The crumbling castle.

ON the 33rd anniversary of Roe v. Wade, the decision seems simultaneously to have become more sacrosanct than ever, and more imperiled than ever. It is supposedly well-settled as a matter of law. Even some former opponents of the decision believe that it has survived for so long that it should not be overruled. The op-ed pages are full of liberals who allow that Roe was never a well-reasoned inference from the Constitution, but say it's too late to let it go.

Since polls find the decision to be popular, it seems to be protected by both legal and political fortifications. Yet we are constantly warned that the president is chipping, chipping away at Roe, and that each of his Supreme Court appointments could be the last vote to overturn it.

There are ways of resolving the apparent contradiction. Maybe the explanation is that conservative Republicans are radicals, attempting to unsettle what's settled and zealous enough that they might just succeed. Or it could be that pro-abortion groups are crying wolf about the danger to Roe: Surely each nominee to the Supreme Court can't be the deciding vote against it. (Given the mercy an end to Roe would show to the unborn, perhaps "crying lamb" would be a more appropriate phrase?)

But we think that the best explanation is that Roe's apparent strength is largely illusory. Take those polls. Do they really mean that 66 percent of the public (to use the figure from a December NBC/Wall Street Journal poll) don't want the Supreme Court to allow state legislatures to be able to prohibit third-trimester abortions? Surely not: Polls find that even larger majorities want such prohibitions. Many people believe, wrongly, that Roe protects only first-trimester abortions, a misimpression that most polls (including the NBC/Journal poll) go out of their way to foster. Many people also believe, wrongly, that overturning Roe would automatically lead to a national ban on all abortions. What polls on Roe really measure is public opposition to an immediate national ban.

The relevance of these polls to actual political behavior is tenuous. The public's support for Roe does not even translate into opposition to the confirmation of Supreme Court nominees who might vote against it--as public support for John Roberts and Samuel Alito makes clear.

Nor is Roe all that well-settled a precedent, which is perhaps what occasions all the somewhat nervous claims that it is. The last time the Court really reconsidered the issue, in a 1992 case called Casey, ...