Long ago I watched a television serial set in Gurdaspur district in the days around Partition. In one scene, a policeman stops a clean-shaven young man whose dress suggests he is a Muslim and asks his name. “Iqbal,” he replies. “Iqbal what?” the policeman demands. “Iqbal Chand, Iqbal Singh or Iqbal Muhammad?” The policeman did not want to know who Iqbal was. He wanted to know what he was.
In 1947, in a district whose own fate at Partition turned on the arithmetic of religion-wise majorities, the second half of a name could decide which side of the line a man would live on, or whether he would live at all. A name has never been a reliable witness to faith. Too often, it has been used as a verdict.
Seventy-nine years later, the State is asking the same question again, this time of registers rather than of travellers.
The Questionnaire That Came Back Blank
On the morning of 29 September 2026, The Hindu’s Vijaita Singh reported what the High-Level Committee on Demographic Changes (HLCDC, the Home Ministry panel chaired by Justice P.P. Naolekar, retired Supreme Court judge) has asked of the States. It is a 35-point questionnaire. It seeks fifteen years of religion-wise data on births, school enrolment, vehicle registrations and property transactions. It seeks district-wise details of religious structures. It asks which districts show unusual spikes in Aadhaar, voter-ID and ration-card registrations. It asks for intelligence assessments and commission reports on conversions among Scheduled Tribes, and for details of detention centres in operation.
The States, it turns out, are stumped. Their departments do not keep most of this information by religion, and never have. A sub-registrar registering a sale deed does not ask whether the buyer bows at a mandir, a masjid, a church or a gurdwara. A District Transport Officer issuing a registration certificate does not ask the owner’s faith. Most of what the Committee wants does not exist, and the only way to produce it now is to guess each person’s religion from his name, which is profiling.
By the afternoon, Jairam Ramesh had called it a “patently diabolical agenda”. That is politics. The more useful questions are administrative and constitutional.
We Said So in May
On 27 May 2026, the morning after the Home Ministry issued its Resolution of 26 May constituting the Committee, I asked on these pages: Counting People or Targeting Communities? The Resolution promised a “scientific study” of demographic change caused by illegal immigration and “other abnormal reasons”. My reading then was that the Committee had a dual character, scientific on paper and operational in design. The inclusion of the Census Commissioner was the tell. It was a preparatory exercise for action.
Two facts sharpened that reading within weeks. The Union Home Minister personally chaired the Committee’s first working meeting and directed field visits to border districts, metros and industrial towns. And, as The Wire pointed out, the Committee has no demographer. Justice Naolekar himself said, with some candour, that demography and illegal migration were new subjects to him. Among the members is Durga Shankar Mishra, former Chief Secretary of Uttar Pradesh and our 1984 IAS batchmate. He knows better than most how a questionnaire travels down a secretariat. By the time it reaches a Tehsildar, the question has become the answer expected.
Today’s report settles the matter. The Home Minister said at the outset that the Committee would study patterns of abnormal population shift at the level of religious and social communities. The questionnaire does exactly that. A committee whose stated concern is illegal immigration has framed its data request around religion.
That is a category error. Illegal immigration is a question of nationality. A Bangladeshi national who crossed the border unlawfully is equally unlawful whether he is Hindu or Muslim. If the concern is infiltration, the variable is citizenship. Religion becomes the variable only if one has already assumed the faith of the infiltrator. A questionnaire that begins with that assumption will end with that conclusion.
By What Authority?
The Committee’s legal footing is worth setting out, because it explains why the States are within their rights to send back blank columns.
The HLCDC is not a creature of statute. It is a Committee of the Union executive, set up by a Resolution. The Union’s executive power extends to subjects on which Parliament may legislate, and that includes citizenship, foreigners and the census. It does not give the Committee coercive power over the States. Land and its registration, police, public order and school education are State or concurrent subjects. The Committee has no power under any law to require a State to create a record it does not maintain. It can ask. The States can answer truthfully that the data are not kept.
The deeper problem is the constitutional one. The right to privacy, affirmed by a nine-judge Bench in K.S. Puttaswamy v. Union of India (2017), requires any State intrusion on personal information to pass three tests. There must be a law, the aim must be legitimate, and the means must be proportionate. A religion-wise dossier on who bought property, who registered a car and who enrolled a child in school fails the first test, because there is no law behind it. It struggles with the third, because a statutory Census is already under way. Religion-wise data gathered to settle whether one community is buying land “unusually” also sits uneasily with the guarantee that the State shall not discriminate on grounds of religion alone.
