When the polls aren’t going his way, Donald Trump likes to toss them aside by declaring a “silent majority” of Americans will rise up on Election Day and deliver him to the White House. The claim seemed silly just last week, when Trump’s path to an electoral college victory had all but vanished. Since then, headlines about the FBI’s renewed look into Hillary Clinton’s emails and some tightening national polls have high-strung Clinton fans fearing that Trump might be on to something.

At first blush, what Trump is suggesting seems not altogether impossible. If he were to win, he’d owe it to the voters that form his strongest core of support—non-college educated whites—turning out in surprisingly huge numbers. There is plenty of room for them to improve, considering only 60% of the demographic tends to show up to the polls. Comparatively, college-educated whites, who look primed to tip Democratic this year for the first time in the history of modern polling, participate at a much higher rate, with four of five typically casting ballots.

So will a working-class white surge materialize that shocks pollsters and secures a Trump victory? William Frey, a senior fellow at the Brookings Institution, ran the numbers and found that the answer, in short, is no. Even under a worst-case scenario, in which FBI director James Comey’s decision to revive the federal probe of Clinton’s emails fires up Trump’s core, he’d need more support from groups with which he’s consistently underperforming.

First, to establish a baseline, imagine that non-college educated whites turn out this year in the same numbers as 2012 and break for Trump by a 32-point margin, 10 points higher than they went for Mitt Romney in the last election. If Trump and Clinton then split college-educated whites while minorities go Democratic by a 64-point margin (as they did in 2012), Frey finds Clinton would win by 6.6 million votes nationally, or 4.8%. That would translate into an electoral college wipeout for Trump.

Following those same assumptions but boosting turnout among working class whites to match their college-educated counterparts—a nearly incomprehensible 22-point jump in their participation—Trump would eek out a popular vote win of less than 500,000 votes, or roughly .3%, according to Frey. But under that scenario, Trump would only pick up two more swing states—Wisconsin and Iowa—and lose the electoral college, 245 to Clinton’s 293, assuming other states broke as they did in 2012.

Frey’s conclusion: Even under the best-case conditions for the Republican nominee, “an extraordinary turnout of white working class voters, by itself, cannot produce a Trump victory.”

Deep breaths, Clinton supporters.