Larry Downing/Reuters

When Chief Justice John Roberts prepared to decide the constitutionality of the Affordable Care Act in 2012, he faced a moment of truth. In an interview for this magazine in 2007, he stressed the importance of maintaining the legitimacy of the Court by avoiding 5-4 decisions on party lines; now he had to decide whether to preside over a partisan divide in the most closely watched case of his tenure. In the end, Roberts made a brave decision to choose judicial legitimacy over ideological purity, casting a tie-breaking vote to uphold this Affordable Care Act.

By the end of June, Roberts will face two decisions that some commentators are viewing as yet another defining moment for his judicial legacy and his judicial soul. Once again, he has to decide whether to drive a stake into the heart of the Affordable Care Act, and he also must decide whether the Constitution requires states to recognize gay marriage. It’s become a parlor game to speculate whether or not he will be swayed by politics in the Obamacare and gay marriage cases. Will he vote with the conservative justices who oppose the Affordable Care Act and gay marriage on policy grounds or maintain his commitment to keeping the Court above partisan politics?

This time, however, Roberts is unlikely to view the cases in the same overtly political terms as the public at large. Focusing on arguments about law rather than legitimacy, he may be more likely to exercise judicial restraint than to make a strategic intervention. Roberts is less likely to cast the tie breaking vote in either case—Justice Anthony Kennedy appears more likely to vote with the liberals, sparing Roberts the need to avoid a 5-4 decision on party lines. Moreover, in round two of the legal battle over the Affordable Care Act, Roberts may be less torn than he was in round one: He is less of a committed textualist than Justice Antonin Scalia and appears less enamored of the argument that the plain text of the Affordable Care Act precludes tax breaks for state-created exchanges. On marriage equality, Roberts’ previous opinions suggest he may be inclined to uphold same sex marriage bans, but since he would be deferring to democratic decisions rather than second-guessing them, he has the freedom to follow his constitutional impulses without calling the Court’s legitimacy into question. For all these reasons, although both the Affordable Care Act and gay marriage cases may reinforce Roberts’s commitment to judicial restraint, they may not ultimately define his judicial legacy.

The Affordable Care Act case, King v. Burwell comes down to the meaning of four words: “established by the State.” The ACA authorizes the Internal Revenue Service to provide tax subsidies to low-income people who buy insurance on an exchange “established by the State under section 1311.” The Act also authorizes the federal government to create “such exchanges,” when the state fails or elects not to do so. And section 1311(d)(1) provides that an exchange “shall be an ... entity established by the State.” The question in King v. Burwell is whether the IRS can permissibly extend tax credit subsidies to individuals in the thirty-four states that did not create their own exchanges and relied instead on the federal government.

The plaintiffs in the case rely on textualism in arguing that the four words mean what they say and that subsidies are only available to people who purchase insurance from exchanges established by the states. If the Supreme Court accepts this argument, it would gut the ACA: Millions of Americans would lose their health insurance. The government counters that the four words have to be read in the context of the whole statute and that Congress intended to extend the subsidies to exchanges created by both the federal government and the states.

If Chief Justice Roberts finds the case a close call, he might have to wrestle with the question of whether to vote to hobble the Act he rescued three years ago. But he may not find the case a close call. As several commentators have noted, a case decided earlier this term suggests that he is less of a committed textualist than Justice Antonin Scalia. That case is Yates v. United States, involving a provision of Sarbanes-Oxley, which defines criminal obstruction of justice to include anyone who “knowingly alters, destroys, mutilates, conceals, covers up, falsifies, or make a false entry in any record, document, or tangible object.” A fisherman named Yates had tossed an illegal catch overboard to eliminate the evidence of his violation.

Chief Justice Roberts joined a plurality opinion written by Justice Ginsburg which held that “tangible objects” didn’t include fish because they were limited to “objects one can use to record or preserve information, not all objects in the physical world.” The opinion made clear that context matters. “Whether a statutory term is unambiguous ‘does not turn solely on dictionary definitions of its component words.' ... In law as in life, however, the same words, placed in different contexts, sometimes mean different things.”