And the Census 2027 is already in the field. Its reference date is 1 March 2027. It asks every resident’s religion, as every Census has since 1881. It does so under the Census Act, 1948, which makes individual returns confidential and bars their use as evidence. So the Union already has a statutory, confidential and scientifically designed instrument that will produce religion-wise population data within months. Why, then, a parallel and non-statutory trawl through property and vehicle registers? Impatience is one answer. The less comfortable one is that the Census counts people, while this questionnaire watches what particular people are doing.
What Lucknow Struck Off the Register
Is the Allahabad High Court’s order of September 2025 relevant here? Its direction does not bind the Committee. Its reasoning, however, comes uncomfortably close.
On 16 September 2025, Justice Vinod Diwakar directed the Uttar Pradesh government to stop recording caste in police records. He called the practice a legal fallacy and a form of identity profiling. He also rejected the police argument that caste was needed to tell apart accused persons with similar names. Aadhaar, fingerprints, mobile numbers and parents’ names, he said, would serve the purpose. Within a week, the State’s Acting Chief Secretary ordered caste struck out of FIRs, arrest memos and charge-sheets, with parents’ names in its place. Caste signage on police stations and vehicles was to come down, and caste stickers on vehicles became punishable under the Motor Vehicles Act. The one exception was the SC/ST (Prevention of Atrocities) Act, where caste is an ingredient of the offence.
The precedent has limits. The order concerned caste, not religion. It bound Uttar Pradesh alone, and a single judge’s direction to a State government does not bind a Union committee. It dealt with labels attached to individuals in coercive police records. The HLCDC, at least on paper, asks for district-wise totals, which are nearer to statistics than to profiling.
But that last distinction collapses on contact with the facts. The registers carry no religion, so the district totals can only be built by assigning a religion to each buyer, each vehicle owner and each schoolchild, one name at a time. The aggregate would be assembled from exactly the kind of individual labelling that Allahabad struck down. The exemption for atrocities cases states the principle neatly. An identity may be recorded where the law needs it to protect someone, and nowhere else.
The politics are awkward as well. A BJP government in Lucknow has made a caste sticker on a car an offence. A committee of the Union Home Ministry, under the same party, now wants vehicle registrations sorted by religion. Uttar Pradesh goes to the polls in early 2027, and its departments must answer the questionnaire from records that the State has just scrubbed of caste and that never carried religion. What will Lucknow say?
The Column Left Blank
Set this beside the other counting controversy of the season.
In August, the Centre notified the 40 questions of the Census 2027 population enumeration. For the first time since 1931, caste will be recorded. Scheduled Castes and Scheduled Tribes will be coded against their notified lists. Everyone else gets an open column. The enumerator writes down whatever caste the respondent declares, with no pre-compiled list and no drop-down menu. The Congress leadership has written to the Prime Minister seeking the questionnaire’s withdrawal and holding up the Telangana model of a pre-compiled list. Sociologists warn of a repeat of the Socio-Economic and Caste Census of 2011, whose open column threw up more than 46 lakh caste names. Those data were so unmanageable that they were never released.
In fairness, the open column has defenders. They argue that the caste hierarchy is too contested, the synonyms too disputed and the synonymous-caste litigation too live for any list to be neutral. On that view, recording self-declaration verbatim is the honest course. There is something to that argument. But an open column without a published coding frame, a pre-announced method for grouping variants, and an expert body to do the coding in public view can easily become a way of collecting caste data without ever having to use them. The OBCs, whose invisibility was the whole case for the caste census, are not even a coded category.
Put the two exercises side by side. Where the Constitution does want the State to know, it gets vagueness. Caste data feed entitlements under the equality code: reservations, sub-categorisation and the fifty per cent ceiling. Where the Constitution counsels restraint, it gets precision, and a demand for religion-wise registers of private transactions. The religion column is being filled in with zeal. The caste column is being left open, and possibly unused.
For a Sikh, there is a further irony. The Gurus rejected caste; the langar and the sangat were built to dissolve it. Yet the Constitution, through the Presidential Order of 1950 as amended in 1956, recognises Scheduled Caste Sikhs, the Mazhabi and Ramdasia among them. Punjab has the highest Scheduled Caste proportion of any State, and the Census must count them if those entitlements are to mean anything. In March this year, in Chinthada Anand v. State of Andhra Pradesh, the Supreme Court held that Scheduled Caste status survives only within the Hindu, Sikh and Buddhist folds. That was a case where religion data served a constitutional entitlement, and I wrote to the Punjab Chief Secretary asking that SC certificates be scrutinised accordingly. The test I applied then is the one I would apply now. Does the counting serve an entitlement the Constitution confers, or a suspicion the Constitution forbids?
