Mitt Romney has turned in another underwhelming performance in the Republican primaries, somehow managing to come third in contests in both Alabama and Mississippi. Rick Santorum won both contests, narrowly beating out Newt Gingrich in each.

Santorum won the state of Alabama with 34.5% of the vote, a clear lead over both Gingrich and Romney, who were at 29.3% and 29% respectively. In Mississippi, Gingrich came closer with 31.2% to Santorum’s 32.8%, with Romney trailing at 30.6%.

Romney won Hawaii clear with 45% to Santorum’s 25% — although no more than 10,000 votes were cast in the whole island state — and took the territorial islands – Northern Marianas, Virgin and American Samoa — with votes ranging from 66% to 100%.

But it’s the two southern states which matter, together with Kansas — which Santorum won on the weekend with 51% to Romney’s 21% and Gingrich’s 14%. No-one had thought that Romney was a shoo-in for the southern states, but they thought he would be competitive — especially since he continues to outspend Santorum by four-to-one.

Given the weird math of the primaries, Romney will come out of yesterday’s contests with around three less delegates than Santorum, his numbers boosted by the disproportionate share of delegates given to the territories and Hawaii.

The result presents the GOP with a “front-runner”, whom no-one really likes and who has no base, save for places where the GOP is a minor presence — and an “outsider” candidate whose victories are solid, not squeakers, occur in states which have a “heartland” cast to them — Iowa, Michigan — and who is still running on a relatively low spend.

Even Romney’s victories in the two key swing states – Ohio and Florida – count for little; Florida because everyone has given up on the state as anything other than a crapshoot, and Ohio because Romney’s huge spend could net him only a three point win.

Now, the cautious ruminations about a split or brokered convention are getting louder. Initially they were dismissed as “make-news” by bored and desperate cable-news TV. Then after Santorum’s hat-trick of Minnesota, Missouri and, especially, Colorado, there were a few cautious words about the real possibility of such, despite its status as a long shot.

Now, though people continue to note pro-forma that Romney will likely get the magic 1,144 delegates by May, the discussion is getting louder, and the possibility becoming real for many.

Why?

Well it comes down to the now psychotically complicated Republican primary rules, a product of state-by-state complexity combined with the national committee’s too-clever-by-half meddling. In 2008, as hitherto, most Republican primaries were winner-take-all, while a fair slew of Democrat ones were proportional. This was taken as a reflection of the parties — the GOP all steak and Ramada — hooker-bl-wjob winner-take-all alpha types, the Democrats whiny, guitar playing, everyone-gets-a-pony wimps.

Indeed, this trope was so amusing that the GOP didn’t notice that the wrapping up of their process by February, and the Democrats’ Barack-Hillary battle all the way till June, gave the latter literally billions of dollars worth of free coverage, while McCain was speaking to Nebraska Shriners Clubs. Though some of the Barack-Hillary exchanges were a little bruising – Hillary’s attack on Obama’s lack of experience – crucially, neither of them went at each other from the Right, thus maintaining a basic unity of the message.

So this time round the GOP gritted their teeth, and transformed the bulk of their contests into proportional ones, hoping that the hard-right candidate, together with Ron Paul’s insurgent campaign, to keep the kettle whistling but not boiling over. Most of these contests are trigger proportionals — with the delegates counted by congressional district within the state, anyone who wins more than 50%+1, takes all three delegates for that district.

Thus in Virginia, where Santorum and Gingrich — guys who want to control, from Washington, the moon and every American v-gina — couldn’t manage to get on the ballot, Romney won all 46 delegates, despite Ron Paul taking nearly a third of the votes.

That’s the way it was supposed to go in a bunch of states, with enough genuine contests to string it out. What it wasn’t meant to do was foster a right-wing candidate — Santorum as it turned out, but it could have been anyone — delivering so many wins, and near-losses, together with a third runner building up some numbers, that the race became a question of “can the designated front-runner actually gain the support of his party?” Rick Perry was pitied for his embarrassing brain freeze in the debates, when he couldn’t remember which three federal departments he was going to abolish. But that was only fifty seconds.

The humiliation of Mitt Romney is now into its second month. But it’s worse than that. With the sudden realisation that they were in with a chance, both Santorum and Gingrich began attacking from a populist left position, decrying Romney as a vulture capitalist.

This was a major step because it messed with established, barely conscious, rules around primary argy-bargy — that the voting public will more or less wipe the slate clean when it comes to personal insults, accusations around character, specific policy etc, so long as the party gives a simple and consistent message as to where it stands.

