With the Republican nomination race winding down, Nate Silver looks back and argues today that the key date on the Republican calendar this year was February 28, when Mitt Romney won in Michigan and Arizona. Given Michigan, he argues, Ohio, Illinois, and Wisconsin all follow — but had Rick Santorum won in those states, we have a very different race.

I disagree. Not on Ohio and Wisconsin; I think it’s quite possible that those states at least would have followed Michigan, although Illinois was a tougher state for Santorum, since Romney won there by over 10 points instead of the more narrow victories in the others. No, what I think is that a series of modest wins for Santorum in Michigan, Ohio, and Wisconsin would still have left him trailing Mitt Romney by a solid, albeit small, margin.

By Michigan, it was already clear that Santorum was unable to expand his campaign to a proper national operation. That meant no proper polling, which may have caused some real errors in where Santorum deployed the limited resources he had. It meant no proper delegate operation, which had visible consequences when he was bounced off some ballots or didn’t file full delegate slates, but also suggested that he was going to do worse in the caucus states than his caucus-night straw poll results would have predicted. But most important of all, it was reasonably clear by February 28, after Santorum’s victory in Iowa and his big day in Colorado, Minnesota, and Missouri, that conservative leaders were just not eager to jump on his bandwagon. It’s unlikely that flipping Michigan, Ohio, and Wisconsin into his column would have changed that, and therefore unlikely that he could have had the resources he would have needed to win a close elongated battle — much less had anything available to start a real stampede in his favor.

So if Michigan wasn’t the deciding state, what was? Silver says that the other obvious suspects are Florida, Illinois, and last night’s contests. Of those, I’d pick Florida, but I’ve also suggested a much less obvious candidate: South Carolina.

Yes, on the surface that was a terrible state for Mitt Romney, since he was clobbered there. But he was beaten not by potentially plausible nominee Rick Santorum or clearly plausible nominee Rick Perry, but by Newt Gingrich, who had already demonstrated in Iowa that his vulnerabilities made him relatively easy to destroy for any candidate with sufficient resources.

Had Santorum finished ahead of Gingrich in South Carolina on January 21, it would have established him as the clear alternative to Romney two weeks earlier, and with considerably more certainty, then his wins in Colorado and elsewhere two weeks later did. Gingrich would almost certainly have either dropped out or been marginalized at that point, having finished behind Santorum in each of the first three events. Two weeks may not seem like much, but perhaps it gives Santorum time to build a real campaign organization, meaning that he doesn’t give away nearly as many delegates. More importantly, by establishing in January, not in March, that he would dominate the South, Santorum might have received the resources he never got at any point along the way.

Maybe not; it’s possible that Republican party actors simply preferred Romney to Santorum, and did from the start. It’s also not at all certain that even a fully funded, well-organized Santorum campaign would have been able to do much more than shift the dial by a few points and, meanwhile, to at least win all the delegates he was entitled to. Which would have been nice, but not enough to actually win. What I do think is that after South Carolina, Santorum really never had a chance. And that’s why if the nomination was settled in the primaries, my pick for the key state was South Carolina.