criticised the fact that the court would, under the Andrews approach, "have to ask whether legislation involving differential treatment of domestic workers differentiates on the basis of sex because most domestic workers are women ... rather than facing squarely the issue of differential treatment of domestic workers". She went on to state that "categories of discrimination cannot be reduced to watertight compartments, but rather will often overlap in significant measure" and that, accordingly, the fact that most domestic workers were immigrant women, "a subgroup that has historically been both exploited and marginalized in our society", was relevant to the assessment of "the social context of the impugned distinction". However interesting L'Heureux-Dubé J.'s approach, its utility is restricted to a constitutional prohibition on discrimination and one which is, further, like s.15 of the Charter, inherently asymmetrical. Whereas, under a symmetrical approach, protection from less favourable treatment "on grounds of race" applies to white as well as black, a s.15 finding that a person has been subject to "differential treatment based on one or more enumerated and analogous grounds" [66] serves only to trigger the real substantive question: whether they have been "discriminated" against (as distinct from merely subject to unequal treatment or impact) for the purposes of s.15. L'Heureux-Dubé J. is concerned with whether scrutiny of this substantive question ought to be restricted to those cases in which claimants can place themselves in a group defined by reference to specific personal characteristics. In the context of domestic statutory provisions, by contrast, and except where disability, married status and gender reassignment are concerned, the delineation of "protected grounds" is not concerned with targeting protection from "discrimination" towards those who are peculiarly vulnerable to it. "Discrimination" is defined, for the most part, formally. Thus "direct" discrimination consists in formally unequal treatment in the absence of a

9

1

Alternative approaches: lessons from Canada?

recognised, tightly defined, exception. Indirect discrimination, it is true, operates differently in that only those who share specified characteristics with members of a disadvantaged group which is itself defined by reference to those characteristics may challenge practices as indirectly discriminatory. [67] But the *90 prohibition on "direct" discrimination benefits men, whites, heterosexuals and those whose religion, if any, is consistent with the dominant norms. Grounds are selected by reference to disadvantage (that experienced, respectively, by black people, women, gay, lesbian and bisexual people and adherents to minority religions). But the link between grounds and disadvantage does not regulate the legal operation of the prohibition on "discrimination" which is, in the United Kingdom, understood as a thin concept of less favourable treatment to which an express exception does not apply [68] rather than, as in Canada, a nuanced concept which takes into account questions of existing disadvantage and justification. On the one hand, this means that the restriction of the prohibition on discrimination to particular grounds is peculiar (inasmuch as, in the manner of its application, there is no necessary link between protected grounds and disadvantage). Equally, however, the prohibition on (virtually) all differential treatment on these protected grounds means that any move away from a tightly defined list of such grounds risks chaos.

Denise Réaume suggests that statutory regulation of discrimination of the type which currently exists in the United Kingdom is unsatisfactory because it combines a rigid statutory structure with an absence of "grand theory". [69] Réaume contrasts "top-down" Benthamite legislation with "bottom-up" Blackstonian judicial norm creation, and suggests that the combination of a rigid statutory framework with an absence of a "comprehensive moral theory" leaves the rules at once "without adequate moral foundation and seriously incomplete" while their legislative precision "hinder[s] adjudicators from filling the gaps that will inevitably come to light over time". [70] This has the result, as she points out of the statutory rules regulating discrimination in Ontario, Canada, of creating "arbitrary pigeonholes into which complainants must fit their fact situations or fail". [71] Réaume advocates the adoption of an open list, suggesting that cross-sectional claims can be dealt with by recognising that: "[I]n no other civil claim is it necessary to establish that there is a single, sufficient cause or explanation of the harm suffered in order to succeed. Normally, all that matters is that the alleged wrongdoing is a 'but-for' cause or necessary condition; the existence of multiple necessary conditions that come together to create the harm is no bar to recovery, especially when both causes are alleged to be unlawful behaviour and are combined in the *91 behaviour of a single agent. By this standard, provided that a black female can show that had she been white or male, she would have been hired (or her chances would have been improved), she ought to be considered to have established a case for both race and sex discrimination." [72] Réaume's point is a powerful one. But one strong argument for the retention of legislative control over the grounds upon which discrimination can be challenged in the context of statutory discrimination provisions, as distinct from a broad constitutional prohibition (as in the case of s.15), relates to the precision of approach typical of such statutory schemes. It is difficult to imagine how statutory provisions along the lines of (say) the current s.4 of the RRA could be combined even with an open-ended list approach. Whereas "discrimination" as understood for the purposes of s.15 of Canada's Charter implies unfair or unjustifiable treatment, discrimination for the purposes of domestic legislation does not (at least where it takes the direct form) have any such connotations. To treat a person less favourably because she is relatively under-qualified, or because she fails an IQ test, for example, is to discriminate against her on the grounds of her educational performance or her intelligence.

