Ban Them All; Let the Courts Sort Them Out. Saving Clauses, the Texas Abortion Ban, and the Structure of Constitutional Rights

Essay - Online Edition - Volume 100

SB-8—Texas’s new abortion ban—was designed to evade constitutional review. It was drafted to preclude pre-enforcement suits brought in federal court under Ex parte Young. Yet constitutional review is coming, either in Texas’s state courts or in a new federal lawsuit brought by the United States directly against the state itself. The courts will then have to decide: Is this ban constitutional? This substantive question has gotten surprisingly little attention. Most commenters believe that the statute’s broad ban on pre-viability abortion is obviously forbidden, at least as long as Roe and Casey remain good law.

But the statute has one more trick up its sleeve: a saving clause that allows individual defendants to escape liability if they can show that imposing it would create an “undue burden” on abortion access. ‘Undue burden,’ of course, is Casey’s magic phrase—a test used to evaluate the constitutional sufficiency of certain abortion restrictions. The statute thus attempts to mount a tautological defense against unconstitutionality: It forbids all abortions, except those that it would be unconstitutional to forbid. Surely, a statute with that structure cannot pose any constitutional problem. Or so the argument will go.

This Essay shows why the saving clause does not save the statute. There are three reasons. First, the saving clause does not do what it says. Its protection against undue burdens is substantially narrower than Casey requires. Second Roe, not Casey, supplies the correct test here, rendering the saving clause a non-sequitur. Finally, even if Casey supplied the right test, and even if the statute actually implemented it, that itself would doom the statute. SB-8’s saving clause defines the law’s restrictions in terms of what is constitutional. But as will be shown, figuring out what is constitutional first requires knowing what the law restricts. This recursion produces an infinite loop, rendering the statute’s restrictions literally undefined. SB-8 is thus a nullity and unenforceable on that ground. The only way to make the statute a non-nullity would be to ignore the saving clause, which would immediately render the law unconstitutional.

I. Statutory Scheming

Most of the legal commentary so far on SB-8, Texas’s new abortion ban, has focused on its procedural peculiarities. The Supreme Court’s refusal to enjoin the law turned on the procedural requirements for issuing preliminary injunctions.[1] And the Court’s application of the preliminary injunction standard turned on deeper procedural questions of jurisdiction and justiciability.[2] These conundrums did not arise by mistake. The statute was intentionally drafted to throw a wrench into otherwise available processes of pre-enforcement review.[3] And for the moment, the tactic seems to have worked. The Court refused to issue an injunction, and the law went into effect.

Now, however, the United States has sued Texas directly to enjoin enforcement of the law. And in a suit between two sovereigns, the statute’s procedural roadblocks may fall away. Even if they do not, SB-8 will eventually be enforced as the statute intends—in Texas state court—where the defendant will surely raise a constitutional defense. Thus, the courts will soon have to decide whether the law is, on the merits, constitutional.

That question, as compared with the procedural ones, has gotten little legal and scholarly attention. Indeed, most commentators simply assume that Texas’s law is unconstitutional, at least as long as Roe v. Wade[4] and Planned Parenthood v. Casey[5] remain good law.[6] The statute certainly appears to flagrantly disregard those cases’ holdings. It begins by imposing $10,000 in liability on anyone who performs, aids, or abets an abortion of any fetus with a detectable heartbeat.[7] It also requires injunctive relief to prevent defendants from engaging in such acts.[8] Fetal heartbeats generally become detectable at around the sixth week of pregnancy, long before the fetus could survive outside the womb.[9] This is, then, a ban on almost all abortions, including almost all pre-viability abortions. And Roe squarely held that states lack a “compelling interest” sufficient to justify bans on pre-viability abortions. Q.E.D., then; the law is plainly unconstitutional.

Or is it?

SB-8 has one more trick up its sleeve. The statute includes a saving clause allowing any defendant to escape liability if imposing it would place an undue burden on some woman seeking an abortion.[10] The phrase, ‘undue burden,’ is cut and pasted from Casey, where the Supreme Court held that laws imposing such burdens are unconstitutional.[11] Thus—the Texas Legislature clearly hopes—the ban cannot be unconstitutional because it applies only in cases where the constitution permits it.

This argument rests on an appealing logical symmetry. Indeed, the inference appears tautological: How can it be forbidden to ban all abortions except those which it is forbidden to ban?

Yet the argument is wrong. The saving clause does not save the statute from unenforceability. There are at least three reasons for this. Two are straightforward. One requires understanding the deep structure of constitutional rights—specifically, the rights protecting individual activities like speaking, defending oneself with a firearm, and obtaining an abortion.

First, the saving clause fails to save the statute because it does not actually do what its tautological reconstruction promises. While the clause initially appears to eliminate liability pursuant to Casey’s undue burden rule, this is a ruse. The saving clause has its own definition of undue burden, which is much narrower than Casey’s. Thus, the statute’s protections are much weaker than the Constitution, as interpreted in Casey, demands.

Second, Casey’s undue burden framework is the wrong test for evaluating SB-8’s constitutionality. Roe and Casey both hold that pre-viability abortion bans are subject to “strict scrutiny,” and that the government lacks a compelling interest to justify them. Thus, such bans are dead on arrival. Casey’s undue burden rule, by contrast, is used to evaluate less restrictive abortion rules—like medical health and safety requirements—falling short of bans. SB-8, however, is a ban. The statute’s saving clause thus incorporates the wrong constitutional test—perhaps because its drafters knew that the statute would obviously fail the correct one.

