South Africa: The Clause That Came First, and the Room It Cannot Reach

Two things are true of South Africa at once, and neither is contested. It was the first country in the world to write a prohibition on discrimination by sexual orientation into its constitution, and its highest court then used that clause to strike down the criminalisation of gay men in 1998 and to open marriage to same-sex couples in 2005. The Civil Union Act followed in 2006, making South Africa the fifth country to recognise same-sex marriage and the first on its continent. It also lives with rape and with the killing of women by their partners at levels its own scientists have measured three times over, with a care few countries apply to themselves — and the women that clause was written to protect are among the women the violence reaches.

A sentence arrives almost by itself here, and this essay will not use it: that the gap proves the law was never sincere, that a progressive constitution sitting on this much violence is hypocrisy, a document written for an audience abroad. It is satisfying, it can be produced without leaving one’s desk, and it explains nothing. It does not say who the hypocrite is. It predicts nothing. And it is spoken with equal enthusiasm by people arguing that rights of this kind are a foreign import with no local root, and by people arguing that law in general is theatre.

What is left, once it is set aside, is a question this chamber has circled for some time and never had so clean a case for. What can a constitution actually do, what can it not do, and where is the line? South Africa is the sharpest test because the law is not weak here and the violence is not hidden here. The answer argued below is that a constitution changes what can be claimed and who may bring a case, and that it does not by itself change who is alone at a taxi rank at night, who depends economically on the person who hurts her, or whether the charge office takes a statement. Law sets the floor of what is arguable. It does not staff the room.

The interim Constitution in force from 27 April 1994, the day of the first universal election, already barred discrimination on the ground of sexual orientation. The Constitution of 1996 carried the ground into section 9(3), alongside race, gender, sex, pregnancy, marital status, disability, religion, culture and birth. No other country had done it. How it arrived matters later: it was negotiated and drafted, argued for inside the liberation movement and by a lobby organised for the purpose, and adopted by a constituent assembly. It was never put to the public as a question of its own, because nothing in that text was.

The court then did what a court holding such a clause can do. In October 1998, in the case brought by the National Coalition for Gay and Lesbian Equality, it confirmed that the common-law offence of sodomy and the statutory provision beside it were invalid, with effect running back to the day the interim Constitution took force. The judgment is usually filed under dignity and belongs there, but it did something more concrete that this essay returns to: it removed the position in which a man assaulted during a criminal act could not go to the police without describing an offence of his own.

Seven years later the court reached marriage. In Minister of Home Affairs v Fourie, decided on 1 December 2005 in a judgment by Justice Albie Sachs, it held unanimously that excluding same-sex couples could not survive the equality and dignity clauses, and gave Parliament twelve months. Parliament passed the Civil Union Act 17 of 2006, in force from 30 November that year. One provision of it is the most instructive thing in this section. Section 6 allowed a state-employed marriage officer to refuse, on grounds of conscience or belief, to solemnise a union between two people of the same sex. The right existed; the counter could still turn you away. It was repealed only in October 2020. Fourteen years sat between what the Constitution granted and what a person could obtain at a government office — and that gap, at its smallest and most bureaucratic, is this essay’s subject.

A decade after the Civil Union Act, a nationally representative attitudes survey by the Human Sciences Research Council found seventy-two per cent of South Africans regarding sex between people of the same sex as morally wrong, while fifty-one per cent held that gay people should have the same human rights as everyone else. Its authors called the country progressive prudes, which describes it better than either half alone. This is not a public that demanded the clause and received it. It is a public that has largely accepted a settlement it did not itself arrive at, while holding a private view that runs the other way.

That sequence — protection drafted and litigated ahead of the opinion around it — is both the strength of the settlement and the precise location of its exposure. The strength is that a right which does not depend on majority feeling does not evaporate when the feeling turns: it sits in a text, it can be asserted in any forum, and a government that dislikes it has to argue against it in public and lose. The exposure is that such a right has no crowd behind it in the street; it is enforced at the point where somebody goes to court, and between those points it is a sentence in a book that many of the people it governs do not accept. None of which makes a rule that rights must be earned by persuasion first: the countries that waited for majority consent waited a very long time and some are waiting still. A right established ahead of opinion is not weaker than one established behind it. It is differently situated, and what it needs afterwards is not more legitimacy. It is reach.