Chief Justice Roberts pointedly refused to join Justice Alito’s concurring opinion, which also held that fish weren’t tangible objects but based its result on narrower and more hypertextualist grounds, parsing “the statute’s list of nouns, its list of verbs, and its title.”

The chief justice’s departure from traditional textualism is not limited to Yates. As Michael Dorf of Cornell Law School has noted, “He is probably the least textualist, most intentionalist of the current conservatives when it comes to statutory interpretation.” As support, Dorf points to Bond v. United States, a case that involved the 1997 Chemical Weapons Convention. The defendant in that case had used a mild toxin to cause a skin rash on a romantic rival. Based on the plain meaning of the statute, this crime of passion might seem to violate the Chemical Weapons Convention. But the chief justice wrote an opinion for the Court ruling that the statute didn’t apply. Justice Scalia wrote of Roberts’s opinion in Bond:

The Court does not think the interpretive exercise so simple. But that is only because its result-driven anti-textualism befogs what is evident … The ordinary meaning of the term being defined is irrelevant, because the statute’s own definition—however expansive—is utterly clear … The statute parses itself … Once again, contemplate the judge-empowering consequences of the new interpretive rule the Court today announces: When there is “dissonance” between the statutory definition and the ordinary meaning of the defined word, the latter may prevail.

If Roberts is, as Scalia charges, an anti-textualist (or at least not a hypertextualist), he may have no trouble agreeing with the Obama administration even if he doesn’t like the Affordable Care Act on policy grounds. That is because the case revolves around statutory interpretation. A new study by scholars including Dan Kahan of Yale Law School and Danieli Evans, Senior Fellow in Constitutional Studies at the National Constitution Center, found that in cases involving the interpretation of a statute addressing a politically polarizing issue, such as immigration or abortion, judges are far less influenced by their ideological bias than when members of the public are asked to decide the same statutory interpretation question. This indicates that judges are more capable than members of the public at setting their biases aside when engaging in legal reasoning and interpreting statutes. The study tested only cases involving statutory interpretation. By contrast, constitutional cases, which are not constrained by statutory language or legislative intent, may give judges more leeway to express their substantive jurisprudential views. If so, this study may support an observation by Justice Breyer in a 2008 interview at the Sixth and I Synagogue. Justices are more likely to agree in statutory than in constitutional cases, he said, because the answers are more technical and their strong ideological predispositions are less likely to be engaged.

While Roberts maintained an uncharacteristically sphinx-like silence during oral arguments, he asked one question that may provide additional insight. Responding to Solicitor General Donald Verrilli, who had just stated that a doctrine called Chevron deference should support the government, the Chief Justice asked “[I]f you’re right about Chevron, would that indicate a subsequent administration could change that interpretation?”

Chevron deference asks first whether the text is ambiguous. If it is, the court must then decide whether the agency’s interpretation of the text is reasonable. In practice, if a court reaches the second step of Chevron, the agency interpretation wins. Both parties in the ACA case prefer that the court not find the text ambiguous: the plaintiffs, because such a finding means they lose; and the government, for precisely the reason that the chief justice identified. A finding of textual ambiguity would leave the door open for a subsequent administration to enact a different interpretation. A holding along these lines would allow the chief justice to avoid decimating the ACA while simultaneously placing its fate in the hands of the next president of the United States.

Roberts’s vote in the marriage equality cases is harder to predict. Conventional wisdom holds that Justice Kennedy will join the four liberals in upholding same-sex marriage. Although some have speculated that Roberts could provide a sixth vote because he has not yet shown his hand on same-sex marriage, and might find bans on same-sex marriage to be similar to bans on interracial marriage, this seems like wishful thinking: nothing in his jurisprudence suggests a readiness broadly to construe marriage as a fundamental right that has to be extended to all citizens on equal terms.

The clearest indication of Roberts’s approach to same-sex marriage comes from his separate dissent in the Windsor case last year, where the Court struck down the Defense of Marriage Act. Roberts disagreed. “[W]ithout some more convincing evidence that the Act’s principal purpose was to codify malice, and that it furthered no legitimate government interests,” he wrote, “I would not tar the political branches with the brush of bigotry.” Unlike Justices Scalia, Thomas, and Alito, Roberts did not explicitly declare that state same-sex marriage bans are constitutional.  Instead, he kept his options open: “The Court does not have before it, and the logic of its opinion does not decide, the distinct question whether the States, in the exercise of their historic and essential authority to define the marital relation, may continue to utilize the traditional definition of marriage.”