And to the extent that the regulated grounds of discrimination were openended, an employer would face insurmountable difficulties under the current statutory approach which does not recognise any general justification defence. If we were to move from (in essence) a closed list to an open list of protected grounds, it would be necessary to amend the statutory scheme to recognise a general justification defence or, more radically still, to define discrimination in terms of unfair or unjustifiable, rather than simply different, treatment. The same would quite obviously be true if we were to proceed by means of abandoning grounds entirely. This would have some benefits, for the reasons Réaume suggests. But it would involve the provision to the judiciary of a great deal of discretion currently denied it. And it is by no means clear that that body could be relied upon to exercise this discretion in a manner favourable to the disadvantaged. Passing the baton on questions such as when differential treatment may be justified to a judiciary responsible for decisions such as those in Bahl, Pearce and (in the case of the Court of Appeal, though not the House of Lords) A v Secretary of State for the Home Department and R. (on the application of European Roman Rights Centre) v Immigration Officer at Prague Airport is perhaps not the best approach. [73] Given these concerns about uncertainty, on the one hand, and a failure of purposive engagement by significant elements of the judiciary, on the other, the most appropriate way to proceed as far as statutory prohibitions on discrimination are concerned may be on the basis of a closed but expanded *92 list of grounds (extending, for example, to family status), drafted so as to facilitate multiply disadvantaged claimants. Such a list would need to be coupled with a provision such as that in the Canadian Human Rights Act (s.3(1)) which states that "a discriminatory practice includes a practice based on one or more prohibited grounds of discrimination or on the effect of a combination of prohibited grounds". Revised legislation ought also to clarify the role of causation, perhaps with a provision to the effect that a finding of discrimination requires only that the ground or grounds relied upon was a causal factor in the treatment challenged. Thus, for example, a claimant in Dr Bahl's position would be entitled to succeed if she could establish [74] that the treatment complained of was caused wholly or in part by her identity as a black woman (or an Asian woman of a particular age or social class). It should also be clarified that discrimination against sections of groups defined along protected grounds (black men from particular geographical locations, for example, or Asian women engineers) is prohibited. And the centrality of the comparator to the analysis of discrimination must be challenged, given the difficulties with identifying persons whose "relevant circumstances" are the same or materially similar to those of claimants who complain of intersectional discrimination. [75] Any new legislation ought to contain a strong purpose clause which makes specific reference to the need to tackle the structural disadvantage suffered by particular groups, many of them at the intersection of protected and other grounds. Crucially, such legislation ought not to define terms such as "race", "disability", "sex" and "sexual orientation", [76] and it ought expressly to provide that treatment resulting from perceptions about the claimant's membership of any group defined by reference to a protected ground, or his or her (lack of) conformity to characteristics associated with such a group, is covered by the prohibition on discrimination. This would have the effect that someone disadvantaged because he is (wrongly) suspected of having cancer, or because she has diabetes (but is functionally unimpaired) would be protected, as would a Muslim woman who is discriminated against because she does not wear a veil, or because it is assumed, because she is a Muslim woman, that she will adopt a subordinate attitude towards male clients or colleagues. This new approach to statutorily protected grounds ought not to be coupled with an abandonment of detailed statutory regulation of discrimination (as distinct from a broad constitutional prohibition). But the statutory approach must be revised. As Sarah Hannett