The South African Police Service recorded 42,569 rapes in the 2023/24 financial year. Before that figure is used for anything, three things it is not have to be said, because this is where writing about South Africa usually goes wrong. Recorded crime counts cases opened with the police. It is not a count of incidents and not a measurement of how many people were harmed. Reported, recorded and prevalence are three different quantities that move for different reasons — reporting with trust in the police, recording with how a station classifies a case, prevalence with what happens to people — and any one can change while the others do not.

This also disposes of the league table. South Africa is routinely placed at or near the top of international rankings of rape, and those rankings are not sound, for reasons that have nothing to do with protecting anyone’s reputation. Since the Criminal Law (Sexual Offences and Related Matters) Amendment Act of 2007, South African law defines rape as any act of sexual penetration without consent, regardless of the sex of either person: broader than several countries still use, and mechanically productive of a larger count from identical conduct. Countries also differ in whether a course of conduct is one case or many, and in how likely a person is to walk into a police station at all. The United Nations office that publishes the figures says in its own notes that they should not be used for direct comparison between countries.

What can be said is worse rather than better, and comes from work that does not depend on police counts. The South African Medical Research Council has run three national mortuary-based studies of women killed — 1999, 2009, 2017 — reading post-mortem and investigation records directly. Among women aged fourteen and over, femicide fell from 24.2 per 100,000 in 1999 to 12.6 in 2009 and 11.1 in 2017; killing by an intimate partner fell from 9.5 to 6.6 to 4.9. The decline is large and real and this essay is obliged to say so, because an argument that needed the numbers frozen would be an argument rather than a look at the world. The researchers nevertheless described the 2017 intimate-partner figure as more than four times the estimated global rate for intimate partner and family-related homicide. And behind all of it sits the part no official count can see: a household study in Gauteng, published in 2010 and now too old and too local to carry anywhere, estimated that roughly one in twenty-five rapes in that province had been reported to the police at all. The recorded number and the real number are nowhere near each other, and every argument here about what law reaches has to be read with that distance inside it.

The phrase so-called “corrective” rape came out of South African townships and activist organisations to name a specific thing: the rape of a lesbian woman by a man or men who present the act as a correction of her. It carries the offender’s own justification inside it, which is why careful writers keep the quotation marks and some refuse the term altogether. The point here is structural, and it is why South Africa belongs in this chamber at all. The constitutional protection and the violence are not falling on two populations that happen to share a country. They land on the same women.

Two names are in the public record. Eudy Simelane, who had played for the national women’s football team, was killed at KwaThema in April 2008; of the men charged, two were convicted, and the proceedings did not treat her sexual orientation as an element of the offence, because no legal category then existed in which it could be one. Noxolo Nogwaza was killed at the same township in April 2011. A petition with well over a hundred thousand signatures followed, and the Department of Justice set up a national task team on violence against people targeted for their sexual orientation or gender identity later that year. The Prevention and Combating of Hate Crimes and Hate Speech Act was signed into law in May 2024.

One absence in that paragraph is itself evidence. There is no national figure for how often this happens, because it is not a distinct offence in South African law and recorded crime is not broken down in a way that would show it. Everything known about its scale comes from organisations that collect cases, from clinical and shelter records, and from research interviews. The thing the Constitution most specifically promised to protect against is the thing the state has never been able to count — and a protection that cannot be counted against is one whose failure cannot be measured. That is not an unenforced law. It is a law whose enforcement has no denominator.

Put the accusation at full strength first, because a weak version would prove nothing. A country that meant this would not look like this. The clause was a flourish at a founding moment, adopted when a new order needed to show itself the opposite of the one it replaced, and written by an elite for an audience that was partly international. Underneath it nothing was built. Therefore the law is a performance, and the response to a performance is to stop applauding.

The first thing wrong with that is grammatical. Hypocrisy is a property of an agent: one mind that says one thing and does another. There is no such mind here. A constituent assembly adopted the Constitution; a court that commands no police officers enforces it; individual men commit the violence in particular streets and houses. The failure to answer it is not a decision taken anywhere. It is distributed across a charge office short-staffed at two in the morning, a docket that goes missing, a prosecutor’s caseload, a laboratory backlog, a magistrate, a family that advises a woman not to proceed, and a wage she cannot replace. To call that combination hypocritical is to assign an intention to a system with no single place where intentions are held, and it insults the people inside it who did their part properly.

The second thing wrong with it is that it does no work. Suppose it is true: what follows? The reading is compatible with every possible state of the world. If enforcement improves, the law was always sincere and the critic has said nothing; if it does not, the law was always theatre. A reading that cannot be wrong is telling nobody anything about South Africa. Set it aside and what remains is not a smaller question but a harder one, about capacity. Writing a rule and reaching a room are not the same kind of act. They cost different things, they fail for different reasons, and a country can be genuinely excellent at one while being inadequate at the other with no dishonesty involved anywhere.

