After his solid, broadly based victories in New Hampshire and South Carolina, Donald Trump now holds a commanding position in the race for the Republican presidential nomination.

But Trump still faces two “known unknowns,” to borrow the memorable phrase from former Defense Secretary Donald Rumsfeld, an architect of the Iraq War that Trump now excoriates. One is whether Trump has a ceiling of support. The second is whether, even if he does, any of his remaining rivals can unify enough of the voters resistant to him to beat him.

So far the evidence suggests the answers are: maybe, and not yet. Indeed over the first three contests, Trump’s two principal remaining opponents have shown mirror-image weaknesses. Texas Senator Ted Cruz has assembled a coalition of support that is too narrow; Florida Senator Marco Rubio is building a coalition that is too shallow.

As in his New Hampshire win earlier this month, Trump’s support in South Carolina transcended many of the usual fissures in Republican politics, according to exit poll results posted by CNN. The one big exception remained education: In each of the first three contests, including the Iowa caucus, Trump has not run as well among voters with a college degree as with those lacking advanced education. But because those white-collar voters have fragmented among many choices, none of Trump’s rivals is consolidating enough of them to overcome the New Yorker’s dominant position among voters without a college degree. The simple equation that Trump has consolidated blue-collar Republicans while the party’s white collar wing remains divided remains the most powerful dynamic in the race, even as Trump has failed to exceed 35 percent of the vote in any of the initial contests.

On most fronts, the big story in South Carolina was the breadth of Trump’s appeal. Repeating the New Hampshire pattern, Trump in South Carolina ran slightly better among men (36 percent) than women (29 percent). He carried 29 percent of voters who identified as very conservative; 35 percent of somewhat conservative voters; and 34 percent of moderates. That also followed the New Hampshire precedent of little ideological variation in Trump’s support.

In South Carolina, Trump won 33 percent of independents and 32 percent of self-identified Republicans; in New Hampshire he had carried exactly 36 percent of both groups. Trump ran somewhat better last night among voters older than 45 (35 percent) than those younger (26 percent). In New Hampshire, by contrast, Trump’s support varied little by age, though he also performed somewhat better with older voters in Iowa.

Perhaps most encouraging for Trump in South Carolina was his performance among evangelical Christians, who the exit poll said comprised nearly three-fourths of all voters, an unprecedented number. Cruz had spent the past week ardently courting evangelical voters, particularly in “the Upstate” around Greenville and Spartanburg, while insisting that he was the most reliable conservative and the best bet to appoint a proper successor to the late Antonin Scalia on the Supreme Court. Yet Trump on Saturday carried a plurality of voters who identified as evangelicals (33 percent) as well as a plurality of those who did not (30 percent). That matched Trump’s feat in New Hampshire when he topped his rivals with both groups (27 percent with voters who were evangelicals and 38 percent with those who were not). In Iowa, Trump also ran first among voters who were not evangelicals, but Cruz beat him by a solid double-digit margin among voters who were.

Trump’s evangelical performance was even more impressive in South Carolina than New Hampshire, partly because born-again voters comprised so much more of the vote in the former (72 percent) than the latter (25 percent.) And Trump won those religiously conservative voters in the Palmetto State even though Cruz had built a much stronger organization among them than he had in New Hampshire.

But, despite all these other indications of strength, in South Carolina Trump’s core asset—and the key to his performance among evangelicals—remained his extraordinary hold on working-class Republicans. In South Carolina, Trump won fully 42 percent of white GOP primary voters without a college degree-exactly the same imposing percentage he carried in New Hampshire. In South Carolina, Trump won almost exactly as many non-college whites as Cruz (24 percent) and Rubio (17 percent) combined. In New Hampshire, Trump carried more non-college whites than his next three closest competitor combined. Only in Iowa did Cruz narrowly edge Trump among working-class whites.

