1 78 Stat. 253, Pub. L. No. 88–352 (1964), 42 U.S.C. § 2000(e) (1976 & Supp. III 1979). Herein we follow the convention of referring to original sections of the statute rather than to those of the U.S. Code.Google Scholar

2 § 703(a)-(d).Google Scholar

3 This rule of title VII holds good not only when racial discrimination against black people is in volved but also when racial discrimination against white people is at issue. See McDonald v. Santa Fe Trail Transp. Co., 427 U.S. 273 (1976). The per se rule against racial discriminations is qualified by a “benign discrimination” or “affirmative action” exception. See United Steelworkers v. Weber, 443 U.S. 193 (1979). Throughout this paper, when references are made to the per se rule against explicit racial classifications, and more generally when explicit classifications on the basis of race or sex are at issue, we are not to be understood as expressing any views on “benign” discrimination unless the contrary is clearly indicated.Google Scholar

4 See note 21 infra.Google Scholar

5 See Griggs v. Duke Power Co., 401 U.S. 424 (1971). Griggs does not explicate the relationship between “job relatedness” and “business necessity.” The latter concept may encompass a broader class of justifications, since it is possible to conceive of reasons for an employer to engage in a practice, which reasons are justified by strong business interests although not related specifically to an ability to perform a specific task. But see Johnson v. Pike Corp. of America, 332 F. Supp. 490 (CD. Cal. 1971). On the other hand, the idea of “necessity” can be understood to limit the occasions for employer justifications, in that “job related” interests are not sufficient unless they are “necessary.” Cf. Albermarle Paper Co. v. Moody, 422 U.S. 405, 436 (1975). See generally Charles A. Sullivan, Michael J. Zimmer, & Richard F. Richards, Federal Statutory Law of Employment Discrimination 53–56 (Indianapolis: Michie Co., Bobbs-Merrill Co., 1980).Google Scholar

6 § 703(e).Google Scholar

7 See, e.g., U.S. Equal Employment Opportunity Commission, Guidelines on Discrimination Because of Sex (hereinafter EEOC Sex Discrimination Guidelines), 29 C.F.R. § 1604.2 (1980). See also Sirota, Michael, Sex Discrimination: Title VII and the Bona Fide Occupational Qualification, 55 Tex. L. Rev. 1025 (1977); Brilmayer, Lea et al., Sex Discrimination in Employer-sponsored Insurance Plans: A Legal and Demographic Analysis, 47 U. Chi. L. Rev. 505 (1980). See generally Sullivan et al., supra note 5, 137–49. The Supreme Court has expressed the same idea in its opinion in Dothard v. Rawlinson, 433 U.S. 321, 334 (1977), although it is not clear that the result in Dothard is consistent with this statement. See sec. IIC3 infra.Google Scholar

8 Sex Discrimination Guidelines, 29 C.F.R. § 1604.2(a)(2) (1980).Google Scholar

9 See e.g., Brilmayer et al., supra note 7, at 527–29. For a discussion of what “efficiency” means in this context, see sec. IIC infra.Google Scholar

10 Sex Discrimination Guidelines, § 1604.2(a)(1)(iii) (1980).Google Scholar

11 Id. § 1604.2(a)(1)(ii) (1980).Google Scholar

12 Sirota suggests that the proposal to include sex discrimination as a prohibition was made “in jest … in an attempt to make the Bill unacceptable to as many legislators as possible.”Supra note 7, at 1027.Google Scholar

13 Representative Goodell of New York proposed including “sex” in the BFOQ exception, saying:Google Scholar

There are so many instances where the matter of sex is a bona fide occupational qualification. For instance, I think of an elderly woman who wants a female nurse. There are many things of this nature which are bona fide occupational qualifications, and it seems to me they would be properly considered here as an exception.Google Scholar

110. Cong. Rec. 2718 (1964).Google Scholar

The interpretative memorandum of title VII submitted by the floor managers of the bill in the Senate characterized the BFOQ as a “limited right to discriminate.” 110 Cong. Rec. 7213. Although this memorandum has been relied upon in support of the narrowness of the BFOQ provision, see Dothard v. Rawlinson, 433 U.S. 321, 334 n.18 (1977), the example of a bfoq given in the memorandum, “the preference of a professional baseball team for male players,” 110 Cong. Rec. 7213 (1964) is inconsistent with the orthodox view of the narrowness of the exception. See Sirota, supra note 7, at 1029 n.27. Indeed, Sirota, a proponent of an extremely narrow BFOQ, admits that “the meager legislative history of the BFOQ provision indicates that Congress intended it to have broad application in the area of sex discrimination.”Id. at 1026.Google Scholar

14 Dothard v. Rawlinson, 433 U.S. 321, 334 (1977).Google Scholar

15 In this connection, we reject, as question begging, statements like the following:Google Scholar

Congress plainly has decided that sex is more like race (than like age in the respective prohibitions on discrimination which apply to them]. Except for some differences in defenses and a separate section on pregnancy, racial and sexual discrimination are forbidden in identical terms in the same clause of the same statute. Those who argue for different meanings of “discriminate” on the basis of race and on the basis of sex are quarreling with the statute, not interpreting it.Google Scholar

Brilmayer et al., supra note 7, at 536–37 (emphasis added). See also Laycock, Douglas & Sullivan, Teresa A., Sex Discrimination as “Actuarial Equality”: A Rejoinder to Kimball, 1981 A.B.F. Res. J. 221, 223, 224.Google Scholar

It is not all that clear that those who do not share these authors' reading of the statute are the ones who are “quarreling” with it. The assertion that the statute requires that sex and race be placed upon the same footing may or may not be correct. If it is correct, it must be so for some reason other than that this is the policy of the statute: what the statute's policy requires is, precisely, what the inquiry is all about. And in analyzing this question, it might reasonably be thought, differences in available defenses are not the stuff of trifling exceptions but are essential aspects of the statutory scheme.Google Scholar

16 See sec. IIB1 infra.Google Scholar

17 See sec. IIB2 infra.Google Scholar

18 See sec. IIB3 infra.Google Scholar

19 See sec. IIC1 infra.Google Scholar

20 See sec. IIC2 infra.Google Scholar

21 See sec. HE infra. The “disparate treatment” concept of discrimination pertains to intentional discrimination on the basis of sex. Cases in this class may involve the employer's explicit drawing of prohibited lines. See City of Los Angeles Department of Water and Power v. Manhart, 435 U.S. 702 (1978). More commonly the principal issue will be whether an employer acted on the impermissible basis of an employee's race, sex, religion, or national origin. In this latter sort of case, the underlying question is one of an employer's “discriminatory motive.” An inference of such a motive can be drawn from the circumstantial evidence surrounding a particular employment decision, see McDonnell Douglas Corp. v. Green, 411 U.S. 792 (1973), or it can be drawn from disparities between an employer's work force and the labor pool from which this work force is drawn. See International Bhd. of Teamsters v. United States, 431 U.S. 324 (1977); Hazelwood School Dist. v. United States, 433 U.S. 299 (1977). In the “disproportionate impact” concept of discrimination, the employer's motive is irrelevant. If a practice, “neutral on its face,” results in consequences that disproportionately fall on persons of a particular race, sex, religion, or national origin, such a practice is unlawful, irrespective of the employer's motive, unless the practice is demonstrably related to job performance or “business necessity.” See note 5 supra. See Griggs v. Duke Power Co., 401 U.S. 424 (1971); Dothard v. Rawlinson, 433 U.S. 421 (1977). See generally Sullivan et al., supra note 5, at 1–90.Google Scholar

22 435 U.S. 702 (1978).Google Scholar

23 See, e.g., Dothard v. Rawlinson, 433 U.S. 321, 333 (1977); Sex Discrimination Guidelines, 29 C.F.R. § 1604.2(a)(1)(ii) (1980); Sirota, supra note 7, at 1033; Taub, Nadine, Keeping Women in Their Place: Stereotyping Per Se as a Form of Employment Discrimination, 21 B.C.L. Rev. 345, 353–61 (1980).Google Scholar

24 Schlesinger v. Ballard, 419 U.S. 498, 508 (1975).Google Scholar

25 One author apparently concedes that the BFOQ exception allows “generalizations”—that is its “sole purpose.” However, the author appears to believe that only those “generalizations” that are true in 100 percent of the cases qualify for the BFOQ. See Sirota, supra note 7, at 1048. This is an unusual concept of generalization; as we shall show infra, not even the example this author uses (the “authenticity or genuineness”BFOQ for actor or actress) would qualify under this definition of generalization.Google Scholar

