From Jus Soli to Jus Sanguinis Without Anyone Voting

A country can change its theory of citizenship by amending its Constitution after public debate. It can also change its theory of citizenship by passing four statutes over thirty-three years, none of which describes itself as doing what it does. India has taken the second route. The Constituent Assembly chose jus soli — birthright citizenship — over jus sanguinis — descent citizenship — after specifically considering both. Article 11 left the permanent regime to Parliament. Four amendments later, the regime Parliament has built is closer to jus sanguinis than to jus soli, and the last of those amendments has added a religious qualification to the naturalisation track. Each amendment passed through the legislative procedure. None of them generated the constitutional debate the original 1949 settlement received. The cumulative effect is a different country, arrived at without anyone being asked to vote on the question.

I claim that what Parliament has done is constitutionally permissible in the narrow sense — Article 11's grant of legislative competence is plenary — but constitutionally unauthorised in a deeper sense. The Constituent Assembly's choice in 1949 was a structural choice. It was made publicly, debated, recorded, and tied to a specific international stance against racial citizenship laws elsewhere. Article 11 was the means of implementing that choice, not of reversing it. To use Article 11 to reverse it — silently, in four pieces, over three decades — is to use the form of constitutional authorisation to defeat the substance of constitutional choice. That is not what the framers wrote, and it is not what we owe the document.

The 1949 settlement: which theory of citizenship, and why

The Constituent Assembly's deliberations on citizenship took two years. Article 5 — the foundational provision conferring citizenship at the commencement of the Constitution — was, in Ambedkar's recorded description, the Article that gave the Drafting Committee 'such a headache' that he could no longer count how many drafts had been prepared and rejected.¹ The headache was substantive. Two theories of citizenship were available. Jus soli would make every person born on Indian soil a citizen. Jus sanguinis would make citizenship turn on descent. The two principles produce different polities. Jus soli countries — the United States, France, much of the Americas — incorporate immigrants quickly through the second generation. Jus sanguinis countries — Germany before 2000, Japan, the Gulf states — produce long-resident non-citizen populations whose children remain outside the polity.

The Sub-Committee on Fundamental Rights, in March 1947 — before Partition was announced — chose jus soli explicitly. It described jus sanguinis as 'an idea of racial citizenship' and rejected it as inconsistent with 'enlightened, modern civilized' constitutional design.² The choice was made before the Partition introduced communal pressure into the citizenship question. Vallabhbhai Patel told the Sub-Committee that 'the provision about citizenship will be scrutinized all over the world' — a remark whose meaning becomes clearer when one remembers that the Indian National Congress had spent forty years opposing racial citizenship laws against Indians in South Africa, Kenya, Burma, and Malaya.¹⁶ A jus sanguinis Constitution in 1949 would have undermined the diplomatic position India had inherited from the freedom movement. The 1949 settlement was, in this sense, doubly considered: it was both a substantive constitutional choice and a foreign-policy choice. It was not a default.

The Constitution that emerged from this choice has Articles 5 to 10 as a closed sequence dealing with citizenship at the commencement of the Constitution. Article 5 confers citizenship by birth, by descent of one parent in India, or by five years' residence. Articles 6 and 7 deal with Partition-era migration. Article 8 deals with overseas Indians. Article 9 disqualifies those who have voluntarily acquired foreign citizenship. Article 10 preserves rights of citizenship subject to later legislation. Article 11 — the provision under which all subsequent legislation has been passed — reads: 'Nothing in the foregoing provisions of this Part shall derogate from the power of Parliament to make any provision with respect to the acquisition and termination of citizenship and all other matters relating to citizenship.'¹⁰

Article 11 is unusual. It does not grant a fresh substantive power. It says that what Parliament does under that power 'shall not derogate' from the foregoing Articles. The drafting is careful. The Constituent Assembly knew that the Articles 5-10 regime was a snapshot for the moment of commencement, and knew that Parliament would have to legislate the permanent regime. What Article 11 did was to tell Parliament that the permanent regime should sit on the foundation Articles 5-10 had laid — not displace it. DD Basu's commentary records the point with the precision a generation of constitutional lawyers were taught to expect: Parliament's competence under Article 11 is plenary, but the constitutional structure of Articles 5-10 sets the orientation Parliament was expected to honour.¹⁷ That orientation, recorded in the Assembly debates, was jus soli.

The 1955 Act: the orientation honoured

The first Parliament to legislate under Article 11 was the first elected Parliament of the Republic. The Citizenship Act 1955, as originally enacted, was a clean statement of jus soli for the post-Constitution generation. Section 3(1) read: 'Every person born in India on or after the 26th day of January, 1950, shall be a citizen of India by birth.'³ There were two narrow exceptions — children born to fathers with diplomatic immunity and children born during enemy occupation to enemy-alien fathers. Neither carried any sense of being a substantive limitation on the principle. The principle was that India was an inclusive polity by birth.