1

1

Conclusion

has noted, "[u]nder our existing ideology of formal equality, every individual represents a potential claimant of multiple discrimination, regardless of race, sex, etc". Hannett goes on to advocate an *93 approach based on disadvantage, arguing that "[c]ontinuing to frame legislation in terms of difference rather than disadvantage, particularly in cases of alleged intersectional discrimination, obscure[s] the historical and continuing realities of inequality facing the subordinated group within each ground" [77] and "rules out any structural protective or remedial measure that addresses disadvantage". According to Hannett, "a real commitment to addressing intersectional discrimination requires a shift towards a substantive equality paradigm". Perhaps the most important first step in this would relate to questions of proof, and could take place even prior to the adoption of new legislation along the lines suggested above, at least where the discrimination alleged is subject to the "shifted" burden of proof. [78] In Dresdner Kleinwort Wasserstein Ltd v Adebayo, the EAT ruled (Cox J.) that a prima facie case of race discrimination could be established, where a (hypothetical) black claimant complained of non-appointment to a position, where the claimant could show that he was "at least as well qualified" as the white candidate selected for the post. [79] In such a case the burden would shift to the alleged discriminator to provide an explanation for the less favourable treatment in order to avert a finding of discrimination, and failure to prove an absence of unlawful discrimination would result in a finding for the claimant. The decision in Adebayo is to be welcomed for easing the burden on discrimination claimants. But it ought to be acknowledged that every decision as regards appointment or promotion, for example, is capable of spawning multiple discrimination claims by all of those unsuccessful applicants who can distinguish themselves from the successful candidate (or even from a candidate whose application progressed further than their own) by reference to sex, race, sexual orientation, age, religion or belief. The successful candidate will be an amalgamation of characteristics identified by reference to each of these protected grounds as well as to many others. Thus for every young white Protestant lesbian appointed to a position, an easy shift in the burden of proof would permit every unsuccessful candidate who was either older (or younger), not white, not Protestant, not lesbian or not a woman to require the employer to prove that it had not discriminated against them on the ground of sex and/or race and/or sexual orientation and/or age and/or religion or belief. It may be that, in recognition of the significantly expanded list of protected grounds and the associated potential for an exponential increase in litigation, the time has come to draw a formal distinction, for at least these purposes, between those advantaged and those disadvantaged by reference to protected grounds. Thus, for example, where a woman asserts sex discrimination in relation to her non-appointment to a position in a predominantly male firm, or a Muslim man claimed discrimination on grounds of religion in relation *94 to his dismissal from an otherwise non-Muslim workplace, the claimant's identification with a category which is disadvantaged by sex and religion respectively ought to provide for an easy shift in the burden to the employer to disprove discrimination. Where, by contrast, a Muslim applicant is appointed to a position in a predominantly nonMuslim workplace, disgruntled non-Muslims ought not to be permitted to benefit from a shift in the burden of proof on the basis of her religion and their equal qualifications alone. The fact is, of course, that those advantaged rather than disadvantaged by reference to protected grounds are perhaps less likely to suspect discrimination on the basis of that ground in the absence of a circumstance such as a programme of positive action, and it may be that in the absence of such a programme a tribunal might readily accept a nondiscriminatory explanation as satisfying the employer's burden of proof. But a particular benefit of a formal recognition of the relevance of disadvantage in this context would be that it would ease the burden on multiply disadvantaged claimants while resisting a descent into absurdity on the part of the law. [80]

A nuanced application of the rules relating to burden of proof is only one step towards a more satisfactory accommodation of the needs of the multiply disadvantaged. But moving towards an open-ended approach to discrimination, which will be necessary more adequately to address these needs, will have to be coupled with a more thoroughgoing retreat from symmetry if it is to avoid plunging the system into a morass of unmeritorious claims by the variously advantaged. It is crucial that we move towards a definition of discrimination in terms of the exacerbation of disadvantage. How this definition might best be formulated is a question for another time.