Finally, the deeper argument. Even if Casey supplied the right test here, and even if SB-8 implemented that test—and not an alternative, narrower one—the structure of the saving clause would doom the statute. As already described, the saving clause says that the statute forbids all abortions that it is constitutional to forbid. This structure makes sense only if one assumes that the undue burden rule functions by picking out some subset of all abortions and insisting that these, but no others, must be allowed to proceed, full stop.

But that is not how our constitutional protections of any individual activity—speech, gun ownership, abortions—work. They do not protect particular instances of activity against any and all encroachment. Instead, they protect the entire class of activity from particular laws. A law may be unconstitutional if it affects the protected activity but has no legitimate purpose. Or, as with the undue burden test, the law may be unconstitutional if its commands, as a whole, are not well matched with its stated legitimate purpose. As a result, many particular instances of protected activity could constitutionally be forbidden by one law, but not by another.

Thus, performing the undue burden test requires knowing both what a law does and why. But, under SB-8’s saving clause, what the law does depends, in turn, on the outcome of the undue burden test. The scope of the law’s prohibition is thus literally undefined. Trying to figure out what it forbids leads one in an infinite logical loop. The statute is thus a nullity—commanding nothing. And a statute that commands nothing can be enforced by no one against no one. SB-8 has thus done the courts’ work for them, effectively striking itself down.

II. Undue Burdens in Texas and in America

Begin, as the Supreme Court surely will, with the text. The saving clause of the abortion ban offers an affirmative defense to liability if “the defendant demonstrates that the relief sought by the claimant will impose an undue burden on [a] woman . . . seeking an abortion.”[12] But the saving clause does not give ‘undue burden’—the constitutionally operative phrase—free semantic rein. Instead, it goes on to narrow the term, providing conditions under which “a defendant may not establish an undue burden under this section.”[13]

By so narrowing the definition of ‘undue burden,’ the saving clause sheds its facially compelling symmetry with constitutional law. No longer does the law even purport to ban all abortions except those which Casey’s undue burden rule says cannot be banned. Instead, it bans all abortions except those which a different, narrower undue burden rule protects. Put bluntly, ‘undue burden’ means something different in Texas than it means in the rest of the United States.

The saving clause begins by stipulating that SB-8 does not create an undue burden if it “merely . . . prevent[s] women from obtaining support or assistance, financial or otherwise” for an abortion.[14] It is entirely unclear why a sudden deprivation of financial support could never count as an undue burden, as Casey defined it. On the contrary, Casey’s majority and plurality opinions are littered with financial considerations. The majority wrote that a law does not impose an undue burden if it “merely make[s] abortions a little more difficult or expensive to obtain.”[15] But a law that makes obtaining one a lot more difficult or expensive—as the provisions held unconstitutional in Casey did—does impose one.[16] To wit, the plurality noted that, in some circumstances, a factfinder might well “conclude that . . . increased costs [for an abortion] . . . amount to substantial obstacles.”[17] Other financial burdens matter too. The majority held that husbands’ “withdrawal of financial support” was an undue burden imposed by a statutory spousal-notice rule.[18]

SB-8’s narrowing of ‘undue burden’ to exclude lost assistance does not stop with financial assistance. Rather, the clause makes irrelevant the loss of any support or assistance in obtaining an abortion. What about the blind woman who relies on a guide to navigate everywhere, including to the doctor’s office? Because of SB-8’s prohibition on “aiding and abetting” abortions, she will likely lose that assistance. This seems quite burdensome. Could that obstacle really never amount to a constitutionally undue burden under Casey? What about the woman who lives outside town and has no car? Under Texas’s law, she may be unable to obtain even an Uber ride to an abortion clinic.[19] It cannot be the case that Casey’s undue burden rule categorically excludes this fact from the analysis. Yet SB-8 does exclude it. One can imagine many more such scenarios.

Ultimately, this “support or assistance” exception to the saving clause’s definition of ‘undue burden’ threatens to swallow the whole thing. Abortions are medical procedures. With few exceptions, every part of every abortion depends on some kind of “assistance.” Surgical procedures must, obviously, be performed by qualified medical practitioners. But even medical abortions require someone to prescribe the medication and someone else to dispense it. If, under SB-8, neither the loss of “assistance” in booking an appointment (from a receptionist), nor in getting to the appointment (from a bus or taxi driver), nor in being prescribed the treatment (by a doctor), nor in purchasing the medication (from a pharmacist), nor in paying for the doctor’s visit and medication (from an insurer) count toward a statutory showing of undue burden, the question becomes: What, if anything, does count? If the answer is “almost nothing,” then the statute’s whittled-down conception of ‘undue burden’ bears little resemblance to Casey’s version of the rule.

The saving clause aggressively narrows the meaning of ‘undue burden’ along a second axis too. The statute provides that a “defendant may not establish an undue burden” by “arguing . . . that an award of relief against other defendants or . . . potential defendants will impose an undue burden.”[20] This exception atomizes the evaluation of burden on a defendant-by-defendant basis.