Now that Roberts has to decide this question, the test he will apply seems clear. The Chief Justice, in his Windsor dissent, suggested that he would require a finding that “the principal purpose” of gay marriage bans was “a bare desire to harm”—in other words, that they were enacted with “sinister motive.” In Windsor, he stressed that he would not lightly “tar the political branches with the brush of bigotry.” He also emphasized differences between state marriage bans and the federal Defense of Marriage Act that might make them less constitutionally vulnerable, including the fact that their legislative history may not have included similar evidence of bias and that states have traditionally had the authority to define marriage as “the foundation of the State’s broader authority to regulate the subject of domestic relations with respect to the ‘[p]rotection of offspring, property interests, and the enforcement of martial responsibilities.’”

It’s easy to imagine Roberts invoking the tone and analysis of the opinion he will be reviewing, Judge Jeffrey Sutton’s opinion for the U.S Court of Appeals for the Sixth Circuit upholding gay-marriage bans in four states. Sutton is one of the most thoughtful and restrained conservative judges in the country—like Roberts, he voted to uphold the Affordable Care Act, invoking the importance of judicial deference to democratic decisions. And he struck a similar theme in rejecting the argument that voters in four states were primarily motivated by animus when they banned same-sex marriage in the past decade or so:

Some equanimity is in order in assessing the motives of voters who invoked a constitutionally respected vehicle for change and for resistance to change: direct democracy. Just as gay individuals are no longer abstractions, neither should we treat States as abstractions. Behind these initiatives were real people who teach our children, create our jobs, and defend our shores. Some of these people supported the initiative in 2004; some did not. It is no less unfair to paint the proponents of the measures as a monolithic group of hate-mongers than it is to paint the opponents as a monolithic group trying to undo American families. “Tolerance,” like respect and dignity, is best traveled on a “two-way street.” If there is a dominant theme to the Court’s cases in this area, it is to end otherness, not to create new others.

Let’s imagine that Roberts is inclined to uphold the state initiatives defining marriage as between a man and a woman because, like Sutton, he believes in judicial restraint and is unwilling lightly to ascribe animus to the citizens who supported them. Might he set aside his constitutional views and provide a sixth vote to strike down gay marriage bans, to preserve the legitimacy of the Court? This seems unlikely: Roberts is most concerned about legitimacy when the Court strikes down laws on a 5-4 party line vote; the same concerns don’t arise when the court strikes them down with a mix of Democratic and Republican votes. Moreover, in the absence of clear constitutional arguments to the contrary, he prefers deferring to democratic decisions, not second-guessing them. Finally, he has strong views that he has been unwilling to compromise in the past and would hardly join an opinion he believed was fundamentally incorrect because of an amorphous concern about his personal judicial legacy, rather than the legitimacy of the Court as a whole. For all these reasons, unless Roberts concludes that the state initiatives are infected by animus (a plausible conclusion in light of the Court’s lax definition of animus in other gay-rights cases), he seems unlikely to join Kennedy in striking the state-level marriage bans down.

Both conservatives and liberals misunderstood Roberts’s vote to uphold the Affordable Care Act as a political act. Instead, in the spirit of his hero, John Marshall, he proved willing to cast a tie-breaking, strategic vote, to avoid making the Court appear partisan in one of the most closely watched cases of the decade. During his opportunity for a re-do in the most recent healthcare and gay-marriage cases, he is unlikely to face the choice of casting a fifth vote with the other conservatives and will therefore be free to follow his legal views. That may lead him to uphold both the tax subsidies in the Affordable Care Act as well as the state gay-marriage bans—an incongruous pair of rulings from a political perspective, but entirely consistent as a matter of legal interpretation. What the cases are likely to reveal, in other words, is not John Roberts’s soul but the depth of his principled commitment to judicial restraint.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.