1

1

My warmest thanks to Keith Ewing and to Sarah Hannett for their helpful comments on earlier drafts. This paper was written before, although published after, the coming into force of the Employment Equality (Age) Regulations 2006 and makes no reference to them. 1. The current main exception to this is the Disability Discrimination Act 1995. 2. The regimes differ as to coverage (the RRA being most protective, the provisions applicable to gender reassignment and married status least so) and definitions of discrimination adopted. See, for general discussion, A. McColgan, Discrimination Law: Text, Cases and Materials (2nd edn, Hart Publishing, Oxford, 2005), Chs 1 and 2. 3. Race Relations Act 1976 (RRA), s.1. 4. Sex Discrimination Act 1975 (SDA), s.3. 5. SDA, s.2A. 6. And, as of 2006, age discrimination regardless of their age though (in common with sexual orientation, religion and belief, only in the employment context. 7. s.6(2). 8. Smith v Gardner Merchant Ltd [1999] I.C.R. 134; MacDonald v Ministry of Defence, Pearce v Governing Body of Mayfield School [2003] UKHL 34, [2003] I.C.R. 937. 9. [1983] 2 A.C. 548. 10. This prior to the implementation of protection in the employment field and, more recently, in relation to access to goods and services, etc. (the Employment Equality (Religion and Belief) Regulations 2003 (SI 2003/1660) and the Equality Act 2006). 11. s.6(2). 12. s.1. 13. See Discrimination Law: Text, Cases and Materials, .2 above, Ch.8. 14. cf. e.g. Quebec v Montreal [2000] 1 S.C.R. 665. 15. Showboat Entertainment Centre v Owens [1984] I.R.L.R. 7; Weathersfield Ltd v Sargent [1999] I.R.L.R. 94, approved by the Court of Appeal in Redfearn v Serco Ltd [2006] EWCA Civ 659, [2006] I.R.L.R. 623.

16. Smith v Gardner Merchant Ltd [1998] 3 All E.R. 852; Pearce, .8 above. 17. See, e.g. Ulane v Eastern Airlines, Inc 742 F.2d 108, Etsitty v Utah Transit Authority 2005 W.L. 1505610, 2005, Vancouver Rape Relief Society v Nixon [2003] B.C.S.C. 1936. 18. N. Iyer, "Categorical Denials: Equality Rights and the Shaping of Social Identity" (1993) 19 Queen's Law Journal 179, p.181. 19. ibid., p.183. 20. ibid., pp.191-192

discrimination which is "neutral" as to sex: see Smith, .28 above, p.243 and P. Smith, "Forging Our Identity: Transformative Resistance in the Areas of Work, Class, and the Law" (2000) 33 UC Davis Law Review 1105, p.1110. 31. S. Hannett, in "Equality At The Intersections: The Legislative And Judicial Failure To Tackle Multiple Discrimination" (2003) 23 O.J.L.S. 65, distinguishes between "additive" and "intersectional" discrimination, citing E.W. Shoben, "Compound Discrimination: The Interaction of Race and Sex in Employment Discrimination" (1980) 55 NYU Law Review 793, p.794 and M. Eaton, "Patently Confused: Complex Inequality and Canada v Mossop" (1994) 1 Review of Constitutional Studies 203, p.229. 32. See, e.g. H. Carby, Reconstructing Womanhood: The Emergence of the AfroAmerican Woman Novelist (1987); P. Smith, "Separate Identities: Black Women, Work, And Title VII" (1991) 14 Harvard Women's Law Journal 21, pp.72-74.

21. ibid., pp.185-186. 22. R. Abel, "Critique of Torts" (1990) 37 UCLA Law Review 785, p.805. Note the assertion by Brennan J. for the Supreme Court in Frontiero v Richardson 411 U.S. 677 (1973), p.685 that "Blacks had gotten the right to vote--before women". According to Cathy Scarborough ("Conceptualising Black Women's Employment Experiences" (1989) 98 Yale Law Journal 1457, p.1463): "The Court saw only two categories: Black men, who had the right to vote, and white women, who were imprisoned in a paternalistic cage. Where do Black women fit into this picture?".

33. [2003] I.R.L.R. 640.

23. See, e.g. M. Minow, "Learning to Live with the Dilemma of Difference: Bilingual and Special Education" (1985) 48 Law and Contemporary Problems 157; "Justice Engendered" (1987) 101 Harvard Law Review 10.

36. Though note that on the facts of the instant case there would have been no actual women or black comparators.

34. [2004] EWCA Civ 1070; [2004] I.R.L.R. 799. 35. Crenshaw, .30 above, p.151, uses the metaphor of a basement in which only those who are least different from the "norm" can fight their way up to escape out of a trap door-only those at the top, privileged "but for" a single protected factor, can get out.