26 The Sex Discrimination Guidelines acknowledge that employers are required by some states to provide separate rest rooms for employees of each sex. The guidelines provide, however, that such statutes are no excuse for employers to exclude members of one sex, no matter what expense might be entailed by cor§ance with the ban on unisex lavatory facilities. For example, an employer whose work force had hitherto been only of one sex would be required to provide separate but equal lavatories regardless of expense, rather than being allowed to exclude a single otherwise qualified member of the sex for whom such facilities were not already in place. See Sex Discrimination Guidelines, 29 C.F.R. § 1604.2(b)(5) (1980). The current EEOC approach to this issue is less permissive than is its earlier position. Initially, the commission had expressed the view that the costs associated with providing separate facilities for men and women (e.g., lavatories, shipboard berths) could, at least in situations where very great costs would be entailed, justify a finding that sex was a BFOQ. See Sirota, supra note 7, at 1053–54. But the current EEOC position does defer to state laws that require “separate but equal” facilities for men and women—in contrast to the position taken with respect to “state protective laws,” e.g., laws prohibiting women from working certain hours that men may work or from performing certain jobs that are permitted to men. Such laws “conflict with and are superseded by title VII.” See Sex Discrimination Guidelines, 29 C.F.R. 1604.2 (b)(1) (1980); Rosenfeld v. Southern Pac. Co., 444 F.2d 1219 (9th Cir. 1971) (weight-lifting restrictions on female employees); Kober v. Westinghouse Elec. Corp., 325 F. Supp. 467 (W.D. Pa. 1971), aff'd, 480 F.2d 240 (3d Cir. 1973) (hours of work). See generally Sullivan et al., supra note 5, at 213–15. There can be little question that, were they not unconstitutional in any case, state laws that required “separate but equal” facilities for members of different races would “conflict with” and be “superseded by title VII.”.Google Scholar

27 Brilmayer et al., supra note 7, at 537 n.152, citing Brown v. Board of Educ., 347 U.S. 483 (1954), and John Hart Ely, Democracy and Distrust: A Theory of Judicial Review 150 (Cambridge, Mass.: Harvard University Press, 1980).Google Scholar

28 Richard Wasserstrom proposes that race and sex segregation of lavatories are essentially different. Race segregation, he argues, is premised on the prevention of contamination by blacks, the segregated class, of whites. Sex segregation is different, he suggests, in that it carries no implication of contamination but rather includes the mystique of sex privacy as a technique, one among many, “to maintain the primacy of heterosexual sexual attraction central to that version of the patriarchal system of power relationships we have today.”Wasserstrom, Richard A., Racism, Sexism, and Preferential Treatment: An Approach to the Topic, 24 UCLA L. Rev. 581, 594 (1977). We suggest that the distinction Wasserstrom points to is a bit indistinct. Indeed, the idea of “contamination” in the race context is not a clear one and probably diverges to some extent from the ordinary meaning of the word. In the race context, the idea of contamination is probably best understood as kindred in sense to the use of the term “taint” in the seminal article of Black, Charles L. Jr., The Lawfulness of the Segregation Decisions, 69 Yale L. J. 421, 426 (1960)—a notion that refers metaphorically to status definition rather than to any literal contagion. Lavatories are, after all, intrinsically democratic places for the common denominator of purpose that unifies those who repair to them. But the equality of patrons that lavatories imply is sometimes denied in social context: e.g., officers and enlisted personnel do not use the same latrines, and—closer to home—faculty and student lavatories are sometimes segregated even in law schools and always, to our knowledge, in primary and secondary schools. It would be difficult if not impossible to demonstrate specific harms flowing from lavatories segregated by race. And indeed, neither the language nor the logic of Brown v. Board of Educ., 347 U.S. 483 (1954), would apply to lavatories at all unless the harm identified in Brown is taken, as realistically it must be and has been, to find a per se social harm in a law and custom which contributes—even if only in a small way—some quantum of affirmation to a larger norm which the policy of the Constitution forbids. The norm that the Constitution uproots declares that blacks and whites are different and should be held separate from one another. The “contamination” associated with race-integrated lavatories is precisely the contamination invidiously associated with race mixing generally or even with assertions of social equality between the races. The sex segregation of lavatories does not, on this analysis, appear very different. Wasser-strom concedes that “patriarchal system[s] of power relationships” do imply status hierarchy; and he argues quite explicitly that reinforcing the notion that men and women are relevantly different for purposes of bathroom privacy does contribute to maintaining those norms. Why this is not “taint” in Black's sense is not apparent.Google Scholar

29 Indeed, this is precisely the basis on which rests the argument that sex and race classifications are to be considered on the same footing.Google Scholar

30 See Janet Saltzman Chafetz, Masculine, Feminine or Human? An Overview of the Sociology of Gender Roles 68–113, esp. 88–89 (2d ed. Itasca, Ill.: F. E. Peacock Publishers, 1978).Google Scholar

31 Brown, Barbara A. et al., The Equal Rights Amendment: A Constitutional Basis for Equal Rights for Women, 80 Yale L.J. 871, 901–2 (1971).Google Scholar

32 Chafetz, supra note 30, at 130; Wall Street J., July 10, 1980, at 19 (Linda Batway, executive director of Womenspace, on the dismissal of Jane Cahill Pfeiffer as chair of the board of National Broadcasting Co.: “We don't use the same locker room, so we're often excluded from the power and the decision making”).Google Scholar

33 Toward Job Equality for Women 5 (1969), cited in Sirota, supra note 7, at 1060 n.212.Google Scholar

34 eeoc Decision No. 71–2410, [1973] eeoc Compl. Man. (CCH) § 6282, at 4500 (1971) (privacy needs of female nursing home patients could be met through assignment practices).Google Scholar

35 Three federal cases have considered the issue of sexual privacy in the retail clothing business. In Roberts v. Union Co., 487 F.2d 387 (6th Cir. 1973), the employer was charged with sex discrimination for failure to hire women in its men's clothing department. The court of appeals remanded for a determination whether sex was a BFOQ given the employer's claim that its customers would be embarrassed by having women fit their clothes, as this task necessitated touching intimate parts of a customer's body. Id. at 388. The other cases involved challenges to pay differences for salesmen and saleswomen as violative of the Equal Pay Act. The courts determined in each case that sex was a BFOQ for the job of salesperson, because physical contact by a salesperson of the opposite sex would embarrass the customer and inhibit sales. Hodgson v. Robert Hall Clothes, Inc., 326 F. Supp. 1264, 1269 (D. Del. 1971); Brennan v. Collin & Williams for Men & Boys, 8 Empl. Prac. Dec. § 9816, at 6427 (W.D. Ark. 1974). The Robert Hall opinion suggests the importance of privacy-related duties to the job:. Often the salesperson is required to assist with opening zippers; to touch the body of a customer near private parts in connection with the measurement of the crotch, seat, waist, chest or inseam; to touch other areas of the body while assisting in the try-on of a garment, and to observe the customer in various stages of undress in connection with try-ons.Google Scholar

326 F. Supp. at 1269.Google Scholar

36 Ludtke v. Kuhn 461 F. Supp. 86 (S.D.N.Y. 1978).CrossRefGoogle Scholar

37 Id. at 98. However, this premise was evidently mistaken; players deprived of privacy at their assigned lockers would be required to dress and undress in the shower rooms proper, or in the trainer's room, an inconvenience that only the most modest Yankees would endure. See Sparky Lyle & Peter Golenbock, The Bronx Zoo 289 (New York: Dell Publishing Co., 1980). A few days after the Ludtke decision, according to Sparky Lyle, “there must have been 30 women reporters in the locker room because every newspaper and TV station decided it would be great to send a woman to cover the Yankees…. [T]he media decided to make it a circus, so they sent the first woman they could find.”Id. at 291.Google Scholar