Suppose a doctor’s receptionist is sued under the statute for abetting an abortion by scheduling an appointment. Perhaps this doctor also allows online scheduling. Then, maybe the burden just of forbidding the receptionist from taking appointments is relatively modest.[21] But if the woman moves forward to schedule her abortion, her doctor will surely be sued, too, and perhaps enjoined from performing the procedure. Alternatively, the doctor might preemptively withdraw care, justifiably fearing liability. But under the statute, in the suit against the receptionist, even the absolute certainty of the doctor’s withdrawal of care is irrelevant. Only the modest inconvenience of online scheduling may be considered.

Suppose that the doctor is then sued, and the plaintiff seeks an injunction against the doctor’s performance of the abortion. The doctor argues that this surely imposes an undue burden on the woman. But, the plaintiff argues, there are many doctors in Texas, so merely enjoining this doctor, on its own, does not impose an undue burden. The woman seeking an abortion can, the plaintiff contends, simply find another doctor. Of course, the statute will almost certainly deter—by injunction or by threat of liability—essentially all of those doctors too. Yet under the saving clause’s definition of ‘undue burden’ none of that matters.[22]

This was not how things worked in Casey, nor is it the norm in constitutional litigation over abortion rights. In Casey, for example, the majority considered the effect of Pennsylvania’s spousal notice rule on all women seeking abortions in that state.[23] It did not confine its analysis to the law’s effect just on Planned Parenthood (the petitioner) and its patients. Similarly, in Whole Woman’s Health v. Hellerstedt,[24] the Court held that the closure of many abortion clinics in addition to the petitioner’s caused an undue burden.[25]

Thus, the saving clause in Texas’s abortion ban is not what it first appears. It parrots the language of Casey and thus purports to contain the statute’s effect within the constitution’s boundaries. But the statute’s protections against undue burdens are much narrower than the protections that Casey enshrined. Indeed, once both of the saving clause’s narrowing provisions are accounted for, almost none of the circumstances that may constitute an undue burden under the U.S. Constitution can constitute one under Texas law.

III. Undue Burden is the Wrong Framework


The saving clause purports to avoid unconstitutionality by narrowing SB-8 to within the limits Casey set. As just discussed, it does not actually do that. But even if it did, that would be no good. Casey does not give us the test for evaluating abortion bans. Roe does.

Roe dealt with one of Texas’s earlier abortion laws. Because that law did not merely make getting an abortion more difficult or expensive, but banned it, the Court applied strict scrutiny.[26] The law therefore needed to both be justified by a “compelling” state interest and be “tailored to” that interest.[27] Texas asserted two interests: protecting “potential life” and ensuring women’s health.[28] The Court acknowledged that these interests are “important and legitimate.”[29] But it held that it is only “as the woman approaches term . . . [that] each [interest] becomes ‘compelling.’”[30] Thus, pre-viability abortion bans are unconstitutional from the get-go, at least if the state’s justification for them is protecting potential life or women’s health.

Those are exactly the interests that Texas again asserts to justify SB-8’s ban. The text of the statute dutifully recounts the state’s “compelling interests from the outset of a woman’s pregnancy in protecting the health of the woman and the life of the unborn child.”[31] Alas, simply calling an interest compelling does not make it so. Texas’s asserted justifications are no more compelling today than they were in 1973, when Roe was decided. SB-8 is thus unconstitutional not because it imposes an undue burden, but because it lacks the compelling interest that strict scrutiny demands.

Casey did not change this picture. True, Casey overruled Roe in one narrow respect. It refined Roe’s picture of precisely when a state’s interest in protecting potential life becomes compelling. Where Roe had drawn the line rigidly at the third trimester of pregnancy, Casey recognized that it was not the third trimester per se that mattered. Rather, the intuition was that a state’s interest became compelling once a fetus could live independently outside the womb.[32] The Casey Court therefore tied the constitutionality of abortion bans directly to fetal viability, casting off the trimester proxy.[33]

It is also true that the Casey Court applied the undue burden test—not strict scrutiny—to evaluate the abortion regulations before it.[34] But that was because the Pennsylvania law at issue there was not a ban. It instead required women seeking abortions to be supplied with certain information, to give consent in light of that information, and to inform their husbands of their intended abortion.[35] It also required most minors to obtain parental consent for an abortion.[36]

One of those requirements was unconstitutional, but none was a ban which, per Roe, would trigger the strictest judicial review. Instead, these less restrictive rules were evaluated under a less exacting test. They did not require a compelling interest to justify them. Instead the merely “legitimate interests” of “protecting the health of the woman and the life of the fetus that may become a child” sufficed.[37] Likewise, these more moderate restrictions on abortion failed to trigger strict scrutiny’s requirement that a law’s means must be “narrowly tailored” to its ends. Instead, the Court instituted a more relaxed means-ends inquiry, asking whether the law’s burdens were “undue” in light of their efficacy in realizing the state’s stated ends.[38]

One might be tempted to argue that Texas’s new abortion law does not impose a ban, and thus that Casey’s undue burden standard is the correct one here. This is a nonstarter. First, the law’s $10,000 statutory penalty functions as a sanction, not as traditional civil damages. It is designed to punish defendants for violating a prohibition, not to compensate plaintiffs for their actual losses.[39] This is obvious from the fact that Texas’s law allows anyone, except government officials, to collect the penalty, regardless of their relationship to the defendant.[40]

If that were not enough, in addition to imposing a hefty penalty, the law literally operates as a ban. It provides that the “court shall award . . . injunctive relief sufficient to prevent the defendant from violating this subchapter.”[41] If a law requiring judges to forbid the performance of abortions on penalty of contempt is not a ban, I do not know what is.

