On 26 September 2018, a five judge Constitution Bench of the Supreme Court of India delivered the Aadhaar judgment 2018 researchers, lawyers and administrators still argue over, upholding the Aadhaar scheme and its parent statute by four votes to one in Justice K. S. Puttaswamy (Retd.) and Another v Union of India. Justice A. K. Sikri wrote the majority opinion for himself, Chief Justice Dipak Misra and Justice A. M. Khanwilkar. Justice Ashok Bhushan concurred in a separate opinion. Justice D. Y. Chandrachud dissented alone. The combined record ran to roughly 1,448 pages, and its length has encouraged two lazy readings. One reading reports the outcome: the scheme survived, Section 57 fell in part, some provisions were read down. The other reading reports the dissent: Justice Chandrachud would have struck the whole statute down, and his opinion is treated as the verdict’s true centre. Both readings miss the event. The verdict’s durable work was architectural. For the first time after the nine judge privacy bench of 24 August 2017 had recognised privacy as a fundamental right, a Constitution Bench took the proportionality test that the 2017 opinions had assembled and ran it, provision by provision, across a working administrative system used by hundreds of millions of residents. Each section of the Aadhaar (Targeted Delivery of Financial and Other Subsidies, Benefits and Services) Act 2016 was asked the same four questions in the same order: legitimate aim, suitability, necessity and balancing. Section 7 passed all four. Section 57 failed. Sections 33(2) and 47 survived only after the bench rewrote their operating conditions. That patterned application, rather than any single outcome, is what later legislatures, litigants and benches inherited, and it is the pattern this article reconstructs from the record.

How Aadhaar Was Built Between 2009 and 2016

The scheme began as an executive project, not a statute. The Unique Identification Authority of India was created by a notification of the Planning Commission on 28 January 2009, and Nandan Nilekani took charge as its chairman in July 2009 with the rank of a cabinet minister. The first Aadhaar number was issued on 29 September 2010 to Ranjana Sonawane, a resident of Tembhli village in Maharashtra’s Nandurbar district, at a launch attended by Prime Minister Manmohan Singh and Congress president Sonia Gandhi. Enrolment proceeded through registrars, most of them state governments and banks, and the number was linked from the start to a stated administrative purpose: a resident who lacked documents, and whose name appeared in no reliable register, could acquire a portable proof of identity that service providers across India would accept. That purpose carried the project’s political defence for a decade, and the majority opinion of 26 September 2018 returned to it repeatedly.

The legal ground under the project stayed thin for seven years. The National Identification Authority of India Bill 2010, introduced in the Rajya Sabha in December 2010, was examined by the Parliamentary Standing Committee on Finance chaired by Yashwant Sinha, which rejected the bill in a report tabled in December 2011, citing the absence of a privacy law, the experience of the United Kingdom’s abandoned identity card scheme and doubts about the technology. The executive project continued regardless, funded through appropriations. By the time the statute arrived in 2016, enrolments had crossed one billion. The sequence matters for everything the 2018 bench decided. India built the database first and legislated for it afterwards, so the constitutional challenge reached the Court as a challenge to a functioning system on which welfare delivery had already been made to depend in part, a fact the petitioners treated as a reason for stricter review and the Union treated as a reason for restraint.

Opposition inside the constitutional structure surfaced early. On 23 September 2013, a three judge bench in the first Puttaswamy proceedings ordered that no person should suffer for want of an Aadhaar number and that enrolment was voluntary. The Union and several agencies returned to the Court seeking modification, because subsidy programmes in cooking gas, kerosene and the public distribution system had begun to insist on the number in practice. On 11 August 2015, Justices J. Chelameswar, S. A. Bobde and C. Nagappan referred the privacy question to a larger bench, recording that the Union’s own argument, built on M. P. Sharma (1954) and Kharak Singh (1962), denied that the Constitution contained a fundamental right to privacy at all. A five judge bench order of 15 October 2015 confined compulsory use to a short list of schemes and repeated that no benefit could be denied for want of the number. The litigation therefore arrived at 2018 with two layers already set: a contested right awaiting definition, and a paper trail of interim orders that the final bench would have to honour, qualify or displace.

The 2016 Act and the Money Bill Decision

The Aadhaar (Targeted Delivery of Financial and Other Subsidies, Benefits and Services) Act received presidential assent on 25 March 2016. The government of Prime Minister Narendra Modi introduced the bill in the Lok Sabha on 3 March 2016 as a money bill, a classification certified by Speaker Sumitra Mahajan under Article 110(3) of the Constitution. That classification had hard procedural consequences. The Rajya Sabha could only recommend amendments, and on 16 March 2016 it recommended five, moved by Congress member Jairam Ramesh, touching privacy protection, opt out provisions and the breadth of Section 7. The Lok Sabha rejected all five the same day, and the bill was deemed passed. A statute creating a national identity platform, an authority with enrolment and authentication powers, offences and penalties, and disclosure provisions extending far beyond the Consolidated Fund of India thus became law without the upper house’s consent being constitutionally required.

The money bill route is the seed of Justice Chandrachud’s dissent and of the loudest institutional argument in the case. The petitioners’ contention, developed in the hearing by senior counsel Shyam Divan, was that Section 7, the provision tying the Act to expenditure on subsidies, benefits and services, was one thread in a much wider fabric, and that a bill whose pith and substance ranged across identity, authentication, private sector use and criminal liability could not be squeezed into Article 110’s enumerated heads. Attorney General K. K. Venugopal answered that the Act’s core was the targeted delivery of expenditure charged on or drawn from the Consolidated Fund, and that the remaining provisions were incidental to that core. The majority accepted the Union’s framing. Justice Sikri’s opinion held that Section 7 was the Act’s heart, that the other provisions served it, and that the Speaker’s certificate, though open to judicial review, disclosed no constitutional error on this statute. Justice Chandrachud held the opposite with unusual force, a holding examined in its own section below.

The statute the bench reviewed had a definite shape, and the article follows that shape. Section 3 entitled every resident to an Aadhaar number. Section 4 gave the number its evidentiary character, expressly denying that it was proof of citizenship or domicile. Section 6 permitted alteration of demographic and biometric attributes. Section 7 empowered the central and state governments to require the number, or alternative identification where a person had not yet been assigned one, as a condition of receiving a subsidy, benefit or service funded from the Consolidated Fund. Chapter IV built the authentication machinery: a requesting entity could seek authentication of a holder’s identity, the Authority would respond with a yes or no confirmation or with e-KYC details, and Section 29 fenced the resulting data, barring the sharing of core biometric material in any circumstance. Section 33 allowed disclosure of information and authentication records under a court order, and, in its second limb, on the direction of an officer not below the rank of Joint Secretary in the interest of national security. Section 47 confined cognisance of offences under the Act to complaints made by the Authority or an officer it authorised. Section 57, the widest provision in the Act, permitted the number and the authentication service to be used for establishing identity for any purpose, by the state, by a body corporate or by any person, under a law or a contract. Section 59 validated everything done under the pre-statute executive regime. Each of these provisions received a separate constitutional answer in September 2018, which is precisely why the outcome reading, one scheme upheld or condemned, flattens the verdict.

The 2017 Privacy Judgment Supplied the Test

On 24 August 2017, nine judges of the Supreme Court held unanimously that privacy is protected as an intrinsic part of the right to life and personal liberty under Article 21 and of the freedoms guaranteed by Part III of the Constitution. The lead opinion of Justice D. Y. Chandrachud, written for four judges, overruled M. P. Sharma and Kharak Singh to the extent that each had denied a constitutional right to privacy, and located the right across dignity, autonomy and the protection of intimate decision. Justice S. K. Kaul’s concurring opinion addressed informational privacy directly and named the state’s data gathering ambitions as a standing constitutional concern. The 2017 bench did not rule on Aadhaar itself. It built the instrument with which Aadhaar would be ruled on. The sequence, right first and scheme second, is traced in the 2017 privacy judgment’s right-architecture-established framework, and the 2018 case is unintelligible without it.

The instrument was proportionality. Justice Chandrachud’s 2017 plurality described a threefold requirement for an invasion of privacy: legality, meaning a basis in law; need, meaning a legitimate state aim; and proportionality, meaning a rational nexus between the object and the means adopted, with procedural guarantees against abuse. Justice Kaul added the language of balancing drawn from the German and Canadian traditions and from the Court’s own decision in Modern Dental College and Research Centre v State of Madhya Pradesh (2016), where a five judge bench had set out four limbs for testing restrictions on the freedom of trade: a legitimate goal, a suitable connection between measure and goal, the absence of a less restrictive equally effective alternative, and a balance between the measure’s effect on rights and the goal’s importance, a formulation the Court had also used in K. S. Puttaswamy’s own earlier rounds and in the internet shutdown and surveillance orders of the same decade. In September 2018 the Aadhaar majority consolidated these strands into a working four limb test and applied it as a checklist to a statute. That consolidation is the operationalisation this article names. Before 2018, proportionality in Indian privacy law was a doctrine stated at the level of principle. After 26 September 2018, proportionality had a decided application, limb by limb, section by section, with outcomes that turned visibly on which limb did the work.

What are the four limbs of the proportionality test the bench applied?

The four limbs ask whether a measure pursues a legitimate aim, whether it suits that aim, whether a less restrictive and equally effective alternative exists, and whether its gain outweighs its cost to the right. The bench applied them in that order to each contested provision. Section 7 survived all four; Section 57 failed; two provisions were reshaped.

The pedigree of that checklist reaches back through the founding text itself. The Constitution’s rights chapter, whose drafting politics is traced in the constitutional rights framework, qualified each freedom with its own restriction clause, inviting courts to weigh the restriction against its stated ground. Proportionality gave that weighing a discipline the earlier reasonable restriction case law had lacked. It also connected the Aadhaar review to the deepest doctrinal line in Indian constitutional law. The insistence that state power, however valid its source, must not exceed what its purpose can justify is a cousin of the basic structure limitation on amending power, whose establishment in Kesavananda Bharati is examined in the basic-structure-doctrine establishment. Where basic structure polices the Constitution’s outer frame, proportionality polices the interior fit between a law’s means and its ends, and the 2018 bench used it to do interior work on an administrative statute of unusual reach.