That is something the attacks on Romney — for making a lot of money in a “free” market — have sundered. When added to his unforced errors and gaffes, it has created the second part of this paradox — a front-runner who has that status, not because he actually polls better against Obama (both Santorum and Paul nip him to that about half the time), but because he polls as the candidate people think is the most electable — quite a different thing. That’s a double whammy that many Republicans might find too difficult to ignore. And that’s where some difficult delegate maths come in. At the moment Romney is still on track to get the magic 1144 delegates before the convention — he has 491 delegates, with Santorum on 233, Gingrich at 157, and even Ron Paul taking 78.

Romney is hoping to consolidate that fairly dismal total with a series of solid metropolitan and northern states — Illinois, Maryland, Wisconsin, New York and Connecticut coming up, then Indiana – with a solid 320 or so delegates on offer. After that there’s another 225 or so in California and New Jersey.

You could give Romney 500 of the 600 or so votes, and say fifty per cent of the proportional contests he’s facing tough races in — Texas, Pennsylvania, West Virginia, North Carolina, Arkansas and Kentucky — which is about 150 out of 300. That would nudge him just over the 1144 mark — but only by June. If he hasn’t reached it by California, there’s Montana — with a strong Mormon population — and New Mexico with a Latino base, and, finally, Utah, to carry him over, with about 1250 delegates.

However there’s two problems here. The first is that there are very few genuine state-wide winner-take-all competitions here. Most — including the big scores of NY and California — are district-by-district. With a commanding candidate, such states usually become a clean sweep. But the more fractured the race gets, the more that separate congressional districts may diverge — backwoods upstate New York from NYC, inland California from the coast and so on. If you take away a hundred or so delegates from Romney’s count, then he would be struggling right up to Utah, and even that might not take him over.

Could it get any worse? Why yes. Because the 491 delegate count is what we call the ‘soft’ delegate count. It includes both delegates pledged to a candidate by the state rules of the party convention — but also the delegates assumed to be voting for the candidate, but not fully bound or pledged. Thus, for example, after the Iowa caucus, both Romney and Santorum were given six delegates each, based on their vote. But the party convention is not bound by those votes — it is delegates to the state convention who have been chosen, not delegates to the national convention. The delegates gained the vote by arguing for candidate X or Y, but the state convention — it varies state by state — can throw the vote to anyone they want. Furthermore, come the convention, some of them are only nominally pledged to the state’s choice of delegate in any case.

When you strip out these votes, you get the hard delegate count: Romney 374, Santorum 160, Gingrich 133, Paul 23 — and Uncommitted 250. That puts Romney a lot further away from a lock down pre-convention win. What he’s relying on, and aiming for, is the rhetorical victory of a soft delegate count win.

But there’s a reason why the real — hard — delegate count is worth paying attention to, and that is the conditions of release.

Fully pledged delegates can’t vote for anyone other than their candidate, unless their candidate bows out and “releases” his/her candidate. But assumed delegates, or whole states, could break away at the first ballot, if a front-runner has not emerged.

In the case of Romney, it may break if polling and performance is so appallingly bad that enough delegates, or leaders of state delegations — the real power — decide to jump a different way. Indeed one of his cornerstones of victory — Illinois — is what’s known as a “loophole” primary — ostensibly, the whole state delegation is tied to a specific candidate, but can be released instantly due to an obscure rule.

The nightmare for the Republican “establishment” — what remains of the rational political calculators at the centre of the party — is that Santorum or Gingrich would strike a deal, and spill the convention immediately, if Romney’s numbers were in the no-man’s land between the soft and hard pledged delegates.

That would set the national convention up for three days of chaos — indeed, given that it has been shortened by a day, there is the prospect that the whole thing would sail close to chaos. Once again, hard-headed political operatives do everything to avoid such a crisis, and say it will never happen.

But the people milling around Santorum and Gingrich aren’t hard-headed — they’re a political cult, and they can taste the chance to do what they had a crack at with 2008’s Huckabee campaign, which is to bang a wedge into the party and then pick up the pieces.

They would not care about losing to Obama — so long as they could keep hold of the party through the next several election cycles, they could last out a “New Deal” season of 16-20 years of Democratic Presidents, and wait until the party is so discredited by power that the US public will take a plausible conservative.

Were that victory to coincide with control of the House — and if they were lucky, death or departure of a liberal supreme court justice or two — then they would have a chance to truly remake post-New Deal US once and for all. As the IRA used to say, “we only have to be lucky once”.

So for all these reasons, it’s worth keeping a close eye on the primaries. Romney is still the favourite, but one would suggest that it’s coming in fast. He’s uniquely vulnerable on numbers, on political grounds and on personal issues. If there were going to be a full-bore crisis in the party, this year offers a unique syzygy, and after the ‘bama and ole Miss, it’s on like a gong.