38 See, e.g., Reynolds v. Wise, 375 F. Supp. 145 (N.D. Tex. 1974) (sex a BFOQ for prison work assignment but not for hiring); Forts v. Ward, 434 F. Supp. 946 (S.D.N.Y. 1977) (preliminary injunction granted to female prison inmates protesting assignment of male corrections officers to duties in housing and hospital units); Fesel v. Masonic Home, 447 F. Supp. 1346 (D. Del. 1978), aff'd, 591 F.2d 1334 (3d Cir. 1979) (BFOQ for sex in hiring where small nursing home could not accommodate privacy needs of female residents through juggling job assignments). See also Carey v. New York State Human Rights Appeal Bd., 61 A.D. 2d 804, 402 N.Y.S.2d 207 (1978) (sex a BFOQ under state human rights law for corrections officer position in a women's prison); City of Philadelphia v. Pennsylvania Human Relations Comm'n, 7 Pa. Commw. Ct. 500, 300 A.2d 97 (1973) (sex a BFOQ under state law for position of youth center supervisor; both “rehabilitative” and privacy reasons were relied upon); Sterling v. Cupp, 22 Empl. Prac. Dec. § 30,797 (Or. Ct. App.) (upholding male inmate's challenge to assignment of female corrections personnel to duties involving frisks). In some of these cases it is not the employer who is asserting that sex is a BFOQ but rather an inmate or resident asserting a privacy right which requires that persons of the other sex be excluded. There may be other situations as well in which privacy concerns will have ramifications for the employment opportunities of people of one or the other sex. The routine police practice of “strip-searching” people following their arrest is sensitive partly because of the claim for intersexual privacy. See, e.g., Ill. Rev. Stat. ch. 38, § 103–1(e) (Cum. Supp. 1980): “(e) All strip searches conducted under this Section shall be performed by persons of the same sex as the arrested person and on premises where the search cannot be observed by persons not physically conducting the search.” Very similar concerns might influence the employment—and deployment—of police officers. When a police officer observes a suspicious-looking character on the street, there may be grounds for further inquiry, even though there is no probable cause upon which to base an arrest. The officer may hail the subject and question him, and is entitled during this interview to assurance that the subject is unarmed. In the resulting frisk, “the officer must feel with sensitive fingers every portion of the prisoner's body. A thorough search must be made of the prisoner's arms and armpits, waistline and back, the groin and area about the testicles, and entire surface of the legs down to the feet.” Terry v. Ohio, 392 U.S. 1, 17 n.13 (1968) (quoting Priar, & Martin, , Searching and Disarming Criminals, 45 J. Crim. L.C. & P.S. 481 (1954)). We are unaware of any cases specifically raising this issue, but a potential difficulty is clearly there, and it is rooted in the modesty claim.Google Scholar

39 Chafetz, supra note 30, 21–25, 99.Google Scholar

40 Furthermore, it could well be argued that abandoning the custom of sex segregation—at least in public lavatories—would not, after all, significantly diminish the privacy of people using public lavatories. Casually imagining a public lavatory in the abstract may fail to evoke just how public it can be. It would be better, instead, to remember—for example—the public rest room at the head of the K Gates in O'Hare Airport some busy Sunday afternoon. The lavatory—we speak only for the men's room now—although large, is jam-packed with, perhaps, 100 men and boys, salesmen, policemen, knots of students, a group of safety patrols from Missouri, milling in and out ad libitum. There is no more privacy than on the floor of a stock exchange. Anyone and everyone can walk through the door—Saul Bellow, Tiny Tim, the Vienna Boys' Choir—so long as she or they are not female. It is difficult to rationalize a privacy interest that is triggered when a man must share an enormous public lavatory with one, lone female but not triggered when he must share the facilities with the entire National Football League.Google Scholar

41 Willingham v. Macon Tel. Publishing Co., 507 F.2d 1084 (5th Cir. 1975) (en bane), vacating 482 F.2d 535 (5th Cir. 1973); Fagan v. National Cash Register Co., 481 F.2d 1115 (D.C. Cir. 1973); Knott v. Missouri P.R.R. Co., 527 F.2d 1249 (8th Cir. 1975); Dodge v. Giant Food, Inc., 488 F.2d 1333 (D.C. Cir. 1973); Baker v. California Land Title Co., 507 F.2d 895 (9th Cir. 1974); Earwood v. Continental Southeastern Lines, Inc., 539 F.2d 1349 (4th Cir. 1976); Barker v. Taft Broadcasting Co., 549 F.2d 400 (6th Cir. 1977); Longo v. Carlisle DeCoppet & Co., 537 F.2d 685 (2d Cir. 1976).Google Scholar

42 Willingham v. Macon Tel. Publishing Co., 507 F.2d at 1091.Google Scholar

43 E.g., in Allen v. Lovejoy 553 F.2d 522 (1977) the Sixth Circuit held that an employer could not require women to use their husbands' last names, though names are not “immutable.” Similarly, in Sprogis v. United Air Lines, 444 F.2d 1194, cert, denied, 404 U.S. 991 (1971), the Seventh Circuit held that an employer could not require female flight attendants to be unmarried, though marital status is also not immutable.Google Scholar

44 Willingham v. Macon Tel. Publishing Co., 507 F.2d 1084, 1091.Google Scholar

45 George Rutherglen, Sexual Equality in Fringe-Benefit Plans, 65 Va. L. Rev. 199, 206–9 (1979).Google Scholar

46 Fountain v. Safeway Stores, Inc., 555 F.2d 753 (9th Cir. 1977).Google Scholar

47 E.g., in Carroll v. Talman Fed. Sav. & Loan Ass'n, 604 F.2d 1028 (7th Cir. 1979), cert, denied, 445 U.S. 929 (1980), an employer was held to have unlawfully discriminated by allowing men to wear “appropriate business attire” while requiring women to wear a prescribed uniform. An item from the Wall Street Journal arrived as this article was going to press and is worth quoting in full:. A nursing home told a male nurse he couldn't keep wearing a female nurse's uniform while he was awaiting a sex-change operation that had been postponed for three years. An arbitrator said the nurse had to be rehired if he agreed to wear the proper uniform until he became a she. Wall Street J., July 14, 1981, at 1, col. 1.Google Scholar

48 See Fagan v. National Cash Register Co., 481 F.2d 115, 1124–25 (D.C. Cir. 1973):. Perhaps no facet of business life is more important than a company's place in public estimation. That the image created by its employees dealing with the public when on company assignment affects its relations is so well known that we may take judicial notice of an employer's proper desire to achieve favorable acceptance. Good grooming regulations reflect a company's policy in our highly competitive business environment.Google Scholar

49 Sex Discrimination Guidelines, 29 C.F.R. 1604.2(a)(1)(iii) (1980).Google Scholar

50 The “taste for discrimination” may be understood as a person's preference not to associate with members of some other group. In order to indulge a taste for discrimination, a person may be willing to suffer pecuniary costs if it will allow him to avoid the intangible costs flowing from the undesired contact. See Richard A. Posner, Economic Analysis of Law 525 (2d ed. Boston: Little, Brown & Co., 1977).Google Scholar

51 See St. Cross v. Playboy Club, App. No. 773, State Human Rights Appeal Bd. (N.Y. 1971); Chamberlain v. Indian Valley Realty Corp., App. No. 743, Case No. CS-21209-70, N.Y. Human Rights Appeal Bd., aff'd per curiam, 38 A.D.2d 890 (1970).Google Scholar

52 Sirota, supra note 7, at 1067.Google Scholar

53 EEOC, Toward Job Equality for Women 5 (1969), quoted in id. at 1066.Google Scholar

54 442 F.2d 385, 388 (5th Cir.), cert. denied, 404 U.S. 950 (1971).Google Scholar

55 In Diaz, the airline claimed that women possessed psychological and personality attributes better suited to fulfill the important function of creating a comfortable and secure atmosphere for passengers. The trial court had accepted the airline's evidence on the “psychological” submission and had found a BFOQ for sex on that basis. Diaz v. Pan American World Airways, 311 F. Supp. 559, 566 (S.D. Fla. 1970). The court of appeals neither accepted nor rejected the district court's factual finding. Rather, it held that even if women had superior “psychological” credentials to be flight attendants, there could be no BFOQ because the BFOQ provision imposes a “business necessity” test, requiring that any asserted job qualification go to the “essence” of the employer's business. 442 F.2d 385, 388–89. It is difficult to escape the feeling that the airline's motivation in Diaz had to do with the sex appeal of stewardesses; its case on the psychological facts, even though the district court credited it, seems almost patently ridiculous.Google Scholar

56 Cf. Guardian Capital Corp. v. N.Y. State Div. of Human Rights, 46 A.D.2d 832, 360 N.Y.S.2d 937 (1974); appeal dismissed, 36 N.Y.2d 806, 369 N.Y.S.2d 1027 (1975) (restaurant that fired waiters and replaced them with seductively dressed waitresses could not claim a BFOQ without showing a direct connection between employee sex appeal and economic benefit).Google Scholar

57 A case somewhat similar to our hypothetical was recently decided by the United States District Court for the Northern District of Texas. Southwest Airlines, the nation's most profitable air carrier in recent quarters, claimed a BFOQ for sex, not only for flight attendants but also for ticketing clerks. Southwest heavily promoted itself as the goatish businessman's answer to intrastate travel: it dressed its personnel in hotpants, styled its in-flight snacks “love bites,” and cultivated the overall ambiance of a high-class massage parlor. According to Howard Putnam, the airline's president, these sexual overtones deserved considerable credit for the airline's outstanding performance in a generally lagging industry. See Wall Street J., June 18, 1981, at 25, col. 4. The court rejected this claim as “speculative at best” and threw out the BFOQ claim. Wilson v. Southwest Airlines Co., Civil Action No. CA-3-80-0680-G, slip op. at 21 (N.D. Tex., filed June 18, 1981). The case differs from our hypothetical in at least one crucial respect: Southwest has a monopoly on service between Love Field and Hobby Field, the in-town airports of Dallas and Houston. It might be expected that a monopoly like that would earn a profit for any airline, irrespective of its marketing tactics. However, this monopoly will end soon, if it has not already. Muse Air, a new entrant, will also offer service between Love Field and Hobby Field but without sexual overtones. See Wall Street J., June 18, 1981, at 25, col. 6. The existence of this competition should test whether Southwest's “love” campaign plays a substantial part in the airline's success.Google Scholar