The 1955 Act sat alongside provisions for citizenship by descent (section 4 — children born outside India to Indian fathers), citizenship by registration (section 5 — persons of Indian origin and certain spouses), citizenship by naturalisation (section 6 — ordinary statutory route with a five-year residence requirement among other conditions), and citizenship by incorporation of territory (section 7 — Sikkim, Goa, Pondicherry). The architecture was symmetric: a primary jus soli rule with subsidiary descent and procedural routes. It was the architecture the Constituent Assembly had chosen.

For thirty-one years the regime held. The case law of those years dealt almost entirely with the Partition-era questions Articles 6 and 7 had left open. Central Bank of India v Ram Narain in 1955 established the factum-plus-animus rule for domicile during the Partition years.¹¹ Kulathil Mammu v State of Kerala in 1966 took a broad view of migration, holding that movement between India and Pakistan during the Partition did not require an intention to settle.¹² These cases were inclusive in temper. They read citizenship widely against the backdrop of the abnormal conditions of 1947-49. The post-Partition Court did not approach citizenship as a gate to be closed; it approached it as a status to be construed in favour of the person who claimed it. The contrast with later case law is sharp.

1986: the silent reversal begins

The Citizenship (Amendment) Act 1986 was passed in the wake of the Assam Accord of August 1985. The Accord, signed between the Government of India, the Government of Assam, and the All Assam Students' Union, set three cut-off dates for migrants from East Pakistan and Bangladesh: those who had migrated before 1 January 1966 were to be treated as citizens; those who had migrated between 1966 and 25 March 1971 were to register as foreigners and wait ten years; those who had migrated after 25 March 1971 were to be treated as illegal.¹⁴ The Accord ended a six-year agitation that had included the Nellie massacre of February 1983, in which over two thousand people were killed. The political pressure on Parliament to legislate the Accord into the Citizenship Act was enormous.

The 1986 Amendment did three things. First, it inserted section 6A into the 1955 Act to implement the Accord's three-tier framework for Assam. Second, it rewrote section 3 — the citizenship-by-birth provision — to add a parental-citizenship qualification. From 1 July 1987 onwards, a person born in India would be a citizen by birth only if 'either of his parents is a citizen of India at the time of his birth'.⁴ Third, it tightened section 5 on registration. The first of these was Assam-specific. The third was administrative. The second was a fundamental structural change.

The change to section 3 turned the 1955 Act's jus soli rule into a hybrid. Birth in India still mattered — it remained the place of birth that mattered for the new section 3 too — but birth alone no longer sufficed. The parent had to be an Indian citizen. The implication was that a child born in India to two non-Indian parents was no longer a citizen by birth. This was a significant move. It introduced descent into the foundational provision of the Act. It did so without describing itself as doing so. The legislative discussion of the 1986 Amendment focused on the Assam crisis. Section 3 was changed almost as a side-effect of the Assam-specific legislation. There was no debate framing the question as a choice between jus soli and jus sanguinis. The choice the Constituent Assembly had debated for two years was reversed in passing.

I should be careful here. The 1986 Amendment did not eliminate jus soli. It modified it. A child born in India to one Indian-citizen parent and one non-citizen parent remained a citizen. The rule was that one parent had to qualify. This is still a more inclusive rule than full jus sanguinis would be — full jus sanguinis would require both parents, or a single qualifying parent for matrilineal or patrilineal descent. The 1986 rule was, at the time, an intermediate position. The structural significance lay not in what the 1986 Act did in isolation but in what it opened the door to. Once descent was admitted as a qualifying criterion in section 3, the question for the next amendment was not whether descent should be relevant but how much weight it should carry.

2003: the door opens further

The Citizenship (Amendment) Act 2003, which came into force in December 2004, hardened the parental-citizenship requirement.⁵ Section 3(1)(c), in its post-2003 form, provides that a person born in India on or after the commencement of the 2003 Amendment is a citizen by birth only if either (i) both parents are citizens of India, or (ii) one parent is a citizen of India and the other is not an 'illegal migrant' at the time of birth. The 'illegal migrant' definition came in alongside, in a new section 2(1)(b): an illegal migrant is a foreigner who has entered India without valid travel documents or who, having entered legally, has overstayed.