So Texas has again banned pre-viability abortions. SB-8’s saving clause purports to make its ban constitutional by bringing it within the boundary of Casey’s undue burden test. But pre-viability bans are not evaluated under that test. They are subject, as in Roe, to strict scrutiny. And under that framework, the interests Texas offers here—women’s health and prenatal life—are insufficient. The law is doomed from the outset.

IV. Either Unconstitutional or Undefined

A. Undefined, incomprehensible, and thus a nullity

SB-8’s has one final, and equally fatal, flaw. Assume that the above arguments are wrong so that the statute is not unconstitutional for the reasons already described. Imagine that Casey’s undue burden was the right test for evaluating SB-8’s constitutionality. And imagine further that SB-8’s saving clause did not narrow ‘undue burden’s’ meaning but rather gave it exactly the same meaning as the Casey Court did.[42]

Then, when all of SB-8’s provisions—its ban and its saving clause—are read together, the statute’s command is undefined, in the mathematical sense. It is thus unenforceable because it is a nullity—not requiring anything in particular. The statute begins with a broad ban on abortions, which, read on its own, blatantly violates the rules of both Roe and Casey.[43] The statute’s saving clause modifies this broad and facially-unconstitutional ban, narrowing the prohibition. No problem so far.

But the way in which the saving clause narrows the ban produces an intractable problem. The saving clause defines the scope of the statute’s prohibition in terms of what is constitutional.[44] But, as we shall see, what is constitutional in this context depends, in turn, on the scope of the statute’s prohibition.

Consider the basic structure of virtually every constitutional protection of individual activity—protections for speech, owning a gun, or obtaining an abortion. Observe first what these protections do not do. They do not pick out particular instances of the activity and grant those instances unassailable protection against all encroachment. A constitutional right to do something does not consist of a right to do it full stop. Instead, it consists of a right to be free from certain laws restricting it.[45] The government may not, for example, ban the burning of an American symbol because of the ideas such burning expresses.[46] But it can ban the burning of an American symbol because the symbol happens to be an important government record.[47] The object of constitutional analysis is thus the law regulating an activity, not the activity itself.

If the Constitution protects against certain kinds of laws, then the legal tests implementing its protections must likewise be designed to examine laws. And so they are. Consider strict scrutiny, an exemplar trans-substantive framework for evaluating laws’ constitutionality. Strict scrutiny first asks what the goal of the law is—what state interest it serves.[48] When a law is constitutionally suspicious enough to have triggered strict scrutiny, only a “compelling” interest can justify it.[49]

Strict scrutiny then asks whether the law’s actual operation—the scope of its mandates or prohibitions—serves its stated interest.[50] Strict scrutiny demands narrow tailoring, which means that the means–ends match must be quite precise. The tailoring requirement serves two purposes. First, it weeds out laws that impose constitutional burdens that are too heavy, as compared with even the legitimate social benefits they generate.[51] Second, the tailoring inquiry flushes out ulterior motives. If a law’s means do not match its stated ends, then one suspects that the stated ends might be pretext, concealing animus toward the constitutionally protected activity itself.[52] As I have argued elsewhere, in addition to strict scrutiny, essentially every constitutional test protecting individual activity—“intermediate scrutiny,” “actual malice,” and even ‘undue burden’—shares this basic structure.[53]

Since SB-8’s saving clause invokes the undue burden test, let us pause for a moment to make sure that it, too, works this way. In Casey, the Supreme Court began by affirming that the state interests that the law there served were “important and legitimate.”[54] And while strict scrutiny would demand a “compelling” state interest, important and legitimate is good enough under the undue burden rule.[55] Then the Court turned to tailoring. It determined that certain provisions of the law produced burdens that were quite heavy and, ultimately, undue in light of the extent to which they served legitimate interests.[56] Recall that, as discussed above, both Casey and Hellerstedt asked whether the law’s total burdens were undue as compared with the total extent to which it served legitimate interests.[57] They did not limit the inquiry to the burdens and benefits to the parties before the Court.

Now we can see clearly why SB-8 collapses into intractable circularity. The statute begins by tentatively forbidding all abortions of fetuses with a heartbeat. But this is not the prohibition’s actual scope. Instead, the law’s command is modified by the saving clause, which narrows it. How much does the saving clause narrow the command? The clause says that it narrows the command to the extent required by the undue burden rule.

So to figure out what the statute actually forbids, one must perform the undue burden analysis.[58] But as just discussed, the undue burden analysis includes a tailoring test. And that test asks whether the law’s effect—the total scope of its command—sufficiently matches its stated ends. Thus, there is simply no way to perform the undue burden inquiry without first knowing what the law, as a whole, demands. This leads us right back where we started, since under SB-8’s saving clause, determining what the law demands requires one to first perform the constitutional test! And so on, and so on, ad infinitum.

The statute thus has no semantic content. It is vacuous. A nullity. And therefore unenforceable. This, on its own, is reason enough for the courts to ignore it. If one wished to invoke a constitutional concept, one could say that SB-8 was straightforwardly void for vagueness.[59] But the situation here is much worse than with most vague statutes. Usually, the problem is that a statute’s prohibition has too much content, such that its words can easily be read to outlaw innocuous everyday activities.[60] Such laws invite arbitrary enforcement.[61] But here we have the opposite problem. By saying nothing, the statute prohibits nothing. Its command is literally impossible to enforce. However, if anyone tried, such an exercise of power would surely be even more arbitrary than those which we already know the constitution forbids.

