I. Throwing Down a Gauntlet
Professor Melissa Murray is right about one thing. Laws banning trait-selection abortion — prohibitions on abortion when had solely because of the race, sex, or specific disability of the child that otherwise would be born — pose a direct challenge to the Supreme Court’s constitutional abortion-law doctrine under Roe v. Wade1×1. 410 U.S. 113 (1973). and Planned Parenthood v. Casey.2×2. 505 U.S. 833 (1992). Such laws prohibit abortion based on a specific reason for having an abortion. And under current judicial doctrine, the state can’t do that: Roe and Casey held that a pregnant woman has a constitutional right to obtain an abortion; that that right may be exercised for essentially whatever reason the woman sees fit; and that the state may not make or enforce laws that have the purpose or effect of inhibiting the woman’s abortion choice.3×3. See Roe, 410 U.S. at 164–65; Casey, 505 U.S. at 845–46. The state may not forbid a reason — any reason at all — for which an abortion is committed or obtained.
There is a little more to it than that. But not much. The Court’s cases make the right to abortion plenary — explicitly so, pre-viability, and essentially so, as the result of a sweeping “health exception” even post-viability.4×4. Roe, 410 U.S. at 163–65; Casey, 505 U.S. at 845–46. Briefly stated: Roe makes the abortion right plenary in the first trimester of pregnancy and makes second-trimester regulation allowable only to further interests in the “health of the mother,” not to restrict the mother’s plenary freedom of abortion choice. Roe, 410 U.S. at 163–64. Roe also requires that abortion be permitted for “health” reasons even in the third trimester, after the point when the child could live outside his or her mother’s womb. Id. at 164–65. Roe’s companion case, Doe v. Bolton, 410 U.S. 179 (1973), defines health for purposes of the abortion right to include “all factors . . . relevant to the well-being” of the pregnant woman, id. at 192, including “emotional, psychological, [and] familial” factors, id., a formulation that renders the abortion choice essentially plenary in the third trimester as well.
Casey modified Roe’s trimester framework, replacing it with a simpler pre-viability/viability line, Casey, 505 U.S. at 872–73, 878–79, and adjusted the threshold of when state laws imposing an “undue burden” on the right to have an abortion before viability are presumptively invalid, id. at 877–78. But Casey otherwise carried forward the full Roe-Doe abortion liberty, including the broad definition of when abortion must be allowed for “health” reasons, if the pregnant woman so chooses. Id. at 879. The result is that Roe and Casey recognize a right to abortion throughout pregnancy for essentially any reason the pregnant woman chooses — explicitly so before viability and, in practical effect, by operation of the broad health justification after viability. Decisions subsequent to Casey have maintained its framework and substance. See Gonzales v. Carhart, 550 U.S. 124, 161 (2007); Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292, 2309–10 (2016).
For a fuller explication of abortion-law doctrine, see Michael Stokes Paulsen, The Worst Constitutional Decision of All Time, 78 Notre Dame L. Rev. 995, 995 n.4 (2003) [hereinafter Paulsen, Worst Decision], and Michael Stokes Paulsen, Five Provocative Pro-life Proposals, 35 Quinnipiac L. Rev. 661, 677–83 (2017). For a narrower reading of the “health exception,” see Stephen G. Gilles, Roe’s Life-or-Health Exception: Self-Defense or Relative-Safety?, 85 Notre Dame L. Rev. 525 (2010). As a result, under Roe and Casey, abortion must be permitted as a matter of constitutional right, to be exercised for virtually any reason, throughout all nine months of pregnancy, right up to the point of live birth.
If Roe and Casey are right, then, abortion constitutionally must be allowed for any reason. This includes the race of the child to be born, the fact that she is a girl, or the fact that the child would be born with a disability. Under Roe and Casey, trait-selection abortion bans are “unconstitutional” (to accede for the moment to a familiar but improper use of the term). For the Court to uphold such bans would require it to repudiate, or greatly revise, the premises and substance of its current abortion-law doctrine.5×5. See infra Part III, pp. 426–30.
Professor Murray is right about another, related thing: that that is at least part of the point of enacting such trait-selection abortion bans. Surely, part of the purpose of laws banning abortion when had for such eugenics reasons is to pose a more or less head-on challenge to the legal premises and deadly logic of Roe and Casey. If those decisions do indeed yield a constitutional right to kill Black human fetuses for being Black, girls for being girls, and the disabled for their disabilities, that result is truly monstrous and rightly discrediting. The ultimate goal of trait-selection bans is — probably just as Murray fears — to lay the groundwork for (or to attain straightaway) the overruling of Roe and Casey, by attacking their constitutional reasoning in a particularly sympathetic and persuasive context that lays bare where their logic leads.
Trait-selection bans throw down a gauntlet: Does the Constitution really confer a right to abortion when the sole reason is the race, or sex, or disability of the unborn human child? Is there a constitutional right to kill a living human fetus because he or she is a child of color, because he or she has Down syndrome, or because (in a perverse reversal of a traditional expression of joy) “it’s a girl!”?
If the answer to these questions is yes, Roe is perhaps revealed in new ways, to fresh eyes, for the legally radical, extremist decision its critics have long claimed it to be: Roe and Casey recognize a constitutional right to abortion even when had for eugenics purposes. And if the answer is no, the Court — eyes wide open, feet to the fire — might well feel compelled to reevaluate, revise, or even repudiate Roe.
At a simple doctrinal level, trait-selection bans thus present a square challenge to the Court’s abortion decisions. And at a deeper level, trait-selection bans tend to undermine both the legal and moral assumptions underlying the judicially created constitutional right to abortion in a unique way: they refute the “it”-ness of the human fetus. The unborn human fetus has human traits, qualities, capacities — in short, a distinctive human identity. Trait-selection abortion bans force fair-minded people (including judges) to confront and wrestle with the assumed “it”-ness of the human fetus in light of its — his or her — undeniable human characteristics. And that wrestling tends to produce a moral intuition: that the unborn human fetus is part of our common humanity. The moral intuition is a powerful one, uniting the otherwise sometimes differing moral instincts of the traditionalist right and the progressive and feminist left.6×6. As progressive feminist theorist Judith Butler has argued, sex and gender — and, I would add, race and disability — are not merely traits that human beings happen to have. They are qualities that lie at the heart of what makes us human, as opposed to mere objects. See Judith Butler, Gender Trouble: Feminism and the Subversion of Identity 111 (1990). “[T]he moment in which an infant becomes humanized is when the question, ‘is it a boy or a girl?’ is answered,” Butler writes. Id. (invoking feminist philosopher Simone de Beauvoir for the proposition that “being sexed and being human are coextensive and simultaneous”).