58 See Sirota, supra note 7, at 1026 n.8; Arthur Larson, 1 Employment Discrimination § 14.10 (New York: Matthew Bender, 1980); but see note 84 infra.Google Scholar

59 Under title VII, of course, the choice of any narrowing criterion is subject to “disproportionate impact” analysis. If the criterion falls more heavily on one group along prohibited lines such as race or sex, it must be justified by “business necessity.” See notes 5 and 21 supra and text at notes 121–22 infra.Google Scholar

60 One economic explanation for many types of discrimination is the supposed utility of a prohibited decision criterion (sex, race, etc.) in reducing information costs. See Kenneth Arrow, The Theory of Discrimination, in Orley Ashenfelter & Albert Rees, eds., Discrimination in Labor Markets 25–32 (Princeton, N.J.: Princeton University Press, 1973); Rutherglen, supra note 45, at 252–53.Google Scholar

61 For our purposes, this decision criterion is “perfect,” even though we recognize two qualifications on its perfection. First, it will not always be true that applicants who cannot lift 200 kilos during a tryout are incapable of lifting it in other circumstances. Second, it will not always be true that those capable of lifting the weight now will always be capable of lifting it at all times in the future. These imperfections would appear to be present for most, if not all, decision criteria; thus we treat them as constants for purposes of our analysis.Google Scholar

62 Brilmayer et al., supra note 7, at 509–11, 526–29; Larson, supra note 58, at §§ 1600, 1700. Sex Discrimination Guidelines, 29 C.F.R. § 1604.2(a)(1)(i), (ii) (1980):. (i) The refusal to hire a woman because of her sex based on assumptions of the comparative employment characteristics of women in general. For example, the assumption that the turnover rate among women is higher than among men. (ii) The refusal to hire an individual based on stereotyped characterizations of the sexes. Such stereotypes include, for example, that men are less capable of assembling intricate equipment; that women are less capable of aggressive salesmanship. The principle of nondiscrimination requires that individuals be considered on the basis of individual capacities and not on the basis of any characteristics generally attributed to the group. These guidelines are consistent with the orthodox view if they are taken to preclude the use of any generalizations about differences between the sexes, no matter how strong their empirical basis. The statement that individualized testing is always required suggests that the commission does indeed take the orthodox view. For our view that the possibility of individualized testing is itself a part of the efficiency inquiry, see text at pp. 607–9 infra.Google Scholar

63 400 U.S. 542 (1971) (per curiam).Google Scholar

64 Id. at 544.Google Scholar

65 Id. at 545.Google Scholar

66 433 U.S. 321 (1977).Google Scholar

67 Id. at 334.Google Scholar

68 See Sirota, supra note 7, at 1047 n.133; Sullivan et al., supra note 5, at 145.Google Scholar

69 Or age—a number of closely related cases deal with age discrimination, which, like sex discrimination, is a subject in which the BFOQ exception plays an important part. See Sullivan et al., supra note 5, at 144, for the proposition that a unitary standard will be applied to age and sex cases. Contrast the Brilmayer authors' contrary view, supra note 7, at 536.Google Scholar

70 As we discuss infra, formulation of this factor is misleading. What is important is not the absolute correlation between sex and a required trait; it is the magnitude of the difference between male and female populations respecting the possession of the trait.Google Scholar

71 Weeks v. Southern Bell Tel. & Tel. Co., 408 F.2d 228 (5th Cir. 1969).Google Scholar

72 Phillips v. Martin Marietta Corp., 400 U.S. 542, at 544 (1971).Google Scholar

73 433 U.S. 321 (1977).Google Scholar

74 When courts or commentators refer to an employer's ability to test for a given trait, what is usually meant is the use of a sex-neutral decision criterion as a means of securing information about the trait. It is important to recognize, however, that making decisions without regard to the trait is a form of “individualized testing” too—an employer who hires without regard to the trait will secure perfect information about the trait by observing how any given employee actually turns out. Individualized testing, then, is never literally impossible but only more or less costly. Individual testing by means of a decision criterion may be impossible—there may be no instrument available to predict for the trait—but the option of seeing how things turn out in practice is a form of individualized testing that always exists.Google Scholar

75 It is a stock answer, when classifications are made along sex lines, that the employer is entirely free to insist on a certain job qualification by describing it in a gender-neutral way. For example, in Phillips v. Martin Marietta Corp., 400 U.S. 542 (1971), Justice Marshall suggested that an employer could certainly choose its employees based on whether an applicant had made satisfactory child care arrangements, thus (partly) assuring consistent attendance at work. 400 U.S. at 544–45 (concurring opinion of Marshall, J.). The Brilmayer authors make a similar point when they note that title VII, while prohibiting an employer from using the (female) sex of a prospective employee as a surrogate for whether she could repeatedly lift heavy weights (but see discussion in text supra), nevertheless would allow an employer to exclude from consideration every applicant who could not lift the requisite weight—even though this gender-neutral way of proceeding would exclude women disproportionately versus men. Brilmayer et al., supra note 7, at 510. Such a view rejects the notion of cost as relevant to the choice of decision criteria; furthermore, it presupposes the existence of another criterion in all cases. This view essentially wishes away hard problems that arise in one of two circumstances: (1) where a sex-neutral decision criterion carries unacceptable risk costs when compared to the use of sex directly, and (2) where the only alternative to sex as a decision criterion would be hiring (or making other decisions) without regard to the sought-for trait. The EEOC'S position similarly presupposes the existence of alternative decision criteria: “The principle of nondiscrimination requires that individuals be considered on the basis of individual capacities and not on the basis of any characteristics generally attributed to the group.” Sex Disorimination Guidelines, 29 C.F.R. § 1604.2(a)(1)(ii) (1980).Google Scholar

76 Another aspect of “importance to the employer” is that the trait relate to the “essence” of the business. See Diaz v. Pan American World Airways, Inc., 442 F.2d 385, 388 (5th Cir. 1971). See sec. IIB3 supra.Google Scholar

77 408 F.2d at 235 n.5. See also Sullivan et al., supra note 5, § 2.4, at 140.Google Scholar

78 Hodgson v. Greyhound Lines, Inc., 499 F.2d 859 (7th Cir. 1974), cert. denied, 419 U.S. 1122 (1975) (bus driver); Usery v. Tamiami Trail Tours, 531 F.2d 224 (5th Cir. 1976) (bus driver); Harriss v. Pan American World Airways, Inc., 649 F.2d 670 (9th Cir. 1980) (flight attendant). In the Tamiami Trail Tours case, the court applied the Weeks-Diaz standard for a BFOQ, suggesting that the term has the same meaning in both title VII and in the Age Discrimination in Employment Act. In the Harriss case, the Ninth Circuit also applied the Weeks-Diaz rule, thereby apparently eliminating a supposed division among the circuits between that rule and a narrower rule of “authenticity or genuineness.” See Rosenfeld v. Southern Pac. Co., 444 F.2d 1219, 1225 (9th Cir. 1971), and discussion of this division in Sullivan et al., supra note 5, § 2.4, at 139.Google Scholar

79 In Tamiami Trail Tours, the court stated the Weeks test as allowing a BFOQ if the employer can demonstrate either (1) that there is a factual basis for believing that all or substantially all women or older workers cannot perform the duties of the job safely and efficiently; or (2) that it is impossible or impractical to test on an individual basis. See 531 F.2d at 235. Moreover, “[t]he greater the safety factor, measured by the likelihood of harm and the probable severity of harm in case of an accident, the more stringent may be job qualifications designed to insure safe driving.” Id. at 236. See also Arritt v. Grisell, 567 F.2d 1267, 1271 (4th Cir. 1977); In re National Airlines, Inc., 434 F. Supp. 249, 262 (S.D.Fla. 1977) (adopting Tamiami Trail Tours version of BFOQ test). The Harriss court applied this analysis to the BFOQ for sex as well. 649 F.2d 670 (9th Cir. 1981). As with the Diaz case, however (see note 55 supra), it is hard to refrain from speculating that the airline's actual concern was not safety so much as sex appeal.Google Scholar

80 Sex Discrimination Guidelines, 29 C.F.R. § 1604.2(a)(2) (1980). See also Sirota, supra note 7, at 1059–60; Sullivan et al., supra note 5 § 2.4, at 148.Google Scholar

81 See, e.g., G.B. Harrison, ed., Shakespeare: The Complete Works 59–60 (New York: Harcourt Brace Jovanovich, Inc., 1968).Google Scholar

