In short, every congressional attributes a number of delegates -- between three and eight. It is very difficult to get an extra delegate in an even-numbered district. For example, in a 4 delegate district, the candidate split the delegates 2-2 unless one of them breaks 62.5%, in which case the allocation is 3-1. In a 6 delegate district, the allocation is 3-3 unless a candidate breaks 58%, in which case it's 4-2. In odd delegate district, a one vote margin in that district gives the victor an extra delegate. So if a candidate gets 50.01% of the vote in a 5 delegate district, that already c reates a delegate gap.

The delegate rules means that Illinois and New York will be big stories tonight: How will each candidate fare in the other's home state? They are both likely to cross 15% of the vote everywhere (though there is a chance that Clinton fails to cross that in the 3 African-American districts of Chicago, which are some rare 8 delegate districts, and that could give some massive leads to Obama), but will they cross the 32% or 27% thresholds that will enable them to force a split of delegates in even-numbered districts?

Obama could be stronger at this game, as he looks strong enough in New York that he is likely to cross 30% in a lot of districts that Clinton will win handily. In other words, Clinton could get 65% of the vote in a district... but only 50% of its delegates! For district-by-district (literally, all of them) analysis and number crunching, take a look at Cook Political's wonderful chart of the delegate allocation scenarios. I have used this chart in my analysis, and will definitely refer to it throughout the night to keep tabs on what is going on.

In short, Clinton has more to lose in this strange delegate game. And this for a very simple reason: The constituency that is breaking the most heavily for a candidate is the African-American community. This means that in predominantly black districts -- and there are many throughout the country -- Obama is likely to come out with more delegates, even if the district has even number of delegates since Obama is getting more than 70% of the black vote. On the other hand, even if Clinton wins more consistently across other districts, she is far less likely to break 60% and get that extra delegate. She could win the state overall but lose the district-level delegate count.

Let's take a look at Tennessee to illustrate this strange reality. Tennessee has 9 congressional districts, that attribute 44 of the state's pledged delegates. Most polls have Clinton up big in this state (about 20%), one of her safest strongholds. But a district-by-district look reveals that Clinton has got to be worried.

• The 9th district which is 59.7% black (which means an even higher proportion of Democratic primary voters). It is also a 6 delegate district, and Obama should be able to break the 59% he needs for a 4-2 split.
• The first, second and third districts all allocate four delegates. And while Clinton is likely to win big in all of them, she needs to break 62.5%. That will obviously be very hard for her to accomplish, especially given that Obama can count on a minimal black base in all those places, so that she could be forced to come out of these 3 districts with a 6-6 split.
• The 5th district is 23% black and has 6 delegates. Obama is likely to win there, and he could be forced to split the delegates. But he is more likely to break 59% than Clinton is to break 62% in the first 3 districts.
• That leaves Clinton with the 4th, 6th, 7th districts to get any sort of delegate lead. All of these have 5 delegates, which makes it near impossible for her to win by a big enough landslide to get a 4-1 edge (once again, the black vote should provide Obama with a strong base). In other words, these three districts combined could give Clinton a 3 delegate lead -- one each.
• Finally, the 8th district looks to be more disputed between the candidates. It is a 5 delegate district, so that the winner will get an extra delegate even if he only wins by a vote.
  
• The bottom line: Obama could lose the state by 15% or more but tie Clinton on the district delegate count. By opening up a 2 delegate margin in the 9th and perhaps the 5th district while forcing Clinton to split the delegates in her own strongholds (the first 3 districts), Obama would at worst trail Clinton by 1-2 delegates, at best lead her by as much. And all of this while losing the popular vote by 15%.

Clinton should still get a delegate lead out of Tennessee since there also are 24 delegates that will be attributed statewide proportionally, and Clinton's victory margin will be big enough to get her a few extra delegates there. But the odd/even district reality will make it hard for her to get a clear delegate lead out of Tennessee even if it's one of her strongholds!

Now, this is in great part due to which districts are odd and even. If the 1st, 2nd and 3rd were odd numbered and the 9th was even numbered, Clinton would look much better in Tennessee. And it is absurd to think that tonight's winner could be determined by such minor factors as to which districts have 3 or 5 delegates, and which have 4 or 6.

And Obama could get in his own share of trouble. Consider Delaware's Wilmington section (Deleware is one of the only 2 states that has specially designed jurisdictions to allocate delegates). Wilmington is 56.4% African-American, so even more among Democratic primary voters but... it has only 2 delegates. In other words, if Clinton crosses the viability threshold of 15% (which she could very well do given that most polls have her getting 20% of the black vote, and she could also depend on the district's 10% Hispanic voters to supply her with whatever else she needs to pass 15%)  she will get one delegate, as many as Obama. That's right, Clinton could get 15% of the vote and force a 50-50 delegate split!

I will do my best to keep you updated on the delegate count throughout the night, thought it will not be easy to get an idea of what is going on considering that all votes really need to be counted for delegate allocations to happen. It is not enough to know who won the district; it is more important to know whether the winner crossed 59% or 62.5%! And that is not something we are likely to know early in the evening.

In fact, California officials are now warning that it will be impossible to get an idea of the district-by-district delegate allocations without having all votes counted, and that might not happen until Friday given the high number of absentee ballots. No definite delegate numbers from California until Friday? Think about that.

Analysis cross-posted on my blog, at Campaign Diaries.