It changes what can be claimed, and who may walk into a court and claim it. That sounds thin until one sees what depends on it. Before 1994 a gay South African had no claim to make: the state was not obliged to justify its treatment of him, because there was no ground on which the treatment could be challenged. After section 9(3) the burden reverses. The state is the party that has to explain itself, in public, on the record, against a standard it does not control. That is the difference between a grievance and a case.

Watch the mechanism in the decisions themselves. Fourie could not have been argued without the clause; the whole of it consists in taking a listed ground seriously and asking the state to justify an exclusion it could not justify. The Home Affairs official who would rather not register the marriage now has no lawful way to refuse, and since October 2020 not even the statutory objection he once had. The 1998 judgment is more concrete still: while an act is a crime, a person cannot report being assaulted during it without confessing to something, so criminalisation does not merely stigmatise — it removes access to the police for a whole class of people, and decriminalisation gives that access back. None of these are cultural changes. They are changes in what is available at a counter and in a courtroom, and nothing but law produces them.

Hence the formulation. Law sets the floor of what is arguable. A claim with constitutional standing does not disappear when the mood turns against it; it waits, it can be picked up by anyone with standing in any decade, and the other side has to meet it. Which is why the sequence in the earlier section matters: a protection that ran ahead of opinion is precisely one that does not need opinion in order to keep existing. This is the thing a constitution does that nothing else does, and any reading that calls it theatre has failed to notice that the theatre is enforceable against the government.

It does not staff the room. A national study of 3,952 rape cases reported to South African police in 2012 followed them through the system and found about nine in a hundred ending in a conviction; in Gauteng the figure was 6.5 in a hundred, against 6.2 in the earlier Tracking Justice study of cases from 2003 using comparable method. Both begin with cases that were reported, which the earlier section established is a small fraction of what happens. The Constitution says, with complete clarity, that everyone has the right to freedom and security of the person, including freedom from all forms of violence. It says nothing about how many detectives are on duty, whether the docket survives, or whether the laboratory result arrives before the trial date.

Three ordinary things sit outside the clause entirely, and they decide most of the outcome. The first is geography and hour: who crosses open ground to get home, who waits at a taxi rank after dark, who has a door that locks. The second is money. In the second quarter of 2025, Statistics South Africa recorded unemployment at 35.9 per cent among women against 31.0 per cent among men, and labour force participation at 54.9 per cent against 65.6 per cent. Those are conditions of dependence, and dependence is the mechanism by which a woman stays in a house she should leave — not because she does not know her rights, but because leaving costs a roof. The third is the charge office: whether the person behind the desk at two in the morning writes the statement down, believes it, and opens a case. No bill of rights determines any of the three, and all three happen before any right becomes relevant.

Against the essay’s own framing it should be said that the state has not treated this as unreachable. South Africa adopted a national plan on gender-based violence and femicide in 2020, and in January 2022 the President signed three statutes together: one reworking bail, evidence by closed-circuit television and complainants’ participation in parole proceedings; one widening the national register of sex offenders; and one extending the definition of domestic violence to coercive behaviour and to a child’s exposure to it. These are attempts to build the room rather than to restate the rule, and it is too early to say what they reached. What can be said is that the mortuary studies show intimate-partner killing falling by roughly half between 1999 and 2017, which establishes the thing the cynical reading has to deny: the room is not immovable.

This Library has now written the same problem from three positions, and setting them beside each other is what the chamber is for. In the essay on the vacuum where a law should be, the difficulty was that a whole field of reproductive life in Japan is governed by no statute at all, so that nobody in it — not the woman carrying the pregnancy, not the intended parents, not the child — has anything to hold up. In the essay on the movement made of disclosure, the difficulty was the opposite: Japan revised its Penal Code substantially in 2023, replacing a framework built on resistance with one built on whether consent was genuinely possible, and the social permission to say aloud what had happened barely moved. South Africa is the third position and the extreme version of the second. Its rule is not merely adequate; on its own terms it is the most advanced text of its kind in the world, and the distance between that text and the room it governs is the largest this chamber has had to describe.