The 2003 change is structurally more significant than the 1986 change. Two consequences follow from it. First, a child born in India to two non-citizen parents is now ineligible for citizenship by birth, regardless of how long the parents have been resident. Second — and this is the consequence that has drawn less attention — a child born in India to one Indian-citizen parent and one parent who is an 'illegal migrant' is also ineligible. The disqualification by parentage now extends not just to descent but to legal status of one parent. A child born to an Indian-citizen mother and a Bangladeshi-illegal-migrant father is, on the face of the statute, born stateless. The statute leaves no path for that child to become a citizen by birth.

This is a substantial departure from the 1955 Act's principle. The 1955 Act's jus soli rule had two exceptions — diplomatic immunity and enemy occupation — both of which were near-impossible in ordinary peacetime conditions. The 2003 amendment's exceptions are not narrow exceptions. They define a substantial category of persons, concentrated geographically in eastern and northeastern India, whose children are excluded from the foundational citizenship provision. The 2003 amendment did not call itself a reversal of the jus soli principle. It was framed administratively, as a tightening of identification procedures and as a follow-on to the Sarbananda Sonowal litigation.⁸ Politically, the framing was about illegal migration from Bangladesh, not about the structural shape of Indian citizenship. The structural shape changed anyway.

The Supreme Court's judgment in Sarbananda Sonowal v Union of India in 2005 is part of the same arc. The Court struck down the Illegal Migrants (Determination by Tribunals) Act 1983 — which had reversed the burden of proof for alleged illegal migrants in Assam — on Article 14 grounds. The Court accepted, in its reasoning, a Governor's 1998 report describing Bangladeshi migrants as 'almost exclusively Muslims' and treated the migration as a threat of 'external aggression' under Article 355. The judicial reasoning ratified what the legislative reasoning had been doing: it tied the citizenship question to a specific religious-identity worry without naming the religion explicitly. The 1986 and 2003 amendments operated through religion-neutral text. The Sonowal judgment cited religion-naming reports as part of its reasoning. The asymmetry is significant. The statute book was kept clean. The constitutional reasoning was not.

In 2004, separately, the Citizenship Rules 1956 were amended to insert rule 8A, devolving citizenship-registration power to district collectors in Rajasthan and Gujarat specifically for 'minority Hindus with Pakistan citizenship'.⁹ This was the first openly religion-named provision in Indian citizenship law. It applied to a narrow administrative question — which officer would process applications from a specific community in two specific States — but its drafting broke the convention that religion would not be named in the citizenship statute book. The rule was an administrative amendment to subordinate legislation, not an amendment to the Act. The Act itself remained religion-neutral on the surface. The convention, however, was broken. The Delhi High Court's decision in Talat Jamal Siddiqui v Union of India a few years later illustrates the asymmetric operation: a woman with both parents born in pre-Independence India was denied a PIO card because she had once held a Pakistani passport.¹⁵

The case law inverts as the statute inverts

The case law of the 1990s and 2000s should be read alongside the statutory changes. The change in the Court's posture between the Partition-era cases and the post-1990 cases is striking and runs parallel to the change in the statute.

Louis de Raedt v Union of India in 1991 illustrates the change.¹³ The petitioner was a Belgian missionary who had been resident in India since 1937 — fifty-four years of continuous residence at the time of the case. He claimed citizenship under Article 5(c), which conferred citizenship on persons ordinarily resident in India for five years immediately before the commencement of the Constitution. The Supreme Court rejected the claim. Residence alone was insufficient; intention to make a permanent home — animus manendi — had to be established. The Court found that de Raedt had applied for a one-year extension of his residence permit in 1980, and treated this as inconsistent with an intention to reside permanently.

The 1991 reasoning would have been hard to sustain in 1955. The Partition-era cases, dealing with Indians who had moved between India and Pakistan in the chaos of 1947-49, had construed intention generously. Central Bank of India v Ram Narain said explicitly that in the conditions of Partition 'men's minds were in a state of flux' and that intention requirements should be applied with care. Kulathil Mammu went further and held that intention to settle was not even required for the meaning of 'migration' in Articles 6 and 7. The early Court treated intention as something to be construed in favour of citizenship.

By 1991, when the petitioner was a Belgian missionary rather than a person displaced by Partition, intention had become a high bar. By the time the post-2000 case law on Bangladeshi migrants developed, intention was being used the other way: a person who had acquired a ration card, an electoral identity card, or even an Indian passport could be told that documentary citizenship was potentially fraudulent and that intention to settle in India was suspect. Niraja Gopal Jayal's account of this inversion in Citizenship and Its Discontents is the most careful in the literature.⁷ The same concepts — intention, domicile, residence — were used in 1955 to confer citizenship and in the 2000s to deny it. The doctrinal vocabulary stayed constant. The direction of doctrinal force reversed.