Murray is concerned that trait-selection abortion bans produce “cognitive dissonance” for abortion rights advocates by putting “the social justice community’s predisposition toward abortion rights in conflict with laws that ostensibly prevent discrimination on the basis of race and sex.” Melissa Murray, Race-ing Roe: Reproductive Justice, Racial Justice, and the Battle for Roe v. Wade, 134 Harv. L. Rev. 2025, 2068 (2021). “Cynically,” Murray writes, “one might argue that in framing its opposition to abortion in terms of race, sex, and disability discrimination, the pro-life movement is not only using antidiscrimination norms opportunistically, it is doing so in a way that divides the coalition of pro-choice advocates and activists.” Id.
But why is there anything “cynical” or improper about this? The pro-life social justice movement seeks to advance positions that might appeal to feminists, disability-rights advocates, racial justice advocates, and political progressives in general, and not just social conservatives. It is exactly the point to call into question a supposed “predisposition toward abortion rights” by framing difficulties with the logic and implications of such a position.
This is potentially game changing. If the intuition of the wrongness of trait-selection abortion has moral salience — the intuition that it is simply wrong to kill a fetus for reasons of race, sex, or disability — it is because of the implicit recognition of the humanity of the fetus. If killing a fetus because she is female (or Black, or disabled) is thought horrible, it can only be because the human fetus is thought to possess moral status as human — because “it” is a baby girl or a baby boy, a member of the human family.
Constitutional law tends to follow moral intuitions. And the legal intuition that tends to follow from recognizing that the fetus has human characteristics — a distinctive, individual human identity — is that it should not be legal to kill a fetus on the basis of such human qualities.7×7. This intuition contrasts sharply with the implicit ideology underlying current abortion law: Roe treats the unborn human fetus as merely “potential” human life. See Roe, 410 U.S. at 150. If the human fetus has no human moral status unless the pregnant woman chooses to give it one (the position assumed by current law), there is nothing at all wrong with sex-selection abortion or with race-based or disability-based abortion. The fetus is, on this view, not really “a girl,” or “Black,” at all. She is only a “potential girl” or “potentially Black.”
This in turn can have further implications — producing something like an “aha!” moment: If it’s wrong to kill a girl because she is a girl, isn’t it wrong to kill a girl for some other reason? Isn’t the girl still a girl, irrespective of the reason for which an abortion is had? Whatever makes it intuitively wrong — and not a matter of constitutional entitlement — to kill a human fetus because of his or her race, sex, or disability strongly suggests that it is also wrong to kill that same human fetus for most other reasons.8×8. I put to one side tragic situations of abortion when a pregnancy presents a serious danger to the life or physical health of the mother. Paulsen, Worst Decision, supra note 4, at 1020–21. There are situations in which one might recognize the humanity and moral worth of the human fetus but nonetheless defend a right to abortion of that life, for the sake of protecting other values or interests, judged to be of equal or greater value.
Some arguments for abortion rights in general take this form — conceding, arguendo, the moral status and presumptive entitlement to life of the human fetus but arguing that autonomy rights of various kinds trump the human rights of the fetus. See id. (Judith Jarvis Thomson’s classic argument for abortion is an argument of this sort. See generally Judith Jarvis Thomson, A Defense of Abortion, 1 Phil. & Pub. Affs. 47 (1971).). Recognition of the human moral status of the fetus, however, “should affect the range of values that might plausibly be thought to justify killing that human life.” Paulsen, Worst Decision, supra note 4, at 1020.
Trait-selection abortion bans thus pose hugely important, stark, and seemingly unavoidable legal and moral challenges to the constitutional legal regime of Roe. These are the fundamental questions posed in the Box v. Planned Parenthood9×9. 139 S. Ct. 1780 (2019) (per curiam). case and that are discussed in Justice Clarence Thomas’s important concurrence in denial of certiorari in that case — the opinion that is the topic of Murray’s article. And yet these are the questions that Murray avoids entirely.
Instead, Murray trains her energies on a specific feature of Justice Thomas’s opinion — its discussion of race specifically as among eugenicist justifications for abortion. Thomas noted, as relevant background to modern efforts to ban abortion when had specifically for eugenics purposes, the disturbing history of racist and disability-eugenics arguments for abortion made in the (disturbingly recent) past.10×10. See infra Part II, pp. 421–26. Murray’s article proposes to “contextualize”11×11. Murray, supra note 6, at 2029. and provide a more “nuanced”12×12. Id. at 2030. view of this historical evidence. Yet, as we shall see, much of Murray’s evidence supports Thomas’s conclusion.13×13. See infra Part II, pp. 421–26.
Murray’s main point, however, is to criticize — to warn against — what she sees as the dangerous rhetorical and legal implications of Justice Thomas’s efforts to draw a connection between the abortion-rights legal regime and abortion-as-an-instrument-of-eugenics arguments and outcomes. Such an attempted connection, Murray argues, uses race to undermine Roe — a consequence she finds objectionable and dangerous. Moreover, she fears, racial justice arguments might supply the “special justification” needed to overrule prior precedent.14×14. Murray does not address whether similar concerns might not specially justify overruling precedent because of the harms it inflicts, or supports, on the basis of sex or disability. I submit that they do. See infra Parts III and IV, pp. 426–33. She believes that this is the project afoot in Thomas’s Box concurrence.
This Response offers a straightforward but harsh critique of Murray’s opus: it misses the point. For all of its analysis, Murray’s article simply fails to address the central legal questions posed by trait-selection abortion bans: Are they constitutional or not? Does the Roe right really embrace the freedom to kill a fetus because it is Black, female, or disabled? And if trait-selection bans are constitutional, why — and doesn’t the answer to that question deservedly undermine the legitimacy of Roe and Casey?
This Response proceeds in three Parts.
Part II discusses Murray’s claim that Justice Thomas’s concurrence in Box v. Planned Parenthood is somehow misguided or mistaken in suggesting a historical relationship between abortion rights and racial eugenics arguments. (Murray gives far less attention to disability-based abortion and essentially none at all to sex-selection abortion.)
