When will the Supreme Court rule on marriage laws? While that question is debated, another brews beneath the surface. It may affect not just the outcome of any ruling, but how we think of support for marriage—and related policies—whatever the Court decides.

The question—which can determine who has the burden of proof in marriage cases—is whether laws defining marriage as a male-female union should be treated as forms of sex discrimination.

Some background first. Courts have created tests to enforce the constitutional right to equal protection of the laws. One asks if a policy involves a “suspect classification.” If so, it’s presumed unconstitutional, and the government has to show that it clears one of two bars.

Policies that classify by race or ethnicity, for example—like school segregation and interracial marriage bans—are struck down unless they’re narrowly tailored to serve compelling state interests. Those that classify based on sex get “intermediate scrutiny”: they have to be “substantially” related to an “important” state interest.

Start your day with Public Discourse

Sign up and get our daily essays sent straight to your inbox.

Other policies just have to have some conceivable rational basis—a very low bar that effectively shifts the burden of proof to the policies’ opponents.

That important difference explains why Justice Anthony Kennedy, widely seen as a swing vote on the marriage issue, asked in the Proposition 8  argument:

JUSTICE KENNEDY: Do you believe this can be treated as a gender-based classification?

MR. COOPER: Your Honor, I -­

JUSTICE KENNEDY: It’s a difficult question that I’ve been trying to wrestle with it.

Many distinguished commentators would answer Kennedy in the affirmative—and not just out of reflexive social liberalism. The great and independent-minded constitutional scholar (and a close mentor of mine at Yale) Akhil Reed Amar and even some right-leaning scholars like Steve Calabresi (Northwestern) and Ilya Somin (George Mason) all think male-female marriage laws should indeed be subject to the scrutiny applied to sex-based discrimination. If they’re right, states will have a harder time arguing their case.

Marriage laws do of course require state officials to consider people’s sex. But it isn’t entirely clear that all policies that do so (e.g., sex-segregated public bathrooms) must get heightened scrutiny under the Supreme Court’s current doctrine. And even if they do, this can change: if judges created the test, they can tweak it.

Either way, I submit, the Court should decide not to extend to marriage laws the scrutiny applied to sex classifications. It should reject what would be (at best) a flatfooted application of the tests it has devised. Such a rejection would cohere perfectly well with its concerns in sex-discrimination cases, and it would make eminently good sense.

In short, I’ll show, the kind of classification used in marriage laws is meaningfully different from just about every other, and certainly from those deemed “suspect.” And you don’t have to agree with me about marriage to see this; it’s based on structural features of the classification, whether ultimately just or not.

Commentators concerned with democratic self-governance—like Amar, Calabresi, and Somin—should not obscure this point. And if our legal system overlooks it, we’ll be likelier as a culture to see even the most modest affirmation of sex differences as an affront to human dignity and equality—with bad effects beyond our state marriage codes.

History of Sex Discrimination Cases

Those who disagree would say that laws that define marriage as the union of a man and woman do just what the laws did that were struck down in the major sex-discrimination cases of Frontiero v. Richardson and Weinberger v. Wiesenfeld. The first made benefits unequally available to Air Force members’ dependents. Wives of male members were automatically eligible, but a female member like Sharron Frontiero could obtain benefits for her husband only if she could show that he depended on her for more than half his living. Under the provision in Wiesenfeld, Steven Wiesenfeld, as a widowed father, was ineligible for certain Social Security benefits available to widowed mothers.

Some courts have tried to distinguish such policies from state marriage laws by pointing out that they treat men and women differently, while marriage laws do not. The usual reply is that the interracial marriage ban in Loving v. Virginia was defended on similar grounds, as applying equally to blacks and whites. In response, it’s often pointed out that the history and structure of the ban in Loving showed that it perpetuated white supremacy (as the Supreme Court itself noted), while laws limiting benefits to opposite-sex couples do not perpetuate male supremacy.

Even the most expansive sex-discrimination case, United States v. Virginia—in which the Court struck down the Virginia Military Institute’s male-only admissions policy—takes for granted the differences between racial and sex-based classifications. As Justice Ruth Bader Ginsburg wrote for the Court:

The heightened review standard our precedent establishes does not make sex a proscribed classification. Supposed “inherent differences” are no longer accepted as a ground for race or national origin classifications. Physical differences between men and women, however, are enduring: “[T]he two sexes are not fungible; a community made up exclusively of one [sex] is different from a community composed of both.”

“Inherent differences” between men and women, we have come to appreciate, remain cause for celebration . . . But such classifications may not be used, as they once were, to create or perpetuate the legal, social, and economic inferiority of women.

This highlights one telling structural difference between marriage laws and the policies struck down in these sex-discrimination cases—a difference that courts and commentators so far have all missed.

Vive La Difference

Sharron Frontiero had to clear a higher legal bar to earn benefits for her spouse simply because she was a woman. Wiesenfeld was ineligible for certain benefits just because he was a man.