82 This same fact also pertains to police decoys, see Button v. Rockefeller, 76 Misc. 2d 701, 351 N.Y.S.2d 488 (Sup. Ct. 1973). This is apparently the only case so far decided on grounds of “authenticity or genuineness.”.Google Scholar

83 Such a judgment was evidently reflected by Chicago Mun. Code § 192–8, which provides criminal penalties for “[a]ny person who shall appear in a public place … in a dress not belonging to his or her sex, with intent to conceal his or her sex.” This ordinance was recently held unconstitutional by the Illinois Supreme Court as violative of the unspecified right to the private choice of a harmless life style. City of Chicago v. Wilson, 75 Ill. 2d 525, 389 N.E.2d 522 (1978).Google Scholar

84 It is conventionally said that a BFOQ limited to the above-described situation only allows sex classifications where intrinsic sex differences are of concern to the employer. Manifestly this response does not explain the BFOQ for actors and actresses, because the mere presence of a Y chromosome does not, in itself, render a performer unable to portray Desdemona. The same observation holds good, we were startled to learn, for another stock example of “authenticity or genuineness”—limiting the job of wet nurse to females. See Hormone-taking Man Breast-fed Baby, Doctor Says, Chicago Sun-Times, Feb. 20, 1980, at 27, col. 1. The subject of the story was a man who wanted an equal share in the rearing of his newborn daughter. The man's physician, Dr. Leo Woll-man, of Brooklyn, N.Y., injected the patient with pitocin, a pituitary hormone, to induce lactation. There appear to be no other reported instances of deliberately induced lactation in normal male human beings, but there are at least a few reports in the medical literature of induced lactation in male rats. See, e.g., Neumann, F., Elger, W., & Berswordt-Wallrabe, R., The Structure of the Mammary Glands and Lactogenesis in Feminized Male Rats, 36 J. Endocrinol. 353 (1966); Berswordt-Wallrabe, R. & Neumann, F., Reactivity of Feminized Male Rats to Progesterone/Estrogen Administration and Induction of Lactation, 204 Arch. Gynaekol. 207 (1967); Arai, Y. & Suzuki, Y., Biphasic Lactogenic Response of Male Rat Mammary Glands After a Single Injection of Reserpine, 50 J. Endocrinol. 697 (1971); Arai, Y., Kubokura, A., Suzuki, Y., & Masuda, S., Initiation of Milk Secretion in the Male Rat by Single Injections of Reserpine, 17 Endocrinol. Jap. 441 (1970). Assuming the Brooklyn report to be credible, a hospital would be allowed to limit its search for wet nurses to women only if courts recognized that employers were entitled to make the value judgment that babies should not be nursed by men or that employers were entitled to predict that so few men will be willing to undergo the necessary course of hormone injections that it would be a waste of time for the employer to seek out or entertain applications from men. It is important at this point to appreciate the contrast, where efficiency ideas are implicated, between the treatment of race and sex by title VII. The question of a parallel BFOQ for race—say, for example, by limiting the casting of the part of Othello to black actors—was explicitly considered and rejected by Congress in the debates surrounding the original enactment of title VII. Rather than considering only blacks, according to these debates, the producer must consider men who appear to be black: “[A] director of a play or movie who wished to cast an actor in the role of a Negro, could specify that he wished to hire someone with the physical appearance of a Negro—but such a person might actually be a non-Negro.” 110 Cong. Rec. 7217 (1964) (statement of Senator Clark). For a discussion of possible reasons why sex and race have been treated differently, see sec. IID infra.Google Scholar

85 As distinct from hazards that arise as a consequence of these functions. See Harriss v. Pan American World Airways, Inc., 649 F.2d 670 (9th Cir. 1981).Google Scholar

86 See generally Lynne Darcy, Birth Defects Caused by Parental Exposure to Workplace Hazards: The Interface of Title VII with OSHA and Tort Law, 12 J.L. Ref. 237, 238–41 (1979).Google Scholar

87 Id. at 239.Google Scholar

88 Id. at 240 and sources cited at n.20.Google Scholar

89 Id. at 238.Google Scholar

90 Id. at 240.Google Scholar

91 Proposed, 45 Fed Reg. 7514 (1980), withdrawn 46 Fed. Reg. 3916 (1981) (hereinafter Reproductive Hazards Guidelines).Google Scholar

92 There was no explicit statement to this effect in the guidelines themselves, but the commission's Explanatory Note indicated that this was its intent. See 45 Fed. Reg. 7515 (1980). The proposed guidelines recognized that not all reproductive hazards can be eliminated by employers. The possibility of eliminating hazards “where feasible” was relevant to the evaluation of such cases. Reproductive Hazards Guidelines § 2(d)(8), 45 Fed. Reg. 7517 (1980). This is in contrast to the claims of some that the cost of eliminating hazards is irrelevant. See Sirota, supra note 7, at 1059. And it must be recognized that some sorts of hazards probably cannot be eliminated without abandoning the enterprise entirely, e.g., lead mines. Dicta in one case suggests that the secretary of labor may actually have this power under OSHA. AFL-CIO V. Brennan, 530 F.2d 109, 121 (3d Cir. 1975). Cf. American Textile Mfrs. Inst., Inc. v. Donovan, 452 U.S.—, 49 U.S.L.W. 4720 (1981).Google Scholar

93 Reproductive Hazards Guidelines § 2 (d)(6), 45 Fed. Reg. 7516 (1980).Google Scholar

94 Arbitration of Firestone T. & R. Co. and Amalgamated Local No. 336, United Rubber Workers (Schorn grievance) (decided Aug. 14, 1977), slip op. at 2–3 n.1.Google Scholar

95 Id.Google Scholar

96 A woman who chose to work in an environment that subjected her to reproductive hazards would presumably “assume the risk” and be deemed to have waived any tort claims on her own behalf. William L. Prosser, Handbook of the Law of Torts 440 (4th ed. St. Paul: West Publishing Co., 1971). But there is a substantial question whether a woman could waive the tort claim of her unborn child. See Darcy, supra note 86, at 253. Thus, arguments that the exclusion of women subject to fetal risk is paternalistic miss the point, because such arguments assume that waiver of tort claims by a mother-to-be will adequately safeguard the employer's interest. See, e.g., Sirota, supra note 7, at 1059. There is no doubt that personal injury actions are available in some jurisdictions to children for prenatal injuries suffered by them. See, e.g., Evans v. Olson, 550 P.2d 924 (Okla. 1976); Sylvia v. Gobeille, 101 R.I. 76, 220 A.2d 222 (1966); Smith v. Brennan, 31 N.J. 353, 157 A.2d 497 (1960). There are substantial barriers to recovery, such as a requirement that the injury occur when the fetus is viable (Evans v. Olson, at 926 n.1), that the employer has acted negligently (unless strict liability is imposed), and that causation be shown. See discussion in Darcy, supra note 86, at 253–57. To say that the law in this area is unsettled, however, does not diminish the employer's legitimate concern. Indeed, the uncertainty may exacerbate the concern by making planning difficult.Google Scholar

97 An employer might have another reason, too, for excluding “persons who might be pregnant,” which would ordinarily escape notice because it would almost always coincide with fears about tort liability. An employer might declare it as company policy to safeguard the welfare of fetal life. Whether an employer should be entitled to let anything turn on the firm's attitudes toward fetal life, of course, is not a short subject and is fraught with implications for first principles that deserve extended reflection. An argument that employers' policies should be limited to looking after its own profits and not the spiritual welfare of the community including those yet en ventre sa mere will run up against the current corporate law orthodoxy that sees the legitimate sphere of corporate influence and concern to run far beyond the shareholders' bottom line. Thornton Bradshaw & David Vogel, eds., Corporations and Their Critics (New York: McGraw Hill Book Co., 1980). But if the enactment of laws against sex discrimination means anything, it precludes employers' imposing their values about sex roles on their employees. See sec. IV infra. In a sense, to attempt to distinguish the employer's concern with fetal life from its other value judgments is merely to recapitulate the debate between the “prochoice” camp and the “prolife” camp on the issue of abortion. If the “fetal life” reason is deemed illegitimate, it would be open to aggrieved women employees to attempt to show that the employer's proffered efficiency reason for excluding them was merely a “pretext” for this illegitimate reason. See McDonnell Douglas Corp. v. Green, 411 U.S. 792 (1973).Google Scholar

98 See note 91 supra. However, the commission has yet to share with the public the principles that it proposed to apply on the “case by case” basis.Google Scholar

99 Reproductive Hazards Guidelines, Explanatory Note, 45 Fed. Reg. 7516 (1980).Google Scholar

100 § 703(e).Google Scholar

101 See Sirota, supra note 7, at 1052–54, 1059; Sullivan et al., supra note 5, § 1.5(b), at 43–44; Brilmayer et al., supra note 7, at 518–29; Rutherglen, supra note 45, at 248–50; Larson, supra note 58, § 1700.Google Scholar