My contention is that Justice Thomas has it mostly right and that Murray’s critique falls wide of the mark. While abortion is not a eugenics conspiracy — a deliberate plot to reduce the size of the African American, female, or disabled populations — it remains an undeniable fact that the aborted are disproportionately racial minorities, female, and those with disabilities.15×15. See infra Part II, pp. 421–26. Abortion, in short, has a markedly disparate impact along lines of race, sex, and disability. This in itself is troubling, even if it is not proof of deliberate design — which Thomas never claims it is. Thomas does fairly highlight, however, the disturbing history of eugenics arguments for the desirability of liberalized abortion.
More to the point, it is undoubtedly the case that abortions are sometimes had, today, for eugenics reasons — fairly often, even, for sex-selection and disability-elimination.16×16. See infra pp. 424–26. In short, abortion is used as an instrument of eugenics. In some or many instances, this is the result of deliberately eugenic motivations of individuals exercising the abortion choice. In other instances, it is simply the incidental — though perhaps predictable — consequence of patterns of exercise of the abortion right due to other motivations.
Part III builds on this discussion to return to the critical substantive question: Where abortion is shown to be motivated by race, sex, or disability, may it constitutionally be prohibited on that ground? My position is that the constitutionally correct answer is yes and that this answer does indeed undermine Roe’s and Casey’s core legal premises, exactly as Murray worries. While trait-selection bans indeed conflict with current doctrine, I offer a road map as to how the Court nonetheless might sustain them — a route that inevitably would undermine or repudiate current doctrine in one or more important respects.
Part IV concludes with a word about stare decisis. As noted, Murray regrets Justice Thomas’s invocation of the racial-eugenics history of the right recognized in Roe because she fears that considerations of racial justice might provide the “special justification” she presumes is needed to warrant overruling Roe. The short answer (which I have developed elsewhere and will not rehearse at length in this essay) is that the premise is wrong: the doctrine of stare decisis — in the strong sense of deliberate adherence to a decision a court is otherwise fully persuaded is an erroneous, unfaithful interpretation of the Constitution — is simply incompatible with written constitutionalism.17×17. See, e.g., Michael Stokes Paulsen, The Intrinsically Corrupting Influence of Precedent, 22 Const. Comment. 289, 289–91 (2005) [hereinafter Paulsen, The Intrinsically Corrupting Influence]; Michael Stokes Paulsen, The Irrepressible Myth of Marbury, 101 Mich. L. Rev. 2706, 2731–34 (2003) [hereinafter Paulsen, The Irrepressible Myth]; Michael Stokes Paulsen, Abrogating Stare Decisis by Statute: May Congress Remove the Precedential Effect of Roe and Casey?, 109 Yale L.J. 1535, 1548–50 (2000) [hereinafter Paulsen, Abrogating Stare Decisis by Statute]. Murray’s only mildly contrived theory of racial justice–necessitated exceptions to stare decisis — and her fear of its introduction into the abortion field — presumes the need for a special exception to an otherwise strict rule of stare decisis, entrenching Roe. That need does not exist. Moreover, even if one recognizes the force of precedent in constitutional cases as a general rule, most everyone agrees that decisions that are seriously wrong on enormously important questions can and should be overruled. That category certainly includes serious errors on racial justice matters. But it is not limited to such cases.
II. Abortion for Eugenics: Conspiracy or Simple Consequence?
How one answers the question whether abortion is a tool of racial, gender, or disability eugenics depends very much on how the question is asked. Is legalized abortion a eugenicist conspiracy — a deliberate plot on the part of those favoring abortion rights to reduce the number of people of a given race, sex, or disability? Surely not. At the very least, such motivations form no part of the modern argument for abortion rights. Does unrestricted legal abortion-choice produce a disparate impact resulting in disproportionate numbers of abortions ending the lives of minority, female, and disabled fetuses? Undeniably. The aborted are disproportionately Black, female, and disabled. Is the right to abortion sometimes used, by those exercising the abortion-choice, for eugenics purposes — specifically for the purpose of aborting on the basis of race, sex, or disability? Unquestionably. Some — but not all — of the abortion–disparate impact is attributable to intentional decisions to abort based on a trait of the baby that otherwise would be born.
These are three different questions. Justice Thomas’s concurrence in Box keeps them distinct. Murray’s article, in attempting to critique Thomas, tends to smush these separate questions together in a mildly confusing way.
Begin with Justice Thomas’s Box concurrence itself. Thomas’s opinion compiles an impressive and rightly disturbing narrative of evidence that family planning and abortion advocates in the past embraced the desirability of abortion as an instrument for achieving racial eugenics and for culling persons with disabilities from the population. (There appears to be no evidence that early abortion advocates ever favored abortion for gender-eugenics purposes — aborting girls because they are girls.18×18. This is probably most simply explained by the fact that the technology for discerning the fetus’s sex before birth was not readily available until relatively recently. See, e.g., Juan Stocker & Lorraine Evens, Fetal Sex Determination by Ultrasound, 50 Obstetrics & Gynecology 462, 465 (1977). )
“Many eugenicists . . . supported legalizing abortion,” Justice Thomas writes, “and abortion advocates — including future Planned Parenthood President Alan Guttmacher — endorsed the use of abortion for eugenic reasons.”19×19. Box v. Planned Parenthood of Ind. & Ky., Inc., 139 S. Ct. 1780, 1784 (2019) (Thomas, J., concurring). Thomas notes how recent some of these eugenics-as-a-reason-for-abortion observations are, extending to the late 1960s and early 1970s.20×20. Id. For Justice Thomas’s collection of evidence of abortion-rights advocates’ use of eugenics arguments for abortion, see id. at 1787–90.
Neither Justice Thomas in his Box concurrence nor Murray in her article discusses Justice Ruth Bader Ginsburg’s notorious comments in a 2009 New York Times Magazine interview, reflecting on how she had been “surprised” by the Supreme Court’s 1980 decision in Harris v. McRae, 448 U.S. 297 (1980), upholding the constitutionality of statutory restrictions on government funding of abortions, because the result was not in accord with commonly articulated eugenics arguments for abortion in currency at the time. In Justice Ginsburg’s words: “Frankly I had thought that at the time Roe was decided, there was concern about population growth and particularly growth in populations that we don’t want to have too many of. So that Roe was going to be then set up for Medicaid funding for abortion.” Emily Bazelon, The Place of Women on the Court, N.Y. Times Mag. (July 7, 2009), https://www.nytimes.com/2009/07/12/magazine/12ginsburg-t.html [https://perma.cc/5BFU-DPGP] (emphasis added) (interview with Justice Ginsburg). If part of the point of abortion rights was concern over unchecked population growth among certain populations, particularly the poor, then a lack of government funding for abortion would impair that goal.
