Geographic Severability and Linguistic Severability

By Mike Dorf


Last week's Supreme Court decision in Perry v. Perez has been widely portrayed as a victory for Republicans in Texas, and thus in Congress, and so it probably is.  But it is also a neat little puzzle that may be useful for exploring the broader notion of severability.

To oversimplify, the Supreme Court held in Perez that the district court erred in drawing its own electoral districts by giving insufficient weight to the districts that the Texas legislature had enacted based on 2010 census data.  It's true, the SCOTUS acknowledged, that parts of the enacted plan could violate Section 2 of the Voting Rights Act and/or the Constitution.  And it's also true that the enacted plan has not yet survived pre-clearance under Section 5 of the Voting Rights Act, pursuant to a separate proceeding in DC.  But--and this is the core holding of Perez--the district court still should have given greater weight to the lawful aspects of the plan drawn by the legislature.

How would that work?  Suppose for simplicity that Texas has ten districts and that the eastern half of the state is homogeneous, whereas the suspected hanky-panky has all occurred in the western half of the state.  Thus, in this simplified hypothetical version, the districting plan drawn by the legislature looks like this:
I understand the Supreme Court in Perez to be saying that the district court ought to accept districts 1 through 5 and then adjust the border lines among districts 6 through 10 to undo suspected unlawfulness.  As I read the Court's opinion in Perez, the district court would be required to leave districts 1 through 5 as is, and even in re-drawing districts 6 through 10, try to retain as much of the original map as it can, without retaining the unlawful bits.  Thus, acceptable legislative judgments -- like a decision by the legislature not to split San Antonio into multiple districts -- would have to be respected in the judicially re-drawn map, even though San Antonio falls in district 7, which is problematic in other respects.  Exactly how the district court will accomplish this goal remains to be seen, but I want to focus on the implications of Perez for other sorts of cases

Note that the remedial move the SCOTUS makes in Perez  is closely analogous to another move in constitutional law: The notion that, when a court finds a law unconstitutional, it should try to preserve as much of the law as possible by severing -- that is, cutting off -- the invalid portions of the law, and leaving the valid portions in effect.  But here's the thing.  When courts do that, typically they don't then re-write the invalid portion.  They just declare it invalid and leave any fix for the legislature.

Let me explain what I mean with another schematic hypothetical.  Suppose Texas had the following law:
Texas Criminal Sodomy Law 
It shall be a felony for any person to commit sodomy with another person: 
a) If he or she lacks the consent of the other person; or 
b) If he or she has the consent of the other person.
Suppose the law is challenged by John Shmawrence and Tryon Shmarner, two adult men who face criminal charges under Part (b) for engaging in consensual sodomy.  Suppose, moreover, that their case makes it to the U.S. Supreme Court.  Under Lawrence v. Texas, Part (b) is unconstitutional, so Shmawrence and Shmarner have a good constitutional defense.

What would be the consequences of the Supreme Court ruling in Shmawrence v. Texas?  Ordinarily, even in a so-called "as-applied" rather than "facial" challenge, the precedent set by the Shmawrence case would be that Part (b) is unenforceable, while Part (a) would remain enforceable.  That result makes sense both because Part (a) is not at issue in the Shmawrence case and because Part (a) looks valid as a prohibition on a form of rape.

But now consider what would happen if the approach the Supreme Court adopted in Perez were to apply in a case like Shmawrence.  Now the Court -- after declaring Part (b) unenforceable against Shmawrence and Shmarner in the circumstances of their case -- would have to see whether it could re-write Part (b) in a way that preserves the valid goals the legislature was trying to accomplish with Part (b).  Are there any?  Perhaps.  We might infer from the overall statute that the Texas legislature doesn't like sodomy at all, but especially doesn't like non-consensual sodomy.  Suppose Part (b) were re-written as follows:
b') If he or she has the consent of the other person, where that other person is a minor and the person committing the offense is not a minor.
Now Part (b) looks like it can be upheld as a criminal prohibition of statutory rape.  (I am putting aside equal protection concerns that might arise from the legislature's singling out of non-consensual sodomy and sodomy with minors rather than involuntary sexual acts and sexual acts with minors more generally.  Let's assume that other Texas laws criminalize those other sexual acts to the same degree.)

But in fact, the Supreme Court's severability doctrine does not require courts, upon finding a particular statutory provision invalid and severable, to re-write the invalid provision so that it is valid and preserves as much of the legislative intent as possible.  Quite the contrary, the doctrine pretty clearly condemns such a move as a judicial usurpation of the legislative role.

So why does the Court in Perez mandate in the geographic realm what it forbids in the linguistic realm?  The difference is not the difference between maps and words.  The difference appears to be one of timing.  After a court invalidates a provision like b) in a case like Shmawrence, it's up to the Texas legislature to re-write b) as b') if it so chooses.  But in a case like Perez, with the election nearly upon us, there isn't time to send the Texas legislature back to the drawing board to try to come up with a new, legal, map.  There must be some map in place on election day (which is April 3 for the primaries), and if the court doesn't draw one, then there is too great a risk that there simply won't be one in time.

Are there circumstances in which a similar problem results from the invalidation of statutory language?  Perhaps.  Suppose that some bit of statutory language is unconstitutional because enacted for an illicit purpose (racial or gender bias, say), but that invalidating that language would leave a legal vacuum of the sort that the NY Court of Appeals worried about in People v. Liberta, where giving the law's challenger what he wanted would have meant there was no valid law on the books forbidding rape until the legislature re-enacted the rape prohibition.  Under such circumstances, one solution would be for the court to abandon the general prohibition of re-writing the invalidated provision to fill the void, because the legislature, if it disagrees with the judicially-created replacement, can always revise the law afterwards.  The court in Liberta did something very much like this, quite sensibly in my view.

Once we recognize from cases like Perez and Liberta that sometimes courts should be permitted to re-write maps and laws, we may have reason to doubt the more general practice by which courts simply invalidate the offending provisions, without trying to reconstruct a law that would accomplish at least some of what the legislature was trying to accomplish.

To take just one currently salient example, suppose the Supreme Court were to rule the minimum coverage provision (MCP) invalid in the pending health care litigation.  So far, of all the judges to have considered the issue, only one, Judge Vinson, found that the MCP was invalid and non-severable from the whole of the law.  But given the arguments that the Justice Dep't advances for the necessity of the MCP to many other aspects of the law, it is quite possible that the SCOTUS, if it finds the MCP invalid, would find it non-severable from at least some substantial portion of the rest of the law.  Yet, if the logic of Perez and Liberta were to apply, then the Court, upon finding the MCP invalid, would be obliged to "re-draw the map," and come up with a version of the MCP -- one that is indisputably a tax, say -- that would be valid.  I don't believe the Court would actually do this, but I do think that, in light of the logic of Perez and Liberta, the case against doing so is weaker than ordinarily assumed.