B. If not null, then unconstitutional

The structure of SB-8’s legal command is unusual but not completely unprecedented. On the contrary, for nearly 150 years, the Supreme Court has dealt with similar problems arising from certain statutes’ severability clauses.

SB-8’s saving clause is not a severability clause. Severability clauses lie dormant until a court has ruled that a portion of the statute is unconstitutional and therefore unenforceable. Such clauses then tell the court to give effect only to the law’s constitutional portions. SB-8’s saving clause, by contrast, operates from the get-go. It narrows the statute’s prohibition as soon as a defendant asserts the clause’s affirmative defense, rather than activating only upon a judicial finding of unconstitutionality. There are thus differences between the two categories of statutory provision.

Nevertheless, as with SB-8’s saving clause, severability clauses sometimes operate to produce a law with a null command.

In United States v. Reese,[62] the Court ruled that a statute “punish[ing] . . . all persons, who . . . hinder [or] delay . . . any person from qualifying or voting” was unconstitutional.[63] The Court held that Congress lacked the power to protect all voters from disenfranchisement. Instead, the Fifteenth Amendment gave lawmakers only the power to outlaw explicit race-based disenfranchisement.[64] Having found the as-written rule unconstitutional, the Court was asked to decide whether the statute could “be limited by judicial construction so as to make it operate only on that which Congress may rightfully prohibit and punish.”[65] The Court refused to limit it. And its explanation of why sounds almost as if it could have been written about SB-8:

It would certainly be dangerous if the legislature could set a net large enough to catch all possible offenders, and leave it to the courts to step inside and say who could be rightfully [punished], and who [could not]. This would, to some extent, substitute the judicial for the legislative department of the government. . . .

To limit this statute in the manner now asked for would be to make a new law, not to enforce an old one. This is no part of our duty.[66]

Why would severing the unconstitutional portion of this statute—and enforcing the rest—constitute an exercise of legislative, rather than judicial, power? Because the law’s entire command was unconstitutional. Thus, if the unconstitutional portion were severed, there would be nothing left. The severed statute would be null.

As the Reese Court wrote, it was “not able to reject a part [of the statute] which is unconstitutional, and retain the remainder,” because no such remainder existed.[67] The Court could not, for example, “strik[e] out or disregard[]” just the unconstitutional “words that are in the section.”[68] There were no separate, constitutional commands. Instead, the only way for the statute to emerge from the severability analysis having any content at all would be if the court “insert[ed] [commands] that are not now there.”[69]

Thus, Reese is not hostile to all severability clauses. Often, a law has multiple, separable commands, some of which are perfectly constitutional. Then, federal courts are happy to enforce the constitutional commands while ignoring the unconstitutional ones. It is only when severing the unconstitutional portions of a law means severing the entire law that problems arise. At that point, the law’s remaining commands are a null set. And the only way for the law to continue requiring anything is for the Court to make up a new rule from scratch. That is the legislative work that the Constitution’s separation of powers rules forbid federal courts from doing.

This principle from Reese—that a federal court may not give content to a null command by making up a brand-new rule—has persisted. In Reno v. ACLU,[70] the Supreme Court refused to rework a broad, unconstitutional restriction on speech. The Court would not, it said, invent a brand-new speech restriction that applied only to some set of “‘persons or circumstances’ that [the Court determined] might be constitutionally permissible.”[71] Citing Reese, the Reno Court held that it “w[ould] not rewrite a . . . law to conform it to constitutional requirements” because the writing of law was the responsibility of “the legislative department of the government.”[72]

Hellerstedt said the same about abortion laws. There, the Court found that a broad suite of putative health-and-safety regulations, acting together, forced a large number of abortion clinics to close.[73] And since those regulations, taken together, produced a “virtual absence of any health benefit,” they imposed an unconstitutional undue burden.[74] The statute contained a severability clause, but citing the Reese principle, as restated in Reno, the Court refused to apply it.[75]

Again, the problem was that, severed of its unconstitutional portions, the statute’s command was empty. No single health regulation caused the multitude of clinic closures, nor did any single rule, on its own, cause the statute to fail to produce health benefits. Rather, it was their cumulative effects that were out of balance. Striking out the unconstitutional portion of the law meant striking out everything. And the Court was neither willing nor able to invent a brand-new set of lightly burdensome, health-promoting regulations from scratch.

There were two potential remedies for the null-set problems arising from severability clauses. First, the Court could have ignored the severability clauses because they asked the Court to do something beyond its power.[76] This is what the Court in Reese, Reno, and Hellerstedt actually did.[77] The formal result was that the statutes persisted as written and were therefore entirely unenforceable as unconstitutional. But the other option would have been to read the severability clauses strictly and do exactly what they asked, but no more. Severability clauses literally ask a court to sever the unconstitutional portions of a statute. So when the entire statute was unconstitutional, the Court could have just severed everything. Then, the formal result would have been that the entire statute was severed, that the severed parts should be ignored, and thus that the statute was unenforceable as null.