Justice Ginsburg’s remarks suggest the widespread understanding and expectation of many, at least at the time of the Court’s decision in Harris v. McRae, that liberalized and subsidized abortion access would achieve reduced numbers of disfavored groups — “populations that we don’t want to have too many of.” See id. It is not clear from Ginsburg’s comments whether she personally shared such contemplated eugenics goals for liberalized abortion or (more probably) was merely describing views she believed to be commonly held at the time among abortion-rights advocates. At all events, Ginsburg’s observations — coming from someone who was, throughout the late-twentieth and early twenty-first centuries, an extraordinarily prominent abortion-rights figure — is certainly evidence in further support of Justice Thomas’s opinion. Indeed, Thomas may have recognized it as such but chosen to address Ginsburg’s comments sub silentio. See Box, 139 S. Ct. at 1787 (Thomas, J., concurring) (noting, at the conclusion of compilation of evidence of eugenics motivations, that “support for the goal of reducing undesirable populations through selective reproduction has by no means vanished”).
Murray’s article does not discuss Justice Ginsburg’s remarks, but it does discuss Harris v. McRae in a fashion consistent with Ginsburg’s observations about the effects of that case on abortion among certain populations. Specifically, Murray embraces the critique that a practical consequence of the decision in Harris was to reduce the ability of women of color to obtain abortions. See Murray, supra note 6, at 2091–92; see also id. at 2051–52 (making a similar point).
But Justice Thomas stopped short of claiming that legal abortion is a racist plot to reduce the African American population — even while noting that some prominent African American leaders have so argued over the years.21×21. Box, 139 S. Ct. at 1790–91 (Thomas, J., concurring). Thomas’s point was narrower: Abortion “is an act rife with the potential for eugenic manipulation,” he wrote.22×22. Id. at 1787. Moreover, “[t]he use of abortion to achieve eugenic goals is not merely hypothetical,” he observed,23×23. Id. at 1783. with considerable history on his side. And more to the present-day point, “[t]echnological advances have only heightened the eugenic potential for abortion, as abortion can now be used to eliminate children with unwanted characteristics, such as a particular sex or disability.”24×24. Id. at 1784. Thus, abortion is now capable of being used as “a disturbingly effective tool for implementing the discriminatory preferences that undergird eugenics.”25×25. Id. at 1790.
The right to abortion at one time had eugenicist proponents (among others). The right to abortion can be employed for eugenicist purposes (among others). Indiana’s trait-selection law was a response in part to this history and this reality. That is all Justice Thomas was saying.
Nothing in Murray’s article refutes Justice Thomas on these points. Murray demonstrates that, historically, different parties have acted from different motives and expressed different reasons for supporting abortion rights. This is an important but not a particularly surprising observation. It does not seriously impeach Thomas’s analysis. In fact, much of Murray’s historical discussion of race and abortion actually supports Thomas’s position.26×26. Murray offers an excellent, poignant account of abortion’s connection with our nation’s history of slavery. She provides a particularly sensitive description of the realities of oppression, violence, tragedy, and killing-out-of-desperation framed by that awful epoch. See Murray, supra note 6, at 2033–34. Murray is also to be commended for being scrupulously honest in acknowledging and presenting evidence of the racist goals of the eugenics movement, beginning more than a century ago, and of the efforts of early advocates of birth control and abortion to align their rhetoric with, and to justify their specific policy positions with reference to, then-popular racial- and disability-eugenicist views. Id. at 2035–41. She provides such evidence even though it is at odds with her thesis. Finally, Murray offers striking evidence — evidence Justice Thomas’s opinion largely omits but which provides indirect support for his ultimate conclusion — of important mid-to-late-twentieth century Black leaders’ condemnation of abortion and birth control as instruments of “race suicide,” id. at 2040, or “racial genocide,” id. at 2041–42. Thus, while Murray contends that “the history of race and abortion is more nuanced and complicated,” id. at 2062, than Thomas’s concurrence, her evidence does not actually contradict Thomas’s description.
Justice Thomas never claimed that abortion is a eugenicist conspiracy. Nor did Thomas claim that evidence of the disparate racial incidence of abortion proves a racially discriminatory purpose behind the abortion-rights position. Indeed, Thomas was careful to disclaim any such argument, maintaining consistency with his disparate-impact-does-not-establish-discriminatory-intent position in other discrimination law contexts.27×27. See, e.g., Tex. Dep’t of Hous. & Cmty. Affs. v. Inclusive Cmtys. Project, Inc., 135 S. Ct. 2507, 2526–32 (2015) (Thomas, J., dissenting). Murray’s critique of Thomas here is badly off the mark. She charges Thomas with being “opportunistic and inconsistent” because — supposedly — “Justice Thomas had no trouble associating disproportionately high rates of abortion in the Black community with eugenics and the desire to limit Black reproduction” even though “he rejects the notion that racism is to blame for racially imbalanced outcomes.”28×28. Murray, supra note 6, at 2097. This simply misrepresents and mischaracterizes Thomas’s position, which is almost exactly the opposite of Murray’s characterization.29×29. Far from suggesting that disparate impact is indicative of intentional discrimination, Justice Thomas suggested — somewhat startlingly — that disparate impact analysis and eugenics theory share a common logical flaw, in that both ways of thinking wrongly tend to assume that a single presumed dominant factor can fully explain disparities in outcomes. For Thomas, such an assumption collapses the distinction between correlation and causation, wrongly stereotypes individuals according to statistical groups, and ignores real-world complexities. Box, 139 S. Ct. at 1786–87 & n.4 (Thomas, J., concurring). One need not agree with Thomas’s distinctive view in order to recognize that it is not the position that Murray criticizes. Thomas’s position may be idiosyncratic and even iconoclastic, but it is not inconsistent.
But what is one to make of the glaring evidence of a shockingly disparate impact of abortion? Even if abortion is not designed for eugenics purposes, it unquestionably has disproportionately eugenicist effects. The abortion right is not a racial, or gender, or disability classification, in terms. It does not depend on eugenicist intentions and is not the product of a eugenics conspiracy. But those possessing the abortion right exercise it in a fashion so as to abort more Black babies, more girls, and more “defective” children.