The Hearing Record and the Evidence of Exclusion

The Constitution Bench heard argument for 38 days between 17 January 2018 and 10 May 2018, the second longest hearing in the Court’s history after Kesavananda Bharati. The petitioners’ side was led by senior counsel Shyam Divan, with Gopal Subramanium, Kapil Sibal, P. Chidambaram, Anand Grover, K. V. Viswanathan and others appearing across the connected petitions, which numbered over thirty and included challenges by former Karnataka High Court judge K. S. Puttaswamy, by civil liberties organisations, by a serving army officer and by petitioners who had lost benefits in the field. Attorney General K. K. Venugopal led for the Union of India, Additional Solicitor General Tushar Mehta appeared for the Unique Identification Authority, and senior counsel Rakesh Dwivedi argued the Election Commission’s and the authority’s administrative case at length. The hearing record, preserved at the Supreme Court Registry, is the under used primary source for the case: it shows each side conceding and contesting the architecture in detail, months before the opinions rearranged those exchanges into holdings.

Two evidentiary contests ran through the hearing and decided more of the outcome than any abstract proposition. The first concerned exclusion. Jean Drèze, the economist who had worked on the public distribution system in Jharkhand for two decades, filed research based affidavits and wrote, with Reetika Khera and others, that authentication failures in ration shops were translating into cancelled entitlements for households that existed, were eligible and had nowhere else to go. A 2017 study by Drèze, Nazar Khalid, Khera and Anmol Somanchi of Jharkhand’s Aadhaar based biometric authentication in the public distribution system found that the system had not reduced quantity fraud measurably, while genuine households faced fresh transaction costs and, in documented cases, deletion from the rolls. The State of Aadhaar reports, produced by IDinsight for 2018 and 2019 from large household surveys, found exclusion from welfare services attributable to the Aadhaar system at lower rates than the field studies suggested, while recording that a small percentage of a very large beneficiary population still amounted to a large absolute number of affected people. The death of Santoshi Kumari, an eleven year old girl in Jharkhand’s Simdega district in September 2017, whose family said their ration card had been cancelled for failure to link it to Aadhaar, was placed before the Court as the human end of the exclusion argument; the district administration disputed the cause of death, and the dispute itself was part of the record.

The majority’s handling of this evidence shaped its Section 7 holding. Justice Sikri’s opinion accepted that instances of exclusion had occurred, treated them as failures of implementation to be corrected through grievance mechanisms and the proviso to Section 7, which required alternative identification where a number had not been assigned, and refused to convert documented field failure into a finding that the enabling provision was disproportionate. Justice Chandrachud read the same record the other way, holding that where a system conditions food on a fingerprint that manual labour wears away, failure is a feature of the design as applied, and that the state had not discharged its burden of showing the absence of less exclusionary alternatives. The disagreement was therefore evidentiary before it was doctrinal, and any account of the verdict that skips the affidavits cannot explain why the two opinions diverge where they do.

The second contest concerned surveillance. The petitioners argued, on affidavits by technology researchers including Usha Ramanathan and on the Authority’s own documents, that authentication logs, recording the time, place and requesting entity of each transaction, allowed reconstruction of a resident’s movements and habits, and that a central database with biometric identifiers created a permanent capacity for profiling regardless of the intentions of any one government. The Union answered on affidavit that the Authority stored no purpose or location data beyond the transaction record, that biometric data sat in a segregated Central Identities Data Repository, and that Section 29’s bar on sharing core biometrics answered the architectural objection. Rakesh Dwivedi’s argument for the Authority pressed a distinction between a database that verifies identity and a database that tracks conduct. The majority largely accepted that distinction, recording that the design, on the material before the Court, was one of minimal data collection for a limited purpose. Justice Chandrachud rejected the distinction as the case’s central factual error, holding that metadata generated by every authentication request, retained in aggregate, was itself the profiling instrument, and that the architecture could not be saved by reading the statute’s silence as a safeguard.

Decision One: Section 7 and the Welfare Core Survive the Test

The majority began where the statute began, with Section 7, and its reasoning on that section is the template for the whole opinion. The legitimate aim limb was satisfied by the text of the section itself: subsidies, benefits and services financed from the Consolidated Fund of India or of a state are public money, and the Union’s stated aim, that such money reach the person entitled to it rather than a duplicate, a ghost or an intermediary, was accepted as legitimate and pressing. The suitability limb was satisfied by the mechanism: a unique number, de duplicated against biometric identifiers held by the Authority, was rationally connected to the detection of duplicates in a way that paper ration cards and locally maintained lists were not. The necessity limb drew the bench’s closest attention. The petitioners proposed alternatives, among them smart cards carrying beneficiary data on the card rather than in a central repository, and the strengthening of existing audit machinery in the public distribution system. The majority held that the alternatives did not achieve the aim with equal effectiveness across programmes, and that a court could not prefer its own instrument choice to the legislature’s where the chosen instrument passed the rights balance. The balancing limb then weighed a recognised but, on the majority’s findings, bounded privacy intrusion, confined by Section 29’s fences, data minimisation in the enrolment form and the absence of any purpose recording in authentication, against the delivery of food, fuel and pensions to identified beneficiaries. The balance, Justice Sikri held, favoured the section.

Two features of that template repay attention. The first is the role of the proviso. Section 7’s own text stated that where an Aadhaar number had not been assigned, the applicant would be offered alternate and viable means of identification, and the Union gave an assurance, recorded in the judgment, that no benefit would be denied for want of enrolment and that exemption mechanisms would cover residents whose biometrics failed. The majority treated the proviso and the assurance as part of the law under review, so the statute that passed the necessity limb was the statute as softened by its own fallback. The second feature is the treatment of dignity. Petitioners had argued that conditioning subsistence on authentication inverted the relationship between citizen and state. The majority answered within the welfare frame it had adopted: a resident’s dignity, on its reading, was served by a system that delivered her entitlement without the discretion of the local dealer, and the Constitution Bench that had decided the 2017 case had itself recognised that the right to privacy coexisted with the state’s duty to provide welfare to the marginalised. The framing contest here is the same framing contest visible in the integrated three-tensions framework: whether a large state platform is read primarily as an instrument of delivery or as an instrument of control. The majority read delivery. The dissent read control. Proportionality supplied the shared grammar in which the two readings could be compared limb by limb, which is why the verdict, for all its length, reads less like a settlement of values than like a worked example of a test.

Why did Section 7 survive proportionality review when other provisions did not?

Section 7 survived because its justification sat inside its own text. It named a bounded class of payments drawn on the Consolidated Fund, and the Union could point to duplication and leakage in those payments as the aim the number served. Its proviso supplied an alternative route for a person not yet enrolled, satisfying the necessity limb.

Other holdings rested on the Section 7 foundation. Section 139AA of the Income Tax Act 1961, inserted in 2017 to require the number in tax returns and permanent account number applications, was upheld, the majority relying in part on the earlier three judge decision in Binoy Viswam v Union of India (2017) and holding that the measure served the aim of curbing duplicate permanent account numbers and false returns, a ground the bench treated as distinct from the privacy analysis of the parent Act. Section 59, which validated acts done under the executive regime between 2009 and 2016, was upheld on the ground that the legislature could cure the absence of a governing law for past administrative action once the governing law existed. The Prevention of Money Laundering rules requiring the linking of the number to bank accounts, by contrast, failed, for reasons that belong to the Section 57 analysis: a delegated rule, made without the anchoring justification of a statute addressed to a defined welfare purpose, could not survive the necessity and balancing limbs merely by borrowing the parent Act’s authority.

Decision Two: Section 57 and the Private Perimeter Fall

Section 57 authorised the use of the number and of authentication for establishing the identity of any person for any purpose, whether by the state or by any body corporate or person, pursuant to any law or any contract. In the years before the verdict, that clause had done the scheme’s expansionary work. Telecom companies required the number for new mobile connections, under a Department of Telecommunications circular of 24 March 2017 that leaned on an order of 6 February 2017 in Lokniti Foundation v Union of India, a case about verifying existing subscribers. Banks and payment companies built e-KYC onboarding on authentication. Private employers, schools and examination bodies demanded the number. The clause converted a welfare delivery identifier into a general identity infrastructure available to any counterparty strong enough to demand it, and the petitioners placed it at the centre of their architecture argument: a number demanded everywhere cannot be refused anywhere, and consent to its use becomes fiction.

The majority struck down the body corporate and person limb of Section 57, and with it the contractual basis of private authentication. Its reasoning proceeded by the limbs. A legitimate aim could be asserted for state purposes, but the clause as written required no aim at all for private use; identity establishment for any purpose whatsoever, under private contract, delegated the scheme’s reach to the market. Suitability could not repair that defect, because the question at the necessity and balancing stages was whether the privacy surrender involved in making a biometric backed identifier the routine key to private transactions was justified by a public gain the state had identified. The majority held it was not. The portion of Section 57 permitting use under a law survived in a narrowed field, the offending words enabling corporate and private use were severed, and the bench added that any future law authorising private use would itself have to satisfy the Constitution. The Department of Telecommunications circular fell with the clause, and the requirement linking the number to mobile connections was held unconstitutional. The money laundering rule mandating bank account linking fell on the same reasoning: the state had not shown why a measure of that sweep was necessary to its anti laundering aim, and a rule framed by the executive could not borrow Section 57’s authority after Section 57 itself had been cut down.

This is the decision the outcome reading reports, and reported alone it looks like a trimming exercise. Read through the test, it is the verdict’s clearest piece of doctrine. The same number, the same database and the same authentication call were constitutional inside Section 7 and unconstitutional inside Section 57, and the entire difference lay in the justification supplied at the first, third and fourth limbs: a named public purpose, an absence of equally effective alternatives, a balance that could be struck. The verdict thereby announced a rule for the platform age in Indian constitutional law. Identity infrastructure is reviewed purpose by purpose, not as a whole, and each fresh purpose must bring its own justification or fail. The administrative context in which that rule operates, a state whose delivery and regulatory programmes are designed at the centre and executed through platforms, is mapped in the post-2014 PMO architecture context.