2019: religion in the text

The Citizenship (Amendment) Act 2019 — the CAA — completed the move that the 1986, 2003, and 2004-Rules amendments had begun. The 2019 Act did two things. First, it amended the definition of 'illegal migrant' in section 2(1)(b) of the 1955 Act to exclude persons belonging to the Hindu, Sikh, Buddhist, Jain, Parsi, or Christian communities from Afghanistan, Bangladesh, or Pakistan who entered India before 31 December 2014. Second, it inserted section 6B to provide a special naturalisation track for such persons, reducing the residence requirement from twelve years to five.⁶

The 2019 Act differs from its predecessors in one important respect. The 1986, 2003, and 2004-Rules changes operated through religion-neutral text — or, in the case of the 2004 Rules, through subordinate legislation that named the religion but in a narrow administrative context. The 2019 Act named religion in the principal Act. The classification was: six named religious communities (excluding Islam, but the statute does not say so — it operates by listing the included communities) from three named countries within a specified time window. The drafting is precise. The political effect of the precision is to create a religion-conditioned naturalisation track in the body of the Act.

I do not, in this essay, take a position on whether the 2019 Act survives Article 14 scrutiny. That question is pending before a Constitution Bench. The question I am taking a position on is narrower. The 2019 Act, whatever its constitutional fate, completed a thirty-three-year transformation of Indian citizenship law from a jus-soli regime to a hybrid regime in which descent, parental legal status, country of origin, and now religion all condition the operation of the citizenship rules. The transformation was achieved through four statutes, three of them drafted in religion-neutral language, the fourth naming religion explicitly. None of them was passed after a public debate framing the choice between jus soli and jus sanguinis as a constitutional choice. The choice was made, in pieces, by a Parliament that had the authority to make it under Article 11 — but did not, in any single moment, hold up the choice for what it was.

The counter-argument: Parliament was authorised, and the Constitution does not require jus soli

The strongest counter-argument is constitutional rather than political. It runs as follows. The Constituent Assembly's debates are not the Constitution. The Constitution is the text. The text in Articles 5 to 10 is silent on the choice between jus soli and jus sanguinis as a long-term principle — those Articles deal with citizenship at the commencement of the Constitution, not with the regime for persons born after. The regime for persons born after was, by Article 11, left entirely to Parliament. Parliament has used the power Article 11 conferred. The 1955 Act, the 1986 Amendment, the 2003 Amendment, and the 2019 Amendment are all valid exercises of that power, unless any specific provision in them violates a substantive constitutional right. The 1986 and 2003 Amendments raise no Article 14 issue because they apply uniformly across India. The 2019 Amendment raises an Article 14 issue, but a narrow one. The structural claim — that Parliament has reversed the jus soli principle without authorisation — fails because there was no constitutional principle requiring jus soli in the first place.

This is the argument the Union of India makes, in substance, whenever the citizenship amendments are challenged. It is a serious argument. It is correct on its narrow point. Article 11 does confer plenary legislative competence. Articles 5 to 10 do address citizenship at the commencement of the Constitution and do not, in their text, prescribe a permanent regime. The Constituent Assembly debates are persuasive authority on intention but not binding on Parliament. Each of the four amendments was passed by the prescribed legislative majority. Each remains on the statute book. The 2019 Amendment is the only one currently under serious constitutional challenge, and even that challenge is narrow.

The narrow point being correct does not, however, dispose of the structural claim. The structural claim is that constitutional choice can be made not only in the text but in the considered drafting history of the text, and that a Parliament exercising a power conferred by the Constitution should honour the orientation in which the conferring provision was placed. This is a stronger claim than the textualist counter-argument allows for, but it is not an exotic claim. It is the claim that lies behind, among other things, the basic-structure doctrine. The basic-structure doctrine holds that there are features of the Constitution so fundamental that they cannot be amended even by Article 368. The features are not listed in the text. They are derived from the structure of the document, from the constitutional history, and from the values the Court reads into both. If structural derivation works for Article 368 — to constrain the express amendment power — it should work, in some attenuated form, for Article 11 too.

The attenuation matters. I am not arguing that the 1986 and 2003 Amendments are unconstitutional. I am arguing that they were, on the Constituent Assembly's recorded view, the kind of constitutional change that should have been made through a constitutional amendment — through Article 368 — rather than through ordinary legislation under Article 11. The framers' choice in 1949 was to write a Constitution that committed itself to jus soli through the means of the substantive Articles read together with Article 11's grant of legislative competence. Reversing that commitment is a constitutional act. It was performed as a legislative act. The form was permissible. The form chosen evaded the debate the substance would have required.