There is no blinking this reality. Consider race: The abortion rate is nearly three-and-a-half times higher among Black women than among white women.30×30. Ctrs. for Disease Control & Prevention Morbidity & Mortality Wkly. Rep., Abortion Surveillance — United States, 2018, at 8 (2020). (The abortion rate is 1.7 times higher for Hispanic women than it is for white women.31×31. Id. ) Black babies are aborted alarmingly more frequently than white babies — a point Justice Thomas made dramatically in his Box opinion.32×32. Box, 139 S. Ct. at 1791 (Thomas, J., concurring). This is almost certainly not attributable to a pattern of deliberate decisions to abort a child because of his or her race, but to socioeconomic factors. Some such factors may be statistically correlated with race, but that does not mean that such abortions are had because of the race of the child. (One suspects that few, if any, pregnancies are aborted specifically because the baby to be born is Black.33×33. One must of course allow for the possibility of unusual and (one would hope) aberrant situations of race-motivated abortion, such as race-based abortion to conceal the identity of an unintended biological father from another domestic partner. There are also reported cases involving alleged discriminatory encouragement (or coercion) of minors who are members of racial minorities to obtain abortions. See, e.g., Arnold v Bd. of Educ. of Escambia Cnty., 880 F.2d 305, 316 (11th Cir. 1989) (sustaining a cause of action against a public school district based on allegations that “the defendants counselled Black students to have an abortion and white students to carry the fetus to term”). ) That does not alter the reality of a significant disparate racial impact to abortion, but it might suggest that something other than racism is the cause of racial disparities in abortion rates.34×34. This is consistent with Justice Thomas’s skepticism of disparate impact analysis generally, as noted above. See supra note 29.
Consider sex: In 1990, Nobel Prize–winning Harvard economist Amartya Sen, in an arrestingly titled article, documented the statistical reality that, worldwide, More than 100 Million Women Are Missing35×35. Amartya Sen, More than 100 Million Women Are Missing, N.Y. Rev. Books (Dec. 20, 1990), https://www.nybooks.com/articles/1990/12/20/more-than-100-million-women-are-missing [https://perma.cc/R4SA-X2VP]. — a result, Sen argued, so far from the baseline norm expected in nature as to be statistically explainable only on the premise of some form of culpable human intervention, likely including severe medical neglect, deliberate female infanticide, and the predictable impact of China’s “one-child family” policy given a strong cultural preference for boys.36×36. Id. Sen’s analysis was focused on Asia and Africa. In 2011, journalist Mara Hvistendahl, in her book Unnatural Selection, reported that the number of missing (and presumed dead) women and girls in Asia alone had reached 160 million and counting.37×37. Mara Hvistendahl, Unnatural Selection: Choosing Boys Over Girls, and the Consequences of a World Full of Men 5–6 (2011). This didn’t just happen. Hvistendahl convincingly demonstrates that the reason for the large demographic disparity in the male-female birth ratio is sex-selection abortion.38×38. Id. Hvistendahl noted that “[s]ex selection defies culture, nationality, and creed.” Id. at 6. Women have abortions of female fetuses because the technology has become widely and inexpensively available to learn the sex of an unborn baby and then to kill her if she is a girl. As Justice Thomas’s Box opinion detailed, the United States is not exempt from the phenomenon of sex-selection abortion.39×39. Box v. Planned Parenthood of Ind. & Ky., Inc., 139 S. Ct. 1780, 1791 (2019) (Thomas, J., concurring) (collecting authorities). Unlike the situation of racial disparities in abortion, there is no way to explain the statistical reality of grossly disparate sex-differential birth rates, at least in some nations or among certain groups, other than by the hypothesis of abortions specifically being obtained because of the female sex of the child.
Consider disability: As Justice Thomas set forth in his Box concurrence, the abortion rate for children diagnosed with Down syndrome approaches 100% in Iceland, exceeds 90% in several other European nations, and is 67% in the United States.40×40. Id. at 1790–91. There is no question but that this is the deliberate killing of children with a disability because of their disability.
Swallow hard and acknowledge the truth of what these numbers reveal, particularly for abortions predicated on sex or disability. Abortion is often employed for eugenics purposes — especially for sex selection and disability elimination. Abortion is used by women and men to kill girls because they are girls. Abortions are obtained in substantial numbers in order to cull the disabled, before birth, because they are disabled. And, while it seems unlikely that an abortion would be obtained specifically because the child is Black, the exercise of the right to legal abortion definitely has a racially disparate eugenic impact.
Is that not disturbing? Might it not supply a judicially recognized “compelling interest” qualifying, or overriding, the judicially fashioned constitutional right to abortion — the point of Justice Thomas’s opinion in Box? Might it not rightly furnish a (further) critique of the validity of Roe v. Wade and Planned Parenthood v. Casey and thereby serve to undermine these decisions’ legal and moral premises? Where race, sex, or disability can be shown to be the reason for a particular abortion, can such an abortion constitutionally be prohibited? Is the Court irrevocably committed to a doctrine that would compel toleration of eugenics-motivated abortion as a constitutional right? Or is there a sensible road back?
III. Thinking Outside the Box: A Roadmap for Upholding Trait-Selection Bans and Reforming Current Abortion Doctrine
As noted at the outset of this Response, laws prohibiting trait-selection abortion conflict in principle with Roe and Casey’s establishment of a constitutional right to abortion for any reason the woman chooses. Is there nonetheless a plausible route for sustaining such prohibitions, even under current constitutional doctrine? I believe there is. But that route requires rereading Roe and Casey, limiting their holdings and narrowing their reasoning fairly substantially, and embracing premises that undermine the doctrinal foundations of those decisions at critical junctures. A thorough doctrinal analysis of these points would require an article of its own.41×41. For an excellent contribution to the literature in this regard, see Thomas J. Molony, Roe, Casey, and Sex-Selection Abortion Bans, 71 Wash. & Lee L. Rev. 1089 (2014). But the broad outlines of the argument can be sketched briskly.
First, rights have reasons. Even unwritten, judicially fashioned rights have reasons. For a right whose origins derive from common law–like judicial decisions rather than constitutional text, the rationale of judicial decisions in effect supply the relevant legal “text” and should permit subsequent common law–like adjustments to better conform rule to rationale. Where the formulation of a right — and especially of an unwritten right — outruns the reasons proffered for that right, there is good reason to limit or narrow it. Roe offers reasons for the abortion right. None of those reasons supports the extension of such a right to situations of deliberately eugenics-motivated abortion.