The Three-Decision Proportionality Architecture Operationalisation Map

The verdict’s three kinds of answer, upholding, striking down and reading down, are set out together in the Three-Decision Proportionality Architecture Operationalisation Map, which a reader can cite as the summary form of the majority’s method. The map keeps the limb at which each provision was decided in view, because that detail, not the bare result, is what later benches and legislatures have had to work with.

Provision	Question before the bench	Decision of 26 September 2018	Limb or ground that decided it
Section 7	May the state require the number for subsidies, benefits and services funded from the Consolidated Fund?	Upheld	All four limbs satisfied; proviso and exemption assurance counted within the measure reviewed
Section 57, private limb	May bodies corporate and persons use the number and authentication for any purpose under contract?	Struck down	Legitimate aim absent for unbounded private use; necessity and balancing failed
Section 33(2)	May a Joint Secretary level officer direct disclosure in the interest of national security?	Read down	Survived only with a judicial element added to the decision and a hearing for the affected person under Section 33(1)
Section 47	May only the Authority complain of offences under the Act?	Read down	Read to permit a complaint by an aggrieved person whose data or rights were violated

The map also records what the binary readings of the verdict cannot: the bench’s third technique. Between upholding and striking down lies reading down, the construction of a provision so that it operates within constitutional limits it did not state for itself. The 2018 bench used that technique on the two provisions where the state’s need was real but the safeguard was missing, and supplied the safeguard by judgment.

Decision Three: Sections 33(2) and 47 Read Down

Section 33 governed disclosure. Its first limb allowed disclosure in pursuance of an order of a court not inferior to that of a district judge. Its second limb allowed disclosure of information and authentication records, including biometric material within limits, in the interest of national security, on the direction of an officer not below the rank of Joint Secretary to the Government of India. The petitioners attacked the second limb as executive self authorisation over the most sensitive data class in the statute, issued without judicial involvement, without a hearing and without any obligation to record reasons that could be tested. The majority declined to strike the provision down, accepting that a national security ground could be legitimate, but read it down in two respects. A direction under Section 33(2) would require the participation of a judicial officer in the decisional process, so that the disclosure decision carried a judicial element the text had omitted. And under Section 33(1), the person to whom the information related had to be given an opportunity of hearing before a court ordered disclosure, a procedural guarantee the bench located in the proportionality framework’s demand, inherited from the 2017 plurality, that state action invading privacy carry safeguards against abuse.

Section 47 presented the mirror image defect. It barred courts from taking cognisance of any offence under the Act except on a complaint by the Authority or an officer authorised by it. A resident whose biometric data was misused by an enrolling agency, a requesting entity or an official therefore had no route of her own to set the criminal law in motion; the gatekeeper was the institution whose own conduct might be in question. The majority read the section down to permit a complaint by an aggrieved person, describing the original bar as a provision that left victims of data misuse remediless under a statute that criminalised that misuse in Sections 38 to 42. Together the two readings down display the operationalised test at its most characteristic. Proportionality, as the bench used it, was not only a guillotine for provisions that failed. It was a drafting instrument, identifying the safeguard whose absence made a provision excessive and writing that safeguard into the provision’s operation.

Around these three decisions the majority arranged a set of bounded holdings that the outcome summaries usually merge. Authentication records were not to be retained beyond six months, and the regulation permitting a five year archive of transaction records was struck down to that extent, answering the metadata argument in part on the retention question while rejecting it on the profiling question. The Authority was directed not to store authentication purpose and transaction metadata in a form that permitted tracking. For children, the bench held that enrolment rested on parental consent, that a child could exit the system on attaining majority, and that the number could not be demanded for school admission; nor could it be made compulsory for examinations conducted by bodies such as the Central Board of Secondary Education, the National Eligibility cum Entrance Test or the University Grants Commission, since those demands lacked the statutory anchoring Section 7 supplied to welfare delivery. The majority also recorded that the number was not proof of citizenship, following Section 4’s own text, and that a denial of benefits for authentication failure would defeat the Act’s stated purpose, directing that exemption mechanisms in the regulations be operated so that no person’s entitlement failed on a fingerprint alone.

How did reading down differ from striking down in the 2018 verdict?

Striking down removed a provision, as happened to Section 57’s private limb and to the mobile and bank linkage rules. Reading down kept a provision alive on conditions the bench supplied: disclosure acquired a judicial participant, Section 33(1) acquired a hearing, and Section 47 acquired a complainant beyond the Authority. The remedy was matched to the defect in each case.

Justice Ashok Bhushan’s separate concurrence, running to over two hundred pages, agreed with the majority’s conclusions while restating the ground in his own terms. He held that the Act’s provisions did not violate the right to privacy as expounded in 2017, emphasised the enrolment and authentication safeguards printed in the regulations, treated the money bill question as answered by the character of Section 7 and the Speaker’s certificate, and differed from Justice Sikri’s opinion in emphasis on several subsidiary points, among them the evidentiary weight of the Union’s affidavit material on exclusion and the degree to which the Authority’s data handling could be tested on the record. The concurrence matters to the count of the court on the statute’s validity, making the operative result four to one, and to the verdict’s later life: a reader who wants the state’s case stated at its fullest judicial strength will find it in Justice Bhushan’s opinion rather than in the majority’s more compressed treatment.

Justice Chandrachud’s Dissent: The Architectural Alternative

Justice D. Y. Chandrachud’s dissent, running to over five hundred pages, is the longest and most fully argued of the four opinions, and it earns that length by refusing the majority’s architecture at every level. Where the majority reviewed purpose by purpose, the dissent reviewed the system as a system. A central biometric database, compulsory in expanding fields, generating an authentication trail at every use, presented in his analysis a single constitutional problem that could not be dissolved into sections: the problem of a state able, in principle, to assemble a profile of each resident’s transactions across welfare, banking, telephony and travel, and of private actors invited into the same infrastructure. On that premise, his proportionality application inverted the majority’s at each limb. The aim, he held, had never been specified with the precision the test requires, because de duplication of welfare rolls could not justify an identity platform of general application. Suitability was not demonstrated on the evidence, because the Union had not shown that leakage in the public distribution system arose from identity fraud rather than from quantity fraud in the supply chain, a distinction Drèze and Khera’s field research put squarely in issue. Necessity failed because less invasive instruments existed, from smart cards holding data on the card to offline verification, that the state had not seriously attempted. Balancing failed because the burden fell hardest on those least able to bear it: manual labourers whose fingerprints fade, the elderly, residents of districts with thin connectivity, the same population Section 7 claimed to serve.

His doctrinal surroundings were correspondingly wider. He held that the passage of the Act as a money bill was a fraud on the Constitution, a phrase he used deliberately and repeated: the certification bypassed the Rajya Sabha’s constitutional role, the Act’s provisions ranged far beyond Article 110’s enumerated heads, and to sustain the classification would license any future government to enact any law as a money bill by attaching a spending provision. He held that the entire Act failed for that reason alone, and would have struck it down on the threshold ground before reaching privacy at all. Reaching privacy, he held that the architecture violated it: biometric identifiers are irrevocable in a way passwords are not, the authentication record created a surveillance capacity incompatible with the 2017 judgment he had himself authored, and consent extracted as the price of rations, wages under the rural employment guarantee or a mobile connection was not consent in any sense the Constitution could recognise. He held Section 7 itself unconstitutional, reasoning that to condition the receipt of entitlements, which are rights rather than largesse, on enrolment in a system prone to documented exclusion inverted the social welfare commitments of the Constitution’s Directive Principles. He dissented, finally, from the majority’s treatment of the interim orders, holding that the repeated orders of 2013 and 2015 recording voluntariness had been treated by the state as obstacles to be drafted around rather than as holdings to be obeyed.

The dissent’s strength, stated at its highest, is that it tests the majority’s own premises. If proportionality is a discipline of justification, the dissent asks, what does the state have to prove, and did it prove it? If authentication failures at scale were in the record, on what basis did the burden of those failures rest on residents rather than on the system’s designers? If the money bill certificate is reviewable at all, what remains of the review if a statute of this breadth survives it? Those questions did not command the Court in 2018. They have structured the argument in every Aadhaar case since, and in the money bill reference discussed below they moved from dissent towards an open institutional question.

What did Justice Chandrachud mean by calling the money bill passage a fraud on the Constitution?

The phrase named a bypass, not a motive. Article 110 confines money bills to enumerated fiscal subjects and gives the Rajya Sabha only a recommending voice. Justice Chandrachud held that an Act containing an identity platform, offences and private use far outside those heads could not travel by that route, because certification would defeat the bicameral design the article protects.

The government’s case, stated at its strongest across the majority and the Bhushan concurrence, deserves the same precision. The Union’s position, argued by Attorney General K. K. Venugopal, was that a Constitution framed for a welfare state could not be read to disable the state’s principal instrument for reaching the poor with their entitlements; that duplication and ghost beneficiaries in subsidy rolls were documented in official reports over two decades; that enrolment was demand driven and had crossed 1.22 billion numbers by 2018 because residents found the number useful; that the Act’s fences, Section 29’s biometric bar, the segregation of the repository, the criminal penalties in Chapter VII, were real safeguards and not decorations; and that the petitioners’ surveillance account described a capability the statute forbade and the Authority’s affidavits disproved. On the money bill, the Union argued that expenditure from the Consolidated Fund was not an attachment to the Act but its reason for existing, every other provision being machinery for that expenditure’s integrity. A reader who cannot state that case fairly cannot measure what the dissent had to overcome, or what the majority accepted.