102 A BFOQ broadened to include issues relevant to hiring still would probably not include the sort of efficiency justification offered by the employer in City of Los Angeles Dep't of Watej & Power v. Manhart. See discussion infra. Thus, in order to accept the argument that sex-segregated actuarial tables are permitted under title VII, it might be necessary to recognize that title VII includes an implied efficiency justification beyond the limits of the BFOQ. The EEOC apparently recognized such a notion in its proposed treatment of the reproductive hazards issue; if such an implied justification is rejected, overruling Manhart would require an amendment to the statute.Google Scholar

103 See text at note 50 supra.Google Scholar

104 See sec. IIB3 supra.Google Scholar

105 See discussion of “stereotype,” sec. IIA supra.Google Scholar

106 For a discussion of a third constraint, see sec. IV infra.Google Scholar

107 See McDonnell Douglas Corp. v. Green, 411 U.S. 792, 804 (1973). The McDonnell Douglas pattern of proof is applicable where, for example, a woman claims to have been discriminated against because of her sex and the employer counters with the assertion that there exists a “legitimate, non-discriminatory reason” for its behavior. In such a case, the ultimate factual question is whether the employer's decision was based on sex or on the sex-neutral reason offered by the employer. Once the nondiscriminatory reason is offered, the plaintiff can rebut the employer's explanation by showing that the reason is a “pretext”—that the employer has a history of discriminatory behavior, that in similar cases the employer has behaved inconsistently with its present actions. Any such showing is admissible to impeach the proposition that the employer's motive was licit. In BFOQ cases, the employer's use of sex is admitted, but such cases are subject to the same analysis. The explicit use of sex places the burden of justification on the employer, which can satisfy that burden by showing that sex is a BFOQ. NOW, the plaintiff can attempt to show that the asserted reason is not the real reason for the employer's actions. In other words, title VII, because of its underlying purpose of eradicating “tastes for discrimination” prohibits employment decisions—which would otherwise be permissible—if they are motivated by illicit uses of race, sex, etc. This analysis also extends to the problem of “affirmative action”; even if an employer would be authorized under the statute to base decisions on race or sex for “benign” reasons, see United Steelworkers v. Weber, 443 U.S. 193 (1979), nonbenign reasons for the use of race or sex are still prohibited; McDonald v. Santa Fe Trail Transp. Co., 427 U.S. 273 (1976). See also Texas Dep't of Community Affairs v. Burdine, 101 S. Ct. 1089 (1981); Board of Trustees v. Sweeney, 439 U.S. 24 (1979); Furnco Constr. Co. v. Waters, 438 U.S. 567 (1978).Google Scholar

108 In United States v. Carroll Towing Co., 159 F.2d 169 (2d Cir. 1947), Judge Learned Hand introduced the notion that the negligence issue can be thought of as a function of three variables, B (the burden of risk avoidance), P (the probability of injury), and L (the magnitude of injury). In that case, he stated that negligence can be found when B<PL. In a subsequent opinion, Hand recognized the limitations of the algebraic model he had formulated. In Moisan v. Loftus, 178 F.2d 148, 149 (2d Cir. 1949), he wrote:. The difficulties are in applying the rule … they arise from the necessity of applying a quantitative test to an incommensurable subject-matter; and the same difficulties inhere in the concept of “ordinary” negligence. It is indeed possible to state an equation for negligence in the form, C = P § D, in which the C is the care required to avoid risk, D, the possible injuries, and P, the probability that the injuries will occur, if the requisite care is not taken. But of these factors care is the only one ever susceptible of quantitative estimate, and often that is not. The injuries are always a variable within limits, which do not admit of even approximate ascertainment; and, although probability might theoretically be estimated, if any statistics were available, they never are; and, besides, probability varies with the severity of the injuries. It follows that all such attempts are illusory, and, if serviceable at all, are so only to center attention upon which one of the factors may be determinative in any given situation.Google Scholar

109 See Bobbitt, Philip C., Constitutional Fate, 58 Tex. L. Rev. 69S, 708 (1980).Google Scholar

110 See note 50 supra.Google Scholar

111 Burt, Robert A., The Constitution of the Family, 1979 Sup. Ct. Rev. 329, 381–82. Lester C. Thurow suggests (N.Y. Times, Mar. 8, 1981, at 2F, col. 3) that the difficulties experienced by women in establishing high-status, high-paying careers may be explained by observing that in the decade between a person's twenty-fifth and thirty-fifth birthdays—the decade when the foundations of career distinction must be laid down—many women invest substantial time in child bearing and family. According to Thurow, people who have not “made it” in the working world by the time they are 35 years old are increasingly unlikely to do so as they grow older. If this sort of explanation is accepted for the observed average differences in the compensation of males and females in the work force, it might suggest that there should be no society-wide duty to redress the imbalance, on the ground that women who bear children “assume the risk” of being less competitive than other workers who have not borne children. Or it may, on the other hand, suggest that the society's basic ethical responsibility is to do something about popular attitudes that make so much turn on what happens on the twenty-fifth through thirty-fifth year of a person's life. Age discrimination laws may recognize this responsibility; but here there is probably even less consensus about the scope of society's remedial obligations than exists with respect to sex differences. As to the correctness of Thurow's hypothesis, see Hill, Martha S., The Wage Effects of Marital Status and Children, 14 J. Human Resources 579 (1979); Corcoran, Mary & Duncan, Greg J., Work History, Labor Force Attachment, and Earnings Differences Between the Races and Sexes, 14 J. Human Resources 3 (1979); and Corcoran, Mary, The Structure of Female Wages, 68 Am. Econ. Rev. Papers & Proc. 165 (1978).Google Scholar

112 Rutherglen, supra note 45, at 251–52 (1979).Google Scholar

113 Id. at 252–54.Google Scholar

114 Id. at 252–55.Google Scholar

115 Id. at 213, 251–54.Google Scholar

116 Brilmayer, supra note 7, at 522.Google Scholar

117 See authorities cited at note 80 supra.Google Scholar

118 See authorities cited at notes 34–39 supra.Google Scholar

119 An example might be the situations in which a constitutional “right of privacy” has been thought to require exclusion of members of one sex from a particular job or job assignment. See cases cited in note 39 supra.Google Scholar

120 According to Kimball, the use of unisex actuarial tables for insurance purposes inevitably discriminates on the basis of sex because it treats men and women as equal in respect to longevity when, in point of fact, they are not equal in this regard. To treat men and women the same when they are actuarially different will have a predictably differential impact on one group or the other (assuming the longevity data are accurate). Kimball's objection to this sort of treatment, after all, is that treating men and women as equals for all purposes in a pension plan (i.e., equal premiums, equal periodic benefits) will result in a higher average per capita payment to women than to men. This is a disproportionate impact argument. See Kimball, Spencer L., Reverse Sex Discrimination: Manhart , 1979 A.B.F. Res. J. 83, 103.Google Scholar

121 Brilmayer et al., supra note 7, at 510.Google Scholar

122 Id. See also Weeks v. Southern Bell Tel. & Tel. Co., 408 F.2d 228 (5th Cir. 1969); Bowe v. Colgate-Palmolive Co., 416 F.2d 711 (7th Cir. 1969).Google Scholar

123 Brilmayer et al., supra note 7, at 508–11, 522.Google Scholar

124 As the Supreme Court held in Dothard v. Rawlinson. 433 U.S. 321 (1977). In the context of racial discrimination, Griggs v. Duke Power Co., 401 U.S. 424 (1971), is such a case. There, the employer used a written test and a high school diploma requirement (the neutral decision criteria) in making hiring decisions. These criteria were correlated racially—a much lower percentage of black than white applicants passed the test and possessed a high school diploma. Although passing the test and possessing the diploma were claimed by the employer to be related to applicant's ability to perform the job, the Court held that no demonstration had been made that these decision criteria were predictive of job performance. Thus, the employer was required to discontinue the use of diplomas and written tests in its hiring process. In race cases, the employer is limited to the second and third options (neutral decision criterion or nothing), because there is no BFOQ for race. See note 84 supra.Google Scholar

125 Fully disclosing all risks, of course, to the female applicants and leaving to them the choice to assume these risks. See note 96 supra.Google Scholar

126 This would, in a sense, also be a disproportionate impact argument. The man would be complaining that the employer is treating men as a group less favorably than they would be treated if women were excluded entirely. Of course, women could point out, in parallel to the observation in the case of an individual weight-lifting exam, that treating them as individuals (hiring without regard to sex) disproportionately injures them because they are peculiarly susceptible to the work place hazard. Similarly, women can argue that an individual weight-lifting test has a disproportionate impact on them in comparison to hiring without regard to weight-lifting ability; men can argue that an individual weight-lifting test has a disproportionate impact on them when compared to the total exclusion of women from consideration. In the array of three choices we have described, then, an argument that the middle choice, individualized testing, has a disproportionate impact, can be asserted by anyone, depending on the point of view.Google Scholar