Central to the reasoning underlying the recognition of an abortion right in Roe was the premise, building on Eisenstadt v. Baird42×42. 405 U.S. 438 (1972). the year before, that the decision “whether to bear or beget a child” is an aspect of the assumed constitutional “right of privacy.”43×43. Id. at 453 (emphasis added) (citation omitted); see Roe v. Wade, 410 U.S. 113, 153 (1973). The implication, but by no means a clear one, is that what Roe seeks to protect is the decision whether to have a child at all. It is the freedom to be, or not to be, a parent. In the case of an abortion had solely because of the race, sex, or disability of the child that would be born, there is no interference with that choice. Rather, an abortion chosen for those reasons would implicate the different decision of whether to have a particular kind of child. To ban sex-selection (or race-motivated, or disability-based) abortion is not to interfere with the choice of whether or not to have a child — the right Roe and Casey seem to have in view — but to interfere with the choice to abort where continued pregnancy and childbirth otherwise would be chosen, but for reasons of the traits of the child to be born.
Similarly, Roe refers to “the distress, for all concerned, associated with the unwanted child.”44×44. Roe, 410 U.S. at 153. But, with trait-selection abortion bans, it is not the child that is unwanted; it is a child of a particular type that is “unwanted.” It is an unwanted girl. Or a child unwanted because of his race or disability. That is a materially different situation from the one contemplated in Roe.
The same point can be made with respect to “autonomy” or “bodily intrusion” or “burdens of pregnancy” or “physical constraints” arguments for the right to abortion, prominent in both Roe45×45. See id. (stating that “[t]he detriment that the State would impose upon the pregnant woman by denying this choice altogether is apparent” and listing medical and psychological burdens of pregnancy generally; the possible “distressful life and future” occasioned by “[m]aternity, or additional offspring;” the “distress, for all concerned, associated with the unwanted childhood;” and in some cases the “stigma of unwed motherhood”). and Casey.46×46. Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 852 (1992) (“The mother who carries a child to full term is subject to anxieties, to physical constraints, to pain that only she must bear.”) Whatever the intrusions or burdens occasioned by pregnancy and childbirth, they are burdens that exist independently of the race, sex, or disability of the child. The very real burdens of pregnancy do not justify eugenics abortions specifically. In short, the right to abortion as formulated in Roe (and in Casey) overshoots the reasons the Court offers in support of the right.
Second, the “undue burden” standard of Casey can be understood in a similar sense, as limited by the scope of what is thought the relevant right. Casey’s holding was that the abortion right is violated, pre-viability, by any law that imposes an “undue burden”47×47. Id. at 876. on the right to choose abortion — that is, by any law that has the “purpose or effect” of placing a “substantial obstacle in the path of a woman seeking an abortion of a nonviable fetus” so as to interfere with the “woman’s right to make the ultimate decision” as to whether or not to have any particular abortion.48×48. Id. at 877. Taken literally, this language would forbid trait-selection bans on the abortion choice. Such bans directly prohibit the ultimate decision, for abortions had for certain reasons. Such laws literally impose a direct legal obstacle to that particular abortion choice.
It seems fair to observe, however, that the “undue burden” formula is vague, amorphous, malleable, and adjustable. In a scheme of judicial development of an unwritten right, it should be open to the Court to conclude that prohibitions on trait-specific abortions do not burden in a relevant way the right to abortion itself, but only the right to trait selection of born children through the means of abortion. Such laws do not directly burden the decision to have an abortion, but only one reason for having an abortion. Just as employment antidiscrimination laws do not forbid hiring and firing, but only certain illegitimate reasons for a particular employment decision, trait-selection abortion bans do not “strike at the right [to abortion] itself.”49×49. Id. at 874. Thus, while Casey’s specific doctrinal formulation would appear to preclude trait-selection bans, Casey does not appear to be all about the doctrine. More realistically, it is a case concerned with a supposed “balancing” of interests, as Roe also purported to be. Where that balance needs to be adjusted, the Court should adjust it.50×50. A similar argument can be made for modification of the Court’s health-exception doctrine. As noted above, the health exception requires that abortion be permitted, even after viability, as a matter of constitutional right, for “emotional, psychological,” and “familial” reasons. See supra note 4. If this exception is construed so as to embrace a right to late-term abortion for reasons of trait selection (on the theory that these fit within the description of abortions had for “emotional,” “psychological,” or “familial” health considerations), the Court certainly would be justified in revising its health-exception doctrine to avoid such a seemingly outlandish and doubtless unintended result.
Third, prohibitions of eugenics-motivated abortion protect distinctive “compelling interests” that plausibly override previous balances struck by the Court’s abortion jurisprudence. In theory, the Court’s judicially fashioned presumptive abortion liberty is subject to being overridden by a sufficiently “compelling interest,” including the compelling interest in protecting viable human fetal life. Even though the Court has not, to date, credited the interest in protecting human fetal life at all stages of fetal development — treating pre-viability living fetal life as only “potential life”51×51. Casey, 505 U.S. at 886 (plurality opinion). — an argument can certainly be made that, even under current doctrine, the state possesses a compelling interest in protecting such unborn human life from destruction for eugenics reasons. For example, a prohibition on sex-selection abortion — where a child otherwise would be born but for her sex — furthers compelling interests in eliminating sex bias in society generally, in affirming the equal moral worth of women and girls, and in preventing the creation of troubling gender imbalances in the population of those who are allowed to be born.52×52. Cf. Roberts v. U.S. Jaycees, 468 U.S. 609, 623 (1984) (finding the elimination of sex discrimination to be a compelling state interest). Likewise, race-based or disability-based abortion bans further compelling interests in forbidding discrimination on these grounds, in condemning these forms of highly socially undesirable animus, in affirming the equal human dignity of all, and preventing eugenicist ends.53×53. Cf. Burwell v. Hobby Lobby Stores, Inc., 573 U.S. 682, 733 (2014) (“The Government has a compelling interest in providing an equal opportunity to participate in the work force without regard to race . . . .”); Tennessee v. Lane, 541 U.S. 509, 522 (2004) (holding that the Fourteenth Amendment prohibits “irrational disability discrimination”). While these interests are not necessarily more compelling in principle than the interest justifying the protection of preborn human life from destruction generally, they are distinctive to the eugenic abortion situation in particular.