37. Or where a hierarchy of protection exists the claimant might be restricted to challenging discrimination on the basis of the less protected ground.

24. Iyer, .18 above, p.190. 25. "Race and Essentialism in Feminist Legal Theory" (1989) 42 Stanford Law Review 581, p.588. 26. For discussion of the single-dimensional approach to black women in Canada see N. Duclos (subsequently Iyer), "Disappearing Women: Racial Minority Women in Human Rights Cases" (1993) 6 Canadian Journal of Women and the Law 25.

38. [1993] 1 S.C.R. 554. 39. See D. Kropp, " 'Categorial' Failure; Canada's Equality Jurisprudence-- Changing Notions of Identity and the Legal Subject" (1997-98) 23 Queen's Law Journal 201, p.221. 40. See now the Employment Equality (Sexual Orientation) Regulations 2003 (SI 2003/1661) and the Equality Bill 2006.

27. 413 F. Supp. 142 (1976). 28. As Pamela Smith points out in "Part II: Romantic Paternalism--The Ties that Bind" (1999) 3 Journal of Gender Race and Justice 181, p.217: "Black women as a class already existed ... The question was whether the court would preclude discrimination that was targeted at this class." 29. The appeal was rejected by the Eighth Circuit Court of Appeals on different grounds (558 F. 2d 480 (1977)), the appeal court not dealing with the question whether black women as such could seek protection under Title VII.

41. Oncale v Sundowner Offshore Servs, Inc 83 F.3d 118. See generally M. Case, "Disaggregating Gender from Sex and Sexual Orientation: The Effeminate Man in the Law and Feminist Jurisprudence" (1995) 105 Yale Law Journal 1. 42. Price Waterhouse v Hopkins 490 U.S. 228.

30. For K. Crenshaw, "Demarginalizing the intersection of Race and Sex" (1989) University of Chicago Legal Forum 139, p.145, white women's "race does not contribute to the disadvantage for which they seek redress. The view of discrimination that is derived from this grounding takes race privilege as a given." The position for black men is more complicated. For the most part it is they who are protected by a prohibition on race

43. Jones v Pacific Rail Services 2001 W.L. 127645 (District Court, Illinois, 2001); Higgins v New Balance Athletic Shoe, Inc 21 F.Supp.2d 66 (1st Circuit, 1999); Simonton v Runyon 232 F.3d 33 (2nd Circuit, 2000); Bibby v Phila Coca Cola Bottling Co 260 F.3d 257 (3rd Circuit, 2001); Smith v City of Salem 256 F.3d 864 (6th Circuit, 2001); Spearman v Ford Motor Co 231 F.3d 1080 (7th Circuit, 1999), Schmedding v Tnemec Co, Inc 187 F.3d 862 (8th Circuit, 1999); Nichols v Azteca Restaurant Enterprises Inc 378 F.3d 566 (9th Circuit, 2004) and James v Platte River Steel Co 113 Fed Appx. 864 (10th Circuit, 2004). cf. however the Texas Court of Appeal in Willborn v Formosa Plastics Corp of Texas 2005 W.L. 1797022, July 28, 2005.

1

1

44. See, for a typical decision, Spearman v Ford Motor Co 231 F.3d 1080. The case law is usefully discussed by M. Sachs, in "The Mystery Of Title VII: The Various Interpretations Of Title VII As Applied To Homosexual Plaintiffs" (2004) 19 Wisconsin Women's Law Journal 359. See also G.S. Trotier, "Dude Looks Like a Lady: Protection Based on Gender Stereotyping Discrimination as Developed in Nichols v Azteca Restaurant Enterprises" (2002) 20 Law and Inequality 237; F. Valdes, "Queers, Sissies, Dykes and Tomboys: Deconstructing the Conflation of 'Sex,' 'Gender,' and 'Sexual Orientation' in Euro-American Law and Society" (1995) 83 California Law Review 1 and D. Hutchinson, "Identity Crisis: 'Intersectionality,' 'Multidimensionality,' and the Development of an Adequate Theory of Subordination" (2000-2001) 6 Michigan Journal of Race & Law 285, p.303.