After September 2018: The Amendment Act and the Money Bill Reference

The legislature’s answer arrived in 2019. The Aadhaar and Other Laws (Amendment) Act, assented to in July 2019, rebuilt parts of the field the verdict had cleared, on the verdict’s own conditions. It inserted a regime of voluntary use: a resident could choose to establish identity through Aadhaar authentication or through offline verification, using means the Authority would notify, and no person could be denied a service for declining authentication where an alternative identification was offered. It omitted Section 57 altogether, removing the clause the bench had severed rather than leaving a judicially edited text in force. It added an express provision that a child enrolled on parental consent could, within six months of attaining eighteen years of age, cancel the number, converting the majority’s direction into statutory text. It strengthened the Authority’s civil penalty powers for non compliance by entities, created an adjudication and appeal route through the Telecom Disputes Settlement and Appellate Tribunal, and amended the Indian Telegraph Act and the Prevention of Money Laundering Act so that banks and telecom providers could accept the number as one voluntary identification document among others, restoring through primary legislation, and on a voluntary basis, part of the private verification space that Section 57 had opened by contract and the verdict had closed. Litigators and teachers who track these changes circulate the three texts side by side through an online presentation viewer when a class or a chamber needs the struck, restored and amended clauses on a single screen.

Read against the complication this article must face, the 2019 Act cuts both ways, and the honest treatment is to say so. The complication holds that the amendment restored, through alternative drafting, part of the private use the verdict had struck down, suggesting that constitutional review of platform architecture is a speed bump which legislative drafting drives around, and that the pattern worth studying is political persistence rather than judicial method. The record supports a narrower conclusion. The restored use is voluntary where the struck use was compulsory in effect; it rests on primary amendments to three statutes where the struck use rested on a single sweeping clause and a departmental circular; it carries the child’s exit and the offline option the bench had demanded; and it nowhere restores the Joint Secretary’s disclosure power without the judicial element the bench had written in. The legislature accepted the test’s terms and drafted inside them. That is what it means to say the verdict operationalised proportionality as an institutional pattern: not that outcomes froze, but that every later move, the state’s included, has had to be expressed in the test’s vocabulary of aim, suitability, necessity and balance, and drafted to survive it.

The dissent’s afterlife ran on a separate track. On 13 November 2019, a five judge bench deciding Rojer Mathew v South Indian Bank Ltd, a challenge to tribunal reforms enacted through the Finance Act 2017, recorded its doubt about the Aadhaar majority’s money bill reasoning and referred the interpretation of Article 110 to a bench of seven judges. The reference reasoned that the Aadhaar majority had not supplied a workable test for which provisions of a composite statute must answer to Article 110’s heads, and that so large a question about the Rajya Sabha’s place in legislation required the larger bench. The reference leaves the Aadhaar money bill holding standing but expressly unsettled, an unusual posture in which the verdict’s most institutional component is simultaneously law and open question. Doctrinal contests of this kind, where a bench’s framework outlives the certainty of its own applications, are the pattern visible in the most-consequential-constitutional-moment framework: the 1973 bench’s holding mattered less, decade by decade, than the review structure it left behind for later courts to operate. Read alongside the long-arc constitutional-erosion-trajectory, the Aadhaar sequence supplies the complementary case, a doctrine built to discipline executive and legislative design in advance rather than to repair a settlement after the fact.

Why did the 2019 amendment delete Section 57 instead of leaving the edited clause in force?

A clause read down survives as a judicial text layered over a legislative one, and administrators must reconcile the statute book with the law reports at every step. By omitting Section 57 and writing voluntary use, offline verification and the child’s exit into fresh provisions, Parliament gave officials a single clean text to administer and challengers one text to test.

The Machine in the Courtroom: Enrolment, Repository and Authentication

Proportionality review of a database statute required the bench to form a view of a machine, and the opinions repay reading as technology findings. Enrolment, as the Authority’s chief executive Ajay Bhushan Pandey explained to the bench in a presentation spanning several sittings in March 2018, collected a resident’s name, date of birth, gender, address and, optionally, mobile number and email, together with ten fingerprints, two iris scans and a facial photograph. The packet travelled encrypted to the Central Identities Data Repository, where the biometric set was compared against every set already stored, the de duplication step on which the scheme’s claim to uniqueness rests, and a number was issued. The statute called the stored material identity information and divided it into core biometric information, which Section 29 placed beyond sharing or publication, and demographic information, which could be shared within the Act’s channels. Authentication was a separate event from enrolment. A requesting entity, a ration shop’s point of sale device, a bank branch, a telecom store, transmitted the number and a biometric or one time password to the repository and received back a digital yes or no, or, in e-KYC mode, a packet of demographic data and photograph with the holder’s consent. The Authority’s evidence was that it recorded the requesting entity, the time and the response, and did not record the purpose of the transaction or its value.

Each side built a constitutional argument out of that design. For the Union, the design’s modesty was the argument: the repository did not know why a fingerprint had been presented, only that it had been presented and matched, so the system was a verifier of identity and nothing more, and comparisons with surveillance systems that collect content were misplaced. For the petitioners, the design’s spread was the argument: a yes or no answer delivered at the ration shop, the bank, the school gate and the telephone store generates, across millions of daily calls, a skeleton of a resident’s life, and the absence of purpose recording in the central system was beside the point because each requesting entity knew the purpose of its own requests and the statute left the joining of those records under regulated private control. The majority adopted the Union’s description as its factual premise. It held that the information collected was minimal for the purpose, that the repository’s segregation and the Section 29 fences were genuine, and that a possibility of misuse, unsupported by proof of a design for tracking, could not condemn a statute whose text forbade the feared use. Justice Chandrachud adopted the petitioners’ description. His dissent treated the authentication log as the design’s centre of gravity rather than as residue, held that a structure which permits profiling cannot be validated by assurances that profiling is not intended, and added that biometric compromise, unlike a leaked password, cannot be remedied by reissue, so the risk calculus of the balancing limb had to count irreversibility. The verdict’s architecture reading holds both descriptions in view: the test the bench applied is indifferent to which description a future case proves, because the limbs ask the same questions of either, and the split of September 2018 was over the facts to which an agreed test was applied.

The Srikrishna Committee and the Missing Data Protection Law

The Aadhaar case was argued in the shadow of an unfinished statute. On 31 July 2017 the Union had constituted a committee under retired Supreme Court judge B. N. Srikrishna to frame a data protection law, and the committee submitted its report and a draft Personal Data Protection Bill on 27 July 2018, weeks before the Aadhaar verdict. The petitioners used the committee’s existence as an admission: the state’s own process conceded that the legal regime governing the largest biometric database in the world lacked the ordinary apparatus of a data protection framework, an independent regulator with enforcement powers over the state itself, purpose limitation enforceable at a resident’s instance and penalties proportioned to harm. Shyam Divan’s argument pressed the sequencing point. The privacy judgment of August 2017 had made safeguards a component of proportionality itself, not a policy garnish, and a scheme collecting biometrics from 2010, legislated in 2016 and still operating in 2018 without a data protection statute asked the Court to find safeguards in the Aadhaar Act’s internal fences and in regulations the Authority wrote for itself.

The majority’s answer on this point was measured and consequential. It declined to hold the Act hostage to a future statute, treating the Act’s own provisions as the safeguards the test required and adding safeguards by construction where it found gaps, as in Sections 33 and 47. Yet the opinions repeatedly noted the Srikrishna process, and the verdict’s structure is best understood as a bridge: the bench supplied, by interpretation, a provisional version of the protections a data protection law would supply by legislation, on the understanding that the legislative field remained open. The 2019 amendment’s adjudication machinery and penalty provisions were early repayments on that understanding. For the article’s claim, the Srikrishna sequence is further evidence of operationalisation: once proportionality had been converted into a checklist, the absence of a safeguarding statute became a legible, arguable defect with a defined remedy, and both the committee’s draft and the amendment were written in direct response to the checklist’s items, data minimisation, purpose specification, retention limits and an enforcement route for the person affected.

Exclusion, Leakage and the Burden of Proof on Section 7

Beneath the doctrinal contest over Section 7 lay a contest over numbers, and the numbers came from the state’s own reform history. The case for the welfare core rested on findings accumulated over two decades: the Planning Commission’s evaluations of the public distribution system had estimated, for 2004 and 2005, that over half the grain issued for the poor did not reach them; successive committees, including the Wadhwa Committee appointed by the Court itself in the right to food litigation, had documented ghost cards and duplicate entries in state rolls; and states that reformed delivery, Chhattisgarh under a computerised and monitored distribution model being the standard exhibit, showed that plugging identity and quantity leaks could transform a broken programme. The Union’s affidavits carried this record into the Aadhaar case and added programme data: de duplication against the repository had identified duplicate and ghost entries in the public distribution system, in cooking gas connections under the PAHAL programme and in rural employment rolls, and direct transfer of the cooking gas subsidy to bank accounts identified by the number had, on the government’s figures, removed crores of duplicate connections. The bench was not asked to audit these claims to an econometric standard. It was asked whether they were capable of grounding the legitimate aim and suitability limbs. The majority held they were.

The petitioners’ numbers answered in kind. The Jharkhand field studies measured a public distribution system in which biometric authentication at the shop had been made compulsory in 2016 and found exclusion concentrated among households whose members’ fingerprints no longer matched, whose enrolment data contained errors they could not get corrected, or whose ration cards had been cancelled in de duplication drives for want of seeding with the number. Their estimate that a substantial share of cancelled cards belonged to genuine households was contested by the state government, which attributed cancellations to duplicates and deaths, and the contest itself was never resolved by an independent count before the bench. The State of Aadhaar survey programme, surveying tens of thousands of households, reported that most welfare users found the system convenient and that outright denial of a benefit for want of the number was uncommon, while also recording error rates which, applied to a beneficiary population in the hundreds of millions, yielded absolute figures in the millions. Into that evidentiary gap walked Santoshi Kumari’s case, argued not as statistics but as the foreseeable end of a chain: a cancelled card, an unseeded household, a death in September 2017 in Simdega district which the family attributed to hunger and the administration attributed to illness. The majority treated each incident as a failure of administration curable within the statute’s own exemption and grievance provisions, and warned that the Act could not be administered so as to deny benefits where authentication failed, recording the Union’s assurance to that effect. The dissent treated the same incidents as the predictable output of the design, reasoning that a safeguard which depends on the machinery that failed is not a safeguard within the necessity limb. The split is the proportionality era’s characteristic judicial disagreement, and it explains why the verdict satisfies neither camp: the test disciplines the questions, and then the questions turn on evidence that litigation assembles imperfectly and judges weigh differently.