Fourth, the situation of trait-specific, eugenic abortion turns purported “gender equality” or “equal protection” arguments for abortion rights back on themselves. The “sex discrimination” or “women’s equality” argument for abortion has always been doctrinally flawed and analytically weak: abortion restrictions do not classify on the basis of sex; they regulate the conduct of men and women; they are designed not to subjugate women but rather to protect fetal human life; and they are drawn to address that interest specifically.54×54. For a fuller presentation of this position, see Paulsen, Worst Decision, supra note 4, at 1008 n.34 & 1008–10, and Michael Stokes Paulsen, Paulsen, J., Dissenting, in What Roe v. Wade Should Have Said 196, 204–07 (Jack M. Balkin ed., 2005) [hereinafter Paulsen, Paulsen, J., Dissenting] ; see also Paulsen, Paulsen, J., Dissenting, supra at 206 (“Those who invoke women’s equality to create a constitutional right to abortion in the end permit mothers (influenced, perhaps, by others) to selectively kill their unborn children because of the child’s sex.”). Nonetheless, the claim retains a certain superficial, intuitive appeal because the burden of pregnancy, and thus of abortion prohibitions, falls uniquely on women as a consequence of human biology.55×55. The disparate social and economic burdens of pregnancy and parenthood on women are, however, socially (and legally) constructed and susceptible to remedies that more equally impose such burdens on men and women. See Paulsen, Paulsen, J., Dissenting, supra note 54, at 216 n.9; see also Michael Stokes Paulsen, Men, Abortion, Sin, and Salvation, Pub. Discourse (May 1, 2014), https://www.thepublicdiscourse.com/2014/05/13080 [https://perma.cc/A8XS-DSGH]. The presumed remedy is (on this view) to give women the freedom to abort their pregnancies, so that women as a class are not adversely impacted.
Sex-selection abortion prohibitions pose a challenge to this intuition. There is something deeply wrong when a right to abortion, championed in the name of female gender equality, produces a constitutional right to abort human embryos and fetuses for being female. Standard-issue feminist arguments for abortion rights, whatever their merit in general, simply do not work in this setting. To whatever extent Roe and Casey rest on sex-equality premises, those premises fail to supply a justification for eugenics-based abortions specifically.
None of this is to deny that, for prohibitions on trait-selection abortion to be upheld, there would need to be fairly major revisions in the Court’s current abortion jurisprudence. The legal premises necessary to sustain such a ban would almost inevitably undermine seriously, if not contradict outright, the premises and doctrines on which Roe and Casey depend. For openers, as noted above, to uphold a law forbidding abortion based on a fetus’s human traits would at least implicitly involve recognition, at some level, of the humanity of the fetus — a subtle but fundamental assault on Roe’s underlying premises. Upholding trait-selection prohibitions would also require holding that abortion, even of a nonviable fetus, can be prohibited when done for at least some reasons (for instance, eugenics reasons), or it would require holding that such a prohibition is not an “undue burden” on the abortion choice — both significant changes. It might also involve holding that a “compelling interest” exists in prohibiting at least some abortions even pre-viability — a different major change in doctrine. Lastly, it may even involve holding that the unborn, living human embryo and fetus are legally entitled to the status and constitutional rights of a legal “person” — a hugely significant change.56×56. See generally Michael Stokes Paulsen, The Plausibility of Personhood, 74 Ohio St. L.J. 13 (2013) (setting forth the textual, structural, and historical arguments supporting this view and reaching the conclusion that it is entirely plausible and defensible, but not incontrovertible).
These are major changes. It is hard to view them as entirely consistent with current abortion doctrine, even if that doctrine is plausibly distinguishable or capable of being reformed to accommodate such shifts. Murray is right in her basic instinct. Trait-selection bans pose a direct challenge to the legal regime of Roe and Casey. It is that threat that she wishes to ward off, by invoking the shield of stare decisis and criticizing Justice Thomas’s discussion of race and disability as an effort to do an end run around that supposed shield.
IV. Staring Down Stare Decisis
Do principles of stare decisis require striking down trait-selection abortion bans? Does stare decisis forbid making major revisions to —or outright repudiation of — the Court’s abortion jurisprudence? Does stare decisis require adherence to Roe? Does stare decisis even require adherence to Casey’s version of stare decisis, and to Casey’s purported reliance on the doctrine to justify adhering (mostly) to Roe? Is there, as Murray suggests, an implicit “race exception” to an otherwise fairly strict rule of stare decisis? And is some such exception really needed in order to justify overruling Roe and Casey (as Murray argues it is, and fears Thomas is trying to create)?
The answer to all these questions is No. Start with the easiest and narrowest point: notwithstanding the doctrinal tensions between trait-selection bans and the abortion regime of Roe and Casey, nothing in those decisions out-and-out decided the specific issue of a claimed right to abortion for eugenics purposes. Under almost any conception of the force of precedent, trait-selection bans present a different and distinguishable issue. Roe and Casey protect the right to choose not to have a child — the right to be (or not to be) a parent — not the right to choose to have a child but not one of a particular race, sex, or disability. Justice Thomas made this obvious point in his Box concurrence, and Judge Easterbrook made the point in his partial dissent from denial of en banc rehearing in the court below.57×57. As Justice Thomas wrote in Box: “Whatever else might be said about Casey, it did not decide whether the Constitution requires States to allow eugenic abortions.” Box v. Planned Parenthood of Ind. & Ky., Inc., 139 S. Ct. 1780, 1792 (2019) (Thomas, J., concurring). And as Judge Easterbrook crisply put it:
Casey and other decisions hold that, until a fetus is viable, a woman is entitled to decide whether to bear a child. But there is a difference between “I don’t want a child” and “I want a child, but only a male” or “I want only children whose genes predict success in life.” . . . None of the Court’s abortion decisions holds that states are powerless to prevent abortions designed to choose the sex, race, and other attributes of children.
Planned Parenthood of Ind. & Ky., Inc. v. Comm’r of the Ind. State Dep’t of Health, 917 F.3d 532, 536 (7th Cir. 2018).