127 Neither should a woman be heard to complain that men are being treated equally to her when in fact they are not equal. The analysis in text applies to any sort of sex classification, regardless of whose ox is gored. Such an analysis rejects the notion of “protected classes,” which would give special deference to the claims of women; however, we are not to be understood as stating a view on issues of “affirmative action.” Cf. McDonald v. Santa Fe Trail Transp. Co., 427 U.S. 273, 281 n.8 (1976); see also note 128 infra.Google Scholar

128 See Oliver Wendell Holmes, Jr., Speeches 96 (Boston: Little, Brown & Co., 1913), the source of this figure.Google Scholar

129 Posner, Richard A., The DeFunis Case and the Constitutionality of Preferential Treatment of Racial Minorities, 1974 Sup. Ct. Rev. 1, 9–15.Google Scholar

130 E.g., Regina v. Dudley & Stephens, 14 Q.B.D. 273 (1884).Google Scholar

131 It should be recognized that the argument in the text has implications for the thorny and sensitive problem of “reverse discrimination” or “affirmative action,” depending upon one's point of view. Both the advocates of “affirmative action” and the opponents of “reverse discrimination” can draw from our arguments. The advocates of “affirmative action” will argue that the powerful symbolism contained in fair employment laws (especially as applied to race, but also as applied to sex) represent society's commitment to right past wrongs and improve the lot of previously dispossessed minorities. Those who follow this line, if they accept our view of the relative strength of the color-blind norm and the sex-blind norm, would conclude that affirmative action on behalf of racial minorities stands on higher moral ground than does affirmative action on behalf of women. (Many feminists would undoubtedly disagree, but this disagreement merely recapitulates the question whether sex discrimination should be legally equated to race discrimination.). Opponents of “reverse discrimination” will draw a different lesson from our discussion. They will say that because the norm of color-blindness is entitled to greater sancity than the norm of sex-blindness, the symbolic message must not be tainted in any way. To them accordingly, drawing race lines, even if for “benign” purposes, is more pernicious than drawing sex lines. Reverse discrimination along race lines is unlawful per se; reverse discrimination along sex lines, however, might be acceptable under some circumstances. For horseback empirical confirmation that the proposition advanced in our discussion would favor affirmative action for, e.g., blacks over affirmative action for women, the reader is invited to compare the opinions of Justice Brennan et al. in Regents of the Univ. v. Bakke, 438 U.S. 265, 324 (1978), and in United Steelworkers v. Weber, 443 U.S. 193 (1979), to that of Justice Brennan in Kahn v. Shevin, 416 U.S. 351, 357 (1974). For the converse proposition, that our discussion points toward affirmative action for women more clearly than to affirmative action for blacks, compare the positions taken by Justice Rehnquist in both Bakke, 438 U.S. 265, 408 Coining the opinion of Justice Stevens), and Weber 443 U.S. at 219, with his position in Kahn v. Shevin, 416 U.S. 352 (joining the opinion of the Court).Google Scholar

132 Indeed, the current law of “state action” recognizes the difference between what government permits and what it requires. Moose Lodge No. 107 v. Irvis, 407 U.S. 163 (1972). We recognize as a constitutional matter for good and sufficient reasons that when the state discriminates, the consequences are likely to be more objectionable than when private parties do, because, we would assert, in addition to the countervailing interests that might be asserted by private parties, when the state acts, its actions carry a symbolic clout that is absent in connection even with very large and powerful private interests.Google Scholar

133 E.g., Califano v. Goldbfarb, 430 U.S. 199 (1977); Craig v. Boren, 429 U.S. 190 (1976).Google Scholar

134 435 U.S. 702 (1978).Google Scholar

135 The Manhart employer required female employees to make higher weekly contributions than men; this resulted in smaller weekly paychecks to women. The decision that this practice violated title VII does not necessarily indicate that all employer uses of sex-segregated actuarial tables would violate the statute; the Court carefully limited its decision, stating:. Although we conclude that the Department's practice violated title VII, we do not suggest that the statute was intended to revolutionize the insurance and pension industries. All that is at issue today is a requirement that men and women make unequal contributions to an employer-operated pension fund. Nothing in our holding implies that it would be unlawful for an employer to set aside equal retirement contributions for each employee and let each retiree purchase the largest benefit which his or her accumulated contributions could command in the open market. Nor does it call into question the insurance industry practice of considering the composition of an employer's work force in determining the probable cost of a retirement or death benefit plan. Finally, we recognize that in a case of this kind it may be necessary to take special care in fashioning appropriate relief. 435 U.S. at 717–18 (emphasis added). See Kimball, supra note 120, at 91–92, and Bernard D. Meltzer, Letter to Director, Office of Federal Contract Compliance Programs (Oct. 20, 1978). See also, EEOC v. Colby College, 439 F. Supp. 631, 638 (D. Me. 1977).Google Scholar

136 See Kimball, supra note 120, at 131–36. The Court's “open market” qualification (see quote, note 135 supra) will apparently not protect employers who deal with insurers using sex-segregated tables (as opposed to self-insurers like the city of Los Angeles), but will only allow lump sum payments to employees who can then choose to purchase their own annuities. See EEOC v. Colby College, 589 F.2d 1139, 1141, 1144 (1st Cir. 1978), reversing 439 F. Supp. 631 (D. Me.), and discussion in Kimball, supra, at 126–28.Google Scholar

137 Kimball, supra note 120, at 106–8.Google Scholar

138 Id. at 108–9.Google Scholar

139 See Id. at 107–8.Google Scholar

140 Brilmayer et al., supra note 7, at 531–33.Google Scholar

141 Id. at 532.Google Scholar

142 Id. at 531.Google Scholar

143 Michael B. Shimkin, Research on Causes and Nature of Cancer, in Juan A. del Regato & Harlan J. Spjut, eds., Ackerman & del Regato's Cancer: Diagnosis, Treatment and Prognosis 15 (St. Louis: C. V. Mosby Co., 1977).Google Scholar

144 An especially strong inference, to be sure. See Ackerman & del Regato's Cancer, supra note 143, at ch. 10, esp. at 370–71.Google Scholar

145 Chafetz, supra note 30, at 27; Ashley Montagu, The Natural Superiority of Women ch. 5 (New York: Macmillan, 1952); Graney, Marshall J., An Exploration of Social Factors Influencing the Sex Differential in Mortality, 28 Soc. Symp. 1, 5 (1979);Harrison, James, Warning: The Male Sex Role May Be Dangerous to Your Health, 34 J. Soc. Issues 65, 73 (1978). See generally Madigan, F. C., Are Sex Mortality Differentials Biologically Caused? 35 Millbank Memorial Fund Q. 202 (1957). The Brilmayer authors, supra note 7, discount the biological explanations but maintain that the causal argument is irrelevant to their position. We agree; moreover it is irrelevant to the position taken here. If the data are sufficiently convincing and use of sex as a predictor is substantially more efficient than the use of other factors that may explain some part of the “sex differential in mortality,” the use of sex as a predictor is justified. To be sure, actuarial tables are based on past experience, and thus may be dated to the extent that changes in life style associated with changing sex roles have any impact on longevity. If such changes do affect longevity, the actuarial tables will eventually reflect this change. In the interim, assuming that the burden of our argument on the efficiency of the sex-segregated tables is accepted, the employer/insurer should be entitled to use the best evidence available—recent actuarial experience. See Kimball, Spencer L., Reprise on Manhart , 1980 A.B.F. Res. J. 915, 918–20.Google Scholar

146 There is one circumstance where we can properly grant that a reliable sex-linked correlation should be considered “spurious,” and consequently rejected as a predictor. That is the case of the “self-fulfilling prophecy,” where the predictive variable is recognizably a cause of the effect as well as the effect of the cause. This point is made more explicitly in our discussion of limiting principles, sec. IV infra.Google Scholar