What the four amendments together describe

Take the four amendments together. The 1986 Amendment introduced parental citizenship as a qualifying condition for birthright citizenship. The 2003 Amendment hardened the qualification by adding the legal status of the non-citizen parent. The 2004 Rules amendment named religion in subordinate legislation. The 2019 Amendment named religion in primary legislation. The trajectory is consistent: each step introduces a new criterion that has the effect of excluding categories of persons from the citizenship regime; each step is framed administratively or with reference to a specific local pressure; none of the steps describes itself as a structural transformation. The structural transformation is visible only when the four are read together.

By 2024, the operating regime is this. A child born in India is a citizen by birth only if (a) both parents are citizens, or (b) one parent is a citizen and the other is not an illegal migrant. An illegal migrant is a foreigner without valid travel documents or one who has overstayed. The illegal-migrant definition has a religious carve-out: a person of certain named religions from certain named countries within a certain time window is not an illegal migrant for these purposes. The carve-out has both administrative effect (the person can be naturalised on a shorter track) and structural effect (a child born in India to such a parent and an Indian-citizen parent is a citizen by birth; a child born to a Bangladeshi-Muslim parent and an Indian-citizen parent is not).

This is a different regime from the 1955 regime. The 1955 regime treated citizenship-by-birth as a near-universal entitlement for those born on Indian soil. The 2024 regime treats citizenship-by-birth as conditional on parental status, with parental status itself conditioned on religion and country of origin. To call the 2024 regime jus soli is to use the term in a way that erases its content. The current regime is, in its operating shape, predominantly jus sanguinis, with an additional religious filter on the descent path.

The Constituent Assembly did not choose this regime. The 1949 Constitution does not require this regime. Parliament chose this regime, in four pieces, over thirty-three years. The constitutional position is that Parliament was within its competence under Article 11 to do so. The constitutional honesty would be to admit that what Parliament has done is a structural reversal of a structural choice, and that the reversal was effected without ever being held up for debate as a structural reversal. The textbook habit, again, occludes this. The textbook treats the 1955 Act as the implementation of Articles 5-11 and treats the 1986, 2003, and 2019 amendments as administrative refinements. The amendments are not administrative refinements. They are the constitutional choice the Constituent Assembly thought it had foreclosed, being reopened by stealth.

Where this leaves us

The position I take is not that the four amendments should be struck down. The 1986 and 2003 amendments would not, on any plausible reading of the existing case law, fall on Article 14 or any other Part III ground. The 2019 amendment is under challenge, and the Constitution Bench will decide that question on the narrower grounds appropriate to it. The structural argument I have made in this essay does not provide a doctrinal handle for the Court to use, and I do not pretend that it does. The argument operates at a different level. It is an argument about how a constitutional system should respect the choices made by the document that constituted it.

What I would press for, instead, is two things. First, the citizenship question should be debated as a citizenship question — that is, as a constitutional question about what theory of citizenship the country has chosen and intends to keep — rather than as a recurring administrative question about specific migrant populations. The 1949 framers held the debate at that level. The 1986, 2003, and 2019 Parliaments did not. The next time citizenship legislation comes before Parliament, the debate should be at the constitutional level, not the administrative one.

Second, the case for using Article 368 rather than Article 11 for further structural changes should be acknowledged. Article 368 is the express constitutional amendment power. Article 11 is a delegated legislative power. The two are not interchangeable when the legislation in question is itself a structural choice. The Constituent Assembly's careful drafting of Article 11 — the explicit non-derogation clause — was an attempt to mark this distinction. The four amendments have, between them, eroded the distinction. A Parliament that wishes to make further structural changes to the citizenship regime should accept the procedural burden of doing so under Article 368, with the supermajority and (in appropriate cases) State-ratification requirements that Article 368 imposes. That is what the constitutional architecture asks for. It is what the silent path through Article 11 has avoided.

The Constitution India has is not the Constitution India had. Most of the change has been openly debated — the Preamble insertions, the right to property, the Ninth Schedule expansions, the basic-structure doctrine's own elaboration. The change to the theory of citizenship has not. We arrived at a different country by a route that was constitutionally permissible and constitutionally evasive at the same time. The route may not be reversible. The honesty about how we got here can still be recovered. The Constituent Assembly chose jus soli, considered the alternative, and rejected it for reasons it recorded. The Republic that emerged from that choice is not the Republic that emerged from the four amendments. Whether we should mind this is a separate question. That we should know it is not.

Get the week's biggest decisions, distilled.

A weekly digest for lawyers, students and the legally curious. Every Friday morning. Free.