All week in South Carolina, Trump courted those working-class white voters by sharply raising the volume on his criticism of free trade and his warnings about the threat from Islamic terrorism. Both of those issues somewhat overshadowed his usual promises to both build a massive wall at the Mexican border and to deport the estimated 11 million undocumented immigrants. And, in fact, as in New Hampshire, only a minority of primary voters said in the exit poll that they supported deporting undocumented immigrants: In both states, most said they should be allowed to obtain legal status (though not necessarily citizenship.) Again, as in New Hampshire, Trump’s proposed temporary ban on Muslims entering the U.S. scored better, with about three-fourths of GOP primary voters backing it. Trump won 47 percent of the voters who backed deportation (double his closest rival) and 41 percent of those who supported the Muslim ban (almost equal to Cruz and Rubio combined).

Critically, Trump’s success with blue-collar voters crossed the religious boundary. In South Carolina, according to figures provided by CNN’s polling director, Jennifer Agiesta, Trump carried a stunning 44 percent of evangelical voters without a college degree, as much as the combined vote for Cruz (29 percent) and Rubio (15 percent). Among blue-collar voters who were not evangelicals, Trump (at 38 percent) also buried Cruz (16 percent) and Rubio (13 percent), with Ohio Governor John Kasich actually leading both (at 17 percent).

That largely followed the pattern from New Hampshire where Trump beat Cruz by 13 percentage points among blue-collar voters who were evangelicals, and by a crushing 36 points among those who were not. In Iowa, Trump had also comfortably carried the working-class voters who were not evangelicals, but Cruz had amassed a solid double-digit lead among blue-collar Republicans who were.

Trump, in other words, has now carried in all three states voters who fit the historic description of “Reagan Democrats”: blue-collar voters who are not evangelicals. If Trump can also maintain the advantage among the non-college evangelicals he’s established in the past two contests, that will make him very tough to beat in upcoming Southern (Georgia, Alabama, Mississippi), border (Oklahoma) and even Midwest states (Ohio and Missouri) that contain large numbers of both working-class and born-again voters.

In all these ways, Trump in South Carolina loomed large. But he continued to display one weak link. Among voters with at least a four-year college degree, Trump carried only 25 percent. That placed him slightly behind Rubio, who won 27 percent of those Republicans with a four-year degree or more; Cruz came in third at 19 percent. Drilling down even more precisely, Trump beat Rubio by five percentage points among South Carolina voters with a four-year college degree, while Rubio outpaced Trump by more than double that among those with graduate degrees.

Trump’s showing among all college-educated South Carolina voters fell between his performance in Iowa (22 percent) and New Hampshire (where he carried a 30 percent plurality). Comparing his showing among college- to non-college voters, Trump has now run five points weaker in Iowa, 12 in New Hampshire and fully 17 in South Carolina. And though the margins have been thin, Rubio has now carried the most college-educated Republicans in two of the first three contests.

If Trump does face a ceiling in the race, it is likely constructed from the relatively greater resistance he is confronting among these white-collar largely suburban and often professional Republicans—who provided critical support for the past two nominees, John McCain in 2008 and Mitt Romney in 2012. In polls these voters have been somewhat less likely to support Trump’s signature proposals to deport undocumented immigrants and temporarily ban Muslims from entering the U.S.; perhaps more significant, those white-collar Republicans have also been more likely to question whether Trump has the personality, skills and above all temperament to succeed as president. As Jane Kizer, a computer programmer, told me while waiting to enter a conservative rally in Greenville last week: “He says what needs to be said, and what a lot of people are thinking. Now a lot of people are saying, ‘It’s been said, and we need a leader.’”

But after the first three contests, it’s an open question whether any of Trump’s remaining rivals can consolidate enough of the voters still skeptical of him to overtake him, particularly given his dominant position with the party’s blue-collar wing.

So far, Cruz’s coalition has been too narrow to threaten Trump. Throughout the first three contests, Cruz has been lopsidedly dependent on evangelical Christian voters—and even that foundation cracked on him in South Carolina.

In Iowa, Cruz won 34 percent of voters who identified as evangelicals but only 18 percent of those who did not. In New Hampshire, Cruz won 23 percent of evangelicals, but just eight percent of those who did not identify with that faith tradition. In South Carolina, Cruz carried 27 percent of those who identified as evangelicals and only 13 percent of those who did not.