55. The Second Sex, p.267. And see K. Sheenan, "Caring for Deconstruction" (2000) 12 Yale Journal of Law and Feminism 85, pp.116-117. Even those differences which are regarded as biological givens are often not so. Women, for example, are encouraged to exaggerate the size of their breasts (whether by clothing or more permanent enlargement), to dress so as to accentuate waist and bottom (with high heels), to curl and primp their hair, etc. while men are encouraged to exaggerate and/or enhance their muscularity. See also K.M. Franke, "The Central Mistake of Sex Discrimination Law: The Disaggregation of Sex from Gender" (1995) 144 University of Pennsylvania Law Review 1; Farrell, ibid, pp.614-615; K.L. Karst, "Constitutional Equality As A Cultural Form: The Courts And The Meanings Of Sex And Gender" (2003) Wake Forest Law Review 513 and D.B. Cruz, "Disestablishing Sex and Gender" (2002) 90 California Law Review 997.

45. 615 F.2d 1025. See also Arnett v Aspin 846 F. Supp. 1234 (1994) (sex plus age); Chambers v Omaha Girls' Club 834 F.2d 697 (8th Circuit, 1987) (sex plus pregnancy); Trezza v Hartford Inc (1998) 78 Fair Empl. Prac. Cas. (B.N.A.) 1826 (sex plus parenthood); Longariello v School Board of Monroe County Florida 987 F.Supp 1440 (1997) (sex plus marital status).

56. L. Outlaw, "Towards a Critical Theory of Race" in Anatomy of Racism (David Goldberg, ed., 1990) pp.58, 68. See also A. Harris, "Equality Trouble: Sameness and Difference in Twentieth-Century Race Law" (2000) 88 California Law Review 1923. At p.1923 Harris highlights the role of law in the US in regulating the "formation, recognition, and maintenance of racial groups, as well as ... the relationships among these groups".

46. .28 above, pp.215-216.

57. A.K. Wing, "Brief Reflections Toward a Multiplicative Theory and Praxis of Being" (1990-1991) 6 Berkeley Women's Law Journal 181.

47. See, for discussion Hannett, .31 above. 48. .32 above, p.46. 49. 708 F 2d 475 (1983). See also Richardson v Coopers & Lybrand 25 Fair Empl. Prac. Cas. (BNA) 1129 discussed by P. Smith, .31 above, p.55. 50. Similarly in the case of a black man seeking representative status in a race discrimination suit and claiming to represent black women also. For a recent paper on class actions and complex claimants see M.E. Powell, "The Claims of Women of Color Under Title VII: The Interaction of Race And Gender" (1996) 26 Golden Gate University Law Review 413. 51. "The intersection Between Gender and 'Race' in the Labour Market: Lessons for Antidiscrimination Law" in A. Morris and T. O'Donnell, Feminist Perspectives on Employment Law (Cavendish, London, 1999) p.139, pp.142-146. See also K. Abrams, "Title VII and the Complex Female Subject" (1993-1994) 92 Michigan Law Review 2479, p.2498; K. Crenshaw, "Mapping the Margins: Intersectionality, Identity Politics, and Violence against Women of Color" (1991) 43 Stanford Law Review 1241, p.1252; M. Kline, "Race, Racism and Feminist Legal Theory" (1989) 12 Harvard Women's Law Journal 115. Smith points out, .28 above, that only five "sex plus race" cases decided since Jefferies were decided in the US between 1980 and 1991. Among the few reported cases of "sex plus race" are Lam v University of Hawaii 40 F.3d 1551, and Andrews v County of Sacremento 898 F. Supp 1435 (1995).

