Turkey: The Convention Named for the City That Left It

The Council of Europe convention on preventing and combating violence against women and domestic violence was opened for signature in Istanbul on 11 May 2011, and has been known ever since by the name of that city. Turkey signed it on the day it opened, first of all the states present. Its parliament ratified it that November without a single vote against, from any party. In March 2012 Turkey deposited its instrument and became the convention’s first state party.

Nine years later, in the early hours of Saturday 20 March 2021, a presidential decision appeared in the Official Gazette announcing that Turkey was leaving. Under the convention’s own rules the departure took effect on 1 July 2021. The convention still carries the name of the city.

That sequence is the spine of this essay. This chamber has spent most of its length reading societies that were moving — at different speeds, by different routes — towards something. Turkey demonstrates that the road runs in both directions. A protection can be argued for, written into statute, submitted to an international monitoring body, and then removed by one signature before dawn.

But "Turkey went backwards" is not the finding, and an essay that stopped there would be doing something dishonest. A country is not the direction its government is facing. The decision was met in the streets of Istanbul within hours, and challenged by more than two hundred applications to the country’s highest administrative court, brought by women’s organisations, lawyers and almost every bar association in the country. To describe Turkey by the decree alone is to erase the people the decree was aimed at, which is exactly what a decree issued at four in the morning would like you to do.

It matters that Turkey was first, and not incidentally first. The convention was finalised under Turkey’s chairmanship of the Committee of Ministers, the drafting had Turkish participation throughout, and the signing ceremony was held in Istanbul. The Grand National Assembly ratified it on 24 November 2011 unanimously — the governing party and the whole of the opposition voting the same way on a question about violence against women. Turkey deposited its instrument in March 2012, before any other state, and the convention entered into force internationally in August 2014 once enough others had followed.

The commitment was not only external. On 8 March 2012 — the date was chosen — Turkey passed Law No. 6284 on the protection of the family and the prevention of violence against women, drafted as the domestic implementation of the convention. It built the machinery: protection and prevention orders available quickly and without the applicant having to produce evidence first, removal of the perpetrator from the home, shelters, enforcement. Whatever the gaps in its operation, the framework was among the more developed in its region, and it had been asked for, loudly and for years, by women’s organisations rather than handed down as a favour.

The gaps were recorded by the body the convention created to record them. GREVIO, its group of experts, published a first baseline evaluation of Turkey on 15 October 2018, acknowledging progress on protection, naming impunity as a persistent concern, and pointing to something that will matter again here: insufficient data on the investigation, prosecution and sentencing of perpetrators. That is the ordinary work of a monitoring relationship — an outside body naming what the state cannot show. It is also the thing a withdrawal ends.

The convention did not arrive in a country with no history of this. In 1926 the Republic replaced Ottoman family law wholesale with a civil code adapted from the Swiss one, abolishing polygamy and unilateral repudiation and giving women equal standing in divorce, custody and inheritance. Women voted in municipal elections from 1930 and in national elections from 1934. For decades this was the Republic’s showpiece, cited abroad as evidence of what it had made of itself.

The second phase was not a showpiece and was not run from the top. In 1990 the Constitutional Court struck down the civil code provision requiring a married woman to have her husband’s permission to work, after a campaign that had made it impossible to defend; the same year the penal provision reducing sentences where the woman raped or abducted had been working as a prostitute was removed. A new Civil Code adopted on 22 November 2001 ended the husband’s legal status as head of the household and made the default matrimonial regime a sharing of property acquired during the marriage, putting a monetary value on unpaid domestic work for the first time. In 2004 a constitutional amendment added to Article 10 that women and men have equal rights and that the state must make that equality real in practice.

This trajectory needs its exclusions attached, or it becomes a flattering story about a state rather than an accurate one about a country. Much of the early phase was legislated from above by a state that was also suppressing other things, and the equality it granted was easier to reach for some women than others: Kurdish women, women who wore headscarves during the decades when that barred them from universities and public employment, rural women and poor women each stood in a different relation to a code written in the capital. And a right that arrives from above without being argued out in public has a specific fragility, which is this essay’s subject. What one administration can grant, another can find it easy to take back.