One could suggest these same two alternatives for SB-8. The statute’s saving provision renders its command an undefined nullity. Above, I argued that this, on its own, makes the statute unenforceable. But in the Reese cases, the remedy for a statutory provision’s having made the statute a nullity was to ignore that provision. One could similarly ignore SB-8’s saving clause. But that does the statute no good. Then, as with the Reese-style statutes, all that remains is the statute’s unconstitutional command—the ban on all post-heartbeat abortions. Thus, if SB-8 is saved from logical indeterminacy, it slides immediately into unconstitutionality.

c. No way out

In Reese and its progeny, the states that enacted completely unconstitutional laws requested a remedy for the resulting null command that their severability clauses produced. They wanted the federal courts that had found the laws unconstitutional to invent brand-new laws to fill the void. The Supreme Court held that federal courts cannot make up new laws because constitutional separation-of-powers principles forbid it. So the request for a remedy was denied.

However, SB-8 is, as noted above, specifically designed to evade review in federal court. The statute’s procedural provisions funnel cases into the Texas state courts. And state courts are not bound by the U.S. Constitution’s rule that the legislature makes laws while courts just interpret them. On the contrary, many states have robust traditions of judicial common law rulemaking.

Perhaps, then, this is the answer to the recursive paradox arising from SB-8’s saving clause: The Texas state courts should just do what the federal courts could not—make up a new legal rule from scratch.[78]

This will not work under SB-8—at least not under SB-8’s saving clause. That clause, unlike a severability clause, does not ask courts to define what the statute says. Indeed, the saving clause is not addressed to courts at all. The clause instead operates as an affirmative defense and is thus directed to private parties. It asks the defendant in a case to show that imposing liability would place “an undue burden on [some] woman . . . seeking an abortion.”[79] As already discussed, showing that a law imposes an undue burden first requires showing what the law does. The saving clause thus invites defendants, if anyone, to define the law’s scope.

The problem is that defendants, acting as private parties in a lawsuit, have absolutely no power to make law. A state court might theoretically take a null statute and, by fiat, redefine it as enshrining some new rule of general applicability. But no matter what a defendant says that SB-8’s undefined command requires, mere assertions do not make it so. Thus, unlike a severability clause, SB-8’s saving clause raises no opportunity for anyone to give the law meaning via post-legislative reinterpretation.

To concretize this observation, consider how SB-8’s saving clause would be litigated in a state-court suit. A defendant sued under the statute might argue that it was unconstitutional or, in the alternative, did not forbid her abortion. The two arguments would run parallel to one another.

The defendant might begin by arguing that, when the statute’s broad ban, plus its saving clause, were read together, the statute operated to forbid X class of abortions. Suppose that X includes the defendant’s abortion. If banning X imposes a serious burden but does little to serve legitimate goals like promoting women’s health, then the law fails the undue burden test. Then, the statute is unconstitutional. Or maybe this is just proof that the statute, because of the saving clause, cannot forbid X and so does not apply to the defendant.[80]

As discussed above, the statute’s command is, in actuality, undefined. So the defendant’s contention that it forbids X is just that—a contention, made in litigation, carrying no legal weight.

Thus, to counter the defendant’s argument for unconstitutionality, the plaintiff might simply assert that SB-8 instead forbids Y class of abortions. If Y is lightly burdensome and quite beneficial—consider, e.g., a ban on surgical abortions where the doctor did not first wash their hands—then it is constitutional. As already established, class X and Y could both include the defendant’s abortion, but rule X could be unconstitutional, while rule Y was constitutional.[81] The plaintiff’s interpretation is just as right—or wrong (or neither)—as the defendant’s; both are coloring on a blank canvas.

Where do we land in the end? Exactly where we started. Who wins? Who loses? Who is right about the law’s meaning, and thus its scope and/or its constitutionality? It is literally impossible to say. SB-8’s saving clause invites private parties to spin out their own interpretations of the statute. But unlike state courts’ interpretations, private parties’ interpretations, no matter how finely spun, simply cannot become law. Before, during, and after litigation, the statute’s command remains resolutely undefined.

Thus, SB-8’s saving clause does not contain a mechanism by which anyone could make sense of the nonsensical law. However, in addition to its saving clause, SB-8 also contains a severability clause.[82] So maybe if, as argued in this Essay, the saving clause cannot save the statute, the severability clause can.

It is at least conceivable that Texas courts could, relying on SB-8’s severability clause, make up an entirely new law to replace the currently undefined one. The operation would go as follows: The court would observe that the law’s post-heartbeat abortion ban was, read on its own, unconstitutional. It would observe that the saving clause could not produce a new, well-defined, and constitutional rule to supplant the unconstitutional one. Drawing on Reese, it would thus ignore the saving clause. But contra Reese, the court would not ignore the severability clause. Instead, it would use the clause to simply make up some new rule and call that SB-8’s command.

There are multiple reasons that the Texas state courts should not do this. First and foremost, the severability clause does not permit them to. Instead, the clause says that “[a]ll constitutionally valid applications of this chapter shall be severed from any applications that a court finds to be invalid, leaving the valid applications in force.”[83] But as in Reese and subsequent cases, when a law is entirely unconstitutional, refusing to enforce the unconstitutional parts leaves no “applications” remaining. Nowhere does the clause suggest that, in this situation, Texas courts—or any courts—should invent a brand-new rule.