The Majority’s Money Bill Holding in Full

The money bill component of the majority opinion is compact beside the dissent’s treatment, and its compactness became part of the later controversy. Justice Sikri’s reasoning proceeded in three steps. The first step concerned reviewability. Article 110(3) declares the Speaker’s decision on whether a bill is a money bill to be final, and earlier decisions, Mohd. Saeed Siddiqui v State of Uttar Pradesh (2014) among them, had treated the certificate as placed beyond judicial scrutiny. The Aadhaar majority stepped back from that position: a certificate could be examined for constitutional compliance, because a decision the Constitution makes final for the House’s internal purposes cannot authorise the House to expand a constitutional category. The second step identified the test. A bill is a money bill if it contains only provisions dealing with the matters enumerated in Article 110(1), and the bench read that requirement through the Act’s centre of gravity: Section 7, tying the identity platform to expenditure on subsidies, benefits and services from the Consolidated Fund, was the Act’s operative core, and the enrolment, authentication, authority and penalty provisions were machinery for that core, provisions dealing with matters incidental to the enumerated heads. The third step applied the test and found it satisfied. Justice Bhushan’s concurrence agreed, adding that the Rajya Sabha’s recommendations had in fact been considered and rejected by the Lok Sabha in accordance with Article 109’s procedure, so the special procedure had been followed on its own terms.

The dissent’s answer attacked the second step. Justice Chandrachud held that the word only in Article 110(1) does the constitutional work: a bill containing provisions that cannot be traced to the enumerated heads, or to genuine incidence upon them, is not a money bill, whatever its fiscal core. On his reading, large parts of the Act, the creation of a general identity infrastructure, the authorisation of private use under Section 57, the offences chapter, the disclosure provisions, had no destination in the Consolidated Fund and would govern conduct in fields where the Rajya Sabha’s equal legislative voice was the constitutional default. The test, he argued, cannot be centre of gravity, because any omnibus statute can be given a fiscal core by drafting, and a Speaker’s certificate would then conclude a question the Constitution reserves to its own text. The 2019 reference in Rojer Mathew fastened on exactly this divergence, noting that the majority had not articulated how a provision’s fit with Article 110 is to be judged where a statute mixes fiscal and non fiscal matter, and that the question bore on every Finance Act of the preceding years. Until the seven judge bench answers, the Aadhaar money bill holding operates as binding law with an asterisk recorded against it by a coordinate bench, a posture Indian constitutional law has accommodated before, most famously for the property cases of the 1960s, and one that keeps the dissent alive as a working brief inside the system rather than as a monument outside it.

Proportionality’s Lineage and What Operationalisation Changed

The four limb structure the bench applied had travelled. Its modern form was assembled in German administrative law, restated by the Supreme Court of Canada in R v Oakes (1986), absorbed into the jurisprudence of the European Court of Human Rights, and carried into Indian law through the service and trade cases before Modern Dental College consolidated it in 2016 as the test for restrictions on Article 19 freedoms. The 2017 privacy opinions drew on that lineage consciously: Justice Chandrachud’s plurality cited the German Federal Constitutional Court’s census jurisprudence on informational self determination, and Justice Kaul’s concurrence set out the four part inquiry with the safeguards requirement attached. What remained undone after August 2017 was the application of the structure to a statute whose every operative section raised the inquiry in a different register, fiscal, administrative, criminal and private law at once. That is the gap the Aadhaar verdict filled, and it is why teachers of the subject reach for the case despite its unwieldiness. A student who works through the majority opinion watches the same four questions yield four kinds of answer: a provision sustained whole, a provision destroyed in part, two provisions reconstructed, and a threshold institutional question, the money bill, answered on a different doctrinal track altogether, marking the boundary of the test’s jurisdiction.

Operationalisation also fixed the test’s default burdens in privacy cases against the state. After September 2018, a government defending an identity or data measure in an Indian court knows the order of the examination it will face and the record it must file: the aim specified at the level of the impugned provision, not of the programme in general; the suitability evidence derived from the programme’s operation rather than its brochures; the alternatives considered and the reasons they were rejected; and the balance struck with the safeguards stated, because the bench had shown it would supply missing safeguards by construction or sever the use that lacked them. The petitioners’ side gained a matching template, which the dissent’s opinion sharpens: attack the aim’s specification, test suitability against field evidence, propose the less restrictive instrument in working detail, and place exclusion and irreversibility into the balance. Both templates are the verdict’s real bequest to litigation practice, and both are invisible to the outcome reading, which can say only who won.

The Decade of Interim Orders, 2012 to 2017

The final hearing did not begin the case. Retired Karnataka High Court judge K. S. Puttaswamy filed the first writ petition in 2012, challenging a scheme that then rested on an executive notification and a lapsed bill, and the docket grew by accretion for six years. The order of 23 September 2013, that no person should suffer for not having a number and that authorities must check whether applicants possessed one only as a voluntary matter, set the pattern the state spent years contesting: every expansion of compulsory use produced a return to court. In 2014 a bench restrained the sharing of biometric data with investigative agencies in a Goa criminal investigation reference, holding that the Authority could not part with core data without the holder’s consent and a court’s order. The August 2015 reference order recorded the Attorney General’s submission, made by Mukul Rohatgi for the Union, that the Constitution guaranteed no privacy right, an argumentative position the Union abandoned after August 2017. The October 2015 Constitution Bench order confined the number’s use, on the Union’s assurance, to the public distribution system and to cooking gas distribution, and repeated the non denial direction. When the 2016 Act and the 2017 tax and telecom measures multiplied compulsory uses, petitioners argued that the interim regime was being outflanked by legislation and rule making faster than the Court could convene. The decision in December 2017 to constitute the five judge bench, and the January 2018 hearing date, were the Court’s answer.

That prehistory shaped the verdict’s remedies. Because benefits had been linked, unlinked and relinked under interim orders for five years, the majority was deciding not on a blank field but on a settled administrative expectation among beneficiaries and departments that the number would be asked for in welfare delivery, and its Section 7 reasoning repeatedly weighs that settled field. Because the tax provision and the telecom circular had been enacted while the challenge was pending, the bench faced the further question whether interim protection should freeze legislative development, and its answers differed by instrument: the statute’s tax limb survived on its own merits, the executive’s telecom limb fell with Section 57’s private use clause. The interim decade also explains the dissent’s sharpest institutional passage. Justice Chandrachud treated the sequence of assurances given and programmes announced as a single course of conduct, and measured the state’s compliance with the 2013 and 2015 orders against the enrolment and seeding drives of those years. On his account, a court whose protective orders can be waited out will find its final judgment received as one more interim event, and the passage reads as a warning addressed as much to future benches as to the Union.

What the Verdict Left Open

A judgment of 1,448 pages still leaves a defined set of questions open, and mapping them is part of reanalysing the verdict accurately. The majority sustained the use of the number under a law for purposes beyond Section 7 in principle, striking the contractual and corporate limb of Section 57 while leaving Parliament free to legislate a private or public use that could itself pass proportionality, so the boundary between permissible and impermissible uses was relocated to future statutes and future challenges rather than fixed in 2018. The bench did not have before it, and did not rule on, the design of state level databases that draw on the number, the National Register of Citizens processes then under way in Assam under separate orders of the Court, or the electoral roll purification proposals the Election Commission had aired in earlier years; the Lokniti Foundation order’s verification concerns were addressed only through the fall of the telecom circular. The verdict did not strike the collection of mobile numbers and email addresses at enrolment, treating them as optional fields, and it left the Authority’s grievance and data correction machinery to be judged by its operation, a deference the petitioners criticised because a resident locked out by a data error experiences the correction queue, not the statute. Nor did the bench fix a compensation route for a resident harmed by wrongful denial or data misuse, beyond opening the criminal complaint route through its reading of Section 47 and leaving civil claims to ordinary law. Each open question became a site of later contest: the voluntary use architecture of the 2019 amendment, the money bill reference of 2019, and continuing challenges to programme level mandates in the High Courts all proceed inside openings the 2018 bench deliberately or necessarily left. A reanalysis that treats the verdict as a full stop misreads its genre. It is better read as a framework judgment, nearer in kind to the 1973 basic structure decision than to a self executing code: it allocates burdens, names the test, supplies worked examples and leaves the system to iterate.

Administration After the Verdict: Circulars, Compliance and Correction

The verdict’s first effects were administrative rather than jurisprudential, and they test the operationalisation claim from the ground. Within weeks, the Authority issued circulars withdrawing the fields in which the number could be demanded as a condition: schools were told admissions could not require it, examination bodies withdrew it from application requirements, and the telecom department’s verification regime built since March 2017 was dismantled, with operators directed to delink numbers collected under the fallen circular where subscribers so chose. Banks adjusted account opening and re verification drives to the fall of the money laundering rule, treating the number as one document among others pending the legislative repair that arrived in 2019. Departments administering Section 7 programmes received the opposite instruction: seeding and authentication in welfare delivery continued, with reiterated directions that exemption mechanisms, fingerprint failure protocols, one time password authentication and manual exceptions be operated so that the assurance recorded in the judgment was honoured at the ration shop and the worksite. Two sets of circulars, moving in opposite directions within a single administrative system in a single season, are the pattern of the verdict made visible in paper form. The same number was withdrawn from the school gate and retained at the ration shop, not because administrators had changed their view of the technology, but because the test had sorted the uses by their justifications, and administration by circular followed the sorting. The compliance record also kept the dissent’s questions alive: reports from Jharkhand and Rajasthan in 2019 of continued authentication failures and contested deletions, documented by the same research networks that had supplied the case record, were argued in the High Courts and in the press as evidence that the majority’s proviso based answer depended on a correction machinery whose performance the judgment had assumed rather than verified. The verdict, that is, did not end the evidentiary argument. It institutionalised it, giving administrators a rule, residents a text to invoke and researchers a standard against which each season’s field reports could be measured.