The reform that best shows what the second phase was doing is the penal one, and it is worth slowing down for, because its substance is grammatical. The code Turkey used until 2005 dated from 1926 and placed rape, sexual assault and abduction in a division concerned with felonies against public decency and family order. In the architecture of the code they were crimes against society. The injured party was public morals and the honour of a family; the woman was where the injury happened rather than the one to whom it happened.

That placement had consequences that were not symbolic. Sentences could be reduced according to whether the woman was a virgin or was married. Prosecution could, in certain configurations, be extinguished if the man married the woman, because the injury to the family’s honour had then been repaired. A code that locates the wrong in a family’s standing keeps producing remedies addressed to a family’s standing, and it did.

The new Turkish Penal Code, Law No. 5237, was adopted on 26 September 2004 and came into force on 1 June 2005. It moved sexual offences into the part of the code dealing with crimes against persons, under a heading naming them as crimes against sexual inviolability, and defined sexual assault as a violation of a person’s bodily inviolability. Marital rape was criminalised. The reductions tied to virginity and marital status went, as did the one available for killings described as defending honour. More than thirty amendments were carried, and a proposal to recriminalise adultery, floated in the same period, was dropped.

None of it was drafted by a government on its own initiative. It was the product of a campaign running from 2002 to 2004 through a platform of more than thirty organisations coordinated by the group Women for Women’s Human Rights, which worked the text clause by clause, held meetings in cities across the country, and put a version of the code in front of the legislature that legislators then had to argue against in public. It was won in the ordinary, tedious, unglamorous way: by people who read a statute closely enough to know which sentence was doing the damage.

The move is worth naming because this Library has watched a version of it elsewhere. Japan’s 2023 revision, which this chamber described in its essay on the movement made of disclosure, was a change in the test rather than in the heading: it replaced a framework built around whether the victim had resisted with one built around whether consent was genuinely possible, and raised an age of consent that had sat at thirteen since 1907. Different mechanism, same underlying question — who is the wronged party, and what is the law actually protecting. Turkey answered that in 2005, which is one reason the matrix calls it a country that had led regionally.

The withdrawal was Presidential Decision No. 3718, published in the small hours of 20 March 2021. Its claimed legal basis was a provision of a presidential decree that the executive read as giving the President power to terminate international agreements. There was no parliamentary debate. The treaty being terminated had been ratified by that parliament, unanimously, nine years earlier.

Two days later the Presidency’s communications directorate gave reasons. It said the convention had been hijacked by a group attempting to normalise homosexuality, that this was incompatible with Turkey’s social and family values, and it noted objections raised by other European states and steps begun by Poland. The convention’s text does not mention homosexuality; what it contains is a definition of gender as socially constructed roles, and an obligation not to discriminate in protection on grounds including sexual orientation and gender identity. The dispute, stated plainly, was over whether protection from violence could be owed to everyone without qualification.

It is worth being exact about what the decision did and did not do, because the loose version of this story is wrong in a way that matters. Leaving the convention did not repeal Law No. 6284 and did not touch the 2005 penal code. A woman in Turkey on 2 July 2021 could apply for the same protection order she could have applied for on 30 June. What ended was an international commitment, an obligation to report, and an outside body with a mandate to publish what the state could not show. The reversal was of the frame and of the accountability, not of the remedy.

That cuts both ways. The frame is not nothing: it is what an organisation cites in court, what a lawyer points to when a prosecutor declines a file, what a monitoring report measures the state against. And the parties that pressed hardest for the withdrawal have since called for the repeal of Law No. 6284 itself, which is still in force and still being defended.

The decision appeared before dawn on a Saturday, which is a choice that tells you what was expected. What happened instead was that by that evening there were crowds in Istanbul, and then in Ankara, İzmir and elsewhere, many carrying the banners of the We Will Stop Femicide Platform — founded in 2010, and occupied ever since with counting women killed by men in Turkey and naming them. The protests continued through the spring and resumed on 1 July, the day the withdrawal took effect.

The legal response was larger and slower. More than two hundred applications were filed with the Council of State, the country’s highest administrative court, seeking a stay of execution and annulment. They came from women’s organisations, from individual lawyers and from bar associations; by the count of women’s groups, seventy-seven of Turkey’s seventy-nine bar associations took the position that the decree was unconstitutional. The Istanbul bar wrote to international bodies and filed its own application.