Across all three states, Cruz’ performance among voters who were not evangelicals varied little from the meager showings by Rick Santorum (in 2012) and Mike Huckabee (in 2008), two previous evangelical favorites who failed to expand from that faction sufficiently. Relative to Santorum and Huckabee, Cruz was supposed to be better positioned to add economic and national-security conservatives to his evangelical base. But so far, Cruz hasn’t broadened more effectively than they did.

Viewed through the ideological lens, Cruz’ support is just as narrow: Although last night he carried a 35-percent plurality of voters who identified as very conservative, he attracted just 17 percent of somewhat conservative voters and 7 percent of moderates. In Iowa and New Hampshire, Cruz also won at least twice as much support from “very” as “somewhat” conservative voters; his support among moderates has not cracked double-digits in any of the three states.

And even as Cruz is relying excessively on evangelical voters, he saw that strength eroded in a two-front squeeze in South Carolina. On the one hand, as noted above, Trump crushed Cruz among evangelicals who lacked a college degree. Almost as striking, Rubio (at 28 percent) narrowly bested Cruz (at 24 percent) and Trump (22 percent) among college-educated evangelicals, according to the figures from the CNN polling unit. Cruz had comfortably carried those better-educated, born-again voters in both Iowa and New Hampshire.

If Cruz’s support is too narrow, Rubio’s backing so far has been too shallow. Recovering from his New Hampshire tailspin, Rubio in South Carolina won a little bit of everything. But even with the support of Governor Nikki Haley, Senator Tim Scott, and Representative Trey Gowdy, the Florida senator didn’t crack 30 percent with any key group.

Rubio in South Carolina won 22 percent of men and 23 percent of women; 23 percent of Republicans and 19 percent of independents; exactly the same 22 percent among voters who were evangelical and those who were not. Measured by ideology his support varied only between 19 percent (very conservative) 23 percent (moderate) and 25 percent (somewhat conservative.) Despite his message of generational change, Rubio ran about as well among voters older than 45 (22 percent) as those who were younger (25 percent). Education marked the most important gap in Rubio’s support: reversing Trump, Rubio ran better among Republicans with at least a four-year degree (27 percent) than those without one (17 percent). That almost exactly matched his split within the two groups in Iowa; in New Hampshire, Rubio plummeted among both.

Rubio’s relatively better performance with the college-educated Republicans more skeptical of Trump points toward the path most open to the Florida senator as the race proceeds. With Jeb Bush joining Chris Christie in ending his campaign, Rubio seems best positioned of the remaining contenders to consolidate more support from white-collar, mainstream conservative voters. Victor Poole, a retired quality engineer from Anderson, South Carolina, who attended a Rubio rally this week, personified his opportunity. Poole is uncertain about Cruz’s honesty and Trump’s volatility, and he finds Rubio simultaneously fresh and reassuring. “I feel like he would be our best choice,” Poole said. “I just really like all of his values. I look into his eyes and I trust him.”

But although Rubio can effectively appeal to better-educated voters with his promise of electability rooted in generational and demographic change, he continues to face competition for them from John Kasich. While Kasich so far looks like the moderate version of Cruz—relying too heavily on centrist voters—the Ohio governor can still pull away support Rubio desperately needs, particularly in the Midwest where Kasich is concentrating. Even more important, Rubio continues to target his message, with its strong religious conservative tone and sharp partisan attacks on President Obama, somewhat to the right of the voters who appear most available to him—on the theory that those in the center will ultimately prefer him over Trump and Cruz anyway.

The intersection of this electoral demography with the race’s geography points toward sequential showdowns for Trump. The cascade of contests through mid-March are dominated by Southern, border and Midwestern states where both evangelicals and blue-collar voters represent a major component of the electorate. That frames those states as likely the decisive face-offs between the front-runner and Cruz. The job for Rubio and (less plausibly) Kasich is to hold out by winning enough support to survive, particularly in the relatively more white-collar states voting over that period, including Massachusetts, Vermont, Georgia, Virginia, North Carolina, and Illinois.

After March 15, the calendar bends more toward more affluent, better-educated (and Democratic-leaning) states, especially along the coasts, where Trump might be more vulnerable. But if Rubio can’t deepen his coalition before then, and if Cruz and Kasich can’t broaden theirs, the only unknown in the race will be when Trump, the improbable front-runner, declares his victory.