One Faith, Many Folds
Even if the data existed, a spreadsheet column treats religion as one tidy variable. In India, every faith is divided many times over.
Islam in India is Sunni and Shia. Within them are the Bohras, the Khojas and the Ismailis. Then there are the Ahmadiyya of Qadian, whom I came to know as District Magistrate of the Batala police district in 1989–90. They consider themselves Muslims, and Pakistan, by constitutional amendment since 1974, declares that they are not. Jains are Digambara or Shvetambara. India’s Buddhists include Theravada and Mahayana traditions, though most trace their faith to Dr. Ambedkar’s conversion at Nagpur in 1956.
Sikhs are sorted twice over. One sorting is by the castes the Gurus rejected: Jat, Khatri, Arora, Ramgarhia, Mazhabi and Ramdasia. The other is by observance: Amritdhari, Keshdhari, Sahajdhari and patit. And who decides? For the purposes of SGPC elections, the question of who counts as a Sikh has been settled by notification, unsettled by the High Court, and settled again by Parliament with retrospective effect. The Sahajdhari vote has gone in and out of the electoral roll on the strength of executive and legislative fiat.
The Census, wisely, stops at the principal religion and accepts the respondent’s self-declaration. The HLCDC questionnaire cannot.
The policeman’s question in Gurdaspur, “Iqbal what?”, is the question every Tehsildar will now have to put to a sale deed, and Iqbal will not be there to answer it. Asked to reconstruct fifteen years of property buyers by religion from names alone, he has no reliable way to separate a Sahajdhari from a Hindu Khatri, an Ahmadi from a Sunni, or a Mazhabi Sikh from a Christian convert who has kept his Sikh name, the phenomenon my May letter to the Chief Secretary flagged. Whoever does the guessing becomes, in effect, the arbiter of other people’s faith.
That is a function the Constitution deliberately left with the citizen. Freedom of conscience includes the freedom to say what one believes and to which fold one belongs. Once a State begins certifying religious identity for its own purposes, the Ahmadiyya experience across the border shows where that road ends. Self-declaration is the only constitutionally safe rule, and it is the one thing a proxy-based questionnaire cannot deliver.
Why Should a Secular State Care?
Which leaves the question the States have implicitly raised: in a secular republic, why should it matter to the State what faith its citizens profess, or at what rate each faith grows?
Indian secularism has never required the State to be blind to religion. Where the Americans built a “wall of separation”, India chose principled distance. The State keeps away from religion where keeping away protects liberty, and engages with it where engagement protects equality. It counts religion to protect minority educational institutions, to design programmes for backward minorities, and to know whether the Sachar Committee’s findings of 2006 still hold. That is counting in the service of the citizen.
What the State has no business doing is treating the natural growth of any community as a threat to be managed. Fertility is falling across every religious community in India. The National Family Health Survey shows the gap between Hindu and Muslim fertility narrowing, survey after survey. Demographers explain this through education, income and the age at which women marry, and faith has little to do with it. A State worried about “unnatural” demographic change should be looking at Bihar’s fertility rate and Kerala’s, not at whose name is on a sale deed in a border tehsil.
So, does it matter to the secular State? Yes, to the extent that knowing helps it keep its promises. No, to the extent that knowing is meant to sort the citizenry into those whose numbers are welcome and those whose numbers are suspect. The Committee’s questionnaire, as reported, leans the wrong way.
Signal, Not Noise
The HLCDC’s report is due by May 2027, extendable by six months. Between now and then, three things would help.
First, the States should respond honestly and on record, stating which data they maintain and which they do not, rather than improvising proxies. A Deputy Commissioner asked to classify property buyers by religion should decline to guess, and say so in writing.
Second, the Committee should publish its questionnaire and its method, and co-opt a demographer. A “scientific study” whose questions reach the public through a newspaper exclusive, and whose members include no demographer, contradicts its own name.
Third, the Registrar General should publish, before enumeration begins in February, the coding frame by which the open caste column will be turned into usable data. If the caste count is meant to be used, show how. If it is not, say so.
Counting, done honestly, is a service of the republic. Done with a conclusion already in mind, it becomes a charge-sheet. The Census Commissioner sits on the Committee, and we should watch which of the two he is asked to produce.