The constitutional argument was narrow and serious. Article 90 of the Constitution provides that international agreements duly put into effect have the force of law, and that where an agreement on fundamental rights conflicts with domestic law the agreement prevails. On that reading a treaty parliament had ratified had entered the legal order as law, and the executive could not remove a law by administrative act. The counter-reading, which prevailed, was that the power to conclude implies the power to terminate.

It prevailed slowly and not unanimously. In June 2021 the Tenth Chamber declined to suspend the decision, three votes to two, with a dissent arguing that withdrawal was not within the President’s competence. In 2022 the Council of State’s own public prosecutor gave the opinion that the decision was unlawful. On 19 July 2022 the Tenth Chamber rejected the annulment anyway, again by a majority, holding that there was no illegality as to authority or elements. On 2 January 2023 the Plenary Session of the Administrative Law Divisions upheld that ruling, and the domestic route closed.

They lost. The counter-thesis this essay is required to hold is not that they secretly won; it is that the loss is one item in a record that mostly does not read that way. In November 2016 a bill that would have suspended sentences for men convicted of sexual offences against girls whom they subsequently married was pulled hours before the vote, after protests and a march on parliament. A version of the same idea was raised again in 2020 and did not become law; in August 2020 more than three hundred women’s and LGBTI+ organisations founded a joint platform in anticipation of exactly that fight. In 2021 a prosecutor applied to have the We Will Stop Femicide Platform dissolved for acting against law and morality; the case ran through 2022, was rejected on 13 September 2023, and became final when the prosecution did not appeal.

That is what contested looks like. Not a country moving in one direction, but a fight whose score is mixed and whose outcome is not yet written, conducted by people who are the subject of these sentences rather than the background to them.

Any essay about this has to say where its numbers come from, and here the answer is uncomfortable and belongs in the body. The continuous record of women killed by men in Turkey is kept by the We Will Stop Femicide Platform, compiled from news reports, court records and information from families, and published monthly and annually. It is one of the few such records anywhere maintained by a non-governmental organisation. There is no complete official series of femicides published as such against which it can be checked.

The first consequence is evidential. A count assembled from reports will miss deaths never reported as such and will include some whose classification a court would contest; its compilers say so. Its year-to-year movements can reflect changes in what gets reported, and in the platform’s own reach, as well as changes in what happens. This essay therefore reads no trend out of it and makes no claim about whether more or fewer women have been killed since July 2021.

The second consequence is political, and sharper. When the only continuous count is kept by a campaigning organisation, the count and the campaign are the same object, and a state that wishes to dispute the number can proceed against the counter instead. That is not hypothetical: the closure case described above was brought against the organisation that keeps the count.

The official picture is no firmer. Turkey has run national surveys on violence against women — one by Hacettepe University’s Institute of Population Studies for the responsible ministry in 2008 and again in 2014, and a newer one by the Turkish Statistical Institute with Marmara University, fieldwork from November 2024 to January 2025, published in 2025. The newer survey reports figures substantially lower than the older ones, and it was conducted by a different institution, with a different instrument, over a differently defined population. Those cannot be laid end to end to produce a trend, and this essay does not lay them end to end.

So what is this essay reading? A legal sequence, documented to the day: a signature, a unanimous ratification, a code reform, a decision in a gazette, court rulings with recorded vote splits, a closure case with a recorded outcome. That is the evidence it has, and its argument is one a legal sequence can carry. It is not an argument about how many women are harmed, because the material to make that argument honestly does not exist.

Here is the mirror, and the reason this country is in this chamber rather than in a news summary. Japan’s gains in this area are recent. The 2023 revision of the Penal Code is the largest change to the law of sexual offences in over a century, and this Library has treated it as a floor that had finally been set — after a hundred and sixteen years in which the age of consent sat at thirteen and nobody could find the force to move it. The word floor carries an assumption: that a floor, once laid, stays where it is.

Turkey is the case that says a floor can be lifted again. Not by revolution or collapse, but administratively, at four in the morning, by a state exercising a power it says it always had. The instrument that had been the proof of commitment — a treaty, signed first, ratified unanimously, named for the country’s largest city — turned out to be removable by the same kind of act that had produced it.

Set beside that the other case in this chamber where consent moved into law. France, as the essay on the freedom to importune and the trial that moved the shame describes, argued the cultural question in public for seven years, held a trial that changed what the argument could sound like, and only then wrote consent into its legal definition of rape in 2025. It got the instrument last, after the argument. Japan got the instrument almost without the argument. Turkey had both, early, and then lost one.