147 City of Los Angeles Dep't of Water & Power v. Manhart, 435 U.S. 702, 705 (1977).Google Scholar

148 The precision with which the premium can reflect the probabilities inherent in the statistical prediction has a suggestive parallel in the hiring context. If an employer, on our analysis thus far, would be able to make out a BFOQ on grounds of efficiency, might the employer in preference to refusing to hire members of one sex (the binary decision), choose instead to transfer the costs of the differential risk onto the excludable sex? For example, in a “reproductive hazards” case, could an employer, instead of excluding pregnant women from jobs they wished to have, require such women to insure the employer against the risk of future liability, or to indemnify the employer, or (what is identical in economic terms) to accept lower wages to reflect the higher risk they pose to the employer's bottom line? The Equal Pay Act of 1963 would seem to bar the employer from taking any of the above steps. See Pub. L. No. 88–38, § 3, 29 U.S.C. § 206(d)(1) (1976). But such actions would seem to be consistent with a conception of “equal pay for equal work,” which considers risk to be an element of “equal work”—e.g., our notions of equal pay for equal work are not offended if two people, hired for the same job, are paid different salaries on the basis of level of education. If our analysis of an efficiency-based BFOQ is correct, it is not clear why an employer should be put to the choice of excluding all members of one sex or eating all the costs of not doing so. A middle ground, a “less restrictive alternative,” exists which could serve the efficiency needs of the employer without the overinclusive interference with employment opportunities that outright disqualification of one sex inevitably represents.Google Scholar

149 See discussion supra at sec. IIE.Google Scholar

150 Brilmayer et al., supra note 7, at 510–11.Google Scholar

151 See Kimball, supra note 145, at 916.Google Scholar

152 See Kimball, supra note 120, at 134 n.134. For a discussion of some anomalies in the administration of unisex tables, see id. at 134–35.Google Scholar

153 As is required by the Equal Pay Act, 29 U.S.C. § 206(d)(1) (1976).Google Scholar

154 At the bottom of every Deep Pocket, there is sure to be a Free Lunch.Google Scholar

155 In the present Congress, legislation is pending that would prohibit the use of sex (and race, religion, and national origin) in any insurance context. H.R. 100 (H.R. Rep. No. 100, 96th Cong., 1st Sess. (1979)). The sex integration of all insurance tables would have a considerable adverse impact on women, because men as a class have 70 percent more automobile accidents than do women. See Regulation, Jan./Feb. 1981, at 6–8.Google Scholar

156 See Meltzer, letter, supra note 135. There is an even stronger reason for permitting the use of unisex tables under title VII, and this is to be found in the Bennett Amendment. With this much of the Manhart case we can agree. The Bennett Amendment to title VII, § 703(h), immunizes compensation schemes that would otherwise be vulnerable to a disproportionate impact analysis. See Manhart, 435 U.S. at 710 n.20; see also Brilmayer et al., supra note 7, at 516–20. The Brilmayer authors' further assertion that the Bennett Amendment requires that sex-segregated tables be prohibited merely revisits the claim that everything not prohibited is mandatory. See discussion supra. Additionally, the Court has suggested (although it disclaims having decided) that claims brought under § 703(a)(1) of title VII are not subject to “disproportionate impact” analysis. General Elec. Co. v. Gilbert, 429 U.S. 125, 137 (1976); Nashville Gas Co. v. Satty, 434 U.S. 136, 144 (1977). The Court has further noted that § 703(a)(1) “appears to be the proper section of Title VII” under which to analyze employer practices that have “no direct effect upon either employment opportunities or job status.” id. at 145. In other words, title VII challenges to fringe benefit programs are to be considered under § 703(a)(1); if disproportionate impact claims are not permitted under that section, the use of unisex tables would not be prohibited. This is the result in any case under the appropriate reading of the Bennett Amendment, see supra; but the Bennett Amendment applies only to sex discrimination. Placing reliance on § 703(a)(1) would have broader implications, in that it would preclude disproportionate impact challenges to fringe benefit programs where race or other proscribed categories were at issue. There is little support for the proposition that § 703(a)(1) embodies a different conception of “discrimination” than does § 703(a)(2). See Sullivan et al., supra note 5, at 153.Google Scholar

157 The Pregnancy Discrimination Act of 1978 does not support a contrary view. In this statute (Pub. L. No. 95–555, 92 Stat. 2076 (1978)), Congress created in title VII a new § 701(k), which provides that the terms “because of sex” and “on the basis of sex” shall include “because of or on the basis of pregnancy, childbirth, or related medical conditions.” It further provides that “women affected by pregnancy, childbirth, or related medical conditions shall be treated the same for all employment-related purposes, including receipt of benefits under fringe benefit programs, as other persons not so affected but similar in their ability or inability to work.” This amendment to title VII was the congressional response to the Supreme Court's decisions in General Elec. Co. v. Gilbert, 429 U.S. 125 (1976). In Gilbert, the Court held that title VII was not violated by an employer's failure to include pregnancy-related disabilities in an otherwise comprehensive disability insurance package. The Court said that discriminations based on pregnancy were not on the basis of sex, because “[t]he program divides potential recipients into two groups—pregnant women and nonpregnant persons. While the first group is exclusively female, the second included members of both sexes.” 429 U.S. at 135, quoting Geduldig v. Aiello, 417 U.S. 484, 496–97 n.20 (1974). The “non-discrimination” rationale of Gilbert was called into question, however, in the Court's subsequent decision in Nashville Gas, which held that an employer was guilty of sex discrimination under title VII in failing to grant carryover seniority credit to women on disability maternity leave, while granting seniority credit in connection with other disability leaves. The Court's attempts to explain away the apparent inconsistency between Gilbert and Nashville Gas were unconvincing (see Sullivan et al., supra note 5, § 2.6) and confusing. Congress attempted to clarify this muddle by means of the Pregnancy Act, which requires that “pregnancy discrimination” be treated as a form of explicit sex discrimination. But the Pregnancy Act went even further and required that pregnancy be treated equally with other disabilities for fringe benefit purposes. This latter aspect of the amendment was necessary to overrule the result in Gilbert, an alternate holding of which was grounded in the fact that the disability insurance program at issue in that case, with pregnancy coverage excluded, was apparently actuarially equivalent for men and women. This congressional rejection of an actuarial equivalence (i.e., efficiency) justification for what is now, under the statute, an explicit sex classification in a fringe benefit package, is said to demonstrate that Manhart was rightly decided. See Brilmayer et al., supra note 7, at 521–22. But this conclusion would follow, if at all, only on the view that the Pregnancy Act restates the law under title VII prior to the amendment. Although there are statements in the legislative history of the act which indicate that it was intended to clarify pre-existing law, it is hardly uncontroversial that one Congress can authoritatively pronounce on the meaning of a statute passed by a different Congress. Moreover, the fact that Congress saw fit not only to clarify the meaning of “on the basis of sex” but also to legislate specifically on the issue of actuarial equivalence, strongly supports the inference that without this latter step, the actuarial equivalence of the disability insurance program in Gilbert might have been a defense to the claim of sex discrimination.Google Scholar

158 See Underwood, Barbara D., Law and the Crystal Ball: Predicting Behavior with Statistical Inference and Individualized Judgment, 88 Yale L.J. 1408, 1414–15 (1979):“When the predicted fact is not subject to individual control, then predicting that fact is less threatening to the value of respect for autonomy.” Id. at 1415.Google Scholar

159 See, e.g., Rutherglen, supra note 45, at 250–51.Google Scholar

160 Id. at 252.Google Scholar

161 Graney, supra note 145.Google Scholar

162 See secs. IIB1, IIB2 supra.Google Scholar

163 See Kimball, supra note 145, at 919. For all we know, the safe driving differences between young men and young women result from young men's driving more miles rather than less safely. Our hypothetical is assumed to control for this possibility.Google Scholar

164 We assume, for purposes of this hypothetical, that the sex of a prospective delivery driver is the only predictive variable to which the pizza parlor will have feasible access; thus, the risk costs of using sex as a decision criterion are to be compared with the risk cost involved in hiring at random.Google Scholar

165 Of course, if it can be shown that the employer's efficiency justification is merely a “pretext” for an improper motive (including a desire to impose an impermissible normative judgment on members of one sex, see supra), a plaintiff may show this in rebuttal of the employer's defense. See McDonnell Douglas Corp. v. Green, 411 U.S. 792, 804 (1973). See text at note 107 supra.Google Scholar

166 This hypothetical is probably on the edge of the BFOQ exception as commonly understood, see sec. 11C2 supra, but because we consider that in any event a broader efficiency justification either is or ought to be a part of the statute, we put this problem to one side.Google Scholar

167 Unless, of course, the employer's behavior is motivated by this stereotype, rather than by the observed differences in the behavior of men and women that may flow from their acceptance of the stereotype. See note 165 supra.Google Scholar

168 Chafetz shows that men and women tend to quit at similar rates when their jobs are held constant. As it is, women quitters tend to be low-level employees. Chafetz, supra note 30, at 132–33. A wage differential (see note 148 supra) to take account of the extra risk posed by the employment of women would also have “self-causing” overtones, if of smaller magnitude, and would tend to push women in the direction of being secondary wage earners.Google Scholar

169 Prosser, supra note 96, at 244.Google Scholar

170 See, e.g., Raymond Motor Transp., Inc. v. Rice, 434 U.S. 429 (1978).Google Scholar

171 See Allied Structural Steel Co. v. Spannaus, 438 U.S. 234 (1978).Google Scholar