Even if Texas’s courts could invent a brand-new law to replace SB-8’s undefined command, they probably should not. Even in states that embrace common-law rulemaking, inventing new, fully formed laws in a single stroke is not the kind of thing courts usually do. In the common law process, all parts of a law—its ends and its means—emerge slowly. They accrete from innumerable individual adjudications, each tied to the legal principles of the past.

Thus, a state authorizing the kind of lawmaking contemplated here would be endowing its courts with legislative power far broader than the traditional common law role. It is highly questionable whether courts are well-suited to this kind of lawmaking. At a minimum, requiring courts to do all of that, on top of their normal work, would, as the Hellerstedt Court said, “inflict enormous costs.”[84] Moreover, such a state might have to ask itself what the point was of having separate legislative and judicial branches.

Finally, it is worth noting that, even if Texas’s courts did decide to invent a new rule to replace SB-8, that could well be a huge win—not a loss—for abortion rights advocates. The new rule could be anything, but unless the courts wanted it, too, to be struck down, the rule would have to be constitutional. It could not ban abortions before viability. It could not impose burdensome “health” regulations that produced no health benefits. Thus, most of the new rules that a court would be likely to impose would be relatively anodyne. If the new rule simply required, for example, handwashing before surgical abortions, who would object?

In this case, SB-8 would formally remain on the books. But functionally, it would be as if the law had been repealed and replaced with a far less restrictive one. This is, in the end, the best-case scenario for those favoring abortion access. After all, even if SB-8 were struck down, Texas’s legislature could, just as easily as the courts, impose a new, modest, and constitutional rule to replace it.


SB-8 is thus unenforceable for any number of reasons. It is unconstitutional because it bans pre-viability abortions. It is not rendered constitutional by its saving clause, which deploys the wrong constitutional test and deploys it badly. And even if it were not unconstitutional, the statute is literally impossible to enforce. Its command is undefined, and thus, the statute forbids nothing.