The comparative finding follows, and it is about durability rather than direction: a right that arrived by decree, treaty or ruling can leave the same way, and what makes it durable is not the instrument. It is whether enough people would notice its absence — whether the thing has been argued for in enough ordinary rooms that taking it away costs something to the person who takes it.

The Turkish case immediately complicates that, and the essay would be cheating not to say so. People did notice. Two hundred lawsuits is what noticing looks like when it is organised and well-advised, and it did not stop the withdrawal. So noticing is necessary and not sufficient. The honest form of the finding is a pair: the instrument is the thing a society can hold a state to, and the argument is what makes holding possible. An instrument without an argument behind it, as the Republic’s early reforms sometimes were, can be withdrawn quietly. An argument without an instrument, as France had for years, has nothing to point at in a courtroom. Turkey is the case where a country possessed both and found that possessing both is still not a guarantee.

This is not a prediction about Japan, and the essay declines to make one. The conditions are not the same, the instruments are not the same, and a reader who leaves having learned to be frightened has learned nothing useful. What the case offers is a different question to ask about any protection, in any country: not is it in place, but who would be inconvenienced by its removal, and how many of them are there.

The essay refuses three readings. The first is that this is a story about Islam. The record does not permit it to be. The 1926 code, the 1990 court decision, the 2001 code, the 2004 penal reform and the 2021 withdrawal all happened in the same country with the same majority religion, and a factor that is constant cannot explain a variable that moves. The stated reason for the withdrawal was contested inside Turkey by observant Muslims as loudly as by anyone else. A reader who leaves with a thesis about a faith has replaced an argument with a prejudice, and has let a specific government off the hook by attributing its decision to a civilisation.

The second is that Turkey stands for a region. It does not. It is one country with its own constitutional order, its own women’s movement with its own history, and its own quarrel about what a republic is for. This Library keeps a separate account of every country it writes about precisely so that none has to carry the weight of a neighbourhood, and this essay makes no claim about any other state.

The third is that the women lost. They lost a case, and this essay has said so without softening it. They also stopped a bill in 2016, stopped its successor in 2020, kept Law No. 6284 in force through years of calls for its repeal, survived an attempt to dissolve the organisation that keeps the country’s only continuous count, and turned a decision taken before dawn into a public argument still running in 2026.

Its limits are these. The femicide counts here are compiled by a non-governmental organisation and no official series exists to check them against; the essay reads a legal sequence and does not count harms. The two official survey programmes are not comparable and no trend is drawn from them. The constitutional dispute is treated in outline. The Kurdish women’s movement and LGBTI+ organising in Turkey are named rather than treated, and both deserve their own account rather than a clause in somebody else’s. Dates and vote splits come from public records and contemporaneous reporting, and where reports differ the narrower claim has been preferred. And it is written from outside, by a house in another country, about a fight it is not in.

This house sells evenings organised around a written agreement and a word that ends anything without explanation, and it should state against its own interest what this essay does to that product. A private guarantee is precisely the kind of instrument the essay has just called fragile. It holds because the house has decided it holds. There is no parliament behind it, no court would hear a case about it, and no outside body publishes a report on whether it is being honoured. It is not a right. It is a promise with good manners, and the difference is the whole subject of this piece.

The house also benefits, commercially, from a reader concluding that public protection is unreliable and private arrangement the sensible substitute. That conclusion is worth money here and is not supported by anything above. The women described in this essay did not respond to a withdrawn treaty by arranging better private guarantees. They filed two hundred lawsuits, kept a count, defended a statute, and went on saying in public what they wanted the law to be.

What this essay does not claim: that Japan will lose anything, or that any particular protection anywhere is at risk; that countries can be ranked, or that one sequence of reform is better than another; that any named organisation will or will not prevail; anything about how a reader should vote, donate or affiliate; that the number of women harmed in Turkey has risen or fallen since 2021, which the available evidence cannot establish; and that the people who filed those applications and lost are a cautionary tale. They are the reason the question here is interesting rather than settled.

What one evening in this house can honestly offer is small, local and unenforceable outside its own door: a few hours in which the exit is written down before anything starts and nobody has the standing to reinterpret it. That is worth something. It is not a floor, it is not a right, and if this essay has done its work the difference between those three things is now harder to lose track of than it was.