58. This has been recognised by the EU (see recital 6 of Council Directive 2000/43). According to the European Network of Independent Experts in the non-discrimination field, (Developing Anti-Discrimination Law in Europe (2005), pp.19-20), some Member States have avoided using the words "race" or "racial" in transposing the Directive on the basis that their use "reinforces the perception that humans can be distinguished according to 'race', whereas there is no scientific foundation for such categorisation." 59. This being the category his assumed membership of which caused him to be suspected of being a terrorist. See P. Currah, "Defending Genders: Sex And Gender NonConformity In The Civil Rights Strategies Of Sexual Minorities" (1997) 48 Hastings Law Journal 1363, pp.1365-1366. 60. This is not invariably true. If, instead of being shot, Jean Charles de Menezes had merely been subject to a tirade of abuse about Islam, or conduct calculated to insult him as a Muslim (discussions about provocation invariably involve the throwing of a pigskin slipper), his lack of self-identification as a Muslim would draw much of the sting from the insult. 61. And in the cases discussed at .43 above. 62. "Difference and Equality: A Critical Assessment of the Concept of 'Diversity"' [1993] Wisconsin Law Review 105, p.109. 63. [1989] 1 S.C.R. 143.

52. "Feminist Legal Methods" (1990) 103 Harvard Law Review 829, p.848. 64. Whether by reference to intention or impact. 53. See J. Conaghan, "Reassessing the Feminist Theoretical Project in Law" (2000) 27 Journal of Law and Society 351. 54. Also the US cases at .43 above.

65. [1995] 2 S.C.R. 513. 66. This being the test set out in Law v Canada [1999] 1 S.C.R. 497 to replace the Andrews approach.

1

1

67. Though once a practice (say a requirement for full-time work) is recognised as indirectly discriminatory it cannot be disapplied to those of the disadvantaged group (say women) but applied to those of the advantaged group (say men). This would amount to direct discrimination against men. Rather, the practice would have to be abandoned entirely or applied only to those who were not rendered unable to comply with it for a reason which disproportionately affected one sex (women in the example).

May 23, 2006, unreported) in which he suggested that the Adebayo approach was most suitable where there were only two candidates for a position. His Lordship also made some interesting remarks on the limits of the decision in Bahl in its application to "complex claimants". UKPL 2007, Spr, 74-94 END OF DOCUMENT © 2007 Thomson/West. N

68. Or indirect discrimination. 69. D. Réaume, in "Of Pigeonholes and Principles: A Reconsideration of Discrimination Law" (2002) 40 Osgoode Hall Law Journal 113, writes about Ontario Canada, but the point could equally be made about the UK. 70. ibid., p.116. 71. ibid., p.116. 72. ibid., pp.135-136. 73. Respectively [2002] EWCA Civ 1502; [2004] Q.B. 335 and [2003 EWCA Civ 666; [2004] Q.B. 811. See also the recent (obiter) post-Roma "wobble" of various members of the House of Lords in R. (on the application of Gillan) v Commissioner of Police for the Metropolis [2006] UKHL 12; [2006] 2 W.L.R. 537. Note also the reluctance properly to engage with the justifiability of differential treatment manifested by the courts up to and including the House of Lords in R. (on the application of Carson) v Secretary of State for Work and Pensions [2005] UKHL 37; [2005] 2 W.L.R. 1369. 74. On the basis of the split burden of proof required by the Employment and Race and amended Equal Treatment Directives (Council Directives 2000/78, 2000/43 and 2002/73 respectively). 75. This is a complex point and space forbids proper consideration of it here, but it is explored by the author in "Cracking the Comparator Problem: Discrimination, 'Equal' Treatment and the Role of Comparisons" [2006] E.H.R.L.R. 650. 76. The European Network of Independent Experts in the non-discrimination field, (.53 above) report that most Member States have adopted the approach taken by Directives 2000/43 and 2000/78 in choosing not to define the grounds of discrimination protected thereby ("racial or ethnic origin", sexual orientation, disability, religion or belief and age). 77. Hannett, .31 above, citing C. Sheppard, "Grounds of Discrimination: Towards an Inclusive and Contextual Approach" (2001) 80 Canadian Bar Review 893, p.908. 78. See DDA, s.17A(1C); RRA, ss.54A and 57ZA; SDA, ss.63A and 66A, and reg.29 of the Sexual Orientation and Religion or Belief Regulations. Note however that the burden remains on the claimant where the discrimination is on grounds of colour or nationality, or concerns sex, disability or sexual orientation outside the employment sphere. 79. [2005] I.R.L.R. 514. 80. Note the decision of Elias J. in Griffiths-Henry v Network Rail Infrastructure Ltd (EAT, 1

2