But more broadly and fundamentally, the judicial doctrine of stare decisis cannot — constitutionally cannot — preclude the Supreme Court from reconsidering and rejecting doctrines and past decisions that it is persuaded conflict with what the Constitution actually provides and permits. Simply put: if the proper task of constitutional interpretation is to faithfully interpret and apply the document itself, then past judicial decisions at odds with the document simply cannot be followed as binding authority in a subsequent case. This is the simple principle of constitutional supremacy — the very principle that fuels the argument for judicial review in The Federalist No. 78 and Marbury v. Madison. It follows that precedent can inform, guide, persuade — and perhaps even furnish a baseline from which a subsequent interpreter must justify departure — but it cannot revise the Constitution itself.58×58. See Michael Stokes Paulsen, Originalism: A Logical Necessity, Nat’l Rev. (Sept. 13, 2018, 11:20 AM), https://www.nationalreview.com/magazine/2018/10/01/originalism-a-logical-necessity [https://perma.cc/B9RC-7TZK]. Any version of the doctrine of stare decisis that would purport to bind the Court to past holdings and doctrines in conflict with the Constitution is quite literally unconstitutional.59×59. This is a telescoped version of arguments I have made in other writing. See supra note 17. See especially Paulsen, The Irrepressible Myth, supra note 17, at 2731–34, making the argument that the premises and logic of Marbury conflict with the judicial doctrine of stare decisis, if understood to require or counsel deliberate adherence to a precedent the Court is otherwise persuaded is wrong; Paulsen, The Intrinsically Corrupting Influence, supra note 17 (arguing that stare decisis conflicts with every theory of constitutional interpretation). Numerous other scholars and jurists have made the same or similar constitutional argument against according binding prospective force to erroneous judicial precedents. See generally, e.g., Gary Lawson, The Constitutional Case Against Precedent, 17 Harv. J.L. & Pub. Pol’y 23 (1994) (concluding that the Supremacy Clause and Marbury forbid judges to follow precedent in conflict with the Constitution); Caleb Nelson, Stare Decisis and Demonstrably Erroneous Precedents, 87 Va. L. Rev. 1 (2001) (concluding that demonstrably erroneous precedent cannot constitutionally be regarded as authority displacing the meaning of the Constitution itself). Justice Thomas has adopted a version of this position. Gamble v. United States, 139 S. Ct. 1960, 1984 (2019) (Thomas, J., concurring) (concluding that “demonstrably erroneous” precedents must be corrected, whether or not other factors support overruling the precedent). For an important argument that principles of due process constitutionally forbid a strict conception of stare decisis that would preclude the legal rights and interests of persons who were not joined as parties to a prior proceeding, see Amy Coney Barrett, Stare Decisis and Due Process, 74 U. Colo. L. Rev. 1011 (2003).
The Court itself acknowledges that the doctrine of stare decisis is not in any way required by the Constitution but is simply a matter of general judicial policy and practice.60×60. Paulsen, Abrogating Stare Decisis by Statute, supra note 17, at 1537 & n.1, 1543–51 (citing the Court’s formulations to this effect). The doctrine is not required by any rule of law fairly traceable to the Constitution’s text, structure, or history — and the Court has repeatedly said it is not (including in Casey).61×61. Id. at 1570–82. It is commonplace that the Court frequently overrules its past decisions. (Casey overruled two decisions.) The judicial doctrine of stare decisis remains very much in a state of flux, with different Justices struggling to articulate their own formulations of and alterations to the doctrine.62×62. Just last Term, two notable cases produced multiple competing opinions of the Justices concerning the proper understanding and application of the doctrine. See June Med. Servs. LLC v. Russo, 140 S. Ct. 2103 (2020); Ramos v. Louisiana, 140 S. Ct. 1390 (2020).
In light of these first principles, and a realistic description of the limitations of the doctrine even as a practical matter, it is difficult to know what to make of Murray’s theorized implicit “race exception” to her otherwise fairly strict conception of stare decisis. One can certainly see, as Murray does, in decisions like Brown v. Board of Education63×63. 347 U.S. 483 (1954). (repudiating Plessy v. Ferguson64×64. 163 U.S. 537 (1896). in substantial part) and last Term’s decision in Ramos v. Louisiana,65×65. 140 S. Ct. 1390 (2020). that concerns over racial injustice — and a history of racist motivations behind legislative policies cast in ostensibly neutral terms but designed to further racial subjugation — figured prominently in the Court’s reasoning on the merits of those controversies. But the simpler explanation of those decisions is that the Court felt it necessary to overrule past erroneous precedent because of the gravity and clarity of the constitutional errors involved.
There is no reason to believe that the Court’s obligation to overrule serious constitutional errors is limited to contexts involving race. The Court’s obligation to overrule serious constitutional errors is the Court’s obligation to overrule serious constitutional errors. Racial injustices forbidden by the Constitution but perpetuated by judicial precedent are just one illustration of a broader principle: the Court must get the Constitution right, and the imperative of doing so feels most pressing on matters that loom large. Brown was clearly one such case.
But Roe and Casey — the issue of abortion generally and of trait-selection abortion bans in particular — clearly pose questions of similar constitutional and moral enormity. And they do so whether race is an overlay or not. On any view of the abortion issue, Roe and Casey are constitutional decisions of enormous consequence. It is important that the Court get the answer right.
Legal doctrine can be a useful tool. In the end, however, what matters are not technical legal nuances or clever doctrinal moves — on questions like whether stare decisis is an otherwise strict rule (but with a race loophole) or whether trait-selection bans can be squeezed within current “undue burden” analysis (or require doctrinal reformulation or repudiation). The doctrines are mere instruments, tools of decision. What matters are the realities and the results.
The reality is that our current constitutional law allows abortion for any reason, including a eugenics reason. The result is that abortions can be obtained, and are obtained, on account of the race, sex, or disability of the child that otherwise would be born.
If this matters, it has enormous consequences for the constitutional law of abortion. That is the ultimate point of Justice Thomas’s opinion in Box. It is a point that Professor Murray would prefer to ignore.
† Responding to Melissa Murray, Race-ing Roe: Reproductive Justice, Racial Justice, and the Battle for Roe v. Wade, 134 Harv. L. Rev. 2025 (2021).
* Distinguished University Chair & Professor of Law, The University of St. Thomas. This essay is dedicated to my nieces and nephews and the diversity of race, sex, ability, and nations of birth they reflect. My thanks to Stephen Gilles, Guy-Uriel Charles, Elizabeth Schiltz, and Teresa Collett for comments on an earlier draft. All errors and opinions are mine.