The Four Opinions Read Together

The verdict is conventionally counted four to one, but the count flattens a finer distribution of agreement that litigants mining the opinions must respect. All five judges accepted the 2017 privacy holding as the governing premise and accepted proportionality as the method of review. Four judges, the three of the majority and Justice Bhushan, sustained the Act’s core: enrolment, Section 7, Section 139AA and Section 59. On Section 57 the four were together against the private limb, though their routes differed in emphasis, Justice Bhushan resting more heavily on the provision’s text and the majority on the absence of a legitimate aim for unbounded private use. On Sections 33(2) and 47 the reading down technique commanded the same four. Justice Chandrachud stood alone on the threshold, the design and the remedy: alone in holding the money bill certification fatal, alone in finding the architecture itself surveillance prone beyond saving construction, alone in holding Section 7 unconstitutional and the Act inseverable, so that in his hands the statute fell entire. The distribution matters for the doctrine’s future because the points on which the bench divided are the points on which its test left calibration open: the intensity of evidence a state must file at the necessity limb, the weight of executive assurance against field record, and the severability of an identity platform into uses. A later bench revisiting any of those calibrations will find, inside the 2018 record itself, a fully argued position at each setting of the dial, the majority’s at the deferential end and the dissent’s at the demanding end, with the concurrence marking a middle position on the evidence. Framework judgments of this kind do not merely decide cases. They stock the shelf from which later courts choose.

Identity Systems Before Other Constitutional Courts

India’s was not the only constitutional court asked to judge a national identity platform in the period, and the comparisons sharpen what the proportionality operationalisation achieved. The United Kingdom had abandoned its identity card scheme by legislation, the Identity Documents Act 2010 repealing the Identity Cards Act 2006 and ordering the destruction of the National Identity Register, a political resolution that left no doctrine behind. Jamaica’s Constitutional Court, in Julian Robinson v Attorney General of Jamaica, decided on 12 April 2019, struck down the National Identification and Registration Act 2017 in its entirety, holding its compulsory biometric registration disproportionate to its stated aims and its safeguards inadequate, a judgment whose reasoning tracks the four limbs closely and whose remedy, total invalidation, matches the Aadhaar dissent rather than the Aadhaar majority. Kenya’s High Court, ruling in January 2020 on the Huduma Namba registration, permitted the system to proceed subject to an adequate data protection framework being enacted first, a sequencing remedy the Aadhaar petitioners had sought and not obtained. The three outcomes, annihilation in Kingston, conditional suspension in Nairobi, calibrated reconstruction in New Delhi, display the range of answers a proportionality discipline can produce on materially similar statutes, and they locate the Indian verdict’s distinctiveness not in its leniency or severity but in its granularity: no other court in the set rebuilt the statute section by section from the bench. That granularity is the operationalisation reading’s core exhibit beyond India’s borders.

The Act’s Protective Chapters in the Majority’s Reading

Two chapters of the Act did quiet work in the majority’s balance and deserve record, because later argument has treated them as decoration. Chapter VI, Sections 28 to 31, imposed confidentiality obligations on the Authority, restricted the circumstances of access to the repository, required the Authority to keep identity information secure and obliged it to respond to a resident’s request to inspect and correct her own record, with Section 29’s absolute bar on sharing core biometric information and its prohibition on using identity information for any purpose outside the Act standing at the chapter’s head. Chapter VII created offences calibrated to the architecture’s feared abuses: impersonation at enrolment, unauthorised access to the repository, tampering with its data, unauthorised collection and disclosure by requesting entities, each punishable with imprisonment and fines, with Section 42 supplying a general penalty and Section 43 excluding the liability of courts acting in good faith under colour of the Act. The majority read these chapters into the proportionality balance as the safeguards component the 2017 plurality had required, treating the criminal fences as evidence that the legislature had addressed misuse in the design rather than leaving it to afterthought. It also read the regulations the Authority had framed under the Act, on enrolment procedure, authentication transactions and data security, as part of the operative scheme, while cautioning that executive regulations could be revisited as the scheme evolved. The dissent’s reply, that safeguards written and enforced by the institution they restrain answer a different question from the one proportionality asks, identifies the genuine soft point of the majority’s balance, and the 2019 amendment’s creation of an external adjudication route for penalties can be read as Parliament’s partial concession to it.

The Petitioners’ Coalition and the Shape of the Challenge

The challenge that reached the Constitution Bench was a coalition’s work, and its composition explains the case’s unusual evidentiary richness. Justice K. S. Puttaswamy’s 2012 petition attacked the scheme at its root, as a compulsory identity imposed without law. Later petitions added distinct fronts. Former army officer S. G. Vombatkere and activist Bezwada Wilson challenged the expanding mandates. Mathew Thomas, a former defence scientist, contested the reliability claims made for biometric de duplication and the enrolment contracts awarded to private agencies. Petitioners from Jharkhand and other states brought the ration shop record. The Internet Freedom Foundation and civil liberties litigants developed the surveillance and data protection fronts. A separate set of petitions attacked Section 139AA’s tax mandate, and the Binoy Viswam decision of June 2017 had upheld that provision against an Article 19(1)(g) challenge while expressly leaving the privacy question to the larger bench, a division of labour the 2018 bench respected. Because the petitions were connected and heard together, the bench received the scheme whole: its welfare promise, its fiscal claims, its field failures and its architectural risks argued by counsel who had spent years on separate parts of the record. Shyam Divan’s opening in January 2018 organised that record into the structure the dissent later adopted, beginning with the architecture, moving to the absence of a governing law for most of the scheme’s life, and closing on the interim orders’ disobedience. The Union’s reply, organised by Attorney General K. K. Venugopal across the hearing’s second half, met the coalition front by front, and the Authority’s presentation by its chief executive, walking the bench through enrolment screens, encryption and the repository’s access logs, was a procedural novelty the opinions cite by name. A reader tracing the verdict’s method should notice how the coalition’s shape maps onto the opinion’s shape: provision by provision review is what a court produces when a statute is attacked provision by provision by specialists, and the proportionality checklist was the instrument that let one bench hold the coalition’s many fronts inside a single analysis.

The Fiscal Claims and Their Critics

Between the hearing’s close in May 2018 and the verdict in September, the public argument over the scheme’s savings ran on figures that also sat, in affidavit form, inside the record. The government’s case for the suitability limb leaned on estimates that de duplication and direct transfer had saved large sums: official statements in 2018 placed cumulative savings across subsidy programmes in the tens of thousands of crores, and a World Bank estimate in its World Development Report 2016, suggesting that the identification platform could yield savings on the order of eleven billion dollars a year if applied across programmes, was quoted in the Union’s written submissions. Finance Minister Arun Jaitley defended the scheme in Parliament and in a published note of January 2019, after the verdict, presenting the number as the backbone of a reformed welfare state in which the middle man had been designed out. The critics’ answer, developed by Drèze, Khera and the economist Himanshu among others, was methodological: the official savings figures, they argued, conflated the deletion of duplicate and ineligible entries with savings attributable to the identity platform, counted reductions driven by unrelated price and eligibility changes, and treated every cancelled card as a fraud prevented rather than asking how many cancelled cards had belonged to genuine households. The majority did not adjudicate the econometrics. It held that the existence of duplication and leakage on a scale the official record documented was enough to ground the aim and suitability limbs, and that the precise quantum of savings was a legislative judgment it would not re perform. That abstention is a calibration choice of the kind the dissent refused: Justice Chandrachud held that where the state’s justification is quantitative, the state must prove its quantities, and that suitability asserted without proof fails the test the majority purported to apply. The fiscal argument thus survives the verdict unresolved in its numbers and resolved in its law: after September 2018, a government defending an identity mandate on savings must file its figures and survive cross examination on them, because the dissent stands in the reports as the standing objection to assertion in place of proof.

Severability: How the Bench Could Save the Act in Parts

The verdict’s provision by provision method rested on a doctrine Indian courts had refined in the first decade of the republic: severability. In R. M. D. Chamarbaugwalla v Union of India (1957), the Court had laid down that where the valid and invalid parts of a statute are separable, and the valid residue can operate as the legislature would have intended, the residue survives the excision of the part that falls. The Aadhaar petitioners contested severability at the level of the whole statute, arguing that the Act was one architecture whose parts drew meaning from each other, so that the failure of Section 57 or the money bill route had to bring the structure down. Justice Chandrachud accepted a version of that argument: on his analysis the Act’s provisions formed a single scheme for a single database, the money bill defect infected the enactment itself, and nothing remained to sever because the falling part was the Act’s foundation. The majority proceeded on the opposite premise. It treated the Act as a bundle of authorisations addressed to distinct uses, welfare delivery, taxation support, disclosure, prosecution and private verification, each capable of standing on its own constitutional footing, and it severed accordingly: the private limb out of Section 57, the longer retention out of the regulations, the offending rules out of the telecom and money laundering frameworks, and the rest left standing.

That choice, more than any single holding, determined what kind of event the verdict was. A total invalidation on the Jamaican pattern would have thrown a billion enrolments, the welfare payment rails built since 2013 and the tax system’s identifier into legal suspension on a single morning, and the majority’s reluctance to produce that result is legible across its pages, in its treatment of settled administrative expectation and in its preference for construction over destruction wherever a provision could bear a constitutional reading. Critics, Bhatia most systematically, read that reluctance as the verdict’s original sin, a remedial caution that disciplined the analysis backwards: because the bench would not contemplate the system’s fall, each provision’s review was softened until the system survived. Defenders answer that severability and reading down are not softness but the ordinary craft of a court reviewing a statute of mixed purposes, and that a bench which destroyed the welfare core because the private limb failed would have committed the opposite error, sacrificing a justified use to punish an unjustified one. The proportionality architecture holds the two readings apart with precision. The test’s unit of analysis is the provision and its purpose. A court that accepts that unit has already accepted severability in substance, because purposes fall one at a time. Justice Chandrachud’s deepest move was therefore not his surveillance analysis but his refusal of the unit: his insistence that the architecture, not the authorisation, was the thing to be judged. The disagreement between the opinions is, at bottom, a disagreement about the size of the object constitutional review should take, and the 2018 verdict will be cited on both sides of that question for as long as platform statutes are challenged.

