While I’m on the subject of caucuses, it’s worth taking a quick look at a story in today’s Des Moines Register explaining why the Iowa caucuses—like all the others—aren’t quite yet settled.

I’ve hit on this before, but caucus nights (the night the campaign press calls a state for one candidate or another), are merely, to use the party’s term, the “first determining step” towards divvying up a state’s national convention delegates. Tomorrow, Iowa’s precinct level delegates (won way back on caucus night; January 3) will gather at county conventions, where they’ll recaucus and send delegates to the state convention and congressional district conventions, where the Denver-bound delegates will finally be chosen via, essentially, another caucus.

Now technically those delegates are allowed, at any step along the way, to switch horses. But they rarely do since they’re likely to be the most devoted supporters of their candidate. Because of that stability, candidates and the press can make decent—though not rock solid—projections of the final national delegate count, even though, in Iowa, they won’t officially be determined until June 14.

But this time, the Register writes, things could be different. John Edwards, long out of the race, won 4,207 delegates from the precinct caucuses. While people once affiliated with his campaign are urging those delegates to “stand up for John,” both Clinton and Obama are working to win them over and goose a handful of extra national delegates.

The DMR story does a pretty good job of explaining what’s at stake as this all wends its way forward. But given that level of understanding, the mistake in the below graf isn’t just puzzling—it undermines the entire point of the article, namely that the national delegate allotment remains up in the air:

Obama, an Illinois senator, won the caucuses on Jan. 3, capturing the equivalent of 38 percent of Iowa’s national Democratic delegates, ahead of former North Carolina Sen. John Edwards with 30 percent and Clinton, a New York senator, with 29 percent.

Again, nobody won any “national Democratic delegates” on January 3. They only won precinct level delegates to the county conventions. And those are the percentages that the DMR is reporting, even though it’s labeling them incorrectly. Admittedly, this is complicated. But that’s and obvious error, and if any outlet should be expected to get all the details right, you’d think it would be The Des Moines Register.

Now, if you were to actually look at the projected “national Democratic delegates” that emerged from caucusing on January 3, you’d get a rather different set of percentages: 35.5% for Obama, 33% to Clinton, and 31% to Edwards.

That’s right. The widely reported numbers, repeated and mislabeled in today’s DMR article, suggest that Clinton came in third. And, for whatever it’s worth, she did back on January 3. But in projected Denver delegates, what actually determine the nomination, she came in second. (The estimates were Obama 16, Clinton 15, and Edwards 14.) This is because Iowa gives more national delegates to some congressional districts than it does to others. So depending on how the math breaks, delegates won in certain districts can be worth more than others.

No one has a real idea of what the “popular vote” was on January 3, as the Iowa Democratic Party refuses to release the data. So the national press widely reports the county delegate numbers as a sort of proxy popular vote. It isn’t accurate, but for momentum-focused reporting, I guess they thought it was good enough.

A very similar thing happened after Nevada’s “first determining step” on January 19, when Clinton edged Obama in the precinct caucuses, but because he had greater support in the right places, he edged her in projected national-delegates. While it still wasn’t explained well, it was at least noticed.

Clint Hendler is the managing editor of Mother Jones, and a former deputy editor of CJR.