If Sanders is about to lose Washington, he’s in deep trouble — and he knows it. Politico calls it a sign that Sanders faces a “total crumbling of his base”:

In Washington, a progressive-leaning state that Sanders carried by a massive margin in 2016, two recent polls — both fielded after Super Tuesday — show Biden erasing the Vermont senator’s advantage. Biden led narrowly in each, though within the margin of error.

If a loss in Michigan would suggest a weakening of Sanders in the Rust Belt, an upset in Washington hours later would suggest a total crumbling of his base.

“I could see it going either way,” said Sandeep Kaushik, a Democratic consultant who is based in Washington and unaffiliated in the presidential campaign. “The last week has completely reshaped this race. …the momentum coming out of Super Tuesday.”

Sanders is still widely considered likely to win. But a Biden upset – which once seemed almost unthinkable – would no longer be shocking, according to more than half a dozen strategists working in Washington or with experience in that state. And the race is likely to be closer than originally expected, no matter who carries the state.

The two polls in question are from SurveyUSA (Biden 36/35), whose previous iteration had Sanders edging Biden 26/21 in a broader-field questionnaire in late January, and Democratic pollster Data for Progress (Biden 47/44). DfP did a pretty good job in its Super Tuesday polls, and in this case the two polls tend to corroborate each other. Both polls are more accurately described as virtual ties, at least on the toplines.

In the demos, however, the two polls have some striking differences. In the DfP survey, there is a wide gender gap between the two candidates, where Biden wins women 56/36 and Sanders wins men 55/37. That gap doesn’t show up at all in SUSA; Biden wins men 37/32 and Sanders edges out a lead with women, 37/35. That’s a very curious and significant difference between the two polls, part of which might be explained by the “some other Democrat” option in the SUSA poll, which gains 13% of all respondents.

The two surveys are more consistent on the age gap, which is yuuuuge. DfP splits their sample at age 45, and Biden loses the younger demo 15/73 while winning the older demo 63/30. SUSA splits it at 50, and Biden also loses the younger demo 29/48 and wins the older demo 44/21. That might point to a big problem for Democrats when it comes to the general election with Donald Trump, where young people might be their best opportunity to beat the incumbent. Bernie might not be turning out the youth vote like he promised, but Biden isn’t exciting younger voters at all.

Still, the toplines come down to a virtual tie, as noted above, but that’s no longer good enough for Sanders. Had Sanders won a couple of more states last week or gotten the kind of massive delegate advantage in California some had predicted earlier, a virtual tie in Washington would have been merely a minor annoyance that still maintained a status quo that would have favored Sanders in a multi-candidate race to the nomination. Now, however, Sanders really needs a big delegate haul out of Washington to have a chance to catch up to Biden. These numbers don’t necessarily show a “total crumbling of his base,” as Politico remarks, but it’s showing a narrowing down to Sanders’ base when confronted by a binary choice in the primaries.

That doesn’t mean it’s over, not even in Washington. Sanders has a better ground organization, and this time he won’t get caught with his pants down in any of the six states tomorrow. Sanders could still eke out a win in Washington, but probably not on the scale he needs to keep Biden from building an insurmountable advantage in both numbers and momentum.