By contrast, Ms. Windsor’s ineligibility for a tax break in the DOMA case, for example, hinged on her sex only derivatively. Unlike the benefits in Frontiero and Wiesenfeld, the tax break she sought was equally available to men and women: being a woman left the eligibility question open. In that sense, Windsor’s eligibility did not depend solely on her own (or any individual’s) sex. The basic criterion was a characteristic of a pair of persons: the couple to which she belonged. Specifically, it was the couple’s sexual composition: same-sex or opposite-sex.

Of course, determining a couple’s sexual composition requires determining individuals’ sex; it involves treating sex as Virginia’s ban in Loving had treated race. But a concern with sexual composition differs from one with racial composition; from a direct concern with individual sex, as in Frontiero and Wiesenfeld; and from other kinds of grouping.

After all, male and female are not just any two sexes, as black and white are just two races. Maleness and femaleness, and a certain social purpose, are necessarily inter-defined: one can’t even fully explain either concept without reference to the other and to a certain social good. What differentiates male and female are not just anatomical or genetic features, but—at a deeper level of explanation—their joint (basic) physical potential for a biological task: reproduction. And this task, its social value, and its link to sexual composition—to the male-female pair—are certainly not mere social inventions.

Note that I am not simply skipping ahead to the question of whether the classification involved in male-female marriage laws is ultimately justified. For all I have said, it might turn out not to be. My point is about the best level of scrutiny, still a question of presumptions. It is that any particular racial (or ethnic, or religious) grouping is prima facie arbitrary—and its relevance to policy presumptively in need of justification—as the male-female sexual grouping is not. In none of the typically suspect groupings (racial, ethnic, etc.) do the very categories of the grouping have any inherent positive (or negative) connection to a legitimate political end. They have such a connection, if at all, only by virtue of contingent and changeable social or cultural goals. Those goals have often been malign. So it makes sense not to presume their legitimacy, and to devise rules for scrutinizing them—as the Court has done.

But the male-female sexual grouping is necessarily linked, by the concepts involved, to a social purpose we did not simply invent and can scarcely do without: society’s reproduction. In this way, the classification in marriage laws differs from any racial grouping, even from other sex classifications.

If our very concepts of African-American and white, for example, ever suggested any particular social harm (or benefit), alone or in combination, it was only because we had created or invented, by our conventions, the harm or the link or both. The same goes for perceived links between maleness or femaleness and most particular professions—but not between the male-female pair and social reproduction. (Of course, our concepts of the sexes are culturally shaped—but they aren’t pure social constructs.)

In light of this, the Court might well decide to leave in place heightened scrutiny of individual sexual and racial classifications, and of racial-composition classifications (as in Loving), while at least presuming the constitutionality of laws that classify by opposite-sex composition.

Such a scheme would best vindicate Justice Ginsburg’s judgment in the VMI case that male-female “inherent” and “physical” differences—unlike alleged interracial differences—are a cause for “celebration” (though not oppression or limitation).

Indeed, what could possibly hew more closely to Ginsburg’s standard than a rule that heightened scrutiny of any sex classification except one targeting (1) a necessarily “celebrat[ed]” social end to which physical sex differences are (2) “inherently” linked?

Constitutionalized Stereotyping?

Some might fear that this move would plant deep in constitutional doctrine diseased ideas that the Court has been moving to eradicate: “outmoded” gender notions like the “pervasive sex‐role stereotype,” repudiated by the Rehnquist Court, that “caring for family members is women’s work.” Or it could entrench what Yale’s Jack Balkin calls a “system of social meanings” that keeps patriarchy in place by “defin[ing] masculinity and femininity in terms of complementary traits and attraction to the opposite sex.”

This is an important concern. We cannot deny that gender stereotypes can be pretexts for subjugating women or unjustly limiting their liberty. Whether all generalizations about sex or gender should be repudiated has been disputed. Against this idea, some feminists (liberal and conservative) have suggested that trying to uproot even the most physically grounded ideas about sex would actually harm and demean women, by holding up the “unencumbered, wombless male” body as the ideal by which all are judged.

But however that dispute might be resolved, my proposal would not entrench stereotypes about what constitutes proper gender identity or behavior. Its premise is not that men are by definition or essence those attracted to women or fatherhood, so that gay or unattached or childless men are abnormal—nor, mutatis mutandis, are women.

It is simply that maleness and femaleness are conceptually specified by men and women’s basic physical potential—not moral obligation or psychological need but radical physical potential—to advance together an interest of any society. This is the kind of “undeniable difference”—like the fact that only mothers give birth—that under the Court’s cases can justify a policy without constituting a stereotype.

This highlights a point as important for broader social issues (and the health of our cultural debates) as it is for the litigation on marriage: The primary question regarding the definition of marriage is not whether any particular class of individuals (gay, straight, male, female) has a special link to the common good, but whether certain couples do. And it shifts the burden of proof onto those who would find no such link.

But that burden, as my coauthors and I and others have argued, is one that can’t be met.