The Scholarly Adjudication and the Claim

Three readings compete for the verdict, and the scholarship states them cleanly. The outcome reading dominates reference works: a divided court upheld the scheme, trimmed its private reach and read safeguards into its disclosure and complaint provisions. The dissent reading dominates commentary: Gautam Bhatia, in The Transformative Constitution (2019), treats the Aadhaar case as a missed transformation, arguing that the majority’s deference on the evidence and its section by section method dissolved a structural challenge into a series of small administrative questions, and that Justice Chandrachud’s dissent better honoured the 2017 judgment’s premises about the relationship between the individual and the data state. Suhrith Parthasarathy, writing in the Economic and Political Weekly after the verdict, pressed the money bill point and the burden of proof point together, contending that a test which asks the state to justify itself cannot be satisfied by the state’s own assertions about its system’s design. Pratap Bhanu Mehta’s essays on the verdict and on the Court’s broader pattern in the same period situated the case within a judicial reluctance to confront the executive on large questions of state architecture, reading the outcome as institutional caution in doctrinal dress.

The proportionality operationalisation reading defended in this article does not dismiss those accounts; it relocates their object. Bhatia’s critique of deference is, in its detail, a critique of how the majority filled the necessity and balancing limbs with the Union’s affidavit material, and it bites hardest exactly where the test was filled thinly, on exclusion. Parthasarathy’s objection is a limb objection: he argues the aim was underspecified and the evidence untested, which is to say the test was applied at a lower intensity than its own logic demands. Mehta’s institutional caution thesis explains why a bench might apply a demanding test at low intensity to a system on which welfare delivery already depended. Each critic, that is, argues inside the architecture the verdict built, disputing the calibration of its limbs rather than proposing a different instrument. That convergence is the strongest available evidence for the claim this article defends: the 2018 Aadhaar verdict operationalised proportionality architecture for the first major application after Puttaswamy, and the architectural pattern, a purpose by purpose discipline of aim, suitability, necessity and balance, applied to sections rather than to schemes, with reading down as its characteristic remedy and the dissent preserved inside it as a stated alternative calibration, shapes the trajectory that followed, from the 2019 amendment drafted to its terms to the seven judge reference that will revisit its money bill threshold. The verdict outcome told India what survived in September 2018. The operationalisation tells India how every successor question will be asked.

A Note on Method: Judgments Stated by Holding

Because the four opinions run to 1,448 pages and disagree in calibration while agreeing in method, this article states each position by its holding and attributes each characterisation to its holder. The majority’s holdings are taken from Justice Sikri’s operative conclusions, Justice Bhushan’s from his concurring conclusions, and the dissent’s from Justice Chandrachud’s concluding summary of the grounds on which he would have allowed the petitions. Where the opinions use a phrase that has become canonical, the fraud on the Constitution in the dissent, the phrase is reported as the dissent’s own and is not adopted in the article’s voice. Where the record is contested, the welfare savings figures, the cause of contested deletions in Jharkhand, the exclusion rates in competing surveys, the contest is reported as a contest, with each side’s source named. That discipline is not decoration. A verdict whose entire method consists of asking who must justify what, and on what evidence, can only be reanalysed fairly by an article that submits to the same discipline in its own sentences.

Frequently Asked Questions

Q: What did the 2018 Aadhaar verdict decide?

On 26 September 2018 a five judge Constitution Bench upheld the Aadhaar Act by four votes to one in Justice K. S. Puttaswamy (Retd.) and Another v Union of India. Justice A. K. Sikri’s majority sustained the scheme and Section 7, struck down the limb of Section 57 that allowed bodies corporate and persons to use the number for any purpose under contract, read down Section 33(2) to add a judicial element to national security disclosure, and read down Section 47 to let an aggrieved person complain of offences. It upheld the passage of the Act as a money bill and the tax provision in Section 139AA. Justice Ashok Bhushan concurred. Justice D. Y. Chandrachud dissented, holding that he would have struck the Act down entirely. The holdings are stated in full in the four opinions running to roughly 1,448 pages.

Q: What is Section 57 of Aadhaar Act?

Section 57 was the Aadhaar Act’s general use clause. It provided that nothing in the Act prevented the use of the Aadhaar number for establishing the identity of a person for any purpose, whether by the state or by any body corporate or person, under any law in force or under any contract. In the years before September 2018 the clause underwrote the spread of the number into private life: telecom operators, banks and other companies built enrolment and e-KYC verification on it, and the Department of Telecommunications circular of March 2017 pressing the number for mobile connections leaned on the atmosphere the clause created. The 2018 majority struck down the body corporate and person limb, holding that identity establishment for any purpose under private contract had no stated public aim to put before the proportionality test. Use authorised by a law survived in narrowed form. The Aadhaar and Other Laws (Amendment) Act 2019 then omitted Section 57 from the statute altogether.

Q: What is Section 7 of Aadhaar Act?

Section 7 is the Aadhaar Act’s welfare delivery provision and, on the majority’s reading, the statute’s core. It empowers the central government or a state government to require a person receiving a subsidy, benefit or service funded from the Consolidated Fund of India or of a state to furnish proof of possession of an Aadhaar number, or to undergo authentication, as a condition of receipt. Its proviso protects a person to whom no number has yet been assigned: such an applicant must be offered alternate and viable means of identification, and the Union assured the bench that benefits would not be denied for want of enrolment. The majority tested the section against the four limbs of proportionality, legitimate aim in the integrity of welfare expenditure, suitability in de duplication, necessity against the alternatives proposed, and balance against the privacy cost as fenced by the Act, and upheld the section on each limb.

Q: What was Justice Chandrachud’s dissent?

Justice D. Y. Chandrachud’s dissent, the longest of the four opinions delivered on 26 September 2018, would have struck down the Aadhaar Act in full. He held first that the Act’s passage as a money bill bypassed the Rajya Sabha on a statute ranging far beyond Article 110’s enumerated subjects, and he called that certification a fraud on the Constitution. On the merits he held that the state had not proved its case under proportionality: the aim was underspecified for a general identity platform, suitability was unproven because leakage in welfare programmes was not shown to be identity fraud, less restrictive alternatives had not been tried, and the balance failed because authentication records created a surveillance capacity and exclusion fell on the poorest beneficiaries. He held Section 7 itself unconstitutional, treating entitlements as rights that could not be conditioned on enrolment in a failure prone system, and he recorded that the state had treated the Court’s interim orders of 2013 and 2015 as obstacles rather than as holdings.

Q: What is the Money-Bill question?

The Money-Bill question asks whether the Aadhaar Act could constitutionally be enacted through the money bill procedure of Articles 109 and 110. A money bill, certified by the Speaker of the Lok Sabha, can only be recommended upon by the Rajya Sabha, not amended or rejected, and Article 110 confines the category to bills containing only provisions on enumerated fiscal subjects such as taxation, borrowing and expenditure from the Consolidated Fund. The Aadhaar bill was certified as a money bill in March 2016; the Rajya Sabha’s five recommended amendments were rejected by the Lok Sabha the same day. The 2018 majority upheld the certification, holding that Section 7’s welfare expenditure was the Act’s core and the remaining provisions incidental machinery. Justice Chandrachud held the Act ranged far outside Article 110. On 13 November 2019 the Rojer Mathew bench doubted the majority’s reasoning and referred the interpretation of Article 110 to a seven judge bench, leaving the holding standing but expressly unsettled.

Q: How does proportionality apply to Aadhaar?

Proportionality is the four limb test the bench applied provision by provision. A measure invading privacy must pursue a legitimate aim; the means chosen must be suitable, rationally connected to that aim; no equally effective and less restrictive alternative may be available; and the measure’s benefit must outweigh its cost to the right, with procedural safeguards against abuse. The test came into the case from the nine judge privacy judgment of 24 August 2017 and from Modern Dental College (2016). Applied to Aadhaar it produced different answers for different provisions. Section 7 passed all four limbs on the strength of its welfare aim, its de duplication mechanism and its proviso for the unenrolled. Section 57’s private limb failed at the aim, necessity and balancing limbs because unbounded private use stated no public purpose. Sections 33(2) and 47 were read down, the bench supplying the safeguards, a judicial element and a victim’s complaint route, whose absence had made each provision excessive as written.

Q: Did 2019 amendments restore Aadhaar uses?

In part, and on the verdict’s terms. The Aadhaar and Other Laws (Amendment) Act 2019 omitted Section 57, the clause the bench had severed, and wrote a fresh voluntary architecture in its place: a resident may choose Aadhaar authentication or offline verification to establish identity, may not be denied a service for declining where an alternative is offered, and a child enrolled on parental consent may cancel the number within six months of turning eighteen. Amendments to the Indian Telegraph Act and the Prevention of Money Laundering Act allowed telecom providers and banks to accept the number as one voluntary identification document among others, restoring part of the private verification space the verdict had closed to compulsion and contract. The amendment also gave the Authority civil penalty powers and an adjudication route. What it did not restore was compulsion: mandatory mobile and bank account linking, struck down in September 2018, was not re enacted.

Q: What did the verdict decide about linking the number to mobile phone connections?

The majority held the linkage requirement unconstitutional. The requirement rested on a Department of Telecommunications circular of 24 March 2017, which had directed the verification of all mobile subscribers against the number, leaning on an order of 6 February 2017 in Lokniti Foundation v Union of India concerning the verification of existing subscribers. With the private use limb of Section 57 struck down, the circular lost its foundation: a general identity platform for every telephone connection had not been justified under the proportionality limbs, and the state had not shown why the verification aim required biometric backed linkage of the entire subscriber base. The circular was quashed in consequence. Telecom operators dismantled the linkage regime that had operated since 2017, and subscribers who had linked under compulsion could seek delinking. The 2019 amendment later restored the number only as a voluntary document a subscriber may offer for a new connection.