  1. . See Whole Woman’s Health v. Jackson, No. 21A24, slip op. at 1 (U.S. Sept. 1, 2021).
  2. .Id. at 1–2.
  3. .Cf. id. (“The State has represented that neither it nor its executive employees possess the authority to enforce the Texas law either directly or indirectly.”).
  4. .410 U.S. 113 (1973).
  5. .Planned Parenthood of Southeastern Pennsylvania v. Casey, 505 U.S. 833 (1992).
  6. .The Supreme Court will soon hear argument in Jackson Women’s Health Org. v. Dobbs, 945 F.3d 265 (5th Cir. 2019), cert. granted in part, No. 19-1392 (U.S. May 17, 2021). There, the Court has been asked to upend the settled constitutional doctrines protecting abortion rights. First, it may overturn Roe and‌/or Casey, in which case, SB-8 would face few, if any, constitutional hurdles. Alternatively, it may hold that Casey, but not Whole Woman’s Health v. Hellerstedt, 579 U.S. 582 (2016), correctly applied the “undue burden” standard. It is not exactly clear what this would entail, since Hellerstedt purported to apply Casey. But it should not affect the arguments presented here, since this Essay draws on both cases as examples of how the standard should be applied.
  7. .Tex. Health & Safety Code Ann. § 171.208(b)(2).
  8. .Id. § 171.208(b)(1).
  9. .Jane Chertoff, How Early Can You Hear Baby’s Heartbeat on Ultrasound and by Ear?, Healthline (Sept. 26, 2021), https://‌‌/health‌/pregnancy‌/when-can-you-hear-babys-heartbeat#takeaway [].
  10. . Health & Safety § 171.209(b)(2).
  11. .Planned Parenthood of Southeastern Pennsylvania v. Casey, 505 U.S. 833, 874 (1992) (plurality opinion).
  12. .Health & Safety § 171.209(b)(2).
  13. .Id. § 171.209(d).
  14. .Id. § 171.209(d)(1).
  15. .See Casey, 505 U.S. at 893 (emphasis added).
  16. . See id. at 893–94, 901 (“While at some point increased cost could become a substantial obstacle, there is no such showing on the record before us.”).
  17. .See id. at 886 (plurality opinion).
  18. .Id. at 893 (majority opinion).
  19. .She could attempt to deceive the driver by concealing her destination. But faced with $10,000 in potential liability, drivers, and perhaps even Uber itself, will be vigilant.
  20. . Health & Safety § 171.209(d)(2).
  21. .At least for women with easy access to the internet.
  22. .William Baude has suggested to me that, while the most natural reading of section 171.209(d)(2) would, indeed, forbid consideration of the other doctors’ unavailability, courts may try to read it more generously to avoid constitutional problems. I agree on both counts, with the caveat that I think alternative readings would strain the text quite badly. However, if courts do read section 171.209(d)(2) as more inclusive in its consideration of burdens, then that particular provision produces fewer constitutional problems than argued above. This would not affect my arguments about section 171.209(d)(1).
  23. .Casey, 505 U.S. at 893 (“The spousal notification requirement is thus likely to prevent a significant number of women from obtaining an abortion.”).
  24. .136 S. Ct. 2292 (2016).
  25. .Id. at 2312–13.
  26. .Roe v. Wade, 410 U.S. 113, 156, 165–66 (1973).
  27. .Id. at 165–66.
  28. .Id. at 162–63.
  29. .Id. at 162.
  30. .Id. at 162–63.
  31. .Health & Safety § 171.201(3).
  32. .Casey, 505 U.S. at 846; id. at 872, 875–77 (plurality opinion).
  33. .Id. at 875–76 (plurality opinion).
  34. .Id. at 874.
  35. .Id. at 844 (majority opinion).
  36. .Id.
  37. .Id. at 846.
  38. .See Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292, 2318 (2016) (holding that a law imposes an undue burden when, despite justifying itself by reference to health and safety, its burdens did little or nothing to improve health and safety).
  39. .See Robert Cooter, Prices and Sanctions, 84 Colum. L. Rev. 1523, 1523 (1984).
  40. .Health & Safety § 171.208(a).
  41. .Id. § 171.208(b)(1) (emphasis added).
  42. .Even for readers convinced by the arguments in Parts II and III, this Part is worth reading. It shows, as a general matter, why laws with the structure “ban everything except that which cannot be constitutionally banned” are unenforceable. Even if SB-8 does not actually have this structure, it is entirely conceivable that, with SB-8 as a model, future statutes may. Moreover, the arguments here may, to some extent, apply even to SB-8 as actually written. The core circularity arises when knowing the scope of the law depends on knowing whether it passes X test, but knowing whether it passes X test requires knowing the scope of the law. SB-8’s version of the undue burden test, while narrower than Casey’s, arguably still has this structure.
  43. .Seesupra Part III.
  44. .As argued above, the clause does not actually do this, since it chooses the wrong constitutional test and narrows the test it chooses beyond recognition. However, in this Part, I aim to show the problems the statute would have even if undue burden were the right test and the saving clause gave that term the same meaning as it had in Casey.
  45. .See Richard H. Fallon, Jr., The Nature of Constitutional Rights 48 (2019).
  46. .See Texas v. Johnson, 491 U.S. 397, 420 (1989).
  47. .See United States v. O’Brien, 391 U.S. 367, 381–82 (1968).
  48. .E.g., United States v. Playboy Ent. Grp., Inc., 529 U.S. 803, 813 (2000).
  49. .Id.
  50. .Id.
  51. .See Peter N. Salib, The Pigouvian Constitution, 88 U. Chi. L. Rev. 1081, 1115 (2021).
  52. .See Joseph Blocher, Bans, 129 Yale L. J. 308, 369–70 (2019).
  53. .See Salib, supra note 51, at 1098–1111.
  54. .Planned Parenthood of Southeastern Pennsylvania v. Casey, 505 U.S. 833, 871 (1992) (plurality opinion) (quoting Roe v. Wade, 410 U.S. 113, 162 (1973)).
  55. .Id.
  56. .Casey, 505 U.S. at 893 (holding law unconstitutional because it did not “merely make abortions a little more difficult or expensive to obtain”).
  57. .See supra notes 23–25 and accompanying text.
  58. .Again, as argued in Parts II and III, SB-8’s “undue burden” analysis is nothing like Casey’s. But here, I steelman the statute by supposing that they are identical.
  59. .See Papachristou v. Jacksonville, 405 U.S. 156, 162 (1972).
  60. .See id. at 162–63, 65 (1972); Kolender v. Lawson, 461 U.S. 352, 357–58 (1983).
  61. .See Papachristou, 405 U.S. at 162–63, 65; Kolender, 461 U.S. at 357–58.
  62. .92 U.S. 214 (1875).
  63. .Id. at 220–22.
  64. .Id. It is worth noting and emphasizing here that this holding operated to decimate America’s Reconstruction-era strides toward political equality. Reese is a very bad case, for that reason. But its holdings on severability are mostly orthogonal to those shortcomings. As described above, they have persisted in modern, less objectionable case law.
  65. .Id. at 221.
  66. .Id.
  67. .Id.
  68. .Id.
  69. .Id.
  70. .521 U.S. 844 (1997).
  71. .Id. at 883.
  72. .Id. at 884–85, 884 n.49 (internal quotations omitted).
  73. .Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292, 2312–13 (2016).
  74. .Id. at 2313.
  75. .See id. at 2319.
  76. .See id. (noting that “our cases have never required us to” comply with a severability’s request).
  77. .See United States v. Reese, 92 U.S. 214, 221 (1875); Reno, 521 U.S. at 884–85, 884 n.49; Hellerstedt, 136 S. Ct. at 2319.
  78. .As already noted, the United States has sued Texas directly in federal court to enjoin enforcement of the law. It may be the case that this federal lawsuit is permitted. Or it may not be. This section proceeds by assuming the best-case scenario for SB-8—that the federal suit is improper or, even if proper, will not preclude all enforcement in state court.
  79. .Health & Safety § 171.209(b)(2).
  80. .Note again the maddening circularity: Xunconstitutional~Xconstitutional. The statute is difficult to even write about coherently.
  81. .Suppose that the defendant, a doctor, did not wash her hands before performing an abortion, but instead sanitized them via an equally beneficial chemical process. Then, she would be liable under a hand-washing rule. The fact that the hand-washing rule is very slightly overbroad, penalizing a few certain sterile-handed surgeons, does not necessarily make it unconstitutional. It is no Second Amendment defense to a run-of-the-mill gun licensing rule to argue that you, in particular, would safely keep and use your gun, even if unlicensed. See Salib, supra note 51, at 1115–16. Put another way, even “narrow tailoring” does not require perfect tailoring.
  82. .Health & Safety § 171.212.
  83. .Id. § 171.212(b).
  84. .Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292, 2319 (2016).