Which forces the question the essay has been building toward. What is a law for, if it cannot reach the room? Two answers are available and both are wrong. The cynical one says nothing: it is theatre, judge a society by its outcomes and ignore its documents. That is refuted by the mechanisms of the previous sections — an official who cannot lawfully refuse, a person who can report an assault without confessing a crime, a state that must justify itself rather than merely act. The naive one says everything: pass the right law and the rest follows. That is refuted by six or seven convictions in a hundred reported cases, and by the fourteen years in which a marriage officer could still turn a couple away from a counter the Constitution had already opened.

The answer proposed here sits between them and is not a compromise. A law is a standing argument that outlives the mood that made it. Its function is to be available — to the person who needs it, at the moment everything else has failed, in a form the other side has to answer. It is the thing a woman can hold up; it does not do the holding, and never did. Everything that determines whether she is in a position to hold it up, the hour and the wage and the door and the officer at the desk, is built by other means and paid for out of other budgets. For a reader in Japan the consequence cuts both ways. The 2023 revision is not nothing and it is not the end of anything: it is a standing argument that did not exist before and will still be there when the attention that produced it has gone. But the work remaining after it is not legislative and will not be done by the next amendment. It is the work of the room, and South Africa is the proof that this second kind of work does not follow from the first, even where the first has been done better than anywhere on earth.

The strongest objection is that the charitable framing works as an alibi. South Africa has had thirty years, a capable civil service in parts, a research establishment that has documented this problem better than most countries document anything, and a budget. Under-resourced charge offices and lost dockets are not weather. They are the results of choices about what to fund and whom to promote, made by people with names, over three decades, and an essay that dissolves those choices into the general difficulty of reaching a room has helped the people who made them. The objection is serious and this essay does not defeat it. It answers only that the difficulty is real in every country, including wealthy ones with low recorded violence, and that saying so is not the same as excusing anyone.

A second objection is that the essay extends a charity it would withhold elsewhere: would it describe a state that failed lesbian women this badly with this much structural sympathy if the constitution in question were not one it admires? The test is uncomfortable, and the only defence is that the same analysis has been applied in this chamber to Japan, where the sympathy is no warmer. A third is that an outsider comparing two countries he lives in neither of is doing tourism with a thesis. That is true as a risk, and the answer is method: hold every claim to a source, refuse rankings, and say where the essay is reading rather than counting.

The limits are these. Recorded crime figures count cases opened and are used as such. The femicide studies are mortuary-based and blind to deaths outside official mortuaries, and perpetrator information was missing in a rising share of cases, so the split between intimate and non-intimate killing rests partly on imputation and is an estimate rather than a tally. No national count of so-called “corrective” rape exists and none is claimed. The attrition studies follow a cohort from 2012 and cannot speak to the statutes of 2021. The causes of South Africa’s levels of sexual violence are a substantial and genuinely contested literature — the labour migration system of the apartheid economy, poverty and crowding, alcohol, norms of masculinity, the aftermath of political violence — and this essay adjudicates none of it, because its argument does not require a cause and would be dishonest if it chose one for convenience. And no claim is made about the character of any nation, which is not a thing this chamber believes exists.

This house sells evenings in which the terms are written down before anything starts, in which a word ends everything without explanation, and in which the person providing the evening is bound by those terms and has no standing to reinterpret them. Against its own interest first: the house benefits directly from the thing this essay describes. A person who has learned — from her own history, or simply from watching how a stated rule behaves when it meets a room — that a rule can be true and still not reach her is a person for whom a room where the rule is guaranteed to hold has a price. That is the market, and it should be said plainly rather than dressed as a philosophy.

What the house can honestly claim is narrow. It can build one room in which the distance described here is zero, because the room is small enough that the person who wrote the terms is the person inside it. That is a real thing to be able to offer, and it is also the reason it does not scale: the mechanism that makes it work is exactly the mechanism a state does not have. A country cannot be a small room, and anyone who reads this essay as suggesting that private arrangements are a model for public ones has read it backwards.

What this does not claim. Not that South Africa’s Constitution is a failure, not that constitutional protection of sexual orientation is a foreign import, not that law in general is theatre; the essay argues at length against all three. Not that Japan should copy South Africa, or that South Africa should have waited for its public to agree. It ranks no country by recorded rape and asks its reader not to, for the reasons given. It offers no diagnosis of any reader, no safety instruction and no advice to anyone in danger, because an essay is not a plan. It takes no position on South African policy, which is not this Library’s to take. It names no living person except from the public record, and attributes no motive to anyone beyond what a court or an official response established. And it does not claim that understanding the difference between a rule and a room changes anything by itself — which is, after all, exactly what it has been saying about laws.