Q: Why did the bank account linking requirement fail in the verdict?

The requirement that every bank account be linked to the holder’s Aadhaar number came not from the Aadhaar Act itself but from the Prevention of Money Laundering (Maintenance of Records) Rules, amended in June 2017 by executive notification. The majority struck the rule down as disproportionate. A delegated rule had imposed, on pain of accounts being frozen, a universal linkage between the banking system and the biometric platform, and the Union had not demonstrated that so sweeping a measure was necessary to the anti money laundering aim or that less restrictive identification routes could not serve it. The bench also held that the rule could not borrow authority from Section 57 after that section’s private limb had been struck down. Account holders who had linked under the rule’s compulsion were left free of the condition, and the Reserve Bank’s know your customer framework continued on other documents. The 2019 amendment to the money laundering statute later admitted the number as a voluntary identification document, a route the verdict’s reasoning left open to Parliament.

Q: What did Section 33(2) permit before the bench read it down?

As enacted, Section 33(2) allowed the disclosure of a resident’s identity information and authentication records, in the interest of national security, on the direction of an officer not below the rank of Joint Secretary to the Government of India. The provision required no court order, no judicial participation, no hearing for the person affected and no recorded process beyond the direction itself, in contrast to Section 33(1), which routed other disclosures through an order of a court not inferior to a district judge. The petitioners attacked the limb as executive self authorisation over the most sensitive class of data in the statute. The majority declined to strike it down, accepting that a national security ground could be legitimate, but read it down so that a direction under the provision would carry a judicial element in the decisional process. It separately read Section 33(1) to require that the person concerned be heard before a court ordered disclosure against her.

Q: How did the verdict change Section 47 on who may file a complaint?

As enacted, Section 47 barred any court from taking cognisance of an offence under the Aadhaar Act except on a complaint made by the Unique Identification Authority or by an officer or person authorised by it. The design left a resident whose data had been misused, by an enrolment agency, a requesting entity or an official, without a route of her own to start a prosecution under the very statute that criminalised the misuse in its offences chapter. The majority read the section down to permit a complaint by an aggrieved person, reasoning that a provision which made the institution the sole gatekeeper of complaints, including complaints about the institution’s own ecosystem, left victims remediless and failed the safeguards component of proportionality. The reading did not rewrite the offences or their penalties. It widened the door to the court, so that the criminal fences the majority had counted in the statute’s favour could be invoked by the people those fences existed to protect.

Q: What limits did the verdict place on retaining authentication records?

Each authentication generates a record held by the Authority: the requesting entity, the time and the response. The regulations framed under the Act had permitted the Authority to archive authentication transaction records for five years, a retention the petitioners attacked as the raw material of profiling. The majority accepted part of that argument in remedial form. It held that authentication records were not to be retained beyond six months, and struck down the archival regulation to the extent that it authorised the longer period. It also directed that the Authority not store the purpose of a transaction or build a repository of metadata capable of tracking a resident across requesting entities, recording the Authority’s own assurance that its systems captured no purpose data. Justice Chandrachud’s dissent treated the six month concession as insufficient, because records distributed across requesting entities and regenerated at each fresh authentication could not be contained by a retention cap on the central store alone.

Q: What did the verdict hold about children enrolled in the scheme?

Children were enrolled on the consent of a parent or guardian, and the petitions had raised both the taking of a child’s biometrics and the uses to which schools put the number. The majority held that enrolment of a child rested on parental consent, that the consent framework required the option for the child, on attaining majority, to exit the system and have the number cancelled, and that the number could not be demanded as a condition of school admission. It further held that bodies conducting examinations and admissions processes, naming the Central Board of Secondary Education, the National Eligibility cum Entrance Test and the University Grants Commission in the demands brought before it, could not make the number compulsory for candidates, because those demands lacked the statutory anchoring that Section 7 gave to welfare delivery. Departments and examination bodies withdrew admission and application requirements in circulars issued after September 2018, and the exit on majority was written into the statute by the 2019 amendment.

Q: Why did Justice Bhushan write a separate concurring opinion?

Justice Ashok Bhushan agreed with the majority’s conclusions and reached them by a route he set out in his own opinion of over two hundred pages. Concurring opinions in Constitution Bench practice serve two purposes: they record that the agreement is considered rather than borrowed, and they let a judge develop reasoning the lead opinion compresses. Justice Bhushan restated the privacy analysis from the 2017 judgment, examined the enrolment and authentication safeguards in the Act and its regulations at greater length than the majority had done, gave the Union’s affidavit material on welfare delivery fuller evidentiary weight, and addressed the money bill question through the character of Section 7 and the Speaker’s certificate. His conclusions tracked the majority’s on the Act’s validity, on Section 57, and on the readings down, which is why the operative result of 26 September 2018 is counted as four to one rather than three to one with an abstention.

Q: What did the verdict decide about Section 139AA and income tax returns?

Section 139AA of the Income Tax Act 1961, inserted by the Finance Act 2017, required a person eligible for an Aadhaar number to quote it in income tax returns and in applications for a permanent account number. A three judge bench had upheld the provision in Binoy Viswam v Union of India (2017) against a challenge under the freedom of trade, while leaving the privacy challenge to the Constitution Bench. In September 2018 the majority upheld Section 139AA on the privacy front as well, holding that the measure served the legitimate aim of curbing duplicate permanent account numbers and false returns, that quoting a number the state had issued for identification was a suitable and bounded means, and that the intrusion weighed less than the measures struck down elsewhere in the case because it compelled disclosure of the number to the tax authority rather than authentication across open ended settings. The requirement therefore survived the verdict intact, and tax Aadhaar linkage continued on its statutory timetable.

Q: How did the Rojer Mathew reference affect the money bill holding?

In Rojer Mathew v South Indian Bank Ltd, decided on 13 November 2019, a five judge bench examined tribunal reforms enacted through the Finance Act 2017 and confronted the Aadhaar majority’s money bill reasoning directly. The bench recorded that the Aadhaar decision had not articulated a clear test for judging whether a composite statute’s provisions answer to Article 110’s enumerated heads, doubted the breadth of the incidental provisions approach on which the Aadhaar Act had been sustained, and referred the interpretation of Article 110 to a bench of seven judges. The reference did not overrule the Aadhaar holding, which continues to govern the Act’s validity, and the 2018 verdict’s privacy and proportionality holdings were not in question in Rojer Mathew at all. The effect is a holding in force with a recorded doubt against its reasoning: the money bill component of September 2018 stands as law while the Court’s larger bench prepares to restate the doctrine, an outcome that preserves Justice Chandrachud’s dissent as a live argument inside the system.

Q: What did the verdict say about Aadhaar as proof of citizenship?

The majority treated the point as settled by the statute’s own text. Section 4(3) of the Aadhaar Act provides that an Aadhaar number, or its authentication, is not by itself proof of citizenship or domicile. The number certifies identity on the basis of the enrolment record, which rests on documents or introducer based enrolment for residents, a category defined by residence in India for 182 days in the preceding twelve months rather than by nationality. The bench recorded that position without disturbance, and its reasoning throughout treats the number as an identity instrument for service delivery, not as a status document. That holding bounded the verdict’s field: arguments about citizenship registers and documentation of nationality, live in other proceedings of the period, were not decided by the Aadhaar bench, and the verdict neither authorised nor forbade the use of enrolment data in such processes, which would have to be tested, if proposed, on their own statutory footing.

Q: Which judges sat on the Constitution Bench that decided the case?

The bench of five was headed by Chief Justice Dipak Misra, sitting with Justice A. K. Sikri, Justice A. M. Khanwilkar, Justice D. Y. Chandrachud and Justice Ashok Bhushan. The composition carried a thread into the case’s prehistory: Justice Chandrachud had authored the lead plurality opinion in the nine judge privacy judgment of 24 August 2017, so the judge who had written privacy into the Constitution’s structure at its fullest was a member of the first bench to apply that right to a working state system. Justice Sikri wrote the majority opinion for himself, the Chief Justice and Justice Khanwilkar. Justice Bhushan concurred separately. Justice Chandrachud dissented alone. The bench had been constituted in December 2017 after the interim order years, heard argument for 38 days between 17 January and 10 May 2018, and delivered judgment on 26 September 2018, two days before Chief Justice Misra’s retirement on 2 October 2018.

Q: What changes did the 2019 amendment make for offline verification?

The Aadhaar and Other Laws (Amendment) Act 2019 created offline verification as a statutory alternative to online authentication. Under the amended framework, a resident may establish identity by producing Aadhaar data in a verifiable offline form, through means the Authority notifies, such as the digitally signed secure code the Authority issues, without an authentication call to the central repository and therefore without generating an entry in the authentication record. The amendment made the choice the resident’s: use of the number, online or offline, was written as voluntary across the Act, a service could not be denied for refusal where another identification was offered, and entities performing offline verification were barred from collecting, using or storing the number itself beyond the notified process. The device answered a strand of the 2018 argument directly, because a verification that leaves no central log narrows the surveillance objection the dissent had pressed, and the majority’s necessity analysis had treated the absence of such alternatives as a point the state needed to meet in future designs.

Q: What did the verdict hold about Section 59 and acts done before the Act?

Between the Authority’s creation in January 2009 and the Act’s assent in March 2016, the scheme enrolled over a billion residents under an executive notification, with no governing statute. Section 59 validated that past: it provided that anything done or any action taken under the notification of 28 January 2009 would be deemed to have been done under the corresponding provisions of the Act. The petitioners attacked the deeming as a legislative cure for seven years of unconstitutional collection, arguing that the absence of law during collection was the defect the privacy judgment’s legality limb condemned. The majority upheld Section 59, holding that a legislature enacting a valid governing statute could validate past administrative action taken under the regime the statute replaced, and that the enrolments stood on residents’ applications rather than on compulsion the record established for the pre Act period. The holding kept the entire enrolment base legally intact, which the scheme’s operation required whichever way the other provisions were decided.