Sen. Mary Landrieu (D-La.) faced an uphill fight for reelection from the outset. Polling averages had her leading last week's free-for-all jungle primary forever, but have consistently also shown that Landrieu would lose to Republican Bill Cassidy if the race came down to the two of them.

Now it has come down to the two of them.

Landrieu won last week's race, in the sense that she got more overall votes. She cleaned up in the populous parishes (or, as Americans call them, "counties"), including demolishing Cassidy in New Orleans. The parish-by-parish county-by-county map at left below shows the relatively close match-up across the state, which Landrieu won by under 20,000 votes.

But keep an eye on that second map.

2014 candidate results

2014 party results

The so-called "jungle primary" (or "primary," as Americans call it) included a number of candidates. The most prominent non-major candidate was the Tea-Party-backed Rob Maness, who pulled in thousands of votes in some areas. So the second map adds up all of the Republican and Democratic votes in each county and compares them. There, Landrieu comes up very short, as you can tell from the second map getting quite a bit more red.

Even if Landrieu were a Republican incumbent, last Tuesday was bad news. Usually when people pull the "X percent of the state opposed Candidate Y," it's because Candidate Y's opponent is trying to put a good face on getting beaten badly. But in a race with a well known incumbent, that so many people voted for someone else is not a good sign.

Another way to look at it is below. It compares the per-county data above with the per-county split in 2010, when Sen. David Vitter (R) ran for reelection. The overall party split last Tuesday looks a lot like Vitter's county splits. He won by almost 240,000 votes.

Click through. Blue counties flip red.

But but but! you say, this is percentages, not population! Plus turnout in the run-off will be different! That's true; you're pretty smart.

So let's figure out how turnout will change. The answer is: It will go down. In 2012, a House run-off race saw a 69 percent drop in turnout. But that was from a presidential race. If you figure that turnout last week was only 73 percent of 2012 turnout statewide, we can assume that turnout in the runoff will be about 50 percent lower than last week. (Well, 42 percent, specifically, but you knew that, smart guy.)

If you were to flatly assume that every voter who voted for a Republican last week will vote for Cassidy and every voter who backed a Democrat will vote for Landrieu, Cassidy would win by 180,000 votes. If you lop half of the turnout off of that you get ... Cassidy winning by 90,000. (Makes sense, right?) That's assuming that the turnout is the same party split. It's probably more likely to be more-frequent voters, meaning older voters. Which probably means more Cassidy voters.

Can Landrieu make up the difference with turnout? Sure, in theory. In theory, anything could happen. But given that the Senate Dems are bailing on the race, the odds that that particular anything would happen are very, very low.